Methods Used to Adjust for Cross-Over in Clinical Trials: Insights From Scandinavian Health Technology Assessments

OBJECTIVES: In clinical trials, cross-over—where patients switch from the control group to the treatment group—can confound the assessment of treatment efficacy, posing a significant challenge for health technology assessments (HTAs). This project investigates the use of methods to adjust for cross-over in clinical trials within HTAs in Scandinavia (Sweden, Norway, and Denmark), utilising a proprietary database of Nordic HTA decisions.

METHODS: We analysed HTA reports from Sweden, Denmark, and Norway, focusing on decisions using one of three common methods to adjust for cross-over: rank-preserving structural failure time (RPSFT) models, inverse probability of censoring weights (IPCW), and the two-stage estimation method (TSE). Decisions were identified using a keyword search in the Nordic HTA database, NMAi. This proprietary database provides a robust dataset of HTA decisions, enabling detailed comparisons across different methodologies and healthcare systems.

RESULTS: A total of 73 decisions mention using one of the three methods for cross-over adjustment. The most used method was RPSFT (n=36), followed by IPCW (n=27). Only ten decisions mentioned the use of TSE (n=10). Cross-over adjusted analyses generally yielded more favourable efficacy estimates for innovative treatments, but they were not always accepted by the HTA agency, and the agencies often requested additional and supporting analysis when a cross-over adjustment method was used. Preferred methods and critiques differed between the three countries included in the analysis.

CONCLUSIONS: The study highlights the importance of standardised adjustment methods to improve the accuracy and comparability of HTA evaluations. Further research is recommended to standardise these methodologies and promote their integration into HTA processes across the Nordics, especially with the introduction of a Nordic joint HTA process.