OBJECTIVE: We investigated the reasons and rates of attrition in knee osteoarthritis trials through a systematic review of randomized, placebo-controlled, clinical trials. METHODS: Randomized trials were identified by searches conducted in MEDLINE, SCOPUS, and the Cochrane Central Register of Controlled Trials. We then attempted to identify and describe the reasons for attrition and their associated themes. For each theme, we calculated the rate of patients who discontinued a trial from the total number of dropouts in each trial. The rates obtained with different trials were combined using a random effects model. We also performed a random effects meta-regression analysis to identify sources associated with the rates. RESULTS: Overall, 259 studies consisting of 266 trials and 13,593 patients were included in the analysis. From these, we short-listed 54 attrition reasons and identified 21 key themes. "Ineffectiveness" and "adverse event" were the reasons frequently reported by >5% of the dropouts. On further investigation of the theme ineffectiveness, the attrition rate was associated with delivery routes of treatment, trial duration, flare design, prohibition of usual analgesics, and allowing the use of escape medication. In cases of adverse events, we found that the treatment type and delivery route affected the attrition rate. CONCLUSIONS: Our findings not only support the importance of the intention-to-treat analysis, but also suggest the possibility of controlling the attrition at the study level.
OBJECTIVE: We investigated the reasons and rates of attrition in knee osteoarthritis trials through a systematic review of randomized, placebo-controlled, clinical trials. METHODS: Randomized trials were identified by searches conducted in MEDLINE, SCOPUS, and the Cochrane Central Register of Controlled Trials. We then attempted to identify and describe the reasons for attrition and their associated themes. For each theme, we calculated the rate of patients who discontinued a trial from the total number of dropouts in each trial. The rates obtained with different trials were combined using a random effects model. We also performed a random effects meta-regression analysis to identify sources associated with the rates. RESULTS: Overall, 259 studies consisting of 266 trials and 13,593 patients were included in the analysis. From these, we short-listed 54 attrition reasons and identified 21 key themes. "Ineffectiveness" and "adverse event" were the reasons frequently reported by >5% of the dropouts. On further investigation of the theme ineffectiveness, the attrition rate was associated with delivery routes of treatment, trial duration, flare design, prohibition of usual analgesics, and allowing the use of escape medication. In cases of adverse events, we found that the treatment type and delivery route affected the attrition rate. CONCLUSIONS: Our findings not only support the importance of the intention-to-treat analysis, but also suggest the possibility of controlling the attrition at the study level.