All that Glitters is Not Gold: Examining Cost Effectiveness Analyses in Radiation Oncology

Abstract

Cost effectiveness analyses (CEA) provide data for health policy decisions in resource constrained environments. These are important in Radiation Oncology as infrastructure and delivery costs increase and indications expand. The purpose of this study was to systematically review methodologic quality and trends in CEAs involving radiotherapy (RT). A systematic review was performed on cost effectiveness/utility studies involving RT, querying PubMed and Embase from inception to September 2020. Non-English, reviews, abstracts and cost-only studies were excluded. Independent reviewers screened and abstracted study demographics, economic parameters and methodological details. After screening 1652 abstracts, 214 met criteria. The first publication was in 1995, and more than half (n = 113, 53%) were published after 2014. Author institutions were from North America (n = 128, 60%), Europe (n = 49, 23%) and Asia (n = 30, 14%) with most reporting in US$ (n = 143, 67%). A majority utilized a decision model (n = 164, 77%), healthcare payer perspective (n = 171, 80%) and a finite time horizon (n = 108, 50%). Publications spanned 96 unique journals, most commonly International Journal of Radiation and Oncological Biology and Physics (n = 35, 16%). Treatment intent was curative in 171 studies. Disease sites included breast (n = 34, 16%), genitourinary (n = 31, 14%), and gastrointestinal (n = 31, 14%). RT was mostly used as primary treatment (n = 144, 67%), followed by adjuvant (n = 70, 33%) and neoadjuvant (n = 10, 5%). Emerging topics included stereotactic RT (n = 45, 21%), immunotherapy (n = 6, 3%), oligometastasis (n = 4, 2%), and heavy particles (n = 23, 11%). RT was compared to other RT (n = 136, 64%), surgery (n = 43, 20%), drugs (n = 14, 7%) and observation (n = 31, 17%). Incomplete reporting was common. Missing elements included analysis perspective (n = 13, 6%), time horizon (n = 38, 18%), discounting of utilities (n = 71, 33%) or costs (n = 54, 25%), and willingness-to-pay threshold (n = 59, 28%). Furthermore, 27 studies did not perform sensitivity analyses, 36 did not evaluate incremental cost-effectiveness ratio and only 60 explicitly utilized recognized reporting guidelines. Conflict of interest statements were found in 63%, with sponsor statements in 59%; 25% were industry sponsors. Outcome parameters were obtained from primary (author institution/trial data) sources in 33%, including randomized trials (RCTs) (n = 20, 9%), retrospective data (n = 20, 9%) and population data (n = 9, 4%). The remainder utilized secondary sources including RCTs (n = 71, 33%), retrospective data (n = 35, 16%) or meta-analyses (n = 11, 5%). Outcomes included quality adjusted life years (n = 158, 74%), life-years (n = 30, 14%) or toxicity (n = 26,12%). 31% utilized author generated utilities; of literature derived only 49% were matched to disease and clinical context. While CEAs are increasingly common in RT, reporting and methodologic rigor must improve. Greater use of published guidelines will improve data quality for decision makers.