|
TRENDSTwelve-Year Trends In Health Insurance Coverage Among Latinos, By Subgroup And Immigration Status
We examine twelve-year trends in the Latino uninsured population by ethnic subgroup and immigration status. From 1993 to 1999, most Latino subgroups, particularly Puerto Ricans, had large decreases in Medicaid coverage. For some subgroups these were offset by increases in employer coverage, but not for Mexicans, resulting in a four-percentage-point increase in their uninsured population. During 2000–2004, Medicaid/SCHIP expansions benefited most subgroups and mitigated smaller losses in employer coverage. However, during 1993–2004, the percentage of noncitizen Latinos lacking coverage increased by several percentage points. This was attributable to Medicaid losses during 1993–1999 and losses in employer coverage during 2000–2004.
LATINOS CONSTITUTE the largest racial or ethnic minority group in the United States and have the highest proportion of people lacking health insurance. In 2004, one-third of the 40.4 million Hispanics residing in the United States were uninsured.1 Also, although Latinos constitute 14 percent of the U.S. population, they account for 30 percent of the nation’s uninsured residents. Hispanics are a heterogeneous population, representing more than fifteen countries of origin. Although Latinos share many common experiences, there are notable subgroup variations in education, socioeconomic status, age, immigration status, and geographic localization that could be masked in analyses that aggregate Latinos.2 Subgroup differences in access to health care also exist.3 For example, in 2002, 36 percent of Mexican Americans lacked coverage, versus 18 percent of Puerto Rican Americans.4 By immigration status, 50 percent of immigrant Latinos who are not U.S. citizens lack coverage, versus 23 percent of U.S.-born Latinos.5 Despite several cross-sectional studies that examine insurance data among Latino subgroups, less is known about longitudinal trends among these subgroups or by immigration status.6 U.S. Census Bureau tabulations show that the number of uninsured Latinos more than doubled, from 6.0 million in 1987 to 13.7 million in 2004.7 However, because the overall percentage of Latinos without coverage has remained fairly stable at around 33 percent, this increase has been attributed to demographic shifts and has not been the focus of much additional investigation. We hypothesize that several events since the early 1990s might have had a differential impact on Latino subgroups. First, the 1996 Welfare Reform Act delinked Medicaid coverage from welfare. Thus, Medicaid was no longer automatically provided to those receiving welfare benefits. The legislation also barred recently arrived immigrants from enrolling in Medicaid and the State Children’s Health Insurance Program (SCHIP).8 Second, several state and federal initiatives aimed at the uninsured, such as SCHIP, might have improved coverage among some subgroups.9 Third, and most notable, is the economy. The mid-to-late 1990s was a period of economic expansion in which the rise in the number of uninsured Americans was more limited than it had been earlier. More recently, the recession at the start of the 2000s and continued declines in employer coverage have resulted in a marked increase in the number of uninsured adults, which might have differentially affected Latino subgroups.10 In this report we use data from the Current Population Survey (CPS) to analyze trends in Latino health insurance coverage from 1993 to 2004. We examine trends among Latino subgroups and by immigration status.
Sample and data source. We analyzed annual data from the March Supplements of the CPS. The March CPS is an in-person household survey representative of the non-institutionalized U.S. population, with over-sampling of Hispanic households. For Spanish-speaking households, bilingual interviewers are used. The coverage rate for the CPS (the extent to which the total population that could be selected for sampling covers the survey’s target population) is about 93 percent for whites versus 83 percent for Latinos.11 We used a hierarchical ranking to assign respondents a primary source of insurance.12 We report data on insurance coverage from 1993 to 2004 (data from the 1994–2005 CPS). Changes to the CPS during this period include the "Uninsured Verification Question" and use of an updated sample framework based on the 2000 decennial census.13 Both of these changes are reflected in our data for 2000.14 Therefore, differences between 1999 and 2000 need to be interpreted with caution, because they likely reflect methodological changes. Variables: race, ethnicity, and immigration status. Ethnic subgroups were created for the three groups for which self-identification was available: Mexican, Puerto Rican, and Cuban. To identify the fourth-largest Latino subgroup in the United States, Dominicans (a relatively recent immigrant group, with most arriving after the 1960s), we used information on nativity and parental nativity. Hispanics who did not belong to one of these four groups are listed as "other Hispanic."15 Latinos were also categorized by immigration status at the person level as U.S.-born Latinos, naturalized citizens, or immigrants who have remained noncitizens.16 The CPS does not ask immigrants about their legal status. Population estimates. Population estimates were derived using weights provided by the Census Bureau, which account for the complex survey design and nonresponse. These weights also account for Census Bureau estimates of year-to-year population changes, including immigration. As an estimate of the sampling error associated with numbers and percentages, we also provide 90 percent confidence intervals for the 2004 data.17
Characteristics. As shown in Exhibit 1  , Latinos were younger, were poorer, and had lower educational attainment than non-Hispanic whites. For example, 61 percent of Latinos had not completed high school, compared with 30 percent of non-Hispanic whites. Among Hispanic subgroups, Cubans were older, while Dominicans had the lowest incomes and the least educational attainment. Approximately a third of Mexicans, Cubans, Dominicans, and "other Hispanics" were not U.S. citizens.
Insurance trends. The number of uninsured Latinos increased from 8.4 million in 1993 to 13.7 million in 2004 (Exhibit 2 ).18 Although non-Hispanic whites constitute the largest group of uninsured people in the United States, the overall number of uninsured non-Hispanic whites in 2004 remained below 1993 levels. During this period, Hispanics accounted for a growing proportion of the uninsured (21.2 percent in 1993; 30.4 percent in 2004), while non-Hispanic whites represented a decreasing proportion (57.4 percent in 1993; 48.0 percent in 2004).
Mexicans, the largest Latino subgroup, had the largest rise in uninsurance (Exhibit 2  ). This represents a small increase in the proportion of uninsured Latinos who are of Mexican origin, from 67 percent in 1993 to 72 percent in 2004. By immigration status, there was little change in the proportion of uninsured Latinos who were noncitizens—50 percent in 1993, versus 53 percent in 2004.
Overall trends in percentage of uninsured Latinos.
For non-Hispanic whites, there was virtually no change in the percentage uninsured from 1993 through 1998 (Exhibit 3
Analyses by Latino ethnic subgroup. Data among the Latino subgroups show variations that differ from the overall trends noted above. From 1993 to 1998, Mexicans and "other Hispanics" saw increases in uninsurance of four percentage points or greater (Exhibit 3 ). In contrast, Dominicans and Cubans tended to have decreases in uninsurance rates. Both of these two groups saw some losses in Medicaid coverage, but such losses were offset by large increases in the percentage with employer coverage. Puerto Ricans had the greatest drop in Medicaid coverage—thirteen percentage points from 1993 to 1999. However, like Cubans and Dominicans, they also experienced a much larger gain in the proportion covered by employer-sponsored insurance, so that uninsurance rates were relatively stable for Puerto Ricans from 1993 through 1999. More recent changes in insurance coverage during 2000–2004 among Latino subgroups track those of the overall Latino population, with most groups having losses in employer coverage of two to three percentage points and similar increases in the percentage covered by Medicaid.19 Thus, the proportion uninsured among the subgroups did not change during 2000–2004. Analyses by immigration status. U.S.-born Latinos had a five-percentage-point increase in uninsurance rates during 1993–1998, which was due to Medicaid losses that were not offset by gains in employer coverage. Since 2000, U.S.-born Latinos have actually seen improvements in coverage, primarily because of recent Medicaid/SCHIP expansions. Immigrant Latinos who became naturalized citizens saw slight improvements in employer coverage from 1993 to 1999, and coverage rates have held steady since 2000. Their rates of Medicaid coverage remained stable and were similar to those of non-Hispanic whites.
In contrast, Latino noncitizens experienced a continuous increase in the proportion uninsured. The increases during 1993–1999 were due to losses in Medicaid coverage, which were disproportionately greater among noncitizens than among U.S.-born Latinos. During 2000–2004, losses in employer coverage for noncitizen Latinos were similar to those of U.S.-born Latinos. However, the gains in Medicaid/SCHIP coverage among Latino noncitizens were less than those seen among U.S.-born Latinos. As a result, Latino noncitizens were the only Latino group to have seen an overall increase in the percentage uninsured from 1993 to 2004 (Exhibit 3
Prior work has examined the reasons for the health insurance gaps among Latino subgroups.20 For all Latino subgroups, the main barrier to being insured is lack of employer-based coverage (Exhibit 3 ). Although employment rates and insurance take-up rates among Latinos and non-Hispanic whites are similar, Latinos are disproportionately more likely to work in situations that do not provide health coverage, including low-wage jobs, small businesses, and certain business sectors such as agriculture and the service industry. Public coverage partially addresses insurance gaps for some groups. For example, Puerto Ricans and Dominicans have traditionally been concentrated in Northeastern states with more favorable Medicaid/SCHIP qualification criteria; thus, they both have higher rates of program participation than other Latino subgroups (Exhibit 3). Unfortunately, the onerous requirements for enrollment and the yearly recertification processes are major limitations of these programs for many otherwise income-eligible Latinos, particularly those with additional barriers such as language. Undocumented, noncitizen Latino immigrants are an especially vulnerable group. Not only are they ineligible for most government insurance programs, but they are also often forced to work in "off-the-books" occupations that offer no health benefits. Impact of welfare reform. In this paper we have advanced the existing body of knowledge by reporting twelve-year trends in health insurance coverage among Latino subgroups and providing data among Latinos stratified by immigration status. Our most striking findings were among Latino noncitizen immigrants, who had a several-percentage-point increase in uninsurance. Some of this was attributable to Medicaid losses, which had begun to occur even before welfare reform. Our data cannot determine whether Medicaid losses after 1996 were attributable to the delinking of Medicaid from welfare and economic expansions or the immigrant provisions of the welfare reform legislation itself. However, our finding that Medicaid losses were greater among noncitizens than among U.S.-born Latinos after 1996 suggests at least some role for the immigrant provisions of welfare reform. Prior studies suggest that the misinformation and fear generated by federal and state anti-immigrant initiatives in the late 1990s might have played a greater role than the statutory provisions of the legislation.21 Impact of economic ups and downs. We also found that gains in employer coverage during the economic prosperity of the 1990s were smaller for noncitizens than for U.S.-born Latinos and that losses in employer coverage for this group were greater during the recession of 2000–2002. This suggests that during economic downturns, noncitizens suffer more and yet share less of the prosperity during economic expansions. Study limitations. In addition to the changes in survey methodology and inability to identify the undocumented Latinos, other caveats apply. CPS insurance estimates approximate the number uninsured at a specific point in time. Studies that examine trends in spells without coverage among Latino subgroups are also needed. However, the sample size of the existing national longitudinal surveys does not permit detailed Latino subgroup analysis. Second, our data do not include people who are underinsured, a group that is also vulnerable to impaired access to care. Lastly, despite the CPS sampling of more than 230,000 people, we have only a limited ability to provide stratified subgroup data such as by age and income. Policy implications. From a policy perspective, our study has two main findings. First, for policy analysts, our study again reiterates that Latinos should not be considered as one uniform group. To the extent possible, researchers need to consider differences among Latino ethnic subgroups and immigration status. Second, the continued rapid growth in the uninsured Hispanic population and the disproportionate growth in uninsured noncitizen immigrants necessitate targeted attention. One proposal that would address these existing barriers and provide health insurance to all uninsured Latinos is single-payer national health insurance.22 Such a program would provide permanent coverage for all U.S. residents, including those who are not yet citizens. By delinking coverage from employment, it would overcome the failings of an employer insurance system that covers less than half of all Latinos and that differentially affects certain Latino subgroups during economic swings. In addition, automatic and irrevocable coverage would eliminate existing barriers associated with current Medicaid/SCHIP programs, such as administrative and statutory regulations for enrollment and re-enrollment. Further, Medicaid disproportionately provides services for less politically influential groups such as minorities and the poor and has always been a relatively easy target for funding cuts. It would be much more difficult for policymakers to enact legislation negatively affecting a program that covers all U.S. residents. An added benefit of national health insurance would be that it has repeatedly been shown to be the only insurance proposal that would cover all U.S. residents yet at the same time decrease overall U.S. health spending.23 Although national health insurance is often labeled a radical proposal lacking any meaningful political support, the National Health Insurance Act, H.R. 676, currently has more than seventy congressional cosponsors, including nine of the twenty-one members of the congressional Hispanic Caucus.24 At a recent congressional briefing on the Latino insurance crisis, several key Latino health policy leaders expressed support for H.R. 676.25 The immediate beneficiaries of such a program would include the nearly fourteen million uninsured Latinos as well as the other thirty-one million non-Hispanics in the United States who also lack coverage.
Sarita Shah is an assistant professor of medicine in the Division of General Internal Medicine, Albert Einstein College of Medicine, in the Bronx, New York. Olveen Carrasquillo (oc6{at}columbia.edu) is an associate professor of medicine and health policy and management at Columbia University Medical Center in New York City. This work was supported by grants from the Health Resources and Services Administration (HRSA) (D08-PE-50011-08) and the National Center on Minority Health and Health Disparities (MD00206 P60). The authors thank Douglas Gould and Bisundev Mahato, who served as the project data analysts, and Andrea Guerra, who helped prepare the data tables during the initial phases of the project.
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Study Data And Methods
Study Results
