as“

Diet and IronStatus, A Studyof Relationships:United States,1971-74.This reporl presents findings of the first

National Health and Nutrition Examination

Survey on the iron status of the U.S.

population 1-74 years of age during

1971-74. The relationship between sever-

al dietary, health, and socioeconomic vari-

ables and iron status also are presentedin this report for selected population

groups.

Data From the National HealthSurveySeries 11. No. 229

DHHS Publication No. (PHS) 83-1679

U.S. Department of Health and Human

Services

Public Health Service

National Center for Health Statistics

Hyattsvil Ie, Md.

December 1982

National Center for Health Statistics

ROBERT A. ISIWEL, Acting Director

JACOB J. FELDMAN, Ph. D., Associate Director forAnalysis and Epidemiology

GAIL F. FISHER, Ph. D., Associate Director for theCooperative Health Statistics System

GARRIE J. LOSEE, Associate Director for DataProcessing and Services

ALVAN O. ZARATE, Ph. D., Assistant Director forItzternational Statistics

E. EARL BRYANT, Associate Director for Interviewand Examination Statistics

ROBERT L. QUAVE, Acting Associate Director forManagement

MONROE G. SIRKEN, Ph. D., Associate Director forReseawh and Methodology

PETER L. HURLEY, Associate Director for Vital andHealth Care Statistics

ALICE HAYWOOD, Information Oficer

Interview and Examination Statistics Program

E. EARL BRYANT, Associate Directtir

MARY GRACE KOVAR, Special Assistant forPolicy and Analysis

Division of Health Examination Statistics

ROBERT S. MURPHY, Director

Data

SIDNEY ABIWHAM, ChieJ Nutrition StatisticsBranch

JEAN ROBERTS, Chief Medical Statistics Branch

KURT R. MAURER, ChieJ Survey Planning andDevelopment Branch

Data Processing and Services Program

GARRIE J. LOSEE, Associate Director

ALAN K. KREGER, ChieJ Computer Users Stay

Division of Data Services

JAMES C. JACKS, Ph. D., Director

PHILLIP R. BEATTIE, Deputy Director

DAVID L. LARSON, ChieJ Health Examination FieldOperation Branch

Cooperation of the U.S. Bureau of the Censu~

Under the legislation establishing the National Health Survey, the Put,licHealth Sewice is authorized to use, insofar as pnssible, the services or facilitiesof other Federal, State, or private agencies. In accordance with specificationsestablished by the National Center for Health Statistics, the U.S. Bureau ,Df the

Census participated in the design and selectinn of the sample and carried outthe household interview stage of the data colkclion and certain parts of thestatistical processing.

Foreword andacknowledgments

The National Health and Nutrition ExaminationSurvey is the only source of general U.S. populationdata that provides a direct link between indicators ofhealth and nutritional status and reported dietaryintake information. Congress provided resources in the“Departments of Labor, Health and Human Services,and Related Agencies Appropriation Bill, 1980” to theNational Center for Health Statistics to fund aninitiative to undertake more detailed analyses ofnutrition-related health problems measured in the firstNational Health and Nutrition Examination Survey.As part of this initiative, the Division of HealthExamination Statistics and the Division of Nutrition,Center for Health Promotion and Education, Centersfor Disease Control, funded a contract (no.233-79–209 1) with Abt Associates, Inc. in Cambridge,Mass., to examine relationships among dietary intakeof iron and measures of anemia and to relate theprevalence of anemia to various socioeconomic anddemographic variables.

The approach and depth of analysis presented inthis report differ from most reports from the Divisionof Health Examination Statistics. This report is basedon a statistical rather than a descriptive presentation ofthe data.

Cognizant that the underlying assumptions oftraditional statistical analyses are violated to someextent, the degree of which is unknown, the authorsand National Center for Health Statistics staff jointlydetermined that the assumptions made in the analysespresented in this report are reasonable in light ofpresent knowledge. In addition, the authors havepresented material throughout the text and technicalappendix that concerns appropriate qualificationsreaders should consider in interpreting the results andconclusions presented.

Analyses for this report were undertaken by thestaff of Abt Associates, Inc., with the assistance ofseveral consultants in the fields of nutrition, hematolo-gy, and biochemistry. Dr. Linda Meyers, Dr. Jean-Pierre Habicht, and D. Yvonne Jones of Cornell

University investigated relationships between socioeco-nomic status and iron status. Drs. Meyers and Habichtalso helped to select appropriate definitions of anemiaand to clarify the overall process of iron metabolism.Dr. James Cook of the University of Kansas MedicalCenter was instrumental in developing an algorithmwhen applied to dietary intake data-estimated avail-able iron. Dr. Peter Dallman of the University ofCalifornia, San Francisco, School of Medicine contrib-uted especially to the knowledge of pediatric hematol-ogy, prevalence of anemia, and racial differences iniron status. In addition, Dr. Jack Smith of theSwanson Center for Nutrition, Omaha, Neb., providedoverall technical guidance, particularly in the area ofbiochemistry. All of these consultants also helped toselect variables to be investigated and to determinesubgroups on which to focus the analyses. Theirassistance throughout the investigation is greatlyappreciated.

Several Abt Asociates, Inc. staff members alsogave invaluable support. In particular, CatharineBarclay directed the data processing and performedanalyses for children ages 1–3 years; Dr. BarbaraDillon Goodson provided statistical support and per-formed analyses for menstruating women ages 12–54years; and Dr. Patricia Granahan undertook a reviewof the literature as well as analyses for elderly personsages 65–74 years. Additional computer support wasprovided by Dr. Charles Cole, who developed aprogram to estimate absorbable iron. Dr. John Himesand Mary Kay O’Neill Fox helped to prepare andreview the final manuscript and tables. The support ofthese persons was instrumental to the successfulcompletion of the project.

A very special acknowledgment goes to Ms. PennySpingam, who provided invaluable assistance in thesubstantive editing and revising of this report.

Robert S. MurphyDirectorDivision of Health Examination Statistics

...Ill

Contents

Foreword and acknowledgments ...................................................................................................................................................

Introduction ......................................................................................................................................................................................

Highlights ..........................................................................................................................................................................................

Background—Review of the literature ...........................................................................................................................................Anemia and iron status ............................................................................................................................................................Iron nutrition in the individual ..................................................................................................................................................Groups at risk of iron deficiency and anemia .........................................................................................................................

Preschool children ............................................................................................................................................................Women of childbearing age .............................................................................................................................................The elderly ........................................................................................................................................................................

Methodology Sources and limitations of the data ........................................................................................................................Survey procedure .....................................................................................................................................................................Biochemical determinations ....................................................................................................................................................Assessment of iron status .......................................................................................................................................................Calculation of dietary intake and absorbable iron .................................................................................................................Relational analyses: Analytical subgroups, independent variables, and tests of statistical significance .........................

Hndings ............................................................................................................................................................................................Iron status .................................................................................................................................................................................

Age and sex ......................................................................................................................................................................Race ..................................................................................................................................................................................Pregnancy .........................................................................................................................................................................

Iron intake and iron status .......................................................................................................................................................

Age, sex, and race ............................................................................................................................................................Pregnancy .........................................................................................................................................................................

Relational anaIyses ..................................................................................................................................................................Children ages 1-3 years ..................................................................................................................................................Menstruating women ages 12-54 years .........................................................................................................................Elderly persons ages 65-74 years ..................................................................................................................................

Dscussion ........................................................................................................................................................................................Iron status in the United States ................................................................................................................................................Diet ............................................................................................................................................... ......................... ....................Socioeconomic, demographic, and health factors ................................................................................................................

References .......................................................................................................................................................................................

List of detailed tables ......................................................................................................................................................................

.,.Ill

1

2

3345566

777789

121212122222222222232527

28202929

31

34

iv

Appendixes

1. Statistical notes ................................................................................................................................................ ......................... ..H. Definitions of terms ........................................................................................................................................................ .............

List of text figures

1. Prevalence of low hemoglobin levels, by sex and age: United States, 1971.74 ......... .......................................................2. Prevalence of low hemoglobin levels among males, by race and age: United States, 1971-74 ......................................3. Prevalence of low hemoglobin levels among females, by race and age: United States, 1971 -74 ...................................4. Prevalence of low transferring saturation levels, by sex and age: United States, 1971-74 ................ ................................5. Prevalence of low transferring saturation levels among males, by race and age: United States, 1971-74 .......................6. Prevalence of low transferring saturation levels among females, by race and age: United States, 1971–74 ....................

7. Prevalence of low hemoglobin and low transferring saturation levels, by sex and age: United States, 1971 -74 .............

8. Prevalence of low hemoglobin and low transferring saturation levels among males, by race and age: United States,1971-74. . ...................................................................................................................................................... ............... .............

9. Prevalence of low hemoglobin and low transferring saturation levels among females, by race and age: United States,1971.74 ........................................................................................................... .........................................................................

10. Mean iron intake for males ages 1-74 years, by age and hemoglobin levels: United States, 1971 -74 ............................11. Mean iron intake for females ages 1-74 years, by age and hemoglobin levels: United States, 1971 –74 ........................

12. Mean iron intake for males ages 1-74 years, by age and transferring saturation levels: United States, 1971 -74 ........ ....13. Mean iron intake for females ages 1-74 years, by age and transferring saturation levels: United States, 1971 -74 ........

List of text tables

A. Age- and sex-specific hemoglobin reference standards ..................................................................... ...................................B. Z-statistics for comparison of iron intake for low and adequate iron status for pregnant women 1844 years, with

mean, and standard error of the mean, by iron status indicator and race: United States, 1971 -74 ...................................

75

81

1314

15161710

19

20

2123242526

8

22

Symbols used in tables

. . . Data not available

. . . Categow not applicable

Quantity zero

0.0 Quantity more than zero but lessthan

0.05

z Quantity more than zero but lessthan

500 where numbers are rounded tothousands

● Figure does not meet standards ofreliability or precision

# Figure suppressed to comply with

confidentiality requirements

Diet and Iron Status, AStudy of Relationshipsby Judith D. Singer, M.A., Patricia Granahan, SC.D., NancyN. Goodrich, Ed.D., Abt Associates, Inc., Linda D. Meyers,Ph.D., Cornell University and Clifford L, Johnson, M.S,P.H.,Division of Health Examination Statistics

Introduction

This report presents the results of a study of ironstatus and selected dietary, health, socioeconomic, anddemographic correlates. Findings are based on datafrom the first National Health and Nutrition Examina-tion Survey; the CataZog of Publications of the NationalCenter for Health Statistical should be consulted ifprevious reports are of interest to the reader.

The first National Health and Nutrition Examina-tion Survey was one in a series of related programscarried out by the National Center for Health Statis-tics under authorization by Congress in the NationalHealth Survey Act of 1956. The programs characteris-tically are national in scope, based on probabilitysampling, and used to collect a broad range ofmorbidity data and related health information. Theessential differentiating characteristic of the healthexamination surveys is their primary concern withhealth-related data obtained only (or at least optimal-ly) from specially standardized direct medical exami-nations, including tests and other procedures used inclinical practice. Examinations given to persons select-ed in the scientific sample permit estimates of the totalprevalence of specifically defined diseases in the U.S.population. They also permit estimation of the distri-bution within the population of a broad variety ofhealth-related measurements, including physical meas-urements, such as height, weight, and various skinfoldsas well as physiological measurements, such as diastol-ic blood pressure, serum cholesterol level, and psycho-logical measurements.

The first National Health and Nutrition Examina-tion Survey and earlier examination studies conductedby the Center are unique in that specially designed andconstructed mobile examination centers were moved tothe selected sample areas so that standard physicalexaminations could be given under standard condi-tions. The team of interviewers and examiners alsomoved with the examination centers.

Standard physical examinations of a probabilitysample of the population have several advantages overother means of collecting data about the health of thepopulation:

The examination can be designed to focus uponspecified conditions and furnish more reliableestimates of the total prevalence of the conditionsthan can be obtained by interview. Cases unknownto the examinee can be identtied, and the criteriafor diagnoses can be specified carefidly.

The use of mobile examination centers and a smallteam of specially trained personnel reduces thevariability that is found in studies using localfacilities and personnel.

The instruments are recalibrated, and interviewingand examining are monitord so the only varianceshould be the person examined.

The probability design reduces the potential biasthat might be found in populations selected forother specific purposes.

The first National Health and Nutrition Examina-tion Survey was conducted from April 1971 to June1974 and included medical, dermatological, and dentalexaminations; body measurements; dietary interviews;medical histories; and demographic interviews of anational probability sample. A detailed description ofthe specific content and plan of operation of thesurvey, inclu~ing the sample design and forms used tocollect the dap has been published. 2

In this rep@t, data from the biochemical determi-nations were used to determine the prevalence of threecategories of iron status—low hemoglobin, low percenttransfernn saturation, and low hemoglobin combinedwith low percent transferringsaturation. The dietaryinterviews provided data to investigate the iron intakeof persons in these,categories. Other analyses exploredthe association between selected demographic, health,and dietary characteristics and the three categories ofiron status defined in this report.

1

Highlights

The analyses of the data from this national surveyby various subpopulations showed that:

Iron-deficiency anemia, measured by the combina-tion of low hemoglobin and low transferring satura-tion levels, is not widespread in the United States.

The prevalence of low hemoglobin is most com-mon in young children, pregnant women, elderlymen, and the black population.

The prevalence of low hemoglobin, low transferringsaturation, and the combination of these twomeasures is marked among children ages 1–3years.

Of the three measures of iron status, low transfer-ring saturation, which indicates that iron status isless than optimal but has not necessarily deterio-rated to the point of causing anemia, is the mostprevalent. It is most common in young childrenand women of childbearing age.

Reported iron intake is below recommended allow-ances for many subpopulations. The relationship be-tween reported dietary iron intake and the three ironstatus categories was examined for persons ages 1–74years and was analyzed by age, race, sex, and pregnan-cy status.

●

●

●

Except among young children, there is no strongrelationship between reported iron intake and ironstatus as measured by hemoglobin levels or trans-ferring saturation levels.

Among children ages 1–3 years, for whom report-ed iron intake and iron status are associated, meaniron intake and an estimate adjusted for availabili-ty were examined, and the adjustment for avail-ability did not affect the association.

Failure to find associations between reported iron

intake and iron status may be due to problems inthe method used to collect dietary information orto problems inherent in the use of referencestandards to classify persons into iron statuscategories. Only a small portion of ingested iron isabsorbed, which may compound the problems.

Dietary, health, socioeconomic, and demographiccharacteristics that may affect iron status in threehigh-risk groups-children ages 1–3 years, menstruat-ing women ages 12–54 years, and persons ages 65–74years—were examined in multivariate logistic ar[aly-ses.

●

●

Two socioeconomic variables, a poverty index andlow education of the head of household, areassociated with poor iron status.

Low iron intake is significantly associated with lowhemoglobin levels among children ages 1–3 years.

Previous pregnancy and diagnosis of anemia areassociated with poor iron status among menstruat-ing women; the use of oral contraceptives isassociated with adequate iron status. None of’ thevariables related to dietary habits is associated ‘withany of the iron status categories for this group.

Poor general health and trouble eating are thevariables most associated with poor iron statusamong persons ages 65–74 years. However, theassociation was not consistent in all categories, Novariables related to dietary habits are associatedwith iron status in this population.

Cross-sectional survey data cannot be interpretedas showing cause and effect relationships. However,the analyses between iron status and the selecteddietary, socioeconomic, and health variables indicatedfactors in addition to reported iron intake that shouldbe investigated further.

Background—Review of theliterature

Anemia and iron status

Anemia generally is believed to be the majornutritional deficiency problem in the United Statestoday.3-s However, estimates of the prevalence ofanemia vary widely;~ for example, estimates of theprevalence of anemia range from 17 to 44 percentamong preschool children’ and from 5 to 25 percentamong women ages 1544 years. X-lI Problems inassessing anemia as a public health problem arise fromtwo causes: (1) there is no general agreement on adefinition of anemia and (2) the commonly usedcriteria do not necessarily reveal the cause of anemia.

Generally, anemia is a condition in which aninadequate amount of oxygen is delivered to cellsthroughout the body. Lack of oxygen at the cellularlevel creates metabolic slowdown, causes muscles(such as the heart) to operate less effectively, andplaces stress on the circulatory system as it tries tooxygenate tissues. The physiological symptoms ofanemia tend to be nonspecific, making the conditiondifficult to isolate. Mild unemia, for example, may beassociated with dizziness, weakness, fatigue, and irrita-bility. A person with iron-deficiency anemia mayexperience breathlessness, tachycardia, dysphagia, andpalpitation on exertion. A diagnosis of anemia usuallyis based on a biochemical determination of thecapacity of an individual’s red blood cells to carryoxygen throughout the body and make it available tocells.

The only parts of the red blood cell to whichoxygen can be attached und transported are the iron-containing porphyrin complexes within the proteinhemoglobin. Thus hemoglobin levels reflect anemiaand iron status.

The hemoglobin level represents a relationshipbetween the production of red blood cells and theirdestruction or loss. Low hemoglobin concentrationsmay result if red blood cell synthesis is decreased or ifdestruction or loss increases, and these conditions areinfluenced in turn by several factors. Dietary compo-

nents such as iron, folic acid, and vitamin BIz tiect therate of synthesis and successful maturation of redblood cells. The most common cause of red blood cellloss is hemorrhage from trauma, surgery, childbirth,or internal bleeding. Genetic diseases such as sicklecell disease, thalassemia, and gIucose-6-phosphatasedehydrogenase deficiency can cause red blood celldestruction. A low hemoglobin level may result fromany of these factors, but the major cause is irondeficiency.l,d

Although determination of low hemoglobin con-centration or hematocrit (volume of packed red bloodcells) may indicate an anemic condition reflecting avariety of factors involved in the synthesis, destruc-tion, or loss of red blood cells, more specific tests canbe used to look at the critical factor of iron status. Freeerythrocyte protoporphyrin measures the portion ofthe hemoglobin molecule to which iron could beattached; thus free erythrocyte protoporphyrin levelswill rise proportionally as the amount of iron inhemoglobin declines. Iron is also transported in bloodplasma. Serum iron, total iron-binding capacity, andpercent transferring saturation are measures of this“circulating iron.’’3,J Iron stores are found in otherbody tissues, such as the liver, spleen, and bonemarrow. These iron reserves can be measured bydetermining levels of iron storage proteins—marrowhemosiderin or serum ferritin.

When iron intake satisfies body needs in a healthyindividual, circulating iron is maintained at an optimallevel. However, when iron needs are not met by dietaryiron, the body’s iron stores become progressivelydepleted, circulating iron levels are affected and,ultimately, hemoglobin levels drop because the iron isinsufficient to support the proper synthesis of redblood cells. Thus iron status is a point on a continuumthat ranges from adequate tissue levels and stores todepleted tissue levels and stores.

Three stages generally are identified in the processof iron depletion. In the first stage, iron stores arereduced, measured by a decrease in marrow hemosid-

3

erin and serum ferritin. During this stage, the absorp-tion of dietary iron may become more efficient, as thebody tries to compensate for increased iron losses orincreased demands for iron.

The second stage involves a severe reduction iniron stores and impaired red blood cell production.Serum iron levels generally decline, and the bodycompensates by increasing the amount of transferring,the protein that transports serum iron. Thus the ratioof serum iron to iron-binding capacity, or transferringsaturation, decreases markedly. In this stage, levels offree erythrocyte protoporphyrin increase, but hemo-globin levels are in the normal range.

It is diilicult to monitor this second stage of irondepletion because serum iron and total iron-bindingcapacity are affected by several factors, not just byiron. For example, serum iron levels exhibit diurnalvariation, with the highest values usually found in themorning and the lowest in the evening. Infection altersserum iron and total iron-binding capacity levels. Bothof these conditions could result in a lower percenttransferringsaturation that might be interpreted as irondeficiency. .

In the third stage of iron depletion, circulating ironlevels decline, and iron stores may be depleted. This isreflected by lower hemoglobin concentrations, lowerhematocrit values, and lower percent transferringsatu-ration.

km nutrition in the individual

Each individual’s iron nutrition is a state ofbalance between the body’s supplies of iron and needsfor iron. Iron supplies are affected by diet and health;iron requirements change with age and physiologicalconditions such as pregnancy and illness.

Dietary sources of iron include foods, iron supple-ments, and unintentional sources, such as iron fromcast-iron pots. The amount and chemical form of ironingested as well as the individual’s iron needs affect thebody’s capacity to absorb iron.

The average American diet contains 6 milligrams(mg) of iron per 1,000 kilocalories.’’’” In general, dailyiron intake is about 3 mg in infancy and increasessteadily until adolescence. For males, intake remainsstable at 18 mg throughout adulthood, but females’intake declines from 15 mg among adolescents to 11mg among adults. This pattern suggests that total ironintake is related directly to total caloric intake.14

These figures for daily iron intake do not indicatehow much iron is absorbed. Iron occurs in two forms,which differ in their availability to the body. Readilyabsorbable “heme iron” comes from animal tissues,such as meat, fish, and poultry. As the name implies, itis the organic iron bound to myoglobin in muscletissues. “Nonheme iron,” which is less available to thebody, also is found in animal products, such as eggs,

and foods from nonanimal origin such as grains,vegetables, and legumes.

The estimated heme iron intake is only 1–3 mg perday in the United States.1S17Nevertheless, heme ironis an important contribution to iron status because :27to 37 percent of the heme iron ingested is abs-orbed. 16,1‘Jg Moreover, heme iron is absorbed directlyinto the mucosal cells of the small intestine, which takeup the entire complex without releasing the iron frclmits bound form.20,21Once inside the mucosal cell, hemeiron is released. Other dietary factors are not known toaffect heme iron absorption.5

Nonheme iron is the more abundant dietaryform.’z Because nonheme iron is bound in inorganicforms and iron-protein compounds, it must be releasedand reduced from a ferric to the more soluble ferrousform by substances in the intestines before it can beabsorbed.zo,” In addition, the absorption of nonhemeiron is affected greatly by other components of themeal in which it is consumed. Thus it is difficult toestimate or predict what portion of ingested nonherneiron is absorbed. 16,24

At least two factors enhance the uptake of non-heme iron by as much as fourfold. These are the “meatfactor” and vitamin C. 12116,24,Z5Iron absorption tests inwhich adult females ate a typical mixed Americrmmeal revealed that beef, lamb, chicken, pork, and fishhad identical enhancing effects on iron absorption.5Therefore, animal tissue is doubly valuable because itprovides highly available heme iron and also contrib-utes an enhancing factor that improves the availabilityof nonheme iron. This enhancing factor has not belenidentified specifically, although studies to date haveruled out protein or animal protein. 17,26There is someevidence that individual amino acids present in largeramounts in animal tissue protein than in vegetableprotein may be responsible.15 Animal protein fromeggs and dairy products does not have an enhancingeffect and may lower the availability of ingested ironby as much as 60 to 80 percent.s,lz,lg

Vitamin C (ascorbic acid), the other dietamyconstituent known to enhance nonheme iron absorp-tion, acts by maintaining iron in a reduced, solubleform that is absorbed readily. Vitamin C also enhancesiron uptake by forming a chelate with iron thatremains soluble (and thus more absorbable), even inthe basic environment of the small intestine.~,ZATheenhancing effect of vitamin C appears to be limited to 4hours after iron ingestions It has been suggested thatthe effects of animal tissue and vitamin C consumedtogether are additive and that 1 gram (g) of meat and 1mg of ascorbic acid may produce roughly equivalenteffects.‘A

Other dietary factors may enhance iron absorp-tion, but most have not been studied sufficiently topermit an assessmentof their effects.s’ls-lTOn the otherhand, there are several known inhibiting factors,including tannic acid in tea, which appears to create

4

insoluble iron complexes. ~’ Egg yolk phosvitin,’7 ant-acids, lb calcium and phosphate salts,28 bran,29 andethylene diamine tetraacetic acid (EDTA)30 also havebeen shown to inhibit iron uptake.

Absorption of heme and nonheme iron also isaffected by an individual’s iron status. As iron storesbecome depleted, absorption increases; as more iron isstored, absorption decreases.l@J~4 Iron status, in turn,varies according to an individual’s iron requirements.

Iron requirements are determined primarily by theamount of iron lost in the normal turnover of bodycells, the amount lost from normal and abnormallosses of blood, and the amount needed to make newcells and tissues, which may vary during the life cycle.Normal iron loss in the adult male has been estimatedto be about 1 mg per day.~1 This loss occurs mainly inthe gastrointestinal tract-O.4 mg as red blood cells,0.2 mg from bile, and less than 0.1 mg throughexfoliated cells. 15JZDaily losses also include about 0.2mg through skin as exfoliated cells and in sweat andabout 0.1 mg in urine.j~

Females require additional iron to compensate forthat lost in menstrual blood and to meet the needs ofpregnancy. An average of 0.5 mg per day of additionaliron is required to replace that lost in menstrualblood.]z Full-term pregnancy places an additionaldemand for 700 to 1,000 mg of iron on the mother, aneed that generally cannot be met without

~~~u Body size and blood volumesupplementation. 1---0-also cent ribute to differences in normal iron losses andaverage iron requirements among individuals andbetween males and females.~~

Another factor that affects iron requirements,especially among adults, is blood loss. A blooddonation of 450 milliliters entails a loss of 225 mg ofiron, the equivalent of a loss of about 0.6 mg per dayover 1 year. }3Surge~, accidents, and chronic illnessessuch as ulcers and hemorrhoids also account forsizable iron losses.

Drugs may increase or decrease iron loss, eitherdirectly or through their effects on blood 10SS. Oralcontraceptives, for example, reduce menstrual bloodloss.~’J7 Other drugs, such as aspirin, act as gastric orenteric irritants and increase blood 10SS.38

The dietary and health factors that affect anindividual’s iron status often are influenced by social,cultural, and economic conditions. People with limitedfinancial resources may not be able to purchase manyiron-rich foods such as meat or foods rich in vitaminC, such as fresh fruits and vegetables. Inadequatemedical care, which is at least partially a function ofincome level, may increase the occurrence or durationof diseases associated with low iron stores or bloodloss. People with limited education may have troublechoosing a diet that provides enough iron. Studies haveshown, for example, that mothers’ educational level ispositively correlated with the quality of their children’sdiets.~’,w Cultural practices associated with geographic

regions or ethnic goups also may affect iron status.Pica, the eating of nonfood items (such as clay orlaundry starch), can inhibit iron absorption by bindingavailable iron and forming insoluble complexes, andthis practice is more common in some cultural and agegroups than in others.

Race is a factor that may be associated with ironnutrition and anemia. Several large surveys havereported a higher prevalence of low hemoglobinamong black persons than among white per-sons. MLAZ These differences may be a result ofdifferences in cultural eating patterns or in income.However, there is increasing evidence that differencespersist even when iron status measured by percenttransfernn saturation4M8 and socioeconomic status4WSare taken into account.

Groups at risk of iron deficiency andanemia

Because so many dietary, health, and socioeco-nomic factors interact in determining iron intake, ironrequirements, and iron loss, there is great individualvariation in iron status. Despite this variation, it isreported commonly in the literature that severalpopulation groups—preschool children, adolescents,women of childbearing age (including pregnant wom-en), and the elderly—are most likely to develop irondeficiency and anemia.20’49 The analyses of iron statusderived from NHANES I data directed attention tothree of these groups: children ages 1–3 years,menstruating women ages 12–54 years, and personsages 65–74 years.

Preschool children

The level of iron endowment at birth varies widely.It generally is believed that most infants are born withstores sufficient only for the first 4 to 6 months of life,at which time they become dependent upon dieta~ orsupplemental iron to maintain body stores and meetthe demands imposed by growth.jo

Several factors make it difficult for very youngchildren to develop and maintain adequate iron status.Human milk and cow’s milk, the major foods in thefirst year of life, are poor sources of iron, althoughbreast milk at least supplies iron in a form that isabsorbed easily and may thus afford some protectionagainst development of iron-deficiency anemia.j Iron-fortified formulas and dry cereals have been usedincreasingly in the past 10 years in an effort to boostthe iron in the daily diet of infants. Nonetheless, by thetime they reach the age of 1 year, many children mayhave inadequate iron status.so

Children over 1 year old whose diet is composedalmost entirely of cow’s milk are at particular risk ofiron deficiency not only because milk is very low iniron but also because it may be substituted for other

5

foods that are better sources of iron. In addition, milkmay interfere with iron absorption through the forma-tion of insoluble iron complexes. It may even promoteiron loss by causing gastrointestinal bleeding (oftencalled cow’s milk anemia).3,51

Women of childbearing age

The combination of menstrual blood loss, in-creased iron needs in pregnancy, and eating habits,including dieting, place women of childbearing age at ahigh risk of iron deficiency and anemia.

As a continual cause of blood loss, menstruation isresponsible for an average daily iron loss of 0.5 mg,added to the average normal physiological iron loss of1.0 mg per day. Some women average daily iron lossesfrom menstrual flow of as much as 1.4 mg per day,bringing their total average monthly loss to 2.4 mg perday.~z

Assuming that only 10 percent of dietary iron isabsorbed, women would need to eat 18 mg of irondaily to meet their physiological needs, and many findthis difficult to achieve. It would take approximately3,000 calories to provide a daily intake of 18 mg ofiron, far beyond the recommended dietary allowancefor calories for women. As a result, even if a womanhas adequate iron stores when she attains menarche,the continual additional demand for iron to replacethat loss during her reproductive years throughmenstruation may place her at risk of iron deficiencyor anemia.

In addition, pregnancy often exacerbates thiscumulative iron loss by increasing the iron needs of awoman who may be bordering on iron deficiency.Increased demands for iron arise from a 25-percentincrease in maternal hemoglobin mass and the need toprovide tissue iron and develop iron stores in the fetus.Because the iron needs of the fetus will be met before

those of the mother, a pregnant woman may experi-ence further drain on her iron stores and may not beable to compensate for this loss after giving birth.52-SO

The elderly

With increasing age, people are more prone tohealth problems. Many chronic conditions, such asarthritis, which are more prevalent in the elderly andresult in disability or discomfort, may influence eatingpatterns.ss Dental and other impairments that mayinfluence diet are also more prevalent among olderpersons. For example, the percent of persons ages65–74 years without natural teeth is higher (about 45percent), than that of persons ages 18–64 years,5bandthe rate of visual and paralytic impairments is muchhigher among persons ages 65 years and over thanamong younger people. STA study of NHANES I datashowed that a higher proportion of persons ages 65--74years with natural teeth, ingested foods that aresources of iron, vitamin C, and protein at levels at orabove the NHANES standards than persons of similarages without natural teeth or whose teeth wereindicated for extraction did.SE

Studies have reported that many senior citizenshave iron intakes below recommended allowances, thatiron intake decreases with increasing age, and thatprotein sources may change from red meats to lessexpensive nonmeat substitutes.sg,wLow-cost, high-en-ergy foods such as cereals, pastas, and legumes havelower iron content and less available forms of iron thananimal protein foods. A slight decrease in iron intakeand lower meat consumption have been documentedfrom the NHANES I data for persons ages 65--74years.b1,b2

The dietary habits of this group, as with allsubpopulations, are influenced by many social, cultur-al, and economic conditions .bJ,64

Methodology: Sources andlimitations of the data

Suwey procedureMedical and demographic data were collected for

NHANES I from April 1971 to June 1974. Teams ofinterviewers and examiners moved with speciallydesigned and constructed mobile examination centersto each of the 65 selected sample areas so that standardphysical examinations could be given under standardconditions. Of the 28,043 persons ages 1-74 years whowere selected in the national probability sample,20,749 (74 percent) were examined.’

Information available from this sample includedinformation from general medical, dermatological, anddental examinations; medical history questionnaires;dietary interviews consisting of a 24-hour recall and afood frequency questionnaire; and laboratory analysesof blood and urine samples. Body measurements weretaken by trained technicians, and demographic datawere obtained by interviewers from the U.S. Bureau ofthe Census. In addition, a detailed component of thesurvey consisting of a general medical history supple-ment, supplements for arthritis and for respiratory andcardiovascular conditions, a health care needs ques-tionnaire, a general well-being questionnaire, and moreextensive medical and laboratory examinations wasadministered to a subsample of 3,854 adults ages25–74 years. The analyses mesented in this reDort are.based & the 20,749 ex-&nin~dpersons.

Biochemical determinations

Blood specimens were obtained by venirmncturefor adults %d by venipuncture or ‘finger~tick forchildren ages 1–3 years.b Hemoglobin levels weredetermined in the mobile examination centers using aCoulter Hemoglobinometer, which employs a beam of

‘Appendix I inchsdex a more detailed account of the selection of thenational probability sample.

%e method used to obtain the blood specimen for each child was notrecorded.

light and a photoelectric measuring device to comparea measured reference solution to the blood sample. Thehemoglobinometer was checked daily with commer-cially available hemoglobin reference solutions. Hemo-globin concentration was computed automatically anddisplayed in grams per deciliter (g/dI). The test wasrun twice and samples had to agree within 0.2 g/all.Readings of less than 11.0 g/all or more than 18.5 g/allwere retested for accuracy and then called to theattention of a physician.

Serum samples were frozen and sent to theNutrition Biochemistry Laboratory at the Centers forDisease Control, where serum iron and total iron-binding capacity were measured using a modificationof the automated Technicon AAII 25 method.bs Thetechnique involved measuring the intensity of theviolet-colored complex formed in the reaction betweenFerrozine and Fe(II) in pH 5.0 buffer at 562 nm. Totaliron-binding capacity was determined by dilutingserum specimens and quality control samples with aniron-saturating solution, removing excess iron, centri-fuging, and analyzing the supematent for iron usingthe serum iron method. Determinations were repeatedfor serum iron below 50 and above 250 p.g/dl and fortotal iron-binding capacity below 250 and above 500pg/dl. Percent transferringsaturation then was calcu-lated. A more detailed descriptionmethods used to obtain these dataIished.bs

Assessment of iron statusThe biochemical determinations

of biochemicalhas been pub-

available fromNHANES I included three measurements that com-mordy are used to assess iron status—hemoglobinconcentration, serum iron, and total iron-bindingcapacity.’ They were used to create three clas-sifications of iron status—low hemoglobin (low Hb),

‘3-Iematocrit was also determined, but it was not used in this study becauseit is highly correlated with hemoglobin levels.

7

Table A. Age- and sex-specific hemoglobin reference standards!

Age SexHemoglobin value

(g/all)

1-5 years ............................6-11 years ..........................12-14 years ........................12-14 years ........................15-17 years ........................15-17 years ........................16-74 years ........................18-74 years ........................

Male and femaleMale and female

MaleFemale

MaleFemale

MaleFemale

11.011.512.512.013.012,014.012.0

~See references66 and67.

low percent transferringsaturation (low TS), and lowhemoglobin combined with low percent transferringsaturation (low Hb/TS) in this report.

Hemoglobin levels are the most widely usedmeasure of anemia. In this study, persons withhemoglobin levels below the reference standard fortheir age and sex (table A), were classified as low Hb.This category describes persons who are below agenerally accepted cutoff point for anemia. The hemo-globin levels could reflect the presence of severalfactors that affect the synthesis, maturation, destruc-tion, and loss of red blood cells.

As serum iron levels decline, the capacity totransport iron, or total iron-binding capacity (TIBC),increases as the body attempts to compensate bycreating more iron-carrying transferringproteins. Thusthe ratio of serum iron to TIBC, called percenttransfernn saturation (TS), declines markedly. Percenttransferringsaturation is widely considered a betterindicator of circulating iron status than serum iron orTIBC alone.bg However, in chronic infections, TSdecreases but TIBC does not increase and may evendecrease slightly. Thus to exclude from the analysespersons with low TS values due to infection, this studyused TS and TIBC to categorize persons with lowcirculating iron. Persons with a TS below 16 percentand a TIBC above 250 pg/dl were classified as low TS.

Persons with hemoglobin levels below the standardfor their age and sex in conjunction with low circulat-ing iron (defined as transferringsaturation below 16percent and total iron-binding capacity above 250p.g/dl) were placed into the third category, lowHb/TS. Instead of a reading of low hemoglobin alone,this category reflects an anemic condition that can beattributed to lack of iron or iron-deficiency ane-miam6S,69

Persons not in these categories were referred to ashaving adequate Hb, adequate TS or adequate Hb/TS.However, “adequate” cannot be interpreted as aclinical judgfnent of iron status. The categories of ironstatus used in this report are defined according toreference standards, which is only one approach todetermining an anemic condition.

Above all, although the measurements used toclassify the data were chosen to reflect stages ofgeneral anemia, iron deficiency, and iron-deficiency

anemia, they do not represent the prevalence of diseasebut rather the prevalence of biochemical levels that arebelow the chosen reference standards, Because individ-ual iron requirements vary widely, analyses reportingthat a certain number of persons are in the low Hbcategory do not mean that this number of cases ofanemia exist, nor does the statement that a certainnumber of persons have adequate Hb mean that noneof these persons could be diagnosed as anemic. Forexample, a 30-year-old female with a hemoglobin levelof 12.0 g/all, the reference standard for her age, maynot have a sufficient iron supply to meet her bc~dy’sneeds, while another female of the same age with ahemoglobin level of 11.5 g/all may not need or even beable to absorb additional iron. A further consequenceof using this approach to classify iron status is that theprevalence estimates calculated in this study aredependent on the reference standards used. Revisingthe standards would alter the prevalence estimates,although the conditions measured by the biochemicaldeterminations do not change.

Other approaches-clinical trials (response to irontherapy intervention), multiple iron parameters, anddistribution analysis—have been used to estimate theprevalence of anemia in populations. The first ap-proach was not applicable to this survey data, and thesecond approach was of limited use because only someof the iron-related determinations were done inNHANES I. The third method, distribution anal!ysis,has been used to look at the iron status of women ages1844 years.qj However, the large sample sizes re-quired by this approach made it less desirable for usein this general study. ,Therefore, the use of cu~toffvalues was the most appropriate method for the studyof iron status using NHANES I data.

Calculation of dietary intake andabsorbable iron

Dietary intake and food consumption patternswere determined for each examinee using two meth-ods: (1) a 24-hour recall questionnaire and (2) a fc)odfrequency questionnaire. Interviews were conducted bytrained staff who had bachelor’s degrees in food amdnutrition or who were registered dietitians.

The recall covered the 24-hour period ending atmidnight the day before the examination. Examina-tions were performed Tuesday through Saturday, sothat weekends, when eating patterns may be different,were excluded. The examinee (or parent of a childunder age 12) was asked to report all foods, beverages,and supplements consumed on a meal-by-meal basis(breakfast, lunch, supper, snack). To help the exami-nee estimate portion size, three-dimensional fc~odmodels were used. The data were coded, edited, amdcomputer analyzed at NCHS for nutrient value per100-gram portion of food. A computer model to

8

analyze the types of foods consumed in each ingestionperiod was designed and implemented at Abt Asso-ciates, Inc.

The food frequency questionnaire, which was usedto collect information on how often different types offoods were consumed during the preceding 3 months,was administered after the 24-hour recall. In additionto providing data about usual food consumptionpatterns, the food frequency questionnaire helped theinterviewer determine whether the foods consumed onthe day reported were typical of those consumed over alonger period. The frequency questionnaire also helpedthe interviewer identify and clarify inconsistent re-sponses.

Both of these instruments have methodologicalproblems. The 24-hour recall is limited to one day;however, intake for any one day may be neither usualnor representative of a person’s overall diet. Moreover,the 24-hour recall is designed to provide an estimate ofthe quantities of food consumed, and individuals varygreatly in their ability to remember what they ate.People also err in estimating the amounts of food theyate. Finally, error is introduced into the 24-hour recallbecause methods of food preparation vary greatlyamong respondents, and it is difficult to determine thenutrient content of many commercial or mixed dishes.Thus the available nutrient breakdowns provide onlyrough estimates of the nutritive value of the foodsreported.

The food frequency questionnaire is subject to thesame limitations as the 24-hour recall because therespondents must remember and report the widevariety of foods and beverages consumed over a 3-month period, as well as estimate how frequently eachitem was consumed. Because the quantities consumedare not reported, nutrient intake cannot be calculatedaccurately from the food frequency questionnaires.Therefore, data from the food frequency question-naires were not included in the analyses for this report.

Mean iron intake values, which are the mostwidely used estimates of dietary iron intake, werecomputed from the 24-hour recall data to examine therelationship between iron intake and the iron statuscategories investigated in the prevalence analysis.However, mean iron intake does not represent accu-rately the amount of iron available to or absorbed bythe individual. In the relational analyses, which alsoconsidered iron intake as one factor contributing toiron status, iron intake from the 24-hour recall datawas adjusted according to an experimental methoddeveloped by Monsen and Cook. M-17

This calculation of absorbable iron took intoaccount the consumption of heme iron, two facilitators(the meat factor and vitamin C), and one inhibitor(tea). Meals were calculated separately and totalled forthe 24-hour period. A two-stage classification systemwas employed.

First, the iron ingested was classified into a low,

medium, or high availability category according toenhancing and inhibiting factors in the meal, asfollows:

Low availability-Meals that contained less than24 mg of vitamin C or less than 6 g of protein frommeat, fish, or poultry.

Medium avai/abili~Meals that contained morethan 25 mg but less than 75 mg of vitamin C or morethan 6 g but less than 18 g of protein from meat, fish,or poultry.

High availabili~—Meals that contained more than75 mg of vitamin C or more than 18 g of protein frommeat, fish, or poultry or from 25–75 mg of vitamin Cplus from 6-18 g of protein from meat, fish, or poultry.

If more than 225 g of tea was consumed during theingestion period, the iron intake was dropped onecategory unless it was classified as low availability.

The Monsen algorithm on which this step is basedemployed grams of meat, fish, and poultry rather thang-rams of protein. In this model, it was assumed thatlean meat contains approximately 25 percent protein;thus 6 grams of protein corresponds to approximately25 grams of meat, fish, or poultry.

In the second step, iron intake was divided into itsheme and nonheme components. It was assumed that40 percent of the total iron present in meat, fish, andpoultry was heme iron. The availability categorydetermined in step one, according to the enhancing orinhibitory nature of the meal, affected the percent ofheme and nonheme iron that would be absorbed. Forexample, it was assumed that only 3 percent ofnonheme iron would be absorbed in a low availabilitymeal, but 8 percent would be absorbed in a highavailability meal. lA,lT.~

Low availability meal: Absorbable iron = (hemeiron x O.110) + (nonheme iron X 0.030)

Medium availability meal: Absorbable iron =(heme iron X O.122) + (nonheme iron X 0.050)

High avaiIabiiip meal: Absorbable iron = (hemeiron X O.14) + (nonheme iron X 0.080)

Relational analyses: Analyticalsubgroups, independent variables, andtests of statistical significance

The relational analyses were designed to examineassociations between iron status and dietary, health,socioeconomic, and demographic variables. Separatebut parallel analyses were performed on threegroups-children ages 1–3 years, women ages 12-54years between menarche and menopause, and personsages 65–74 years. These groups generally are ccmsid-ered to be at risk of iron deficiency and anemia. Theyalso were chosen for closer study because analyses of

9

the NHANES I data revealed a sufficient prevalence oflow Hb, low TS, or low Hb/TS to suggest that theremight be a public health problem in these groups andbecause the sample ,sIzeswere large enough to drawmeaningful statisticalconclusions.

The subpopulation of children ages 1–3 yearscomprised black and white children and both sexes.The group of menstruating women was composed ofblack and white women who had attained menarcheand had not reached menopause (ages 12–54 years),excluding pregnant women and women who had beenpregnant within the past year. The elderly subpopula-tion was defined as persons ages 65–74 years, male andfemale and black and white. Adults with a history ofulcers, who might be suffering from large blood losses,or have diabetes, were excluded from the analyses.These groups were excluded because of the knownassociation of the condition with iron status or dietaryintake status. As noted earlier, the sample sizes withineach subgroup varied because of missing data for someof the independent variables,

For each group, relationships between iron statusand an extensive set of independent variables wereconsidered. Approximately 200 independent variables,hypothesized to be associated with iron status or ironintake, were selected for study from the availableNHANES I data tapes. Preliminary statisticalanalyseswere conducted to collapse these variables into asmaller set of analytical constructs for use in subse-quent bivariate and multivariate analyses. Variablesthat had a high proportion of missing data, did notvary across the sample, or were not easily quantifiedwere set aside. The food frequency questionnaire data,for example, were eliminated for the last reason. Inaddition, variables that addressed the identical con-struct were collapsed into a single variable; forexample, several questions about aspirin were col-lapsed into a single measure of regular aspirin use.Many other potential variables were excluded on thebasis of low frequency (e.g., leukemia), the expectationof little relationship to iron status (e.g., dermatologyexam), or poor measurement properties (e.g., physicalactivity).

The resulting set of “core variables” to be testedwas composed of the 21 variables listed below. Thedefinitions of these variables are presented in appendixH.

Total iron intake (mg)(log) Total iron intake (log mg)Total animal tissue iron intake (mg)Total non-animal tissue iron intake (mg)Total protein intake (mg)Total animal tissue protein intake (mg)Total non-animal tissueprotein intake (mg)Estimated available iron (mg)Estimated available iron with tea (mg)Vitamin C intake (mg)Pica

Regular vitamin useRegular mineral useRegular vitamin and mineral useRegular aspirin useAgePoverty income ratio (a poverty index)Estimated family incomeNumber of persons in the householdAge of head of householdYears of education of head of household

In addition, a few variables of particular interestwithin each age group were added to the appropriatesubpopulation. For children ages 1–3 years, additicmalvariables included history of breastfeeding and vari-ables related to the child’s birth. The analyses formenstruating women and the elderly included vari-ables dealing with general health, reproductive history,and two additional demographic -scribed below.

Children ages 1–3 yearsEver breast-fedMother’s age at child’s birthWhether child was the first bornChild’s birthweightWhether child was prematureWhether child is presently ill

Women ages 12–54 and personsSpecial dietAge menstruation beganEver pregnantEver diagnosed as anemicAny trouble eating

variables, as de-

ages 65–74 years

General findings on the general medical examEver had abdominal surgeryMarriedEducation of examineeCurrent or recent use of oral contraceptives

(women ages 12-54 years only)

For each group, bivariate analyses were conductedto examine the relationship between iron status amdeach variable in the appropriate core set. Two-tailedZ-tests with a significance level of 0.05 were conductedtaking into account the complex survey design (seeappendix I).

As a further step to understanding the relationshipof these dietary, health, socioeconomic, and demo-graphic variables to iron status, a limited set ofvariables was placed into a multivariate logistic framew-ork, These multivariate analyses were designed todetermine which combination of variables was mostassociated with the various iron status categories ineach subgroup, This approach was concerned withdiscerning associations that might warrant closerstudy and does not show cause-and-effect relationshipsbetween any variable and iron status.

As a general guideline, the dietary and healthvariables that were significantly associated with iron

10

status in the bivariate analyses were chosen for themultivariate analyses. As a final test of iron intake,however, (log) mean iron intake and estimated absorb-able iron were placed in the multivariate framework,regardless of the results of the bivariate test. Inaddition, a few variables of special interest—thosecommon in a particular age group or those holdinggeneral theoretical interest—were included in themultivariate set.

Furthermore, the socioeconomic and demographicvariables included in the core bivariate set could not beexamined simultaneously in a single multivariatemodel because many of them were highly correlated. Aseries of logistic analyses was conducted to select asmaller set of these variables that were relativelyunconfounded and related to iron status. These analy-ses were designed to assess whether the logarithmictransformation of a poverty index called the povertyincome ratio (PIR), was a better variable than PIRitself and whether any additional demographic vari-ables (other than age, race, and sex of the examinedperson) contributed to the analyses when used inconjunction with PIR. (The poverty index and how itis calculated are discussed in appendix IL) Theanalyses, conducted simultaneously for all three

groups for each iron status category, showed that thelogarithmic transformation of PIR did not predict ironstatus better than PIR. In addition, the only socioeco-nomic variable that had a significant effect whenentered with PIR was years of educaton of the head ofthe household. Because these results were consistentacross all three analytic groups and for all measures ofiron status, PIR and years of educaton of the head ofhousehold were entered as socioeconomic covariates inall multivariate analyses.

Initially, a simple forward stepwise solution to thelogistic model was obtained without taking intoaccount the complex sample design. If any variablewas not included because of collinearity between it andthe remaining variables in the model, the variable wastested without including its collinear measures. Two-tailed t-tests with a significance level of 0.05 wereconducted. Each variable that was significant underthis approach (t > 1.96) was retained for the finalmodel. If at that point, dietary iron intake was notentered in the model, it was “forced in, ” to yield a finaltest of its significance in relation to iron status whencontrolling for all other significant variables. (Adetailed presentation of the statistical approach isincluded in appendix I.)

11

Findings

Iron statusThe mean values and distribution of the measures

of iron status collected in NHANES I (hemoglobin,hematocrit, serum iron, total iron-binding capacity,and percent transferring saturation) are shown in tables1-1o.

These measures were used as described in thepreceding section to calculate prevalence rates forthree categories of iron status—low Hb, low TS, andlow Hb/TS. Prevalence estimates for various agegroups, males and females, and black and whitepersons are presented in tables 11–13 and summarizedin figures 1–9. Z-statistics for selected comparisons ofprevalence of iron status are presented in table 14.Prevalence rates for pregnant and nonpregnant womenand Z-statistics are presented in table 15. Differenceswith a Z-value greater than 1.96 (p < .05) are noted as“significant” in the text.

Age and sex

The prevalence of low Hb decreased rapidly frominfancy through the preschool years, from a peak of19.3 percent for males and 12.6 percent for females atage 1 to 1.7 percent and 1.9 percent by ages L%5years(table 11 and figure 1). A brief rise in prevalenceoccurred for children entering adolescence (ages 6-11years), which continued for males into ages 12–14years. The prevalence was higher for boys ages 12–14years (7.6 percent) than for girls of similar ages (3.4percent), although the difference was not statisticallysignificant.

Prevalence rates of low Hb were relatively con-stant at approximately 5 percent for adults (ages 18–54years) of both sexes. With advancing age, however, theprevalence rate rose markedly for males, beginning atages 55–64 years. At ages 65–74 years, males had aprevalence of 14.9 percent and females only 4.3percent, a difference that is statistically significant.

The prevalence of iron deficiency, as defined bypercent transferring saturation and total iron-bindingcapacity, generally was higher than that of low

hemoglobin, except among adult males. LOIW TSpeaked at age 1 with a prevalence of 50.8 percent formales and 38.3 percent for females (table 12 and figure4). During the childhood years, the prevalence de-creased rapidly for both sexes, dropping to 8,6 percentfor boys and 13.1 percent for girls ages 12–14 years.Beginning in adolescence, values for females roseslightly and reached 16.4 percent for ages 1844 years,while the prevalence rate for males dropped to 3.6percent. This difference is statistically significant.

Beginning at the age group 45–54 years, ]preva-lence rates for males and females became similalr anddecreased to approximately&7 percent by ages 65-74years.

The combination of low Hb and low TS showedthe lowest prevalence of the three categories. Ex~ept atage 1, when it was especially marked (17.7 percent formales and 9.9 percent for females), the prevalence oflow Hb/TS for any age and sex group never exceeded4 percent (table 13 and figure 7). Within this low level,however, the pattern was similar to that obserwed forthe other two categories of iron status.

Race

In many instances, the prevalence of low Hb wassignificantly higher for black than for white personsamong males and females and at all ages. Exclept atages 55–64 years among males and ages 45 yearsamong females, the prevalence of low Hb was 2 to 10times as high among black persons. For example, theprevalence among white women ages 1844 years was3.4 percent; it was almost five times as high (15.8percent) among black women of the same ages (table11 and figures 2 and 3).

Differences between black and white persons gen-erally were not as great for low TS and low Hb/TS norwere the rates for black persons consistently higher(tables 12 and 13, figures 5 and 6, and figures 8 and 9).Furthermore, although black persons tended tal haveslightly higher prevalence rates in both categories,most differences were not statistically signi:licant.

12

13

14

@J~

.=5:

IIKl

,...-../.

m.

u-l

auacuad

15

6

+j

_.m

E

m.

.

N‘$5>*‘Tm

II

II

II

II

0‘s

0f-

C9

0m0

0w

m0

z0

17

0[i

00

bo

wm

o0

=i-

0m

00

m

!uam

ad

18

--” u.

m.

mw-@3 m

“

0

0.

mr

m.

o

m,

0)

.&a

0

21

These findings are consistent with previousresearch,4145 which indicates that a racial variation iniron status appears to be prominent for low hemoglo-bin but not for other indicators of iron status.

Pregnancy

Because iron status measures may change duringpregnancy, pregnant women were analyzed as aseparate sample. Table 15 shows that pregnant womenages 1844 years were six times as likely to have lowhemoglobin as women in this age range who are notpregnant. This significant difference was observedamong black women and white women; the prevalenceof low hemoglobin for pregnant women was threetimes and eight times as high. No significantdifferences between pregnant and nonpregnant womenwere found for the low TS or low Hb/TS categories.

Iron intake and iron status

Mean iron intakes (mg per day) of persons in twoof the iron status categories analyzed by age, sex, andrace, are presented in tables 16 and 17. Results for thetotal sample classified as low or adequate Hb and lowor adequate TS are shown in figures 10-13. Thesefigures are presented to show the overall relationship.Selected subgroups have small sample sizes thataccount for large variability in the mean iron intakes.(Mean iron intakes for pregnant women are presentedin table B.) Associations with a Z-statistic of greaterthan 1,96 are noted as “significant” in the text.

Age, sex, and race

In general, males with low hemoglobin levels hadmean iron intakes similar to males with adequatehemoglobin levels. Similar findings were observed forfemales with low and adequate hemoglobin levels.

With one exception, the differences in mean ironintake for males classified as low and adequate TS were

nonsignificant. No significant differences were foundfor females.

One marked difference between the sexes was thatfemales’ iron intake decreased between the ages 12 and17 years, while males’ intake increased. This patternwas found in all the low and adequate Hb and TScategories of iron status, although it was most pro-nounced in the low TS group. Too few individualswere classified as low Hb/TS to present findings abouttheir iron intake.

Similarly, the sample sizes for black persons weretoo small to substantiate any racial pattern ofdifferences in iron intake. In general, the iron intiake ofblack persons in the various iron status categories wassimilar to that of the total sample.

Overall, mean daily iron intakes were lower thanthe recommended allowances for each age and sex. 12

Pregnancy

As shown in table B, no statistically significantdifferences were found in the iron intake of pre,gnantand nonpregnant women for low and adequate hemo-globin levels. Although the difference for black womenwas larger, the small sample sizes precluded testing thedifference for statistical significance.

The difference in iron intake of pregnant andnonpregnant women with low and adequate transferringsaturation levels was significant for all races combinedand for white women. For black women, the differencewas in the same direction, but small sample sizes againled to unreliable estimates. Too few pregnant womenhad low hemoglobin combined with low transferringsaturation values to present findings for this iron statusindicator.

Relational analyses

For each subpopulation-children ages 1–3 :years,menstruating women ages 12–54 years, and personsages 65–74 years—bivariate analyses were conductedon a core set of dietary, health, socioeconomic and

Table B. Z-statistics for comparison of iron intake for low and adequate iron statusl for pregnant women 1844 years, with mean, andstandard error of the mean, by iron status indicator and race: United States, 1971–74

Low Adequate

Iron status indicator and race Number of Standard Number of Standardexamined Mean error of

Z-staifisticexamined Mean error of

persons mean persons mean

Hemoglobm

All racesz ..... ...... ...... ...... ....... ...... ...... ...... 63 11.1 0.67 133 10.9 0.46 0.2!5

White , ..... ..... . . . . . . .. . . .. . .. . 39 11.4 0.85 113 10.6 0.49 0.E12Black ....................................................................... 22 “10.3 ‘1.11 14 *13.7 ●1.66 . . .

Transfernn saturationAll racesz ..... .... ...... ....... ....... ....... ........ ........ 40 9.0 0.66 155 11.5 0.43 -3.10

White ....................................................................... 30 8.6 0.77 117 11.4 0.48 -3.09Black ....................................................................... 9 ‘10.4 ●1.40 31 12.7 1.13 .,, .

I Lowandadequatslevels as defined in this report.z Includes data for races which are not shown as separate categories.NOTE: Estimatss preceded by an asterisk do not meet this report’s standards of reliability (i.e., the number of examined persons is less than 25).

22

18

17

16

15

14

13

12

11

10

9

8

7

— Adequate hemoglobin

---- Low hemoglobin

–H”

J,,, ,,,6

5

0 I I I I

1 2-34-5 6-11 12-1415-17 16-44 45-54 55-64 65-74

Age in years

Figure 10. Mean iron intake for males ages 1-74 years, by age and hemoglobin levels: United States, 1971-74

demographic variables. Multivariate logistic analysesalso were conducted on a smaller set of these variables.Because data were missing on some of these inde-pendent variables for certain sample persons, thesample sizes in the analyses vary within the subgroups.All associations discussed in the text are significant ata value of 1.96 (p < .05), unless specifically noted as“not statistically significant.”

Children ages 1-3 years

Analyses examining the relationship between ironstatus and a core set of dietary, health, socioeconomic,and demographic variables were performed only forchildren classified as having low or adequate hemoglo-bin, because about 45 percent of the serum iron andtotal iron-binding capacity data were missing fromNHANES I for this age group. Appendix I discussessome of the reasons this large amount of blood datawere missing in this survey. Males and females werenot analyzed separately because sexual differences iniron status are not noted until adolescence, whenmenstruation begins in females, and both sexes experi-ence growth spurts.TO

Bivariate analyses were conducted for 27 variables,

and results for the total sample (table 19) showed 16statistically significant associations. Many of theseassociations were for dietary variables-most of theiron intake variables, animal protein intake, vitamin Cintake, and regular vitamin and mineral use. Twovariables related to dietary habits also were significant;37.5 percent of the children in the low Hb categoryhad a history of pica compared with 18.3 percent ofthe children in the adequate Hb category. In addition,only 16.9 percent of the children with low Hb had beenbreastfed, compared with 26.5 percent of those withadequate Hb.

The only health variable for which differences weresignificant was “presently ill,” but there were foursignificant findings for the socioeconomic and demo-graphic variables. Children with low Hb were younger,with a mean age of 2.19 years compared with 2.58years for the children with adequate Hb. Low hemo-globin also was associated with a lower poverty indexratio, more people in the household, and lowereducational level of the head of household.

Tables 20 and 21 present the results of analysesconducted for black children and white children.Among white children with low compared withadequate Hb, significant differences were found for the

23

18 r — Adequate hemoglobin

t

---- Low hemoglobin17

14 1-%

-Ic, ,~,: 13&.-

= 12

Tg 11.-

: 10. 1- -\. \.4md.4. e.--= ------------- =---

9 ,/”.-.-

.-m\

89%.

\“w

7 i6 ,/’

?5

012-3 4-5 6-11 12-14 15-17 18-44 45-54 55-64 65-74

Age in years

Figure 11. Mean iron intake for females ages 1-74 years, by age and hemoglobin levels: United States, 1971-74

same variables that were simificant in the total samrde, variables that were significantly associated with hamo-.-except for vitamin C intake, breastfeeding, and pres-ently ill, and with the addition of total animal tissueiron intake. The same trends were evident for blackchildren as well, although the only statistically sig-nificant difference was for age, This may be due to thesmall sample sizes for black persons. In the whitepopulation, 105 children had low Hb, and 960 childrenhad adequate Hb; in the black population the samplesizes were 72 children with low Hb, and 285 childrenwith adequate Hb. Because standard errors weresubstantially higher for the estimates for the blackpopulation than for the white population, a differencethat was statistically significant for white children maynot be significant for black children. When results forthe sample of black children were examined for each ofthe seven variables found to be significantly associatedwith low hemoglobin in the analyses for the totalsample and the white sample, the differences betweenmeans for each variable were quite consistent. Thisalso suggests that the lack of statistical significancewithin the analyses of black children was due primari-ly to higher standard errors associated with the smallnumber in the sample.

Three of the socioeconomic and demographic

globin status in the b~variate a~alyses for the total andwhite samples were chosen for analysis in the multi-variate framework—PIR, years of education of thehead of household, and age. Most dietary variablesthat were significant in the bivariate analyses werecorrelated with iron intake, so (log) total iron intakeand estimated available iron were retained for multi-variate analysis, with vitamin and mineral use and -pica. Race was the eighth variable included in themutivariate analysis.

The results of the multivariate approach (table 22) “showed that age, race, PIR, and education of the hleadof household still were significantly associated withlow hemoglobin. Children who were younger, black,from poorer homes, and from homes in which the headof the household had fewer years of education weremore likely to have low hemoglobin.

When these socioeconomic and demographiceffects were taken into account, pica and iron intakealso were associated with low hemoglobin, indicatingthat two factors related to diet—adequate iron intakeand the absence of pica—make it more likely that achild’s iron status will be adequate regardless ofsocioeconomic situation. The contributions of (log)

24

18

17

16

15

14

B

7

— Adequate transferringsaturationi’”x.

- — -- Low transferringsaturation

i ‘.N.i ‘\.

\---------------

.\*\

●\. \ -+●-. +. -

—#’

J,,, , ,,6

5

0 I I I I

1 2-3 4-5 6-11 12-14 1517 16-44 45-54 55-64 65-74

Aga in years

Figure 12. Mean iron intake for males ages 1-74 yeare, by age and transferring saturation levels: United States, 1971-74

total iron intake and estimated available iron wereabout equal in the final logistic equation, suggestingthat adjusting iron intake for bioavailability does notoffer any advantage in explaining low hemoglobin inchildren.

Memtruating women ages 12-54 years

Because of cumulative iron losses due to menstrua-tion and pregnancy, women of childbearing age havebeen considered particularly susceptible to irondeficiency and anemia.lz The prevalence estimatescalculated from NHANES I data suggest that theseconditions may not be as widespread as has beenpostulated; however, they are prevalent enough towarrant closer examination.

A large sample with complete data permittedanalyses of possible correlates of all three iron statuscategories for menstruating women ages 12-54 years.Bivariate analyses of 31 variables were conducted foreach category for the total sample of women and forblack women and white women separately. Results arepresented in tables 23-31.

The bivariate analyses for the total sample (tables23–25) produced few significant associations. Only age,

previous pregnancy, and years of education of the headof household were associated with all three iron statuscategories. Analyses of the sample of white women(tables 26-28) yielded similar statistical results; analy-ses of the sample of black women (tables 29–31)showed differences that were similar in magnitude butwere not statistically significant, probably because ofsmall sample sizes. Women who were older, had everbeen pregnant, and came from homes where the headof the household had less education, were more likelyto have poor iron status measured by Hb, TS, or both.

Several other variables were significantly associ-ated with two iron status categories. The dietaryvariable of vitamin and mineral supplementation wasassociated with TS or Hb combined with TS for thetotal sample and for white women. Previous diagnosisof anemia, which often is related to previous pregnan-cy, was associated with low Hb or low Hb combinedwith low TS for the same two samples. Women whoreported using oral contraceptives, which may reduceiron losses by reducing menstrual blood flow, weremore likely to be in the adequate TS and adequateHb/TS categories; this association was signiikant forthe total, white, and black samples. Of the demograph-ic variables, age of the head of household was

25

18— Adequate transferring saturation

17 ---- Low transferring saturation

16

15

14

-aE 13ms.——= 12E

~11

m

E.-C 10 ~------ ------- ---0L

9 “\.

8\,

7

6 –i

5/ 5

0’ I I I I I I I —1 2-3 4-5 6-11 12-14 15-17 18-44 45-54 55-64 65-74

Age in years

Figure 13. Mean iron intake for females ages 1-74 years, by age and transferring saturation levels: United States, 1971-74

associated with Hb or Hb combined with TS in thetotal and white samples. In addition, a few variablesshowed significant associations with only one ironstatus category (tables 23–3 1).

Twelve variables were selected for examination inthe multivariate framework. Two demographic vari-ables—age and years of education of the head ofhousehold—that were consistently significant in thebivariate analyses were retained for the multivariateanalyses. Race also was included, as was PIR. Of thedietary variables, regular vitamin use was a significantcorrelate in the bivariate analyses; (log) total ironintake and estimated available iron were automaticallyincluded; and pica was included for theoretical inter-est. The health variables chosen were previous preg-nancy, previous diagnosis of anemia, and oral contra-ceptive use, which were significantly associated withiron status in the bivariate analyses, and age ofmenarche, which showed a high correlation.

Because this set of variables was larger andcontained more intercorrelated variables than the setof variables chosen for multivariate analysis for chil-dren, the analytic strategy was more complex. First, aseries of models was tested to (1) examine thesignificance of the individual independent variables, (2)

ascertain which variables were confounded, and (3)identify the combination of variables that was mostassociated with iron status, These analyses yielded afinal model of seven variables that were consistentlyand significantly associated with iron status, as slhownin tables 32–34.

These seven variables included the four socioeco-nomic and demographic variables that were significantcorrelates of iron status in children—PIR, educationof the head of household, age, and race. Persons frompoorer homes and homes where the household headhad less education were more likely to have low Hb.Black women were six times more likely to have lowHb and almost three times more likely to have lowHb/TS than white women. The older a woman was,the more likely she was to have poor iron status asmeasured by Hb, TS, or both.

The health variables were the other major falctorsaffecting iron status over and above socioeconomicstatus, age, and race. Women who were using oralcontraceptives were only half as likely to have low Hb,low TS, or low Hb/TS as women who did not use oralcontraceptives. On the other hand, women whcl hadever been pregnant were more likely to be in the lowHb or low TS categories, and women who had been

26

diagnosed as anemic were twice as likely to have lowHb or low Hb/TS as women who had never beendiagnosed as anemic. Most diagnoses of anemia hadbeen made during or immediately after pregnancy,which further suggests that pregnancy has long-termeffects on iron status.

In fact, it is difficult in this approach to separatethe effects of previous pregnancy, previous diagnosisof anemia, and age. Women who have been pregnantare more likely to have been diagnosed as anemic, andthey also are generally older than women who havenever been pregnant. When all three variables areentered in one model, only one or two remainsignificant. Therefore, it was concluded that theconfiguration of these variables is associated with ironstatus.

Elderly persons ages 65-74 years

The sample of the elderly included only a fewpersons who were classified as low Hb\TS, so theanalyses of the relationship between iron status andpossible correlates focused on low Hb and low TS.

Bivariate analyses of 30 variables revealed fewsignificant differences between persons in the low oradequate Hb and low or adequate TS categories (tables35 and 36). Low hemoglobin was significantly associ-ated with lower PIR, more people in the household,and less education for both the examinee and the headof household in the analyses for the total sample.Significant health variables were previous diagnosis ofanemia and, for women, previous pregnancy. Use ofvitamin and mineral supplements was associated withHb and vitamin C intake with TS. Low TS also wasassociated with aspirin use and with age of the head ofhousehold.

Separate analyses conducted for white persons andblack persons yielded similar results (table 37-40). Inmost cases, the magnitude of the differences between

groups was similar, but the differences usually werenot significant within the racial groups because of thesmall sample sizes.

A set of 12 variables was selected for multivariateanalysis. This set included five socioeconomic anddemographic varibles (PIR, years of education of thehead of household, age, race, and sex) and four dietaryand health variables that were significant correlates inthe bivariate analyses (regular vitamin use and regularvitamin and/or mineral use, previous diagnosis ofanemia, and previous pregnancy). Estimated availableiron and (log) total iron intake were included automat-ically. Two health variables that were reported bymany people in the elderly subgrou~findings on thegeneral medical examination and trouble eating—wereincluded in the multivariate analysis, although theywere not significant correlates in the bivariate analyses.

In the multivariate approach, six variables weresignificantly associated with low Hb, including raceand sex; black persons were six times more likely tohave low Hb than white persons, and men were seventimes more likely to have low Hb than females. As wasfound in the analyses for menstruating women, thehealth variables were associated with hemoglobin, butdietary factors were not. Elderly persons who had ahistory of anemia, who said they had trouble eating,and who had health problems noted on their generalmedical exams were two to three times more likely tohave low hemoglobin than persons without thesedifficulties.

Only one variable, years of education of the headof household, was associated with low TS in the elderlysubgroup. However, this association was not sig-nificant in the bivariate analyses, nor were the associa-tions between trouble eating and findings on themedical exam and low hemoglobin. Although themultivariate analyses indicated that socioeconomicand health factors are important contributors to ironstatus among the elderly, this approach did not revealany clear-cut relationships.

27

Discussion

Iron-deficiency anemia is believed to be the majornutritional deficiency problem in the United Statestoday s-s Estimates of the prevalence vary, but there isgeneral agreement that the populations at greatest riskare infants, adolescents, females during the reproduc-tive years, and pregnant women. IZ.Z0,19The primarycause of anemia usually is attributed to nutrition, thatis, the increased physiological need for iron either tosupport growth or to compensate for iron losses thatexceed intake and absorption.qq Small-scale and na-tional nutrition studies have shown that iron intake isbelow recommended levels for the American popula-tion in general and for these high-risk groups i!nparticular. 1r6111,71

The results of this study generally support theprevious research on iron status and iron intake. Twosomewhat unexpected findings emerged, however.

First, the study shows that iron deficiency asmeasured by low transferringsaturation levels is themost prevalent of the three measures of iron status inthe United States. The prevalence of iron-deficiencyanemia, in which the condition can be attributed toiron lack based on a combination of low transferringsaturation and low hemoglobin levels, is limited; thisiron-deficiency anemia appears to be a problem onlyamong very young children.

Second, although the dietary iron intake of chil-dren with low hemoglobin is significantly lower thanthat of children with adequate hemoglobin, there is nostrong relationship between iron intake and iron statusin the population as a whole,

Iron status in the United StatesLow hemoglobin, perhaps the most widely used

indicator of iron status, is most common in youngchildren, pregnant women, older men, and the blackpopulation. At age 1, the prevalence of low hemoglo-bin in the total sample population is 19.3 percent formales and 12.6 percent for females, and at ages 2–3years, it is approximately 9 percent for both sexes.

After early childhood, the prevalence rarely exceeds 5percent, except for males ages 65–74 years, when itrises to almost 15 percent. At all ages, however, blackmales and females are 2 to 10 times more likely to havelow hemoglobin than white persons. Approximately 32percent of the pregnant women in NHANES I had lowhemoglobin, with a prevalence of 28.4 percent amongwhite women and 52.9 percent among black women.

These findings are in general agreement with theresults of other more limited studies. The Ten-StateNutrition Survey,8 the Preschool Nutrition Sw-vey,72and the Nutrition Surveillance System of the Centersfor Disease ControlTs all found low hemoglclbin inchildren ages 1–3 years. The presence of low hemoglob-in among pregnant women has been reportedl wide-ly,52-54

The high proportion of males ages 55–74 yearswho have low hemoglobin also has been repo]led inother studies, notably in the Ten State NZJtritionSurvey.8 However, although men of this age lhave ahigh prevalence of low hemoglobin, no other evidencerelates these levels to iron lack. It has been suggestedthat the phenomenon is attributable to changes intestosterone levels occurring with advancing age,74inwhich case the reference standard used to definehemoglobin status might be inappropriate for elderlymen.

A high prevalence of low hemoglobin among blackmales and females of all ages also has been repolrtedinmany recent studies of iron status. In the Ten-StateNutrition Survey, a consistent 1 g/all difference in meanhemoglobin levels between black persons and whitepersons of both sexes was found regardless of age,income, or region. ~ Mean hemoglobin levels of blackchildren in the Preschool Nutrition Survey were 0.5g\dl lower than mean values for white children, evenafter controlling for transferringsaturation.~,1’ Otherstudies, including the collaborative perinatal project,12the Bogalusa heart study,% and several sma~ll-scalestudies,45~8found persistent differences in mean,hemo-globin levels and hence, prevalence rates of low

28

hemoglobin between black persons and white persons,using the same reference standards.

Although it is clear that the true prevalence ofanemia is higher in the black population, differences inhemoglobin levels between black persons and whitepersons are evident among healthy Persons.ds Severalresearchers have suggested that lower mean hemoglo-bin values found for black persons are normal healthyiron status.a Because little difference in the TS valuesof black persons and white persons was found in thisstudy and in others, the problem may lie with applyinghemoglobin cutoff values traditionally used for whitesubjects to data obtained from black subjects.

In contrast to low hemoglobin, which was notcommon throughout the population, low circulatingiron measured by transferringsaturation and total iron-binding capacity is widespread and is equally prevalentamong black persons and white persons. Groupsidentified in this study as at the greatest risk of irondeficiency are children ages 1–5 years, adolescents ages12–14 years, and women ages 1844 years.

The finding that low TS is more prevalent than lowhemoglobin is in agreement with earlier studies.Prasad, for example, cites six studies that reported lowTS without low Hb, and he states that iron deficiencyis two to three times more common than iron-deficiency anemia.zo

Despite the high prevalence of low transferringsaturation (iron deficiency), iron-deficiency anemia(low Hb/TS) is not widespread according toNHANES I data. Although the condition seems to bea problem among l-year-old children, only a smallproportion of individuals over age 2 were classified aslow Hb/TS. Apparently, iron intake adequately meetsthe iron needs associated with growth in adolescenceand with menstrual blood loss during the femalereproductive years, although the low values found fortransferringsaturation and total iron-binding capacityindicate that iron nutrition is not optimal.

DietThe recommended di~’ary allowances (RDAs) for

iron intake are set at ,U mg/day for infants up to 6months of age, children ages +10 years, males ages 19and over, and post-menopausal women ages 50 andover. A higher intake of 15 mg/day is recommendedfor children 6 months to 4 years of age. An intake of18 mg/day is recommended during adolescence (ages11–18 years) and the female reproductive period (ages18–50 years).lz

Previous analyses of NHANES I data showed thatiron intakes are well below RDAs for much of the U.S.population. 1,2Groups whose intakes most often aredeficient include those at greatest risk of anemia andiron deficiency-infants and preschoolers, adolescents,and females during their reproductive period. Thisstudy shows that low hemoglobin and low circulatingiron are common in these groups, which suggests a

clear association between iron intake and iron status.However, a comparison of mean dietary iron intakeand iron status did not support that hypothesis; therewas a significant association between low iron intakeand poor iron status only among children ages 1–3years.

The study did not detect relationships betweeniron intake and iron status. Although such relation-ships may not exist or may be confounded by otherfactors, the 24-hour recall method used to collectdietmy intake data may have been inaccurate orunreliable or individuals may have been misclassifiedas to their iron status.

The 24-hour recall is a method used to collectinformation about a single day’s dietary intake. It maynot reliably reflect daily dietay intake52,73nor intakeover a longer period. The relationship between ironintake and iron status may not lie with dietary ironintake alone, but with a more refined measure (perhapsestimates adjusted for bioavailability). Yet when ironintake was adjusted experimentally in the multivariateanalyses, the adjusted estimate was not a betterpredictor of low hemoglobin in children than (log)total mean iron intake.

Misclassification of persons with adequate ironstatus into low iron status categories may have tiectedthe ability to uncover a relationship between ironintake and iron status. If, for example, the referencestandard for hemoglobin were inappropriate, it wouldbe difficult to differentiate anemic and nonanemicgroups. Even if this reference standard did identifycorrectly the majority of healthy individuals (a highlyspecific diagnosis), dilution of the anemic group couldoccur if the prevalence of low hemoglobin was low, asit was in this study. Thus the unreliability of thedependent variable-iron status—may have dimin-ished the power to detect effects.

The measures of iron status also maybe too broad.Although hemoglobin and transferringsaturation meas-urements can reflect the presence of iron deficiency,neither gives any clue to the cause—whether irondeficiency results from decreased iron intake or in-creased iron requirements. Yet iron loss due tobleeding has different implications from other causesof increased requirements, because bleeding lossesoften cannot be covered by dietary iron intakes.

Finally, the consistent relationship between report-ed iron intake and iron status among young childrenmay reflect a greater real association between ironintake and iron status. This may be due to severalfactors associated with this age group-increased ironneeds due to growth, maternal awareness and controlover what the child eats, or a more limited diet.

Socioeconomic, demographic, andhealth factors

The multivariate logistic analyses were designed topoint to associations between iron status and selected

29

dietary, socioeconomic, demographic, and health fac-tors. In this approach, race, age, and sex are allcorrelates of iron status. Race was significantly associ-ated with iron status in three high-risk groups—children ages 1–3 years, menstruating women ages12–54 years, and persons ages 65–74 years. Age was asignificant correlate for children and women, and sexwas a significant correlate for the elderly.

Two socioeconomic factors were significant evenwhen dietary and health variables, race, age, and sexwere taken into account. Women and children frompoor homes, as measured by the poverty income ratio,are most likely to have poor iron status. The educationof the head of the household was significantly associ-ated with iron status in all three groups.

Educational attainment generally is regarded asassociated with income, and years of education of thehead of household was correlated with iron status.However, the relationship of education to iron status isnot simply a function of income. Even when PIR wastaken into account in multivariate analysis, years ofeducation of the head of household had a distinctsignificant association with iron status.

Several health variables also were associated withiron status among adults in this study. For menstruat-

ing women ages 12–54 years, oral contraceptive use,reproductive history, and prior diagnosis of anemia allwere important. Individuals who have ever beendiagnosed as anemic were more likely to have pooriron status later in life. Anemia routinely is treatedwith oral iron preparations, but after the treatmentperiod is over, the dietary or health conditions thatcontributed to anemia may exert their effect again. Therole of oral contraceptives in reducing blood losses andof pregnancy in draining iron stores are welll estab-lished as factors affecting iron status.52

For persons ages 65–74 years, findings on thegeneral medical examination and problems related toeating were associated with iron status in the multivar-iate approach used in this study. However, theseassociations did not occur across all analyses, so theconclusions for the elderly are less certain.

These multivariate analyses are only one approachto looking at correlates of iron status in the UnitedStates. They are not designed to show any cause-and-effect relationships, but the findings for the socioeco-nomic, demographic, and health variables did indicatethat other factors besides iron intake should beinvestigated in future studies of iron status.

30

References

‘National Center for Health Statistics: Catalog of Publications ofthe National Center for Health Sfatirtics. DHEW Pub. No. (PHS)81-1301. Public Health Service. Washington. U.S. GovernmentPrinting Office, Sept. 1981.

‘National Center for Health Statistics: Plan and operation of theHealth and Nutrition Examination Survey, United States,1971-1973. Vital and Health Statistics. Series 1-Nos. 10a and 10b.DHEW Pub. No. (HSM) 73–13 10. Health Resources Administra-tion. Hyattsville, Md, Feb. 1973.3p.R D~~m, M-A. Siimes, and A. Stekel: Iron deficiency ininfancy and childhood. Am. J. Clin. Nutr. 33:86-118, 1980.

4J.D. Cook and C.A. Finch: Assessing iron status of a population.An. 1 C]in.ZWr. 32:2115-2119, 1979.

5J.D. Cook: Absorption of food iron. Fed. Proc. 36(7):202&–2032,1977.

6C v- Mmre: Iron, in N.S. Goo&ti and M.E. Shils, eds., M~ern

Nutrition in Health and Direase. Philadelphia. Lea and Febiger,1976. p. 312.

7P. ~fiowsky: Iron Dejciency . . . A Public Health Problem.Paper presented at a Symposium on Hematologic Disease inChildren for the City of New York, Department of Health, Bureauof Child Health, 1975.

@n~m for Dis~e Gntrol: Ten-~tate ~Ul~-fiOn ~WVeY:

1968-1970. DHEW Publication No. (HSM) 72–8 132. Washington.U.S. Government Printing Office, 1972.

9JD. ~k, C.A. Finch, and N.J. Smith:’ Evaluation of the irons&us of a population. Blood 48:449455, 1976.

1~.M. o’N~, O.G. Abrahms, M.B. Korhs, and others: Theincidence of anemia in residents of Missouri. Am_ L Clin. Nutr.29:1158-1166, 1976.

1‘H. White and E. Johnson: Hemoglobin concentrations of 2263university women. 1 Am, College Health Assoc. 17:255-256, 1969.

lz~~tt= on Diem flowmces, Food and Nutrition Botid:Recommended Dietary ’Ailowancex Washington, National Acade-my of Sciences, 1980

ISM.S. ch~ey ~d M.L. ROSS: Nutrition. Boston. Houghton-MifHin, 1971. p. 152.

14c A Finch: Iron Mebbolism in Nutrition Reviews’ pre.Sent. .Knowledge in Nutrition. New York. The Nutrition Foundation,1976. pp.28&289.

ISJ.D. ~k ~d C.A. Finch: Iron nutrition. West ~ Med-122:474481, June 1975.

16”A group of experts”; Report to the Department of Agricultureconcerning the calculation of available dietary iron. WorkshopReport. Seattle, Washington. Sept. 13–14, 1976.lTER. Monsen L. Ha]lberg, M. Layrisse, and others: Estimation ofav;lable dieta& iron. Am. J. Clin. Nutr. 31:134141, 1978.lL?EBjom-Rasmussen, L. HWberg, B. Isaksson, and B. Arvidsson:

Food iron absorption in man: applications of the two-pool extrinsictag method to measure heme and nonheme iron absorption fromthe whole diet. J Clin Inv. 53:247–255, 1974.19M. @rrisse ~d C. Ma&ez-Tom=: Model for measuring

dietary absorption of heme iron: test with a complete meal. Am. J.Clin. Nutr. 25:401+11, 1972.

Z13Sprasad: Chapter 5, Iron in Trace Elements and Iron in HumanMetabolism. New York. Ple~um Medical Book Co., 1978.21A Tumbul], F. Cleton, and C.A. Finch: Iron absorption. IV. Theabsorption of hemoglobin iron. Z Clin. Inv. 41(10):1897-1906,1962.ZZER. Monsen: Simplfied method for calculating available dietary

iron. Food and Nutrition News. 51(4): 1-4, 1980.

ZSJ.W. Harris and R.W. Kellermeyer: Iron metabolism andironlack anemia. The Red Cell Production, Metabolism, Destruc-tion: Normal and Abnormal. Cambridge. Harvard UniversityPress, 1970. pp. 64-148.MM Layrisse C. Maflhez-Tomes, and M. Gonzalez: Measure-ment of the total daily dietary iron absorption by extrinsic tagmodel. Am. 1 Ciin_Nutr. 27: 152–162, 1974.MJD cook M. Lay&e, C. Martinez-Torres, R. Walker, and. .others: Food iron absorption measured by an extrinsic tag. J. ClinInv. 51:805–815, 1972.

Z6J.D. ~ok ~d E.R. Monsen: Food iron absorption in humansubjects. III. Comparison of the effect of animal proteins onnonheme iron absorption. Am_ J Clin. Nutr. 29:859-867, 1976.

zTpB Disler, L3.R.Lynch, R. W. Charlton, and others: The effect oftea-on iron absorption. Gut 16:193-200, 1975.

ZBE-R. Monsen ~d J.D. Cook: Food iron absorption in humansubjects. IV. The effects of calcium and phosphate salts on theabsorption of nonheme iron. Am. J. Clin. Nutr. 29: 1142–1 148,1976.

Z9E, Bjom.R~m~sen: Iron absorption from wheat bread—inlluence of various amounts of bran. Nutr MetaboI. 16, 1974.

30J.D. cook amj E. Monsen: Food iron absorption in human

subjects. II. The effect of EDTA on absorption of dietary nonhemeiron. Am. J. Clin. Nutz 29:614620, 1976.

3IR. Green, R.w, Charlton, T.H. Both well, and others: Body ironexcretion in man: a collaborative study. Am. J. Med. 45:33&353,1968.

3ZI+N. Munro: Iron absorption and nutrition. Fed. Proc.

36(7):2015-2023, 1977.33cA- Finch: Iron metabolism, in present Knowledge in Nutrition

(4th cd.). New York. The Nutrition Foundation, 1976.

34K,M. HambridgeandA.M. Mauer: Trace elements, lnLabOra~o-

ry Indices of Nutritional Status in Pregnancy, Committee onNutrition of the Mother and Preschool Child. Food and NutritionBoard, Washington, National Research Council, National Acade-myof Sciences, 1978.

35c.v. Moore: Iron, in N.S. Goodhart and M-E. Shils, eds.,

Modern Nutrition in Health and Disease. Philadelphia. Lea andFebiger, 1976. p. 310.

36L. NilsSon and L. Solvell: Clinics] studies on oral COntracep-

tives—a randomized double blind crossover study of four differentpreparations (AnovlarR mite, LyndiolR mite, OvulenR and Volo-dam). Acts Obstet, Gunecol. Scan. 46(Suppl.) (8), 1967.3TM.Thein, G.M. BeatOn,H. Milne, and M. Veen: Oral contracep-tive drugs: some observations on their effect on menstrual loss andhematologic indices. Canad. Med. Assoc. J 101, 1969.

3EH.J. Weiss: Aspirin — a dangerous drug? L4MA 229(9):122 l–1222, 1974.

39M.D. Caliendo, D, Sanjur, J, Wright, and G. Cummings:Nutritional status of preschool children. 1 Am. Dietet. Assoc.71:20-26, 1977.

‘D. Sanjur, M. LaChapell, and L. Parker: Feeding patterns ofpreschool children among rural families. Research Monograph.New York State College of Human Ecologj. New York, CornellUniversity, 1973.

41S.M, Garn and D, C, Clark: Problems in the nutritional assess-

ment of black individuals. AJPH. 66(3):262–267, 1976.

42s.M. Gain, H.A. Shaw, and K. McCabe: Black-white hemoglo-

bin differences during pregnancy. Ecol Food Nutri. 5:99, 1976.

13L.D. Meyers, J.p, Habicht, C. L. Johnson: C02’13PO12entsof the

difference in hemoglobin concentrations in blood between blackand white women in the United States. Am. J. Epidemiol.109:539-549, 1979.

‘G-M. Owen and A. Yanochik-Owen: Should there be a differentdefinition of anemia in black and white children? AJPH67(9):865-866, 1977.

45P,R. Dallman, G.D. Barr, C. M.. Allen, and H.R. Shinefield:Hemoglobin concentration in white, black and oriental”children: isthere a need for separate criteria in screening for anemia? Am. J.Clin. Nutr. 31:377–380, 1978.

46R.R. Frerichs, L.s. W.ebber, S.R. Srinivasan, and G.S. Berenson:

Hemoglobin levels in children from a biracial southern community.Am, J Pub. H., 67:841, 1977.

17G.M. Owen, A,H. Lubin, and P.J. Garry: Hemoglobin levels

according to age, race and transfemin saturation in preschoolchildren of comparable socioeconomic status. J. Peal. 82:85&851,1973.

~S.M. Garn, N.J. Smith, and D.C. Clark: Lifelong differences inhemoglobin levels between blacks and whites. J. Natl. Med. Amn.75(2):91, 1975.

qgrnternationa] Nutritional Anemia Consultative Group: Guide-

lines for the Eradication of Iron-De@ciency -Anemia. Report ofMeeting, June 1977.

50S.J,Foman: Jnfant Nutrition. Philadelphia. W.B. Saunders, 1974.

slM,A. Koerper and P.R. Dallman: Serum iron concentration and

transferring saturation in the diagnosis of iron deficiency inchildren: normal developmental changes. Y. Peal.91:870, 1977.

52Committee on Maternal Nutrition/Food and NutritionBoard/National Research Council: Maternal Nutrition and theC’ourxeof Pregnancy. Washington. National Academy of Sciences,1970.

5ZG, Christ&is, (cd.): Nutritional assessment in health programs.

Reprinted from the American Journal of Public Health, Part 2,Vol. 63, N’ov. 1973.

51B,S, Woflhington, J. Vermeersch, and S. Williams: Nutrition inPregnancy and Lactation. St. Louis. C.V. Mosby Co., 1977.

jfNational Center for Health Statistics, S.S. Jock and P-w. Ries:

Current estimates from the National Health Interview Survey:United States, 1979. Vital and Health Statistics. Series 1OI-NO.136.DHHS Pub. No. (PHS) 81-1564. Public Health Service. Washing-ton. U.S. Government Printing Office, Apr. 1981.

jANational Center for Health Statistics, C.E. Burnam: Eclentulous

persons. Vital and Health Statistics. Series 1O-NO.89. DHIEW Pub.No. (HRA) 74-1516. Public Health Service, Washington. U.S.Government Printing Office, June 1974.

sTNational Center for Health Statistics, B.A. Feller: pP%dHICe of

selected impairments: United States, 1977. Vital and HealthStatistic.r. Series 1O-NO. 134. DHHS Pub. No. (PHS) 81-1562.Public Health Service. Washington. U.S. Government PrintingOffice, Feb. 1981.

j8National Center for Health Statistics, B.A. hti, S.A. Eklund,

JR. Landis, and others: Diet an{ dental health, a study ofrelationships: United States, 197 1–74. Vital and Health Statistics.Series 11-No.225. DHHS Pub. No. (PHS) 82–1675. Public HealthService. Washington. U.S. Government Printing Office, 1981.59R B. McGandy: Nutrient intakes and energy expenditures in men

of different ages. J. Gerontol. 21, 1966.

‘M.V. Dibble: Evaluation of the nutritional status of elderlysubjects with a comparison between spring and fall. J. Am. GeriatSOC. 15, 1967.

GINationalCenter for Health Statistics, S. Abraham, M. D,, ~rroll,

C.M.V. Dresser, and C.L. Johnson: Dieta~ Intake Source Data:United States, 1971–74. DHEW Pub. No. (PHS) 79–1221, PublicHealth Service. Washington. U.S. Government Printing Office,Sept. 1979.

GzNational Center for Health Statistics, C.M.V. Dresser, M. D-

Carroll, and S. Abraham: Food consumption profiles of white andblack persons ages 1–74 years: United States, 197 1–74. Vital andHealth Statistics. Series 11-No. 210. DHEW Pub. No. (PHS)79–1 658. Public Health Service. Washington. U.S. GovernmentPrinting Office, May 1979.

G3E.D. Schlenker: Accessibility of food to the aged. New England

Conference on Food, Nutrition, and Health. Final Report, Vol. 2:The Position Papers. Boston. Sept. 1979. pp. 343401.

‘National Nutrition Policy Study—1974: Hearings Before theSelect Committee on Nutrition and Human Needs of the U.S.Senate. (93rd Congress). Part III—Nutrition in Special Groups.Nutrition and the Elderly. June 19, 1974. pp. 985–992.

GsNation~ Center for Health Statistics: HANES I, hematology andclinical procedures developed or utilized by the Centers for DiseaseControl, Bureau of Laboratories, 197 1–1975. NCHS InstructionManual, Part 16, Hyattsville, Md. Public Health Service, 1980.

GGp,R, Dallman and M,A. Siimes: Percentile curves fOr hemoglobin

and red cell volume in infancy and childhood, J. Peal. 94:2631,1979.67JB Miale: Laborato~ Medicine—Hematology. 5th Ed. St. Louis.. .Mosby. p. 444, 1977.

32

‘D.F. Bainton and C.A. Finch: The diagnosis of iron-deficiencyanemia. Am J. Med., 37:62-70, 1%4.

@T.H, Bothwe]l and C.A. Fksch: Chapter 8, Iron metabolism, in

Iron-Dejciency Anemia. Boston. Lhtle Brown and Co., 1962.

mcon~ultint panel for the Study of Dietary Intake and Food

Patterns Related to Nutritional Anemia Personal Communication.

71E-R. Monsq I.N. Kuhn, and C.A. Finch: Iron stitlls of

menstruating women. Amer. .l Cii~ Nutr, 20:842–849, 1967.

72G.M. Owen, K.M. Kram, P.J. Garry, J.E. Lowe, and A.H.Lubin: A study of nutritional status of preschool children in theU. S., 1968-1970. Pediatr. 53(4, Part II): 597-646, 1974.

T3Centers for Diseme Contio]: Nutrition Surveillance Annual

Summary 1978. DHEW Public Health Service.

Tqp-R. D~lm~: Personal Communication.

TSU.S. Depafiment of Health, Education, and Welfare: A Compari-

son and Analysis of Examined and Unexamined Persons onMedical History Characteristics for the First Round of the Healthand Nutrition Examination Study. HSM—1 10-73-371. Preparedby Westat Inc. Rockville, Md. Jan. 24, 1974.

7%. Press and S. Wilson: Choosing between logistic regression anddiscriminant analysis. 1 of the .4merican Statistical Association.73:699-705, 1978.

TTDR Cox: The Analysis CJfBinary Data. London. Methuen & CO.,Ltd:, i970.

TUM.Go]dstein and W. Dillon: Discrete Discriminant Analysis. New

York. John Wiley and Sons, 1978.

‘~. Gordon: Hazards in the use of the logistic function withspecial reference to data from prospective cardiovascular studies. J.of Chronic Diseases. 27:97–102.80M HalPenn, w-c. Blackwelder, and J.I. Verter: Estimation of

the multivariate logistic risk function: a comparison of thediscriminant function and maximum likelihood approaches. 1 ofChronic Diseases 24:125–158, 1971.

SIJ. T~ett, J. Comfie]d, and W. Karmel: A mukivariate analysis of

the risk of corona~ heart disease in Framingham. 1 of ChronicDiseases 20:5 11–524, 1967.

SZW. Hauck: Developing the multinominal logit for use in dah

analysis. Harvard University, Department of Statistics, TechnicalReport No. CP-19. 1975.

HH.L. Gray ad W.R. Schucany: The Generalized Jackkn!Ye

Statistic. New York. Marcel Dekker, Inc., 1972.

sJR MilleC Jack~fing variances. Annals of Mathematical Statis-

tics; 39:567–582, 1968.

8SJ-A. Ande~on: Sepaate sample logistic discrimination. Biometri-

cs. 59:19-35, 1972.

E6N. M~tel: SYnthetic retrospective studies and related topics.

Biometrics. 29:479486, 1973.

STN-Breslow and W. powers: Are there two logistic regressions for

retrospective studies? Biometrics. 34:100-105, 1978.

NU.S. Bureau of the Census: Poverty increases by 1.2 million in

1970. Current Population Reports. Series P.60, No. 77, Washing-ton. U.S. Government Printing Office, May 7, 1971.

~gM Orshansky: counting the poor: another look at the Povefiy

profile. Social Security Bulletim 28(l): Jan. 1965.

‘M. Orshansky: Who’s who among the poor a demographic viewof poverty. Social Security BuIIetin. 28(7): 3–31, July 1965.

91u.s. Bureau of the Census: Revision in pOVerty s~tktic$

1959-1968. Current Population Repotis. Series P-23, No. 28,Washington. U.S. Government Printing Office, Aug. 12, 1969.

~Zcircular No. A-46, Transmittal Memorandum No. 9, Executive

Office of the President, Bureau of the Budget, Aug. 29, 1969, andExhibit L (rev.).

33

List of detailed tables

1.

2.

3.

4,

5,

6.

7.

&

Hemoglobin levels for males 1-74 years, with mean, standarderror of the mean, and selected percentiles, by race and age:United States, 1971.74 .................................................................. 36Hemoglobin levels for females 1-74 years, with mean, stan-dard error of the mean, and selected percentiles, by race andage: United States, 1971.74 .......................................................... 37Hematocrit levels for males 1–74 years, with mean, standarderror of the mean, and selected percentiles, by race and age:United States, 1971-74 .................................................................. 38Hematocrit levels for females 1-74 years, with mean, standarderror of the mean, and selected percentiles, by race and age:United States, 1971.74 .................................................................. 39Serum iron levels for males 1–74 years, with mean, standarderror of the mean, and selected percentiles, by race and age:United States, 1971.74 .................................................................. 40

Serum iron levels for females 1–74 years, with mean, standarderror of the mean, and selected percentiles, by race and age:United States, 1971.74 .................................................................. 41

Total iron binding capacity levels for males 1-74 years, withmean, standard error of the mean, and selected percentiles, byrace and age: United States, 1971–74 .......................................... 42

Total Iron binding capacity levels for females 1-74 years, withmean, standard error of the mean, and selected percentiles, bvrace and age: United States, 1971 -74 ................... ....................... 43

9. Transferringsaturation levels for males 1–74 years, with mean,

10,

11,

12,

13.

14.

15.

34

standard error of the mean, and selected percentiles, by raceand age: United States, 1971–74 .................................................. 44

Transferring saturation levels for females 1–74 years, withmean, standard error of the mean, and selected percentiles, byrace and age: United States, 1971-74 .......................................... 45

Percent and standard error of percent of persons ages 1-74years, with low hemoglobin levels, by sex, race, and age:United Statesl 1971-74 .................................................................. 46

Percent and standard error of percent of persons ages 1-74years, with low transfernn saturation levels, by sex, race, andage: United States, 1971.74 .......................................................... 47

Percent and standard error of percent of persons ages 1-74years with the combination of low hemoglobin and lowtransferring saturation levels, by sex, race, and age: United

States, 1971-74 .............................................................................. 48

Z-statistics for comparison of percent of persons with low ironstatus indicators for sex within age and race within sex and agecomparisons: United States, 1971-74 .......................................... 49

Percenf and standard error of the percent for women 18-44years with low hemoglobin, low transferringsaturation, and thecombination of low hemoglobin with low transferringsaturation,with Z-statistic for comparison of pregnancy status, by race:United States, 1971.74 .................................................................. 50

16.

17.

10.

19.

20.

21.

22.

23.

24.

25.

26.

27.

2a.

Iron intake for persons 1-74 years, with mean, and standarderror of the mean, by sex, hemoglobin status, race, and age:United States, 1971.74 .................................................................. 51Iron intake for persons 1-74 years, with mean, and standarderror of the mean, by sex, transferringsaturation status, race,and age: United States, 1971-74 ..................................................Z-statistics for comparison of iron intake for persons wit!h lowor adequate hemoglobin and for persons with low or adequatetransferringsaturation, by specied age and sex-race categories:United States, 1971.74 ..................................................................Z-statistics for comparison of low and adequate hemoglobinstatus for children 1-3 years, with statistic and standard errorof the statistic, by selected characteristics: United States,1971.74 ...........................................................................................Z-statistics for comparison of low and adequate hemoglobinstatus for white children 1–3 years, with statistic and standarderror of the statistic, by selected characteristics: United States,1971.74 ...........................................................................................Z-statistics for comparison of low and adequate hemoglobinstatus for black children 1–3 years, with statistic and standarderror of the statistic, by selected characteristics: United States,1971.74 ............................................................................................Logit analysis statistics showing the relationship betweenselected characteristics and hemoglobin status for children1-3 years: United States, 1971.74 .................................................Z-statistics for comparison of low and adequate hemoglobinstatus for menstruating women ages 12-54 years, with statisticand standard error of the statistic, by selected characteristicsUnited States, 1971-74 ......................................................... ........Z-statistics for comparison of low and adequate transferringsaturation status for menstruating women ages 12-54 years,with statistic and standard error of the statistic, by sefectedcharacteristics: United States, 1971-74 .............................. ........Z-statistics for comparison of low and adequate hemoglobincombined with transferring saturation status for menstruatingwomen ages 12–54 years, with statistic and standard error ofthe statistic, by selected characteristics: United States,1971.74 ...........................................................................................Z-statistics for comparison of low and adequate hemoglobinstatus for whhe menstruating women ages 12-54 years, with

statistic and standard error of the statistic, by selectedcharacteristics: United States, 1971-74 .......................................Z-statistics for comparison of low and adequate hemoglobinstatus for black menstruating women ages 12-54 years, withstatistic and standard error of the statistic, by selected

characteristics: United States, 1971-74 .......................................Z-statistics for comparison of low and adequate transferringsaturation status for white menstruating women ages 12-54years, with statistic and standard error of the statistic, by

52

53

54

55

56

57

57

58

59

60

61

selected characteristics: United States, 1971 -74..., .................... 62

29. Z-statistics for comparison of low and adequate transferringsaturation status for black menstruating women ages 12-54years, with statistic and standard error of the statistic, byselected characteristics: United States, 1971 –74 ........................ 63

30. Z-statistics for comparison of low and adequate hemoglobincombined with transferringsaturation status for white menstru-

ating women ages 12–54 years, with statistic and standarderror of the statistic, by selected characteristics: United States,1971.74 ........................................................................................... 64

31. Z-statistics for comparison of low and adequate hemoglobincombined with transferringsaturation status for black menstru-ating women ages 12-54 years, with statistic and standarderror of the statistic, by selected characteristics: United States,1971.74 ........................................................................................... 65

32. Logit analysis statistics showing the relationship betweenselected characteristics and hemoglobin status for menstruat-ing women ages 12–54 years: United States, 1971 -74 ............... 66

33. Logit analysis statistics showing the relationship betweenselected characteristics and transferring saturation status formenstruating women ages 12-54 years: United States,1971-74 ........................................................................................... 66

34. Logit analysis statistics showing the relationship betweenselected characteristics and hemoglobin and transferringsatu-ration status for menstruating women ages 12–54 yearsUnited States, 1971–74 . ........... ........ ................ ........ ............... 66

35. Z-statistics for comparison of low and adequate hemoglobinstatus for persons ages 65-74 years, with statistic andstandard error of the statistic, by selected characteristics:United States, 1971.74 .................................................................. 67

36. Z-statistics for comparison of low and adequate transferringsaturation status for persons ages 65-74 years, with statisticand standard error of the statistic, by selected characteristics:United States, 1971.74 .................................................................. 68

37. Z-statistics for comparison of low and adequate hemoglobinstatus for white persons ages 65-74 years, with statistic andstandard error of the statistic, by selected characteristics:United States, 1971.74 .................................................................. 69

36. Z-statistics for comparison of low and adequate hemoglobinstatus for black persons ages 6S74 years, with statistic andstandard error of the statistic, by selected characteristics:United States, 1971.74 .................................................................. 70

39. Z-statistics for comparison of low and adequate transferringsaturation status for white persons ages 65-74 years, withstatistic and standard error of the statistic, by selectedcharacteristics: United States, 1971-74 ..... .......... ...................... 71

40. Z-statistics for compa;son of low and adequate transferringsaturation status for black persons ages 65-74 years, withstatistic and standard error of the statistic, by selectedcharacteristics: United States, 1971-74 ....................................... 72

41. Logit analysis statistics showing the relationship betweenselected characteristics and hemoglobin status for personsages 65-74 years: United States, 1971 -74...............................,.. 73

42. Logit analysis statistics showing the relationship beWeenselected characteristics and transferring saturation status forpersons ages 65-74 years: United States, 1971 -74 ................... 73

35

mm

bu)m.

nmtnw

=f

00

00

0.0

00

00

0000”

00”0000

mmkcocoom

mm-t

oo

oo

o~o

oo

.o

00000000”

00”

-b-m--m

mN

ocue%

-o~-.oo~~0000000000

wo.cyr-~

a.m.~

qq

Ocumlmlm=fmm

%-a

-------.

..

m.-

nmnq

:!::::::::::::::::::

(r

36

y~oqqoqqyo+

..ojcNslm

mc9m

m.

..s-....

..

~coqcyqqw

caoq

...oJm

ltNolN

m(N

7!-.7-..7..

.

((.“

?’“

37

0000000000l-*m

wco

ooJN

..

@Y

mm

c9m***ee

1=

Oooooooooc

WC

9*U-)W

ON

N.C

Olm

mm

mm

***a

..I-o*kwuJm

mOo)lnbmaabwo)

ym

-m-0-yN-m-0.l-.

%

-nlcuommr-~m

*-

u

0.000.0.0.0.0.0.(

Om

”o-o

?Ym

=rm

[W

C5W

-*V9W

T.

O-%-ooool-o-f

comr=mmm”m”-o”.

c9ccl@Y

c9cclm@

-Jev.

00.0.0,

0.0.0.0.0.0.

mmcabmmmooo

@)m

@’)eY

cocY@

-J-*v

-.0.00.0.0.0.0,

0.(

=tln

m-b

mm

moa’lc

C9m

mm

mm

c?lwm

c

%-0

000

00

00c

C5u-lu-)m

batimr-r

@)elc9c9c9crll?Y

mm

c

l-oo~~

ooooo

-=f-rm

mw

mm

mm

mC

omm

mlm

cclcoc’)mm

oooo,oo~ooo

Wm

ln(or-r=r-m

amI?Y

@Y

@Y

cclm.?)m

mcnm

ooooo~oooo

e-=fm

mhw

mrnm

”ai@

Ym

lc.-Jl?)c’)mm

crlmm

00000.00000

Nm

emfafar-r-r-h

mcc.m

m.m

mm

mnm

o~l-oo.000o(

m4cum

-lnu-J*.cDm

.~C

Tl@

YcT

l@Y

mm

rJmm

c

6

%3m..a

:)1

.I-&.

39

0000.7.-.0.00.0,

a-t-mkomb.mm

cNw

c9w

bm

mh

br-

o.o.o-ooo~~?-o

wm

Ln@

lmN

e*oo)

ml=

r@Y

u-JuJkwcD

r-u3

0.0.0.0,0.0,0.0,0,

r.

wm

med*m

o*7km

lmm

*lnlnwu7uY

uY

u-)rd.mm

cf’)bor=~

?-wy~

a.w,oo.y

0..23..l-.m

JN-N

-r

y~.~

lqqul~=

fclwm

ur--3mcJ-w

cim-

wm

wm

oa~ooo

.T-l-l-l-.

7-m

-l--

emN

--

40

0.0.0.0.0.

r.r.

0.0.~

.m

mc.m

wm

mm

cuom

Jwlnu-Ju3k-)lnw

cDw

Iqm

.~~q~qrncuww

mm

cDuJ.-m

l-r=l=mwmo)oomom---

L1-J

wln

cclwocd

lnr-cl

obmmmmmwn.n

mo-

m-u-

o-m-o---

0-o.-

cdm.om

~r-:h

,rI’

::::::.::..:;:::,:::

::::::::::::

;:::.

:::::;::

.:;:::::::::::::::.::

Aci;

ui(k.-

AL

,

mm(uta(

a)a)a)w(

%%

>%

:

l-oooJoqoJ~0.

tim*m

mw

m”

Nco.

N1-N

cuclT

r9-mlc

-----------

0.0.0.0,

0.r.

0.0.0.c

minm

wecum

ao.-?

C9-tm

mm

wm

wu

C911I-N

.

OW

--N-O

WW

UY

c.Jkw

ww

l-!mcu

--cu-.-m

--

t

41

—

CD

oyqQyom

,qq.

mT

-w

0.

s-m-d

om

mrm

--mm

mm

Ulv-fm

**mm

mm

rd.

a:..?L

%

Y5i

tic1,(

x:

ommmmooomw

mmm~mwmmoo

vm-h

-w--m

~-~

-m-v-w

-N

am

mc

uu

mC

D-

br=momobmxfm

Lnmocuammm-k

N*bc9cT

lw-cD

uYeY

-el

.

::::::::::::::;:

lno~ym.~o.t-<

T-c.J

@Y

eweu

)-U

-J--i-nm

lwln..

0C

flcY@

l@)ccl@

lmm

@)

mo’-cu.-ocDco!n

wm

ww

.t-mw

--N

-?-m-l-

1I

43

Ic

xvm=

?6a)>

1

b.ym

.yo-q

o.~

-.l00N

uYW

am-m

lC

oC3m

’mm

mm

mn,

u_)O

lIY@

YC

D0N

KIb

@3

mm

mm

mooom

wo+m

mm

mnw

mm

mom~w-ommo~ofx

c9aJlnN

cY

wm

moo

*Nhul

---N

-m-r-w

-cvm

-rw-m

~-:-:-h--

--m-m

-uim:~m

a

?Urr

::~:

:::::::::::::

;:::::::.::::,::::::;:.

:::::::::

45

Table 11. Percent and standard error of percent of persons ages 1–74 years, with low hemoglobin Ievels,l by sex, race, and age:United States, 1971-74

Male Female z

Race and age Number of Estimated Standard Number of Estimated Standardexamined population Percent error of the examined population Percent error of thepersons in thousands percent persons in thousands percent

All racess

1 year ......................................2-3 years, ...............................4-5 years, ...............................6-11 years ..............................12–”4 years,,, .........................15-17 years, ...........................18-44 years, ...........................45–54 years ............................55-64 years ............................65–74 years ............................

White

1 year ......................................2-3 years ................................4-5 years ................................6-11 years ..............................12-14 years ............................15–17 years ............................18–44 years.., .........................45–54 years, ...........................55-64 years, ...........................65–74 years ............................

Black1 year, .....................................2–3 years ................................4-5 years ................................6–11 years ..............................12–14 years .,,.........................15-17 years ............................18-44 years ............................45-54 years .,.....,,,......,,.......,,.55-64 years, ...........................65-74 years ............................

272577549974519487

2,128740569

1,501

199425419734396373

1,786607484

1,293

70138127229121108307126

77270

1,692 19.33,538 8.73,379 1.7

11,913 3.06,348 7.66,196 2.3

35,425 5.211,156 5.38,993 13.15.495 14.9

1,401 14.92,991 7.42,857 0.9

10,070 1.65,434 4.05,297 1,1

31,376 4.510,045 4.6

8,195 12.94,969 12.4

281 42.1473 18.7495 6,4

1,732 11,5877 30.2009 8.6

3,529 12.01,059 12.0

707 17.0482 40.4

3.591.760.830.621.651.020.721.242.121.34

3.791.910.690.691.400.810.741.282.291.37

0.854.983.263,166.264,052.784.346.424.4a

254535571974534457

4,613788639

1,728

179401418734401357

3,679665531

1,426

7012014823412898

065118105294

1,6253,3903,301

11,4556,0435,932

36,99812,15910,0547,275

1,3282,8572,7779,6305,1465,132

32,24310,876

9,1156,600

260501506

1,745831775

4,3541,235

868651

12.68.31.93.23.44.94.94.72.04.3

10.57.21.72.12.23.43.44.20.73.1

21.815.3

2.89,1

11.514.815.89.1

12.416.6

3.121.790.860.051.181.510.481.130.830.73

3,441,940,950,,791,,101,“440,,451,,170,,540,,69

7,404,772,032.824.235.381.663.974.023,26

! Age- and sex-speclflc hemoglobin standards are shown m table A.

2 Does not include data for pregnant women,

3 Includes data for races which are not shown as separate categories

46

Table 12. Percent and standard error of percent of persons ages 1-74 years, with low transferring saturation Ievels,’by sex, race, and age:United States, 1971-74

Male Femalez

Race and age Number of Estimated Standard Number of Estimated Standardexamined population Percent error of the examined population Percentpersons

error of thein thousands percent pareons in thousands percent

All racesa1 year ......................................2-3 years ................................4-5 years ................................8-11 years ..............................12-14 years ............................15-17 years ............................13-44 years ............................45-54 years ............................55-64 years ............................6%74 years ....... .....................

White

1 year ......................................2-3 years ................................4-5 years ................................6-11 years ..............................12-14 years ............................15-17 years ............................1644 years ............................45-54 years ............................55-64 years ............................65-74 years ............................

Black

1 year ......................................2-3 years ................................4-5 years ................................6-11 years ..............................12-14 years ............................15-17 years ............................1844 years ............................45-54 years ............................55-64 years ............................65-74 years ............................

113342552979526485

2,126715556

1,545

87251411719393360

1,753579464

1,232

2682

130250131119337130

05294

7242,2033,372

11,9256,2606,248

35,44911)137

8,9905,496

6251,B772,859

10,0865,3435,354

31,39810,026

8,1924,969

99275488

1,738883813

3,5191,073

715488

50.827.023.116.9

8.66.33.67.27.74.2

47.727.522.217.67.96.83.67.47.94.3

70.924.327.313.212.93.74.25.75.53.0

7.053.602.691.801.831.650.611.451.700.77

8,034.233.272.132.041.990.671.631.880.87

13.367.105.696.874.392.601.643.053.711.49

77313571988535462

4,661789632

1,701

56233405720389350

3,637658514

1,375

2076

161262141110956126115318

5052,0703,295

11,4338,0365,980

37,05212,16510,037

7,283

4381,7932,7589,6255,1335,160

32,29010,880

9,1036,606

59264515

1,735818796

4,3731,243

848656

38.324.218.012,813.116.316.4

9.58.17.1

37.121.017.312.712.416.315.8

9.17.26.9

‘53.043.922.912.916.417.220.713.117.6

9.4

8.313.632.411.592.192.580.811.571.630.93

9.684.002.821.862.512.960.911.681.711.03

.

8.544.973.114.685.401.974.515.332.45

I Persons with a transfernn salutation below 16.0 percent snd a TIBC above 250 ~g/dl were classified as low treneferrin saturation.ZDoes not include data for pregnant women.3Includes data for races which are not shown as separate categories.

NOTE Asterisks indicate estimates that do not meet this report’s slandards of reliability.

47

Table 13. Percent and standard error of percent of persons ages 1-74 years, with the combination of low hemoglobin and low tra(nsferrin

saturation Ievels,l bv sex, race, and acre: United States, 1971–74

Male Femalez

Race and age Number of Estimated Standard Number of Estimated Standardexamined population Percent error of the examined population Percent error (of thepersons in thousands percent persons in thousands percent

AH races3

1 year ......................................2-3 years .... . .............. .......4-5 years ......................... .......6–1 1 years ..............................12–14 years...,, ........ ...............15–17 years..., ........................18-44 years ............................45-54 years ............................55–64 years .,, .........................65-74 years, ...........................

White

1 year ......................................

2–3 years ................................4–5 years ................................6–1 1 years ..............................12–14 years ............................15–17 years ............................

18-44 years..., ........................

45–54 years ............................55-64 years...,,, ......................65-74 years...,, .......................

Black

1 year ......................................

2-3 years ................................4-5 years ................................6–1 1 years ..............................12-14 years ... .........................15–17 years.., ........ .................18-44 years.,, .........................45–54 years.,, .........................

55-64 years .,, .........................

65-74 veals ............................

109333524927497

2,013690527

1,469

84

247

3946B9374342

1,672558

4431,181

2577

127

228121108306126

77

270

7222,1773,372

11,9246,2956,237

35,50111,1429,0595,495

6231,8642,850

10,0835,3795,339

31,45210,0320,2614,969

99264497

1,745881810

3,5401,074718483

17.71.60.90.51.50.20,11.32.21.8

17.00.90.70,40.90.00.11.32.41.0

22.26.42.01.55.31.20.30.80.81.5

5.491.040.610.340,820.310.100.650.960.52

6.150.900.630.360.730.000.120.721.090.58

12.474.181.861.213.061.570.471.191.521.11

72296547931510431

4,397741602

1,607

53227395691377331

3,469618494

1,305

1865

14823412898

863118105294

4792,0153,303

11,4466,0565,947

36,96912,1619,9687,279

4211,7502,7809,6215,1575,152

32,20610,8769,0346,601

49252509

1,746813772

4,4011,241843655

9.93.30.30.61.02.02.82.40.81.6

11.21.30.30,40.61.62.22.40.61.3

‘0.017,30.21.43.14,17.42.02.74.1

5,;~B

1.!560.:350.:380.661.010.:370.840.!540.47

6.!501.”130.410.:360.[301.030.:37().92

0.!520.47

.

7.04O.!js

1.”152.:303.001.341.{)3

2.:?71.73

~Age- and sex-speclflc hemoglobin stendards are found in Table A. Persons with a transferring saturation below 16.0 percent and a TIBC above 250 ~g/dl were classified aslow transferring saturation.

ZDoes not include data for pregnant women.3Includes data ~orraces which are not shown as separate categories,

NOTE: Asterisks indicate estimates that do not meet this raport’s standards of reliability.

48

Table 14. Z-statisticsi for comparison of percent of persons with lowiron status indicators for sex within age and race within sex and agecom~arisons: United States. 1971-74

Iron status indicator

Low

Comparisons LowLow hemoglobin

transferringhemoglobin saturation and low

transferringsaturation

Males versus females

1 year .................................2-3 years ...........................4-5 years ...........................6-11 years .........................12-14 years .......................15-17 years .......................18-44 years .......................45-54 years .......................55-64 years .......................65-74 years .......................

Black males versuswhite males

1 year..., .............................2-3 years ...........................4-5 years ...........................6-11 years .........................12-14 years .......................15-17 years .......................1B44 years .......................45-54 years .......................55-64 years .......................65-74 years .......................

Black females versuswhite females

1 year .................................2-3 years ...........................4-5 years ...........................6-11 years .........................12-14 years .......................15-17 years .......................16-44 years .......................45-54 years .......................55-64 years .......................65-74 veals .......................

1.410.16

-0.17-0.17

2.00-1.42

0.350.364.876.93

-2.63-0.75-1.65-3.064.07-1.02-2.61-1.63-0.60-5.98

-1.38-1.57-0.49-2.39-2.13-2.05-6.40-l.l B-2.41+.06

Z-statistics

1.150.551.411.59

-1.58-3.26

-12.62-1.00-0.17-2.10

-1.49-0.85-0.79

1.07-1.03

0.95-0.34

0.490.580.75

*-2.43-0.98-0.06-0.75-0.15-2.264.83-1.86-0.63

1.02-0.91

0.84-0.19

0.48-1.70-6.96-1.03

1.270.59

-0.37-1.29-0.66-0.87-1.40-0.76-0.41

0.360.050.24

*

-2.240.15

-0.03-1.05-0.79-3.75

0.19-0.664.78

I Significant values are z & 1.96 based on two-tailed test (P=o.05).NOTE Asterisks indicate statisl!cs that do not meat this report’s standards of

reliability,

49

Table 15. Percent and standard error of the percent for women 16-44 years with low hemoglobin,l low transferringsaturation,z and the cornbioation oflow hemoglobin with low transferringsaturation, with Z-statistic for comparison of pregnancy status, by race: United States, 1971-74.

Pregnant Non-pregnant

Race and iron status indicators Number of Standard Number of Standardexamined Percent

Z-statisticerror of the examined Percent error of the

persons percent persons percent

All racesa

Low hemoglobin ...................................................... 196 31.9 4.99 4,613 4.9 0.48 !j.as

Low transferringsaturation ...................................... 195 22.1 4.46 4,661 16.4 O.B1 1.26Low hemoglobin and low transferring

saturation ................................................................ 1B5 6.5 2.72 4,397 2.8 0.37 ‘1.35

WhiteLow hemoglobin ...................................................... 152 28.4 5.49 3,679 3.4 0.45 4.54Low transferring saturation ...................................... 147 21.7 5.10 3,637 15.8 0.91 ‘1.14LOW hemoglobin and low transferring

saturation ................................................................ 141 7.0 3.22 3,469 2.2 0.37 ‘1.40

BlackLow hemoglobin ......................................................Low transferringsaturation ......................................Low hemoglobin and low transferring

saturation ................................................................

36 52.940 27.6

36 5.0

12.4810.60

5,45

865 15.8 1.86 ,2.94956 20.7 1.97 10.64

863 7.4 1.34 10.43

1Age. and sex-specific hemoglobin standards are found in table A.ZPersons with a transferringsaturation below 16.0 percent and a TIBC above 250 pg/dl were classified as having low transferringsaturation.3Includes data for races which are not shown as separate categories.

50

.u3*m

cuw

-cQl-w

(Qcycyw

-m.@

Yyycr+

~0000000000

mwh7wr-o

mow

-w

.w-+

=fyoyqw

~00000-0000

“%wc1

g=<

if91

T-.=$.

?1

akiigig!

51

.r-om

mm

cooo~btn-abm

cuacocO

ooo.o”o-ooc

Iqo.yo)*qqqq*.

mt-moio.mootnm--

@Y

1-*u-Jl-mcu*cY

el*mwowmr-wmr-

I-I?’Jlomm

cqu)w!--

.-

mmawm.momw

~lqo-)~

mlq-qm

.u-)o

0000.CWF--7

m>,rr

l-d.

N*

●

m-iuacu.=:<:

~,,

T

al-aEL?.

53

Table 19. Z-statistics for comparison of low and adequate hemoglobin statusl for children 1-3 years, with statistic and standard error ofthe statistic, by selected characteristics: United States, 1971-74

Low hemoglobin Adequate hemoglobin

Characterktics Number of Standard Number of Standardexamined Statistic

Z!-statisticerror of the examined Statistic error of the

persons statistic persons sta fistic

Dietary intake

Total iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .... . ....Total nonanimal tissue iron intake (mg) ...........Total protein intake (mg) ...... ....... . ........ ...... . .Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) .....Estimated available iron (mg) ........................ ....Estimated available iron with tea (mg) .............Vitamin C (mg) .......................)..............................

Socioeconomic

Age of mother at child’s birth (years) ....... ......Birth weight (pounds) ...........................................Age (years) .............................................................Poverty income ratio (index) ...................... ........Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........

Other2Pica ..........................................................................Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Ever breastfed .......................................................Is child the first born ...........................................Premature birth ......................................................Present illness ........................................................Regular aspirin use, ..............................................

177177177177177177177177177177

173175177173173177177173

177177177177177142175159177

Mean Mean

6.2073.63

1.504.70

49.1311.9537.16

0,650.65

60.04

24.017.172.191.57

4,253.935.74

32.4210.72

Percent

37.4930.2610.1542.4616.9016.43

7.6745.4441.00

0.492.780.190.412.231.201.820.060.067.49

0.560.130.090.11

297.530.291.020,37

1,2451,2451,2451,2451,2451,2451,2451,2451,2451,245

1,2321,2291,2451,2271,2261,2451,2441,201

7.4782.55

1.855.61

52.5114.6837.82

0.820.82

76.93

25.237.312.562.22

3,877.034.76

31.7711.94

Percent

0.210.940.080.200.930.550.680.030.033.47

0.240.050.040.06

111.170.060.370.14

-2.39-3.05-1.73-2.02-1.40-2.06-0.33-2.63-2.66-2.05

-1.94-0.104.15-5.21

1.193.250.60

-3.12

5.465.183.425.574.234.673.025.925.55

1,2451,2451,2451,2451,241

9361,2411,0721,245

16.2744.5620.7152.9326.5017.70

7.1632.9532.07

1.642.111.722.121.881.671.102.151.98

3.37-2.56–2.76-1.61-2.06-0.25

0.161.981.52

! Age-and sex-specific hemoglobin standards are found in table A.ZStatistics [or Ihese characteristics represent only those persons who answered yes on the questionnaire.

54

Table 20. Z-statistics for comparison of low and adequate hemoglobin statusl for white children 1-3 years, with statistic and standard error

of the statistic, by selected characteristics: United States, 1971-74

Low hemoglobin Adequate hemoglobin

Characteristics Number of Standard Number of Standard Z-sfatisticexamined Statistic error of the examined Statistic error of thepersons statistic persons sta fistic

Dietaty intakeTotal iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (mg) ...........Total protein intake (mg) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) .....Estimated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ......................................................

SocioeconomicAge of mother at child’s birth (years) ..............Birth weight (pounds) ...........................................Age (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........

OtherzPica ..........................................................................Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Ever breastfed .......................................................Is child the first born ...........................................Premature birth ......................................................Present illness ........................................................Regular aspirin use ...............................................

105105105105105105105105105105

102105105103103105105102

10510510510510584

10593

105

5.9173.14

1.304.61

49.6510.9538.70

0.600.59

61.86

24.117.402.191.73

4,268.045.62

31.7810.92

Percent

37.6330.67

9.7945.7618.0013.524.75

43.3545.41

Mean

0.453.200.200.372.891.462.360.060.06

10.52

0.690.140.110.15

391.630.361.240.49

7.096.764.357.295.625.603.117.717.29

960960960960960960960960960960

951951960948949960960925

960960960960957719951815960

7.4982.77

1.605.69

52.8714.3638.51

0.810.81

78.77

25.447.392.592.36

3,909.254.63

31.3812.22

Percent

18.0645.8221.3956.0728.2018.04

5.7231.7432.16

0.251.060.080.231.060.610.760.030.034.06

0.270.060.040.06

128.740.080.390.16

1.862.411.991.972.182.151.132.452.26

-3.10-2.86-2.36-2.46–1 .05-2.16

0.06-3.00-3.33–1 .50

–1 .800.07

-3.42-3.87

0.872.700,31

–2.51

2.67-2.08-2.43-1.37–1 .69-0,75-0.29

1.441.16

1Age- and sex-specific hemoglobin standards are found in table A.ZStatistics for these characteristics represent only those persons who answered yes on the questionnaire.

55

Table 21. Z-statistics for comparison of low and adequate hemoglobin status’ for black children 1-3 years, with statistic and standard errorof the statistic, by selected characteristics: United States, 1971-74

Low hemoglobin Adequate hemoglobin

Characteristics Number of Standard Number of Standard Z-statisticexamined Statistic error of the examined Statistic error of thepersons statistic persons sta fistic

Dietary intakeTotal iron intake (mg) .... ......................................Log total iron intake (log mg) ...... ................ ..Total animal tissue iron intake (mg) .. ... .Total nonanimal tissue iron intake (mg) ....... .Total protein intake (ma) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) .. .Estimated available iron (mg) .............................Estimated available iron with tea (mg) . .........Vitamin C (mg) ......................................................

Socioeconomic

Age of mother at child’s birth (years) ..............Broth weight (pounds) ...........................................Age (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........

OtherzPica ... ...... ...... ...... ...... ........ ....... . . .... .Regular vitamm use,, ............................................Regular mineral use ..............................................Regular vitamin and mineral use...,, ..................Ever breastfed .......................................................Is child the first born ...........................................Premature bifih ......................................................Present illness........................................................

72727272727272727272

7170727070727271

727272727250706672

Mean

7.0475.02

2.104.94

47.6514.8632.77

0.020.82

54.76

23.746.502.191.13

4,212.496.10

34.2710.10

Percent

37.0928.5111.1832.8913.9024.7616.2651.5628.23

1.195.520.391.003.512.122.630.120.128.29

1.020.240.150.12

445.460.521.830.48

8.547.985.578.316.128.506.619.23

,7.96

2652B52652852052852B5285285285

281278285279279285284276

2852852852852B4217278257285Regular aspirin use...,, ..........................................

Mean

7.3281.14

2.185.14

50.1816.7033.41

0.860.86

65.13

23.876.802.521.31

3,664.615.61

34.2710.16

Percent

19.6736.5216.3039.2415.3015.5016.5640.0731.46

0.392.120.180.282.181.251.490.070.075.78

0.580.130.070.10

202.350.230.980.28

3.534.283.264.343.203.693.354.594.13

-0.22--1.03--0.18--0.19-4.61--0.78-0.21-0.28--0.28--1.03

--0.11--1.09--2.01-1.14

1.120.870.00

-0.14

1.80.4.88-0.79.4.68-0.20

1.00-0.04

1.12-0.36

IAge- andsex-specifichemoglobmstandardsare foundintable AZStatisticsforthese characteristicsrepresentonlythosepersonswhoansweredyes onthe questionnam

56

Table 22. Logit analysis statistics showing the relationship between selected characteristics and hemoglobin status for children 1-3 years:United States, 1971-74

Characteristicsbgit analysis statistics

Coefficierrt Sfandard error t-fesf stafisfic 1 Odds rafio 2

Constant ............................................................................................. 1.494 0.672Race ...................................................................................................

. . . . . .0.891 0.249 3.580 2.45

Age ...................................................................................................... -0.505 0.097 -5.216 0.60Poverty income ratio ........................................................................ -0.243 0.104 -2.336 0.78Education of head of household (years) ..................................... -0.036 0.018 -2.018 0.96Hca ..................................................................................................... 0.712 0.211 3.376 2.05Log total iron intake ........................................................................ -0.015 0.004 -3.664 0.98

10nly characteristicswitha f-test statistic > 1.96 (p < .05) are shown.~See appendixI for discussionof oddsratios.

Table 23. Z-statistics for comparison of low and adequate hemoglobin status! for menstruating women ages 12-54 years, with statistic andstandard error of the statistic, by selected characteristics: United States, 1971–74

Low hemoglobin Adequate hemoglobin

Characteristics Number of Standard Number of Sfandard Z-stafisticexamined Statisfic error of the examined Statistic error of thepersons sta fistic persons sfatistic

Dietary intake

Total iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (mg) ...........Total protein intake (mg) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) .....Estimated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ......................................................

SocioeconomicAge (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

Marriedz ............................................ ......................

OtherzPica ..........................................................................Regular vitamin use ..............................................Regular mineral use............................................l.Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special diet .............................................................Oral contraceptive usage in pasf 6 months ....Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General fintings .....................................................Abdominal operations ................... ......................

Acte beaan menstruating Wears) .......................

259259259259259259

259259

259246240259259253250

220

259259259259259259259259259259258

259

Mean

9.5095.16

3.595.91

63.2631.1932.07

1.341.30

68.15

31.102.30

3,908.124.64

4,23410.1B11.01

Percent

65.30

1.8521.7313.4336.9449.38

5.7615.6063.7037.7029.10

0.8324.50

Mean

12.85

0.421.870.290.272.802.171.580.080.086.88

1.100.13

248.430.231.120.350.27

4.69

1.263.043.186.064.662.173.304.484.524.230.854.01

0.13

3.6463,6463,6463,6463,6463,6463,6463,6463,6463,646

3,6463,5423,5423,6463,6453,5693,627

3,117

3,6463,6463,6463,6463,6463,6463,6463,6463,6463,6463,6393,646

Mean

9.9996.09

3.756.23

65.0931.0234.06

1.411,36

80.27

27.722.67

3,687.134.28

39.3911.6711.55

Percent

65.70

1.2326.9119.9848.9946.37

7.0522.4153.9625.7233.00

0.8920.73

Mean

0.120.520.080.060.760.630.450.020.022.22

0.270.04

65.340.050.290.080.07

1.28

0.271.130.991.701.240.671.041.241.091.170.231.01

3,646 12.61 0.04

-1.12-0.48-0.54-1.12-0.62

0.06-1.21-0.66-1.00-1.68

2.98-2.63

0.861.562.56

-4.72-1.94

-0.08

0.46-1.79-1.97-1.91

0.62-0.92-1.93

2.092.56

-0.91-0.07

0.91

1.73

t Age-andsex-specifichemoglobinstandardsare foundintableA.2Statisticsforthese characteristicsrepresentonlythosepersonswhoansweredyes onthe questionnaire.

57

Table 24. Z-statistics for comparison of low and adequate transferring saturation statusl for menstruating women ages 12–54 years, withstatistic and standard error of the statistic, by selected characteristics: United States, 1971-74

Low transferringsaturation Adequate transferringsaturation

Characteristics Number of Standard Number of Standardexamined Statistic

Z-statisticerror of the examined Statistic error of the

persons statistic r3ersons sta fistic

Dietary intake

Total iron intake (mg) ................ ........ ........ ......Log total iron intake (log mg) ............ ...... .......Total animal tissue iron intake (mg) . .. . .Total nonanimal tissue iron intake (mg) ..........Total protein intake (mg) ........... ................. ......Total animal tissue protein intake (mg) . ....... ..Total nonanimal tissue protein intake (mg) .. ..Estimated available iron (mg) . ..... ....... ........ .Estimated available iron with tea (mg) ........ ..Vitamin C (mg) . ...... ......................... ...................

Socioeconomic

Age (years) .............................................................Poverty income ratio (index) ........ ........ .............Estimated family income (dollars) ...... ....... .....Number of persons in household (units) ..........Age of head of household (years) ... ...... ........Education of head of household (years) ..........Education of examinee (years) ......... ........ .......

Marriedz ...................................................................

OtherzPica ............,,,,....,.,. ....................... ...... ....... ....... ,.Regular vitamin use,, ............................................Regular mineral use., ............................................Regular vitamin and mineral use ......................Regular aspirin use ...............................................Special diet .............................................................Oral contraceptive usage in past 6 months...,Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General findings .....................................................Abdominal operations ................. .............. ........

Age began menstruating (years) .... ...... ....... ...

664664664664664664664664664664

664648648664663640662

571

664664664664664664664664664664662664

664

Mean

9.5194.57

3.555.96

61.802,94032.40

1.331.30

73.75

29.242.45

3,727.554.61

40.4211.0511.23

Percent

64.00

2.1121.3615.6839.2346.38

8.5314.0860.8530.7234.08

5.3923.55

Mean

12.61

0.271.190.170.201.661.331.030.050.055.07

0.650.09

152.990.120.670.200.15

3.01

0.842.392.124.022.901.632.032,842.692,761.322.47

0.08

3,2413,2413,2413,2413,2413,2413,2413,2413,2413,241

3,2413,1423,1423,2413,2413,1743,223

2,774

3,2413,2413,2413,2413,2413,2413,2413,2413,2413,2413,2353,241

3,241

Mean

9.9996.11

3.766.24

65.5831,0734.51

1.411.38

80.81

27.702.67

3,670.174.23

39.4011.9611.59

Percent

0.130,550.090,090.840.680.480.020.022.34

0.290,04

69.340.050.300.080.07

65.60

1.0730.5120.7950.6747.22

7.7623.1053.1725.6433.03

1.5720.71

Mean

1.35

0.271.211.071.891.320.701.111.311.151.240.331.07

12.63 0.04

-1.62-1.17-1.07-1,26-2.01-1.12-1.86-1.49-1.51-1.26

2.14-2.28

0.342.811.39

-4.17-2.22

LO.48

1.1843.42–’2.15-2.57-0.26

13.43-:3.91

,2.45

1.7413.352.821.06

-0.23

~Persons with a transferringsaturation below 16,0 percent and a TI BC value above 250 pg/dl were classified ae having a low transferringsaturation,ZStatistics for these characteristics represent only those persons who answered yes on the questionnaire.

58

Table 25. Z-statistics for comparison of low and adequate hemoglobin combined with transferring saturation status’ for menstruating womenages 12-54 years, with statistic and standard error of the statistic, by selected characteristics: United States, 1971 –74

Low hemoglobirr/ Adequate hemoglobin/transfemn saturation tran.sferrin saturation

Characteristics Number of Standard Number of Standard Z-statistic

examined Statisfic error of the examined Statistic error of thepersons statistic persons statistic

Dietary intakeTotal iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (mg) ...........Total protein intake (mg) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) .....Estimated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ......................................................

Socioeconomic

Age (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units), .........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

Married~ ...................................................................

Other~Pica ..........................................................................Regular vitamin use ..............................................Regular mineral use ....................... ......................Regular vitamin and mineral use ......................Regular aspirin use ...............................................Special diet .............................................................Oral contraceptive usage in past 6 months ....Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .......................... ......................General findings.....................................................Abdominal operations ...........................................

Age began menstruating (years) .......................

143143143143143143143143143143

143139139143143139143

133

143143143143143143143143143143142143

143

Mean

9.7396.55

3.696.04

61.7431.0130.78

1.371.31

62,02

34.412.33

3,903.794.04

44.549.11

10.75

Percent

47.60

2.3014.6011.0027.0047.90

7.509.10

72.3038.7034.00

0.0023.40

Mean

0.612.570.440.394.153.351.900.120.117.50

1.440.17

339.390.311.440.500.37

6.49

18.694.433.937.356.273.303.615.616.115.950.005.32

13.08 0.18

3,7623,7623,7623,7623,7623,7623,7623,7623,7623,762

3,7623,6513,6513,7623,7613,6833,742

3,212

3,7623,7623,7623,7623,7623,7623,7553,7623,7623,7623,75!53,762

3,762

9.9595.97

3.746.21

65.1731.0434.14

1.411.37

79.85

27.672.66

3,678.214.28

39.4411.8511.54

Percent

62.40

1.2020.9019.8040.9047.20

7.8022.4053.8025.9032.80

0.9020.90

Mean

12.61

0.120.510.080.090.760.610.440.020.022.18

0.270.04

64.250.050.280.080.06

1.20

0.271.110.981.751.220.651.021.221.071.150.230.99

0.03

-0.350.22

-0.11-0.42-0.80-0.01-1.72-0.34-0.52-2.28

4.59-1.91

0.651.813.46

-5.47-2.10

-2.24

0.06-3.13-2.18-2.90

0.11-0,09-3.54

3.222.060.20

-3.910.46

2.54

~Age- and sex-apeclflc hemoglobin standards are found in table & persons with a iransfemn saturation below 16.0 percenl and TIBC value above 250 ~g/dl were classn7edas having a low transferringsaturation

~statistics for these charactenst(cs represent only those persons who anewered yes on the questlonnare.

59

Table 26. Z-statistics for comparison of low and adequate hemoglobin status’ for white menstruating women ages 12-54 years, withstatistic and standard error of the statistic, by selected characteristics: United States, 1971-74

Low hemoglobin Adequate hemoglobin

Characteristics Number of Standard Number of Standard Z-stlatisticexamined Statistic error of the examined Statistic error of thepersons statistic persons statistic

Dietary intake

Total iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (ma) ...........Total protein intake (mg) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) .....Estimated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ......................................................

Socioeconomic

Age (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

Marriedz ...................................................................

OtherzPica ..........................................................................Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special diet .............................................................Oral contraceptive usege in past 6 months ....Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General findngs .....................................................Abdominal operations ...........................................

Age began menstruating (years) ........................

133133133133133133133133133133

133125125133133131132

119

133133133133133133133133133133133133

133

Mean Mean

9,8897.70

3.646.04

66.8532.4434.41

1.411.35

70.88

33.282.66

3,939.534.35

42.4010.4211.12

76,10

0.0022.9214.7637.2350.49

5.3515.7367.1045.1329.63

0.0026.96

12,86

0.592.500.450.344.163.372.070.120.11

10.07

1.550.17

362,920.271.540.500.36

5.66

0.005.474.616.446.502.934.746.116.475.940.005.77

0.17

3,0243,0243,0243,0243,0243,0243,0243,0243,0243,024

3,0242,9392,9393,0243,0242,9633,010

2,615

3,0243,0243,0243,0243,0243,0243,0243,0243,0243,0243,0243,024

3,024

10.0696.56

3.766,29

65.6930.9834.72

1.421.36

81.12

27.862.77

3,636.444.18

39.3212.0511.65

Percent

67.60

0.6030.5020.6950.6947.40

6.1422.2853.3625.6132.69

0.9021.39

Mean

12.60

0.130.560.090.100.850.690.500.030.032.44

0.300.04

71.960.050.310.090.07

1.37

0.211.261.111.971.360.751.141.361.191.260.261.12

0.04

-0.30CI.44C1.le

41.7111.27CI.42

41,14

41.08

~1.26

41.99

3.44-0.52

0.020.621.96

-3,20-1.36

1,41

-2.85-1.35-1.29-1.55

0.47-0.92-1.35

2.192.94

-0.50-3.51

0.95

1,55

~Age- and sex-specific hemoglobin standards are found in table A.2Statistics for thase characteristics represent only those persons who answerad yes on tha questionnaire.

60

Table 27. Z-statistics for comparison of low and adequate hemoglobin statusl for black menstruating women ages 12-54 years, withstatistic and standard error of the statistic, by selected characteristics: United States, 1971-74

Low hemoglobin Adequate hemoglobin

Characteristics Number of Standard Number of Standard Z-statisticexamined Statistic error of the examined Statistic error of thepersons statistic persons statistic

Dietary intake

Total iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (mg) ...........Total protein intake (mg) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) .....Estimated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ......................................................

Socioeconomic

Age (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

126126126126126126126126126126

126123123126126122126

109

126126126126126126126126126126125126

126

Mean

0.7490.06

3.005.66

56.0328.6B27.35

1.211.19

62.65

26.701.58

3,847.995.25

42.229.68

10.78

Percent

Marriedz ...................................................................

OtherzPica ..........................................................................Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special diet .............................................................Oral contraceptive usage in past 6 months ....Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General findings .....................................................Abdominal operations ...........................................

Age began menstruating (years) ........................

43.60

5.5719.3310.7636.3647.15

6.5915.3356.8622.7728.04

2.5019.54

Mean

12.84

0.57 6222.76 6220.28 6220.46 6223.63 6222.24 6222.40 6220.09 6220.09 6228.75 622

I .3a 6220.15 603

326.07 6030.39 6211.63 6210.46 6060.37 617

Mean

9.3492.18

3.615.73

59.9931.3728.62

1.341.33

73.04

26.521.78

4,105.145.15

39.9810.3410.70

Percent

7.13 502

3.06 6225.28 6224.14 6228.72 6226.67 6223.31 6224.82 6226.62 6225.60 6226.00 6222.09 6215.30 622

48.00

6.4715.5a12.3334.7541.17

5.1623.4558.9724.9341.17

0.6015.20

Mean

0.22 622 12.74

0.301.410.190.211.901.490.960.050.055.33

0.650.09

151.280.160.730.100.16

3.35

1.482.181.984.002.961.332.552.962.602.960.472.16

0.10

-0.94-0.60–1 .58-0.14-0.97-1.00-0.49-1.26–1 .34-1.01

0.12-1.13-0.72

0.261.25

-1.340.20

4.66

-0.260.66

4.340.170.020.40

-1.49-0.29-0.35-1.96

0.890.76

0.41

1Age- and sex-specific hemoglobin standards ara found in table A.zStatistics for these characteristics represent only those pereons who answered yes on the questionnaire.

61

Table 28. Z-statistics for comparison of low and adequate transferring saturation status! for white menstruating women ages 12-54 years,with statistic and standard error of the statistic, by selected characteristics: United States, 1971–74

Low transferring saturation Adequate transferring saturation

Characteristics Number of Standard Number of Standard Z-statisticexamined Statistic error of the examined Statistic error of thepersons sfa tistic persons statistic

Dietary intakeTotal iron retake (mg) ..........................................Log total iron intake (log mg) ..... ........ ...... . ..Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (mg) ..... ....Total protein retake (mg) .... ...... ........ ........ ....Total animal tissue protein intake (mg) ... ........Total nonanimal tissue protein intake (mg) ....Estimated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ....... ..............................................

Socioeconomic

Age (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

Marriedz ...................................................................

Otherz

Pica ...... . ................................ .............................Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special diet .............................................................Oral contraceptive usage in past 6 months ....Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General findings .....................................................Abdominal operations ...........................................

Age began menstruating (years) ........................

505505505505505505505505505505

505492492505505497504

Mean

9.5294.67

3.585.94

62.5529.4533.10

1.331.29

73.09

29.342,50

3,696,644.46

40.1711.2211.34

Percent

0.311.360.200.231.951.541.190.060.065.70

0,750,10

177.410.140.770.230.17

437

505505505505505505505505505505504505

67.30

1.2822.8516.2539.7846,47

9,0414.4159.7930.9334.44

2.7824.14

Mean

3,37

0.752.802.464.633.331,912.343.273.093.171.102.86

505 12.60 0.09

2,6522,6522,6522,6522,6522,6522,6522,6522,6522,652

2,6522,5722,5722,6522,6522,5972,638

2,297

2,6522,6522,6522,6522,6522,6522,6522,6522,6522,6522,6472,652

2,652

Mean

10.1196.74

3.796,32

66,3431.1435.20

1.431.39

.92.19

27.852,79

3,625.264.13

39.2912.1811.69

67.70

0,4032.1621.8852.4147.90

7.9423,0452.6225.6832.50

1.6421.30

Mean

12.62

0.140.600.100.100.930.750.530.030.032.61

0.320.05

76.920.060.330.090.08

1.46

0.101.361.202.091.460.791.231.451.271.360.371.19

0.04

-1.73-1.39-0.92-1.50-1.76-0.99-1.61-1.44-1.62‘-1 .45

1.83-1.91

0.372.261.05

-3.88-1.91

--0.11

1.14--2.99--2.05--2.48-0.39

0.53--3,26

2.001.570.560.980.92

-0.21

TPersons with a transferringsaturation below 16,0 percent and TIBC value above 250 pg/dl were classified as having a low transferring saturation.ZStatistic for these characteristics represent only those persons who answered yes on the questionnaire.

62

Table 29. Z-statistics for comparison of low and adequate transferring saturation statusf for black menstruating women ages 12-54 years,with statistic and standard error of the statistic, by selected characteristics: United States, 1971–74

Low transferring saturation Adequate transfemin saturation

Characteristics Number of Standard Number of Standard Z-statisticexamined Statistic error of the examined Statistic error of thepersons statistic persons sta tisk

Dietary intakeTotal iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (ma) ...........Total protein intake (mg) .....................................Total animal tissue protein intake (ma) ............Total nonanimal tissue protein intake (mg) .....Estimated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ......................................................

SocioeconomicAge (yearn) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

MarriedZ ...................................................................

OtherzPica ..................................................... ....................Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special diet .............................................................Oral contraceptive usage in past 6 months ....Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General findings.....................................................Abdominal operations ...........................................

Age began menstruating (years) ........................

159159159159159159159159159159

159156156159158151158

134

159159159159159159159159159159158159

159

Mean Mean

9.4693.96

3.406.06

57.1229.1028.02

1.341.31

77.93

28.561.63

3,921.415.55

41.959.92

10.52

Percent

43.50

7.3312.0512.0535.7945.79

5.3611.9767.5329.3831.83

0.2019.82

Mean

0.502.480.310.413.142.531.930.080.08

11.53

1.300.14

289.410.301.340.370.30

6.43

3.103.873.878.015.932.683.865.571.855.540.544.74

589589589589509

5B9589589589

589570570589509577585

477

5895095895895895095095B95895B9588509

12.73 0.18 589

9.1390.89

3.515.62

59.4030.5028.90

1.301.29

69.55

26.401.69

4,032,335.50

40,2610.1610.75

Percent

47.00

6.5717.0411.9036.3941.64

6.3223.5957.6325.2037.33

1.0715.84

Mean

12.73

0.311.470.190.221.951.471.030.060.065,00

0.670.09

156.190,170.760.190.17

3.43

1.532.322.004.193.051.502.623.052.693.000.642.26

0.10

0.561.07

-0.310.95

-0.62-0.48-0.400.400.200.67

1.48-0.36-0.34

0.141.10

-0.57-0.67

-0.48

0.22-1.10

0.03-0.07

0.62-0.31-2.49

1.561.26

-0.87-1.04

0.76

0.00

1Personswitha transferringsaturationbelow t 6.0 percentandTIBC valueabove250 pg/dl were classifiedas havinga lowtransferringsaluralion,ZStatisticsforthesecharacterishcsrepresentonlythosepersonswhoansweredyes on lhs questionnaire.

63

Table 30. Z-statistics for comparison of low and adequate hemoglobin combined with transferring saturation status~ for white menstruatingwomen ages 12-54 years, with statistic and standard error of the statistic, by selected characteristics United States, 1971-74

Low hemoglobin/ Adequate hemoglobin/transferring saturation transferring saturation

Characteristics Number of Standard Number of Standard Z-statistic

examined Statistic error of the examined Statistic error of thepersons statistic persons statistic

Oietary intake

Total iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (mg) ...........Total protein intake (mg) .....................................

Total animal tissue protein intake (mg) ....... .Total nonanimal tissue protein intake (mg) .. .Estimated available iron (mg) .............................Estimated available iron with tea (mg) .......... .Vitamin C (mg) ......................................................

Socioeconomic

Age (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

MarriedZ ...................................................................

Otherz

Pica ........................................................ ....... .......Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special diet .............................................................Oral contraceptive usage in past 6 months ....Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General findngs .....................................................Abdominal operations ...........................................

888B8888808888888888

88840480888788

82

800888888888888880808888

88

Mean

10.2099.45

4.066.14

65.7633.3632.40

1.461.40

59.92

35.752.56

4,054.374.61

44.739.21

10.68

Percent

0.783.140.600.445.534.592.300.160.039.19

1.620.21

451.190.361.900.64

47.81

56.90

0.0015.3011.0025.8045.00

6.6010.5072.2044.6034.10

0.0024.30

Mean

Age began menstruating (years) ........................ 13.04

8.20

0.005.765.019.197.964.034.897.167.967.560.006,86

0.20

3,0693,0693,0693,0693,0693,0693,0693,0693,0693,069

3,0692,9802,9803,0693,0693,0073,054

2,652

3,0693,0693,0693,0693,0693,0693,0693,0693,0693,0693,0633,069

3,069

Mean

10.0396.46

3.766.27

65.7730.9734.80

1.421.36

81.02

27.822.78

3,629.194.26

39.3512.0511.65

Percent

63.10

0.5830.6020.6050.6047.80

8.0022.3053.3025.9032.40

0.9021.50

Mean

12.60

0.130.550.090.100.840.680.490.020.022.42

0,300.04

71.400.050.310.090.07

1.41

0.211.251.101.951.350.741.131.351.191.270.261.11

0.04

0.210.940.46

-0.29-0.00

0.52-1.02

0.250.53

-2.22

4.29–1 .01

0.930.902.80

-4.43-0.02

-0.75

-2.83-2.60-1.91-2.644.35-0.29-2.35

2.592.350.22

-3.530.40

2.15

1Age- and sex-specific hemoglobin standards are found in table A persons with a transferring saturation below 16.0 percent and TIBC value above 250 pg/d! were classifiedas having a low transferring saturation.

ZStatistics for these characteristics represent only those persons who answered yes on the questionnaire.

64

Table 31. Z-statistics for comparison of low and adequate hemoglobin combined with transfemin saturation statusl for black menstruatingwomen ages 12-54 years, with statistic and standard error of the statistic, by selected characteristics United States, 1971-74

Low hemoglobin/ Adequate hemoglobirrliransferrin saturation fransfem”n saturation

Characteristics Number of Standard Number of Standard Z-statistic

examined Statistic error of the examined Statistic error of thepersons statistic persons statistic

Dietary intakeTotal iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (ma) ...........Total protein intake (mg) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) .....Estimated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ......................................................

SocioeconomicAge (yearn) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

Marned2 ...................................................................

OtherzPica ..........................................................................Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special diet .............................................................Oral contraceptive usage in past 6 months ....Ever been pregnant ..............................................Ever been anemic ........................... ......................Any trouble eating .................................................General tindings.....................................................Abdominal operations ...........................................

55555555555555555555

55555555555255

51

555555555555555555555455

55

Mean

8.1806.90

2.405.71

48.6123.2325.38

1.061.03

69.03

29.981.61

3,424.235.58

43.898.76

11.01

Percent

14.90

9.8012.5011.0030.8057.40

9.704.50

72.6018.4033.60

0.1020.60

Mean

0.914.110.390.844.653.363.300.130.13

13.43

2.170.23

464.430.562.090.760.56

7.48

6.016.706.33

12.5210.015.974.199.027.779.550.068.19

693693693693693693693693693693

693671671693692676

560

693693693693693693693693693693692693

Age began menstruating (years) ...... ................. 13.22 0.39 693

9.3092.06

3.615.70

60.4431.5528.89

1.341.32

70.73

26.481.74

4,058.455.09

40.1710.2410.72

Percent

55.80

6.0016.0012.1035.9042.10

5.6023.3057.9025.4036.00

0.9015.70

Mean

12.71

0.281.330.170.201.791.370.930.050.054.87

0.620.08

143.270.150.700.170.15

3.15

1.362.091.863.832.B21.312.412.822.482.740.542.07

0.09

-1.17-1.19-2.62

0.01-2.37-2.29-1.02-2.02-2.064.12

1.55-0.53-1.30

0.641.69

-1.900.50

-5.04

0.62-0.50-0.17-0.39

1.470.67

-3.091.560.66

-0.24-1.40

0.56

1.27

1Ag+ andsex-specifichemoglobinslandardsare foundintableA personswith a Iranafeminasturationbelow16.0 percent and TleC value above 250 pg/dl were classifiedas having a low Iranslemn aalwahon.

ZStatistics for these characteristics represent only those persons who answered yes on the questionnaire.

65

Table 32. Logit analysis statistics showing the relationship between selected characteristics and hemoglobin status for menstruating womenages 12–54 years: United States, 1971-74

CharacteristicsLogit analysis statistics

Coefficient Standard error t-test statistic Odds ratio 2

Constant ............................................................................................. -1.901 0.674 . . . . . .Poverty Income ratio ........................................................................ -0.164 0.057 –2.878 o.i95Race ..:................................................................................................ 1.818 0.179 10.148 6.:34Education of head of household .................................................. -0.060 0.016 -3.091 0.’94Oral contraceptive usage in past 6 months ............................... -0.548 0,284 -1.930 0.:56Age ...................................................................................................... 0.023 0.018 2.269 1.ID2

Ever been anemic ............................................................................ 0.561 0.188 2.966 1.76Ever been pregnant ......................................................................... 0.025 0.301 0.083 1.(D3

~Only characteristics with a t-test statstic > 1.96 (p < .05) are shown.2See appendix I for discussion of odds ratios.

Table 33. Logit analysis statistics showing the relationship between selected characteristics and transferring sat;;ation status formenstruating women ages 12-54 years: United States, 1971-74

Logit ana/ysk statisticsCharacteristics

Coefficient Standard error t-test statistic 1 Odds ratio z

Constant ................... .......................................................................... 0.535 0.427 . . . . . .

Poverty Income ratio ........................................................................ -0.175 0.040 =$.325 0.04

Race ..:................................................................................................ 0.495 0.171 2.696 1.65

Education of head of household .................................................. -0.032 0.010 -3.304 0.97

Oral contraceptive usage m past 6 months ............................... -0.379 0.142 -2.663 0.68

Age ...................................................................................................... 0.003 0.007 0.424 1.00

Ever been anemic ............................................................................ 0.059 0.143 0.416 1.06

Ever been pregnant ......................................................................... 0.609 0.177 3.436 1.84

10nly characteristics with a t-test statistic > t .96 (P < .05) are shown.Z. Seeappendix I for discussion of odds ratios.

Table 34. Logit analysis statistics showing the relationship between selected characteristics and hemoglobin and transferring saturationstatus for menstruating women ages 12–54 years: United States, 1971–74

CharacteristicsLogit analysis statistics

Coefficient Standard error t-test statistic Odds ratio z

Constant ............................................................................................. 0.977 0.886Poverty income ratio ........................................................................

. . . ,,. .0.205 0.079 -2.592 0.61

Race ................................................................................................... 1.017 0.236 4.307 2.70Education of head of household .................................................. 0.073 0.020 -3.655 0.93Oral contraceptive usage in past 6 months ...... ....... ............... 0.929 0.404 -2.297 0.38Age ...................................................................................................... 0.033 0.013 2.510 1.04Ever been anemic ............................................................................ 0.595 0.280 2.129 1.81Ever been pregnant ......................................................................... 0.437 0.427 1.023 1.57

10nly characteristics with a f-test statistic > 1.96 (p < .05) are shown.2~e appendix I for discussion of odds ratios.

66

Table 35. Z-statistics for comparison of low and adequate hemoglobin statusl for persons ages 65-74 years, with statistic and standarderror of the statistic, by selected characteristics: United States, 1971-74

Low hemoglobin Adequate hemoglobin

Characteristics Number of Standard Number of Standardexamined Statistic

Z-statisticerror of the examined Statistic error of the

persons statistic persons statistic

Oietary intakeTotal iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (ma) ...........Total protein intake (mg) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) .....Estimated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ......................................................

SocioeconomicAge (years) .............................................................Poverly income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

Married2 ...................................................................

0ther2

Pica ...................................................... ...................Regular vitamin use ..............................................Regular mineral use..............................................Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special diet .............................................................Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General findings.....................................................Abdominal operations ...........................................

Age began menstruating (years) ........................

249249249249249249249249249249

249237237249249235245

249

247249249249249247

63249249247247

57

Mean

11.2099.42

3.737.47

64.8130.2134.60

1.591.54

06.47

69.292.20

3,534.002.45

67.468.588.41

Percent

65.50

0.8525.7819.1930.8145.0217.9121.9025.2568.91

2.5348.96

Mean

13.44

0.772.300.430.563.012.231.690.140.149.05

0,280.18

210.950.170.830.410.40

4.52

0.884.163.745.154.733.667.824.124.401.504.77

0.33

1,9871,9871,9871,9871,9871,9871,9871,9871,9871,987

1,9871,8981,8981,9871,9871,8401,966

1,982

1,9861,9871,9871,9871,9871,9861,1231,9871,9871,9751,986

1.081

Mean

10.2697.47

3.197.08

60.6426.8633.78

1.381.34

94.34

68.892.5e

3,915.002.02

68.549.859.90

Percent

0.170.680.110.120.990.770.590.030.032.81

0.100.07

78.540.040.220.130.13

61.70

0.4134.5717.9748.1141.7414.2947.6316.6461.80

1.7251.33

Mean

13.64

1.64

0.221.601.292.221.661.182.241.251.630.441.68

0.08

1.200.811.210.681.321.420.461.441,38

-0.83

1.34-2.01-1.69

2.49-1.26–2.92-3.53

0.79

0.49–1 .97

0.31-3.09

0.650.94

-3.162.001.510,52

-0.47

-0.59

IAge andsex-specifichemoglobinstandardsare foundin table A.ZStatistics for these characteristics represent only those persons who answered yea on the questionnaire

57

Table 36. Z-statistics for comparison of low and adequate transferring saturation statusl for persons ages 65-74 years, with statistic andstandard error of the statistic, by selected characteristics: United States, 1971-74

Low transferrirr saturation Adequate transferringsaturation

Characteristics Number of Standard Number of Standard Z-statisticexamined Statistic error of the examined Statistic error of thepersons statistic persons statistic

Dietary intakeTotal iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) ... .............Total nonanimal tissue iron intake (mg) ..... .....Total protein intake (ma) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) ....Estimated available iron (mg) ........................ ....Estimated available iron with tea (mg) .............Vitamin C (mg) ............ ....... ........ ................. . ....

SocioeconomicAge (years) .............................................................Poverly income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ................. .........

MarriedZ ...................................................................

OtherzPica .................................................. ..... ..... . .....Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .... ..................Regular aspirin use..., ...........................................Special diet .............................................................Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General findngs .....................................................Abdominal operations ................................. ........

Age began menstruating (years) ..... ....... ....... .

126126126126126126126126126126

126121121126126120124

126

12512612612612612579

126126126125

74

Mean

9.9393.94

2.996.93

59.5226.7132.61

1.271.22

72.24

68.862.13

3,429.002.44

66.949.018.91

Percent

56.40

1.4735.1020.3040.0059.1021.5055.3023.6064.10

5.4057.30

Mean

13.54

0.773.170.440.564.283.482.440.120.129.33

0.400.21

307.360.260.880.500.48

6.63

1.626.395.378.346.565.518.395.676.683.026.64

0.36

2,1102,1102,1102,1102,1102,1102,1102,1102,1102,110

2,1102,0142,0142,1102,1101,9952,087

2,106

2,1002.1102,1102,1102,1102,1081,1072,1102,1102,0962,108

1.064

10.3697.83

3.277.09

61.2727.3333.93

1.401.36

93.02

66.902.52

3,906.002.03

68.479.739.75

Percent

62.60

0.2434.2018.3046.9042.1014.1044.6017.2063.00

1.6050.90

13.64

0,170.670.110.130.960.750.570.030,032.72

0.090.06

76.310.030.220.130.13

1.59

0.161.551.262.131.611.142.241.231.500.411.63

0.09

-0.55-1.20-0.62-0,27-0.40-0.17-0.45

0.55-1.24-2.14

-0.10-1.76-1.51

1.54-2.19-1.42-1.69

-0.91

0.760.140.36

-0.802.511.311.231.100.161.250.94

-0.27

1Personswitha Iransfeninsaturationbelow 16.0 percentandTIBC valueabove250 Wg/dlwareclassifiedse havinga lowtransferringsaturation.zStatisticsforthese characteristicsrepresentonly those pereonswhoansweredyesonthe questionnaire.

Table 37. Z-statistics for comparison of low and adequate hemoglobin statusl for white persons ages 65-74 years, with statistic andstandard error of the statistic, by selected characteristics: United States, 1971-74

Low hemoglobin Adequate hemoglobin

Characteristics Number of Standard Number of Standardexamined Statistic

Z-sfatisticerror of the examined Statistic error of the

Persons statistic oersons statistic

Dietary intakeTotal iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (ma) ...........Total protein intake (mg) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) .....Estimated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ......................................................

SocioeconomicAge (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

Marriedz ...................................................................

OlherzPica ..........................................................................Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special diet .............................................................Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General findings.....................................................Abdominal operations ...........................................

Aae began menstruating (years) ........................

145145145145145145145145145145

145140140145145136143

145

14414514514514514435

145145143144

34

Mean Mean

10.6199.45

3.487.13

63.9229.1134.02

1.401.43

89.20

69.332.59

3,780.002.30

67.1010.00

9.77

Percent

73.00

1.0330.0821.0033.0945.4520.9522.1728.8369.17

2.3052.10

Mean

13.48

0.672.720.400.463.552.5B2.060.120.129.74

0.360.25

285.750.211.200.450.44

5.48

1.275.715.076.836.205.09

10.625.615.751.B86.25

0.44

1,6831,6831,6831,6831,6831,6831,6831,6831,6831,683

1,6831,6111,6111,6831,6831,5651,670

1,680

1,6821,6831,6831,6831,6831,682

9411,6831,6831,6731,682

919

10.3898.06

3.227.16

61.0926.8734.22

1.391.35

94.61

68.912.67

3,966.001.99

68.5710.1110.12

Percent

62.00

0.3835.3818.5948.6341.2714.3047.4316.6961.25

1.6952.59

Mean

13.67

0.180.730.110.141.070.830.640.030.032.98

0.110.07

85.620.040.230.140.14

1.77

0.231.751.422.411.801.282.441.361.760.471.83

0.09

0.330.490.63

-0.060.760,830.280.710.64

-0.53

1.12-0.31-0.62

1.87-1.20-0.23-0.76

2.05

0.51-0.69

0.46-2.04

0.651.25

-2.312.101.320.31

-0.08

-0.43

1Age- andsex-specifichemoglobmstandardsare foundintableA.2Statisticsfor these characteristics represent only those persons who answered yes on the questionnaire.

69

Table 3EI. Z-statistics for comparison of low and adequate hemoglobin status’ for black persons ages 65-74 years, with statistic andstandard error of the statistic, by selected characteristics: United States, 1971 –74

Low hemoglobin Adequate hemoglobin

Characteristics Number of Standard Number of Standardexamined Statistic

Z-statisticerror of the examined Statistic error of the

persons sfa tistic persons sta fistic

Dietary intake

Total iron intake (mg) ................ ........ ...... .......Log total iron intake (log mg) . .................. ......Total animal tissue iron intake (mg) ....... ........Total nonanimal tissue iron intake (mg) ...........Total protein, intake (mg) ................................... .Total animal tissue protein intake (mg) .. ..... ...Total nonanimal tissue protein intake (mg) .,.,.Estimated available iron (mg) .............................Estimated available iron with tea (mg) ......... ...Vitamin C (mg) ................... .............................. ...

Socioeconomic

Age (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

MarriedZ ...................................................................

Otherz

Pica ................................................................... ......Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special diet .............................................................Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General findings .....................................................Abdominal operations ............................... ........ ..

Age began menstruating (years) ........................

104104104104104104104104104104

1049797

104104

99102

104

103104104104104103

28104104104103

23

Mean Mean

12.6699.33

4.35B.31

67.0032.9434.06

1.851.83

79.72

69.201.16

2,074.002.64

68.355.235.00

Percent

45.20

0.3715.1714.7123.2143.9610,1319.9017.6768.24

3.1040.92

1.804.291.001.375.624.263.010.340.34

16.78

0.470.12

265.620.300.870.590.58

7.33

0.095.285.217.607.304.46

11.325.616.852.557.27

304304304304304304304304304304

304287267304304275296

302

304304304304304304162304304302304

0.9090.45

2.766.13

55.3426.7428.61

1.201.17

91.06

68.571.54

3,302.002.32

68.196.867.15

Percent

57.20

0.7424.9110.6541.9547.4013.2349.9916.1169.34

2.1036.26

Mean

●13.34 * 162 13.35

0.381.910.260.272.652.341.300.070.079.01

0.230.10

182.220.120.600.360,35

4.27

0.743.722.655.864.302.925.563.163.971.244.14

0.19

2.041.891.541.561.881.271.661.891.92

-0.54

1.21-2.49-1.33

0.990.15

-2.36--3.17

--1.42

-0.32--1.51

0.69--1.95--0.41-4.58--2.39

0.24-0.14

0.350.56

0.02

j Age- and sex-specific hemoglobin standards are found in table A.2Statistics for these characterfstice represent only those persons who answered yes on the questionnaire.

70

Table 39. Z-statistics for comparison of low and adequate transferring saturation status! for white persons ages 65-74 years, with statisticand standard error of the statistic, by selected characteristics: United States, 1971-74

Low hemoglobin Adequate hemoglobin

Characteristics Number of Standard Number of Standard Z-statisticexamined Sfatistic error of the examined Sfatistic error of fhepersons stafisfic persons sfa tisfic

Dietary intakeTotal iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (mg) ...........Total protein intake (mg) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) .....Estimated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ......................................................

SocioeconomicAge (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) ....................Education of head of household (years) ..........Education of examinee (years) ...........................

Marriedz ...................................................................

OtherzPica ..........................................................................Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special det .............................................................Ever been pregnant ..............................................Ever been anemic ........... ......................................Any trouble eating .................................................General findings.....................................................Abdominal operations ...........................................

101101101101101101101101101101

1019797

10110197

101

101

10110110110110110163

101101101101

61

Mean

9.9394.22

2.717.23

58.2424.5533.69

1.231.19

76.70

68.792.19

3,408.002.36

66.999 +79.11

Perce,rW

60.90

1.5030.4022.2037.8060.1023.6058.1025.3063.60

2.8056.20

Mean

Age began menstruating (years) ........................ 13.53

0.853.470.390.674.383.282.800.120.12

10.78

0.430.24

334.460.281.260.54Q-.52

7.28

1.797.256.210.767.316.349.326.497.182.467.40

0.40

1,7271,7271,7271,7271,7271,7271,7271,7271,7271,727

1,7271,6541,6541,7271,7271,6041,712

1,725

1,7251,7271,7271,7271.7271,725

9131,7271,7271,7151,725

892

Mean

10.4498.38

3.297.15

61.6627.3234.34

1.411.37

93.60

60.922.63

3,979,002.01

68.4910.0610.06

Percent

63.30

0.2335.2018.9048.1041.6014.1045.0017.3062.30

1.6052.50

Mean

13.67

0.180.720.110.131.060,810.640.030.032.92

0.100.07

84.680.040.240.140.13

1.74

0.171.731.412.361.781.262.471.361.750.461.81

0.09

-0.59-1.17-1.41

0.12-0.76-0.82-0.23-1.40-1.42-1.51

-0.30-1.76-1.66

1.24-1.17-1.59-1.77

-0.32

0.710.430.52

-1.142,461.471.361.210.180.480.49

-0.34

I Personswitha transferringsaturationbelow16.0 percentand TIBC value above 250 pg/dl were classified as having a low tranferrin saturation.ZStatistics for these characteristics represent only those persons who answered yes on the questionnaire.

71

—

Table 40. Z-statistics for comparison of low and adequate transferring saturation statusi for black persons ages 65-74 years, with statistic,and standard error of the statistic, by selected characteristics United States, 1971-74

Low transferringsaturation Adequate transferringsaturation

Characteristics Number of Standard Number of Sfandardexamined Statistic

Z-stiatisticerror of the examined Statistic error of the

persons statistic persons sta fistic

Dietary intakeTotal iron intake (mg) ..........................................Log total iron intake (log mg) ............................Total animal tissue iron intake (mg) .................Total nonanimal tissue iron intake (mg) ...........Total protein intake (mg) .....................................Total animal tissue protein intake (mg) ............Total nonanimal tissue protein intake (mg) ....Est~mated available iron (mg) .............................Estimated available iron with tea (mg) .............Vitamin C (mg) ......................................................

Socioeconomic

Age (years) .............................................................Poverty income ratio (index) ...............................Estimated family income (dollars) ......................Number of persons in household (units) ..........Age of head of household (years) .... .......... ....Education of head of household (years) ..........Education of examinee (years) ...........................

MarriedZ ...................................................................

OtherzPica .......... ....... ....... ......... ... ...... .................Regular vitamin use ..............................................Regular mineral use ..............................................Regular vitamin and mineral use .......................Regular aspirin use ...............................................Special diet .............................................................Ever been pregnant ..............................................Ever been anemic .................................................Any trouble eating .................................................General findings .....................................................Abdommal operations ...........................................

Age began menstruating (years),, ......................

25252525252525252525

25242425252323

25

2425252525241625252524

13

Mean

9.8891.91

5.094.79

68.8542.4226,43

1.531.40

39.57

69.41‘1 .57

‘3,607.003.08

66.62●7.79“6.9B

Percent

1.878.121.740.62

13.8012.743.580.410.35

10.50

1.11●

*0.732.74

*●

23.30

‘ 1.6010.60

6.7056.6052.30‘3.70

‘34.7010.7068.2024.40

●66.70

Mean

12.69

●

9.247.53

24.9315.00

**

9.2713.9812.87

●

‘13.66 ●

383383383383383383383303383383

383360360383383351375

381

383383383383303383194383383381303

172

Mean

9.5792.06

3.076.50

57.1627.4729.69

1.311.30

87.20

68.631.40

3,128.002.33

68.286.356.52

Percent

55.30

0.0423.7012.1034.6047.2013.2041.0015.8069.50

0.8034.50

Mean

13.29

0.501.840.330.362.411.991.240.100.108.17

0.210.08

156.060.110.500.330.32

3.82

0.493.262.504.733,832.605.302.803.530.683.64

0.18

0.16-0.02

1.14-2.38

0.831,16

-0.860.520.28

-3.58

0.690.460.571.01

-0.601.130,38

-2,,41

0,,41-1 ,“34-0.68

0,,870,33

-1,50-0,344,53-0.09

1.832.16

0.51

1Personswitha transferrmsaturationbelow16.0 percentand a TIBC above250 pg/dl were classifiedas havinglowtransfarrinsaturation.ZStatlshcsforthese characteristicsrepresentonly those persons who answered yes on the questionnaire.

72

Table 41. Logit analysis statistics showing the relationship between selected characteristics and hemoglobin status for persons ages 65-74years United States, 1971-74

CharacteristicsLogit analysis ststisfics

Coehkient Standard error t-test statistic 1 Odds ratio 2

Constant ............................................................................................. -0.159 0.778 . . .Race ..................................................................................................-

. . .1.903 0.301 6.317 6.33

Sex ... . ., .... ............................................................................................ -1.939 0.236 -6.220 0.14Education of head of household .................................................. -0.042 0.018 -2.350 0.96General findngs ................................................................................ 1.269 0.645 1.966 3.53Ever bean anmic ...........................................................................- 0.901 0.205 4.395 2.64Any trouble eating ............................................................................ 0.464 0.161 3.011 1.58

10nly characfenaticawitha t-teststalistic> 1.96 (p < .05) are shown,2See appandix I for dkussion of odds ralim.

Table 42. Logit analysis statistics showing the relationship between selected characteristics and transfernn saturation status for personsages 65-74 years: United States, 1971-74

CheractarMksLogit analysis statistic

CoefY7cient Standard error t-test stafisiic 1 Odds ratioz

~nstant ............................................................................................. -0.637 0.993 . . . . . .Race ................................................................................................... 0.364 0.311 1.160 1.44sex. ..... ................................................................................................ 0.430 0.300 1.461 1.57Education. o! head of household .................................................. -0.057 0.020 -2.894 0.94General fltilngs ................................................................................ 1.406 0.065 1.625 3.84Ever been anemic ............................................................................ 0.10.9 0.255 0.425 1.13Any trouble eating ............................................................................ -0.083 0.194 -0.429 0.92

10nly characlaristics with a t-test statistic z 1.96 (p < .05) are shown.2%ea appendixI fw dimmsion of 03ds ratiis.

73

Appendixes

Contents

1. Statistical notes .,... ........................................................................................................... ................................ .......... .......... ...Survey design ............................................................................................................................................... .......... .......... ...Nonresponse. ..................................................................... ......................................................................... ............. ..........Missing data ................................... ................ ............................................................................................................ .........

Small numbers .....................................................................................................................................................................Reliability of estimates of means and prevalence ..........................................................................................................Analytic approach for relational analyses . .. ....................................................................................................................

Il. Definitions of terms ........................................ .................................................................... .....................................................

Demographic and socioeconomic terms ................................. ..........................................................................................Age ...................................................................................................................................................................................Race ...................................................................................................................... ..........................................................Family income .................................................................................................................................................................

Poverty index ................................................................................................................................................................ ..

Other terms used in relational analyses ............................................................................................................................

7575

76

76777779

81

818181

81

8182

List of appendix tables

1. Percent distribution of nonresponse adjustment factors: 1971-74 ..................................................................................... 7711, Number and percent of examined persons in NHANES I with missing hematological, biochemical, and nutrient

intake measurements, by age ................................................................................................................................................. 77Ill. Estimated design effects and correlations of selected health, demographic, and nutrient intake variables: NHANES I 78IV. Number of persons classified as ‘(low” and “adequate” iron status in the original and logistic samples, by iron status

indicator and selected population group: NHANES 1,1971 -74.... ........................................................................................ 80V. Weighted average thresholds at the low-income level in 1971, by farm-nonfarm residence, sex of family head, and

size of the family United States, .................................................................. ........................................................................... 82

Appendix L Statisticalnotes

Survey design

Individuals examined during NHANES I wereselected in a three-stage, stratified probability sampleof loose clusters of persons by geographic location. Thesample was designed to be representative of the civiliannoninstitutionalized population ages 1-74 years livingwithin the coterminous United States; however, allpersons residing upon reservation lands set aside foruse of American Indians were excluded.

In the first stage of the design, the 1960 decennialcensus lists of addresses and the nearly 1,900 primarysampling units (PSU’S) into which the coterminousUnited States is divided were examined. (Each PSU iseither a standard metropolitan statistical area, a singlecounty or two or three contiguous counties.) ThesePSU’S were grouped into 40 strata on the basis ofgeographic region and population density to selecttarget PSU’S for NHANES 1.

Of the 40 strata, 15 were composed of single largemetropolitan areas with more than 2 million persons.These 15 metropolitan areas were chosen for thesample with certainty. A modified Goodman-Kishcontrolled selection technique was then used to choosetwo PSU’S from each of the remaining 25 noncertaintystrata with probability proportionate to its 1960population. In this manner, a total first-stage sample of65 (15 + (2 X 25)) PSU”S or “stands” were selectedfor study.

Within each PSU, a systematic sample of segments(loose clusters of households) was chosen. Selectionwas made using the most up-to-date information oncensus enumeration districts (ED’s) at the time of thevisit—1960 census data for the first 44 stands and 1970data for the remaining 21 stands. To make the samplerepresentative of the current U.S. population, listswere supplemented by a sample of housing units thathad been constructed since the most recent decennialcensus,

ED’s having addresses that could be used (themajority of the areas visited) were divided intosegments containing an average of six households each

(in the first 44 stands) or eight households each (in theremaining 21 stands). The change was made primarilyfor operational advantages and was supported byresearch by the U.S. Bureau of the Census, indicatingthat the precision of estimates would not be tiectedappreciably. For ED’s without addresses that could beused (generally located in rural areas), area samplingwas employed.

Enumeration districts were divided into two eco-nomic classes. The first class, identified as the “povertystratum,” was composed of “Current Poverty Areas”that had been identified by the Census Bureau in the1960 or 1970 census and other ED’s in the PSU with amean family income of less than $3,000 in 1959 (basedon the 1960 census). The second economic class, the“nonpoverty stratum,” comprised all ED’s not desig-nated as belonging to the poverty stratum.

Target segments were selected from each of thetwo strata, All sample segments classified into thepoverty stratum were sampled with probability 1. Forthe first 42 stands, sample segments in nonpovertystratum ED’s were divided into eight random sub-groups, and one of the subgroups was chosen toremain in the NHANES I sample. Ongoing researchthen indicated that the efficiency of estimates could beincreased by changing the ratio of poverty to nonpov-erty segments from 8:1 to 2:1. Therefore, in theremaining 23 stands, the selected segments in thenonpoverty ED’s were divided into two randomsubgroups, and one of the subgroups was chosen toremain in the sample.

After identifying the sample segments, a list of allcurrent addresses within the segment boundaries wasmade, and a person in the household was interviewedto determine the age and sex of each person as well asdemographic and socioeconomic information requiredfor the survey. If no one was at home after repeatedcalls or if the household members refused to beinterviewed, the interviewer tried to determine thehousehold composition from neighbors.

To select the persons in sample segments to beexamined in NHANES I and at the same time to

75

oversimple certain groups at high risk of malnutrition,all household members ages 1–74 in each segment firstwere listed on a sample selection worksheet with eachhousehold in the segment listed serially. The numberof household members in each of the six age-sexgroups shown below were then listed on the worksheetunder the appropriate age-sex group column. Thesample selection worksheets were put in segment-number order, and a systematic random sample ofpersons in each age-sex group was selected to beexamined using the following sampling rates.

Age and Sex

1-5 years (males and females) ...................................6-19 years (males and females) ................................2044 years (males) ......................................................2044 years (females) ...................................................46-64 years (males and females) ..............................65-74 years (males and females) ..............................

Rate

1121/41/41/21/4

1

The persons selected in the 65-stand sample ofNHANES I constituted a representative sample of thetarget population and included 28,043 persons ages1–74 years, of whom 20,749 (74 percent) were exam-ined. When adjustments were made for differentialsampling for high-risk groups, the response rate was 75percent.

All data presented in this report are based on“weighted” observations. That is, data recorded foreach person are inflated to characterize the subuni-verse from which that person was drawn. The weightfor each examined person is a product of the reciprocalof the probability of selecting the person, an adjust-ment for nonresponse cases (i.e., persons not exam-ined), and a poststratified ratio adjustment that in-creases precision by bringing survey results into closealignment with U.S. Bureau of the Census populationfigures for 20 age, race, and sex groups in the UnitedStates as of November 1, 1972, the approximatemidpoint of NHANES 1.

A more detailed description of the survey designand selection technique has been published-z

Nonresponse

In any survey, after the sample is identified and thepersons are requested to participate, the survey meetsone of its more severe problems—nonresponse. Theproblem is more severe in a health examination survey,because often many persons will not participate in theexamination. A potential for bias results if the personsin the sample who do not participate differ from thepersons in the sample examined with respect to thecharacteristics under investigation. Intensive effortswere made in NHANES I to develop and implementprocedures and inducements that would reduce the

NOTE: A list of references follows the text.

76

number of nonrespondents and thereby reduce thepotential bias due to nonresponse.z

Despite these intensive efforts, 25 percent of thepersons in the sample were not examined, comparedwith previous surveys conducted by the Divisicln ofHealth Examination Statistics that had response ratesof more than 86 percent. Consequently, the potentialfor a sizable bias does exist in the estimates in thispublication. Because the response rate for the meldicalquestionnaire was more than 95 percent of the personsin the sample, however, it was possible to examine thecharacteristics of the nonrespondents and the nature ofnonresponse.

An analysis of data on examined and nonexamlined(but interviewed) persons was conducted using datafrom the first 35 stands of NHANES 1.75The twogroups were quite similar with respect to the healthcharacteristics that were compared. For example, 11percent of persons examined reported having an illhm.ssor condition that interfered with their eating, com-pared with 9 percent of persons not examined but whohad a medical history.

As was mentioned earlier, the data in this reportwere based on weighted observations, and one of thecomponents of the weight assigned to an examinedperson was an adjustment for nonresponse. A proce-dure was adopted that mutiplied the reciprocal of theprobabilityy of selection of sample persons who ‘wereexamined by a factor to bring estimates basecl onexamined persons up to a level that would have beenachieved if all sample persons had been examined. Thenonresponse adjustment factor was calculated bydividing the sum of the reciprocals of the probability ofselection for all selected sample persons in each of fiveincome groups within each stand by the sum of thereciprocals of the probability of selection for examinedsample persons in the same stand and income group.The five income groups were (1) under $3,000, (2)$3,000-6,999, (3) $7,000-9,999, (4) $10,OOCL14,999,and (5) $15,000 and over. For sample weightingpurposes, income group was imputed for 5.6 percemtofthe sample persons using educational level of the headof household. To the extent that the income-within-stand classes were homogeneous with respect to thehealth characteristics under study, the adjustmentprocedure was effective in reducing the bias due tononresponse. The percent distribution of the nonre-sponse adjustment factors computed for the 65-standsample of NHANES I is shown in table I.

Missing data

Examination surveys are subject to the loss ofinformation not only through the failure to examine allpersons in the sample, but also from the failure to.obtain and record all items of information for exam-ined persons. For several examinees, one or more ofthe hematological and biochemical measurements

and/or one or more of the dietary interviews were notavailable. The extent of these missing measurements isindicated in table II.

The number of missing hemoglobin concentrationswas small for all age groups. Total iron-bindingcapacity and percent transferringsaturation were avail-able for most persons ages *74 years; the number ofmissing measurements for these two variables forchildren under 4 years of age, however, was large.(Most missing biochemical determinations were aresult of refusal to give blood (especially youngchildren), loss of blood specimens in shipping to thelaboratory, and results that did not meet laboratoryquality control specifications.) Nutrient intake valueswere missing for 479 persons due to unsatisfactorydietary interview results.

For the descriptive information on persons ages4-74 years, presented in tables 1–18 and figures 1–13,imputed hematological, biochemical, and nutrientintake measurements were used. These estimates weremade on the basis of multiple regression type deci-sions, substituting for the missing measurements thoseof an individual who was of the same age, sex, race,and geographic location. (For approximately one-fourth of the missing records, a further adjustment of

Table 1. Percent distribution of nonresponse adjustmentfactors 1971-74

Size of factor Percent distribution

Total ......................................................... 100.0

1.00-1 .24 ................................................. 32.61.25-1 .49 ................................................. 38.5l.5&l.74 ................................................. 18.21.75-1 .99 ................................................. 7.42.0&2.49 ................................................. 2.82.5%2 .99 ................................................. 0.33.007......................................................... 0.3

1A size of 3.00was assignedfor all factors greater than 3.00.The finalposlstratitiedratioadjustmentcorrectsforthistruncation.

Table Il. Number and percent of examined persons in NHANES Iwith missing hematological, biochemical, and nutrient intakemeasurements, bv age

Measurement and age Number Percent

Hemoglobin1-3 years ................................................ 18 1.04-74 years .............................................. 2,019 10.6

Total iron-binding capacity1+ years ................................................ 704 45.54-74 years .............................................. 5,431 28.5

Percent transferring saturation

1-3 years ................................................ 704 45.54-74 yeam...l .......................................... 5,786 30.4

Nutrient intake1-74 years .............................................. 479 2.3

the nutrient intake values was made on the basis of theperson’s body weight.) For children under 4 years ofage, the number of missing hematological values wastoo great to use such an imputation procedure.Therefore, data on persons ages 1–3 years with missinghematological and biochemical measurements areomitted from all detailed tables.

For the relational analyses presented in tables1942, no imputed hematological values were used. Inthis manner, no individual was assigned a particulariron status without full and complete knowledge ofthat individual’s measured status at the time ofexamination. Imputed nutrient intake measurementswere retained for this sample.

Small numbers

In some tables, estimates are shown for cells forwhich the sample size is so small that the estimatesshown, including those of variability, may not bereliable. In such instances, the numbers, if shown, havebeen included only to convey an impression of theoverall table.

Significance tests of mean differences were con-ducted only if there were at least 25 persons in eachgroup. Logit analyses were conducted only if therewere at least 50 individuals in each group.

Reliability of estimates of means andprevalence

Because the statistics presented in this report arebased on a sample, they differ somewhat from thefigures that would have been obtained if the survey hadbeen conducted on the complete population usingprecisely the same procedures. In other words, thestatistics are subject to sampling variability.

The standard error is primarily a measure ofsampling variability, but also may include part of thevariation that arises in the measurement process. Thestandard errors used to calculate the significance ofdifferences shown in tables 1–10 were calculated by atechnique referred to as “balanced repeated replica-tion.” The need for this specialized technique forestimating standard errors arose because of the com-plexity of the sample design of NHANES I. Thiscomplexity made it inappropriate to calculate standarderrors by a technique that did not account for th’emultistage cluster sample design. (Estimates of stand-ard errors are subject to errors that may be large ifthe number of cases upon which the estimates arebased is small, or the number of PSU’S used in thevariance calculations is small.)

However, to reduce the complexity of the effortrequired to produce standard error estimates for all ofthe means and proportions in this report, a “variancesmoothing” approach was used to calculate all otherstandard errors.

77

Under this approach, a standard error estimate fora sample mean (Xi) is produced in two steps. First, thesimple random sample (SRS) estimate of the standarderror (se.) is calculated with the usual formula,

s.e.~.~(~) = Sx,/(72i)z

where SX i is the sample standard deviation and ni isthe size of the sample. Second, the simple randomsample estimate of the standard error is multiplied by adesign effect (defined as the effect that the complexsampling design has on the magnitude of the standarderrors) to produce the standard error estimate of X - i.

All standard errors of means presented in tables11J12 were computed using this technique. The designeffects were estimated by the method of least squares.The replicated half-sample variance estimates for asubset of the variables presented in this report wereused as the dependent variables, and the correspondingsimple random sample variance estimates were used asthe independent variables in the model

se. (~) = (design effect) X s.e.~R~ (~) + e

to produce estimates of the design effects.The estimated design effects and the correlations

between the half-sample variance estimates and the“smoothed” or predicted variance estimates (giving ameasure of the strength of the relationship between thetwo, that is, the effectiveness of the fitting process) aregiven in table 111, For all eight variables examined, thedesign effect never exceeded 1.50. In this report, thegenerally conservative strategy of assuming a designeffect of 1.50 for all variables was used.

The usual test for the difference between twomeans (denoted by 7A and xx ), in which the estimat-ed standard errors are treated as constants and thecovariance is assumed to be zero, may then be used todetermine the significance of a difference as follows:

where se. (7A ) and se. ( ~X ) are the standard errorsfor the two groups reported in the detailed table.

Table Ill. Estimated design effects and correlations of selectedhealth, demographic, and nutrient intake variables: NHANES I

Variable Desicrn effect Correlation

Have you ever had anemia? ........... .Are you or have you ever been

pregnant? .............................................Do you have trouble eating? .............Have you used oral contraceptive

in past 6 months? .............................Age at exam .........................................Poverty income ratio ............................Iron intake ..............................................Protein intake ........................................

1.09

1.100.98

1.490.881.141.251.32

0.01

o.a40.76

0.97o.al0.71o.alo.a5

For example, from table 19, the mean a,ge ofchildren ages 1–3 years with low hemoglobin is 2.19,and the standard error is 0.09; the correspondingstatistics for the adequate hemoglobin group are 2.58and 0.04, respectively. The significance test is, there-fore:

z=(2.19 - 2.58)

[(0.09)2 + (0.04)2]%

~Z.3,96~

This Z-statistic, if looked up in a table, is found tc~havea probability of less than 0.001.

Standard error estimates for sample proportionswere similarly calculated in two steps. First, the simplerandom sample estimate of the standard error wascalculated with the usual formula,

. ,, “ = Q’ij(l - Pij)/ni)z“-2-ZRS:, “ij~

where pi j @ the estimated proportion, and ni is the sizeof the sample. Second, the simple random sampleestimate of the standard error was multiplied by thedesign effect to produce the standard error estimate of

Pij.The following example may be illustrative. From

table 11, 14,9 percer~t o; males ages 65–74 years areestimated to have low hemoglobin; for females, thepercentage is 4.3. The proportions denoted by Prn andpf are, therefore, 0.149 and 0.043. The standard errorsunder simple random sampling were estimatedthrough the following calculation:

s.e.~R~(pm) = (0.149 X 0.851/1581)% = .0089555

s.e.~~~(p~) = (0.043 X 0.957/1728)% = .0048799

The standard errors with design effects are, therefore:

s.e.(p~) = 1.5 X .0089555 = .0134332

s.e.@~) = 1.5 x .0048799 = .0073199

All standard errors of proportions presented in tables11–1 3, 15, and 19-42 were computed using thistechnique.

The usual test for the difference between twoproportions may then be used to determine sig-nificance:

Pm - Pfz=

[(s.e.(pm))2+(s.e.(pf))21%

dThez.~tatistiapresentdin the detailed tables were calculat~ Withstandard error values calculated to three or four decimal places. Thestandard errors presented in the tables are rounded to two decimals.Therefore, Z-statistics calculated using these values may differ slightly fmmthe values presented in the detailed tables.

NOTE A list of references follows the text.

78

For the example described in the previous paragraph,

z.0.149-0.043

[(.0134332)2 + (.0073199)2]%

.106z..0152981

Z = 6.93

This Z-statistic, if looked up in a table, is found tohave a probability of less than 0.001.

Analytic approach for relationalanalyses

Logistic models were used to study the relation-ship between iron status (as measured by a dichoto-mous variable) and the dietary, health, socioeconomicand demographic factors under study. These modelswere developed to estimate the probability (or odds)that an individual will have, for example, low transfer-ringsaturation (TS), given a set of independent vari-ables. Akin to multiple regression, logistic analysisassumes that although only a discrete outcome can beobserved, an increase (or decrease) in each relevantindependent variable increases (or decreases) the likeli-hood that an individual has a low TS level. 76,77Theparticular advantage of logistic models is that theyovercome the well-known problems of inefficiency andpossible misspecification associated with alternativeclassification schemes, such as discriminantanalysis,7%E1

Under a logistic model, the natural logarithm ofthe odds that an individual is iron deficient is ex-pressed as a linear combination of independent vari-ables. Mathematically, this relationship is expressedas:

where: p = the probability of having a lowtransfernn saturation level.

x, . . . x. = the independent variables understudy.

b,.. . b~ = the logistic coefficients associatedwith these independent variables.

e = random error.

Estimated logistic coefficients (the b’s) represent thechange in the natural logarithm of the odds of aperson’s having a low TS that would be produced by a

NOTE A list of references follows the text.

unit change in the particular independent variable. Asthe equation above cannot be solved explicitly, aniterative procedure is used to solve for the b’s. In thepresent study, a Fletcher minimization procedure wasused.’2

Although such an iterative solution is satisfactoryfor estimating the logistic coefficients, the complexityof the NHANES I sample design necessitated use of aninnovative approach to the estimation of the standarderrors of these coefficients. Jackknifing, an all-purposetechnique for determining the variance of any statisti-cal parameter, is one approach. In essence, thetechnique involves estimating the parameter of interest(here, a logistic coefficient) on the full sample, thensystematically setting aside portions of the sample andestimating the logistic coefficient on each of the so-generated subsamples and finally, computing thestandard error of the logistic coefficient by studyingthe variability of the deviations of the statistics in thesubsamples from the full sample estimate.ss,sqAl-though it would be possible to employ this techniqueto estimate the requisite standard errors, such anapproach would be prohibitively expensive, as therewere over 5,000 subjects in some analytic groups.

Incorporating a second technique-sampling bythe dependent variable-into this strategy greatlyreduces the cost burden involved. This technique wasdeveloped for estimating logit coefficients on large datasets in which the prevalence of the particular healthcondition under study is extremely 1ow.8S-B7Therationale is based upon the fact that the relativeamount of information contained in two samples ofsizes nl and nz, where we wish to compare the means inthe two groups, is a function of the harmonic mean ofthe sample sizes [2n1nz/(nl + n2)]. Thus if one group isby nature very small (here, for example, the number ofpersons classified as having low TS), very little isgained by retaining all the individuals in the secondgroup to estimate their mean value (here, the adequateTS group) because it does little to increase theharmonic mean of the sample sizes. Hence a samplecan be drawn from the overall data set, taking all thoseindividuals with low TS and only a small fraction ofthose who have adequate TS levels with small (if any)loss of information. Logit coefficients estimated on asubsample so drawn will be unbiased estimates of thetrue logit coefficients.ss The jackknife technique maythen be used on this substantially reduced subsampleto produce an estimate of the standard error.

This strategy was used for all log-it analysespresented in this report. SubSamples were selectedseparately for persons with low and adequate ironstatus. All persons with low values for the iron statusindicators were chosen with certainty. The sample ofpersons with adequate levels was chosen so that therate of increase in the effective n, which determines thepower to detect effects, was small and constant acrossgroups (an additional adequate iron status individual

79

Table IV. Number of persons classified as “low” and “adequate” iron status in the original and logistic samples, by iron status indicatorand selected population group: NHANES 1, 1971-74

Iron status indicator and population groupOriginal sample Logit sample

Total Low Adequate Total Low Adequ~

Hemoglobin Number

Children, 1+3 years ............................... ........................................... 1,353 170 1,183 760 170 59CIAdult women, 12–54 years ............................................................. 3,707 242 3,465 1,083 240 841Elderly, 65-74 years ................. ...... ................ .......... ...... ....... .. 1,967 222 1,745 993 222 770

Transferring saturation

Adult women, 12–54 yearsl ............... ....... ..... ................ .... ... 3,707 632 3,075 1,264 632 632!Elderly, 65–74 years ................................................................ ........ 1,967 114 1,853 510 114 396

Hemoglobin and transferring saturation

Adult women, 12-54 years ..................................... ........................ 3,707 134 3,573 600 134 466

I For adul[ women, the number with low TS levels was sufficlently large (632 persons) so that only an equal number of women with adequate TS were chosen.

in the sample would have increased the effective n byonly one-tenth). This has the effect of keeping theprevalence constant across groups at 288 per 1,000persons. Table IV presents the sizes of the sample sodrawn.

The odds ratios presented in the logit analysistables are “overall” odds ratios. For the dichotomouscharacteristics (e.g., by sex, ever been anemic), theodds ratio presented in the table represents the

increase or decrease in the likelihood that an individu-al has low iron status for the given indicator with aunit change in the value of the characteristic understudy. A unit change for continuous variables may notbe the most appropriate comparison to make for anyanalysis of the effects on iron status for that character-istic. The overall odds ratios are presented in the tablesfor these variables, but must be interpreted appropri-ately.

Appendix Il. Definitions ofterms

Demographic and socioeconomicterms

The demographic and socioeconomic characteris-tics of the population sampled are defined below:

Age

The age recorded for each examinee was the age atlast birthday on the date of examination. The agecriterion for inclusion in the sample used in this surveywas defined in terms of the examinee’s age at time ofcensus interview. Twenty of those who were 74 yearsold at the time of interview became 75 years old by thetime of the examination. In the adjustment andweighting procedures used to produce national esti-mates, these persons were included in the 74 year oldgroup.

For analyses on children ages 1–3 years, anadditional age variable was computed. This variablerepresents the child’s age, in decimal form, computedfrom her or his date of birth.

Race

For each individual, race was determined byobservation and recorded as “white,” “black,” or“other races.” The last category included AmericanIndians, Chinese, Japanese, and all races other thanwhite or black. Mexican persons were incuded withwhite unless definitely known to be knerican Indianor of another race other than white.

Family income

The income recorded was the total income report-ed during the past 12 months by the head of thehousehold and all other household members related tothe head by blood, marriage, or adoption. This incomewas the total cash income (excluding pay in kind, e.g.,meals, living quarters, or supplies provided in place of

cash wages) except in the case of a family with its ownfarm or business, in which case net income wasrecorded. Also included in the family income figurewere allotments and other money received by thefamily from a member of the Armed Forces whethershe or he was living at home or not.

Poverty index

Income status was determined by the povertyincome ratio (PIR). Poverty statistics published in theCensus Bureau reportsss were based on the povertyindex developed by the Social Security Administration(SSA) in 1965. (For a detailed discussion of the SSApoverty standards, see references 89 and 90.)Modifications in the definition of poverty were adoptedin 1969.91 The standard data series on poverty forstatistical use by all executive departments and estab-lishments was used to determine the poverty level.9z

The two components of the PIR are the totalincome of the household (numerator) and a multiple ofthe total income necessa~ to maintain a family withgiven characteristics on a nutritionally adequate foodplan~ (denominator). The dollar value of the denomi-nator of the PIR is constructed from a food plan(economy plan) necessary to maintain minimum rec-ommended daily nut ritional requirements. The econ-omy plan is designated by the Department of Agricul-ture for “emergency or temporary use when funds arelow.”

As shown in table V, the annual income consid-ered to be the poverty level increases as the family sizeincreases. A family with any combination of character-istics and with the same income as that shown in thetable has been designated as having a PIR (or povertylevel) of 1.0. The same family with twice the incomefound in the table would have a PIR of 2.0. Ratios :Ifless than 1.0 can be described as “below poven

NOTE: A list of references follows the text.

8i

.

Table V. Weighted average thresholds at the low-income level in 1971, by farm-nonfarm residence, sex of family head, and size off familyUnited Statas

Size of familyNonfarm

TotalFarm

Total Male headl Female headl Total Male head~ Female headl

All unrelated indviduals, ...................................... $2,033 $2,040 S2,136 $1,978 $1,727 $1,703Under 65 years ................................................. 2,093

$1,6692,098 2,101 2,017 1,005 1,853

65 years and over...,, ....................................... 1,9311,715

1,940 1,959 1,934 1,652 1,666 1,643

All families .............................................................. 3,700 3,724 3,764 3,428 3,235 3,242 3,0792 persons ........................................................... 2,612 2,633 2,641 2,561 2,219 2,224

Head under 65 years .................................. 2,6992,130

2,716 2,731 2,635 2,317 2,322Head 65 years and over, ........................... 2,424

2,1952,448 2,450 2,437 2,062 2,061

3 persons ........................................................... 3,2072,069

3,229 3,246 3,127 2,745 2,749 2,6274 parsons ........................................................... 4,113 4,137 4,139 4,116 3,527 3,5285 parsons ........................................................... 4,645

3,5134,660 4,884 4,637 4,159 4,159

6 persons ........................................................... 5,4414,1,46

5,489 5,492 5,460 4,668 4,6697 persons or more ........................................... 6,678

4,6566,751 6,771 6,583 5,736 5,749 5,516

~Forunrelatedindividuals,sexof the individual.SOURCE U.S. Bureauof the Census Chamtiefisticsofthelow-incomepopulation1971, CunentPopulationRepofis,SefiesP+O, No,86, p.l B.

ratios greater than 1.0 are described as “above pover-ty.”

Poverty thresholds are computed on a nationalbasis only. These thresholds have not been adjusted forregional, state, or other local variation in the cost ofliving (except for the farm and nonfarm difference).None of the noncash public welfare benefits such asfood stamps or free food commodities are included inthe income of the low-income families reeeiving thesebenefits.

The threshold income values in 1971 for thecombinations listed above are shown in table V.

Other terms used in relationalanalysesThe following terms used in this report are abbrevia-tions of the data items or questions listed underdescription. These characteristics are listed on thevarious NHANES I questionnaires or derived fromdata collected on them.~

Term

Number of persons inhousehold

Marned

Education of examinae

Age of head of house-hold

Education of head ofhousehold

Total iron intake

Log total iron intake

Description

Total number of persons reported onNHANES household questionnaire.

Are you married? (yes or no)

What is the highest grade or year ofragular school the examinee hasever attended?

How old was the head of the house-hold on his or her last birthday?

What is the highest grada or year ofregular school the head of thehousehold has ever attended?

Milligrams of iron from all sources,consumed during a 24-hour period.

Logarithm (base 10) of total iron in-take,

NOTE: A list of references follows the text.

82

Term-Con. Description-Con.

Total animal tissue ironintake

Total nonanimal tissueiron intake

Total protein intake

Total animal tissue pro-tein intake

Total nonanimal tissueprotein intake

Estimated available iron

Estimated available ironwith tea

Vitamin C

Pica

Regular vitamin use

Regular mineral use

Regular vitamin andmineral use

Special diet

Ever breastfed

Age of mother atchild’s birth

Milligrams of animal tissue (i.e., frommeat, fish, and poultry) consumedduring a 24-hour period.

Milligrams of nonanimal tissue (i.e.,from eggs, dairy products, and allplant sources) consumed during a24-hour period.

Grams of protein from all sourcesconsumed during a 24-hour period.

Grams of protein from meat, fish,poultry consumed during a 24-hourperiod.

Grams of protein from all plantsources (grains, fruits, vegetables,etc.); eggs and all dairy productsconsumed during a 24-hour period.

Milligrams of iron from all sources,modified by vitamin C and animaltissue protein intake consumecl dur-ing a 24-hour period.

Milligrams of iron from all sources,modified by vitamin C, animal tissueprotein and tea intake consumedduring a 24-hour period.

Milligrams of vitamin C consum,edduring a 24-hour period.

Yes answers to “does person evereat dirt or clay, starch, paint c~rplaster, or any material that mightbe considered unusual?”

Has person taken vitamins within thelast 30 days and within the lastweek?

Has person taken minerals within thelast 30 days and within the la~tweek?

Does person report that he or shehas taken vitamins and/or mineralswithin the last 30 days and withinthe last week?

Are you on a special diet?

Was child breastfed at any time?

How old was child’s mother whlen(child) was born? years

.\

Term-Con.

1s child the first born?

Birth weight (pounds)

Premature bitth

Oral contraceptive us-age in past 6 months

Age began menstruat-ing

Ever been pregnant?

Ever been anemic?

Any trouble eating?

General findings

Present illness

Regular aspirin use

Abdominal operations

Desctiptiorr-Con.

How many children were born beforethis child?

How much did child weigh when hewas born? poundsounces

Was child born prematurely? (That is,early or not carried the full ninemonths.)

Have you taken birth control pills dur-ing the past 6 months?

How old were you when your periodsor menstrual cycle started?

Are you or have you ever been preg-nant?

Have you ever had anemia, some-times called “low blood?”

Do you have any illness or conditionwhich interferes with your eating?

Any findings on the general medicalexam.

Is child presently ill?

Has person taken aspirin within last30 days and within the last week?

Have you had an abdominal opera-tion?

o U, S. 00VEKNMENT PIWT13JCOFFICE : 7983 361-161/107

83

Vital and Health Statistics series descriptions

SERIES 1. Programs and Collection Procedures.-Reports describing the general programs of the National Center for Health

Statistics and its offices and divisions and the data col-

lection methods used. They also include definitions and

other material necessary for understanding the data.

long-term care, ambulatory care, hospital care, and family planning services.

SERIES 2. Data Evaluation and Methods Research.-Studies of new statistical methodology including experimental tests of

new survey methods, studies of vital statistics collection methods, new analytical techniques, objective evaluations of reliability of collected data, and contributions to sta- tistical theory.

SERIES 14. Data on Health Resources: Manpower and Fecilitias- Statistics on the numbers, geographic distribution, and

characteristics of health resources including physicians, dentists, nurses, other health occupations, hospitals,

nursing homes, and outpatient facilities.

SERIES 15. Data From Special Surveys.-Statistics on health and health-related topics collected in special surveys that are

not a part of the continuing data systems of the National Center for Health Statistics.

SERIES 3. Analytical and Epidemiological Studies.-Reports pre- senting analytical or interpretive studies based on vital and health statistics, carrying the analysis further than the expository types of reports in the other series.

SERIES 4. Documents and Committee Reports.-Final reports of major committees concerned with vital and health sta- tistics and documents such as recommended model vital

registration laws and revised birth and death certificates.

SERIES 20. Data on Mortality.-Various statistics on mortality other then as included in regular annual or monthly reports. Special analyses by cause of death, age, and other demo- graphic variables; geographic and time series analyses; and statistics on characteristics of deaths not available from

the vital records based on sample surveys of those records.

SERIES 10. Data from the National Health Interview Survey.-Statis- tics on illness, accidental injuries, disability, use of hos- pital, medical, dental, and other services, and other health-related topics, all based on data collected in the continuing national household interview survey.

SERIES 11. Data From the National Health Examination Survey and

the National Health and Nutrition Examination Survey.- Data from direct examination, testing, and measurement of national samples of the civilian noninstitutionalized

population provide the basis for (I) estimates of the medically defined prevalence of specific diseasas in the

United States and the distributions of the population with

respect to physical, physiological, and psychological characteristics and (2) analysis of relationships among the various measurements without reference to an explicit

finite universe of persons.

SERIES 12. Data From the Institutionalized Population Surveys.-Dis-

continued in 1975. Reports from these surveys are in-

cluded in Series 13.

SERIES 13. Date on Health Resources Utilization.-Statistics on the

utilization of health manpower and facilities providing

SERIES 21. Date on Netality, Marriage, and Divorce.-Various sta- tistics on natality, marriage, and divorce other than as included in regular annual or monthly reports. Special analyses by demographic variables; geographic and time series analyses; studies of fertility; and statistics on

characteristics of births not available from the vital records based on sample surveys of those records.

SERIES 22. Date From the National Mortality end Netality Surveys.- Discontinued in 1975. Reports from these sample surveys

based on vital records are included in Series 20 and 21, respectively.

SERIES 23. Data From the National Survey of Family Growth.-

Statistics on fertility, family formation and dissolution, family planning, and related maternal and infant health topics derived from a periodic survey of a nationwide

probability sample of ever-married women 1544 years of age.

For a list of titles of reports published in these series, write to:

Scientific and Technical Information Branch

National Center for Health Statistics Public Health Service

Hyattsville, Md. 20782

US DEPARTMENT OF HEALTH AND HUMAN SERVICES Public Health Serwce NatIonal Center for Health Statlsttcs 3700 East.West HIghway HyattwIle, Maryland 20782

OFFICIAL BUSINESS PENALTY FOR PRIVATE USE, $300

POSTAGE AND FEES PAID U.S. DEPARTMENT OF HHS HI-IS 396

THIRD CLASS BULK RATE

DHHS Pub. No. 83-1679, Series 11, No. 229

For a listing of publications in the VITAL AND HEALTH STATISTICS series, call 301436-NCHS