Table 2. Sensitivity Analyses for Difference-in-Difference Model for Primary Care Spending and Use Among Adults Eligible for the ACA Marketplace Relative to Adults With Employer-Sponsored Insurance.
| Outcome | Differences-in-differences adjusted estimate | |
|---|---|---|
| Estimate (95% CI) | P value | |
| Limiting to individuals who gained Marketplace coveragea | ||
| Primary care spending, $ | 43.76 (−28.26 to 115.78) | .23 |
| OOP spending for primary care visits, $ | −8.66 (−25.10 to 7.77) | .30 |
| Any primary care visit, % | −0.01 (−0.07 to 0.06) | .87 |
| No. of primary care visits | 0.19 (0.01 to 0.38) | .03 |
| Using different modelsb | ||
| Primary care spending, $ | 47.76 (−59.42 to 154.95) | .49 |
| OOP spending for primary care visits, $ | −2.44 (−29.77 to 24.86) | .61 |
| Any primary care visit, % | 0.05 (−0.01 to 0.11) | .53 |
| No. of primary care visits | 0.27 (0.08 to 0.46) | <.001 |
| Using linear regression model with log-transformed spendinga | ||
| Primary care spending, log ($) | 0.50 (−0.10 to 1.10) | .10 |
| OOP spending for primary care visits, log ($) | −0.05 (−0.59 to 0.47) | .82 |
Abbreviations: ACA, Affordable Care Act; OOP, out-of-pocket.
Differences-in-differences adjusted estimate was estimated from a linear regression model or a linear probability model that controlled for age, sex, marital status, educational level, family income, area of residence, self-reported health status, 10 chronic conditions (asthma, arthritis, diabetes, emphysema, stroke, heart attack, coronary heart disease, other heart disease, angina, and joint pain), and survey year.
Differences-in-differences adjusted estimate was estimated from a 2-part model, a logistic model, or a negative binomial model that controlled for age, sex, marital status, educational level, family income, area of residence, self-reported health status, 10 chronic conditions (asthma, arthritis, diabetes, emphysema, stroke, heart attack, coronary heart disease, other heart disease, angina, and joint pain), and survey year. Using the marginal effects, the mean values of the outcomes for each group in the pre-ACA and post-ACA period were estimated, and then postestimation tests were conducted to estimate the differences in the outcomes between adults eligible for the ACA Marketplace and adults with employer-sponsored insurance before and after the ACA.