Analytic Considerations for
Collecting Data on Race, Ethnicity,
and Language
Romana Hasnain-Wynia, Ph.D., Northwestern University
November 30, 2010
AcademyHealth

The professional society of health services research and
health policy analysis.

AcademyHealth represents a broad community of people with
an interest in and passion for using health services research to
improve health care. We promote interaction across the health
research and policy arenas by bringing together a broad
spectrum of players to share their perspectives, learn from
each other, and strengthen their working relationships.

www.academyhealth.org
Aetna Foundation
The Aetna Foundation is a national health foundation that
promotes wellness, health, and access to high-quality health
care for everyone. Our giving is focused on addressing the
rising rates of adult and childhood obesity in the U.S.;
promoting racial and ethnic equity in health and health care;
and advancing integrated health care.
Originally founded in 1972 as the independent charitable and
philanthropic arm of Aetna Inc., Aetna and the Aetna
Foundation have contributed over $379 million in grants and
sponsorships since 1980. This work is enhanced by the time
and commitment of Aetna employees, who have volunteered
nearly 2 million hours since 2003. For more information, visit
http://www.aetnafoundation.org.
Learning Objectives

The goals of this session are to:
– Discuss why collecting data to reduce
health care disparities is important
– Address patient/enrollees concerns
– Discuss broad and granular categories
– Address the use of the data and data
standards
The audio and slide
presentation will be delivered
directly to your computer

Speakers or headphones are required to hear the
audio portion of the webinar.

If you are having difficulties with the audio-stream,
please dial (888)-632-5061 and enter the
Conference ID number: 92065340 followed by the
# sign.
Technical Assistance

If you require technical assistance, or are
having difficulties with the audio portion of
this webinar, please click on “Help”
Submitting Questions

Questions, both substantive & technical in nature, may be submitted at
any time during the presentation.

Click on “Ask a Question” below this presentation. Complete the form
and click “Submit.”

Responses will only be sent if related to a technical issue; otherwise
you will not receive a confirmation that your question has been
received.
Faculty
Romana Hasnain-Wynia, Ph.D
Director, Center for Healthcare Equity
Associate Professor, Research
Institute for Healthcare Studies
Division of General Internal Medicine
Northwestern University
Feinberg School of Medicine
Analytic Considerations for
Collecting Data on Race,
Ethnicity, and Language
Romana Hasnain-Wynia, Ph.D.
Northwestern University, Feinberg School of Medicine
Agenda





Why is collecting data to reduce health care
disparities important?
Address patient/enrollees concerns
Broad and granular categories
Use of the data
Questions and comments
Health Care Should Be






Safe
Effective
Patient-Centered
Timely
Efficient
Equitable
Major Reports on Disparities:
Plenty of Evidence that Disparities Exist
The “Usual” Explanations



Patient-Level
 Patient “preferences”: Treatment refusal, clinical
presentation of symptoms, mistrust
 Communication barriers
Provider-Level
 Beliefs/stereotypes re: patient health and behaviors
 Inadequate communication
 Bias/prejudice
Organizational –Level
 Structural and resource differences in where
different groups receive care
Reducing Disparities Within the
Health Care System
Kilbourne AM, et al. American Journal of Public Health. 96; 2113-2121: 2006.
Why Detecting/Understanding is Important For
Health Care Organizations
B. Siegel et al. Journal of Healthcare Quality (2007)
 Hospital and Health Care Leaders--- “NIMBY”
N. Lurie, et al. Circulation (2005)
344 Cardiologists:
 34% agree disparities exist overall
 12% believe disparities exist in own hospital
 5% believe disparities exist in own practice
S. Taylor, et al. Annals of Thoracic Surgery (2005)
208 Cardiovascular Surgeons:
 13% believe disparities occur often or very often
 3% believe disparities occur often or very often in own practice
T. Sequist, et al. 2008, Journal of General Internal Medicine (2008)
169 Primary Care Clinicians
 88% acknowledged that disparities in diabetes care existed in U.S.
 40% acknowledged disparities in own practice
Detecting and Understanding

Quality improvement requires high-quality data.
“The first and most critical step for Expecting Success was helping hospitals
gather data on patient race, ethnicity and primary language so that they had
accurate and complete information about their patients.”

Hospital leaders need to be willing to discuss the possibility
of disparities.
“Physicians and hospital leaders are committed to doing the right thing by their
patients, but there is a troubling reluctance among some leaders to consider gaps
in the quality of care by patient demographics. Hospital staff must be brave
enough to gather data and critically examine the evidence to learn if they are
providing care that is high quality and equitable.”
Source: Robert Wood Johnson Foundation, Expecting Success: Excellence in Cardiac Care Program
The Basic Building Block
To eliminate disparities,
we need to collect reliable data
and then use these data to improve care
The Case for Collection of Race,
Ethnicity, and Language Data
Race, ethnicity, and language data are needed to:

Stratify quality performance metrics

Organize and focus quality improvement and
disparity reduction initiatives

Track progress over time, locally and as a nation
The Case for Standardization
Standardized race, ethnicity and language data:

Support comparisons across organizations and
regions and over time

Support combination of data across organizations
or regions to create pooled data sets

Support reporting of, and replication of, successful
disparity-reduction initiatives
Institute of Medicine, 2009
What do patients and
health plan enrollees think?
“It is important for hospitals and
clinics to collect information from
patients about their race or ethnic
background”
Strongly agree
 Somewhat agree
 Unsure
 Somewhat disagree
 Strongly disagree

43%
37%
6%
10%
4%
Baker DW, Cameron KA, Feinglass J, Georgas P, Foster S, Pierce D, Thompson JA, Hasnain-Wynia R. “Patients'
attitudes toward health care providers collecting information about their race and ethnicity.” J Gen Intern Med. 2005
Oct;20(10):895-900.
“It is important for hospitals and
clinics to conduct studies to make
sure that all patients get the same
high-quality care regardless of their
race or ethnic background”
Strongly agree
 Somewhat agree
 Unsure
 Somewhat disagree
 Strongly disagree

93%
4%
2%
1%
0%
Baker DW, Cameron KA, Feinglass J, Georgas P, Foster S, Pierce D, Thompson JA, Hasnain-Wynia R. “Patients'
attitudes toward health care providers collecting information about their race and ethnicity.” J Gen Intern Med.
2005 Oct;20(10):895-900.
“How concerned would you be that
this data could be used to
discriminate against patients”
Not concerned at all
 A little concerned
 Somewhat concerned
 Very concerned

34%
15%
20%
31%
Baker DW, Cameron KA, Feinglass J, Georgas P, Foster S, Pierce D, Thompson JA, Hasnain-Wynia R. “Patients' attitudes
toward health care providers collecting information about their race and ethnicity.” J Gen Intern Med. 2005 Oct;20(10):895900.
Health Plan Enrollees
•Collection of language data is not worrisome
•Concerns about how health plans would use
race/ethnicity data
•Concerns about privacy/confidentiality
•Despite concerns, could consider ways that
data would be useful to plans
Hasnain-Wynia R, Taylor-Clark K, Anise A. Medical Care Research and Review. October
2010
Addressing
Patients/Enrollees Concerns
Tell People Why You are Asking
“Now I would like you to tell me your Race and Ethnic
Background. We use this to review the treatment patients
receive and make sure everyone gets the highest quality of
care.”
Testing introductions

Of the participants who were not completely
comfortable reporting their race and ethnicity
25.0% said that the quality statement made them
somewhat more comfortable
 25.6% said the quality statement made them much
more comfortable


Far better than the results for the other 3
statements
Baker DW, Cameron KA, Feinglass J, Georgas P, Foster S, Pierce D, Thompson JA, Hasnain-Wynia R. “Patients' attitudes
toward health care providers collecting information about their race and ethnicity.” J Gen Intern Med. 2005 Oct;20(10):895-900
Broad Categories
Granular Categories
Race and Ethnicity
Institute of Medicine
Recommendations

Health care organizations must have data on the
race, ethnicity, and language of those they serve in
order to identify disparities and to provide high
quality care.

Detailed “granular ethnicity” and “language need”
data, in addition to the OMB categories, can
inform point of care services and resources and
assist in improving overall quality and reducing
disparities.
Existing Guidance

OMB Directive – 1997


Hispanic/Latino Ethnicity
5 Race Categories

Progress has been made in incorporating the OMB
categories into many data collection activities – not
all are aligned

The OMB categories are insufficient to illuminate
many disparities and to target QI efforts efficiently
The Affordable Care Act:
Section 4302
Understanding Health Disparities: Data Collection and Analysis
Focuses on federal national data collection efforts and the analysis and
reporting of these data
Race and Ethnicity
 Guided by OMB standards for race and ethnicity
 Consultations with OMB, Dept of Labor, Bureau of Census and other
federal partners
 Informed by recent IOM reports on data granularity

Section 4302 has great potential to improve data collection by




Requiring the DHHS Secretary to establish data collection standards
Calling for the use of the standards in federal data collection
Instructing that the data be used for analyses and that the results be reported
Articulating some important language about funding
Data Standards

Must be for self-reported measures


Or for parents to report for children and guardians
to report for legally incapacitated adults
Must comply with OMB standards

The law states current OMB standards for race and
ethnicity must be used at a minimum
Modified Office of Management and
Budget (OMB) Categories
RACE QUESTION:
Which category best describes your
race?
 American Indian/Alaska
Native
 Asian
 Black or African American
 Native Hawaiian/Other
Pacific Islander
 White
 Multiracial
 Declined
 Unavailable/Unknown
ETHNICITY
QUESTION:
Do you consider yourself
Hispanic/Latino?
 Yes
 No
 Declined
 Unavailable/Unknown
Problems with Splitting Race and
Ethnicity—OMB Categories
Using OMB Categories Without
Splitting Race/Ethnicity
-African American/ Black
-Asian
-Caucasian/White
-Hispanic/Latino/White
-Hispanic/Latino/Black
-Hispanic/Latino/Declined
-Native American
-Native Hawaiian/Pacific Islander
-Multiracial
-Declined
-Unavailable/Unknown
Broad Categories

Process
-Ethnicity first

Ethnicity
 Hispanic or Latino
 Not Hispanic or Latino
 Declined
 Patient unavailable
Source: Rohit Bhalla, MD, MPH
Montefiore Medical Center
Bronx, NY, Expecting Success Site

Race
 American Indian or Alaskan Native
 Asian
 Black or African American
 Native Hawaiian or Other Pacific
Islander
 White
 Multiracial: Asian/Black-African
American
 Multiracial: Asian/White
 Multiracial: Black-African
American/White
 Multiracial: Other combination
 Declined
 Patient unavailable
The Rationale for Granular
Ethnicity Data

Disparities exist within the OMB categories



Differential pap screening rates among Asian subgroups
even when insured
Higher rates of childhood asthma and recent attacks
among Puerto Rican than Mexican ethnic groups
It is still important to use OMB race and Hispanic
ethnicity categories
Institute of Medicine, 2009
Granular Ethnicity - Mammography
Source: University of California, Los Angeles, Center for Health
Policy Research, California Health Interview Survey
Mother’s ethnicity and cesarean rates
all deliveries,* MA, 2004
10%
15%
20%
25%
30%
35%
40%
45%
Brazilian 40%
African 37%
Haitian 35%
Asian Indian 34%
W. Indian/Caribean 32%
European 32%
Dominican 31%
All Other 31%
Oth Port 31%
African Amer 30%
Native Amer 28%
Vietnamese 28%
Other Asian 27%
Middle Eastern 27%
Oth Hispanic 27%
Chinese 26%
Puerto Rican 26%
Oth Central Amer 25%
Cape Verdean 25%
Salvadoran 22%
Cambodian 15%
* A delivery of multiples is counted once
Center for Health Information and Statistics, MDPH
Variation in Breastfeeding Rates by
Asian Ethnicity
91%
Asian Indian
Japanese
Pakistani
Korean
Filipino
Thai
71%
MA TOTAL
Chinese
Vietnamese
Laotian
35%
Cambodian
0%
20%
40%
60%
Source: Asian Births in Massachusetts: 1996-1999; Hispanic Births in Massachusetts: 1996-1999; and
Black Births in Massachusetts: 1997-2000
80%
100%
Definition of Granular
Ethnicity
Ancestry, which the Census Bureau defines as “a
person’s ethnic origin or descent, ‘roots,’ or
heritage, or the place of birth of the person or
the person’s parents or ancestors before their
arrival in the United States” is the ethnicity
concept adopted by the subcommittee as the
level of detail necessary for quality improvement
Institute of Medicine, 2009
Granular Ethnicity



Collect granular ethnicity data as a separate
variable from the OMB race and Hispanic
ethnicity categories
Granular ethnicity categories should be
selected from a national standard list
Lists should include an “Other, please
specify:__” option for additional selfidentification
Institute of Medicine, 2009
Selecting Locally Relevant Granular
Ethnicity Categories
Local circumstances can dictate whether an entity uses
10 or 100 categories from the national standard list;
criteria for selection:
 Health and health care quality issues
 Evidence or likelihood of disparities
 Size of subgroups within the population
 Analyses of relevant data on the service or study
population
Institute of Medicine, 2009
Further investigation needed to:
•Examine different ways of framing the
questions and response categories at the
level of the OMB standards
•Monitor implementation of granular
ethnicity data collection to elicit “promising
practices”
Treating Patients with Limited
English Proficiency



80% of hospitals encounter LEP patients frequently –
63% daily/weekly; 17% monthly
81% of general internal physicians treat LEP patients
frequently – 54% at least once a day or a few times a
week; 27% a few times per month
84% of FQHCs provide clinical services daily to LEP
patients – 45% see more than ten patients a day; 39%
see from one to 10 LEP patients a day.
Collecting Data on Language
Need

Identify language need by determining:


how well an individual believes he/she speaks English
what language he/she needs for a health-related
encounter

Less than “very well” is defined as LEP

Where possible, also could collect language spoken
at home and language preferred for written
materials
IOM Recommendations for standardized collection of
race, ethnicity, and language need
 Hispanic or Latino
 Not Hispanic or Latino
OMB Race
(Select one or more)
 Black or African
American
 White
 Asian
 American Indian or
Alaska Native
 Native Hawaiian or Other
Pacific Islander
 Some other race
Spoken English Language
Proficiency
Language Need
Race and Ethnicity
OMB Hispanic Ethnicity




Very well
Well
Not well
Not at all
(Limited English proficiency is
defined as “less than very well”)
Granular Ethnicity
 Locally relevant choices
from a national standard
list of approximately 540
categories with CDC/HL7
codes
 “Other, please
specify:___” response
option
 Rollup to the OMB
categories
Spoken Language Preferred
for Health Care
 Locally relevant choices from a
national standard list of
approximately 600 categories
with coding to be determined
 “Other, please specify:__”
response option
 Inclusion of sign language in
spoken language needs list
and Braille when written
language is elicited
ALL patients/enrollees should be asked about
their race/ethnicity, and language

Self-reporting is the most accurate source of
information

Self-reporting will increase consistent reporting
within a health care institution
 Patients are more likely to select the same
categories to describe themselves over time
than staff who are assuming or guessing
Systematic Implementation
and Practical Concerns

Conduct education and feedback
sessions with leadership and staff

Define issues and concerns and
identify how you will respond to
them

Training and education components
should include
 Policy context
 Revised policies
 New fields
 Screens
 Leadership-staff materials
 Staff scripts
 FAQs and potential answers
 Specific scenarios
 Staff questions
 Monitoring
Two Approaches to Using Broad Categories
National Health Plan Collaborative: Toolkit to Reduce Racial and Ethnic Disparities in Health Care
Data Collection Methods
National Health Plan Collaborative: Toolkit to Reduce Racial and Ethnic Disparities in Health Care
Staff will feel more comfortable asking
patients/enrollees these questions if:

They understand
How these data will be used
 What disparities are
 Why they occur
 Their vital role in identifying and addressing
disparities and improving the quality of care for
everyone


Invest in Staff Training
Palaniappan, LP et al. Health Services Research, October 2009.
Hasnain-Wynia R, Taylor-Clark, Anise. Medical Care Research and Review. October 2010
What will change?
Standardizing Direct Data Collection





Who: information should always be asked of patients or their caretakers and should
never be gathered by observation alone
When: information should be collected upon admission or patient registration to
ensure that appropriate fields are completed when the patient begins treatment or
for plans, when the individual enrolls (as permitted by state law)
What:
 Questions about the OMB race and Hispanic ethnicity categories (one- or twoquestion format permitted)
 A question about granular ethnicity with locally relevant response categories
selected from a national standard set
 A question to determine English-language proficiency
 A question about language preference needed for effective communication.
Where: data should be stored in a standard format for easy linking to clinical data
How: patient concerns should be addressed when the information is being
obtained, and staff should receive ongoing training and evaluation.
How Organizations Can Use the
Data to Improve Care







Gain knowledge about the patient/enrollee population served
Identify populations at increased risk of adverse outcomes
Strengthen your organizations ability to develop culturally
appropriate materials to improve care
Track progress toward providing equitable care
Stratify distribution, utilization, process, outcome and patient
experiences with care measures
Improve outreach to local communities
Develop better data necessary for P4P and other programs
Weinick R, Flaherty K, Bristol S. Creating Equity Reports, 2008.
Wynia, Ivey, Hasnain-Wynia,. Collection of data on patients’ race and ethnic group by
physician practices. New England Journal of Medicine, 2010; 362:846-50
•Ensure patients from different
backgrounds get proper care
•Devise campaigns that tailor treatment
and counseling to patient ethnicities
•Tactics include
-Multilingual hand-outs
-On-site cooking lessons
-Culturally relevant
discussion of portion sizes
-Tracking patients in
electronic data systems to
target health tips and
patient education
materials
Effective Communication with Diverse Populations
as a Target for QI





Multi-stakeholder consensus report
Established specific measurable
expectations for organizational
performance – including data
collection
To guide quality improvement in
patient-centered communication
Focuses on needs of diverse
populations
Available at: www.EthicalForce.org
Using Assessments of Race/Ethnicity/Language
of Populations for
Within-Institution QI





To monitor quality of care for all groups
To design innovative programs to eliminate
disparities and rigorously test them
To know patients, better meet their needs, and work
with the community to deliver the best care possible
To satisfy legal, regulatory and accreditation
requirements (e.g., Joint Commission, CMS, NCQA
State Mandates, etc.)
To take the leadership position in eliminating health
care disparities and providing patient-centered care
Place-Based Disparities Exist:
Use Data to Make a Case for Resources





Disparities are multi-factorial—who you are and
where you go
Continued segregation in health care
Under-resourced institutions serve minority
communities
Focus incentives toward institutions serving a
large % of minority patients.
Target resources to areas of greatest impact
Implications for Research
Currently race/ethnicity data are on a sandy foundation—implementing
standardized collection methods and categories will strengthen and solidify the
foundation—more confidence in the data and in the ability to draw
conclusions.

Improving collection methods



Using national data sets for disparities research


Direct methods: how to collect race/ethnicity data and what categories to use
(both by providers and through surveys for research purposes)
Indirect methods (e.g., geocoding, surname recognition): specify when these
approaches are applicable and when they are not
Opportunities to link national databases and to conduct regional surveys including
over-sampling for racial/ethnic minorities.
Improving the availability and utility of patient-level administrative
data for disparities research

Ensure privacy and confidentiality and address sample size challenges by pooling
data
Thank You!
r-hasnainwynia@northwestern.edu
www.iom.edu/datastandardization/
Discussion

Questions for our speaker can be submitted
now.

Click on “Ask a Question” below this
presentation. Complete the form and click
“Submit.”
Methods Session: Register Today!
Community Based Participatory Research: Strategies for
Reducing Health Disparities
 Date: Wednesday, December 8, 2010 - 1:00 - 2:30pm
(Eastern)
 Faculty: Nina Wallerstein, Dr.P.H. and Lorenda Belone,
Ph.D., Robert Wood Johnson Foundation Center for
Health Policy at the University of New Mexico; Kevin R.
Shendo, Director of Education, Pueblo of Jemez
Visit www.academyhealth.org/training for more
information.
Thank You!
Please take a moment to fill out the online evaluation.