Retrospective database analysis of the impact of prior authorization for type 2 diabetes medications on health care costs in a medicare advantage prescription drug plan population

摘要

Background: Health plans and pharmacy benefit managers have implemented utilization management strategies for newer type 2 diabetes mellitus (T2DM) medications to control pharmacy expenditures. Little is known about the impact of utilization management strategies on overall health care costs and subsequent use of T2DM medications among members who request, but do not receive, a T2DM medication requiring prior authorization (PA). Objective: To examine the relationship between the receipt of a T2DM medication requiring PA, health care costs, and subsequent treatment for T2DM. Methods: A retrospective cohort study using pharmacy, medical, and laboratory claims data was conducted among Medicare Advantage Prescription Drug plan members with a denied claim for a T2DM medication requiring PA (sitagliptin, a dipeptidyl peptidase-4 inhibitor [DPP-4i], and exenatide, an incretin mimetic) between January 1, 2008, and June 30, 2009. Subjects were required to have 12 months of continuous enrollment both before and after the index date. The entire study period was 24 months in duration, including a 12-month pre-index and 12-month postindex period. Three cohorts were identified: 1 that received a medication requiring PA (denied claim, subsequent fill) and 2 nonfilling control groups. Both control groups requested a medication requiring PA, as evidenced by the denied claim, but neither received the medication, either because the medication was not authorized or the member chose not to fill. Claimsbased estimates were used to infer whether the individual likely met the criteria for PA, with 1 control group designated as having met the claimsbased criteria (qualifying nonfilling cohort) and the other not having done so (nonqualifying nonfilling cohort.) The primary endpoint evaluated was the relationship between PA medication fill status and plan-paid costs (medical [including laboratory] and pharmacy) over the 12-month postdenial period, with generalized linear models adjusting for key covariates including demographics, concomitant medications, pre-index costs, preindex adherence, and comorbidities. The secondary endpoint of T2DM medication use (post-denial) among the 2 nonfilling control groups was also evaluated. Results: There were 1,728 members identified who received medication for T2DM requiring PA (the received authorization cohort) and 2,373 who did not (606 qualifying nonfilling cohort; 1,767 nonqualifying nonfilling cohort.) Cohorts were similar with regard to age and gender, but the nonfilling cohort had more comorbidities. Total unadjusted plan-paid 12-month costs were lowest among the received authorization cohort ($11,739), slightly higher ($11,980) for the qualifying nonfilling cohort, and notably higher for the nonqualifying nonfilling cohort ($12,962), although no differences were statistically significant. After adjusting for key covariates, the difference between the nonqualifying nonfilling cohort ($11,980) and the received authorization cohort ($11,729) was statistically significant (P = 0.034). Large differences in plan-paid medical costs ($10,127 for the nonqualifying nonfilling cohort vs. $8,192 for the received authorization cohort) appeared to drive the overall cost totals and were significant in both the unadjusted (P = 0.005) and adjusted models (P < 0.001). Pharmacy costs were significantly lower for the nonqualifying nonfilling cohort in the adjusted model and for the qualifying nonfilling cohort in both models (all P < 0.001), but the lower pharmacy costs were not offset by the higher medical costs. In examining the use of medication for treatment of T2DM following the denied claim, 10.6% of the qualifying nonfilling cohort and 13.4% of the nonqualifying nonfilling cohort added another oral therapy, 10.2% and 5.8% added insulin, and 11.9% and 7.1% had treatment intensification, respectively. More than half (56.1%) of the qualifying nonfilling cohort, but only 32.1% of the nonqualifying nonfilling cohort