OBJECTIVES: To assess the impact of suboptimal clinical evidence on funding recommendations

METHODS: Using published pan-Canadian Oncology Drug Review (pCODR) reports, 111 unique decisions spanning July 2011 to April 2018 were identified. Thirty-nine independent variables were consistently reported covering clinical, economic, and patient inputs. Quality of trials providing clinical evidence was measured regarding randomization, phase, design (non-inferiority vs. superiority), and presence of control arm. Logistic regression was used to analyze the impact of trial quality in terms of: list vs. do not list (DNL); full vs. conditional list.

RESULTS: Of the 111 decisions, 12 were full list, 74 conditional list, and 25 DNL. Evidence of improved survival and/or response was the main driver of list vs DNL (odds ratio [OR]=77; p<0.001); along with unmet need (OR=9.4; p<0.01) and worse safety profile (OR=0.09; p<0.01). The summary attribute for trial quality showed that submissions supported by high-quality trials tend to be listed (OR =5.6; p=0.06). Subcomponents of trial quality did not demonstrate strong association with the list decision (p>0.15). Among drugs showing improved clinical benefit based on a single trial, listing possibility was similar regardless if phase II or III (86% vs. 85%), although sample was smaller for phase II (14 vs. 61). Trial quality was not a statistically significant factor in full vs. conditional listing, where the driving factor was cost-effectiveness (OR=36; p<0.001). In simulated cross-validation, the model predicted correctly 81% of decisions (specificity=0.83, sensitivity=0.75).

CONCLUSIONS: Using suboptimal clinical evidence reduces the probability of listing but doesn’t affect the funding recommendation which is dependent on efficacy. Other factors, such as improved clinical benefit and unmet need, are also important. Whether this applies to other indications beyond oncology or to other agencies needs to be elucidated.