Estimation of average treatment effects of having a usual source of care on Pap test use by race/ethnicity: A population-based study

Objectives: Although plentiful studies have documented disparities in Pap test use across races/ethnicities, little is known about how the mechanism of race/ethnicity contributes to this disparity. The relationship between having a usual source of care and preventive service usage has also been documented. This paper investigates the average treatment effects of having a usual source of care, as a potential mechanism, on Pap test use by race/ethnicity.

Method: This is a cross-sectional quantitative study using the 2000, 2003, and 2005 waves of the National Health Interview Survey (NHIS). The target population is noninstitutionalized U.S. female residents age 18 years and older. The outcome variable is Pap test use. The variable of interest is having a usual source of care, which is considered endogenous or confounded by subject characteristics in this analysis. To deal with this selection problem, nearest-neighbor matching was utilized to obtain the counterfactual outcome, which cannot be actually observed, for each observation by matching observations with a usual source of care with those without a usual source of care, and vice versa. Covariates used in the matching process include age, marital status, education, income, health insurance, health status, residence region, immigration status, and year. Bias-corrected matching estimators for average treatment effects proposed by Abadie and Imbens (2002) then were utilized to evaluate the impact of having a usual source of care on Pap test use by race/ethnicity.

Results: A bivariate analysis of having a usual source of care and Pap test use showed that they were statistically significantly associated. Results from the matching estimators indicated that having a usual source of care statistically significantly increased the probability of Pap test use for White, Black, and Hispanic respondents at the 0.001 significance level. However, having a usual source of care did not increase the probability of Pap test use among Asian respondents.

Conclusions: Understanding the mechanism contributing to disparities in Pap test use is the first step to crafting an effective policy to improve Pap test use. However, the results of this study demonstrate that a policy encouraging Asian women to have a usual source of care will not increase their Pap test use, even though previous research has documented a positive association between having a usual source of care and Pap test use. Further research to find other potential mechanisms is essential to form an effective policy targeting Asian women to increase their Pap test use.