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

RESULTS

Summary of Dependent and Independent Variables

Table 1 summarizes all weighted dependent and independent variables. The final disposition of each potential predictor is given in the far right column, indicating whether the variable was kept in final models, deleted because of multicollinearity, or deleted because of interim model F values < 1. Variables deleted for F values < 1.0 included the log of family income, "other" race, public health services, bed supply, HMO market penetration rate, and the percentage of the sample covered by public insurance.

Variation in Access and Social Capital

The variation among the 22 MSAs in the weighted percent of persons with access problems was 11.8 to 19.9% (MSA level mean = 16.4, SD = 2.2). The six components of social capital were intercorrelated with a Cronbach alpha of 0.76. The variation among the 22 MSAs in community social capital was 38.7 to 56.4 (MSA level mean = 49.6, SD = 5.1). The three MSAs with the lowest social capital were Las Vegas, Miami, and West Palm Beach. The three MSAs with the highest social capital were Baltimore, Columbus, and Denver. Social capital and access problems at the MSA level correlated at -0.25.

Results of Hierarchical Models

Results of the models estimated to predict access problems are summarized in Table 2. Column A includes only the individual level predictors. Reported access to care problems are associated with females, non-Hispanic ethnicity, lower relative family income, poor child health ratings, younger age, higher education, poorer physical and mental health, HMO enrollment, and lack of health insurance.

Column B summarizes model results when health sector variables are added to the individual predictors. All individual level predictors remain unchanged, and two community-level health care system attributes are significant. Greater access problems are associated with fewer HMO plans and also with more public health--community collaborations. This model represents a significant improvement in fit relative to the column A model (improvement in Satterwaite chi-square = 18.9, df = 0.47, p < .01).

Column C summarizes model results when general community social capital is added to the individual and health sector predictors. Other individual and community effects remain unchanged. The social capital effect is significant; higher social capital is related to fewer reported access problems. Coefficients for individual and community variables change very little relative to model A. Model C relative to model B improves significantly (differences in Satterwaite chi-square = 9.7, df = 0.11, p < .01).

DISCUSSION

The results support the hypothesis that social capital is related to improved health care access. We speculate this occurs because social capital (i.e., trust among citizens, reciprocity, and civic engagement) likely improves the functioning and efficiency of community social institutions (Putnam 1993). The current research extends prior work by demonstrating that the benefits of social capital may extend from local government institutions in general to local health care institutions in particular.