HOW HAVE ECONOMIC EVALUATIONS EVALUATED UNCERTAINTY IN POLYGENIC RISK SCORES?: A SCOPING REVIEW
Author(s)
Dina Hassen, MPP1, Zilin Cheng, MS2, Katrina Romagnoli, PhD, MS, MLIS1, Laney K. Jones, PharmD, MPH1, Tierney Lyons, MLS1, Shangqing Joyce Jiang, PhD, MPH2, Victoria Schlieder, MS1, Alex S. F. Berry, PhD1, Matthew T. Oetjens, PhD1, Josh Peterson, MD, MPH3, Marc S. Williams, MD1, David Veenstra, PharmD, PhD2, Jing Hao, MPH, MS, PhD, MD1;
1Geisinger, Danville, PA, USA, 2CHOICE Institute, University of Washington, Seattle, WA, USA, 3Vanderbilt University Medical Center, Nashville, TN, USA
1Geisinger, Danville, PA, USA, 2CHOICE Institute, University of Washington, Seattle, WA, USA, 3Vanderbilt University Medical Center, Nashville, TN, USA
OBJECTIVES: Polygenic risk scores (PRS) are increasingly being used to predict patients’ risk for many different diseases. Clinical action is typically recommended above a specific risk threshold. Understanding the impacts of uncertainty in test performance and risk prediction is important for informing policy. The objective of this study was to evaluate how economic evaluations have incorporated uncertainty in PRS performance.
METHODS: We developed a literature search strategy to identify cost-effectiveness studies published from January 2018 to October 2025 that used decision analytic modeling and evaluated adult disease screening strategies with at least one incorporating a PRS. We defined uncertainties in PRS performance as arising from two sources: 1) test performance uncertainty, reflecting variability in PRS construction methodology (e.g. differences in SNP sets, algorithms, and populations), and 2) predictive uncertainty, reflecting uncertainty in patients’ predicted disease risk.
RESULTS: We identified 728 studies, of which 27 met the inclusion and exclusion criteria. Disease conditions included cardiovascular disease (n=6), colorectal cancer (n=5), prostate cancer (n=5), breast cancer (n=3), type 2 diabetes (n=1), other cancers (n=5) and other conditions (n=2). All 27 studies incorporated PRS performance in their models, 16 (59.3%) studies also incorporated PRS uncertainty. Three studies (11.1%) addressed test performance uncertainty: two used scenario analyses and one developed and compared multiple PRS models for the population datasets used for the cost-effectiveness model. Thirteen studies (48.2%) assessed predictive uncertainty. Within the 13 studies, standalone approaches were one-way sensitivity analysis (n=5), probabilistic sensitivity analysis (n=2), or scenario analysis (n=3); 3 used combinations - two pairing PSA with another method and one utilizing all three.
CONCLUSIONS: Fewer than 60% of cost-effectiveness studies evaluated PRS uncertainty. The majority only assessed predictive uncertainty with some type of sensitivity or scenario analysis. Improvement in the assessment of PRS uncertainties in economic evaluations is needed.
METHODS: We developed a literature search strategy to identify cost-effectiveness studies published from January 2018 to October 2025 that used decision analytic modeling and evaluated adult disease screening strategies with at least one incorporating a PRS. We defined uncertainties in PRS performance as arising from two sources: 1) test performance uncertainty, reflecting variability in PRS construction methodology (e.g. differences in SNP sets, algorithms, and populations), and 2) predictive uncertainty, reflecting uncertainty in patients’ predicted disease risk.
RESULTS: We identified 728 studies, of which 27 met the inclusion and exclusion criteria. Disease conditions included cardiovascular disease (n=6), colorectal cancer (n=5), prostate cancer (n=5), breast cancer (n=3), type 2 diabetes (n=1), other cancers (n=5) and other conditions (n=2). All 27 studies incorporated PRS performance in their models, 16 (59.3%) studies also incorporated PRS uncertainty. Three studies (11.1%) addressed test performance uncertainty: two used scenario analyses and one developed and compared multiple PRS models for the population datasets used for the cost-effectiveness model. Thirteen studies (48.2%) assessed predictive uncertainty. Within the 13 studies, standalone approaches were one-way sensitivity analysis (n=5), probabilistic sensitivity analysis (n=2), or scenario analysis (n=3); 3 used combinations - two pairing PSA with another method and one utilizing all three.
CONCLUSIONS: Fewer than 60% of cost-effectiveness studies evaluated PRS uncertainty. The majority only assessed predictive uncertainty with some type of sensitivity or scenario analysis. Improvement in the assessment of PRS uncertainties in economic evaluations is needed.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
EE308
Topic
Economic Evaluation
Disease
No Additional Disease & Conditions/Specialized Treatment Areas