abstract
- OBJECTIVE: There appears to be considerable confusion about the interpretation of phase II clinical trial conclusions as contrasted with the alternative (HA) and null (H0) hypotheses. This study was conducted to evaluate whether there is congruence with numerical results of phase II trials and their overall verbal conclusions. METHODS: A literature search of 2006 and 2007 phase II clinical trials was conducted. The alternative and null hypotheses were noted as were point estimates with confidence intervals (CIs). These were compared with the final conclusions and concordance and discordance rates were calculated. RESULTS: A total of 152 eligible analyses were reviewed. The point estimates were below H0 in 42 (27.6%) trials, above HA in 60 (39.4%) trials and between H0 and HA (i.e. the grey zone) in 50 (32.9%) trials. Thirty-three (21.7%) trials reported negative conclusions, 111 (73.0%) reported positive conclusions and 8 (5.3%) were ambiguous. All 60 trials in which the point estimate was greater than HA reported positive conclusions, as did 40/50 (80.0%) of trials with point estimates in the grey zone. CONCLUSIONS: There exist inconsistencies and ambiguities in the conclusions drawn from phase II trials, particularly when results are in the grey zone (greater than H0 but less than HA). This may make the integration of phase II trials in phase III trial development strategies difficult and better understanding of the statistical properties of phase II clinical trials is required.