Estimands in EQ-5D-5L: How to Deal with the “Worse Than Death” and the Intercurrent Events of “Death”

OBJECTIVES: For the analysis of results from EuroQol 5 dimensions 5-level (EQ-5D-5L) questionnaire in the clinical trials, health state of the five dimensions is usually scored into a single value, called “Index Value", using the established country-specific tariffs. Index Value is generally ranged from 0 (state as bad as being dead) to 1 (full health state); however, in some country-specific tariffs, Index Value less than zero could be taken, representing a “worse than death (WTD)” state. In such a case, it may be argued how to handle an “actual death”. There are often occasions where Index Value of patients with actual death is treated to be zero. It results in the fact that Index Value of patients who died is greater than that of patients with WDT state. It means that the health state of patients who died is better than that of patients who are still alive with WDT state. This may cause misinterpretation of the statistical analysis results of Index Value in comparative clinical trials. This research discusses how to deal with WTD and death in comparative clinical trials.

METHODS: Using the estimand framework (ICH E9 (R1) in 2019), the endpoints, intercurrent events (ICEs) which potentially affect the interpretation of the study results (e.g., actual death), and handling of ICEs were systematically categorized for comparative clinical trials where both WTD state in Index Values and actual death could be observed.

RESULTS: In the framework, we were able to demonstrate that the health state of actual death was treated as a worse state than WTD in the handling of the ICEs. The endpoints and possible ICEs along with their handling were also summarized.

CONCLUSIONS: The summary of the endpoints and handling of ICEs for the analysis of EQ-5D-5L data is useful to assess the health state in comparative clinical trials with potentially death and WTD state.