II. Methods and Data

To assess whether race/ethnicity continues to be a factor associated with where an individual obtains medical care, data are examined using two indicators stratified by insurance coverage: 1) the proportion of people with no usual source of care, and 2) the proportion of people whose usual source of care was an office-based provider, a hospital clinic or an OPD, or a hospital ER. We examine the likelihood of having a hospital-based provider as a usual source of care using descriptive and logistic regression analysis. The analysis compares the sources of care of whites, African Americans, and Latinos under age sixty-five. Findings are examined separately for children under age eighteen and adults ages eighteen to sixty-four because these groups differ greatly in their health needs and health insurance coverage.

This study analyzes data from the 1996 Medical Expenditure Panel Survey (MEPS), the third in a series of surveys conducted by the Agency for Healthcare Research and Quality of the U.S. Department of Health and Human Services. The MEPS is a nationally representative survey that collects detailed information on the financing and use of medical care by individuals and families in the United States. Data are collected during multiple interview rounds. The MEPS full-year consolidated data file is used for this analysis. Detailed information on the survey is available.   The unweighted sample size for this analysis was 18,603 persons under age sixty-five. Estimates presented in this analysis are weighted to represent the non-institutionalized U.S. population. The statistical package SUDAAN (professional software for SUrvey DAta ANalysis) was used to obtain weighted population estimates and standard errors. This package calculates weighted estimates to correct for the complex multistage sampling design of the MEPS.

Information on the usual source of care of each family member is obtained from the Access to Care Supplement of the MEPS. The usual source of care is defined from a question that asks: Is there a particular doctor's office, clinic, health center, or other place that (PERSON) usually goes if he/she is sick or needs advice about his/her health? Any family member who has a particular person or place they usually go for care or advice is considered to have a usual source of care. Three categories of a usual source or site of care were created: 1) office-based provider; 2) hospital clinic or OPD; and 3) ER. The categories are self-explanatory *21 except office-based provider, which reflects the health care system's evolving assortment of financing and service delivery arrangements. Included within office-based providers are physicians in solo practice, physicians or other providers in larger group practices, health maintenance organizations or other types of managed care plans, as well as private and public community health clinics. MEPS, unlike its predecessor the National Medical Expenditure Survey (NMES), categorizes persons whose usual source of care is a community clinic or health center as having an “office-based” provider.

Respondents' races/ethnicities are based on self-reported information. Three mutually exclusive racial/ethnic categories were created: white, African American or black, and Latino or Hispanic. All persons of Hispanic origin, regardless of race, are classified as Latino or Hispanic. People reporting their racial/ethnic identity solely as Asian, American Indian, Alaska Native, or “other” are excluded from this analysis. This decision was made, in large part, because there are too few individuals in MEPS who identify themselves as Asian, American Indian/Alaska Native, or “other” for reliable population-specific estimates.

Insurance coverage, one of the major independent variables in this study, is defined based on a series of questions. Since family members can have health coverage from multiple sources, a hierarchical variable was created to define insurance coverage that gave priority to private coverage (employment-based or privately purchased) and then Medicaid. Individuals without private or Medicaid coverage were classified as uninsured. Individuals with “other sources of public coverage,” such as CHAMPUS or Medicare, were excluded from the analysis. Their numbers in the sample were too small for meaningful interpretation of the patterns of care.

Separate logistic regression models for children and adults are used to assess the effects of race/ethnicity on the likelihood of having a hospital-based provider as a usual source of medical care. In addition, two other models are run for each age group. One model evaluates the effects of including persons who identify the ER as a usual source of care, and the other evaluates the effects of excluding this population. Although the ER is not an appropriate usual source of care, we included this population in one of the models since some respondents identify the ER as serving this purpose. Finally, we also tested an interaction term for race and insurance to assess whether the findings observed by race/ethnicity are consistent across all of the insurance categories.