Evaluation of the impact of terminating the reimbursement of a therapeutic class of drugs on the French health system: The example of symptomatic slow-acting drugs for osteoarthritis (SYSADOAs)

Introduction The French social security has had a balanced budget as a recurring objective for several decades, with a specific target on the National Health Insurance (NHI) branch. Over the past several years, efforts have intensified, with a continuing policy on decreasing drug prices, developing ambulatory care to limit hospitalizations or reducing the reimbursement of some specialists. Decreasing the reimbursement rate of drugs reduces, as a direct effect, the prescription and delivery of the targeted drugs. However, the substitution effect in favor of still reimbursed drugs should not be forgotten, because it could minimize the intended effect in economic terms as well as in terms of public health. The recent French decision to terminate the reimbursement of Symptomatic Slow-Acting Drugs for Osteoarthritis (SYSADOAs) is an interesting case study for the impact of terminating the reimbursement a drug from a collective perspective. The aim of the present study was to investigate prescription substitutions in favor of analgesics and non-steroidal anti-inflammatory drugs (NSAIDs), both in terms of volumes and costs, as well as the occurrence of their potentially associated adverse effects (AEs) using a methodology based on access to national administrative databases. Methods To quantify these effects in the most exhaustive manner, French medico-administrative databases were used (EGB - general sample of beneficiaries; PMSI - program for medicalization of hospital information systems). Three study periods were considered to cover a potential historical bias: the primary period was defined around the date of reimbursement termination (2015/03/01), considering one year both pre and post cutoff; the secondary period was defined in the same manner around the date of the reimbursement rate decrease (35% to 15% on 2011/12/01); a control period was considered with a cutoff date (2009/03/01) chosen to reduce seasonal variations impact as comparator of the primary period. We defined the population on each of these periods on two criteria: they were at least 40 and had had at least one SYSADOA delivery in the year prior to the cutoff. We considered subgroups on the delivery persistence and prevalence. We performed comparisons for the following events, quantified in volumes and in costs: NSAID deliveries, analgesic deliveries, hospitalizations for cardiovascular or gastric causes, or for renal insufficiency, and surgery for arthritis. Results The analyses concerned respectively 19,345, 20,066, and 16,200 patients in the control, secondary and primary periods. Among the patients, 4345 were present in all 3 periods. The classification of patients according to delivery persistence remained stable over the three periods. We observed a decrease of incident patients. The volume of NSAID deliveries decreased over the three periods from a total of 40.5 (± 76.3) defined daily dose (DDDs) per patient in 2008 to 29.6 (± 66.4) DDDs in 2015. The number of analgesic DDDs increased constantly over the three study periods, from 70.2 (± 108.9) DDDs in 2008 to 76.9 (± 123.1) DDDs in 2015. Regarding the costs, there was an overall decrease over the three study periods of the amount reimbursed by NHI for all deliveries and for hospitalizations. The effects being quantitatively comparable in all three periods, it was not possible to conclude from these data that terminating the reimbursement of SYSADOAs had an impact on the French NHI. Conclusion The information available in medico-administrative databases makes it possible to study the impact of a change of policy in the reimbursement schemes by allowing the constitution of large representative cohorts at the national level. However, our results underline how the measured effects were restricted to those impacting NHI: OTC deliveries or out-of-pocket expenses were not evaluable with these data sources alone. Thus, the impact in terms of public health can only be partially measured.