A Review of NICE Technology Appraisals in Oncology Including Economic Evaluations With Time-Varying Network Meta-Analysis
Author(s)
Gemma Marceniuk Gooud, BSc, MSc, Ada Adriano, BSc, MPH, MSc, Ross Maconachie, MSc, BSc.
MSD, London, United Kingdom.
MSD, London, United Kingdom.
OBJECTIVES: To review time-varying network meta-analysis (NMA) methodologies used in National Institute for Health and Care Excellence (NICE) technology appraisals (TAs), and how these are justified by submitting companies and evaluated by Evidence Assessment Groups (EAGs) and NICE committees.
METHODS: A targeted review of the NICE website was conducted to identify cancer-related TAs incorporating time-varying outcomes in NMAs. Guidance documents were screened to confirm inclusion, and associated committee papers were reviewed to determine the type of time-varying NMA used and the justification provided. The review also assessed how time-varying NMA results were implemented in economic analyses, the selection and justification of reference treatments, and the handling of extrapolation beyond trial follow-up.
RESULTS: 24 TAs were identified as using time-varying NMAs. 18 exclusively employed fractional polynomial (FP) models, while 6 used parametric, piecewise, or spline-based approaches—either as alternative scenarios to FP models or exclusively. Time-varying NMAs were generally well received by EAGs and committees. Few instances were found where EAGs questioned the exclusion of alternative time-varying NMA approaches. For economic analysis, time-varying hazard ratios (HRs) were typically applied to a reference treatment curve extrapolated using standard survival models. However, the choice of reference treatment was rarely justified. Where constraints on time-varying HRs were applied, these were typically linked to discussions on treatment effect waning.
CONCLUSIONS: FP models are the predominant time-varying NMA approaches used in NICE TAs when the PH assumption is violated or when biological rationale supports their use; limited consideration is given to alternative approaches. When similarity, consistency, and transitivity assumptions within the NMA hold, the choice of reference treatment should matter less, but as longer term hazards are frequently attenuated within the economic model, the choice of reference curve can still be influential. A standardised checklist for time-varying NMAs would improve reporting and methodological quality in future TA submissions.
METHODS: A targeted review of the NICE website was conducted to identify cancer-related TAs incorporating time-varying outcomes in NMAs. Guidance documents were screened to confirm inclusion, and associated committee papers were reviewed to determine the type of time-varying NMA used and the justification provided. The review also assessed how time-varying NMA results were implemented in economic analyses, the selection and justification of reference treatments, and the handling of extrapolation beyond trial follow-up.
RESULTS: 24 TAs were identified as using time-varying NMAs. 18 exclusively employed fractional polynomial (FP) models, while 6 used parametric, piecewise, or spline-based approaches—either as alternative scenarios to FP models or exclusively. Time-varying NMAs were generally well received by EAGs and committees. Few instances were found where EAGs questioned the exclusion of alternative time-varying NMA approaches. For economic analysis, time-varying hazard ratios (HRs) were typically applied to a reference treatment curve extrapolated using standard survival models. However, the choice of reference treatment was rarely justified. Where constraints on time-varying HRs were applied, these were typically linked to discussions on treatment effect waning.
CONCLUSIONS: FP models are the predominant time-varying NMA approaches used in NICE TAs when the PH assumption is violated or when biological rationale supports their use; limited consideration is given to alternative approaches. When similarity, consistency, and transitivity assumptions within the NMA hold, the choice of reference treatment should matter less, but as longer term hazards are frequently attenuated within the economic model, the choice of reference curve can still be influential. A standardised checklist for time-varying NMAs would improve reporting and methodological quality in future TA submissions.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
HTA14
Topic
Economic Evaluation, Health Technology Assessment
Topic Subcategory
Decision & Deliberative Processes, Value Frameworks & Dossier Format
Disease
Oncology