CONFOUNDING AND ADJUSTMENT CONCERNS IN EXTERNAL CONTROL ARM EVIDENCE SUBMITTED TO IQWIG
Author(s)
Katarzyna Jablonska, MSc1, Adrianna Czubin, MSc1, Renata Majewska, MSc1, Emily Robertshaw, MSc2, Sylvaine Barbier, MSc3.
1Inizio Ignite Putnam, Krakow, Poland, 2Inizio Ignite Putnam, Newcastle, United Kingdom, 3Inizio Ignite Putnam, Lyon, France.
1Inizio Ignite Putnam, Krakow, Poland, 2Inizio Ignite Putnam, Newcastle, United Kingdom, 3Inizio Ignite Putnam, Lyon, France.
OBJECTIVES: In Germany, the Institute for Quality and Efficiency in Health Care (IQWiG) places particular emphasis on the identification, justification, and adjustment for confounding factors in its guidance for health technology assessment (HTA) submissions involving single-arm trials and external control arms (ECAs). This study aimed to characterize confounding-related criticisms raised in ECA submissions and to evaluate the adjustment approaches used by applicants.
METHODS: A systematic review of German HTA submissions was conducted via the Federal Joint Committee (G-BA) website to identify submissions involving ECAs based on real-world data sources (including registry studies, observational cohorts, and natural history studies) or historical clinical trial data between January 2021 and April 2026.
RESULTS: Thirty-three submissions incorporating ECA analyses were identified. ECAs were ultimately accepted by IQWiG for benefit assessment in five appraisals, however, only one was deemed methodologically acceptable, while limitations were identified for the remaining four. Across all submissions, the most common criticisms raised by IQWiG related to confounding, including lack of systematic confounder identification (n=15), inadequate selection of confounders (n=7), missing data on confounders (n=7), and insufficient balance after adjustment (n=5). Two-thirds of appraisals employed some form of confounding adjustment, including propensity-score-based approaches such as matching, weighting, and stratification. Notably, increasing methodological sophistication in adjustment methods did not consistently translate into greater acceptance of ECA evidence, reflecting both persisting confounding issues and broader concerns regarding study comparability, data quality, and residual uncertainty beyond confounding adjustment.
CONCLUSIONS: Confounding-related concerns remain a central barrier to ECA acceptance in German HTA assessments, despite the increasing use of sophisticated adjustment methodologies. The frequency of these criticisms suggests that established methodological expectations are not yet being consistently incorporated by the industry into ECA design and evidence generation strategies. Future success of ECA submissions may depend on systematically addressing known confounding-related challenges from the earliest stages of ECA study planning.
METHODS: A systematic review of German HTA submissions was conducted via the Federal Joint Committee (G-BA) website to identify submissions involving ECAs based on real-world data sources (including registry studies, observational cohorts, and natural history studies) or historical clinical trial data between January 2021 and April 2026.
RESULTS: Thirty-three submissions incorporating ECA analyses were identified. ECAs were ultimately accepted by IQWiG for benefit assessment in five appraisals, however, only one was deemed methodologically acceptable, while limitations were identified for the remaining four. Across all submissions, the most common criticisms raised by IQWiG related to confounding, including lack of systematic confounder identification (n=15), inadequate selection of confounders (n=7), missing data on confounders (n=7), and insufficient balance after adjustment (n=5). Two-thirds of appraisals employed some form of confounding adjustment, including propensity-score-based approaches such as matching, weighting, and stratification. Notably, increasing methodological sophistication in adjustment methods did not consistently translate into greater acceptance of ECA evidence, reflecting both persisting confounding issues and broader concerns regarding study comparability, data quality, and residual uncertainty beyond confounding adjustment.
CONCLUSIONS: Confounding-related concerns remain a central barrier to ECA acceptance in German HTA assessments, despite the increasing use of sophisticated adjustment methodologies. The frequency of these criticisms suggests that established methodological expectations are not yet being consistently incorporated by the industry into ECA design and evidence generation strategies. Future success of ECA submissions may depend on systematically addressing known confounding-related challenges from the earliest stages of ECA study planning.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MSR122
Topic
Health Technology Assessment, Methodological & Statistical Research
Topic Subcategory
Confounding, Selection Bias Correction, Causal Inference
Disease
No Additional Disease & Conditions/Specialized Treatment Areas