Premature discontinuation of patients: a potential bias in COPD clinical trials

Premature discontinuation from clinical trials may bias results against effective therapies. In the present study mortality rates were retrospectively reviewed in a 6-month, randomised, placebo-controlled trial in which tiotropium 18 mu g daily was shown to decrease chronic obstructive pulmonary disease exacerbations. Patients participated for 6 months even if trial medication was prematurely discontinued. Exposure-adjusted incidence rates (IRs) were calculated for randomisation-end trial, randomisation-end trial drug (0-ED) and end trial drug-end trial (ED-ET). Of 1,829 patients (forced expiratory volume in one second 1.04 L (36% predicted), mean age 68 yrs, 99% male), 16% tiotropium and 27% placebo patients prematurely stopped trial medication. The number of fatal events for the entire cohort was: 62 all cause, including 16 cardiac and 16 lower respiratory. IRs for fatal events per 100 patient-yrs were higher in the discontinued period: 1.9 (0-ED) versus 23.0 (ED-ET) in the tiotropium group and 1.8 versus 19.0 in the placebo group. Respective IRs for fatal cardiac events were 0.7 versus 2.8 (tiotropium) and 0.5 versus 6.2 (placebo); for fatal lower respiratory events were 0.7 versus 2.8 (tiotropium) and 0.8 versus 5.4 (placebo). Rate ratios (tiotropium/placebo) for fatal events were lower in the discontinued period: 1.4 versus 0.5 for cardiac and 0.9 versus 0.5 for lower respiratory. Higher incidence rates of fatal events occurred following premature discontinuation of study medication. Incomplete information from rate ratios occurs as a result of failure to consider outcomes of patients who discontinue early from clinical trials.