The Perils of Treatment Crossover in Clinical Trials
Author(s)
Maria Roza Nikolopoulou, MSc1, Goda Kijauskaite, MSc2, Clare Jones, MBA, PhD3, Harrison Davis, PhD2.
1Avalere Health, Athens, Greece, 2Avalere Health, London, United Kingdom, 3Avalere Health, Fleet, United Kingdom.
1Avalere Health, Athens, Greece, 2Avalere Health, London, United Kingdom, 3Avalere Health, Fleet, United Kingdom.
OBJECTIVES: Treatment crossover can confound the estimation of a therapy’s true effect on overall survival (OS). While various statistical methods exist to adjust for this bias, how health technology assessment agencies consider crossover in decision-making remains unclear. This study compared the approaches taken by agencies in oncology assessments, focusing on whether they consider results adjusted for crossover.
METHODS: Oncology appraisals, published by the National Institute for Health and Care Excellence (NICE) between 2001 and April 2025 (n=386), were reviewed to identify those discussing statistical methods to adjust for crossover. Corresponding evaluations (same drug, indication, and clinical trial) conducted by the Haute Autorité de Santé (HAS) and the Gemeinsamer Bundesausschuss (G-BA) were retrieved for comparison.
RESULTS: Of n=18 NICE appraisals identified which discussed adjustments for crossover, corresponding assessments were found for all n=18 from HAS and for n=15 from G-BA. Adjustment methods were mentioned in only two HAS, and six G-BA assessments. Two NICE appraisals showed a shift from non-statistically to statistically significant hazard ratios for OS after crossover adjustment. In both cases, NICE accepted the adjusted OS results. In the equivalent submissions to HAS and G-BA, the adjusted OS results were statistically significant in only one case. Notably, HAS provided no commentary on the adjusted data in either case, and G-BA commented in only one case, expressing scepticism regarding the use of a non-pre-specified adjustment method.
CONCLUSIONS: NICE considered the results from adjusted analyses using several different methodologies across multiple assessments, with adjustments influencing the interpretation of OS outcomes in some cases. In contrast, crossover adjustments had limited impact on the HAS and G-BA decision-making process. This difference highlights the importance of planning to mitigate for crossover and another challenge for methodological alignment required for the Joint Clinical Assessment framework, where harmonizing evidentiary standards is essential.
METHODS: Oncology appraisals, published by the National Institute for Health and Care Excellence (NICE) between 2001 and April 2025 (n=386), were reviewed to identify those discussing statistical methods to adjust for crossover. Corresponding evaluations (same drug, indication, and clinical trial) conducted by the Haute Autorité de Santé (HAS) and the Gemeinsamer Bundesausschuss (G-BA) were retrieved for comparison.
RESULTS: Of n=18 NICE appraisals identified which discussed adjustments for crossover, corresponding assessments were found for all n=18 from HAS and for n=15 from G-BA. Adjustment methods were mentioned in only two HAS, and six G-BA assessments. Two NICE appraisals showed a shift from non-statistically to statistically significant hazard ratios for OS after crossover adjustment. In both cases, NICE accepted the adjusted OS results. In the equivalent submissions to HAS and G-BA, the adjusted OS results were statistically significant in only one case. Notably, HAS provided no commentary on the adjusted data in either case, and G-BA commented in only one case, expressing scepticism regarding the use of a non-pre-specified adjustment method.
CONCLUSIONS: NICE considered the results from adjusted analyses using several different methodologies across multiple assessments, with adjustments influencing the interpretation of OS outcomes in some cases. In contrast, crossover adjustments had limited impact on the HAS and G-BA decision-making process. This difference highlights the importance of planning to mitigate for crossover and another challenge for methodological alignment required for the Joint Clinical Assessment framework, where harmonizing evidentiary standards is essential.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
HTA325
Topic
Clinical Outcomes, Health Technology Assessment, Methodological & Statistical Research
Topic Subcategory
Decision & Deliberative Processes
Disease
Oncology