Modeling Treatment Sequences in Pharmacoeconomic Models

Abstract

As the number of interventions available in a therapeutic area increases, the relevant decision questions in health technology assessment (HTA) expand to compare treatment sequences instead of discrete treatments and identify optimal sequences or position for a particular treatment in a sequence. The objective of this work was to review approaches used to model treatment sequences and provide practical guidance on conceptualizing whether and how to model sequences in health economic models. Economic models including treatment sequencing assessed by the National Institute for Health and Care Excellence were reviewed, as these assessments generally provide both policy relevance and comprehensive model detail. We identified 40 treatment-sequence models in the following disease areas: oncology (13), autoimmune (7), cardiovascular (6), neurology/mental health (4), infectious disease (2), diabetes (2), and other (6). Modeling techniques included discrete event simulation (6), individual state-transition (3), decision tree (3) and, most commonly, cohort state-transition with tracking states (28). In most cases, treatment sequencing had been incorporated to reflect either clinical practice or clinical trial design. In other cases, it was used to assess where in a treatment sequence a new treatment should be placed, or to evaluate the addition of more efficacious treatment options to a current treatment sequence. Important considerations for determining how to best model sequences include the number of treatment options, patient heterogeneity, key outcomes, and event risk (time-varying or constant). The biggest challenge is the scarcity of clinical data, as clinical trials do not commonly evaluate different treatment sequences.