WHAT MATTERS BEYOND MODEL ARCHITECTURE? SPONSORSHIP, RISK EXTRAPOLATION, AND FAVOURABLE CONCLUSIONS IN EVALUATIONS OF NEWER TYPE 2 DIABETES THERAPIES
Author(s)
DAN LI, III, PhD1, Carlo A. Marra, BSc, PharmD, PhD2, Amber Young, PhD3, Alesha Smith, PhD3.
1University of Otago, Dunedin, New Zealand, 2Faculty of Health Sciences, Curtin University, Perth, Australia, 3School of Pharmacy and Pharmacology, University of Otago, Dunedin, New Zealand.
1University of Otago, Dunedin, New Zealand, 2Faculty of Health Sciences, Curtin University, Perth, Australia, 3School of Pharmacy and Pharmacology, University of Otago, Dunedin, New Zealand.
OBJECTIVES: Long-term economic models inform reimbursement, guideline, and priority-setting decisions for newer type 2 diabetes therapies. Although model architecture is central to credibility, it may not capture how sponsorship and risk-equation choices shape cost-effectiveness conclusions. This study aims to examine whether favourable conclusions were more clearly patterned by model architecture, sponsorship, or risk-extrapolation approach.
METHODS: A systematic review identified long-term model-based economic evaluations of newer type 2 diabetes therapies. Extracted variables included model architecture, risk-extrapolation approach, funding source, and authors' cost-effectiveness conclusions. Within-study favourable proportions were classified as uniformly favourable, mixed, or uniformly unfavourable. Descriptive patterns were summarised by model architecture, funding source, risk-equation category, and funding-risk-equation strata.
RESULTS: Among 108 eligible evaluations, favourable conclusions were highly concentrated: 90 (83%) were uniformly favourable, 10 (9%) mixed, and 8 (7%) uniformly unfavourable. Mean within-study favourable proportion was 89%; median was 100%. Conclusions varied little by model architecture: uniformly favourable conclusions occurred in 69 of 82 patient-level simulations (84%) and 17 of 22 cohort-level models (77%). Greater descriptive variation was observed by funding source and risk-equation category. Uniformly favourable conclusions were more frequent in industry-funded than non-industry-funded evaluations: 75 of 83 (90%) versus 11 of 19 (58%). By risk-equation category, uniformly favourable conclusions were reported in 51 of 54 evaluations using pre-UKPDS 82 equations (94%), 13 of 18 using UKPDS 82 or later equations (72%), 12 of 15 using mixed UKPDS approaches (80%), and 14 of 21 using other or non-applicable approaches (67%). The clear favourable cluster was industry-funded pre-UKPDS 82 evaluations: 45 of 47 evaluations (96%).
CONCLUSIONS: Favourable conclusions were common and showed a clearer pattern by sponsorship and risk-extrapolation choices than by model architecture alone. Critical appraisal should move beyond model type and examine evidence-governance assumptions through which short-term treatment effects are extrapolated into long-term clinical and economic value.
METHODS: A systematic review identified long-term model-based economic evaluations of newer type 2 diabetes therapies. Extracted variables included model architecture, risk-extrapolation approach, funding source, and authors' cost-effectiveness conclusions. Within-study favourable proportions were classified as uniformly favourable, mixed, or uniformly unfavourable. Descriptive patterns were summarised by model architecture, funding source, risk-equation category, and funding-risk-equation strata.
RESULTS: Among 108 eligible evaluations, favourable conclusions were highly concentrated: 90 (83%) were uniformly favourable, 10 (9%) mixed, and 8 (7%) uniformly unfavourable. Mean within-study favourable proportion was 89%; median was 100%. Conclusions varied little by model architecture: uniformly favourable conclusions occurred in 69 of 82 patient-level simulations (84%) and 17 of 22 cohort-level models (77%). Greater descriptive variation was observed by funding source and risk-equation category. Uniformly favourable conclusions were more frequent in industry-funded than non-industry-funded evaluations: 75 of 83 (90%) versus 11 of 19 (58%). By risk-equation category, uniformly favourable conclusions were reported in 51 of 54 evaluations using pre-UKPDS 82 equations (94%), 13 of 18 using UKPDS 82 or later equations (72%), 12 of 15 using mixed UKPDS approaches (80%), and 14 of 21 using other or non-applicable approaches (67%). The clear favourable cluster was industry-funded pre-UKPDS 82 evaluations: 45 of 47 evaluations (96%).
CONCLUSIONS: Favourable conclusions were common and showed a clearer pattern by sponsorship and risk-extrapolation choices than by model architecture alone. Critical appraisal should move beyond model type and examine evidence-governance assumptions through which short-term treatment effects are extrapolated into long-term clinical and economic value.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
SA2
Topic
Economic Evaluation, Methodological & Statistical Research, Study Approaches
Topic Subcategory
Literature Review & Synthesis
Disease
Diabetes/Endocrine/Metabolic Disorders (including obesity)