The policy making and implementation process is poorly understood by many quantitative health services researchers leading to potential threats to the validity of some studies. There is an extensive qualitative literature on the policy making process and a distinction arises between those that decide on a policy (policy makers) and those that implement a policy (policy actors). As multiple actors implement a policy there are multiple concerns raised including heterogeneity of effect among actors, evolving policies as different policy actors interact amongst themselves, variable implementation times among actors and spillover effects as actors engage with other, non-intended groups. When measuring health policies, each of these factors needs to be considered when designing studies and drawing conclusions to properly inform policy makers of policy effects.
Crowd-out in health insurance occurs when individuals would have private insurance but for the existence of a public insurance program. Policy makers concerned with crowd-out will put barriers to enrollment in public plans which may discourage the neediest people, who the programs are intended to help, from enrolling. Past estimates of children’s crowd-out during expansions in eligibility to Medicaid and the Children’s Health Insurance Program have varied widely and are national in scope. I estimate state-specific levels of crowd-out using a regression discontinuity analysis to estimate crowd-out independent of an expansion for children near the eligibility threshold. I find that among states with similar eligibility levels, there is significant variability among crowd-out levels.
To evaluate how crowd-out levels may change over time, I estimate crowd-out in Ohio from 2004-2012. I find that crowd-out levels were not constant over time and decreased during these years.
Caps on noneconomic damages are viewed by some as a means of lowering the cost of healthcare. A potential unintended effect of these caps is that people with meritorious claims may not be able to find appropriate legal representation and will not be compensated for their injury. A challenge in measuring the effect of these tort reforms is that different states implement the same general policy differently and have different statutes of limitations, meaning the effective date for a policy will vary. By applying state-level, interrupted time series and matched pair difference-in-difference designs, and adjusting for the variable implementation times, I estimate state-specific effects. I find that facially similar noneconomic damage cap statutes led to significantly different effects on the rate of malpractice settlements, with many states seeing no effect from the policy.
Changes in Medicare reimbursement policies are known to affect the behavior of providers serving the Medicare population, but less is known about how such policies may spillover onto the non-Medicare population. I evaluate the change in the rate of bariatric surgeries in the United States after the Centers for Medicare and Medicaid Services passed several National Coverage Decisions using an interrupted time series design. I find a significant decrease in the rate of bariatric surgeries for the Medicare and non-Medicare populations. The timing and magnitude of these changes were nearly identical for both populations.