Child Healthcare Disparities: Findings from the Annual Report and Future Directions

The Annual Report on Child Health is published in this issue of Academic Pediatrics, and its focus this year is on healthcare disparities¹. Child health and health care disparities can be defined as an inequitable difference in health or healthcare based on demographic characteristics such as race, ethnicity, generational status, socioeconomic position, and geography. For these differences to be labeled as disparities, they must be inequitable, systematic, and potentially avoidable². The current paper, which reports on analyses culled from the Medical Expenditure Panel Survey (MEPS) and the Healthcare Cost and Utilization Project (HCUP), breaks down child health care quality indicators based on race/ethnicity, income, and insurance status, and thus provides an interesting look at the interactions among these distinct but interrelated categories.

In the United States as well as other industrialized countries, minority children and families are disproportionately represented in lower socioeconomic strata, so analyses such as these which stratify based on SES or insurance status (often a proxy for SES) provide important information. Do poor minority children (or more importantly, poor minority children of specific racial or ethnic heritages) do worse than poor majority children? If so then evidence for disparities based on race/ethnicity exists. The analyses in this report are interesting in that no distinct pattern of disparity based on race/ethnicity is seen within income or insurance status group, and some of the differences seen are somewhat unexpected. For example, compared to low income white children, low income black children had better profiles with regard to effectiveness of care (e.g., greater advice regarding exercise, healthy eating, use of car seats, and provider time alone with teens), overall satisfaction, and patient safety. While low income Hispanic children, compared to whites, had worse profiles regarding some measures of effectiveness of care, timeliness of routine appointments, and higher asthma admissions and hospital infections, they had better overall satisfaction ratings and lower rates of diabetes complication admissions, accidental punctures and decubitus ulcers among hospitalized patients. Overall, no clearly consistent pattern of racial/ethnic disparity is seen, and some of the findings are contrary to what clinical experience and limited past data would have suggested.

I think that these disparate results highlight some of the weaknesses of using large nationally representative data sets in the analysis of complex health care issues. For example, while there’s “strength in numbers”--especially large sample sizes, many of the findings, although statistically significant, are clinically and epidemiologically irrelevant (e.g., a statistically significant difference between 0.35 and 0.38 on a 0 to 1 continuum). Secondly, although these surveys measure healthcare quality based on the IOM-conceptualized domains of effectiveness, patient-centeredness, timeliness, and safety, how these complex domains are operationalized in large surveys such as these is (necessarily) limited. Third, what’s also limited is the categorization of race/ethnicity. MEPS uses the Office of Management and Budget standards for the classification of race and ethnicity. These standards may be good for census-type analyses, but to tease out the complex relationships between the multiple factors contributing to healthcare disparities, they leave a lot to be desired. For example, according to OMB standards, individuals are categorized as “Asian-Pacific Islanders” if they have origins from the Far East, Southeast Asia, Indian subcontinent, or the Pacific Islands.³ In this scheme, individuals from Pakistan, China, India, Vietnam, Sri Lanka, Cambodia, Japan, and Afghanistan would all be categorized together. To lump such disparate groups together without recognizing the extreme differences in culture, immigration context, relative social position, and interactive styles with healthcare may result in masking significant disparities that might exist for any of these groups. Likewise, lumping together American-born African Americans with West Indian/Caribbean and African immigrants, and Hispanic subgroups (Mexican American, Puerto Rican, Cuban American, Dominican, Central American, etc.) together effectively homogenizes groups that may have significantly different interactions with the healthcare system.

All this is not the fault of the authors of this paper. They can only use the data which are made available to them. They note the important finding that “quality of care is less than optimal for many children, regardless of race/ethnicity, income, and insurance.” Some of the methodological issues touched upon above may contribute to their conclusion that “some of our findings were seemingly inconsistent and others were simply unexpected, with minorities having better quality of care than their white counterparts.” For future research to inform clinical care and public policy, the following suggestions are provided, based in part on the recommendations of the recent conference “Starting Early: A Life Course Perspective on Child Health Disparities: Developing a Research Action Agenda”²:

1. Findings from large national studies need to be placed in context. Are different outcomes in quality indicators such as effectiveness and timeliness due to structural barriers in the healthcare system, individual issues such as health education and literacy, differences in health beliefs and behaviors, or other individual, interpersonal, or institutional factors? We need to learn the story behind the numbers before we can successfully intervene. This will require studies using mixed methodologies (qualitative and quantitative), as well as multidisciplinary collaboration.

2. Traditional census-type categorizing of race and ethnicity needs to be replaced in our research with more meaningful approaches. Categories of convenience can no longer be accepted. A recent IOM report calls for the collection of “granular” ethnicity—a detailed and comprehensive breakdown of all locally relevant ethnicities—in addition to the OMB classifications⁴. Granular data can always be re-aggregated during analysis, but if you don’t obtain this level of specificity during data gathering, you can re-create it during analysis. In addition, self reported race/ethnicity data should be the standard. In the HCUP dataset used in the present report (generated by hospital claims data), race/ethnicity data is based on information recorded by a hospital admissions clerk, and therefore may be self report or based on observation. In 2010, obtaining race/ethnicity data based on third party observation is cause for concern.

3. When working with immigrant groups, generational status and acculturation status provide important dimensions of intra-cultural variability with regard to health and healthcare disparities⁵. Generational and acculturation differences need to be included in analyses in the same way that the authors of the present report stratified their data based on income and insurance status.

4. Finally, while descriptive studies that illustrate the areas where child health and healthcare disparities exist are informative and important, we’re getting close to the point where we don’t need many more studies that tell us what we already know: that poor children and minorities often have worse outcomes. The present paper is interesting, and its unexpected results may lead to future hypothesis-driven studies that help find the reasons behind the results. I hope that the field of health disparities research will soon enter its next “developmental” phase: where we spend less effort describing the problem and more on understanding the processes and mechanisms through which they become manifest, and in identifying and testing promising interventions to eradicate them.