outlineofpresentation(已修改)
2025-07-29 16:32 本頁(yè)面
【正文】 Twostage individual participant data metaanalysis and flexible forest plots David Fisher MRC Clinical Trials Unit Hub for Trials Methodology Research at UCL 2022 UK Stata Users Group Meeting Cass Business School, London Outline of presentation ? Introduction to individual patient data (IPD) metaanalysis (MA) ? IPD vs aggregatedata (AD) MA ? “Onestage” vs “twostage” IPD MA ? The ipdmetan mand ? Basic use。 parison with metan ? Covariate interactions ? Combining AD with IPD ? Advanced syntax ? The forestplot mand ? Interface with ipdmetan ? Standalone use and “stacking” ? Summary and Conclusion Introduction to IPD metaanalysis ? Metaanalysis (MA): ? Use statistical methods to bine results of “similar” trials to give a single estimate of effect ? Increase power amp。 precision ? Assess whether treatment effects are similar in across trials (heterogeneity) ? Aggregate data (AD) vs IPD: ? “Traditional” MAs gather results from publications ? Aggregated across all patients in the trial。 nothing is known of individual patients ? IPD MAs gather raw data from trial investigators ? Ensures all relevant patients are included ? Ensures similar analysis across all trials ? Allows more plex analysis, . patientlevel interactions “Onestage” IPD MA ? Consider a linear regression (extension to GLMs or timetoevent regressions is straightforward) ? For a onestage IPD MA (i = trial, j = patient): ?????? = ???? + ?? + ???? ?????? ? Examples in Stata: ? Fixed effects: regress y x ? Random effects: xtmixed y x || trial: x, nocons where αi = trial identifiers β = overall treatment effect estimated across all trials i (with optional random effect ui) “Twostage” IPD MA ? For a twostage IPD MA: ?? ?? ?? = ??(??) + ??(??)?? ?? ?? ?? 1 ?? = ??(1) + ??(1)?? 1 ?? … for trial 1 for trial i … ? Then: ?? = ??????(??)?? ?????? and ???? = 1???? ??(??) 2 ? Weights wi may be altered to give random effects ? . DerSimonian amp。 Laird, ???? = 1 ???? ??(??) 2 + ??2 ? Straightforward, but currently messy in Stata where ???? ?? = 1 ?????? Treatmentcovariate interactions ? Assessment of patientlevel covariate interactions is a great advantage of IPD ? Arguably best done with “onestage” ? Main effects amp。 interactions (amp。 correlations) estimated simultaneously ? But basic analysis also possible with “twostage” ? Relative effect (interaction coefficient) only ? Same approach (inversevariance) as for main effects ? Ensures no estimation bias from betweentrial effects ? Can be presented in a forest plot, with assessment of heterogeneity etc. ? Discussed in a published paper (Fisher 2022) ?????? = ???? +???????? +???????? + ?????????????? “Onestage” vs “twostage” Onestage Twostage Pros All coeffs amp。 correls estimated simultaneously Flexible amp。 extendable model structure Natural extension of AD MA Easily presentable in forest plots Applicable to any set of effect estimates and SEs (incl. interactions) Negligible difference to 1S in most mon scenarios Cons Requires more statistical expertise Challenging in certain situations, . randomeffects with timetoevent data Not a natural fit with forest plots Only a single estimate can be pooled, which limits plexity (. interactions) Theoretically inferior in (at least) some scenarios Example data ? IPD MA of randomised trials of postoperative r
環(huán)評(píng)公示相關(guān)推薦
鄂ICP備17016276號(hào)-1