Target estimands for population‐adjusted indirect comparisons

Remiro‐Azócar, Antonio

doi:10.1002/sim.9413

Cited by 12 publications

(12 citation statements)

References 93 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is necessary to make a distinction between marginal and conditional effect measures. I argue that the former are more relevant to the HTA setting 1,3 . Another question that arises is whether relative estimands (treatment effects) or absolute estimands (mean outcomes) are more appropriate.…”

Section: Harmonization Between Decision‐making Contextsmentioning

confidence: 99%

“…First and foremost, I would like to thank an anonymous Associate Editor and Prof. Nigel Stallard for arranging a fascinating discussion around my article, “Target estimands for population‐adjusted indirect comparisons.” 1 This is based on a prior exchange with Phillippo et al 2‐4 I extend my gratitude to Russek‐Cohen, 5 Schiel, 6 Senn, 7 Spieker, 8 and Van Lancker et al, 9 for their additional contributions.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Some considerations on target estimands for health technology assessment

Remiro‐Azócar

2022

Statistics in Medicine

Self Cite

View full text Add to dashboard Cite

First and foremost, I would like to thank an anonymous Associate Editor and Prof. Nigel Stallard for arranging a fascinating discussion around my article, "Target estimands for population-adjusted indirect comparisons." 1 This is based on a prior exchange with Phillippo et al. [2][3][4] I extend my gratitude to Russek-Cohen, 5 Schiel, 6 Senn, 7 Spieker, 8 and Van Lancker et al, 9 for their additional contributions.This rejoinder discusses the potential development of an estimands framework in the context of evidence synthesis and health technology assessment (HTA). I consider the following base-case scenario. An evidence synthesis (eg, an indirect treatment comparison or a network meta-analysis) is required for HTA. The evidence synthesis combines the results of multiple randomized controlled trials (RCTs). [10][11][12] Each RCT has been designed for regulatory approval in the premarketing authorization setting, and has target estimands of its own. There are exceptions to this scenario; regulatory decisions are not exclusively based on RCTs and HTA decisions are not exclusively based on evidence synthesis. Nevertheless, it is a common scenario.Two challenges arise. First, different policy decisions will require different research questions; the estimands targeted for regulatory approval will likely not align with those targeted for HTA. Second, within the evidence synthesis, there may be misalignment between the estimands targeted by individual studies. Finally, I reassert that marginal estimands should be preferred in the context of the original discussion. [2][3][4] This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

show abstract

Section: Harmonization Between Decision‐making Contextsmentioning

confidence: 99%

mentioning

confidence: 99%

Some considerations on target estimands for health technology assessment

Remiro‐Azócar

2022

Statistics in Medicine

Self Cite

View full text Add to dashboard Cite

show abstract

“…A recurring theme throughout this article is that the terms “conditional and adjusted (likewise marginal and unadjusted) should not be used interchangeably” because marginal need not mean unadjusted and covariate‐adjusted analyses may also target marginal estimands. 29 , 35 …”

Section: Contextmentioning

confidence: 99%

Parametric G‐computation for compatible indirect treatment comparisons with limited individual patient data

Remiro‐Azócar

Heath

Baio

2022

Research Synthesis Methods

View full text Add to dashboard Cite

Population adjustment methods such as matching‐adjusted indirect comparison (MAIC) are increasingly used to compare marginal treatment effects when there are cross‐trial differences in effect modifiers and limited patient‐level data. MAIC is based on propensity score weighting, which is sensitive to poor covariate overlap and cannot extrapolate beyond the observed covariate space. Current outcome regression‐based alternatives can extrapolate but target a conditional treatment effect that is incompatible in the indirect comparison. When adjusting for covariates, one must integrate or average the conditional estimate over the relevant population to recover a compatible marginal treatment effect. We propose a marginalization method based on parametric G‐computation that can be easily applied where the outcome regression is a generalized linear model or a Cox model. The approach views the covariate adjustment regression as a nuisance model and separates its estimation from the evaluation of the marginal treatment effect of interest. The method can accommodate a Bayesian statistical framework, which naturally integrates the analysis into a probabilistic framework. A simulation study provides proof‐of‐principle and benchmarks the method's performance against MAIC and the conventional outcome regression. Parametric G‐computation achieves more precise and more accurate estimates than MAIC, particularly when covariate overlap is poor, and yields unbiased marginal treatment effect estimates under no failures of assumptions. Furthermore, the marginalized regression‐adjusted estimates provide greater precision and accuracy than the conditional estimates produced by the conventional outcome regression, which are systematically biased because the measure of effect is non‐collapsible.

show abstract

“…It is a pleasure to have the opportunity to discuss Dr. Remiro-Azócar's recent contribution, "Target estimands for population-adjusted indirect comparisons." His insights add to an existing chain of methodological development and corresponding commentary on adjustment methods for indirect comparisons, [1][2][3][4] and more broadly to the ongoing discussion regarding the proper selection of target parameters and the appropriate estimation thereof. While this discussion is of great value in the context of indirect comparisons-in which effects are compared across potentially different populations-the conversation transcends this specific area of application.…”

mentioning

confidence: 95%