Background: Some retrospective analyses have suggested that participation in clinical trials is associated with better outcome. However, it is not clear to what extent selection bias contributes to this observation.
Patients and methods: We evaluated the reasons for non-enrolment of ovarian cancer patients in clinical trials. All patients with ovarian cancer not enrolled in clinical studies and treated in 2001 in the participating centres were documented retrospectively and compared with patients enrolled in clinical trials at the same institutions during the same time period.
Results: Two hundred and seventy-four patients with advanced ovarian cancer (FIGO stage IIB-IV) were included, of whom 139 (51%) and 135 (49%) patients were enrolled in this study and in prospective clinical trials, respectively. Ninety-four of 274 patients (34%) did not meet the inclusion criteria for clinical trials. Of 180 eligible patients, 28 (16%) refused participation and a further 17 patients (9%) were not recruited although they met the inclusion criteria. The non-study patients were older (66.7 versus 57.2 years; P <0.0001), underwent less radical surgery (hysterectomy, oophorectomy and omentectomy performed: 61.2% versus 80.7%; P = 0.001; rate of lymphadenectomy 30.9% versus 45.2%; P = 0.015) and more frequently had bulky residual disease (residual disease >2 cm: 36.2% versus 20%; P = 0.016). However, 62% of the non-study patients were treated with the same chemotherapy as in the standard arm of the respective clinical studies.
Conclusions: Study patients differ substantially from non-study patients, thus hampering generalisation of study results. Our results suggest that at least some inclusion criteria for clinical trials should be modified to increase study participation without compromising safety.