Placebo-controlled clinical trials on COPD are characterized by premature discontinuation. At present, no clear insight into this phenomenon is available.

To obtain better insight into the phenomenon of premature discontinuation.

We analyzed the pattern of discontinuation in the UPLIFT-trial.

Premature discontinuation was substantial and greater in the placebo than in the tiotropium group (45 vs. 37%, p < 0.001). Patients discontinuing were characterized by more severe COPD (p < 0.0001), greater number of pack years (p < 0.002), smaller pre-bronchodilator and post-bronchodilator FEV₁ (p < 0.0001 for both), and worse SGRQ scores (p < 0.0001). Rates of decline of FEV₁ and SGRQ were greater in non-completers (p < 0.0001 for both). The latter differences increased over time indicating that the evolution of variables in time was related to trial completion. The risks of exacerbations and hospitalizations were greater in non-completers. In logistic regression analysis BMI, post-bronchodilator FEV₁, male gender and treatment with tiotropium were positively related to trial completion, whereas age, worse SGRQ, female gender, current smoking and assignment to the placebo group were negatively related.

Assignment to the control group is related to premature discontinuation. Discontinuation was important and selective in this large trial. Pulmonary function, health-related quality of life and smoking are the most important other variables related to discontinuation. The evolution of variables during the trial is also related to discontinuation. Complete follow-up of discontinued patients may provide better insight into the efficacy of medication in future trials.