Adjusted rates in diabetes-related prescriptions pre-ACA and post-ACA Medicaid expansion between expansion groups
| Prescription | Expansion | Non-expansion |
| Number of states | 25 | 19 |
| Metformin | ||
| Pre-ACA, adjusted rate per 100 enrollees | 17.00 | 8.18 |
| Post-ACA, adjusted rate per 100 enrollees | 16.82 | 7.66 |
| Absolute change pre-ACA to post-ACA | −0.18 | −0.52 |
| Absolute DID (95% CI) | 0.33 (0.323 to 0.344) | Ref. |
| Relative change pre-ACA to post-ACA | −1.1% | −6.3% |
| Relative DID (95% CI) | 1.056 (1.055 to 1.057) | Ref. |
| Long-acting insulin | ||
| Pre-ACA, adjusted rate per 100 enrollees | 7.09 | 4.90 |
| Post-ACA, adjusted rate per 100 enrollees | 6.79 | 4.49 |
| Absolute change pre-ACA to post-ACA | −0.30 | −0.41 |
| Absolute DID (95% CI) | 0.11 (0.098 to 0.113) | Ref. |
| Relative change pre-ACA to post-ACA | −4.3% | −8.3% |
| Relative DID (95% CI) | 1.044 (1.043 to 1.046) | Ref. |
| Novel agents | ||
| Pre-ACA, adjusted rate per 100 enrollees | 1.20 | 0.58 |
| Post-ACA, adjusted rate per 100 enrollees | 1.28 | 0.58 |
| Absolute change pre-ACA to post-ACA | 0.08 | 0.00 |
| Absolute DID (95% CI) | 0.08 (0.079 to 0.086) | Ref. |
| Relative change pre-ACA to post-ACA | 7.1% | 0.5% |
| Relative DID (95% CI) | 1.066 (1.061 to 1.071) | Ref. |
Expansion states included 25 states (Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Hawaii, Illinois, Iowa, Kentucky, Maryland, Massachusetts, Michigan, Minnesota, Nevada, New Jersey, New Mexico, New York, North Dakota, Ohio, Oregon, Rhode Island, Vermont, Washington, and West Virginia) and the District of Columbia. A total of 19 non-expansion states included Alabama, Florida, Georgia, Idaho, Kansas, Maine, Mississippi, Missouri, Nebraska, North Carolina, Oklahoma, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, Wisconsin, and Wyoming. During our study period, 6 states (New Hampshire, Pennsylvania, Alaska, Indiana, Louisiana, and Montana) expanded later and were excluded from this analysis. For each diabetes medication category, we performed generalized estimating equation (GEE) Poisson regression models with the following variables: an indicator denoting expansion status (state expand Medicaid vs did not expand), an indicator for period (pre-ACA vs post-ACA), the interaction terms between expansion status and period, state unemployment rate and state diabetes prevalence. For all GEE models, we specified a Poisson distribution with a log link, an offset equal to log(Medicaid enrollees that year) and assumed an autoregressive correlation matrix of degree 1 to account for the autocorrelation of yearly observations within states. Rates were adjusted for state-level unemployment rate and diabetes prevalence.
ACA, Affordable Care Act; DID, difference-in-differences.