Background: Assessing the quality of included studies is important in systematic reviews. In a given study, the choice of parameters and methods for statistical analyses and sample size calculations often involves an element of subjective and even arbitrary judgment. It is thus recommended that these analyses are pre-specified in protocols in order to prevent biased decisions based on the data.

Objectives: To compare the reporting of sample size calculations and statistical analyses in publications of randomised trials with those specified in the protocol.

Methods: We reviewed protocols, amendments, and publications for parallel group randomised trials approved by the Scientific-Ethics Committees of Copenhagen and Frederiksberg from 1994-95. Publications were identified by literature searches and a survey of authors. From both protocols and publications, we extracted information about sample size calculations; methods of analysis for primary outcomes, subgroups, and adjusted results; and methods of handling protocol deviations and missing data.

Results and Conclusions: 102 published randomised trials have been identified from 274 trials approved by the Committees. 70/102 published trials are of parallel group design, and are included in our study cohort. Final study results will be presented.