ADVANCED NMA TECHNIQUES IN NICE HTA SUBMISSIONS
Author(s)
Garcia A1, van Beekhuizen S2, Gebregergish S2, Gurskyte L3, Ouwens DM4, Heeg B5
1Ingress-Health, rotterdam, The Netherlands, 2Ingress Health, Rotterdam, The Netherlands, 3Ingress-Health, Rotterdam, ZH, The Netherlands, 4Astrazeneca, Mölndal, Sweden, 5Ingress-Health, Rotterdam, The Netherlands
OBJECTIVES: Network meta-analyses (NMA) often form the basis of economic model submissions for reimbursement purposes. Network meta-analyses on hazard ratios (HRs) are most frequently applied, yet this requires an assumption on proportional hazard which is not always supported. According to recent NICE guidelines, applying results of a HR NMA to extrapolated pivotal survival curves with distributions assuming non-proportional hazards is not appropriate for economic modelling. As a result, evidence synthesis methods allowing non-proportional hazards, such as parametric NMA or fractional polynomial NMA are being employed in recent submissions. This study quantifies the approaches applied in submitted NMAs and assesses the effect of the NICE guidelines on the NMA methods in HTA submissions. METHODS: A NICE health technology assessment (HTA) review was conducted on submissions appraised between 2001-2018. The methodologies applied on submissions with NMAs were extracted and evaluated. RESULTS: The review covered 180 HTA submissions. 36 submissions reported NMAs, the results of which were implemented in the economic model. 23 NMAs applied a HR model. The remainder applied more advanced techniques, of which 1 was restricted means NMAs (submitted in 2017), 5 were parametric NMAs (first submitted in 2016) and 7 were fractional polynomial NMAs (first submitted in 2017). The most frequent reason for applying more advanced NMA techniques was violation of the proportional hazard assumption. The methodology of 8 (5 parametric, 3 fractional polynomial) of the 13 submissions with NMAs applying advanced techniques were considered acceptable by NICE. Other NMAs using advanced techniques were critiqued on robustness or uncertainty of results (n=4), or NICE evaluation not reported (n=1). HR NMAs have been critiqued on the appropriateness of assuming constant HRs (n=11). CONCLUSIONS: This study provides insight on the trends in NMA methodologies and other advanced techniques. More complex techniques for dealing with proportional hazard violations have been applied and accepted by HTA reviewers.
Conference/Value in Health Info
2018-11, ISPOR Europe 2018, Barcelona, Spain
Value in Health, Vol. 21, S3 (October 2018)
Code
PRM106
Topic
Methodological & Statistical Research
Topic Subcategory
Modeling and simulation
Disease
Multiple Diseases