Factors associated with off-label (OL) drug use in oncology: The role of cost and financing in a universal healthcare system.

Abstract

e18825 Background: In recent years, OL use in oncology has become widespread, with estimates ranging up to 50-75% of all prescriptions. Most OL use is not publicly reimbursed within universal healthcare systems. We aimed to characterize the financing sources of OL use and to identify predictors for forgoing such treatment. Methods: We studied 708 oncology OL prescription requests submitted for approval to the Institutional Drug Committee in Rabin Medical Center, a large tertiary center in Israel, between January 2016 and December 2018. We included only requests for patients that were alive for more than 60 days following prescription approval. For each indication we extracted the level of supporting evidence at the time of prescription (sufficient, limited, inadequate). We also examined patients’ disease and sociodemographic properties, treatment costs, and financing sources. We used univariable logistic regression to ascertain these variables’ effects on actual OL drug initiation. We then used multivariable logistic regression to explore predicting factors for drug initiation. Results: The median monthly cost of a planned OL treatment was ILS39,928 (approximately US$ 11,500). Approximately one third (31%) of the treatments did not have a financing source at the time of request approval. The primary financing sources were patient access plans (30%) and private health insurance (25%). Of 708 approved OL requests, only 583 (82.3%) treatments were initiated. The proportion of requests that were eventually initiated was higher in the metastatic versus adjuvant setting (84.9% vs. 76.5%; p= .008). The estimated median OS for metastatic patients was 9.9 months, significantly higher in patients that did initiate treatment versus patients that did not (10.4 vs. 7.2 months; p= .048). Although not significant, median costs of initiated treatments were higher compared to treatments that were not (ILS41,686 vs. ILS24,670). Prescriptions for the metastatic setting and immunotherapy as well as ones with sufficient evidence had higher odds for treatment initiation (OR = 1.73; p= .007, OR = 1.69; p= .013 and OR = 1.62; p= .016). Prescriptions with limited supporting evidence and with no planned financing source had lower odds for treatment initiation (OR = 0.59; p= .012, OR = 0.37; p< .0005). A multivariable logistic regression showed that if no financing plan was in place at the time of request, there was a 2.5 times higher likelihood of not initiating the treatment (OR = 2.5; p< .0005). Conclusions: While OL recommendation is widespread and institutional approval is granted, a substantial proportion of treatments are not initiated. Although cost was not associated with treatment initiation, having a planned financing source was a strong predictor for OL treatment initiation. This study elucidates the financing sources of OL treatments in a universal healthcare system and identifies access barriers.