The influence of institution accrual on patient survival in Canadian cancer clinical trials
Purpose: To determine whether participating centre annual accrual and centre-size corrected (“proportional”) accrual are associated with overall survival in Canadian Cancer Trials Group (CCTG) clinical trials. Methods: The association between centre annual accrual and overall survival was analyzed for all Canadian participants enrolled by 88 member sites onto 52 phase III clinical trials led by CCTG between 1975 and 2013. Centre annual accrual rate per year was used as a continuous variable. 30 member centres in the Province of Ontario with available referral/treatment data were used to determine the association of centre-size corrected accrual with overall survival, where centre size is estimated by the number of treated cases seen at the centre, and centre-size corrected accrual is defined as the proportion of patients treated at a single centre accrued to clinical trials. The association of accrual measures with overall survival was estimated by a hierarchical mixed effects Cox-proportional hazards models. Results: Median CCTG annual accrual at 88 member centres was 2.47 patients per year (IQR 1.54 – 3.89). Analysis included 20 353 Canadian participants, and 10 338 Ontario participants enrolled onto 52 phase III trials. Centres with higher annual accrual (per 1 patient / year) had longer patient overall survival (HR 0.98, 95% CI 0.96 – 0.99, p = 0.006) after adjustment for covariates. The median centre-size corrected accrual (i.e. as proportion of total incident cases) of Ontario cancer patients that were recruited into CCTG trials, from member centres, was 0.13% (IQR 0.09% – 0.31%). Centre-size corrected accrual (per 0.10% increase in percent of patients accrued to clinical trials) was marginally significantly associated with patient overall survival (HR 0.95, 95% CI 0.89 – 1.02, p = 0.057). Conclusions: Overall survival is longer for cancer patients on CCTG trials treated by high-accruing centres. When centre-size corrected (“proportional”) accrual is modelled, the association with patient overall survival is marginally significant.