One-stage individual participant data meta-analysis models for continuous and binary outcomes: Comparison of treatment coding options and estimation methods

Titanium-nitride-oxide-coated vs. drug-eluting stents in acute coronary syndromes: an individual patient data meta-analysis

BACKGROUND AND AIMS

In acute coronary syndromes (ACS), vascular healing at the site of implantation of drug-eluting stents (DES) can be delayed. Titanium-nitride-oxide-coated stents (TiNOS) demonstrate faster strut coverage without the excessive intimal hyperplasia observed with bare metal stents. The 5-year outcomes of patients presenting with ACS, randomized to receive either TiNOS or DES, were compared.

METHODS

A systematic review and individual participant data meta-analysis of trials comparing TiNOS with DES for the treatment of ACS was conducted (PROSPERO: CRD42024514342). The primary endpoint was major adverse cardiac events (MACE) at 5 years, a composite of cardiac death (CD), myocardial infarction (MI), and ischaemia-driven target lesion revascularization (TLR). Pre-specified secondary endpoints included CD, MI, TLR, and stent thrombosis. Data were pooled using a mixed-effects Cox regression model with random slope and stratified baseline hazards.

RESULTS

Patient-level data (n = 2743) were obtained from three randomized controlled trials (TiNOS: n = 1620 vs. DES: n = 1123). After a median follow-up of 4.93 years, there was no significant difference in the primary endpoint between TiNOS and DES (12.6% vs. 16.2%; hazard ratio [HR] .82, 95% confidence interval [CI] .67-1.00, P = .051), mainly due to a similar rate of TLR (8.0% vs. 8.1%; HR 1.05, 95% CI .80-1.38, P = .733). However, TiNOS was associated with significantly lower rates of CD (1.5% vs. 3.7%; HR .46, 95% CI .26-.81, P = .007), MI (5.2% vs. 9.6%; HR .56, 95% CI .42-.75, P < .001), and stent thrombosis (1.1% vs. 3.8%; HR .30, 95% CI .17-.53, P < .001).

CONCLUSIONS

In ACS patients, TiNOS was associated with similar rates of MACE and TLR as compared with DES but significantly lower rates of CD, MI, and stent thrombosis.

Patent Foramen Ovale Closure in Older Patients With Stroke: Patient Selection for Trial Feasibility

BACKGROUND AND OBJECTIVES

Whether patent foramen ovale (PFO) closure benefits older patients with PFO and cryptogenic stroke is unknown because randomized controlled trials (RCTs) have predominantly enrolled patients younger than 60 years of age. Our objective was to estimate anticipated effects of PFO closure in older patients to predict the numbers needed to plan an RCT.

METHODS

Effectiveness estimates are derived from major observational studies (Risk of Paradoxical Embolism [RoPE] Study and Oxford Vascular Study, together referred to as the "RoPE-Ox" database) and all 6 major RCTs (Systematic, Collaborative, PFO Closure Evaluation [SCOPE] Consortium). To estimate stroke recurrence risk, observed outcomes were calculated for patients older than 60 years in the age-inclusive observational databases (n = 549). To estimate the reduction in the rate of recurrent stroke associated with PFO closure vs medical therapy based on the RoPE score and the presence of high-risk PFO features, a Cox proportional hazards regression model was developed on the RCT data in the SCOPE database (n = 3,740). These estimates were used to calculate sample sizes required for a future RCT.

RESULTS

Five-year risk of stroke recurrence using Kaplan-Meier estimates was 13.7 (95% CI 10.5-17.9) overall, 14.9% (95% CI 10.2-21.6) in those with high-risk PFO features. Predicted relative reduction in the event rate with PFO closure was 12.9% overall, 48.8% in those with a high-risk PFO feature. Using these estimates, enrolling all older patients with cryptogenic stroke and PFO would require much larger samples than those used for prior PFO closure trials, but selectively enrolling patients with high-risk PFO features would require totals of 630 patients for 90% power and 471 patients for 80% power, with an average of 5 years of follow-up.

DISCUSSION

Based on our projections, anticipated effect sizes in older patients with high-risk features make a trial in these subjects feasible. With lengthening life expectancy in almost all regions of the world, the utility of PFO closure in older adults is increasingly important to explore.

Systemic THrombolysis Randomization IN Endovascular Stroke Therapy (SHRINE) Collaboration: a patient-level pooled analysis of the SKIP and DEVT Trials

OBJECTIVE

To evaluate the non-inferiority of endovascular treatment (EVT) alone versus intravenous thrombolysis (IVT) followed by EVT and to assess its heterogeneity across prespecified subgroups.

METHODS

We pooled data from two trials (SKIP in Japan; DEVT in China). Individual patient data were pooled to assess outcomes and heterogeneity of treatment effect. The primary outcome was functional independence (modified Rankin Scale score 0-2) at 90 days. Safety outcomes included symptomatic intracranial hemorrhage (sICH) and 90-day mortality.

RESULTS

We included 438 patients (217 EVT alone; 221 combined IVT+EVT). The meta-analysis failed to demonstrate the non-inferiority of EVT alone over combined IVT+EVT in achieving 90-day functional independence (56.7% vs 51.6%; adjusted common odds ratio (cOR)=1.27, 95% CI 0.84 to 1.92, pnon-inferiority=0.06). Effect sizes favoring EVT alone were present with stroke onset to puncture time longer than 180 min (cOR=2.28, 95% CI 1.18 to 4.38, pinteraction ≤180 vs >180 min=0.02) and intracranial internal carotid artery ICA occlusions (for ICA cOR=3.04, 95% CI 1.10 to 8.43, pinteraction ICA vs MCA=0.08). The rates of sICH (6.5% vs 9.0%; cOR=0.77, 95% CI 0.37 to 1.61) and 90-day mortality (12.9% vs 13.6%; cOR=1.05, 95% CI 0.58 to 1.89) were comparable.

CONCLUSIONS

The cumulative data of these two recent Asian trials failed to unequivocally demonstrate the non-inferiority of EVT alone over combined IVT+EVT. However, our study suggests a potential role for more individualized decision-making. Specifically, Asian patients with stroke onset to EVT longer than 180 min, as well as those with intracranial ICA occlusions and those with atrial fibrillation might have better outcomes with EVT alone than with combined IVT+EVT.

Two-stage or not two-stage? That is the question for IPD meta-analysis projects

Individual participant data meta-analysis (IPDMA) projects obtain, check, harmonise and synthesise raw data from multiple studies. When undertaking the meta-analysis, researchers must decide between a two-stage or a one-stage approach. In a two-stage approach, the IPD are first analysed separately within each study to obtain aggregate data (e.g., treatment effect estimates and standard errors); then, in the second stage, these aggregate data are combined in a standard meta-analysis model (e.g., common-effect or random-effects). In a one-stage approach, the IPD from all studies are analysed in a single step using an appropriate model that accounts for clustering of participants within studies and, potentially, between-study heterogeneity (e.g., a general or generalised linear mixed model). The best approach to take is debated in the literature, and so here we provide clearer guidance for a broad audience. Both approaches are important tools for IPDMA researchers and neither are a panacea. If most studies in the IPDMA are small (few participants or events), a one-stage approach is recommended due to using a more exact likelihood. However, in other situations, researchers can choose either approach, carefully following best practice. Some previous claims recommending to always use a one-stage approach are misleading, and the two-stage approach will often suffice for most researchers. When differences do arise between the two approaches, often it is caused by researchers using different modelling assumptions or estimation methods, rather than using one or two stages per se.

Benefits and harms of perioperative high fraction inspired oxygen for surgical site infection prevention: a protocol for a systematic review and meta-analysis of individual patient data of randomised controlled trials

INTRODUCTION

The use of high fraction of inspired oxygen (FiO2) intraoperatively for the prevention of surgical site infection (SSI) remains controversial. Promising results of early randomised controlled trials (RCT) have been replicated with varying success and subsequent meta-analysis are equivocal. Recent advancements in perioperative care, including the increased use of laparoscopic surgery and pneumoperitoneum and shifts in fluid and temperature management, can affect peripheral oxygen delivery and may explain the inconsistency in reproducibility. However, the published data provides insufficient detail on the participant level to test these hypotheses. The purpose of this individual participant data meta-analysis is to assess the described benefits and harms of intraoperative high FiO2compared with regular (0.21-0.40) FiO2 and its potential effect modifiers.

METHODS AND ANALYSIS

Two reviewers will search medical databases and online trial registries, including MEDLINE, Embase, CENTRAL, CINAHL, ClinicalTrials.gov and WHO regional databases, for randomised and quasi-RCT comparing the effect of intraoperative high FiO2 (0.60-1.00) to regular FiO2 (0.21-0.40) on SSI within 90 days after surgery in adult patients. Secondary outcome will be all-cause mortality within the longest available follow-up. Investigators of the identified trials will be invited to collaborate. Data will be analysed with the one-step approach using the generalised linear mixed model framework and the statistical model appropriate for the type of outcome being analysed (logistic and cox regression, respectively), with a random treatment effect term to account for the clustering of patients within studies. The bias will be assessed using the Cochrane risk-of-bias tool for randomised trials V.2 and the certainty of evidence using Grading of Recommendations, Assessment, Development and Evaluation methodology. Prespecified subgroup analyses include use of mechanical ventilation, nitrous oxide, preoperative antibiotic prophylaxis, temperature (<35°C), fluid supplementation (<15 mL/kg/hour) and procedure duration (>2.5 hour).

ETHICS AND DISSEMINATION

Ethics approval is not required. Investigators will deidentify individual participant data before it is shared. The results will be submitted to a peer-review journal.

PROSPERO REGISTRATION NUMBER

CRD42018090261.

Effects of remdesivir in patients hospitalised with COVID-19: a systematic review and individual patient data meta-analysis of randomised controlled trials

BACKGROUND

Interpretation of the evidence from randomised controlled trials (RCTs) of remdesivir in patients treated in hospital for COVID-19 is conflicting. We aimed to assess the benefits and harms of remdesivir compared with placebo or usual care in these patients, and whether treatment effects differed between prespecified patient subgroups.

METHODS

For this systematic review and meta-analysis, we searched PubMed, Embase, the Cochrane COVID-19 trial registry, ClinicalTrials.gov, the International Clinical Trials Registry Platform, and preprint servers from Jan 1, 2020, until April 11, 2022, for RCTs of remdesivir in adult patients hospitalised with COVID-19, and contacted the authors of eligible trials to request individual patient data. The primary outcome was all-cause mortality at day 28 after randomisation. We used multivariable hierarchical regression-adjusting for respiratory support, age, and enrollment period-to investigate effect modifiers. This study was registered with PROSPERO, CRD42021257134.

FINDINGS

Our search identified 857 records, yielding nine RCTs eligible for inclusion. Of these nine eligible RCTs, individual data were provided for eight, covering 10 480 patients hospitalised with COVID-19 (99% of such patients included in such RCTs worldwide) recruited between Feb 6, 2020, and April 1, 2021. Within 28 days of randomisation, 662 (12·5%) of 5317 patients assigned to remdesivir and 706 (14·1%) of 5005 patients assigned to no remdesivir died (adjusted odds ratio [aOR] 0·88, 95% CI 0·78-1·00, p=0·045). We found evidence for a credible subgroup effect according to respiratory support at baseline (p_interaction=0·019). Of patients who were ventilated-including those who received high-flow oxygen-253 (30·0%) of 844 patients assigned to remdesivir died compared with 241 (28·5%) of 846 patients assigned to no remdesivir (aOR 1·10 [0·88-1·38]; low-certainty evidence). Of patients who received no oxygen or low-flow oxygen, 409 (9·1%) of 4473 patients assigned to remdesivir died compared with 465 (11·2%) of 4159 patients assigned to no remdesivir (0·80 [0·70-0·93]; high-certainty evidence). No credible subgroup effect was found for time to start of remdesivir after symptom onset, age, presence of comorbidities, enrolment period, or corticosteroid use. Remdesivir did not increase the frequency of severe or serious adverse events.

INTERPRETATION

This individual patient data meta-analysis showed that remdesivir reduced mortality in patients hospitalised with COVID-19 who required no or conventional oxygen support, but was underpowered to evaluate patients who were ventilated when receiving remdesivir. The effect size of remdesivir in patients with more respiratory support or acquired immunity and the cost-effectiveness of remdesivir remain to be further elucidated.

FUNDING

EU-RESPONSE.

Inferring the multiplicity of founder variants initiating HIV-1 infection: a systematic review and individual patient data meta-analysis

BACKGROUND

HIV-1 infections initiated by multiple founder variants are characterised by a higher viral load and a worse clinical prognosis than those initiated with single founder variants, yet little is known about the routes of exposure through which transmission of multiple founder variants is most probable. Here we used individual patient data to calculate the probability of multiple founders stratified by route of HIV exposure and study methodology.

METHODS

We conducted a systematic review and meta-analysis of studies that estimated founder variant multiplicity in HIV-1 infection, searching MEDLINE, Embase, and Global Health databases for papers published between Jan 1, 1990, and Sept 14, 2020. Eligible studies must have reported original estimates of founder variant multiplicity in people with acute or early HIV-1 infections, have clearly detailed the methods used, and reported the route of exposure. Studies were excluded if they reported data concerning people living with HIV-1 who had known or suspected superinfection, who were documented as having received pre-exposure prophylaxis, or if the transmitting partner was known to be receiving antiretroviral treatment. Individual patient data were collated from all studies, with authors contacted if these data were not publicly available. We applied logistic meta-regression to these data to estimate the probability that an HIV infection is initiated by multiple founder variants. We calculated a pooled estimate using a random effects model, subsequently stratifying this estimate across exposure routes in a univariable analysis. We then extended our model to adjust for different study methods in a multivariable analysis, recalculating estimates across the exposure routes. This study is registered with PROSPERO, CRD42020202672.

FINDINGS

We included 70 publications in our analysis, comprising 1657 individual patients. Our pooled estimate of the probability that an infection is initiated by multiple founder variants was 0·25 (95% CI 0·21-0·29), with moderate heterogeneity (Q=132·3, p<0·0001, I²=64·2%). Our multivariable analysis uncovered differences in the probability of multiple variant infection by exposure route. Relative to a baseline of male-to-female transmission, the predicted probability for female-to-male multiple variant transmission was significantly lower at 0·13 (95% CI 0·08-0·20), and the probabilities were significantly higher for transmissions in people who inject drugs (0·37 [0·24-0·53]) and men who have sex with men (0·30 [0·33-0·40]). There was no significant difference in the probability of multiple variant transmission between male-to-female transmission (0·21 [0·14-0·31]), post-partum transmission (0·18 [0·03-0·57]), pre-partum transmission (0·17 [0·08-0·33]), and intra-partum transmission (0·27 [0·14-0·45]).

INTERPRETATION

We identified that transmissions in people who inject drugs and men who have sex with men are significantly more likely to result in an infection initiated by multiple founder variants, and female-to-male infections are significantly less probable. Quantifying how the routes of HIV infection affect the transmission of multiple variants allows us to better understand how the evolution and epidemiology of HIV-1 determine clinical outcomes.

FUNDING

Medical Research Council Precision Medicine Doctoral Training Programme and a European Research Council Starting Grant.

Estimating interactions in individual participant data meta-analysis: a comparison of methods in practice

Medical interventions may be more effective in some types of individuals than others and identifying characteristics that modify the effectiveness of an intervention is a cornerstone of precision or stratified medicine. The opportunity for detailed examination of treatment-covariate interactions can be an important driver for undertaking an individual participant data (IPD) meta-analysis, rather than a meta-analysis using aggregate data. A number of recent modelling approaches are available. We apply these methods to the Perinatal Antiplatelet Review of International Studies (PARIS) Collaboration IPD dataset and compare estimates between them. We discuss the practical implications of applying these methods, which may be of interest to aid meta-analysists in the use of these, often complex models.Models compared included the two-stage meta-analysis of interaction terms and one-stage models which fit multiple random effects and separate within and between trial information. Models were fitted for nine covariates and five binary outcomes and results compared.Interaction terms produced by the methods were generally consistent. We show that where data are sparse and there is low heterogeneity in the covariate distributions across trials, the meta-analysis of interactions may produce unstable estimates and have issues with convergence. In this IPD dataset, varying assumptions by using multiple random effects in one-stage models or using only within trial information made little difference to the estimates of treatment-covariate interaction. Method choice will depend on datasets characteristics and individual preference.

How to observe the principle of concurrent control in an arm-based meta-analysis using SAS procedures GLIMMIX and BGLIMM

Network meta-analysis is a popular method to synthesize the information obtained in a systematic review of studies (e.g. randomized clinical trials) involving subsets of multiple treatments of interest. The dominant method of analysis employs within-study information on treatment contrasts and integrates this over a network of studies. One advantage of this approach is that all inference is protected by within-study randomization. By contrast, arm-based analyses have been criticized in the past because they may also recover inter-study information when studies are modelled as random, which is the dominant practice, hence violating the principle of concurrent control, requiring treated individuals to only be compared directly with randomized controls. This issue arises regardless of whether analysis is implemented within a frequentist or a Bayesian framework. Here, we argue that recovery of inter-study information can be prevented in an arm-based analysis by adding a fixed study main effect. This simple device means that it is possible to honour the principle of concurrent control in a two-way analysis-of-variance approach that is very easy to implement using generalized linear mixed model procedures and hence may be particularly welcome to those not well versed in the more intricate coding required for a contrast-based analysis. This article is protected by copyright. All rights reserved.

Efficacy of acupuncture in subpopulations with functional constipation: A protocol for a systematic review and individual patient data meta-analysis

Background

Several systematic reviews have suggested that acupuncture is effective against functional constipation, but it is unknown whether variations in treatment effect across subgroups remain consistent. Our purpose of this study is to explore the heterogeneity of treatment effect of acupuncture on functional constipation across subgroups.

Methods

We will search eleven English and Chinese electronic databases and three clinical trial registries from inception to December 2021. Randomized controlled trials that evaluate acupuncture compared with sham acupuncture or no treatment for functional constipation will be eligible if they report at least one primary outcome. The primary outcomes will include the change in weekly complete spontaneous bowel movements or spontaneous bowel movements from baseline. Two authors will independently identify the relevant studies, assess the risk of bias using the Cochrane RoB 2 tool and contact the primary researchers of the eligible trials for individual patient data. Individual patient data obtained from the original trial author will be standardized and all trial data will be combined into a single database. Generalized linear mixed effects model will be used to determine possible subgroup effects by adding an interaction term for predefined subgroup and treatment.

Systematic review registration

International Prospective Register of Systematic Reviews (Number: CRD42020188366).

Diet and physical activity in pregnancy to prevent gestational diabetes: a protocol for an individual participant data (IPD) meta-analysis on the differential effects of interventions with economic evaluation

INTRODUCTION

Mothers with gestational diabetes mellitus (GDM) are at increased risk of pregnancy-related complications and developing type 2 diabetes after delivery. Diet and physical activity-based interventions may prevent GDM, but variations in populations, interventions and outcomes in primary trials have limited the translation of available evidence into practice. We plan to undertake an individual participant data (IPD) meta-analysis of randomised trials to assess the differential effects and cost-effectiveness of diet and physical activity-based interventions in preventing GDM and its complications.

METHODS

The International Weight Management in Pregnancy Collaborative Network database is a living repository of IPD from randomised trials on diet and physical activity in pregnancy identified through a systematic literature search. We shall update our existing search on MEDLINE, Embase, BIOSIS, LILACS, Pascal, Science Citation Index, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Database of Abstracts of Reviews of Effects and Health Technology Assessment Database without language restriction to identify relevant trials until March 2021. Primary researchers will be invited to join the Network and share their IPD. Trials including women with GDM at baseline will be excluded. We shall perform a one and two stage random-effect meta-analysis for each intervention type (all interventions, diet-based, physical activity-based and mixed approach) to obtain summary intervention effects on GDM with 95% CIs and summary treatment-covariate interactions. Heterogeneity will be summarised using I2 and tau2 statistics with 95% prediction intervals. Publication and availability bias will be assessed by examining small study effects. Study quality of included trials will be assessed by the Cochrane Risk of Bias tool, and the Grading of Recommendations, Assessment, Development and Evaluations approach will be used to grade the evidence in the results. A model-based economic analysis will be carried out to assess the cost-effectiveness of interventions to prevent GDM and its complications compared with usual care.

ETHICS AND DISSEMINATION

Ethics approval is not required. The study is registered on the International Prospective Register of Systematic Reviews (CRD42020212884). Results will be submitted for publication in peer-reviewed journals.

The effects of dietary and lifestyle interventions among pregnant women with overweight or obesity on early childhood outcomes: an individual participant data meta-analysis from randomised trials

BACKGROUND

The impact of maternal obesity extends beyond birth, being independently associated with an increased risk of child obesity. Current evidence demonstrates that women provided with a dietary intervention during pregnancy improve their dietary quality and have a modest reduction in gestational weight gain. However, the effect of this on longer-term childhood obesity-related outcomes is unknown.

METHODS

We conducted an individual participant data meta-analysis from RCTs in which women with a singleton, live gestation between 10⁺⁰ and 20⁺⁰ weeks and body mass index (BMI) ≥ 25 kg/m² in early pregnancy were randomised to a diet and/or lifestyle intervention or continued standard antenatal care and in which longer-term maternal and child follow-up at 3-5 years of age had been undertaken. The primary childhood outcome was BMI z-score above the 90th percentile. Secondary childhood outcomes included skinfold thickness measurements and body circumferences, fat-free mass, dietary and physical activity patterns, blood pressure, and neurodevelopment.

RESULTS

Seven primary trials where follow-up of participants occurred were identified by a systematic literature search within the International Weight Management in Pregnancy (i-WIP) Collaborative Group collaboration, with six providing individual participant data. No additional studies were identified after a systematic literature search. A total of 2529 children and 2383 women contributed data. Approximately 30% of all child participants had a BMI z-score above the 90th percentile, with no significant difference between the intervention and control groups (aRR 0.97; 95% CI 0.87, 1.08; p=0.610). There were no statistically significant differences identified for any of the secondary outcome measures.

CONCLUSIONS

In overweight and obese pregnant women, we found no evidence that maternal dietary and/or lifestyle intervention during pregnancy modifies the risk of early childhood obesity. Future research may need to target the pre-conception period in women and early childhood interventions.

TRIAL REGISTRATION

PROSPERO, CRD42016047165.

Reducing Self-harm in Adolescents. An individual participant data meta-analysis (RISA-IPD): systematic review protocol

INTRODUCTION

Up to 10% of adolescents report self-harm in the previous year. Non-fatal repetition is common (18% in 1 year), death from any cause shows a fourfold and suicide a 10-fold excess. Despite the scale of the problem, there is insufficient evidence for effective interventions for self-harm. Those who self-harm do so for a variety of different reasons. Different treatments may be more effective for subgroups of adolescents; however, little is known about which subgroups are appropriate for further study. This protocol outlines a systematic review and individual participant data meta-analysis (IPD-MA) to identify subgroups of adolescents for which therapeutic interventions for self-harm show some evidence of benefit.

METHODS AND ANALYSIS

A systematic literature search was conducted in August 2019 (including Cochrane Library, Embase, trial registers and other databases). An update search is planned. Study selection will identify randomised controlled trials examining interventions to reduce self-harm in adolescents who have self-harmed and presented to services. Identified research teams will be invited to contribute data and form a collaborative group. Two-stage IPD-MA will be used to evaluate effectiveness of different therapeutic interventions compared with any active or non-active control on repetition of self-harm, general psychopathology, depression, suicidal ideation, quality of life and death. Subgroup analyses will identify adolescent subgroups in whom different therapeutic interventions may be more effective. Meta-regression will explore moderating study and intervention effects. Sensitivity analyses will incorporate aggregate data from studies lacking IPD and test the robustness of results to methods for handling missing data, within-study clustering, non-adherence and study quality.

ETHICS AND DISSEMINATION

Ethical approval is provided by the University of Leeds, Faculty of Medicine and Health Ethics Committee (18-098). Outcomes will inform research recommendations and will be disseminated internationally through the collaborative group, a service user advisory group, open-access peer-reviewed publication and conference presentations.

PROSPERO REGISTRATION NUMBER

CRD42019152119.