“Collecting Race and Ethnicity Data is Not Enough: Measuring and Reporting Disparities” Karen...

“Collecting Race and Ethnicity Data is Not Enough: Measuring and Reporting Disparities”

Karen Kar-Yee Ho, MHS

Lead Staff, NHDR, Agency for Healthcare Research and Quality

David R. Nerenz, PhD

Director, Center for Health Services Research,

Director, Outcomes Research, Neuroscience

Institute, Henry Ford Health System

Bruce Siegel, MD, MPH

Research Professor, George Washington University School of

Public Health and Health Services

Joseph R. Betancourt, MD, MPH

Director, The Disparities Solutions Center at

Massachusetts General Hospital

Moderator

Methods for the National Methods for the National Healthcare Disparities ReportHealthcare Disparities Report

Karen Ho, MHS

Lead Staff, NHDR

October 16, 2007

2006 National Healthcare Quality and Disparities Reports

Released Jan 11, 2007

How the Reports are Related

NHQR NHDR

Snapshot of quality of health care in America

Snapshot of disparities in health care in American

Quality Quality + Access

Safety, Effectiveness, Timeliness, Patient Centeredness

Safety, Effectiveness, Timeliness, Patient Centeredness + Equity

Variation across states Variation across populations

Types of DataSurveys collected from populations:Surveys collected from populations: AHRQ, Medical Expenditure Panel Survey (MEPS), 2002-2004 CAHPS® Hospital Survey, 2007 California Health Interview Survey, 2001-2005 Centers for Disease Control and Prevention (CDC), Behavioral Risk Factor

Surveillance System (BRFSS), 2001-2005 CDC-NCHS, National Health and Nutrition Examination Survey (NHANES),

1999-2004 CDC-NCHS, National Health Interview Survey (NHIS), 1998-2005 CDC-NCHS/National Immunization Program, National Immunization Survey

(NIS), 1998-2005 CDC-NCHS, National Survey of Family Growth (NSFG), 2002 Centers for Medicare & Medicaid (CMS), Medicare Current Beneficiary

Survey (MCBS), 1998-2003 National Hospice and Palliative Care Organization, Family Evaluation of

Hospice Care, 2005 Substance Abuse and Mental Health Services Administration (SAMHSA),

National Survey on Drug Use and Health (NSDUH), 2002-2005 U.S. Census Bureau, American Community Survey, 2004 National Center for Education Statistics, National Assessment of Adult

Literacy, Health Literacy Component, 2003

Data collected from samples of health care facilities Data collected from samples of health care facilities and providers:and providers:

National Sample Survey of Registered Nurses, 2004 CDC-NCHS, National Ambulatory Medical Care Survey (NAMCS), 1997-

2004 CDC-NCHS, National Hospital Ambulatory Medical Care Survey-Outpatient

Department (NHAMCS-OPD), 1997-2004 CDC-NCHS, National Hospital Ambulatory Medical Care Survey-Emergency

Department (NHAMCS-ED), 1997-2004 CDC-NCHS, National Hospital Discharge Survey (NHDS), 1998-2005 CMS, End Stage Renal Disease Clinical Performance Measures Project

(ESRD CPMP), 2001-2005 American Cancer Society and American College of Surgeons, National

Cancer Data Base (NCDB), 1999-2004 CDC-NCHS National Nursing Home Survey (NNHS), 2004

Data extracted from data systems of health care organizations:Data extracted from data systems of health care organizations:

AHRQ, Healthcare Cost and Utilization Project (HCUP) State Inpatient Databases disparities analysis file,* 2001-2004

CMS, Hospital Compare, 2006 CMS, Medicare Patient Safety Monitoring System, 2003-2005 CMS, Home Health Outcomes and Assessment Information Set (OASIS), 2002-2005 CMS, Nursing Home Minimum Data Set, 2002-2005 CMS, Quality Improvement Organization (QIO) program, Hospital Quality Alliance

(HQA) measures, 2000-2004 HIV Research Network data (HIVRN) data, 2001-2003 Indian Health Service, National Patient Information Reporting System (NPIRS), 2002-

2004 National Committee for Quality Assurance, Health Plan Employer Data and Information

Set (HEDIS®), 2001-2005 National Institutes of Health (NIH), United States Renal Data System (USRDS), 1998-

2003 SAMHSA, Treatment Episode Data Set (TEDS), 2002-2004

Data from surveillance and vital statistics systems:Data from surveillance and vital statistics systems:

CDC, National Program of Cancer Registries (NPCR), 2000-2004 CDC-National Center for HIV, STD, and TB Prevention, HIV/AIDS

Surveillance System, 1998-2005 CDC-National Center for HIV, STD, and TB Prevention, TB

Surveillance System, 1999-2003 CDC-NCHS, National Vital Statistics System (NVSS), 1999-2004 NIH, Surveillance, Epidemiology, and End Results (SEER) program,

1992-2004

Stratified data

By race and ethnicityBy race and ethnicity By incomeBy income By educationBy education By insurance statusBy insurance status

Multi-stratifications:Multi-stratifications: By race/ethnicity and incomeBy race/ethnicity and income By race/ethnicity and educationBy race/ethnicity and education By race/ethnicity and insuranceBy race/ethnicity and insurance Regression models??Regression models??

Table 1aWomen age 40 and over who reported they had a mammogram within the past 2 years,ª by race, United States, 2003

Population group Percent SE Percent SE Percent SE Percent SE Percent SE Percent SE Percent SE

Total 69.5 0.5 69.9 0.6 69.9 1.5 58.2 3.7 * * * * 67.7 5.0

40-64 70.5 0.7 70.9 0.7 71.9 1.7 58.0 4.0 * * * * 62.0 6.665 and over 67.7 1.0 68.1 1.0 65.5 2.8 56.1 7.7 * * * * * *

Private 76.0 0.7 76.0 0.7 80.5 1.9 67.1 4.1 * * * * * *Public 66.4 2.0 67.8 2.3 66.5 4.0 * * * * * * * *Uninsured 42.0 1.8 39.5 2.1 52.2 4.2 * * * * * * * *

Medicare and private 72.6 1.1 72.6 1.2 69.3 4.5 * * * * * * * *Medicare and public 64.9 2.6 67.0 2.9 63.3 6.2 * * * * * * * *Medicare only 58.2 1.9 58.1 2.1 61.9 4.8 * * * * * * * *

Negative/poor 55.4 1.6 55.2 2.0 59.1 2.9 * * * * * * * *Near poor/low 60.3 1.5 59.1 1.6 66.0 3.4 * * * * * * * *Middle 69.7 1.1 69.5 1.1 71.5 3.2 69.1 6.9 * * * * * *High 76.4 0.9 76.6 1.0 78.8 3.5 61.7 5.5 * * * * * *

Less than high school 58.3 1.4 57.9 1.6 59.6 3.5 54.0 7.9 * * * * * *High school graduate 67.2 1.0 66.9 1.2 73.5 2.4 52.7 6.3 * * * * * *At least some college 75.2 0.7 75.6 0.7 73.7 2.4 71.3 4.1 * * * * 78.0 4.7

Urban 69.6 0.6 70.1 0.7 70.7 1.5 57.1 3.8 * * * * 68.3 5.9Rural 69.2 1.0 69.3 1.1 65.2 4.1 * * * * * * * *

Age, not age adjusted

Family incomec

Education

Residence location

Health insurance,b

age 40-64

Health insurance, age 65 and over

Multiple racesAI/ANTotalSingle race

NHOPIAsianBlackWhite



Total 69.5 0.5 69.9 0.6 69.9 1.5 58.2 3.7 * * * * 67.7 5.0

40-64 70.5 0.7 70.9 0.7 71.9 1.7 58.0 4.0 * * * * 62.0 6.665 and over 67.7 1.0 68.1 1.0 65.5 2.8 56.1 7.7 * * * * * *



Negative/poor 55.4 1.6 55.2 2.0 59.1 2.9 * * * * * * * *Near poor/low 60.3 1.5 59.1 1.6 66.0 3.4 * * * * * * * *Middle 69.7 1.1 69.5 1.1 71.5 3.2 69.1 6.9 * * * * * *High 76.4



Total 69.5 0.5 69.9 0.6 69.9 1.5 58.2 3.7 * * * * 67.7 5.0

40-64 70.5 0.7 70.9 0.7 71.9 1.7 58.0 4.0 * * * * 62.0 6.665 and over 67.7 1.0 68.1 1.0 65.5 2.8 56.1 7.7 * * * * * *



Negative/poor 55.4 1.6 55.2 2.0 59.1 2.9 * * * * * * * *Near poor/low 60.3 1.5 59.1 1.6 66.0 3.4 * * * * * * * *Middle 69.7 1.1 69.5 1.1 71.5 3.2 69.1 6.9 * * * * * *High 76.4 0.9 76.6 1.0 78.8 3.5 61.7 5.5 * * * * * *

Less than high school 58.3 1.4 57.9 1.6 59.6 3.5 54.0 7.9 * * * * * *High school graduate 67.2 1.0 66.9 1.2 73.5 2.4 52.7 6.3 * * * * * *At least some college 75.2 0.7 75.6 0.7 73.7 2.4 71.3 4.1 * * * * 78.0 4.7

Urban 69.6 0.6 70.1 0.7 70.7 1.5 57.1 3.8 * * * * 68.3 5.9Rural 69.2 1.0 69.3 1.1 65.2 4.1 * * * * * * * *


Family incomec

Education

Residence location

Health insurance,b

age 40-64


Multiple racesAI/ANTotalSingle race

NHOPIAsianBlackWhite

Table 1bWomen age 40 and over who reported they had a mammogram within the past 2 years,ª by ethnicity, United States, 2003

Population group Percent SE Percent SE Percent SE Percent SE Percent SE

Total 69.5 0.5 70.0 0.6 70.4 0.6 70.0 1.5 65.1 1.7

40-64 70.5 0.7 71.1 0.7 71.6 0.8 72.0 1.8 63.9 1.865 and over 67.7 1.0 67.6 1.0 68.1 1.1 65.4 2.8 69.5 3.3

Private 76.0 0.7 76.3 0.7 76.2 0.8 80.9 1.9 71.5 2.5Public 66.4 2.0 66.0 2.2 67.8 2.7 65.9 4.1 68.5 4.1Uninsured 42.0 1.8 39.3 2.2 34.9 2.6 51.7 4.2 51.4 3.3

Medicare and private 72.6 1.1 72.8 1.2 72.7 1.2 70.5 4.4 59.7 7.2Medicare and public 64.9 2.6 63.7 2.9 66.1 3.4 62.7 6.3 71.3 4.7Medicare only 58.2 1.9 57.2 2.1 57.2 2.3 61.4 4.8 63.9 5.8

Negative/poor 55.4 1.6 54.7 1.9 53.8 2.5 58.5 3.0 59.7 3.1Near poor/low 60.3 1.5 59.6 1.7 57.9 2.0 65.8 3.4 63.9 3.2Middle 69.7 1.1 70.0 1.1 69.8 1.2 71.6 3.2 65.3 3.4High 76.4 0.9 76.6 0.9 76.8 1.0 80.1 3.5 68.4 3.6

Less than high school 58.3 1.4 58.6 1.9 58.0 2.3 58.4 3.6 59.8 2.3High school graduate 67.2 1.0 67.4 1.1 66.9 1.2 73.7 2.4 62.7 3.7At least some college 75.2 0.7 75.3 0.7 75.6 0.8 74.3 2.4 73.2 3.1

Urban 69.6 0.6 70.2 0.7 70.7 0.8 70.8 1.5 65.1 1.8Rural 69.2 1.0 69.4 1.1 69.5 1.1 65.2 4.1 62.4 5.4

Hispanic, all races

Family incomec

Education

Residence location

Non-Hispanic

All races White Black



Total

Health

insurance,b age 40-64



Total 69.5 0.5 70.0 0.6 70.4 0.6 70.0 1.5 65.1 1.7

40-64 70.5 0.7 71.1 0.7 71.6 0.8 72.0 1.8 63.9 1.865 and over 67.7 1.0 67.6 1.0 68.1 1.1 65.4 2.8 69.5 3.3





Urban 69.6 0.6 70.2 0.7 70.7 0.8 70.8 1.5 65.1



Total 69.5 0.5 70.0 0.6 70.4 0.6 70.0 1.5 65.1 1.7

40-64 70.5 0.7 71.1 0.7 71.6 0.8 72.0 1.8 63.9 1.865 and over 67.7 1.0 67.6 1.0 68.1 1.1 65.4 2.8 69.5 3.3





Urban 69.6 0.6 70.2 0.7 70.7 0.8 70.8 1.5 65.1 1.8Rural 69.2 1.0 69.4 1.1 69.5 1.1 65.2 4.1 62.4 5.4

Hispanic, all races

Family incomec

Education

Residence location

Non-Hispanic

All races White Black



Total

Health

insurance,b age 40-64

Comparisons for disparities

Reference group Each race group is compared with data for whites Hispanics are compared with non-Hispanic whites Poor populations are compared with high income populations Uninsured and publicly insured are compared with privately insured

Earliest data year (1999/2000) is compared with most recent data year (2004/2005) available.

Disparities exist if Relative differences are at least 10% and statistically significant with p<0.05,

assessed using z-tests.

Change over time Difference must have p<0.05 and a geometric rate of change of >1% per

year.

Challenges Faced in National Data

Data on specific racial, ethnic, and socioeconomic groups were often – Not collected – Collected in different ways– Sample size could not provide reliable estimates

Race Data Collection

New Office of Management and Budget New Office of Management and Budget guidelines for the collection of race and guidelines for the collection of race and ethnicity in federal data in effect since ethnicity in federal data in effect since 1997.1997.– Expands collection of racial data from 4 groups Expands collection of racial data from 4 groups

to 5 groups to 5 groups – Identify >1 raceIdentify >1 race– Federal agencies had until 2003 to implementFederal agencies had until 2003 to implement– Non-federal data are not subject to OMB Non-federal data are not subject to OMB

standardsstandards

Data gaps

Gaps in data for:Gaps in data for:

Racial groupsRacial groups– Native HawaiiansNative Hawaiians– American IndiansAmerican Indians– AsiansAsians– Mixed raceMixed race

Priority PopulationsPriority Populations ChildrenChildren Rural residentsRural residents People with special People with special

healthcare needshealthcare needs

Why Identify and Track Disparities?

To reduce disparities in health care by:To reduce disparities in health care by:

Informing targeted improvement strategies-Informing targeted improvement strategies- Health care achieved for one population should be Health care achieved for one population should be achievable for all populations.achievable for all populations.

Establishing a national standard-Establishing a national standard- States, communities, and providers can measure their States, communities, and providers can measure their successes and opportunities for improvement in successes and opportunities for improvement in comparison to the nation.comparison to the nation.

Evaluating progress and change over time-Evaluating progress and change over time- Data are needed to identify successful interventions Data are needed to identify successful interventions and opportunities for improvement.and opportunities for improvement.

Collecting Race and Ethnicity Data is Not Enough: Measuring and Reporting

Disparities

October 16, 2007

Bruce Siegel, MD, MPH

Expecting Success Sites

Duke University Hospital

Durham, NC

Mount Sinai HospitalChicago, IL

Sinai-Grace HospitalDetroit, MI

Montefiore Medical CenterNew York, NY

Memorial Regional Hospital

Hollywood, FL

University of Mississippi Medical CenterJackson, MSDelta Regional Medical Center

Greenville, MS

University Health SystemSan Antonio, TX

Del Sol Medical CenterEl Paso, TX

Washington Hospital CenterWashington, DC

Expecting Success Hospital Applicants, 2005

3%

97%

Collects R/E Does Not Collect

5%

95%

Yes QI Initiatives No QI Initiatives

Siegel, Bretsch, Sears, Regenstein, & Wilson. Assumed equity: early observations from the first hospital disparities collaborative. Journal for Healthcare Quality 2007;29(5):11-15.

% Collecting R/E Data% Reported QI Initiatives to

Reduce Care Disparities

n = 4

n = 118

n = 6

n = 112

Collecting standardized R/E/L DataLessons Learned

□ Engage all stakeholders Changing R/E/L fields affects registration staff,

ambulatory sites, patient registries, language services

□ Work with Information Systems – altering fields and testing changes

□ Educate staff on why collecting the data is important

□ Address registration staff anxiety Patients did not “pushback” as expected, esp. when

told “why” up front

Collecting standardized R/E/L DataLessons Learned (cont’d)□ Post-implementation check-in

Focus group three months

□ Monitor individual performance and provide feedback Directly observe patient registration Record pre-registration phone calls Include R/E/L in performance reports and evaluation Regularly provide data on unknowns, declines back to

registration staff Make it routine!!

The many uses of R/E/L data

With reliable R/E/L data, hospitals can:□ Provide person centered care

Target cultural and linguistic competence efforts From menus to interpreters

Understand educational needs, customize materials

□ Analyze service lines Truly understand your market - it may be different from

what you expected

□ Capture changes in hospital demographic trends□ Stratify quality measures and find quality

opportunities

R/E/L: Promises and Challenges

Hospital XHospital XDischarges with "Unknown" Demographics

Q4 2005 and Q4 2006

26

95

45

62

96

0102030405060708090

100

Race Ethnicity Primary Language

Pe

rce

nt o

f Dis

cha

rge

s

4th quarter 2005 4th quarter 2006

A challenge to R/E/L data collection

26%

3%

55%

Hospital DPercent of Patients Declining to Specify Race

Q4 2006

01020304050

60708090

100

All Non-Hispanic Hispanic

Hospital Y Percent of Heart Failure Patients Receiving Discharge Instructions by Ethnicity

2005 Quarter 4 - 2006 Quarter 4

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

2005Q4 2006Q1 2006Q2 2006Q3 2006Q4

Year/Quarter

Per

cen

t o

f P

atie

nts

Hispanic Patients

Not Hispanic Patients

83.3%

65.6%

85.7%

75.3%

92.9%

88.3% 91.3%

90.0%

100.0%

100.0%

Closing the Gap

Transitions and DisparitiesHospital X

Readmissions Within 30 Days by Race

Q4 2005 through Q4 2006

11.0% 10.2%

18.7%

1.3% 1.3% 3.0%

0%

5%

10%

15%

20%

25%

*Readmits HF (HFR) *Patient Readmit HF(HFR-P)

*Patient Readmit (any cause, HFR-P2)

*p<.05

Per

cent

of H

F D

isch

arge

s

Black

White

Measurement caveats

□ If looking at “core” or HQA measures: Large numbers of exclusions Thus small sample sizes and many measures to review

Potential solutions Aggregate data Use “all or nothing” measures:

Larger samples Fewer measures More patient-centered

Hospital Z30-Day Same Cause Readmission

RatesQ4 2006 Discharges

0% 10% 20% 30% 40% 50%

HIV (N=225)

Ped. Asthma

(N=487)

COPD (N=260)

Stroke (N=172)

ALL

RACE

Black

White

ETHNICITY

Hispanic

Non-Hispanic

LANGUAGE

English

Spanish

Broadening the Use of R/E/L Data

www.expectingsuccess.org

Using Data on Race/Ethnicity to Identify Disparities in Quality of Care and to Track Progress of Efforts to Reduce Disparities

David R. Nerenz, Ph.D.Center for Health Services Research

Henry Ford Health System

October 16, 2007

Essential Steps

Obtain Data on Race/Ethnicity

Link to HEDIS/CAHPS Data

Stratify HEDIS/CAHPS Data

Identify “Significant” Disparities

Plan QI Project(s)

Evaluate Impact

No More Disparity!

Examples of HEDIS Data Stratified by Race/Ethnicity at the Individual Health Plan

Level

Asthma: Outpatient Follow-up

After Acute Episodes• Core concept:

Outpatient follow-up after either ER visit or admission

• Children 5-17 years old

• Standard based on national expert panel guidelines

0

10

20

30

40

50

60

70

Follow-up Rate

Caucasian

African-American

Comprehensive Diabetes Care:

Foot Exam Performed

0

20

40

60

80

100

White African American Hispanic Asian Overall

White vs. African American (p<0.001), White vs. Hispanic (p<0.001) and White vs. Asian (p<0.001).

Rate

Multiple Disparities in HEDIS Measures

in Single Health Plan(Six-State Medicaid Project)

0

10

20

30

40

50

60

70

HbA1cTesting

GoodGlycemicControl

AppropriateAsthma

Meds

PrenatalCare

CaucasianAfrican American

Hispanic

Per

cent

Source: Single Health Plan analysis of HEDIS data – 2003, unpublished

Comparison of non-Hispanic/Hispanic Breast Cancer Screening by Commercial, Medicare

Risk, and Medicaid Products in a Single Health Plan, 2000

0

20

40

60

80

100

Commercial Medicare Risk Medicaid

non-Hispanic Hispanic

Rate

P=.001 non-Hispanic population

Asthma Medication ManagementAsthma Medication ManagementReporting Year 2003Reporting Year 2003

African-AmericanAfrican-American CaucasianCaucasian

NumeratNumeratoror

DenominatDenominatoror

RateRate NumeratNumeratoror


RateRate

All Co’sAll Co’s 411411 600600 69%69% 698698 921921 76%76%

AA 189189 272272 69%69% 174174 218218 80%80%

BB 153153 213213 72%72% 375375 499499 75%75%

CC 6969 115115 60%60% 149149 204204 73%73%

Breast Cancer Screening Breast Cancer Screening Reporting Year 2003Reporting Year 2003

African-AmericanAfrican-American CaucasianCaucasian

NumeratNumeratoror


RateRate NumeratNumeratoror


RateRate

All Co’sAll Co’s 11161116 14681468 76%76% 25812581 31683168 81%81%

AA 390390 519519 75%75% 536536 650650 82%82%

BB 435435 561561 78%78% 14151415 17191719 82%82%

CC 291291 388388 75%75% 630630 799799 79%79%

Examples of Tracking Stratified

HEDIS Data over Time

Improvements in Quality of Care for African American Health Plan

Members with Diabetes

HbA1c Testing

0

20

40

60

80

100

African American Members

Percent2003

2004

LDL-C Testing

0

20

40

60

80

100

African American Members

Percent 2003

2004

Another Approach to Evaluating

QI Program Success• Asthma severity

definition involving ER visits and admissions

• Focus on African-American members with asthma

• Used shift in distribution of severity categories as measure of program success

• Statistically significant using Chi-square test

0

10

20

30

40

50

60

70

80

90

100

Mild Moderate Severe

Pre-InterventionPost-InterventionP

erc

en

t

Comparison of Caucasian and African American

HbA1c Testing in a Single Plan

0

20

40

60

80

100

1998 1999 2000

Caucasian African American

Rate

Disparities in Medicare Managed Care (HEDIS)

Measures Over Time• Standard, widely-

used quality measures

• Trends from 1997 or 1999 to present

• Improvements in quality overall, reduction in disparities in some HEDIS measures, but not all

• Trivedi et al, NEJM, August 18, 2005

Childhood Immunization – Combo I – (HEDIS 1999 Definition)

0

5

10

15

20

25

2002 2003 Rolling2003-2004

First Q2004

HispanicTotal

Perc

ent

0

5

10

15

20

25

30

35

Feb Mar Apr May June July Aug Sep Oct Nov

Month in 2006

Mam

mog

ram

s

Number of Mammograms Billed per Month for African American women

Barrier Analysis Survey

Provider focusgroups

Mammogram Blitz

Summary

• There are a number of health plans that have been able to collect race/ethnicity data, link it to HEDIS or other quality of care data bases, and identify disparities in quality of care.

• The same basic methods can be used to repeat analyses in future time periods in order to track progress on reducing disparities.

• In some cases, “supplemental” analyses can be done to identify associations between specific initiatives and changes in process of care measures.

Question and Answer Period

Type your question in the “chat” box on the lower right of your screen, select “host” and click on “send”

to submit your question.

www.mghdisparitiessolutions.org

“Of all the forms of inequality, injustice in healthcare is the most shocking and inhumane.”

–Dr. Martin Luther King, Jr.

To receive email notifications of upcoming free web seminars and other future activities at the Disparities Solutions Center, please go to our website and click on the “sign up” link on our homepage.

“Collecting Race and Ethnicity Data is Not Enough: Measuring and Reporting Disparities” Karen...

Documents

Transcript of “Collecting Race and Ethnicity Data is Not Enough: Measuring and Reporting Disparities” Karen...