Table 6 Example of free-text answers provided by authors regarding changes or clarifications of the statistical analysis of the primary outcome

Changes of methods to impute missing data

“….they asked us to include the multiple imputation data for the primary outcome, which was a sensitivity analysis, in the abstract and up front in the results. We ensured that all of the data were presented and it made almost no difference to the results so we were OK with it.” (case 478) “Requested that we change report of primary outcome to use multiple imputation of missing data.” (case 454)

Change of statistical model/methods

“We were asked to use a different type of regression model and different independent variables than we had stated in our trial protocol.” (case 556)

“They recommended a different statistical approach. In the end we felt it was a better approach, and it did not change the overall intent or purpose of the study, but it was different than what we originally had planned in our protocol.” (case 450)

Analysis populations and adjustment issues

“To include all patients after randomisation in ITT and to delete the per-protocol analysis from the main text. Also, to present the baseline adjusted analysis as primary analysis and the analysis adjusted for unbalanced baseline characteristic as additional analysis.” (case 509)

Clarifications or presentation issues

“The editors requested clarification on a planned non-inferiority analysis.” (case 651)

“Clarify statistical methods used.” (case 563)

“Change mean and SD into median and 95% CI in case of skewed distributions.” (case 462)