IMPACT OF TIME-TO-EVENT EXTRAPOLATION ON 36-MONTH TARGET LESION REVASCULARISATION ESTIMATES FOR COVERED BALLOON-EXPANDABLE STENTS IN AORTOILIAC OCCLUSIVE DISEASE
Author(s)
Kashfa Iqbal, MA1, Lewis Meecham, MD FRCS2, Julie Lyon, BSc3.
1EMEA HEOR, Gore UK Medical Limited, Livingstone, United Kingdom, 2University Hospital of Wales, Cardiff, United Kingdom, 3Gore UK Medical Limited, Livingstone, United Kingdom.
1EMEA HEOR, Gore UK Medical Limited, Livingstone, United Kingdom, 2University Hospital of Wales, Cardiff, United Kingdom, 3Gore UK Medical Limited, Livingstone, United Kingdom.
OBJECTIVES: To evaluate a hazard-based time-to-event extrapolation approach for generating consistent 36-month freedom from target lesion revascularisation (fTLR) estimates for covered balloon-expandable stents (CBES) in aortoiliac occlusive disease (AIOD) with incomplete follow-up.
METHODS: A cohort-based modelling analysis was conducted using published clinical evidence identified through a systematic literature review. Thirty-six-month follow-up data was available for three out of five CBES (CBES-A, CBES-B and CBES-C) whereas only 12-month outcomes were reported for CBES-D and CBES-E. For devices with limited follow-up, 24- and 36-month fTLR estimates were generated using a time-to-event extrapolation framework. Interval-specific hazards for 12-24 and 24-36 months were derived from CBES-B data and applied on the cumulative hazard scale to extrapolate observed 12-month outcomes while preserving the underlying time-to-event structure. CBES-B was selected as the reference dataset because it provided the largest comparator cohort, a broad case mix and lesion complexity, supporting its use as a pragmatic reference for extrapolation. The approach was informed by piecewise exponential modelling methods and standard cumulative-hazard formulations described by Friedman and Collett.
RESULTS: Decreased fTLR rates were observed at 12, 24 and 36 months for CBES-A (95.0%, 94.0% and 92.0%), CBES-B (94.0%, 90.0% and 84.0%) and CBES-C (96.1%, 85.8% and 82.4%), respectively. Published 12-month fTLR estimates for CBES-D and CBES-E were 96.7% and 93.9%, respectively. Using hazard-based extrapolation, corresponding 24- and 36-month fTLR estimates were 92.6% and 86.4% for CBES-D and 89.9% and 83.9% for CBES-E. The extrapolation approach generated continuous time-dependent estimates consistent with event trajectories observed in devices with long-term follow-up.
CONCLUSIONS: Hazard-based time-to-event extrapolation provides a structured and internally consistent method for extending short-term fTLR data to 36 months. By preserving the cumulative event patterns observed in datasets with longer follow-up, this approach enables comparative assessment across devices with heterogeneous follow-up durations.
METHODS: A cohort-based modelling analysis was conducted using published clinical evidence identified through a systematic literature review. Thirty-six-month follow-up data was available for three out of five CBES (CBES-A, CBES-B and CBES-C) whereas only 12-month outcomes were reported for CBES-D and CBES-E. For devices with limited follow-up, 24- and 36-month fTLR estimates were generated using a time-to-event extrapolation framework. Interval-specific hazards for 12-24 and 24-36 months were derived from CBES-B data and applied on the cumulative hazard scale to extrapolate observed 12-month outcomes while preserving the underlying time-to-event structure. CBES-B was selected as the reference dataset because it provided the largest comparator cohort, a broad case mix and lesion complexity, supporting its use as a pragmatic reference for extrapolation. The approach was informed by piecewise exponential modelling methods and standard cumulative-hazard formulations described by Friedman and Collett.
RESULTS: Decreased fTLR rates were observed at 12, 24 and 36 months for CBES-A (95.0%, 94.0% and 92.0%), CBES-B (94.0%, 90.0% and 84.0%) and CBES-C (96.1%, 85.8% and 82.4%), respectively. Published 12-month fTLR estimates for CBES-D and CBES-E were 96.7% and 93.9%, respectively. Using hazard-based extrapolation, corresponding 24- and 36-month fTLR estimates were 92.6% and 86.4% for CBES-D and 89.9% and 83.9% for CBES-E. The extrapolation approach generated continuous time-dependent estimates consistent with event trajectories observed in devices with long-term follow-up.
CONCLUSIONS: Hazard-based time-to-event extrapolation provides a structured and internally consistent method for extending short-term fTLR data to 36 months. By preserving the cumulative event patterns observed in datasets with longer follow-up, this approach enables comparative assessment across devices with heterogeneous follow-up durations.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MSR3
Topic
Clinical Outcomes, Medical Technologies, Methodological & Statistical Research
Topic Subcategory
Missing Data
Disease
Cardiovascular Disorders (including MI, Stroke, Circulatory), No Additional Disease & Conditions/Specialized Treatment Areas