OBJECTIVES : As novel treatment options emerge, evaluation of treatment sequences has become increasingly relevant in clinical and reimbursement decisions. Previous studies have assessed economic models of treatment sequences for chronic diseases such as rheumatoid arthritis and highlighted methodological challenges. This study aimed to review oncology sequence models and to assess the methodological approaches and challenges.

METHODS : Oncology technology appraisals published by the National Institute for Health and Care Excellence as of December 2019 were searched. Cost-effectiveness models that explicitly modeled the efficacy of subsequent lines of therapy were included in the review. Model objectives, structure, data sources, and techniques for modeling treatment sequences were summarized.

RESULTS : Twenty-six oncology sequence models were identified, with 58% published after 2015. The common rationale for sequence modeling was to reflect clinical guidelines and 2-8 relevant sequences were evaluated in each model. Thirteen models focused on solid tumors (such as prostate [N=3] and colorectal [N=3]) and majority (N=8) modeled sequences from first-line metastatic settings. The other 13 models focused on hematology indications including leukemia (N=7) and lymphoma (N=4). The most common model structure was cohort state-transition model (N=22). As clinical trials were not typically designed to compare treatment pathways, efficacy inputs for each line of therapy were usually estimated from the trials in the corresponding lines. Most models used data from different sources to directly model a sequence, except one which adjusted patient characteristics using inverse probability censoring weighted analysis. Seven models conducted indirect treatment comparisons, but only one applied to each modeled line.

CONCLUSIONS : The scarcity of clinical data and the limitations of current modeling approaches have been found to be common challenges for modeling oncology sequences. Future research is required to bridge data gaps and to develop a comprehensive modeling framework for evaluation of treatment pathways in oncology and potentially generalizable for other disease areas.