Research in PHC

37
Research in PHC Research in PHC ممم ممم م م مممم مممم م م مممممم مممممم م م م م ممممم مممممPO Box 27121 – Riyadh 11417 Tel: 4912326 – Fax: 4970847 Introduction to Primary Care: a course of the Center of Post Graduate Studies in FM 1

description

بسم الله الرحمن الرحيم. Research in PHC. Introduction to Primary Care: a course of the Center of Post Graduate Studies i n FM. PO Box 27121 – Riyadh 11417 Tel: 4912326 – Fax: 4970847. Objectives:. Appreciate the various uses of epidemiology in day to day practice - PowerPoint PPT Presentation

Transcript of Research in PHC

Page 1: Research in PHC

Research in PHCResearch in PHC

مممممممممممممممممممممممممممممممممممممممممممم

PO Box 27121 – Riyadh 11417 Tel: 4912326 – Fax: 4970847

Introduction to Primary Care: a course of the Center of Post Graduate Studies in FM

1

Page 2: Research in PHC

Objectives:

• Appreciate the various uses of epidemiology in day to day practice

• Define and distinguish between key measures of disease frequency

• Explain the main features of study designs• Discuss measures of impact and association• Describe what is statistics and types of data.• Describe the average and spread • Explain what is p-value & confidence interval• Research Phobia in Family Medicine

2

Page 3: Research in PHC

Epidemiology:

• Study of distribution & determinants of disease frequency in human population & application of this study to prevention & control health problems.

• Population : groups of people with common characteristics as age, gender, disease …etc.

Last’s Dictionary of Epidemiology.11

• Disease must be clearly defined in order to determine accurately who should be counted.

• Disease definition: physical + pathological exam, diagnostic test results & S/S. e.g.H1N1 definition.

3

Page 4: Research in PHC

Epidemiology:

Cases source: hospital registries, death certificates, surveys & reporting system – cancer, TB …etc

Uses of epidemiology :1. Population or community health assessment -

Measuring disease burden in a population.2. Investigating etiology (causation)3. Determining natural history and identifying

predictors of outcome.4. Evaluation of intervention

Measuring disease frequency:

Health states : “disease present” or “disease absent”

To establish disease presence criteria requires a definition of “normality” & “abnormality.”

4

Page 5: Research in PHC

Population at riskCorrect testimate of number of people under study i.e. people who are susceptible to a given disease . e.g..

population at risk: identified by demographic, geographic or environmental factors. . e.g.

1- cervical disease - women 2- occupational injuries: brucellosis - working on farms & in slaughterhouses

Ratios, Proportions & Rates : = x/y × 10nRatio : values of x & y is completely independent, or x ise included in y : (1) male/female (2) female/all Proportion : ratio in which x is included in y - female/allRate : is often a proportion, measured over time. Rate = population at risk during the same time period # of cases or events occurring during given time period

× 10n

Ratios, Proportions & Rates

5

Page 6: Research in PHC

prevalence : frequency of existing cases in a population at a given point in time

Prevalence = # of people with disease /condition at a specified time ________________________________________ ×10n # of people in the population at risk at specified time

Incidence = # new events in a specified period __________________________________ 10n #persons exposed to risk during this period

Prevalence = incidence × duration of disease

Relationship between incidence & prevalence

Low incidence & a high prevalence – e.g. diabetes – or High incidence & a low prevalence – e.g. URTI. why?

Incidence : rate of occurrence of new cases arising in a given period in a population

Incidence & prevalence of diseases…

6

Page 7: Research in PHC

Prevalence increased by:

Age & disease severity, duration & number of cases..etc

Decreased by:

Factors influencing prevalence :-

1. Longer duration of disease2. Prolongation of life of patients without cureمم new cases - incidence4. In-migration of cases5. Out-migration of healthy people6. In-migration of susceptible people7. Improved diagnostic facilities (better reporting)

1. Shorter duration of disease2. High case-fatality3. rate from diseaseمم new cases - incidence5. In-migration of healthy

people6. Out-migration of cases7. Improved cure rate of cases

Uses of prevalence

1- Assessing need for preventive action, healthcare & planning of health services.

2- Chronic diseases occurrence - DM, rheumatoid arthritis7

Page 8: Research in PHC

Prevalence (unlike) can be influenced by factors unrelated to disease cause not provide strong evidence of causality.Attack rate

Often used instead of incidence during a disease outbreakin a narrowly-defined population over a short period of time.

Attack rate= # of people affected/# of exposed

For example, in the case of a food-borne disease outbreak (food poisoning) , the attack rate can be calculated for each type of food eaten, and then these rates compared to identify the source of the infection.

8

Page 9: Research in PHC

Types of study designs

Experimental studies

Observational studies

CohortXCSEcological CCS

Analytic studiesDescriptive studies

Case report Case series

9

Page 10: Research in PHC

1- Essential prerequisite for all health care levels.

2- Establish priorities.

3- Determine health policy.

7- Effective use of available resources8- Adjustment of health strategies to situations changing.

9- Provide rational foundation for decisions & introducing objectivity into decision-making process (EBM).

Purposes of research

4- Identify causes of diseases

5- Identify risk factors of diseases

6- Identify pts health education needs.

10

Page 11: Research in PHC

Descriptive studies :

Describing the characteristics of a particular situation, event or case. 2 types:-(1) Case report & case series:• Describe in-depth characteristics of one / limited number of ‘cases’. • A case may be, patient, a health centre, or a village.• It can provide quite useful insight into a problem. • Case studies are common in inclinical medicine. e.g., characteristics of a hitherto unrecognized illness may be documented as a case study. It is often first step toward building up a clinical picture of that illness. e.g. HIV diagnosis started as reported cases of similar unusual groups of symptoms on 1980 . However, if to test whether findings pertain to a larger population, a more extensive, XCS is needed. 11

Page 12: Research in PHC

(2) Cross-sectional surveys (XCS) :Descriptive studies :

XCS aim: describe & quantify distribution of certain variables in a study population at one point of time. e.g.:•Physical characteristics: people, materials/environment as— prevalence surveys ( bilharzias, HIV)— evaluation of coverage ( immunization,)•SE characteristics of people as age, education, marital status, number of children & income• Behavior or practices of people & knowledge, attitudes, beliefs, (KAP studies), or •Events that occurred in the population. XCS cover a selected sample of the population. If a XCS covers total population it is called a census.

Looking at present

Small surveys can reveal associations between certain variables, as between having TB & SES.

If describe +compare groups within study population comparative / analytical studies.

12

Page 13: Research in PHC

Comparative or analytical studies

1- XC comparative studies 2- CCS 3- Cohort

Cross –sectional comparative studies

XC surveys focus on describing + comparing groups.e.g. : a survey on malnutrition to establish:• % of malnourished children in a population; • SE, physical variables influence availability of food; • Feeding practices; & knowledge, beliefs & opinions influence these practices. The researcher will :- describe these variables &, by comparing malnourished & well-nourished children, he will determine which SE, behavioral & other independent variables may contributed to malnutrition.

Search for cause & effect Or why & how

e.g.

Smoking - lung ca, Salmonella outbreak-eat shawerma

13

Page 14: Research in PHC

Comparative or analytical studies

In any comparative study, watch out for CONFOUNDING or INTERVENING variables.

Study A found an association between cigar smoking & baldness - The study was confounded by age

Study C found improved perinatal outcomes for birthing centers when compared to hospitals

– The study may be confounded by highly motivated volunteers who select the birthing center option

•Confounding is an apparent association between disease and exposure caused by a third factor not taken into consideration. •A confounder is a variable that is associated with the exposure & independent of that exposure, is a risk factor for the disease. Age is the strongest confounder. e.g.

14

Page 15: Research in PHC

Case-control studies: CCS : longitudinal

CCS Design

Target Population

Exposed

Not Exposed

Exposed

Not Exposed

Diseased

(Cases)

Not Diseased

(Controls)

TIME

Direction of inquiry

Start with:

Enroll gp. of people with disease (or other outcome) (cases) & a gp. without this disease (controls) & compare their patterns of previous exposure to a risk factor

An observational analytic study that identify & compare affected & non-affected subjects to determine risk of association for investigated disease.

15

Page 16: Research in PHC

CCS

CCS provide a relatively simple way to investigate causes of diseases, especially rare diseases.

In a study of causes of neonatal death, investigator will first select ‘cases’ (children who died within first month of life) & ‘controls’ (children who survived their first month of life), then interviews their mothers to compare history of these 2 groups of children, to determine whether certain risk factors are more prevalent among children who died than among those who survived.

Selection of cases

1- Selected cases should represent all cases in study population 2- Cases selected on basis of disease not exposure

e.g.

3- Define diagnostic criteria for disease i.e. :Case definition -clinical criteria as restriction by time, place, & person.

e.g. H1N1 16

Page 17: Research in PHC

CCS

Controls should come from same ‘source’ population. Of cases e.g. from same hospital If not they would not be comparable to cases.

Classic e.g.: discovery of relationship between thalidomide & limb defects in babies born in Germany in 1959 and 1960. The study, done in 1961, compared affected children with normal children.

Control confounding variables by matching the groups

In a study on causes of malnutrition in children-3 yrs match well & malnourished on 1- age ( strongest confounder), 2- economic status of parents.

Selection of control e

.g.

Key is to identify appropriate control or comparison group.

م

17

Page 18: Research in PHC

CCS

Controls should come from same ‘source’ population. Of cases e.g. from same hospital If not they would not be comparable to cases.

Selection of cases & control

Selected cases should represent all cases in studied population Cases selected on basis of disease not exposure

An estimate of the ratio of incidence rates or risks (relative risk) is obtained by calculating an odds ratio (OR)

CCS uses

Exposure

YesNo

Disease

Yesab

Nocd

ad

bcOR =

“2-by-2” table

18

Page 19: Research in PHC

Cohort Study

the whole cohort is followed up & observed over a period of time to if development of the disease (or other outcome) differs between the (E) & (Ē) groups .

longitudinal / incidence/prospective studies

An observational analytic study that identify exposed (E ) & unexposed (Ē) population & follow them prospectively over time to determine rate of specific disease event. Begin by categorizing subjects on basis of exposure to

potential cause (risk factor) – they are free of disease- : study group (E) & control group (Ē).

Smoking lung ca., bronchitis…etcSo directly measure incidence in E & Ē

Source of data :1- Existing records:- medical, employment2- study subjects :Interviews, Qers, physical exam. or a test19

Page 20: Research in PHC

Cohort Study

Target Population

Exposed

Not Exposed

Diseased

Not Diseased

Diseased

Not Diseased

TIME

People withoutthe disease

Direction of inquiry

Cohort study uses

1- best information about causation of disease 2- most direct measurement of risk of developing disease.

can be both prospective and retrospective depending on time of data collection

Framingham Study Since 1948, samples of residents of Framingham, followed up

for risk factors related to occurrence of heart disease. 20

Page 21: Research in PHC

224

176

No CHD

(Controls)

31288Non-smoker

288112SmokerExposure Status

TOTAL

CHD cases

(Cases)

Disease Status

Example: Calculating the Relative Risk

Relative Risk (RR) =

= = a/(a+b)

b/(c+d)

112 / 288

88 / 312

= 1.38

For cohort or CCS

If XCS : approximate RR/ OR OR = ad/bc =

Ie/Iu

RR

112 224

88 176= 1.53

21

Page 22: Research in PHC

CCS Slow Fast

Common dis. Rare dis.

Ethical problems ± signf. Ethical problems minimal Volunteers needed Volunteers: no need Large sample Small sample Attrition problems No attrition problems Less bias susceptible Selection , recall bias Incidence determn No incidence calcult RR accurate RR approximate (OR)

Cohort

Expensive Cheap

Defined population # Undefined population # 22

Page 23: Research in PHC

Intervention / Experimental studiesInvolves a direct comparison of 2or more intervention

Aims 1- Strongest/gold standard test to a hypothesis

3- To determine effective Rx. 2- To determine a causative factor

1- Prospective in nature.

Distinc

tions

5- Ethical considerations: as:

3- Feasibility problem: as4- Cost.c- population selection

a- time b-manpower

2- Investigator manipulates /change / intervenes with E for one group

a- Harmful agents b- Useful Rx or vaccine.

Reference pop.

Non- participants

Comparison gp.

Participants ( study pop.)

Rx gp.

Experimental pop.

Rx allocation

1- Staff training - as WBC nurses to improve their performance2- Health education for obese patients to loss weight.

Intervention conduction at PHC level :-

23

The hallmark : investigator dictates each subject’s exposure.

Page 24: Research in PHC

Experimental followed by cohort, CCs then XCS.

Study types power Strength of hypothesis testing

XCSCCSCohortAt one time , Ask Q “What is happening?” Prevalence.Generate hypotheses

Advantages:

1-Quick 2- cheap 3- available data.

4- Frequent: 1st step in investigating E-O.

5- Correlate DM data

Disadvantages:

1- Inability to link E-O. 2- can not measure E-O

3- confounders

Inquiry :backward time. Ask “ What happened?” Test hypothesis Case detfntion.

Advantages: 1- Timely:: diaseas with long incubation period 2-Cost effective. 3-Rare dis. 4-Multiple exposuers1. 5-Ideal:unwell defined pop. as outbreaks can implicate disease sources > cohort.

Disadvantages:

Bias: recall & selection

Inquiry : forward time. Ask “What will happen? Test hypothesisIncidenceTest causatiob (RR).Advantages 1- E-O temporal sequence 2- Rare E 3-Multiple outcomes2

4- Exam dis. cause. 5- Disease natural history 6- Identify RFs. 7- well defined pop. Disadvantages 1- Time-consuming. 2- Expensive. 3- Bias: loss follow–up/attrition

Page 25: Research in PHC

Basic biostatistics: concepts and toolsNeeded for summarizing and analyzing data

Summarizing data : Data are either numerical or categorical variables.

• Numerical variables : 1- Counts - # children of a specific age 2- measurements - height & weight.• Categorical variables : 1- The result of classifying - individuals can be classified into categories according to their blood group; A, B, O/ AB.2- Ordinal data – which express ranks – as cancer grading.

• Summary numbers include medians, means, ranges, standard deviations, standard errors and variances.

25

Page 26: Research in PHC

Basic biostatistics: concepts and tools

•Tables & graphs : important means of summarizing & displaying data, but seldom prepared with sufficient care. Aim: to display data so quickly & easily understood. Each table / graph should be self-expressing: contain enough information so that it can be interpreted without reference to text.

Figure 1.1 :Distribution of cholera cases in London, August-September 1854 Table 1.1: Mortality from cholera

in e London -July 1854

Districts with Water Supplied

Cholera Death Rate /1,000

Southwark

Lambeth

5.0

0.9

26

Page 27: Research in PHC

Frequency distributions & histograms…

Frequency distribution : organization of a data set into contiguous mutually exclusive intervals.

Displayed : 1- a histogram : no space between bars. 2- Bar chart. 3- Pie chart.

27

Frequency distributions, measures of central location, and measures of dispersion are effective tools for summarizing numerical characteristics such as height, BP, & incubation period..

Page 28: Research in PHC

28

Measures of Central Location & dispersion

Measures of central location are single values that represent center of observed distribution of values. Different ways :-

Arithmetic Mean The most commonly used measure. It is arithmetic average - “mean” or “average.” In formulas Mean = x = Σ xi/n

Mode : Value occuring most frequently in a set of data .

Median i.e. middle

Identifying median from individual data:1. Arrange observations - increasing /decreasing order.2. Find Middle rank = (n +1)/2a. If # of observations (n) is odd: median = middle rankb. If n is even, middle rank falls between 2 observations & median isequal to average of values of those observations. e.g.

Page 29: Research in PHC

Measures of Central Location & dispersion

Odd number of observations : set of data with n = 5: 13, 7, 9, 15, 111. Arrange observations in increasing or decreasing order : either: 7, 9, 11, 13, 15 or: 15, 13, 11, 9, 7. 2. Find Middle rank = (n +1)/2 = 5+1 /2=3 median lies at value of 3rd observation - 11.

Even number of observations: set of data with n = 6: 15, 7, 13, 9, 10, 11 1. Arrange the observations :- increasing or decreasing order - 15,13,11,10,9,7 2. Find Middle rank = (n +1)/2 = 6+1 /2=3.5 median lies halfway between values of 3rd & 4th observations. = average of 13 & 9 = 13+2/2= 10.5. 29

13+7+9+11++13+7/5 : Arithmetic Mean= = Σ xi/n = 60/5= 12. Mode = 7.

Page 30: Research in PHC

p-value & confidence interval

•Assessments of role of chance :- hypothesis testing, which produces a ‘p-value’ – i.e check that this is an unbiased study findings.•Assessment of whether or not findings are ‘significantly different’ or ‘not significantly different’ from some reference value .

approach to statistical significance Threshold value is 0.05 or 0.01.

If the P-value is 0.05, there is a 95% probability that :

– The results did not occur by chance

– The null hypothesis is false

– There really is a difference between the groups i.e. there is a significant effect. 30

Page 31: Research in PHC

31

Measures of Dispersion\ deviationRange, Minimum Values, and Maximum Values.

Standard Deviation (SD) : measures of dispersion of observations around the mean of a distribution.

Show relationships of mean & SD.

Normal distributions bell-shaped

• 68.3% the area under normal curve lies between the mean ± 1 SD.• 95.5% of the area lies between the mean ± 2 SD• 99.7% of area lies between mean ± 3 SD.• 95% of area lies between the mean and ± 1.96 SD.

Page 32: Research in PHC

Population parameter is interfered from sample statistic.

Point Estimate for population mean μ & Error : Sample mean x is a point estimate for population mean μ x for a random sample will not be exact same value as true μ.

Probability note: reality is that population mean is either inside or outside the range we have calculated.

32

Page 33: Research in PHC

95% Confidence Intervals (CI)

33

95% of area under normal curve lies between ±1.96 SD on each side of the mean. So the 95% confidence limits=Lower 95% confidence limit = x −(1.96×SE)Upper 95% confidence limit = x +(1.96×SE)

The true mean has a 95% probability of lying between the limits we find.

•Confidence limits are also calculated for proportions, rates, risk ratios, odds ratios, and other measures when we wish to draw inferences from a sample to population at large. •The interpretation of the confidence interval remains the same: ( narrower interval, more precise our estimate of population value ( & more confidence we have in our study value as an estimate of population value);.

Page 34: Research in PHC

Research Phobia in Family Medicine

Historical background : Past decades : Family doctors (FD) involved in manual practice & distant from Ideas & Theory of Research. But now...

Frequent Discoveries & Health Authorities are often asking us to change our prescribing behavior

We need to study & to work in group with Research tools :Epidemiology. EBM, Qualitative Research

Myths against Research It is necessary to change but FDs still resist hard… We often think:- “We are inferior & very practical “Research is high Theory for academic people” “We have no time” Too much Statistics NO tools for research

34

Page 35: Research in PHC

Behavioral Therapy

First small steps……The Idea..

Do not be afraid of the white empty page…Start from the richness of FM : Informal ideas,problems and feelings connected to daily

practice are the real “steam-engine” of Research

New Development in FM (Group practice, PC, Telematics, not expensive software) can facilitate a change

“cognitive-behavioral” therapy can be useful to break “mental walls” still surviving in our open world

First step: Follow steps of a flow chart

35

Page 36: Research in PHC

Steps of conducting a research project

36

Page 37: Research in PHC

Thank you

تم بحمد تم بحمد اللهالله

37