One-stage individual participant data meta-analysis models: estimation of treatment-covariate interactions must avoid ecological bias by separating out within-trial and across-trial information

Hua, Hairui; Burke, Danielle L.; Crowther, Michael J.; Ensor, Joie; Smith, Catrin Tudur; Riley, Richard D

doi:10.1002/sim.7171

Cited by 75 publications

(141 citation statements)

References 56 publications

Supporting

Mentioning

137

Contrasting

Order By: Relevance

“…The presented methods assume that within‐trial interactions and across‐trial interactions are equivalent. If ecological bias is at play, across‐trial information may be biased and may differ from interactions found within trials . The NMA models that separate within‐trial and across‐trial interaction have been previously proposed to explore biases .…”

Section: Discussionmentioning

confidence: 99%

Network meta‐analysis including treatment by covariate interactions: Consistency can vary across covariate values

Donegan

Welton

Smith

et al. 2017

Research Synthesis Methods

Self Cite

View full text Add to dashboard Cite

BackgroundMany reviews aim to compare numerous treatments and report results stratified by subgroups (eg, by disease severity). In such cases, a network meta‐analysis model including treatment by covariate interactions can estimate the relative effects of all treatment pairings for each subgroup of patients. Two key assumptions underlie such models: consistency of treatment effects and consistency of the regression coefficients for the interactions. Consistency may differ depending on the covariate value at which consistency is assessed. For valid inference, we need to be confident of consistency for the relevant range of covariate values. In this paper, we demonstrate how to assess consistency of treatment effects from direct and indirect evidence at various covariate values.MethodsConsistency is assessed using visual inspection, inconsistency estimates, and probabilities. The method is applied to an individual patient dataset comparing artemisinin combination therapies for treating uncomplicated malaria in children using the covariate age.ResultsThe magnitude of the inconsistency appears to be decreasing with increasing age for each comparison. For one comparison, direct and indirect evidence differ for age 1 (P = .05), and this brings results for age 1 for all comparisons into question.ConclusionWhen fitting models including interactions, the consistency of direct and indirect evidence must be assessed across the range of covariates included in the trials. Clinical inferences are only valid for covariate values for which results are consistent.

show abstract

Section: Discussionmentioning

confidence: 99%

Network meta‐analysis including treatment by covariate interactions: Consistency can vary across covariate values

Donegan

Welton

Smith

et al. 2017

Research Synthesis Methods

Self Cite

View full text Add to dashboard Cite

show abstract

“…One of the big advantages of IPD MA over aggregate data MA is the greater power that it affords for investigating treatment and patient‐level covariate interactions . To explore how the treatment effect may vary in relation to a patient‐level covariate, a TCI can be fitted . The power to detect a TCI will depend on the distribution of the covariate within each study …”

Section: Introductionmentioning

confidence: 99%

“…For example, an IPD MA comparing 2 antiepileptic drugs as monotherapy for controlling seizures originally combined the within‐ and across‐trial interactions and identified an interaction between treatment and age . However, a recent reanalysis separating out the within‐ and across‐trial interactions no longer indicated an interaction between treatment and age on the basis of the within‐trial interaction only …”

Section: Introductionmentioning

confidence: 99%

“…Individual participant data MA models separating out within‐ and across‐trial interactions were first developed for continuous outcomes and later applied to time‐to‐event outcomes . In the IPD NMA setting, models separating out within‐ and across‐trial interactions have been proposed for dichotomous outcomes and time‐to‐event outcomes using the Cox regression model . In this paper, we show how to separate out the within‐ and across‐trial interactions in the IPD NMA setting for time‐to‐event outcomes using the Royston‐Parmar model.…”

Section: Introductionmentioning

confidence: 99%

“…In the pairwise MA case, it is clear that within‐trial interactions are the most clinically relevant estimates, as they are free from ecological bias . In the NMA case, more research is needed to explore how the consistency assumptions of the network applied to within‐trial interactions (as in our model) can help to improve their precision.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A framework for identifying treatment‐covariate interactions in individual participant data network meta‐analysis

Freeman

Fisher

Tierney

et al. 2018

Research Synthesis Methods

View full text Add to dashboard Cite

Background: Stratified medicine seeks to identify patients most likely to respond to treatment. Individual participant data (IPD) network meta‐analysis (NMA) models have greater power than individual trials to identify treatment‐covariate interactions (TCIs). Treatment‐covariate interactions contain “within” and “across” trial interactions, where the across‐trial interaction is more susceptible to confounding and ecological bias. Methods: We considered a network of IPD from 37 trials (5922 patients) for cervical cancer (2394 events), where previous research identified disease stage as a potential interaction covariate. We compare 2 models for NMA with TCIs: (1) 2 effects separating within‐ and across‐trial interactions and (2) a single effect combining within‐ and across‐trial interactions. We argue for a visual assessment of consistency of within‐ and across‐trial interactions and consider more detailed aspects of interaction modelling, eg, common vs trial‐specific effects of the covariate. This leads us to propose a practical framework for IPD NMA with TCIs. Results: Following our framework, we found no evidence in the cervical cancer network for a treatment‐stage interaction on the basis of the within‐trial interaction. The NMA provided additional power for an across‐trial interaction over and above the pairwise evidence. Following our proposed framework, we found that the within‐ and across‐trial interactions should not be combined. Conclusion: Across‐trial interactions are susceptible to confounding and ecological bias. It is important to separate the sources of evidence to check their consistency and identify which sources of evidence are driving the conclusion. Our framework provides practical guidance for researchers, reducing the risk of unduly optimistic interpretation of TCIs.

show abstract

Patient-Level Pooled Analysis of Ultrasound Renal Denervation in the Sham-Controlled RADIANCE II, RADIANCE-HTN SOLO, and RADIANCE-HTN TRIO Trials

et al. 2023

View full text Add to dashboard Cite

ImportanceUltrasound renal denervation (uRDN) was shown to lower blood pressure (BP) in patients with uncontrolled hypertension (HTN). Establishing the magnitude and consistency of the uRDN effect across the HTN spectrum is clinically important.ObjectiveTo characterize the effectiveness and safety of uRDN vs a sham procedure from individual patient-level pooled data across uRDN trials including either patients with mild to moderate HTN on a background of no medications or with HTN resistant to standardized triple-combination therapy.Data SourcesA Study of the ReCor Medical Paradise System in Clinical Hypertension (RADIANCE-HTN SOLO and TRIO) and A Study of the ReCor Medical Paradise System in Stage II Hypertension (RADIANCE II) trials.Study SelectionTrials with similar designs, standardized operational implementation (medication standardization and blinding of both patients and physicians to treatment assignment), and follow-up.Data Extraction and SynthesisPooled analysis using individual patient-level data using linear regression models to compare uRDN with sham across the trials.Main Outcomes and MeasuresThe primary outcome was baseline-adjusted change in 2-month daytime ambulatory systolic BP (dASBP) between groups.ResultsA total of 506 patients were randomized in the 3 studies (uRDN, 293; sham, 213; mean [SD] age, 54.1 [9.3]; 354 male [70.0%]). After a 1-month medication stabilization period, dASBP was similar between the groups (mean [SD], uRDN, 150.3 [9.2] mm Hg; sham, 150.8 [10.5] mm Hg). At 2 months, dASBP decreased by 8.5 mm Hg to mean (SD) 141.8 (13.8) mm Hg among patients treated with uRDN and by 2.9 mm Hg to 147.9 (14.6) mm Hg among patients treated with a sham procedure (mean difference, −5.9; 95% CI, −8.1 to −3.8 mm Hg; P &lt; .001 in favor of uRDN). BP decreases from baseline with uRDN vs sham were consistent across trials and across BP parameters (office SBP: −10.4 mm Hg vs −3.4 mm Hg; mean difference, −6.4 mm Hg; 95% CI, −9.1 to –3.6 mm Hg; home SBP: −8.4 mm Hg vs −1.4 mm Hg; mean difference, −6.8 mm Hg; 95% CI, −8.7 to −4.9 mm Hg, respectively). The BP reductions with uRDN vs sham were consistent across prespecified subgroups. Independent predictors of a larger BP response to uRDN were higher baseline BP and heart rate and the presence of orthostatic hypertension. No differences in early safety end points were observed between groups.Conclusions and RelevanceResults of this patient-level pooled analysis suggest that BP reductions with uRDN were consistent across HTN severity in sham-controlled trials designed with a 2-month primary end point to standardize medications across randomized groups.Trial RegistrationClinicalTrials.gov Identifier: NCT02649426 and NCT03614260

show abstract

One-stage individual participant data meta-analysis models: estimation of treatment-covariate interactions must avoid ecological bias by separating out within-trial and across-trial information

Cited by 75 publications

References 56 publications

Network meta‐analysis including treatment by covariate interactions: Consistency can vary across covariate values

Network meta‐analysis including treatment by covariate interactions: Consistency can vary across covariate values

A framework for identifying treatment‐covariate interactions in individual participant data network meta‐analysis

Patient-Level Pooled Analysis of Ultrasound Renal Denervation in the Sham-Controlled RADIANCE II, RADIANCE-HTN SOLO, and RADIANCE-HTN TRIO Trials

Contact Info

Product

Resources

About