The Regulatory ESIG discussed the implications for clinicals trials of the Russian attack on Ukraine. Most if not all international sponsors have stopped recruitment in both countries. Initially many statisticians felt that the situation is like a ‘COVID 2.0’. However, on closer inspection only some of the methodological issues are similar to COVID while others are new. To date it appears that the conduct of trials in Russia is largely unaffected by the war. In Ukraine however, trialists and sponsors do face a multitude of serious issues, e.g. destroyed source data, missing data, drastically increased rate of patients lost to follow up, patients switching centers or even countries, or investigators unable to answer queries or provide signatures.
The EMA has just published its “Points to consider on the impact of the war in Ukraine on methodological aspects of ongoing clinical trials” (for more information click here) with the following key points for statistics:
• Use the estimands framework to deal with events related to the regional crisis, i.e. revision of pre-specified estimand and/or addition of estimands may be warranted
• Intercurrent events related to the regional crisis to be addressed on the estimand level, i.e. hypothetical strategy to be considered
• Missing data due to the regional crisis to be dealt with on an estimation level, i.e. use of statistical methods aligned with the estimand of interest/MCAR assumption to be considered
• Sensitivity analyses may be required to assess the robustness of results to missing data assumptions.
To our knowledge no other health authority has published a guidance so far. In the experience of the discussants in the ESIG only a few trials of their respective companies were relevantly affected with some trials in Ukraine severely disturbed while others were virtually unaffected by the war. Hence, a trial-by-trial approach seems reasonable. All agreed that at the present state a solid documentation of the reasons for protocol deviations is key, as it is for COVID-19 related issues.