|Home | About | Journals | Submit | Contact Us | Français|
Disparities in healthcare coverage and access have a prominent place on the national health policy agenda. It is, therefore, essential to understand strengths and limitations of national surveys that provide annual or periodic data for population-based healthcare disparities research and monitoring. Importantly, publicly available data on healthcare coverage and access are needed for disparities populations defined by race, ethnicity, or immigrant group (REI).
To document public use data availability for REI groups, insurance coverage, and access to care measures in selected national surveys used for healthcare disparities research.
We examined public use data for general population surveys that collect information on healthcare coverage and access on an annual or periodic basis for the nation. Data sources examined include the following: Current Population Survey, Survey of Income and Program Participation, National Health Interview Survey (NHIS), National Health and Nutrition Examining Survey, National Survey of Children’s Health, Behavioral Risk Factor Surveillance System, and Medical Expenditure Panel Survey-Household Component.
Although each survey has strengths for healthcare disparities research, there is no single survey that has detailed REI group identifiers, comprehensive measures of coverage and access, and geographic identifiers.
Current Population Survey and NHIS have detailed REI identifiers. NHIS and Medical Expenditure Panel Survey-Household Component have comprehensive measures of coverage and access but are limited by smaller samples and no geography. Findings summarized in this article will assist with identifying existing data to examine healthcare coverage and access disparities and highlight areas for improvement in public use data availability.
Disparities in healthcare coverage and access have a prominent place on the national health policy agenda as evidenced by the monitoring efforts of Healthy People 2010 and the annual National Healthcare Disparities Report. Access to healthcare is 1 of 10 Leading Health Indicators (LHI) that is routinely monitored at the national level. Each LHI has Healthy People 2010 objectives associated with it. The objectives used to monitor access to healthcare in the United States are as follows: (1) insurance coverage, (2) having a usual source of care, and (3) prenatal care in the first trimester.1 The National Healthcare Disparities Report is an important vehicle to monitor population-level healthcare disparities in terms of access and quality.2 However, lack of data for some race, ethnicity, or immigrant (REI) groups precludes a comprehensive examination of healthcare disparities.3-5
To identify potential data sources for healthcare disparities research, we reviewed publicly available data from selected national surveys that collect information on healthcare coverage and access. Although others have recently reviewed data availability issues for assessing healthcare disparities,6-8 our review is unique in several ways that complement this previous work. First, we document detailed race and ethnic groups identified in the public use data and corresponding sample sizes. Second, we document the availability of public use data for immigrant groups by national origin. Third, we focus on “access to care” but go beyond the measures represented by the LHI objectives to look at a broader set of variables representing insurance coverage and access to care. Our intent is to document existing data availability and highlight potential areas for improvement in available survey data that will facilitate identifying, monitoring, and ultimately eliminating health disparities.
We examined public use data for general population surveys that collect information on healthcare coverage and access on an annual or periodic basis for the nation. Prenatal care is not measured in any of these national surveys. Data sources that we examined included the following: the Current Population Survey (CPS), the Survey of Income and Program Participation (SIPP), the National Health Interview Survey (NHIS), the National Health and Nutrition Examining Survey (NHANES), the National Survey of Children’s Health (NSCH), the Behavioral Risk Factor Surveillance System (BRFSS), and the Medical Expenditure Panel Survey-Household Component (MEPS-HC). Table 1 provides an overview of each survey. Our review focused on the availability of public use data for 2 content areas: (1) race, ethnicity and immigrant group identifiers; and (2) health insurance coverage and access to care measures. We briefly describe the measures in the following paragraphs.
In 1997, the Office of Management and Budget (OMB) issued a revision to OMB Directive 15, which provides recommendations for the collection and classification of race and ethnicity.9 The OMB guidelines were not developed to define the concepts of race or ethnicity; they were intended only to standardize data collection and publication across federal agencies.
The minimum OMB standard for collecting data on race is to include the following: (1) American Indian or Alaska Native (AIAN), (2) Asian, (3) Black or African American, (4) Native Hawaiian or other Pacific Islander (NHOPI), and (5) White. All surveys conformed to the OMB minimum standards for collection of data on race. However, data collection varied across surveys.
In federal surveys, ethnicity refers to Hispanic ethnicity. The minimum OMB standard for collecting data on Ethnicity includes the following options: (1) Hispanic/Latino and (2) Not Hispanic/Latino. The surveys varied in terms of group-specific detail collected from none to as many as 8 Hispanic subgroups.
We use the term immigrant to broadly represent the distinction between US born and foreign born and more specifically to identify foreign-born groups by country of origin. There are no OMB standards for the collection of data on immigrant groups. We reviewed publicly available variables for each survey to determine whether respondents could be grouped as US or foreign born, as well as for detailed information on country of origin.
We examined public use data looking specifically for measures of healthcare coverage and access. Some surveys have other healthcare related measures, but our review focused on concepts for health insurance and access to care outlined later.
We examined whether publicly available data allow researchers to distinguish types of insurance coverage. Several surveys collected detailed health insurance coverage types. Some even collected information about service providers that often act like “health insurance” like Indian Health Service and the Veterans Administration (VA). For the uninsured, we assessed whether variables allow researchers to distinguish point-in-time uninsurance from past year intermittent uninsurance or all year uninsurance. A critical review of uninsurance measures has been previously documented for several of the surveys.10-12
The concept of access to care is not well-defined and measures representing access are not universally agreed upon.13,14 A critical review of access measures is beyond the scope of this article, but previous authors have addressed how this concept has been measured in surveys.15-17 We examined publicly available variables related to 3 elements: usual source of care, barriers to care, and utilization of care. Source of care was assessed by variables related to the respondents’ usual provider or usual place for healthcare. Barriers to care were assessed by variables related to delayed care and foregone care in the past year. Utilization was assessed by variables related to the interval since the last doctor visit and measures indicating the numbers of past year doctor visits, dental visits, emergency room visits, or hospital overnight stays.
Most surveys reviewed collected detailed race and ethnicity information, although this detail is not necessarily available in the public use files. Table 2 shows REI groups identified in the public use data and un-weighted sample sizes for each.
NHANES provides a single variable combining ethnicity and race, therefore, only distinguishing non-Hispanic White, non-Hispanic Black, and non-Hispanic Other from all Hispanics. SIPP identifies 3 single race categories (White, Black, Asian) and a residual other category that includes AIAN, NHOPI, and multiple race individuals. NSCH identifies 2 single race categories (White and Black) with a residual other for all states. However, AIAN, Asian, and NHOPI can be individually identified for selected states. MEPS identifies all 5 OMB categories for single race individuals and a residual multiple race category. CPS identifies all 5 single race categories and provides detailed identifiers for many multiple race combinations. BRFSS and NHIS provide several race variables, allowing researchers to choose single race categories with a residual multiple race category or a recode that allocates multiple race individuals to a race category based on self-identified preferred race. The NHIS is the only public use data with an expanded race variable, distinguishing some Asian subgroups.
Three surveys (SIPP, NSCH, BRFSS) only distinguish Hispanic from non-Hispanic, while the NHANES distinguishes Mexican from other Hispanic and non-Hispanic. The CPS, NHIS, and MEPS-HC provide additional detail on Hispanic origin so that Hispanic subgroups can be disaggregated.
Immigrant group identifiers are the most limited across the public use data files. Two surveys, BRFSS and MEPS-HC, do not identify respondents’ place of birth. Two surveys (SIPP, NSCH) identify whether the respondent was born in the United States or not, whereas the NHANES distinguishes those born in the United States, Mexico, or elsewhere. The NHIS distinguishes those who are US born from those born in 10 global regions and a residual foreign-born category. The CPS provides the most detailed information, with 149 countries of origin identified for respondents in 2007.
Publicly available data for insurance coverage types and the coverage time period varied across surveys (see Table 3), primarily due to differences in the data originally collected.
With 2 exceptions (BRFSS and NSCH), available data allow for the distinction between basic private (employer/union sponsored and privately purchased) and public insurance coverage (Medicare, Medicaid, State Children’s Health Insurance Plan (SCHIP), and other public programs) with many offering detail on types of coverage. BRFSS only distinguishes those who are covered and not covered, while NSCH identifies whether a child had coverage through Medicaid or SCHIP; there is no additional detail. Public use data for CPS, SIPP, NHANES, NHIS, and MEPS contain variables representing different types of coverage. Those covered by Medicare can be identified in all 5 surveys. Medicaid and SCHIP are distinguishable in 4 surveys (CPS, SIPP, NHANES, NHIS), while they are grouped together in MEPS. Data for those who report Military coverage are aggregated in 3 surveys (SIPP, NHANES, MEPS). In CPS and NHIS, it is possible to distinguish CHAMPUS/TRICARE, CHAMP/VA, and VA/Military. Three surveys (CPS, NHANES, NHIS) have a variable to identify those who report access to Indian Health Service.
The CPS uninsurance variable represents all year uninsurance, whereas the BRFSS variable represents point-intime uninsurance coverage. NHANES and NSCH variables represent point-in-time coverage and past year intermittent coverage. NHIS insurance variables are similar to NHANES and NSCH, but include additional specification of the number of months not covered in the past 12 months allowing estimation of all year uninsurance. Because they are panel surveys and data are collected several times, SIPP and MEPS can be used to estimate point-in-time uninsurance, past year intermittent uninsurance, the length of time someone has been uninsured, and all year uninsurance.
Three elements were reviewed: usual source of care, barriers to care, and utilization of care (see Table 3). Source of care: Variables varied from a single indicator of whether the respondent has a usual doctor (BRFSS) or where the respondent seeks healthcare (SIPP) to multiple questions about usual place and type of place (NHIS, NHANES, MEPS), usual provider and type of provider (MEPS), and reason for no usual place (MEPS). The NSCH has a set of variables related to whether the child has a primary care provider and access to a “medical home.” Barriers to care: Three surveys had no measures of barriers to care. The 4 surveys with measures of barriers varied from a single variable indicating whether care was foregone in the past year (BRFSS) to variables representing types of delayed care (MEPS), reasons for delayed care (NHIS, MEPS), types of foregone care (NHIS, MEPS, NSCH), and reasons for foregone care (MEPS, NSCH). Utilization: The CPS has no utilization measures, and BRFSS has a single variable indicating the amount of time since the last doctor visit. Three surveys (NHANES, NHIS, NSCH) had a measure of the interval since last doctor visit and additional measures of various types and numbers of past year healthcare visits. Both MEPS and SIPP are panel surveys that collect repeated measures of healthcare encounters.
Access to geographic identifiers is critical to local healthcare disparities research. Although a detailed summary is beyond the scope of this article, brief comments are warranted. Four surveys provide state identifiers, some with caveats. CPS and BRFSS include state identifiers for all respondents. The SIPP has state identifiers but was not intended to produce state-level estimates. The NSCH provides state identifiers; however, state-specific analysis is limited for children identified as AIAN (n = 7 states), Asian (n = 5 states), or NHOPI (n = 1 state). Other geographies may be available through Research Data Centers.18
The United States has rich and varied data resources to study health, healthcare access, and health disparities. We recommend CPS and NHIS as surveys with the most detailed REI identifiers. The NHIS and MEPS-HC have the most comprehensive measures of coverage and access, whereas CPS and BRFSS allow estimates at state and local levels. However, there is no single national survey that will support a comprehensive analysis of healthcare coverage and access for race, ethnicity, or immigrant groups. For example, whereas CPS has detailed REI identifiers and local geography, it is not a health survey and only includes health insurance. Although NHIS has detailed REI identifiers and comprehensive measures of healthcare coverage and access, it is limited by small REI samples and no geography.
Many decisions regarding coverage and access to care are made at state and local levels, thus good data to inform practice and policy and to evaluate the impact for local REI populations are critical. However, only 4 of the reviewed surveys provide state-level identifiers. More effort must be made to collect and release data that allow for state-level disparities analysis. The NHIS releases state-level estimates of health insurance coverage for the 20 largest states.19 However, state estimates are not available for REI groups, which minimizes the utility for healthcare disparities monitoring. The State Health Access Data Assistance Center has a data tabulator20 and also documents states that regularly administer surveys on healthcare coverage.21 Yet, state survey data are not publicly available (except California Health Interview Survey). A new data resource, the American Community Survey, will support state and county-level estimates of health insurance coverage and include REI variables comprising detailed categories of race, tribal affiliation, and country of origin.22 The American Community Survey is an annual survey designed to replace the long-form of the Decennial Census, but it is not a health survey so there are no health or healthcare access data.
Previous authors have called for enhancing federal survey efforts or expanding state-federal partnerships to improve state-level health insurance estimates.10,23,24 We echo this call and further argue that efforts must also address the disparate availability of REI group-specific data. Given federal priorities25 and state-level resource constraints,10 we encourage the exploration of innovative approaches including academic partnerships or privately funded initiatives.7 Good data are essential to developing new knowledge for all groups, including First Americans and New Americans. We hope that calling attention to the limited availability of public use data for healthcare disparities research will foster continued dialogue on improving data availability to support the goal of eliminating health disparities for all.
Research for this article was conducted as part of Health Disparities Research Loan Repayment Award (L60 MD002033-01) from the National Institutes of Health (P. J. Johnson).