Enrollment Success, Factors, and Prediction Models in Cancer Trials (2008-2019).

PURPOSE: To investigate the enrollment success rate of cancer clinical trials conducted in 2008-2019 and various factors lowering the enrollment success rate. METHODS: This is a cross-sectional study with clinical trial information from the largest registration database ClinicalTrials.gov. Enrollment success rate was defined as actual enrollment greater or equal to 85% of the estimated enrollment goal. The association between trial characteristics and enrollment success was evaluated using the multivariable logistic regression. RESULTS: A total of 4,004 trials in breast, lung, and colorectal cancers were included. The overall enrollment success rate was 49.1%. Compared with 2008-2010 (51.5%) and 2011-2013 (52.1%), the enrollment success rate is lower in 2014-2016 (46.5%) and 2017-2019 (36.4%). Regression analyses found trial activation year, phase I, phase I/phase II, and phase II (v phase III), sponsor agency of government (v industry), not requiring healthy volunteers, and estimated enrollment of 50-100, 100-200, 200, and >500 (v 0-50) were associated with a lower enrollment success rate (P < .05). However, trials with placebo comparator, ≥5 locations (v 1 location), and a higher number of secondary end points (eg, ≥5 v 0) were associated with a higher enrollment success rate (P < .05). The AUC for prediction of the final logistic regression models for all trials and specific trial groups ranged from 0.69 to 0.76. CONCLUSION: This large-scale study supports a lower enrollment success rate over years in cancer clinical trials. Identified factors for enrollment success can be used to develop and improve recruitment strategies for future cancer trials.

Duke Scholars

Author Herbert Pang Biostatistics & Bioinformatics, Division of Biostatistics

Author Xiaofei Wang Biostatistics & Bioinformatics, Division of Biostatistics