e18179 Background: Drug advertisements marketed to consumers and healthcare providers have been the subject of recent debate, and the efficacy endpoints that underpin these advertisements may sway perceptions of drug efficacy. Methods: Using the AdPharm database we extracted data for oncology drug advertisements printed or aired between March 1, 2017 and September 1, 2018. Our primary objective was to evaluate endpoints from television and print advertisements and endpoints from the corresponding published papers. A secondary objective is to identify the frequency of drug advertisements that are aired or printed while overall survival data is immature. Results: We included 74 advertisements for 34 unique oncology drugs and 22 unique indications. Print ads were most common (n = 66) and most print ads were targeted to healthcare providers (n = 55, 83.3%). The 74 advertisements were underpinned by 48 published clinical trials. The primary endpoint of the 48 underpinning trials was most often PFS (n = 25), followed by OS (n = 11) and ORR (n = 7). Primary endpoints of underpinning trials were not reported in 8 (8/74, 10.8%) advertisements. Statistically significant endpoints were reported more often than nonsignificant endpoints (RR 1.42; 95% CI, 1.26 - 1.60). Thirty advertisements (42.3%) ran while OS data was immature. When OS and PFS were statistically significant, they were reported at a significantly higher rate than when not reported (each P < .001). ORR was not preferentially reported based on its statistical significance (RR 1.33; 95% CI, .94 - 1.87). Conclusions: Oncology drug advertisements appear to preferentially report statistically significant endpoints, which is highly suggestive of selective outcome reporting bias. By reporting only endpoints that achieve statistical significance, advertisers may encourage misconceived notions about a drug’s efficacy profile.
Effects of selective outcome reporting on risk perception
