Access to health care and community social capital

Health Services Research, Feb, 2002 by Michael S. Hendryx, Melissa M. Ahern, Nicholas P. Lovrich, Arthur H. McCurdy

Sample selection occurred primarily through random-digit dialing, supplemented with field samples to represent persons who do not have telephone service. Families within households included the respondent, their spouse, children under age 18, and children age 18-23 who were full-time students. All interviews were conducted by telephone (including cell phones used in the field for persons without telephones) and were conducted in English or Spanish. During the course of the interview, information was obtained on all adults in the family as well as one randomly selected child.

Persons in the 22 MSAs in this study numbered 19,672. The final sample represents a 65 percent response rate. No information was collected from families that refused to participate, and consequently, the potential for bias from survey nonresponse could not be ascertained. However, person-level weights were poststratified to account for nonresponse based on age, sex, race or ethnicity, and years of education. Estimates reported in this article were weighted to be representative of the noninstitutionalized civilian U.S. population as well as representative of each of the 60 sites, using the weights created in the CTS study.

Individual level personal characteristics treated as independent variables in this study include sex, age, race or ethnicity, income, number of persons in the household, years of education, insurance coverage (yes/no), HMO enrollment (yes/no), and self-reported health status. Income was measured in two ways: by the log of reported family income and by calculating a measure of relative income, which was family income divided by the mean family income for the MSA in which the person resided. Health status was measured by two variables for adults: the SF-12 Physical Composite Score and the SF-12 Mental Health Composite Score. For children, health status was assessed by the proxy's rating of general health (SF-12 item 1; a higher score indicates worse health). We also used the Household Survey to create a community-level variable, the proportion of the community covered by Medicaid or other state public insurance. Additional descriptions of the methodology of the CTS Household Survey can be found elsewhere (Cunningham and Kemper 1998; Kemper, Blumenthal, Corrigan, et al. 1996).

The NIHCM (1999) Data Source served as our source for information on 1996-97 MSA level health care supply and managed care characteristics. Five variables were collected from this source: nonfederal primary care physicians per 100,000 persons (primary care defined as family and general practice, internal medicine, obstetrics/gynecology, and pediatrics), nonfederal, nonprimary care specialist physicians per 100,000 persons, hospital beds per 100,000 persons, number of HMO plans, and HMO market penetration (defined as total HMO enrollment divided by total MSA population). HMO variables were obtained by the NIHCM from Interstudy Competitive Edge 8.1. Hospital bed supply was obtained from Hospital Statistics (Healthcare InfoSource, Inc., a subsidiary of the American Hospital Association). Physician supply was obtained from Physician Characteristics and Distribution in the US. (American Medical Association).