*Names in bold indicate Presenter
The data come from a commercial database of prescription and institutional records generated from computer systems that exist in pharmacies and medical institutions to record transactions and allow manufacturers to track their product sales. The ProMetis Market Profiler database contains prescription information and medical institutional records on nearly 80% of the US population. Prescription data fields include product, date written, date filled, prescription writer’s core-based statistical area (CBSA), payment type, days supply, quantity, fill type, and refills. With this rich prescription dataset we are able to examine prescription contraception consumption at a product level over time. These data have been used in prior published research (Ketcham and Simon, 2008).
The population most affected by this mandate is women who are covered by private health insurance, most of which is provided through an employer. This population of women is likely to have higher education and income levels relative to women who do not have insurance or who are insured through public insurance (DeNavas-Walt et al. 2004). Women who are more highly education and have higher incomes are more likely to use contraception, and more effective methods (Martinez et al. 2012). The predicted effect of the policy is then ambiguous. Economic theory suggests insured women may increase their contraception consumption in response to a decrease in out-of-pocket expenses. Or this policy may lead women to substitute higher copayment, more effective methods for lower copayment methods.
Our empirical strategy exploits the predicted differential responses of long-acting reversible contraceptive (LARC) methods to other contraceptive methods. Since LARC methods typically have higher copayments and much longer durations of effectiveness, their demand may be more likely to respond to anticipated future price reductions and more sensitive to the policy change. Preliminary findings suggest the policy change does, in fact, have a stronger effect on the consumption of LARC prescriptions than non-LARC methods, in part due to a decrease in non-LARC methods. Approximately two-thirds of this effect is due to refilled prescriptions, with one-third due to new prescriptions. This effect represents the difference between two groups affected by the policy, and should be interpreted as a lower bound estimate of the total treatment effect of the policy.