Population Health Surveys: Data to Improve Public Health in an Era of Health Care Reform (Text Version)

On September 29, 2010, Richard Brown made this presentation at the 2010 Annual Conference. Select to access the PowerPoint® presentation (2 MB).

---

Slide 1

Population Health Surveys: Data to Improve Public Health in an Era of Health Care Reform

E. Richard Brown, PhD
Director, UCLA Center for Health Policy Research
Professor, UCLA School of Public Health
Principal Investigator, California Health Interview Survey

Agency for Healthcare Quality and Research

September 29, 2010

Slide 2

Why do we need population health survey data?

• Why do we need population health data?
• Health care reform promises a new era of electronic health records.
• What will electronic health records provide for health services research and what will population surveys and other types of population health data provide?
  • Some key examples and what they have to offer health services researchers.
• Combining population health survey data and other data for research on causes of disparities.

Slide 3

Why do we need population health data?

• Data and analysis to assess and track health and social problems—"no data, no problem"
• Population health data needed to:
  • Identify health needs
  • Track changes in population's health
  • Identify disparities in health and health care
  • Analyze causes and consequences of disparities and unmet needs
  • Suggest factors that can improve health

Slide 4

Electronic health records

• Health care reform and ARRA's HITECH act mandate wide adoption of interoperable electronic health records.
• Data generated in process of delivery of clinical services:
  • Information associated with each patient (physician services, other health personnel, labs, prescriptions, etc.
  • Information recorded by health personnel.
  • Information for each service provided (if it is recorded).
• Health services research nirvana:
  • Nation's health system may join the VA, the Netherlands and New Zealand, current leaders in use of electronic health records.
  • Data to help improve quality of care.
  • Excellent research data on clinical processes of care.

Slide 5

Electronic health records

• Evolving system:
  • For next decade, some provider systems and geographic areas likely to be leaders and broad adopters, but others will be limited.
  • Even when electronic health records complete, population health surveys still needed.
• What do other types of population health data provide that still will be needed?

Slide 6

Vital statistics

• Births, deaths, fetal deaths, marriages, divorces:
  • Good for tracking major life events (census of all events).
  • Some have information relevant to health services research (e.g., birth and death certificates), especially as outcomes.
• Limitations?
  • But they have little information for in-depth analysis:
    • Not designed to answer policy research questions, lacking detailed information on key demographics and other factors.
    • Some data validity and reliability limitations (e.g., unreliable recording of cause of death, inaccurate coding of race/ethnicity).
  • National Vital Statistics System [Image: The NVSS logo]:
    • States and localities collect data.
    • Federal and state government efforts to improve data quality.

Slide 7

Administrative data

• Collected in delivering health programs and services:
  • Electronic health records are type of admin data.
  • Public programs have data on enrollees, claims, etc.
• What's it good for in health services research?
  • Good for analyses of who got into programs and who got services.
  • Not very useful for answering questions about who did not get in and did not get services.
• Limitations?
  • Usually lacking in detailed data on demographics and income.
  • Often unreliable recording of race/ethnicity.
  • Usually limited variables for analysis (usually not designed to answer policy research questions).

Slide 8

Administrative data: Hospital discharge data

• National data sets and state data sets on hospitalizations:
  • NCHS's National Hospital Discharge Survey (annual sample of hospitalizations from national sample of hospitals).
  • State-based discharge data on all hospitalizations (e.g., census of all cases):
• Analyses of variations in hospitalization rates:
  • Potentially preventable hospitalizations:
    • Especially for ambulatory care sensitive conditions—e.g., preventable with better access to primary and preventive care.
  • Re-admissions (which identify remediable quality issues)
• Limitations?
  • Limited information on individuals' demographics, etc.

Slide 9

Administrative data: Health programs

• Medicare and Medicaid enrollees:
  • Program enrollment data
  • Claims and service data
• What's it good for in health services research?
  • Good for analyses of:
    • Who is in program
    • Who has gotten services (among those in program)
• Limitations?
  • Not useful for answering questions about who is not in program.
  • Analytic utility depends on information collected during program administration and service delivery.

Slide 10

Population health surveys

• Population health surveys collect self-reported information from defined populations:
  • Probability sample of population—i.e., representative of population (if done properly and well)
• People left out of many programs and admin/clinical data systems:
  • The uninsured and others not in public programs
  • Marginalized populations—e.g., groups often adversely affected by health and health care disparities (defined by social characteristics)

Slide 11

Population health surveys

• Information not systematically included in vital statistics, administrative data, or electronic health records:
  • Health care information not recorded in administrative data—e.g., health insurance coverage, access to health care, use of some services, experiences with health care system.
  • Health behaviors—e.g., substance use, dietary behaviors, physical activity, interaction with physical and social environment.
  • Health indicators not clinically recorded—e.g., general health status, chronic disease symptoms, mental health, etc.
  • Socio-demographic information—e.g., race/ethnicity, immigration status, languages spoken, education, employment, income.
  • Social determinants—e.g., social capital measures, social support.
• Often only data source for self-reported information.

Slide 12

Population health surveys

• Many federal surveys generally provide data on national sample, but often with geographic identifiers.
• National Health Interview Survey (NHIS) [Image: NHIS logo]:
  • NCHS collects data primarily in-person from cross-sectional sample of civilian noninstitutionalized population.
  • High response rate, high quality data.
  • Large national sample with some large subsamples for some states.
  • Principal source of information on health for U.S. population:
    • Annual federal reports—Health, United States; National Healthcare Disparities Report; Healthy People
    • Benchmark for other surveys—comparing estimates to those of NHIS

Slide 13

Population health surveys

• Medical Expenditure Panel Survey (MEPS) [Image: MEPS logo]:
  • AHRQ's large national surveys of families and individuals, their medical providers, and employers:
    • Household Component—data from sample of families and individuals in selected communities, based on representative subsample of households that participated in NHIS.
    • Insurance Component—data from sample of private and public sector employers on health plans offered to their employees.
    • Medical Provider Component—covers hospitals, physicians, home health care providers, and pharmacies used by MEPS respondents.
  • Most complete source of data on use of health care and health insurance coverage and expenditures for them.

Slide 14

Population health surveys

• National Health & Nutrition Examination Survey (NHANES) [Image: NHAMES logo]:
  • In-depth interviews plus clinical examination and lab tests
  • Used to determine actual (distinguished from self-reported) prevalence of major diseases and risk factors
  • Used to assess nutritional status and its association with health promotion and disease prevention
  • Basis for standards for height, weight, and blood pressure
  • Used in epidemiological studies
• Behavioral Risk Factor Surveillance System [Image: BRFSS logo]:
  • CDC-sponsored state-level population health telephone survey conducted by all 50 states, DC, and territories
  • Collects limited information across all states for wide range of health indicators
  • Comparable data for all states

Slide 15

Population health surveys

• Some state surveys provide uniquely valuable data for health services research.
• California Health Interview Survey [Image: CHIS logo]:
  • Very large telephone survey of California population, capturing geographic, ethnic and other social diversity:
    • Since 2001, conducted in six languages, county-level samples.
    • CHIS sampled > 50,000 households in most years.
  • Extensive data on many health and social indicators—e.g., health insurance coverage, access to health care, use of some services, health behaviors, mental health, chronic illness, general health status, socio-demographic information, social determinants.
  • Data files and query system easily accessed—www.chis.ucla.edu

Slide 16

Population surveys + other data

• Combining population health survey data and other data for research on causes of disparities:
  • Link health survey data with electronic medical record and coverage program administrative data
  • Analyses of multiple types of factors that affect health, for example:
    • Health insurance and access to care
    • Clinical processes and quality of care
    • Health behaviors and mental health
  • One-off data sets (increasing concerns about data confidentiality)
    or
    federated data system (grid computing, complex data sharing)

Slide 17

Population surveys + other data

• Simpler example from policy research on obesity
• Combining population health survey data (California Health Interview Survey) and other data on social and built environment:
  • Study by Susan Babey, UCLA Center for Health Policy Research
• How is food environment (types of food outlets) related to prevalence of obesity and diabetes?
  • Fast food restaurants and convenience stores
  • Supermarkets, produce markets and farmers markets

Images: A photograph of roadside signs for various fastfood restaurants, and another photograph showing bins full of fresh citrus fruits are shown.

Slide 18

Population surveys + other data

• Retail Food Environment Index:
  • Ratio of fast-food restaurants and convenience stores vs. grocery stores and produce vendors near a person's home

  RFEI = (Fast Food + Convenience Stores) / (Supermarkets + Produce Markets + Farmers Markets)

• Using data from:
  • 2005 California Health Interview Survey on individual-level data on demographics and health-related variables
  • 2005 InfoUSA Business File on location of several types of food outlets
  • 2000 Census on neighborhood income

Susan Babey, PhD, UCLA Center for Health Policy Research.

Slide 19

Food environment associated with neighborhood income

Image: Bar graph titled "Average Retail Food Environment Index by Neighborhood Income Using Urbanicity-Specific Buffers, Adults, California 2005" shows the following data:

• Low-income Neighborhood: 4.9
• Higher-income Neighborhood: 4.1*

Note: *Significantly different from "Low-income Neighborhood"; p<0.05.

Source: 2005 California Health Interview Survey, 2005 InfoUSA Business Data and 2000 U.S. Census
Courtesy of Susan Babey, PhD, UCLA Center for Health Policy Research.

Slide 20

Association of obesity with food environment

Image: Bar graph titled "Percent Obese as a Function of RFEI Using Urbanicity-Specific Buffers, Adults Age 18 and Over, California, 2005" shows the following data:

• <3: 19.6%
• 3-4.9: 22.6%*
• 5+: 23.8%*

Note: * Significantly different from "< 3"; p<0.05.

Source: 2005 California Health Interview Survey, 2005 InfoUSA Business Data and 2000 U.S. Census
Courtesy of Susan Babey, PhD, UCLA Center for Health Policy Research.

Slide 21

Factors associated with obesity

Factors Associated with Obesity, Adults 18+, California

Variable	OR	95% CI
Retail Food Environment Index
<3	ref
3 - 4.9	1.12*	1.01-1.25
5+	1.18**	1.06-1.31
Individual Household Income
Low-Income (<200% FPL)	1.21**	1.10-1.33
Higher income	ref
Neighborhood Income
Low-Income (<200% FPL)	1.27**	1.17-1.38
Higher income	ref
Race Ethnicity
White	ref
Latino	1.47**	1.32-1.63
Asian	0.34**	1.28-1.41
African American	1.85**	1.60-2.15
American Indian	1.55**	1.14-2.10
Other (incl. Pacific Islander)	1.31*	1.05-1.62

Note: Also adjusted for Age and Gender; * p<0.05, **p<0.01.

Source: 2005 California Health Interview Survey, 2005 InfoUSA Business Data and 2000 U.S. Census
Courtesy of Susan Babey, PhD, UCLA Center for Health Policy Research.

Slide 22

Association of diabetes with food environment

Image: Bar graph titled "Percent Diagnosed with Diabetes as a Function of RFEI Using Urbanicity-Specific Buffers, Adults Age 18 and Over, California, 2005" shows the following data:

• <3: 7.7%
• 3-4.9: 8.5%
• 5+: 9.3%*

Note: * Significantly different from "< 3"; p<0.10.

Source: 2005 California Health Interview Survey, 2005 InfoUSA Business Data and 2000 U.S. Census
Courtesy of Susan Babey, PhD, UCLA Center for Health Policy Research.

Slide 23

Factors associated with diabetes

Factors Associated with Diabetes Adults 18+, California

Variable	OR	95% CI
Retail Food Environment Index
<3	ref
3-4.9	1.15*	0.98-1.36
5+	1.25***	1.06-1.46
Individual Household Income
Low-Income (<200% FPL)	1.62***	1.41-1.87
Higher income
Neighborhood Income
Low-Income (<200% FPL)	1.23***	1.07-1.40
Higher income	ref
Race Ethnicity
White	ref
Latino	2.02***	1.71-2.39
Asian	1.22*	0.98-1.53
African American	1.81***	1.45-2.26
American Indian	2.88***	1.72-4.82
Other (incl. Pacific Islander)	1.56***	1.15-2.10

Note: Also adjusted for Age and Gender
; *p<0.10, **p<0.05, ***p<0.01.

Source: 2005 California Health Interview Survey, 2005 InfoUSA Business Data and 2000 U.S. Census
Courtesy of Susan Babey, PhD, UCLA Center for Health Policy Research.

Slide 24

Retail Food Environment Index (RFEI) by Los Angeles County Service Planning Areas

Image: A color-coded map shows sections of Los Angeles County by RFEI:

• West, Metro, and San Gabriel Valley: 3.0-3.9
• South Bay and San Fernando Valley: 5.0-5.9
• South, East, and Antelope Valley: 6.0-6.9

Statewide average RFEI is 4.2.

Note: RFEI is ratio of fast food restaurants plus convenience stores divided by grocery stores plus produce vendors. Community with RFEI of 2.0 has twice as many fast food restaurants and convenience stores as it does grocery stores and produce vendors.

Source: 2005 California Health Interview Survey, 2005 InfoUSA Business Data and 2000 U.S. Census
Courtesy of Susan Babey, PhD, UCLA Center for Health Policy Research.

Slide 25

Prevalence of overweight or obesity by Los Angeles County Service Planning Areas

Image: A color-coded map shows percentage of overweight/obese persons by sections of Los Angeles County:

• West: 40-49%
• Metro and San Gabriel Valley: 50-54%
• South Bay and San Fernando Valley: 55-59%
• South, East, and Antelope Valley: 60-65%

Source: 2005 California Health Interview Survey. Susan Babey, PhD, UCLA Center for Health Policy Research.

Slide 26

Diabetes prevalence by Los Angeles County Service Planning Areas

Image: A color-coded map shows percentage of diabetic persons by sections of Los Angeles County:

• West: less than 6%
• Metro, South Bay, Antelope Valley, San Fernando Valley, and San Gabriel Valley: 6-7.5%
• East: 7.5-9.9%
• South: 10% and above

Source: 2005 California Health Interview Survey. Susan Babey, PhD, UCLA Center for Health Policy Research.

Slide 27

From policy research to policy action

• This study of effect of Retail Food Environment on obesity and diabetes led to:
  • Legislative efforts at state level to mandate providing consumers with information on calorie content of restaurant food items.
  • When applied to Los Angeles County Service Planning Areas, encouraged LA City to adopt zoning ordinance banning new fast food outlets in South LA (a predominantly Latino and Black low- and moderate-income area).
• All politics (and policy) is local.
• Using population health survey data, health services research can help improve health policy and address health disparities.

Slide 28

CHIS data widely used in policy and research

• Used in research—more than 170 peer-reviewed journal articles published using CHIS data.
• AHRQ uses CHIS data in congressionally mandated National Healthcare Disparities Report:
  • Information on Asian subpopulations in California—Chinese, Filipino, Japanese, Korean, Vietnamese, and South Asian
    (in 2008, 34% of Asians in US lived in California)
  • Information on Hispanic subpopulations in California—Mexican, Puerto Rican, Central American, and South American (in 2008, 30% of Hispanics in US lived in CA)
  • Available by gender, age, income, immigration status, languages spoken, etc.
• CHIS extensively used to develop and advocate for policy in California.

Image: The cover of the National Healthcare Disparities Report is shown.

Slide 29

Keep in touch!

E. Richard Brown, Ph.D.
erbrown@ucla.edu
www.healthpolicy.ucla.edu

California Health Interview Survey [Image: CHIS logo]
visit www.chis.ucla.edu  for:

• Information about CHIS
• Use of free online data query system [Image: A button labeled "Ask CHIS"]
• Data files