pith. sign in

arxiv: 2605.20567 · v1 · pith:X7FUW5OEnew · submitted 2026-05-19 · 📊 stat.ME

Meta-analysis and network meta-analysis of time-to-event outcomes with non-proportional hazards: a Bayesian time-varying hazard ratio approach

Rhiannon K Owen , Keith R Abrams This is my paper

Pith reviewed 2026-05-21 06:01 UTC · model grok-4.3

classification 📊 stat.ME
keywords meta-analysisnetwork meta-analysistime-to-event outcomesnon-proportional hazardstime-varying hazard ratioBayesian methodsCox modelhealth technology assessment
0
0 comments X

The pith

When the proportional hazards assumption fails, a bivariate Bayesian meta-analysis of treatment and treatment-log(time) interaction coefficients from per-study Cox models produces usable time-varying hazard ratios.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper seeks to establish that a treatment-log(time) interaction fitted within each study's Cox model, followed by bivariate meta-analysis or network meta-analysis of the resulting coefficients, yields an overall time-varying hazard ratio that remains interpretable on the familiar hazard ratio scale. This matters in health technology assessment contexts because constant hazard ratio summaries become misleading when treatment effects vary over time, and many alternative approaches have proven too complex for routine decision use. A sympathetic reader would see value in an approach that retains the hazard ratio metric while directly incorporating evidence of non-proportionality from the underlying trials. The method is shown in a pairwise meta-analysis of chemotherapy versus standard care for gastric cancer progression-free survival and in a network meta-analysis of overall survival in BRAF-mutated melanoma.

Core claim

By assuming a treatment-log(time) interaction term within a Cox proportional hazards model for each study and then undertaking a bivariate meta-analysis of the treatment and interaction coefficients, an overall time-varying hazard ratio can be obtained for meta-analysis or network meta-analysis of time-to-event outcomes. In the gastric cancer example the approach produced hazard ratios ranging from 0.83 at 0.5 years to 0.99 at 3.5 years; in the melanoma network the combination immunotherapy showed a hazard ratio improving from 0.37 at one year to 0.24 at five years. The resulting time-varying estimates are presented as intuitive and directly applicable in health technology assessment.

What carries the argument

Bivariate meta-analysis of the treatment coefficient and treatment-log(time) interaction coefficient obtained from per-study Cox models, yielding a time-varying hazard ratio (TVHR).

If this is right

  • The obtained time-varying hazard ratios can be read directly at chosen time points for use in economic models and health technology assessment submissions.
  • The method applies equally to pairwise meta-analysis and to network meta-analysis while preserving the hazard ratio scale.
  • Credible intervals around the time-varying hazard ratio reflect uncertainty arising from both within-study and between-study variation.
  • When non-proportional hazards are detected, the approach avoids reliance on a single constant hazard ratio that misrepresents the data.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same bivariate structure could be extended to individual-patient-data meta-analysis to allow study-specific interaction slopes while borrowing strength across trials.
  • Similar non-proportionality problems arise in long-term survival extrapolation for cost-effectiveness models; the time-varying hazard ratio supplies a direct way to parameterise waning or strengthening effects.
  • Neighbouring methods that use fractional polynomials or restricted cubic splines for the interaction could be compared head-to-head on the same datasets to test robustness of the log(time) choice.

Load-bearing premise

That a treatment-log(time) interaction term within a Cox model for each study sufficiently captures the form of non-proportional hazards present in the included trials.

What would settle it

Re-fitting the same trial data with a more flexible model such as Royston-Parmar splines or time-dependent coefficients and finding that the resulting time profile of hazard ratios differs materially from the log(time)-interaction estimates.

Figures

Figures reproduced from arXiv: 2605.20567 by Keith R Abrams, Rhiannon K Owen.

Figure 1
Figure 1. Figure 1: Gastric cancer meta-analysis - Kaplan-Meier curves for studies with non-proportional hazards. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Advanced BRAF-mutated melanoma NMA - Kaplan-Meier Plots. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Gastric cancer meta-analysis - study-specific time-varying hazard ratios (TVHR) and constant hazard ratios, [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Gastric cancer meta-analysis - pooled time-varying hazard ratio (TVHR) and constant hazard ratio, and [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Melanoma network meta-analysis assuming time-varying and constant hazard ratios (HR), and associated [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Melanoma network meta-analysis - probability best over time using a TVHR NMA model. [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
read the original abstract

Background: Often when undertaking meta-analyses of time-to-event (TTE) outcomes, especially in a Health Technology Assessment context, a hazard ratio (HR) scale is used. However, issues arise when there is evidence of non-proportional hazards in some of the studies included. A number of methods have been advocated, but their use has been limited by either their complexity and/or the ease with which their results can be used in HTA. An alternative approach is to assume a treatment-log(time) interaction within a Cox proportional hazards model for each study, and to then undertake a bivariate meta-analysis of the resulting treatment and interaction coefficients, so that an overall time-varying HR (TVHR) can be obtained. Methods: A TVHR approach was applied to a meta-analysis of chemotherapy compared to Standard of Care for advanced recurrent gastric cancer, and in which Progression-Free Survival (PFS) was an outcome. The approach was also applied to a network meta-analysis (NMA) evaluating overall survival (OS) in advanced BRAF-mutated melanoma. Results: Five trials in the advanced gastric cancer meta-analysis displayed evidence of non-proportional hazards for PFS. Using a TVHR model produced HRs ranging from 0.83 (CrI:0.75-0.91) at 0.5 years to 0.99 (CrI:0.79-1.23) at 3.5 years. Three studies showed evidence of non-proportional hazards in the advanced BRAF-mutated melanoma NMA for OS. Using a TVHR model, nivolumab plus ipilimumab demonstrated consistent superiority from month 7 onwards, with a HR improving from 0.37 (CrI:0.26-0.51) at one year to 0.24 (CrI:0.12-0.45) at five years. Conclusions: A TVHR approach to the meta-analysis or NMA of TTE outcomes when the proportional hazards assumption appears not to hold, produces an intuitive solution which can be readily used in HTA.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes a Bayesian time-varying hazard ratio (TVHR) approach for meta-analysis and network meta-analysis of time-to-event outcomes when the proportional hazards assumption does not hold. It fits a Cox model with a treatment-by-log(time) interaction term to each study, extracts the two coefficients, and feeds them into a bivariate random-effects meta-analysis (or NMA extension) to derive overall time-varying HR trajectories. The method is applied to a meta-analysis of PFS in advanced recurrent gastric cancer (five trials with NPH) and an NMA of OS in advanced BRAF-mutated melanoma (three studies with NPH), producing time-dependent HR estimates presented as suitable for HTA use.

Significance. If the modeling assumptions hold, the approach offers a practical extension of standard Cox and bivariate meta-analysis tools that yields intuitive, time-dependent HRs directly usable in health technology assessment. The two real-data applications demonstrate the method's ability to produce changing HRs (e.g., gastric cancer PFS HR from 0.83 at 0.5 years to 0.99 at 3.5 years; melanoma NMA showing improving superiority for nivolumab plus ipilimumab). Credit is due for grounding the framework in familiar machinery and focusing on HTA applicability rather than purely theoretical derivation.

major comments (2)
  1. [Methods (TVHR model)] Methods section (TVHR construction): The central claim rests on the treatment-log(time) interaction term within each study's Cox model being a sufficient summary of non-proportional hazards. This parametric choice is load-bearing because the extracted coefficients are then pooled via bivariate meta-analysis; if the true NPH pattern is non-monotonic, involves crossing hazards, or varies qualitatively across studies, the coefficients will be misspecified and the resulting TVHR trajectory will propagate that error. The applications report significant interactions but provide no diagnostic evidence (e.g., Schoenfeld residual plots or comparison to alternative forms) that the log-linear interaction matches the observed data signatures.
  2. [Results] Results (applications and model reporting): The manuscript reports plausible numerical TVHR results from the two applications yet omits model diagnostics, sensitivity analyses to Bayesian prior choices on the meta-analysis parameters, and direct comparisons against alternative non-PH methods. These omissions are load-bearing for the claim that the TVHR approach produces a robust, readily usable solution for HTA, as the soundness of the pooled estimates cannot be fully assessed without them.
minor comments (1)
  1. [Abstract and Results] The abstract and results sections would benefit from explicit reporting of MCMC convergence diagnostics or effective sample sizes for the Bayesian bivariate meta-analysis fits.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review of our manuscript. We appreciate the emphasis on methodological robustness and have addressed each major comment below. Where revisions are warranted, we have indicated the changes to be made in the next version of the manuscript.

read point-by-point responses

  1. Referee: Methods section (TVHR construction): The central claim rests on the treatment-log(time) interaction term within each study's Cox model being a sufficient summary of non-proportional hazards. This parametric choice is load-bearing because the extracted coefficients are then pooled via bivariate meta-analysis; if the true NPH pattern is non-monotonic, involves crossing hazards, or varies qualitatively across studies, the coefficients will be misspecified and the resulting TVHR trajectory will propagate that error. The applications report significant interactions but provide no diagnostic evidence (e.g., Schoenfeld residual plots or comparison to alternative forms) that the log-linear interaction matches the observed data signatures.

    Authors: We agree that the treatment-by-log(time) interaction imposes a specific parametric form on the non-proportional hazards and does not capture all possible patterns, such as non-monotonic effects or qualitative differences across studies. This form was selected for its direct interpretability as a time-varying hazard ratio and its compatibility with standard Cox regression and subsequent bivariate meta-analysis. To strengthen the manuscript, we will add Schoenfeld residual plots (or scaled Schoenfeld residuals against log(time)) for each included study in the revised Methods and Results sections to provide diagnostic evidence supporting the interaction term. We will also expand the Discussion to explicitly acknowledge the limitations of this assumption and note that alternative specifications (e.g., time-dependent covariates or spline-based approaches) could be explored in future work when data permit. revision: yes

  2. Referee: Results (applications and model reporting): The manuscript reports plausible numerical TVHR results from the two applications yet omits model diagnostics, sensitivity analyses to Bayesian prior choices on the meta-analysis parameters, and direct comparisons against alternative non-PH methods. These omissions are load-bearing for the claim that the TVHR approach produces a robust, readily usable solution for HTA, as the soundness of the pooled estimates cannot be fully assessed without them.

    Authors: We acknowledge that additional diagnostics and sensitivity checks would improve transparency and support the claim of robustness for HTA applications. In the revised manuscript we will include: (i) Bayesian model diagnostics such as trace plots, Gelman-Rubin statistics, and effective sample sizes for the meta-analysis parameters; (ii) sensitivity analyses varying the prior distributions on the between-study covariance matrix and mean effects, with results presented in supplementary tables; and (iii) a brief comparison of the TVHR results against at least one alternative non-PH approach (e.g., landmark analysis or restricted mean survival time differences) for the gastric cancer example where individual-patient data are available. These additions will be placed in the Results and Supplementary Materials sections. revision: yes

Circularity Check

0 steps flagged

No circularity: standard Cox-plus-bivariate-meta framework with no reduction to fitted inputs

full rationale

The paper presents a modeling workflow: fit a Cox model containing a treatment-by-log(time) interaction term separately in each study, extract the two coefficients, and feed them into a bivariate random-effects meta-analysis (or NMA extension) to produce a time-varying HR trajectory. This chain relies on well-established, externally validated statistical components (Cox partial likelihood and standard bivariate meta-analysis) rather than any derivation that equates its output to its inputs by construction. No self-citations are load-bearing, no uniqueness theorems are invoked, and no ansatz is smuggled via prior work. The reported TVHR values are direct consequences of the chosen parametric form applied to the observed data, not a renaming or re-derivation of the same quantities.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The method rests on the modeling choice that a log-time interaction adequately represents non-proportionality and on standard Bayesian meta-analysis assumptions; no new entities are postulated.

free parameters (1)
  • Bayesian priors on meta-analysis parameters
    Priors for the bivariate random effects are required but not detailed in the abstract.
axioms (1)
  • domain assumption A treatment-log(time) interaction within the Cox model captures the non-proportional hazards present in the data
    This is the central modeling decision invoked to handle non-PH.

pith-pipeline@v0.9.0 · 5927 in / 1184 out tokens · 40329 ms · 2026-05-21T06:01:43.180759+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

47 extracted references · 47 canonical work pages

  1. [1]

    Shannon Cope, Keith Chan, Harlan Campbell, Jenny Chen, John Borrill, Jessica R May, William Malcolm, Sebastien Branchoux, Katrin Kupas, and Jeroen P Jansen. A comparison of alternative network meta-analysis methods in the presence of nonproportional hazards: a case study in first-line advanced or metastatic renal cell carcinoma.V alue in Health, 26(4):465...

    work page 2023

  2. [2]

    Applying trial-derived treatment effects to real-world populations: Generalizing cost-effectiveness estimates when modeling complex hazards.V alue in Health, 27:173–181, 2024

    Ian Koblbauer, Daniel Prieto-Alhambra, Edward Burn, and Rafael Pinedo-Villanueva. Applying trial-derived treatment effects to real-world populations: Generalizing cost-effectiveness estimates when modeling complex hazards.V alue in Health, 27:173–181, 2024

    work page 2024

  3. [3]

    Network meta-analysis of survival data with fractional polynomials.BMC medical research methodology, 11(1):61, 2011

    Jeroen P Jansen. Network meta-analysis of survival data with fractional polynomials.BMC medical research methodology, 11(1):61, 2011

    work page 2011

  4. [4]

    Enhanced secondary analysis of survival data: reconstructing the data from published kaplan-meier survival curves.BMC medical research methodology, 12(1):9, 2012

    Patricia Guyot, AE Ades, Mario JNM Ouwens, and Nicky J Welton. Enhanced secondary analysis of survival data: reconstructing the data from published kaplan-meier survival curves.BMC medical research methodology, 12(1):9, 2012

    work page 2012

  5. [5]

    Network meta-analysis of parametric survival curves

    Mario JNM Ouwens, Zoe Philips, and Jeroen P Jansen. Network meta-analysis of parametric survival curves. Research synthesis methods, 1(3-4):258–271, 2010

    work page 2010

  6. [6]

    Jansen, Sven Klijn, Kevin Towle, Bill Malcolm, and Shannon Cope

    Harlan Campbell, Dylan Maciel, Keith Chan, Jeroen P. Jansen, Sven Klijn, Kevin Towle, Bill Malcolm, and Shannon Cope. One-step parametric network meta-analysis models using the exact likelihood that allow for time-varying treatment effects.Research Synthesis Methods, 16(4):650–671, 2025

    work page 2025

  7. [7]

    Bayesian one-step ipd network meta-analysis of time-to-event data using royston-parmar models.Research synthesis methods, 8(4):451–464, 2017

    Suzanne C Freeman and James R Carpenter. Bayesian one-step ipd network meta-analysis of time-to-event data using royston-parmar models.Research synthesis methods, 8(4):451–464, 2017

    work page 2017

  8. [8]

    Suzanne C Freeman, Alex J Sutton, Nicola J Cooper, Alessandro Gasparini, Michael J Crowther, and Neil Hawkins. Bayesian pairwise meta-analysis of time-to-event outcomes in the presence of non-proportional haz- ards: A simulation study of flexible parametric, piecewise exponential and fractional polynomial models.Re- search Synthesis Methods, 15(5):780–801, 2024

    work page 2024

  9. [9]

    Network meta-analysis of survival out- comes with non-proportional hazards using flexible m-splines.arXiv preprint arXiv:2509.10383, 2025

    David M Phillippo, Ayman Sadek, Hugo Pedder, and Nicky J Welton. Network meta-analysis of survival out- comes with non-proportional hazards using flexible m-splines.arXiv preprint arXiv:2509.10383, 2025

    work page arXiv 2025

  10. [10]

    de Jong, Karel G.M

    Valentijn M.T. de Jong, Karel G.M. Moons, Richard D. Riley, Catrin Tudur Smith, Anthony G. Marson, Mari- nus J.C. Eijkemans, and Thomas P.A. Debray. Individual participant data meta-analysis of intervention studies with time-to-event outcomes: A review of the methodology and an applied example.Research Synthesis Meth- ods, 11(2):148–168, 2020

    work page 2020

  11. [11]

    Treatment effect waning assumptions: a review of national institute of health and care excellence technology appraisals.V alue in Health, 27(8):1003– 1011, 2024

    Laura A Trigg, GJ Melendez-Torres, Ahmed Abdelsabour, and Dawn Lee. Treatment effect waning assumptions: a review of national institute of health and care excellence technology appraisals.V alue in Health, 27(8):1003– 1011, 2024

    work page 2024

  12. [12]

    Using time dependent covariates and time dependent coefficients in the cox model

    Terry Therneau, Cynthia Crowson, and Elizabeth Atkinson. Using time dependent covariates and time dependent coefficients in the cox model. CRAN, Dec 2024

    work page 2024

  13. [13]

    Role of chemotherapy for advanced/recurrent gastric cancer: An individual-patient-data meta- analysis.European Journal of Cancer, 49:1565–1577, 2013

    GASTRIC Group. Role of chemotherapy for advanced/recurrent gastric cancer: An individual-patient-data meta- analysis.European Journal of Cancer, 49:1565–1577, 2013

    work page 2013

  14. [14]

    Federico Rotolo, Xavier Paoletti, and Stefan Michiels. surrosurv: An r package for the evaluation of failure time surrogate endpoints in individual patient data meta-analyses of randomized clinical trials.Computer methods and programs in biomedicine, 155:189–198, 2018

    work page 2018

  15. [15]

    Peter C Austin. Statistical power to detect violation of the proportional hazards assumption when using the cox regression model.Journal of statistical computation and simulation, 88(3):533–552, 2018. 14 Meta-Analysis and Network Meta-Analysis with Time-V arying HRsOwen & Abrams

    work page 2018

  16. [16]

    Network meta-analysis of therapies for previously untreated advanced braf-mutated melanoma.Cancer Treatment Reviews, 74:43–48, 2019

    Michael J Zoratti, Tahira Devji, Oren Levine, Lehana Thabane, and Feng Xie. Network meta-analysis of therapies for previously untreated advanced braf-mutated melanoma.Cancer Treatment Reviews, 74:43–48, 2019

    work page 2019

  17. [17]

    Suzanne C Freeman, Nicola J Cooper, Alex J Sutton, Michael J Crowther, James R Carpenter, and Neil Hawkins. Challenges of modelling approaches for network meta-analysis of time-to-event outcomes in the presence of non- proportional hazards to aid decision making: Application to a melanoma network.Statistical Methods in Medical Research, 31(5):839–861, 2022

    work page 2022

  18. [18]

    Chapman and Hall/CRC, 2023

    David Collett.Modelling survival data in medical research. Chapman and Hall/CRC, 2023

    work page 2023

  19. [19]

    R Foundation for Statistical Comput- ing, Vienna, Austria, 2026

    R Core Team.R: A Language and Environment for Statistical Computing. R Foundation for Statistical Comput- ing, Vienna, Austria, 2026

    work page 2026

  20. [20]

    John Wiley & Sons, 2004

    David J Spiegelhalter, Keith R Abrams, and Jonathan P Myles.Bayesian approaches to clinical trials and health-care evaluation. John Wiley & Sons, 2004

    work page 2004

  21. [21]

    Schmid, Heinz Schmidli, Sibylle Sturtz, Sebastian Weber, and Tim Friede

    Christian R ¨over, Ralf Bender, Sofia Dias, Christopher H. Schmid, Heinz Schmidli, Sibylle Sturtz, Sebastian Weber, and Tim Friede. On weakly informative prior distributions for the heterogeneity parameter in bayesian random-effects meta-analysis.Research Synthesis Methods, 12(4):448–474, 2021

    work page 2021

  22. [22]

    Bayesian methods in meta-analysis and evidence synthesis.Statistical methods in medical research, 10(4):277–303, 2001

    Alex J Sutton and Keith R Abrams. Bayesian methods in meta-analysis and evidence synthesis.Statistical methods in medical research, 10(4):277–303, 2001

    work page 2001

  23. [23]

    Bayesian multivariate meta-analysis with multiple outcomes.Statistics in medicine, 32(17):2911–2934, 2013

    Yinghui Wei and Julian PT Higgins. Bayesian multivariate meta-analysis with multiple outcomes.Statistics in medicine, 32(17):2911–2934, 2013

    work page 2013

  24. [24]

    Prior distributions for variance parameters in hierarchical models.Bayesian Analysis, 1(3):515–533, 2006

    Andrew Gelman. Prior distributions for variance parameters in hierarchical models.Bayesian Analysis, 1(3):515–533, 2006

    work page 2006

  25. [25]

    Network meta-analysis for indirect treatment comparisons.Statistics in medicine, 21(16):2313– 2324, 2002

    Thomas Lumley. Network meta-analysis for indirect treatment comparisons.Statistics in medicine, 21(16):2313– 2324, 2002

    work page 2002

  26. [26]

    Network meta-analysis of multiple outcome measures accounting for borrowing of informa- tion across outcomes.BMC medical research methodology, 14(1):1, 2014

    Felix A Achana, Nicola J Cooper, Sylwia Bujkiewicz, Stephanie J Hubbard, Denise Kendrick, David R Jones, and Alex J Sutton. Network meta-analysis of multiple outcome measures accounting for borrowing of informa- tion across outcomes.BMC medical research methodology, 14(1):1, 2014

    work page 2014

  27. [27]

    JAGS: A program for analysis of bayesian graphical models using gibbs sampling

    Martyn Plummer. JAGS: A program for analysis of bayesian graphical models using gibbs sampling. InProceed- ings of the 3rd International Workshop on Distributed Statistical Computing (DSC 2003), pages 1–10, Vienna, Austria, 2003

    work page 2003

  28. [28]

    R package version 4-17

    Martyn Plummer.rjags: Bayesian Graphical Models using MCMC, 2025. R package version 4-17

    work page 2025

  29. [29]

    Inference from iterative simulation using multiple sequences.Statistical Science, 7(4):457–472, 1992

    Andrew Gelman and Donald B Rubin. Inference from iterative simulation using multiple sequences.Statistical Science, 7(4):457–472, 1992

    work page 1992

  30. [30]

    Rank-normalization, folding, and localization: An improved ˆRfor assessing convergence of MCMC.Bayesian Analysis, 16(2):667– 718, 2021

    Aki Vehtari, Andrew Gelman, Daniel Simpson, Bob Carpenter, and Paul-Christian B¨urkner. Rank-normalization, folding, and localization: An improved ˆRfor assessing convergence of MCMC.Bayesian Analysis, 16(2):667– 718, 2021

    work page 2021

  31. [31]

    CODA: Convergence diagnosis and output anal- ysis for MCMC.R News, 6(1):7–11, 2006

    Martyn Plummer, Nicky Best, Kate Cowles, and Karen Vines. CODA: Convergence diagnosis and output anal- ysis for MCMC.R News, 6(1):7–11, 2006

    work page 2006

  32. [32]

    Bias in identification of the best treatment in a bayesian network meta-analysis for binary outcome: a simulation study.Clinical epidemiology, pages 451–460, 2014

    Taddele Kibret, Danielle Richer, and Joseph Beyene. Bias in identification of the best treatment in a bayesian network meta-analysis for binary outcome: a simulation study.Clinical epidemiology, pages 451–460, 2014

    work page 2014

  33. [33]

    Multilevel network meta-regression for population-adjusted treatment com- parisons.Journal of the Royal Statistical Society Series A: Statistics in Society, 183(3):1189–1210, 2020

    David M Phillippo, Sofia Dias, AE Ades, Mark Belger, Alan Brnabic, Alexander Schacht, Daniel Saure, Zbig- niew Kadziola, and Nicky J Welton. Multilevel network meta-regression for population-adjusted treatment com- parisons.Journal of the Royal Statistical Society Series A: Statistics in Society, 183(3):1189–1210, 2020

    work page 2020

  34. [34]

    Extrapolation beyond the end of trials to estimate long term survival and cost effectiveness.BMJ medicine, 1(1):e000094, 2022

    Nicholas R Latimer and Amanda I Adler. Extrapolation beyond the end of trials to estimate long term survival and cost effectiveness.BMJ medicine, 1(1):e000094, 2022

    work page 2022

  35. [35]

    Analysis of survival data with nonproportional hazards: A comparison of propensity-score-weighted methods.Biometrical journal, 66(1):202200099, 2024

    Elizabeth A Handorf, Marc C Smaldone, Sujana Movva, and Nandita Mitra. Analysis of survival data with nonproportional hazards: A comparison of propensity-score-weighted methods.Biometrical journal, 66(1):202200099, 2024

    work page 2024

  36. [36]

    Meta-analysis using individual participant data: one-stage and two-stage approaches, and why they may differ.Statistics in medicine, 36(5):855–875, 2017

    Danielle L Burke, Joie Ensor, and Richard D Riley. Meta-analysis using individual participant data: one-stage and two-stage approaches, and why they may differ.Statistics in medicine, 36(5):855–875, 2017

    work page 2017

  37. [37]

    A method for reconstructing individual patient data from kaplan-meier survival curves that incorporate marked censoring times.MDM Policy & Practice, 7(1), 2022

    Basia Rogula, Greta Lozano-Ortega, and Karissa M Johnston. A method for reconstructing individual patient data from kaplan-meier survival curves that incorporate marked censoring times.MDM Policy & Practice, 7(1), 2022. 15 Meta-Analysis and Network Meta-Analysis with Time-V arying HRsOwen & Abrams

    work page 2022

  38. [38]

    A spline-based approach to smoothly constrain hazard ratios with a view to apply treatment effect waning.Statistics in Medicine, 44(6):e70035, 2025

    Angus C Jennings, Mark J Rutherford, and Paul C Lambert. A spline-based approach to smoothly constrain hazard ratios with a view to apply treatment effect waning.Statistics in Medicine, 44(6):e70035, 2025

    work page 2025

  39. [39]

    Nicholas R Latimer. Survival analysis for economic evaluations alongside clinical trials—extrapolation with patient-level data: inconsistencies, limitations, and a practical guide.Medical Decision Making, 33(6):743–754, 2013

    work page 2013

  40. [40]

    Network meta-analysis: development of a three- level hierarchical modeling approach incorporating dose-related constraints.V alue in Health, 18(1):116–126, 2015

    Rhiannon K Owen, Douglas G Tincello, and R Abrams Keith. Network meta-analysis: development of a three- level hierarchical modeling approach incorporating dose-related constraints.V alue in Health, 18(1):116–126, 2015

    work page 2015

  41. [41]

    Pmu109 how do nice evidence review groups and ap- praisal committees perceive fractional polynomial methodology in technology appraisals?V alue in Health, 22:S727, 2019

    Thomas Leahy, Megan Besford, and Cormac Sammon. Pmu109 how do nice evidence review groups and ap- praisal committees perceive fractional polynomial methodology in technology appraisals?V alue in Health, 22:S727, 2019

    work page 2019

  42. [42]

    Svenja Petersohn, Bradley McGregor, Sven L Klijn, Jessica R May, Flavia Ejzykowicz, Murat Kurt, Matthew Dyer, Bill Malcolm, S ´ebastien Branchoux, Katharina Nickel, et al. Challenges in conducting fractional poly- nomial and standard parametric network meta-analyses of immune checkpoint inhibitors for first-line advanced renal cell carcinoma.Journal of Co...

    work page 2023

  43. [43]

    Meta-analysis of time-to-event out- comes from randomized trials using restricted mean survival time: application to individual participant data

    Yinghui Wei, Patrick Royston, Jayne F Tierney, and Mahesh KB Parmar. Meta-analysis of time-to-event out- comes from randomized trials using restricted mean survival time: application to individual participant data. Statistics in Medicine, 34(21):2881–2898, 2015

    work page 2015

  44. [44]

    Individual patient data meta-analysis of survival data using poisson regression models.BMC Medical Research Methodology, 12(1):34, 2012

    Michael J Crowther, Richard D Riley, Jan A Staessen, Jiguang Wang, Francois Gueyffier, and Paul C Lambert. Individual patient data meta-analysis of survival data using poisson regression models.BMC Medical Research Methodology, 12(1):34, 2012

    work page 2012

  45. [45]

    Philip Cooney and Arthur White. Extending beyond bagust and beale: Fully parametric piecewise exponential models for extrapolation of survival outcomes in health technology assessment.V alue in Health, 26(10):1510– 1517, 2023

    work page 2023

  46. [46]

    Nicholas Latimer. Nice dsu technical support document 14: survival analysis for economic evaluations alongside clinical trials-extrapolation with patient-level data.Report by the decision support unit, 20, 2011

    work page 2011

  47. [47]

    Therneau.coxme: Mixed Effects Cox Models, 2024

    Terry M. Therneau.coxme: Mixed Effects Cox Models, 2024. R package version 2.2-22. 16 Meta-Analysis and Network Meta-Analysis with Time-V arying HRsOwen & Abrams Supplementary Materials S1 Pairwise meta-analysis - Advanced Gastric Cancer −4 −3 −2 −1 0 1 2 0.021 0.082 0.15 0.23 0.32 0.47 0.71 0.91 Time Beta(t) for tx Study 1 −4 −3 −2 −1 0 1 2 0.08 0.13 0.2...

    work page 2024