Search

Systematic reviews and meta-analyses (MA) are accepted modalities for evidence synthesis in evidence-based medicine. However, as MA uses aggregate data that includes averaging patient characteristics and pooled effect estimates, it has limitations when considering personalised medicine. In contrast, individual participant data meta-analysis (IPD-MA) includes and segregates individual patient data to study new outcomes, identify outcome predictors, and analyse multiple covariate effects on treatments. IPD-MA requires data from multiple investigators, review board approvals, clear communication with collaborators, and statistical recalculation of cumulative data. IPD-MA can be performed as a single-stage process where data from all included studies is pooled and reanalysed or as a two-stage process where additionally the data from individual studies is re-analysed before being pooled. This review aims to orient clinicians about IPD-MA, including the process of performing it, comparing it with other types of meta-analyses and considering the potential barriers in conducting it. [ABSTRACT FROM AUTHOR]

Aim To identify data sharing practices of authors of randomized-controlled trials (RCTs) in indexed orthodontic journals and explore associations between published reports and several publication characteristics. Materials and methods RCTs from indexed orthodontic journals in major databases, namely PubMed® (Medline), Scopus®, EMBASE®, and Web of Science™, were included from January 2019 to December 2023. Data extraction was conducted for outcome and predictor variables such as data and statistical code sharing practices reported, protocol registration, funding sources, and other publication characteristics, including the year of publication, journal ranking, the origin of authorship, number of authors, design of the RCT, and outcome-related variables (e.g. efficacy/safety). Statistical analyses included descriptive statistics, cross-tabulations, and univariable and multivariable logistic regression. Results A total of 318 RCTs were included. Statement for intention of the authors to provide their data upon request was recorded in 51 of 318 RCTs (16.0%), while 6 of 318 (1.9%) openly provided their data in repositories. No RCT provided any code or script for statistical analysis. A significant association was found between data sharing practices and the year of publication, with increasing odds for data sharing by 1.56 times across the years (odds ratio [OR]: 1.56; 95% confidence interval [CI]: 1.22, 2.01; P  < .001). RCTs reporting on safety outcomes presented 62% lower odds for including positive data sharing statements compared to efficacy outcomes (OR: 0.38; 95% CI: 0.17, 0.88). There was evidence that funded RCTs were more likely to report on data sharing compared to non-funded (P  = .02). Conclusions Albeit progress has been made towards credibility and transparency in the presentation of findings from RCTs in orthodontics, less than 20% of published orthodontic trials include a positive data sharing statement while less than 2% openly provide their data with publication. [ABSTRACT FROM AUTHOR]

Systematically missing data in distributed data networks presents practical and methodological challenges. Failure to handle it appropriately can bias statistical inference. Multiple imputations can be used to address systematic missingness. However, when data from different study sites cannot be pooled into a unified file, conventional imputation approaches become unavailable due to the absence of a basis for imputation. To address such challenges, we introduce an imputation method based on conditional quantiles – conditional quantile imputation (CQI) – which involves four steps: (i) estimating 99 quantiles for the systematically missing variable in studies with observed data; (ii) deriving a weighted average of regression coefficients across studies and transmitting it to sites with systematically missing data; (iii) imputing the systematically missing values based on observed data and the set of regression coefficients from step ii; and (iv) combining estimates of the substantive outcome model across imputations using Rubin's rules. We evaluate CQI in different simulation scenarios and illustrate it with an applied data example. We conclude that CQI can be a suitable approach for the imputation of systematically missing data when data from multiple studies cannot be pooled. [ABSTRACT FROM AUTHOR]

Increasing integrity concerns in medical research have prompted the development of tools to detect untrustworthy studies. Existing tools primarily assess published aggregate data (AD), though scrutiny of individual participant data (IPD) is often required to detect trustworthiness issues. Thus, we developed the IPD Integrity Tool for detecting integrity issues in randomised trials with IPD available. This manuscript describes the development of this tool. We conducted a literature review to collate and map existing integrity items. These were discussed with an expert advisory group; agreed items were included in a standardised tool and automated where possible. We piloted this tool in two IPD meta‐analyses (including 116 trials) and conducted preliminary validation checks on 13 datasets with and without known integrity issues. We identified 120 integrity items: 54 could be conducted using AD, 48 required IPD, and 18 were possible with AD, but more comprehensive with IPD. An initial reduced tool was developed through consensus involving 13 advisors, featuring 11 AD items across four domains, and 12 IPD items across eight domains. The tool was iteratively refined throughout piloting and validation. All studies with known integrity issues were accurately identified during validation. The final tool includes seven AD domains with 13 items and eight IPD domains with 18 items. The quality of evidence informing healthcare relies on trustworthy data. We describe the development of a tool to enable researchers, editors, and others to detect integrity issues using IPD. Detailed instructions for its application are published as a complementary manuscript in this issue. [ABSTRACT FROM AUTHOR]

Increasing concerns about the trustworthiness of research have prompted calls to scrutinise studies' Individual Participant Data (IPD), but guidance on how to do this was lacking. To address this, we developed the IPD Integrity Tool to screen randomised controlled trials (RCTs) for integrity issues. Development of the tool involved a literature review, consultation with an expert advisory group, piloting on two IPD meta‐analyses (including 73 trials with IPD), preliminary validation on 13 datasets with and without known integrity issues, and evaluation to inform iterative refinements. The IPD Integrity Tool comprises 31 items (13 study‐level, 18 IPD‐specific). IPD‐specific items are automated where possible, and are grouped into eight domains, including unusual data patterns, baseline characteristics, correlations, date violations, patterns of allocation, internal and external inconsistencies, and plausibility of data. Users rate each item as having either no issues, some/minor issue(s), or many/major issue(s) according to decision rules, and justification for each rating is recorded. Overall, the tool guides decision‐making by determining whether a trial has no concerns, some concerns requiring further information, or major concerns warranting exclusion from evidence synthesis or publication. In our preliminary validation checks, the tool accurately identified all five studies with known integrity issues. The IPD Integrity Tool enables users to assess the integrity of RCTs via examination of IPD. The tool may be applied by evidence synthesists, editors and others to determine whether an RCT should be considered sufficiently trustworthy to contribute to the evidence base that informs policy and practice. [ABSTRACT FROM AUTHOR]

Abstract Background Reference intervals, which define an interval in which a specific proportion of measurements from a healthy population are expected to fall, are commonly used in medical practice. Synthesizing information from multiple studies through meta-analysis can provide a more precise and representative reference interval than one derived from a single study. However, the current approaches for estimating the reference interval from a meta-analysis mainly rely on aggregate data and require parametric distributional assumptions that cannot always be checked. Methods With the availability of individual participant data (IPD), non-parametric methods can be used to estimate reference intervals without any distributional assumptions. Furthermore, patient-level covariates can be introduced to estimate personalized reference intervals that may be more applicable to specific patients. This paper introduces quantile regression as a method to estimate the reference interval from an IPD meta-analysis under the fixed effects model. Results We compared several non-parametric bootstrap methods through simulation studies to account for within-study correlation. Under fixed effects model, we recommend keeping the studies fixed and only randomly sampling subjects with replacement within each study. Conclusion We proposed to use the quantile regression in the IPD meta-analysis to estimate the reference interval. Based on the simulation results, we identify an optimal bootstrap strategy for estimating the uncertainty of the estimated reference interval. An example of liver stiffness measurements, a clinically important diagnostic test without explicitly established reference range in children, is provided to demonstrate the use of quantile regression in estimating both overall and subject-specific reference intervals.

Background: Reference intervals, which define an interval in which a specific proportion of measurements from a healthy population are expected to fall, are commonly used in medical practice. Synthesizing information from multiple studies through meta-analysis can provide a more precise and representative reference interval than one derived from a single study. However, the current approaches for estimating the reference interval from a meta-analysis mainly rely on aggregate data and require parametric distributional assumptions that cannot always be checked. Methods: With the availability of individual participant data (IPD), non-parametric methods can be used to estimate reference intervals without any distributional assumptions. Furthermore, patient-level covariates can be introduced to estimate personalized reference intervals that may be more applicable to specific patients. This paper introduces quantile regression as a method to estimate the reference interval from an IPD meta-analysis under the fixed effects model. Results: We compared several non-parametric bootstrap methods through simulation studies to account for within-study correlation. Under fixed effects model, we recommend keeping the studies fixed and only randomly sampling subjects with replacement within each study. Conclusion: We proposed to use the quantile regression in the IPD meta-analysis to estimate the reference interval. Based on the simulation results, we identify an optimal bootstrap strategy for estimating the uncertainty of the estimated reference interval. An example of liver stiffness measurements, a clinically important diagnostic test without explicitly established reference range in children, is provided to demonstrate the use of quantile regression in estimating both overall and subject-specific reference intervals. [ABSTRACT FROM AUTHOR]

Background: Induction of labour (IOL) is common practice and different methods carry different effectiveness and safety profiles. Objectives: To compare the effectiveness, and maternal and perinatal safety outcomes of IOL with vaginal misoprostol versus vaginal dinoprostone using individual participant data from randomised clinical trials. Search strategy: The following databases were searched from inception to March 2023: CINAHL Plus, ClinicalTrials.gov, Cochrane Pregnancy and Childbirth Group Trial Register, Ovid Embase, Ovid Emcare, Ovid MEDLINE, Scopus and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP). Selection criteria: Randomised controlled trials (RCTs), with viable singleton gestation, no language restrictions, and all published and unpublished data. Data collection and analysis: An individual participant data meta‐analysis was carried out. Main results: Ten of 52 eligible trials provided individual participant data, of which two were excluded after checking data integrity. The remaining eight trials compared low‐dose vaginal misoprostol versus dinoprostone, including 4180 women undergoing IOL, which represents 32.8% of all participants in the published RCTs. Of these, 2077 were assigned to low‐dose vaginal misoprostol and 2103 were assigned to vaginal dinoprostone. Compared with vaginal dinoprostone, low‐dose vaginal misoprostol had a comparable rate of vaginal birth. Composite adverse perinatal outcomes did not differ between the groups. Compared with vaginal dinoprostone, composite adverse maternal outcomes were significantly lower with low‐dose vaginal misoprostol (aOR 0.80, 95% CI 0.65–0.98, P = 0.03, I2 = 0%). Conclusions: Low‐dose vaginal misoprostol and vaginal dinoprostone for IOL are comparable in terms of effectiveness and perinatal safety. However, low‐dose vaginal misoprostol is likely to lead to a lower rate of composite adverse maternal outcomes than vaginal dinoprostone. This article includes Author Insights, a video abstract available at: https://vimeo.com/915799731. [ABSTRACT FROM AUTHOR]

Background Major depressive disorder (MDD) is highly prevalent and burdensome for individuals and society. While there are psychological interventions able to prevent and treat MDD, uptake remains low. To overcome structural and attitudinal barriers, an indirect approach of using online insomnia interventions seems promising because insomnia is less stigmatized, predicts MDD onset, is often comorbid and can outlast MDD treatment. This individual-participant-data meta-analysis evaluated the potential of the online insomnia intervention GET.ON Recovery as an indirect treatment to reduce depressive symptom severity (DSS) and potential MDD onset across a range of participant characteristics. Methods Efficacy on depressive symptom outcomes was evaluated using multilevel regression models controlling for baseline severity. To identify potential effect moderators, clinical, sociodemographic, and work-related variables were investigated using univariable moderation and random-forest methodology before developing a multivariable decision tree. Results IPD were obtained from four of seven eligible studies (N = 561); concentrating on workers with high work-stress. DSS was significantly lower in the intervention group both at post-assessment (d = −0.71 [95% CI−0.92 to −0.51]) and at follow-up (d = −0.84 [95% CI −1.11 to −0.57]). In the subsample (n = 121) without potential MDD at baseline, there were no significant group differences in onset of potential MDD. Moderation analyses revealed that effects on DSS differed significantly across baseline severity groups with effect sizes between d = −0.48 and −0.87 (post) and d = − 0.66 to −0.99 (follow-up), while no other sociodemographic, clinical, or work-related characteristics were significant moderators. Conclusions An online insomnia intervention is a promising approach to effectively reduce DSS in a preventive and treatment setting. [ABSTRACT FROM AUTHOR]

The cornerstone of obesity treatment is behavioural weight management, resulting in significant improvements in cardio-metabolic and psychosocial health. However, there is ongoing concern that dietary interventions used for weight management may precipitate the development of eating disorders. Systematic reviews demonstrate that, while for most participants medically supervised obesity treatment improves risk scores related to eating disorders, a subset of people who undergo obesity treatment may have poor outcomes for eating disorders. This review summarises the background and rationale for the formation of the Eating Disorders In weight-related Therapy (EDIT) Collaboration. The EDIT Collaboration will explore the complex risk factor interactions that precede changes to eating disorder risk following weight management. In this review, we also outline the programme of work and design of studies for the EDIT Collaboration, including expected knowledge gains. The EDIT studies explore risk factors and the interactions between them using individual-level data from international weight management trials. Combining all available data on eating disorder risk from weight management trials will allow sufficient sample size to interrogate our hypothesis: that individuals undertaking weight management interventions will vary in their eating disorder risk profile, on the basis of personal characteristics and intervention strategies available to them. The collaboration includes the integration of health consumers in project development and translation. An important knowledge gain from this project is a comprehensive understanding of the impact of weight management interventions on eating disorder risk. [ABSTRACT FROM AUTHOR]

Background: Few studies have systematically examined the safety and effectiveness of antidepressant versus mood stabilizer monotherapy of bipolar II depression. To date, there are no aggregated or mega‐analyses of prospective trials of individual participant‐level data (IPD) to inform future treatment guidelines on the relative safety and effectiveness of antidepressant or lithium monotherapy. Methods: Data from a series of four independent, similarly designed trials of antidepressant or lithium monotherapy (where longitudinal IPD were available) (n = 393) were aggregated into an IPD dataset (i.e., mega‐analysis). Hierarchical log‐linear growth models were used to analyze primary outcome of change over time in Hamilton Rating Scale for Depression (HRSD) scores; while secondary outcomes examined Clinical Global Impressions severity (CGI/S) and change (CGI/C) scores, and change over time in Young Mania Rating (YMR) scores. Results: Relative to lithium monotherapy, antidepressant monotherapy demonstrated significantly greater symptom reduction on HRSD scores across time (b = −2.33, t = −6.68, p < 0.0001), significantly greater symptom reduction on the CGI/S across time (b = −0.414, t = −6.32, p < 0.001), and a significant improvement in CGI/C across time (b = −0.47, t = −7.43, p < 0.0001). No differences were observed in change over time for YMR scores between antidepressant and lithium monotherapy (b = 0.06, t = 0.49, p = 0.62). Conclusion: Findings from this IPD mega‐analysis of bipolar II depression trials suggest a divergence from current evidence‐based guidelines recommending combined mood stabilizer plus antidepressant therapy. The current mega‐analysis suggests that antidepressant monotherapy may provide superior short‐term effectiveness without clinically meaningful increase in treatment‐emergent hypomanic symptoms compared to lithium monotherapy. [ABSTRACT FROM AUTHOR]

BackgroundOxygen therapy is a cornerstone treatment of critically ill patients in the intensive care unit (ICU). Whether lower oxygenation therapy brings superior survival outcomes to higher oxygenation therapy is unknown.MethodsWe will search electronic databases: PubMed, Embase, Web of Science, the Cochrane Central Register of Controlled Trials (CENTRAL), International Clinical Trials Registry Platform (ICTRP), and ClinicalTrials.gov from inception to 1 January 2024. Two authors will independently screen for all eligible clinical studies. Emails will be sent for individual participant data. The statistical analyses will be conducted using STATA 15.0 software.ResultsWe will evaluate the efficacy of lower oxygenation therapy compared with higher oxygenation therapy based on individual participant data.ConclusionThis study will offer clinical evidence for oxygen therapy in ICU patients.

Abstract Observational data provide invaluable real-world information in medicine, but certain methodological considerations are required to derive causal estimates. In this systematic review, we evaluated the methodology and reporting quality of individual-level patient data meta-analyses (IPD-MAs) conducted with non-randomized exposures, published in 2009, 2014, and 2019 that sought to estimate a causal relationship in medicine. We screened over 16,000 titles and abstracts, reviewed 45 full-text articles out of the 167 deemed potentially eligible, and included 29 into the analysis. Unfortunately, we found that causal methodologies were rarely implemented, and reporting was generally poor across studies. Specifically, only three of the 29 articles used quasi-experimental methods, and no study used G-methods to adjust for time-varying confounding. To address these issues, we propose stronger collaborations between physicians and methodologists to ensure that causal methodologies are properly implemented in IPD-MAs. In addition, we put forward a suggested checklist of reporting guidelines for IPD-MAs that utilize causal methods. This checklist could improve reporting thereby potentially enhancing the quality and trustworthiness of IPD-MAs, which can be considered one of the most valuable sources of evidence for health policy.

Abstract Background The provision of data sharing statements (DSS) for clinical trials has been made mandatory by different stakeholders. DSS are a device to clarify whether there is intention to share individual participant data (IPD). What is missing is a detailed assessment of whether DSS are providing clear and understandable information about the conditions for data sharing of IPD for secondary use. Methods A random sample of 200 COVID-19 clinical trials with explicit DSS was drawn from the ECRIN clinical research metadata repository. The DSS were assessed and classified, by two experienced experts and one assessor with less experience in data sharing (DS), into different categories (unclear, no sharing, no plans, yes but vague, yes on request, yes with specified storage location, yes but with complex conditions). Results Between the two experts the agreement was moderate to substantial (kappa=0.62, 95% CI [0.55, 0.70]). Agreement considerably decreased when these experts were compared with a third person who was less experienced and trained in data sharing (“assessor”) (kappa=0.33, 95% CI [0.25, 0.41]; 0.35, 95% CI [0.27, 0.43]). Between the two experts and under supervision of an independent moderator, a consensus was achieved for those cases, where both experts had disagreed, and the result was used as “gold standard” for further analysis. At least some degree of willingness of DS (data sharing) was expressed in 63.5% (127/200) cases. Of these cases, around one quarter (31/127) were vague statements of support for data sharing but without useful detail. In around half of the cases (60/127) it was stated that IPD could be obtained by request. Only in in slightly more than 10% of the cases (15/127) it was stated that the IPD would be transferred to a specific data repository. In the remaining cases (21/127), a more complex regime was described or referenced, which could not be allocated to one of the three previous groups. As a result of the consensus meetings, the classification system was updated. Conclusion The study showed that the current DSS that imply possible data sharing are often not easy to interpret, even by relatively experienced staff. Machine based interpretation, which would be necessary for any practical application, is currently not possible. Machine learning and / or natural language processing techniques might improve machine actionability, but would represent a very substantial investment of research effort. The cheaper and easier option would be for data providers, data requestors, funders and platforms to adopt a clearer, more structured and more standardised approach to specifying, providing and collecting DSS. Trial registration The protocol for the study was pre-registered on ZENODO ( https://zenodo.org/record/7064624#.Y4DIAHbMJD8 ).

Abstract Background Selective reporting of results from only well-performing cut-offs leads to biased estimates of accuracy in primary studies of questionnaire-based screening tools and in meta-analyses that synthesize results. Individual participant data meta-analysis (IPDMA) of sensitivity and specificity at each cut-off via bivariate random-effects models (BREMs) can overcome this problem. However, IPDMA is laborious and depends on the ability to successfully obtain primary datasets, and BREMs ignore the correlation between cut-offs within primary studies. Methods We compared the performance of three recent multiple cut-off models developed by Steinhauser et al., Jones et al., and Hoyer and Kuss, that account for missing cut-offs when meta-analyzing diagnostic accuracy studies with multiple cut-offs, to BREMs fitted at each cut-off. We used data from 22 studies of the accuracy of the Edinburgh Postnatal Depression Scale (EPDS; 4475 participants, 758 major depression cases). We fitted each of the three multiple cut-off models and BREMs to a dataset with results from only published cut-offs from each study (published data) and an IPD dataset with results for all cut-offs (full IPD data). We estimated pooled sensitivity and specificity with 95% confidence intervals (CIs) for each cut-off and the area under the curve. Results Compared to the BREMs fitted to the full IPD data, the Steinhauser et al., Jones et al., and Hoyer and Kuss models fitted to the published data produced similar receiver operating characteristic curves; though, the Hoyer and Kuss model had lower area under the curve, mainly due to estimating slightly lower sensitivity at lower cut-offs. When fitting the three multiple cut-off models to the full IPD data, a similar pattern of results was observed. Importantly, all models had similar 95% CIs for sensitivity and specificity, and the CI width increased with cut-off levels for sensitivity and decreased with an increasing cut-off for specificity, even the BREMs which treat each cut-off separately. Conclusions Multiple cut-off models appear to be the favorable methods when only published data are available. While collecting IPD is expensive and time consuming, IPD can facilitate subgroup analyses that cannot be conducted with published data only.

Observational data provide invaluable real-world information in medicine, but certain methodological considerations are required to derive causal estimates. In this systematic review, we evaluated the methodology and reporting quality of individual-level patient data meta-analyses (IPD-MAs) conducted with non-randomized exposures, published in 2009, 2014, and 2019 that sought to estimate a causal relationship in medicine. We screened over 16,000 titles and abstracts, reviewed 45 full-text articles out of the 167 deemed potentially eligible, and included 29 into the analysis. Unfortunately, we found that causal methodologies were rarely implemented, and reporting was generally poor across studies. Specifically, only three of the 29 articles used quasi-experimental methods, and no study used G-methods to adjust for time-varying confounding. To address these issues, we propose stronger collaborations between physicians and methodologists to ensure that causal methodologies are properly implemented in IPD-MAs. In addition, we put forward a suggested checklist of reporting guidelines for IPD-MAs that utilize causal methods. This checklist could improve reporting thereby potentially enhancing the quality and trustworthiness of IPD-MAs, which can be considered one of the most valuable sources of evidence for health policy. [ABSTRACT FROM AUTHOR]

Background: The provision of data sharing statements (DSS) for clinical trials has been made mandatory by different stakeholders. DSS are a device to clarify whether there is intention to share individual participant data (IPD). What is missing is a detailed assessment of whether DSS are providing clear and understandable information about the conditions for data sharing of IPD for secondary use. Methods: A random sample of 200 COVID-19 clinical trials with explicit DSS was drawn from the ECRIN clinical research metadata repository. The DSS were assessed and classified, by two experienced experts and one assessor with less experience in data sharing (DS), into different categories (unclear, no sharing, no plans, yes but vague, yes on request, yes with specified storage location, yes but with complex conditions). Results: Between the two experts the agreement was moderate to substantial (kappa=0.62, 95% CI [0.55, 0.70]). Agreement considerably decreased when these experts were compared with a third person who was less experienced and trained in data sharing ("assessor") (kappa=0.33, 95% CI [0.25, 0.41]; 0.35, 95% CI [0.27, 0.43]). Between the two experts and under supervision of an independent moderator, a consensus was achieved for those cases, where both experts had disagreed, and the result was used as "gold standard" for further analysis. At least some degree of willingness of DS (data sharing) was expressed in 63.5% (127/200) cases. Of these cases, around one quarter (31/127) were vague statements of support for data sharing but without useful detail. In around half of the cases (60/127) it was stated that IPD could be obtained by request. Only in in slightly more than 10% of the cases (15/127) it was stated that the IPD would be transferred to a specific data repository. In the remaining cases (21/127), a more complex regime was described or referenced, which could not be allocated to one of the three previous groups. As a result of the consensus meetings, the classification system was updated. Conclusion: The study showed that the current DSS that imply possible data sharing are often not easy to interpret, even by relatively experienced staff. Machine based interpretation, which would be necessary for any practical application, is currently not possible. Machine learning and / or natural language processing techniques might improve machine actionability, but would represent a very substantial investment of research effort. The cheaper and easier option would be for data providers, data requestors, funders and platforms to adopt a clearer, more structured and more standardised approach to specifying, providing and collecting DSS. Trial registration: The protocol for the study was pre-registered on ZENODO (https://zenodo.org/record/7064624#.Y4DIAHbMJD8). [ABSTRACT FROM AUTHOR]

BACKGROUND IVF and IUI with ovarian stimulation (IUI-OS) are widely used in managing unexplained infertility. IUI-OS is generally considered first-line therapy, followed by IVF only if IUI-OS is unsuccessful after several attempts. However, there is a growing interest in using IVF for immediate treatment because it is believed to lead to higher live birth rates and shorter time to pregnancy. OBJECTIVE AND RATIONALE Randomized controlled trials (RCTs) comparing IVF versus IUI-OS had varied study designs and findings. Some RCTs used complex algorithms to combine IVF and IUI-OS, while others had unequal follow-up time between arms or compared treatments on a per-cycle basis, which introduced biases. Comparing cumulative live birth rates of IVF and IUI-OS within a consistent time frame is necessary for a fair head-to-head comparison. Previous meta-analyses of RCTs did not consider the time it takes to achieve pregnancy, which is not possible using aggregate data. Individual participant data meta-analysis (IPD-MA) allows standardization of follow-up time in different trials and time-to-event analysis methods. We performed this IPD-MA to investigate if IVF increases cumulative live birth rate considering the time leading to pregnancy and reduces multiple pregnancy rate compared to IUI-OS in couples with unexplained infertility. SEARCH METHODS We searched MEDLINE, EMBASE, CENTRAL, PsycINFO, CINAHL, and the Cochrane Gynaecology and Fertility Group Specialised Register to identify RCTs that completed data collection before June 2021. A search update was carried out in January 2023. RCTs that compared IVF/ICSI to IUI-OS in couples with unexplained infertility were eligible. We invited author groups of eligible studies to join the IPD-MA and share the deidentified IPD of their RCTs. IPD were checked and standardized before synthesis. The quality of evidence was assessed using the Risk of Bias 2 tool. OUTCOMES Of eight potentially eligible RCTs, two were considered awaiting classification. In the other six trials, four shared IPD of 934 women, of which 550 were allocated to IVF and 383 to IUI-OS. Because the interventions were unable to blind, two RCTs had a high risk of bias, one had some concerns, and one had a low risk of bias. Considering the time to pregnancy leading to live birth, the cumulative live birth rate was not significantly higher in IVF compared to that in IUI-OS (4 RCTs, 908 women, 50.3% versus 43.2%, hazard ratio 1.19, 95% CI 0.81–1.74, I 2 = 42.4%). For the safety primary outcome, the rate of multiple pregnancy was not significantly lower in IVF than IUI-OS (3 RCTs, 890 women, 3.8% versus 5.2% of all couples randomized, odds ratio 0.78, 95% CI 0.41–1.50, I 2 = 0.0%). WIDER IMPLICATIONS There is no robust evidence that in couples with unexplained infertility IVF achieves pregnancy leading to live birth faster than IUI-OS. IVF and IUI-OS are both viable options in terms of effectiveness and safety for managing unexplained infertility. The associated costs of interventions and the preference of couples need to be weighed in clinical decision-making. [ABSTRACT FROM AUTHOR]

Background: Selective reporting of results from only well-performing cut-offs leads to biased estimates of accuracy in primary studies of questionnaire-based screening tools and in meta-analyses that synthesize results. Individual participant data meta-analysis (IPDMA) of sensitivity and specificity at each cut-off via bivariate random-effects models (BREMs) can overcome this problem. However, IPDMA is laborious and depends on the ability to successfully obtain primary datasets, and BREMs ignore the correlation between cut-offs within primary studies. Methods: We compared the performance of three recent multiple cut-off models developed by Steinhauser et al., Jones et al., and Hoyer and Kuss, that account for missing cut-offs when meta-analyzing diagnostic accuracy studies with multiple cut-offs, to BREMs fitted at each cut-off. We used data from 22 studies of the accuracy of the Edinburgh Postnatal Depression Scale (EPDS; 4475 participants, 758 major depression cases). We fitted each of the three multiple cut-off models and BREMs to a dataset with results from only published cut-offs from each study (published data) and an IPD dataset with results for all cut-offs (full IPD data). We estimated pooled sensitivity and specificity with 95% confidence intervals (CIs) for each cut-off and the area under the curve. Results: Compared to the BREMs fitted to the full IPD data, the Steinhauser et al., Jones et al., and Hoyer and Kuss models fitted to the published data produced similar receiver operating characteristic curves; though, the Hoyer and Kuss model had lower area under the curve, mainly due to estimating slightly lower sensitivity at lower cut-offs. When fitting the three multiple cut-off models to the full IPD data, a similar pattern of results was observed. Importantly, all models had similar 95% CIs for sensitivity and specificity, and the CI width increased with cut-off levels for sensitivity and decreased with an increasing cut-off for specificity, even the BREMs which treat each cut-off separately. Conclusions: Multiple cut-off models appear to be the favorable methods when only published data are available. While collecting IPD is expensive and time consuming, IPD can facilitate subgroup analyses that cannot be conducted with published data only. [ABSTRACT FROM AUTHOR]

Background Vaccination of infants with pneumococcal conjugate vaccines is recommended by the World Health Organization. Evidence is mixed regarding the differences in immunogenicity and efficacy of the different pneumococcal vaccines. Objectives The primary objective was to compare the immunogenicity of pneumococcal conjugate vaccine-10 versus pneumococcal conjugate vaccine-13. The main secondary objective was to compare the seroefficacy of pneumococcal conjugate vaccine-10 versus pneumococcal conjugate vaccine-13. Methods We searched the Cochrane Library, EMBASE, Global Health, MEDLINE, ClinicalTrials.gov and trialsearch.who.int up to July 2022. Studies were eligible if they directly compared either pneumococcal conjugate vaccine-7, pneumococcal conjugate vaccine-10 or pneumococcal conjugate vaccine-13 in randomised trials of children under 2 years of age, and provided immunogenicity data for at least one time point. Individual participant data were requested and aggregate data used otherwise. Outcomes included the geometric mean ratio of serotype-specific immunoglobulin G and the relative risk of seroinfection. Seroinfection was defined for each individual as a rise in antibody between the post-primary vaccination series time point and the booster dose, evidence of presumed subclinical infection. Each trial was analysed to obtain the log of the ratio of geometric means and its standard error. The relative risk of seroinfection (‘seroefficacy’) was estimated by comparing the proportion of participants with seroinfection between vaccine groups. The log-geometric mean ratios, log-relative risks and their standard errors constituted the input data for evidence synthesis. For serotypes contained in all three vaccines, evidence could be synthesised using a network meta-analysis. For other serotypes, meta-analysis was used. Results from seroefficacy analyses were incorporated into a mathematical model of pneumococcal transmission dynamics to compare the differential impact of pneumococcal conjugate vaccine-10 and pneumococcal conjugate vaccine-13 introduction on invasive pneumococcal disease cases. The model estimated the impact of vaccine introduction over a 25-year time period and an economic evaluation was conducted. Results In total, 47 studies were eligible from 38 countries. Twenty-eight and 12 studies with data available were included in immunogenicity and seroefficacy analyses, respectively. Geometric mean ratios comparing pneumococcal conjugate vaccine-13 versus pneumococcal conjugate vaccine-10 favoured pneumococcal conjugate vaccine-13 for serotypes 4, 9V and 23F at 1 month after primary vaccination series, with 1.14- to 1.54-fold significantly higher immunoglobulin G responses with pneumococcal conjugate vaccine-13. Risk of seroinfection prior to the time of booster dose was lower for pneumococcal conjugate vaccine-13 for serotype 4, 6B, 9V, 18C and 23F than for pneumococcal conjugate vaccine-10. Significant heterogeneity and inconsistency were present for most serotypes and for both outcomes. Twofold higher antibody after primary vaccination was associated with a 54% decrease in risk of seroinfection (relative risk 0.46, 95% confidence interval 0.23 to 0.96). In modelled scenarios, pneumococcal conjugate vaccine-13 or pneumococcal conjugate vaccine-10 introduction in 2006 resulted in a reduction in cases that was less rapid for pneumococcal conjugate vaccine-10 than for pneumococcal conjugate vaccine-13. The pneumococcal conjugate vaccine-13 programme was predicted to avoid an additional 2808 (95% confidence interval 2690 to 2925) cases of invasive pneumococcal disease compared with pneumococcal conjugate vaccine-10 introduction between 2006 and 2030. Limitations Analyses used data from infant vaccine studies with blood samples taken prior to a booster dose. The impact of extrapolating pre-booster efficacy to post-booster time points is unknown. Network meta-analysis models contained significant heterogeneity which may lead to bias. Conclusions Serotype-specific differences were found in immunogenicity and seroefficacy between pneumococcal conjugate vaccine-13 and pneumococcal conjugate vaccine-10. Higher antibody response after vaccination was associated with a lower risk of subsequent infection. These methods can be used to compare the pneumococcal conjugate vaccines and optimise vaccination strategies. For future work, seroefficacy estimates can be determined for other pneumococcal vaccines, which could contribute to licensing or policy decisions for new pneumococcal vaccines. Study registration This study is registered as PROSPERO CRD42019124580. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Health Technology Assessment programme (NIHR award ref: 17/148/03) and is published in full in Health Technology Assessment; Vol. 28, No. 34. See the NIHR Funding and Awards website for further award information. Plain language summary Pneumococcal disease is a serious illness caused by a bacterial infection that can result in death. Children in the United Kingdom receive a vaccine to prevent this disease that protects against 13 different types of pneumococcal diseases. It is very effective, but other vaccines are also available, such as one that contains 10 types of pneumococcal diseases. Vaccines in the United Kingdom are bought by the government and the choice of which vaccine to provide is based on the cost of the vaccine as well as the benefits to our health. However, there is very little information comparing different vaccines and it is often assumed they are the same. We did a large analysis combining all studies of the two main licensed pneumococcal vaccines to determine which vaccine provides better protection against infection and how this affects costs. We used information from studies published in medical journals, and also data from studies done by the companies that own the vaccines. Our results showed that pneumococcal conjugate vaccine-13 vaccine provided better protection than pneumococcal conjugate vaccine-10 for 5 of the 10 serotypes that are contained in both vaccines. When we used these results to model what might have happened had either of these vaccines been introduced into the United Kingdom vaccination programme in 2006, we found that both vaccines caused a rapid decrease in the amount of disease, but that the decrease in disease was faster with pneumococcal conjugate vaccine-13 than pneumococcal conjugate vaccine-10. This resulted in 2808 cases of diseases prevented over a 25-year time frame with pneumococcal conjugate vaccine-13 compared with pneumococcal conjugate vaccine-10. Our methods can be used to compare other vaccines and we recommend this type of study be done in future when making decisions on vaccine product choice. Scientific summary Streptococcus pneumoniae (pneumococcus) causes severe diseases, including bacterial pneumonia, meningitis and sepsis, leading to substantial morbidity and mortality worldwide, with the highest disease burden being in young children and older adults. Three pneumococcal conjugate vaccines (PCVs) have been widely deployed worldwide in the past two decades: PCV7 (Prevnar; Pfizer, headquartered in New York City, New York, USA), PCV10 (Synflorix; GlaxoSmithKline, headquartered in Brentford, London, UK) and PCV13 (Prevenar 13; Pfizer, headquartered in New York City, New York, USA), resulting in substantial reduction in disease. Between 2009 and 2011, PCV7 was gradually replaced by PCV13 and PCV10 and is no longer available. The World Health Organization (WHO) does not preferentially endorse one PCV over another. Both PCV13 and PCV10 have been shown to provide both direct and indirect protection against pneumococcal pneumonia, invasive pneumococcal disease and nasopharyngeal carriage. Although there are 10 common serotypes in these 2 vaccines, the components of the vaccines differ, with different carrier proteins used in the conjugation process, as well as different amounts of polysaccharide, and these differences may contribute to differences in protection. Large randomised controlled trials directly comparing different PCVs with invasive pneumococcal disease as the primary outcome are not feasible. We previously used ‘seroinfection’ as an outcome for analysis of PCVs, where seroinfection is defined as an increase in antibody levels between the primary vaccination series (typically complete at 5–7 months of age) and the booster dose (typically administered at 9–18 months of age). Seroinfection can be regarded as evidence of exposure to the pathogen and a resultant subclinical infection, given antibody responses wane rapidly during this period otherwise. Seroinfection rates for different vaccines can be compared by calculating the relative risk (RR) of seroinfection, referred to herein as ‘seroefficacy’. We meta-analysed data from studies of PCVs to compare the immunogenicity and seroefficacy of PCV10 with PCV13 for each serotype. We aimed to determine if serotype-specific immune responses were higher for either vaccine and whether this resulted in greater protection again seroinfection. In addition, we explored the overall relationship between the higher immune response and protection against seroinfection in infants. Following this, we show how serotype-specific estimates of seroefficacy can be incorporated in vaccine cost-effectiveness models. Objectives The primary objective of the systematic review was to compare the immunogenicity of PCV10 versus PCV13 for each serotype contained in the vaccines. The secondary objectives were: to compare the seroefficacy of PCV10 versus PCV13 for each serotype contained in the vaccines for PCV10 and PCV13 separately, to estimate immunogenicity and seroefficacy in comparison with the older PCV7 vaccine to determine how the comparisons of immunogenicity and efficacy of PCV10 to PCV13 are affected by the co-administration of different routine vaccines. Methods Systematic review We conducted a systematic review identifying studies that compared the immunogenicity of licensed PCVs in trials which randomised children to one of two different PCVs. The PCVs included in the review were PCV7 (Prevnar; Pfizer), PCV10 (Synflorix; GlaxoSmithKline) and PCV13 (Prevenar 13; Pfizer); PCV7 was included even though no longer available, so that we could compare PCV13 and PCV10 indirectly through them each being compared with PCV7 for the same serotypes. Data sources The databases searched were Cochrane Database of Systematic Reviews and Cochrane Central Register of Controlled Trials, EMBASE, Global Health and MEDLINE. The trial registers searched were ClinicalTrials.gov (https://clinicaltrials.gov/) and WHO International Clinical Trials Registry Platform (https://trialsearch.who.int/). The search comprised title/abstract keywords and subject headings for pneumococcal vaccines and children. A methodological search filter for randomised controlled trials taken from the Cochrane Handbook was used to limit to randomised controlled trials. Pharmaceutical company websites (GlaxoSmithKline and Pfizer) were also hand-searched for relevant studies. No date or language limits were applied. Study selection Randomised controlled trials were included if they provided direct comparisons of either PCV7, PCV10 or PCV13 among infants and children ˂ 2 years of age, and if they provided estimates of antibody responses [serotype-specific anti-pneumococcal immunoglobulin G (IgG) to PCVs for at least one time point of 1] between 4 and 6 weeks after the primary vaccination series and/or 1 month after a booster vaccination. Individual participant-level data were retrieved if available. Aggregate data from publications were extracted if individual participant data were not available. Risk of bias in results of the included studies was assessed independently by two reviewers using the Cochrane Risk of Bias Tool. Data synthesis Each trial with individual participant-level data available was analysed to obtain the log of the ratio of geometric means (log-GMR) and its standard error (SE) for each serotype and time point of interest. The RR of seroinfection was estimated by comparing the proportion of participants with seroinfection between vaccine groups. When no seroinfection occurred in any group (numerator of absolute risk was 0), a small non-zero value (0.5) was added to both sero-infected and sero-non-infected groups to allow estimation of the RR. The log-GMRs, log-RRs and their SEs constituted the input data for evidence synthesis. Only trials supplying individual participant data were included in seroefficacy analyses. For serotypes contained in all three vaccines, evidence could be synthesised using a network meta-analysis (NMA) of all comparisons. For other serotypes, meta-analysis was used for evidence synthesis. To estimate the overall association between antibody geometric mean ratio (GMR) and RR across all serotypes, we fitted a mixed-effect model regressing study-level RRs of seroinfection on GMRs across serotypes, weighted by the sample size of each study. Fixed effects included GMR, serotype and interactions between GMR and serotype (allowing serotype-specific association), while study was included as a random effect. Mathematical modelling and retrospective economic evaluation To illustrate the use of serotype-specific estimates of seroefficacy in modelling vaccine impact and cost-effectiveness, we developed a serotype-specific mathematical model of pneumococcal transmission dynamics to compare the differential impact of PCV10 and PCV13 introduction on invasive pneumococcal disease cases with vaccine serotypes in England and Wales. The model estimated the impact over a 25-year time period from 2006 to 2030. We subsequently assessed the cost-effectiveness of introducing infant vaccination with PCV13 compared with introducing PCV10 from a healthcare payer perspective in England and Wales. More specifically, we retrospectively estimated the additional threshold price per dose below which PCV13 would be more cost-effective than PCV10 had they both been available at the time of introduction of the PCV vaccine programme in England and Wales in 2006. Results Database registry and hand searches identified 4699 publication records of which 47 studies (78 publication reports) satisfied our eligibility criteria. Nineteen studies (24 publication reports) were excluded from the analysis: 6 studies did not provide individual patient or aggregate data and 13 studies (18 publication reports) were studies with the vaccines of interest, but it was not possible to form a loop within the NMA to provide indirect evidence. The remaining 28 studies (54 publication records) from 2009 to 2023 were included in the NMAs. Twenty-two studies provided individual participant data with a further five studies reporting aggregate data. Immunogenicity Geometric mean ratios for comparisons between PCV13 versus PCV10 for any primary series schedule were higher for PCV13 for serotypes 4, 7F, 9V and 23F at 1 month after primary vaccination series, with 1.14- to 1.54-fold higher IgG responses with PCV13. Additional serotypes contained only in the PCV13 vaccine (3, 6A and 19A) also favoured PCV13 as expected. GMRs were similar for the remaining serotypes (1, 5, 6B, 14, 18C and 19F). GMRs favoured PCV7 over either PCV13 or PCV10 for serotypes 4, 6B, 9V, 14 and 23F. There was no difference in GMRs for serotypes 18C and 19F across three vaccines. At the pre-booster time point, data were available from 18 cohorts. IgG responses were lower with PCV13 compared with PCV10 for all PCV7 serotypes except for serotype 14, with the point estimates of GMRs comparing PCV13 versus PCV10 ranging from 0.44 to 0.78. IgG responses were higher for PCV13 for serotypes 1, 5 and 7F. GMRs comparing PCV13 versus PCV7 showed higher IgG with PCV7 for serotypes 4, 6B, 9V, 14 and 23F and higher IgG with PCV13 for serotype 19F. At 28 days post booster, data were available from 26 cohorts. GMRs favoured PCV13 over PCV10 for serotype 6B, 9V, 14 and 23F and favoured PCV10 over PCV13 for serotype 18C. For serotype 1, 5 and 7F, antibody responses were higher in PCV13 compared with PCV10. PCV7 recipients had higher geometric mean concentrations (GMCs) compared with PCV13 for all PCV7 serotypes except 6B for which there was no difference, and 19F, which favoured PCV13. For PCV13-only serotypes (3, 6A and 19A), GMRs favour PCV13 at all three time points. Substantial heterogeneity and network inconsistency were present for most serotypes at all three time points. To explore potential reasons for the observed heterogeneity, we summarised cohort-level GMRs and RRs for each vaccine comparison. These descriptive analyses revealed a lack of consistency in the direction of study-level estimates within each vaccine comparison, resulting in the significant heterogenicity. There was also no observable pattern in any trial-level variable (region, co-administered vaccines, vaccine schedule), from which one might propose a mechanism that would adequately explain this variation in GMRs. Seroefficacy There were 12 studies (15 cohorts) with available individual participant antibody data at both post-primary and prior to the booster dose, allowing serotype-specific estimation of seroefficacy from a total of 5152 participants. Of these 15 cohorts, 6 compared PCV10 versus PCV7, 3 compared PCV13 versus PCV7 and 6 compared PCV13 versus PCV10. Among PCV7 serotypes, the risk of seroinfection was lower with PCV13 than PCV10 for serotypes 4, 6B, 9V, 18C and 23F, while no difference was seen for serotype 14 and 19F. The RRs of seroinfection (PCV13 vs. PCV10) for PCV7 serotypes ranged from 0.32 (95% CI 0.19 to 0.52) for serotype 4 to 1.28 (95% CI 0.95 to 1.74) for serotype 14. For serotypes 1, 5 and 7F, evidence was summarised from six studies directly comparing PCV13 with PCV10. Comparisons between PCV13 and PCV7 favoured neither vaccine over the other, whereas comparisons between PCV7 and PCV10 favoured PCV7 for serotypes 5, 6B, 9V, 18C and 23F. The I2 and p-values indicated some heterogeneity for all PCV7 serotypes except for serotype 4 and 19F. In the mixed-effects model of all serotypes combined, vaccines that produced the same amount of antibody (GMR = 1) had very similar protection (adjusted RR 0.80, 95% CI 0.41 to 1.58). The model estimate indicates that for each twofold increase in antibody response, the risk of seroinfection was halved (GMR of 2.0; RR 0.46, 95% CI 0.23 to 0.96). Mathematical model and economic evaluation Mathematical model results showed that in the absence of any vaccine programme, an increase in invasive pneumococcal disease cases caused by all five serotypes would be seen over the 25-year time frame. With the introduction of either PCV13 or PCV10 vaccine programmes in 2006, case counts would have decreased, achieving near eradication of all serotypes within the time frame modelled. The decrease in cases was most rapid for serotype 6B and least rapid for serotype 4. The decrease in cases was less rapid for PCV10 than for PCV13 due to the lower seroefficacy. The introduction of an infant PCV13 programme was predicted to avoid an additional 2808 (95% CI 2690 to 2925) cases of invasive pneumococcal disease compared with PCV10 introduction between 2006 and 2030. This includes an estimated 326 cases of meningitis, 578 cases of sepsis, 1770 cases of invasive pneumonia and 30,680 cases of non-invasive pneumonia. Under base-case assumptions, this resulted in discounted healthcare savings of £13 million (95% CI £12 to £14 million). Including non-invasive pneumonia increased the savings to £27 million (95% CI £25 to £29 million). Conclusions In our study, we used a novel methodology to define seroinfection from immunogenicity data to compare the relative efficacy of PCVs in preventing infection. Our results using individual-level data from a global meta-analysis provide the first estimates of the comparative protection afforded by different pneumococcal vaccines and show that for many serotypes, carriage events are less common after PCV13 than PCV10, likely due to a higher antibody response. In addition, we quantify the relationship between the immune response to vaccination and protection against infection, measured serologically, and show that higher antibody responses in infants are associated with greater protection from infection. Licensure of new vaccines is based on non-inferiority comparisons with current vaccines and the proportion of antibody responses above the agreed threshold as a minimum requirement. Once a vaccine meets this ‘at-least-as-good-as’ immunogenicity criteria, it has previously not been clear whether exceeding it is of benefit, and the WHO position paper on pneumococcal vaccines states ‘It is unknown whether a lower serotype-specific GMC of antibody indicates less efficacy’. Our results show that lower protection against subclinical infection does indeed follow from lower antibody production and that two vaccines that produce a similar level of antibody will provide similar levels of protection. The implications of these findings are of greatest importance when a new vaccine roll-out is being considered. Lower antibody production or lower seroefficacy for one vaccine product does not necessarily imply limited effectiveness against invasive pneumococcal diseases when considering vaccines such as PCV10 and PCV13 which are highly effective vaccines in many settings. Instead, lower antibody responses lead to less rapidly observed indirect protection after implementation into a national programme as a smaller proportion of transmission events are blocked by the vaccine. This is evident in the mathematical modelling which showed less rapid decreases in the number of cases of invasive disease when introducing PCV10 compared with PCV13. Implications for practice This evidence of differences in serotype-specific protection can be incorporated into cost-effectiveness models used to compare vaccine products. Cost-effectiveness studies have highlighted the lack of evidence of comparative efficacy for different PCVs, resulting in previous cost-effectiveness models that ignore serotype-specific differences and assume equivalent efficacy for all serotypes covered by different PCVs. Our study fills this evidence gap and allows researchers and policy-makers to use more accurate vaccine-specific models in decision-making. Our cost-effectiveness analysis of a hypothetical scenario showed that introducing infant PCV13 was predicted to avert a higher burden of pneumococcal disease compared with PCV10. This would have realised a small saving of £13 million discounted over 24 years. When considering the introduction of new pneumococcal vaccines into the routine immunisation schedule, we recommend that differences in antibody responses for different vaccines be considered in modelling scenarios as higher antibody responses result in reduced transmission and greater impact on invasive diseases. Vaccine-specific threshold prices can then be determined for cost-effective vaccines. Our analysis showed that due to its higher efficacy against some serotypes, a higher threshold price per dose could be paid for PCV13 while remaining cost-effective. Study registration This study is registered as PROSPERO CRD42019124580. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Health Technology Assessment programme (NIHR award ref: 17/148/03) and is published in full in Health Technology Assessment; Vol. 28, No. 34. See the NIHR Funding and Awards website for further award information.

Abstract Background Depression is a highly common and recurrent condition. Predicting who is at most risk of relapse or recurrence can inform clinical practice. Applying machine-learning methods to Individual Participant Data (IPD) can be promising to improve the accuracy of risk predictions. Methods Individual data of four Randomized Controlled Trials (RCTs) evaluating antidepressant treatment compared to psychological interventions with tapering ( $$N=714$$ N = 714 ) were used to identify predictors of relapse and/or recurrence. Ten baseline predictors were assessed. Decision trees with and without gradient boosting were applied. To study the robustness of decision-tree classifications, we also performed a complementary logistic regression analysis. Results The combination of age, age of onset of depression, and depression severity significantly enhances the prediction of relapse risk when compared to classifiers solely based on depression severity. The studied decision trees can (i) identify relapse patients at intake with an accuracy, specificity, and sensitivity of about 55% (without gradient boosting) and 58% (with gradient boosting), and (ii) slightly outperform classifiers that are based on logistic regression. Conclusions Decision tree classifiers based on multiple–rather than single–risk indicators may be useful for developing treatment stratification strategies. These classification models have the potential to contribute to the development of methods aimed at effectively prioritizing treatment for those individuals who require it the most. Our results also underline the existing gaps in understanding how to accurately predict depressive relapse.

BackgroundIn non-randomized studies (NRSs) where a continuous outcome variable (e.g., depressive symptoms) is assessed at baseline and follow-up, it is common to observe imbalance of the baseline values between the treatment/exposure group and control group. This may bias the study and consequently a meta-analysis (MA) estimate. These estimates may differ across statistical methods used to deal with this issue. Analysis of individual participant data (IPD) allows standardization of methods across studies. We aimed to identify methods used in published IPD-MAs of NRSs for continuous outcomes, and to compare different methods to account for baseline values of outcome variables in IPD-MA of NRSs using two empirical examples from the Thyroid Studies Collaboration (TSC).MethodsFor the first aim we systematically searched in MEDLINE, EMBASE, and Cochrane from inception to February 2021 to identify published IPD-MAs of NRSs that adjusted for baseline outcome measures in the analysis of continuous outcomes. For the second aim, we applied analysis of covariance (ANCOVA), change score, propensity score and the naïve approach (ignores the baseline outcome data) in IPD-MA from NRSs on the association between subclinical hyperthyroidism and depressive symptoms and renal function. We estimated the study and meta-analytic mean difference (MD) and relative standard error (SE). We used both fixed- and random-effects MA.ResultsTen of 18 (56%) of the included studies used the change score method, seven (39%) studies used ANCOVA and one the propensity score (5%). The study estimates were similar across the methods in studies in which groups were balanced at baseline with regard to outcome variables but differed in studies with baseline imbalance. In our empirical examples, ANCOVA and change score showed study results on the same direction, not the propensity score. In our applications, ANCOVA provided more precise estimates, both at study and meta-analytical level, in comparison to other methods. Heterogeneity was higher when change score was used as outcome, moderate for ANCOVA and null with the propensity score.ConclusionANCOVA provided the most precise estimates at both study and meta-analytic level and thus seems preferable in the meta-analysis of IPD from non-randomized studies. For the studies that were well-balanced between groups, change score, and ANCOVA performed similarly.

Individual participant data (IPD) meta‐analyses of randomised trials are considered a reliable way to assess participant‐level treatment effect modifiers but may not make the best use of the available data. Traditionally, effect modifiers are explored one covariate at a time, which gives rise to the possibility that evidence of treatment‐covariate interaction may be due to confounding from a different, related covariate. We aimed to evaluate current practice when estimating treatment‐covariate interactions in IPD meta‐analysis, specifically focusing on involvement of additional covariates in the models. We reviewed 100 IPD meta‐analyses of randomised trials, published between 2015 and 2020, that assessed at least one treatment‐covariate interaction. We identified four approaches to handling additional covariates: (1) Single interaction model (unadjusted): No additional covariates included (57/100 IPD meta‐analyses); (2) Single interaction model (adjusted): Adjustment for the main effect of at least one additional covariate (35/100); (3) Multiple interactions model: Adjustment for at least one two‐way interaction between treatment and an additional covariate (3/100); and (4) Three‐way interaction model: Three‐way interaction formed between treatment, the additional covariate and the potential effect modifier (5/100). IPD is not being utilised to its fullest extent. In an exemplar dataset, we demonstrate how these approaches lead to different conclusions. Researchers should adjust for additional covariates when estimating interactions in IPD meta‐analysis providing they adjust their main effects, which is already widely recommended. Further, they should consider whether more complex approaches could provide better information on who might benefit most from treatments, improving patient choice and treatment policy and practice. [ABSTRACT FROM AUTHOR]

Introduction: Numerous arguments have been advanced for broadly sharing de‐identified, participant‐level clinical trial data. However, data sharing in pragmatic clinical trials (PCTs) presents ethical challenges. While prior scholarship has described aspects of PCTs that raise distinct considerations for data sharing, there have been no reports of the experiences of those at the leading edge of data‐sharing efforts for PCTs, including how these particular challenges have been navigated. To address this gap, we conducted interviews with key stakeholders, with a focus on the ethical issues presented by sharing data from PCTs. Methods: We recruited respondents using purposive sampling to reflect the range of stakeholder groups affected by efforts to expand PCT data sharing. Through semi‐structured interviews, we explored respondents' experiences and perceptions about sharing de‐identified, individual‐level data from PCTs. An integrated approach was used to identify and describe key themes. Results: We conducted 40 interviews between April and September 2022. Five overarching themes emerged through analysis: (1) challenges in sharing data collected under a waiver or alteration of consent; (2) conflicting views regarding PCT patient‐subject preferences for data sharing; (3) identification of respect‐promoting practices beyond consent; (4) concerns about elevated risks or burdens from sharing PCT data; and (5) diverse views about the likely benefits resulting from sharing PCT data. Conclusion: Our data indicate unresolved tensions in how to fulfill the expectation to broadly share de‐identified, individual‐level data from PCTs, and suggest that those promulgating and implementing data‐sharing policies must be sensitive to PCT‐specific considerations. Future work could inform efforts to tailor data‐sharing policy and practice to reflect the challenges presented by PCTs, including sharing experiences from trials that have successfully navigated these tensions. [ABSTRACT FROM AUTHOR]

Before being used in clinical practice, a prediction model should be tested in patients whose data were not used in model development. Previously, we developed the AD F ICE_IT models for predicting any fall and recurrent falls, referred as Any_fall and Recur_fall. In this study, we externally validated the models and compared their clinical value to a practical screening strategy where patients are screened for falls history alone. Retrospective, combined analysis of 2 prospective cohorts. Data were included of 1125 patients (aged ≥65 years) who visited the geriatrics department or the emergency department. We evaluated the models' discrimination using the C-statistic. Models were updated using logistic regression if calibration intercept or slope values deviated significantly from their ideal values. Decision curve analysis was applied to compare the models' clinical value (ie, net benefit) against that of falls history for different decision thresholds. During the 1-year follow-up, 428 participants (42.7%) endured 1 or more falls, and 224 participants (23.1%) endured a recurrent fall (≥2 falls). C-statistic values were 0.66 (95% CI 0.63-0.69) and 0.69 (95% CI 0.65-0.72) for the Any_fall and Recur_fall models, respectively. Any_fall overestimated the fall risk and we therefore updated only its intercept whereas Recur_fall showed good calibration and required no update. Compared with falls history, Any_fall and Recur_fall showed greater net benefit for decision thresholds of 35% to 60% and 15% to 45%, respectively. The models performed similarly in this data set of geriatric outpatients as in the development sample. This suggests that fall-risk assessment tools that were developed in community-dwelling older adults may perform well in geriatric outpatients. We found that in geriatric outpatients the models have greater clinical value across a wide range of decision thresholds compared with screening for falls history alone. [ABSTRACT FROM AUTHOR]

Background: Depression is a highly common and recurrent condition. Predicting who is at most risk of relapse or recurrence can inform clinical practice. Applying machine-learning methods to Individual Participant Data (IPD) can be promising to improve the accuracy of risk predictions. Methods: Individual data of four Randomized Controlled Trials (RCTs) evaluating antidepressant treatment compared to psychological interventions with tapering ( N = 714 ) were used to identify predictors of relapse and/or recurrence. Ten baseline predictors were assessed. Decision trees with and without gradient boosting were applied. To study the robustness of decision-tree classifications, we also performed a complementary logistic regression analysis. Results: The combination of age, age of onset of depression, and depression severity significantly enhances the prediction of relapse risk when compared to classifiers solely based on depression severity. The studied decision trees can (i) identify relapse patients at intake with an accuracy, specificity, and sensitivity of about 55% (without gradient boosting) and 58% (with gradient boosting), and (ii) slightly outperform classifiers that are based on logistic regression. Conclusions: Decision tree classifiers based on multiple–rather than single–risk indicators may be useful for developing treatment stratification strategies. These classification models have the potential to contribute to the development of methods aimed at effectively prioritizing treatment for those individuals who require it the most. Our results also underline the existing gaps in understanding how to accurately predict depressive relapse. [ABSTRACT FROM AUTHOR]

Prior research suggests that brief interventions (BIs) for alcohol and other drug use may vary in effectiveness across patient sociodemographic factors. The objective of this individual participant data (IPD) meta-analysis was to explore for whom BIs delivered in general healthcare settings are more or less effective. We examined variability in BI effects by patient age, sex, employment, education, relationship status, and baseline severity of substance use using a two-stage IPD meta-analysis approach. All trials included in a parent aggregate data meta-analysis (k = 116) were invited to contribute IPD, and 29 trials provided patient-level data (12,074 participants). Among females, BIs led to significant reductions in binge alcohol consumption ( g ¯ = 0.09, 95% CI [0.03, 0.14]), frequency of alcohol consumption ( g ¯ = 0.10, 95% CI [0.03, 0.17]), and alcohol-related consequences ( g ¯ = 0.16, 95% CI [0.08, 0.25]), as well as greater substance use treatment utilization ( g ¯ = 0.25, 95% CI [0.21, 0.30]). BIs yielded larger reductions in frequency of alcohol consumption at 3-month follow-up for individuals with less than a high school level education ( g ¯ = 0.16, 95% CI [0.09, 0.22]). Given evidence demonstrating modest BI effects on alcohol use and mixed or null findings for BI effects on other drug use, BI research should continue to investigate potential drivers of effect magnitude and variation. Protocol registration details: The protocol for this review was pre-registered in PROSPERO #CRD42018086832 and the analysis plan was pre-registered in OSF: osf.io/m48g6. [ABSTRACT FROM AUTHOR]

Evidence synthesis involves drawing conclusions from trial samples that may differ from the target population of interest, and there is often heterogeneity among trials in sample characteristics, treatment implementation, study design, and assessment of covariates. Stitching together this patchwork of evidence requires subject-matter knowledge, a clearly defined target population, and guidance on how to weigh evidence from different trials. Transportability analysis has provided formal identifiability conditions required to make unbiased causal inference in the target population. In this manuscript, we review these conditions along with an additional assumption required to address systematic missing data. The identifiability conditions highlight the importance of accounting for differences in treatment effect modifiers between the populations underlying the trials and the target population. We perform simulations to evaluate the bias of conventional random effect models and multiply imputed estimates using the pooled trials sample and describe causal estimators that explicitly address trial-to-target differences in key covariates in the context of systematic missing data. Results indicate that the causal transportability estimators are unbiased when treatment effect modifiers are accounted for in the analyses. Results also highlight the importance of carefully evaluating identifiability conditions for each trial to reduce bias due to differences in participant characteristics between trials and the target population. Bias can be limited by adjusting for covariates that are strongly correlated with missing treatment effect modifiers, including data from trials that do not differ from the target on treatment modifiers, and removing trials that do differ from the target and did not assess a modifier. [ABSTRACT FROM AUTHOR]

To evaluate and optimize brief alcohol interventions (BAIs), it is critical to have a credible overall effect size estimate as a benchmark. Estimating such an effect size has been challenging because alcohol outcomes often represent responses from a mixture of individuals: those at high risk for alcohol misuse, occasional nondrinkers, and abstainers. Moreover, some BAIs exclusively focus on heavy drinkers, whereas others take a universal prevention approach. Depending on sample characteristics, the outcome distribution might have many zeros or very few zeros and overdispersion; consequently, the most appropriate statistical model may differ across studies. We synthesized individual participant data (IPD) from 19 studies in Project INTEGRATE (Mun et al., 2015b) that randomly allocated participants to intervention and control groups (N = 7,704 participants, 38.4% men, 74.7% White, 58.5% first-year students). We sequentially estimated marginalized zero-inflated Poisson (Long et al., 2014) or negative binomial regression models to obtain covariate-adjusted, study-specific intervention effect estimates in the first step, which were subsequently combined in a random-effects meta-analysis model in the second step. BAIs produced a statistically significant 8% advantage in the mean number of drinks at both 1–3 months (RR = 0.92, 95% CI = [0.85, 0.98]) and 6 months (RR = 0.92, 95% CI = [0.85, 0.99]) compared to controls. At 9–12 months, there was no statistically significant difference in the mean number of drinks between BAIs and controls. In conclusion, BAIs are effective at reducing the mean number of drinks through at least 6 months post intervention. IPD can play a critical role in deriving findings that could not be obtained in original individual studies or standard aggregate data meta-analyses. [ABSTRACT FROM AUTHOR]

Introduction: Evidence comparing double‐balloon vs single‐balloon catheter for induction of labor is divided. We aim to compare the efficacy and safety of double‐vs single‐balloon catheters using individual participant data. Material and methods: A search of Ovid MEDLINE, Embase, Ovid Emcare, CINAHL Plus, Scopus, and clinicaltrials.gov was conducted for randomized controlled trials published from March 2019 until April 13, 2021. Earlier trials were identified from the Cochrane Review on Mechanical Methods for Induction of Labour. Randomized controlled trials that compared double‐balloon with single‐balloon catheters for induction of labor in singleton gestations were eligible. Participant‐level data were sought from trial investigators and an individual participant data meta‐analysis was performed. The primary outcomes were rates of vaginal birth achieved, a composite measure of adverse maternal outcomes and a composite measure of adverse perinatal outcomes. We used a two‐stage random‐effects model. Data were analyzed from the intention‐to‐treat perspective. Results: Of the eight eligible randomized controlled trials, three shared individual‐level data with a total of 689 participants, 344 women in the double‐balloon catheter group and 345 women in the single‐balloon catheter group. The difference in the rate of vaginal birth between double‐balloon catheter and single‐balloon catheter was not statistically significant (relative risk [RR] 0.93, 95% confidence interval [CI] 0.86–1.00, p = 0.050; I2 0%; moderate‐certainty evidence). Both perinatal outcomes (RR 0.81, 95% CI 0.54–1.21, p = 0.691; I2 0%; moderate‐certainty evidence) and maternal composite outcomes (RR 0.65, 95% CI 0.15–2.87, p = 0.571; I2 55.46%; low‐certainty evidence) were not significantly different between the two groups. Conclusions: Single‐balloon catheter is at least comparable to double‐balloon catheter in terms of vaginal birth rate and maternal and perinatal safety outcomes. [ABSTRACT FROM AUTHOR]

Background: With the release of the text revision of the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5-TR), criteria for Prolonged Grief Disorder (PGD) were included. This necessitates studying grief trajectories based on these criteria. Objective: This is the first study examining latent trajectories of DSM-5-TR-based PGD symptom levels and testing whether specific risk factors (e.g. cause of death) predicted PGD trajectories. Method: We evaluated latent DSM-5-TR PGD trajectories using pooled existing data collected at 6–12, 13–24, and 25–60 months post-loss in Danish and Dutch bereaved adults (N = 398). Latent Growth Mixture Modelling (LGMM) was employed to determine the trajectories. Multinomial logistic regression analyses were used to examine which risk factors predicted class membership. Results: The four-class LGMM solution with a quadratic term was best-fitting the data. This solution represented four trajectories: High stable PGD (6%), High PGD quick recovery (10%), High PGD slow recovery (35%), and Low PGD symptoms (49%). Participants with a higher educational level were more likely to be assigned to the Low PGD symptoms trajectory compared to High stable PGD and High PGD slow recovery trajectories. Unnatural causes of death increased the likelihood of being in the High stable PGD and High PGD slow recovery trajectories compared to the Low PGD symptoms trajectory. Conclusions: Consistent with prior research, the Low PGD symptoms trajectory was the most common. A significant minority experienced high and stable levels of PGD within five years after the loss. About one-third of participants experienced high acute grief levels that decreased slowly; how slow decreasing symptoms relate to an individual's functioning requires further attention. This study demonstrates that a significant minority of bereaved people develop acute PGD symptomatology that does not diminish within five years post-loss, emphasizing the need for early screening for PGD to prevent long-lasting complaints. This is the first latent trajectory study based on DSM-5-TR Prolonged Grief Disorder (PGD) criteria. Data were analysed using latent growth mixture modelling. Stable high (6%), quick recovery (10%), slow recovery (35%), low symptoms (49%) PGD trajectories arose. Early screening and treatment of PGD seems warranted. [ABSTRACT FROM AUTHOR]

Objective: Identify independent and novel risk factors for late‐preterm (28–36 weeks) and term (≥37 weeks) stillbirth and explore development of a risk‐prediction model. Design: Secondary analysis of an Individual Participant Data (IPD) meta‐analysis investigating modifiable stillbirth risk factors. Setting: An IPD database from five case–control studies in New Zealand, Australia, the UK and an international online study. Population: Women with late‐stillbirth (cases, n = 851), and ongoing singleton pregnancies from 28 weeks' gestation (controls, n = 2257). Methods: Established and novel risk factors for late‐preterm and term stillbirth underwent univariable and multivariable logistic regression modelling with multiple sensitivity analyses. Variables included maternal age, body mass index (BMI), parity, mental health, cigarette smoking, second‐hand smoking, antenatal‐care utilisation, and detailed fetal movement and sleep variables. Main outcome measures: Independent risk factors with adjusted odds ratios (aOR) for late‐preterm and term stillbirth. Results: After model building, 575 late‐stillbirth cases and 1541 controls from three contributing case–control studies were included. Risk factor estimates from separate multivariable models of late‐preterm and term stillbirth were compared. As these were similar, the final model combined all late‐stillbirths. The single multivariable model confirmed established demographic risk factors, but additionally showed that fetal movement changes had both increased (decreased frequency) and reduced (hiccoughs, increasing strength, frequency or vigorous fetal movements) aOR of stillbirth. Poor antenatal‐care utilisation increased risk while more‐than‐adequate care was protective. The area‐under‐the‐curve was 0.84 (95% CI 0.82–0.86). Conclusions: Similarities in risk factors for late‐preterm and term stillbirth suggest the same approach for risk‐assessment can be applied. Detailed fetal movement assessment and inclusion of antenatal‐care utilisation could be valuable in late‐stillbirth risk assessment. This article includes Author Insights, a video abstract available at: https://players.brightcove.net/3806881048001/default%5fdefault/index.html?videoId=6320088103112. Linked article: This article is commented on by J. Jardine, pp. 1071 in this issue. To view this mini commentary visit https://doi.org/10.1111/1471‐0528.17469. [ABSTRACT FROM AUTHOR]

Background: Brief motivational interventions (BMIs) are one of the most effective individually focused alcohol intervention strategies for college students. Despite the central theoretical role of motivation for change in BMIs, it is unclear whether BMIs increase motivation to change drinking behavior. We conducted a two‐step meta‐analysis of individual participant data (IPD) to examine whether BMIs increase motivation for change. N = 5903;59% women, 72% White) from Project INTEGRATE. The BMIs included individually delivered motivational interviewing with personalized feedback (MI + PF), stand‐alone personalized feedback (PF), and group‐based motivational interviewing (GMI). Methods: We included 15 trials of BMI (N = 5903;59% women, 72% White) from Project INTEGRATE. The BMIs included individually‐delivered motivational interviewing with personalized feedback (MI + PF), stand‐alone personalized feedback (PF), and group‐based motivational interviewing (GMI). Different measures and responses used in the original trials were harmonized. Effect size estimates were derived from a model that adjusted for baseline motivation and demographic variables for each trial (step 1) and subsequently combined in a random‐effects meta‐analysis (step 2). Results: The overall intervention effect of BMIs on motivation for change was not statistically significant (standard mean difference [SMD]: 0.026, 95% CI: [−0.001, 0.053], p = 0.06, k = 19 comparisons). Of the three subtypes of BMIs, GMI, which tended to provide motivation‐targeted content, had a statistically significant intervention effect on motivation, compared with controls (SMD: 0.055, 95% CI: [0.007, 0.103], p = 0.025, k = 5). By contrast, there was no evidence that MI + PF (SMD = 0.04, 95% CI: [−0.02, 0.10], k = 6, p = 0.20) nor PF increased motivation (SMD = 0.005, 95% CI: [−0.028, 0.039], k = 8, p = 0.75), compared with controls. Post hoc meta‐regression analysis suggested that motivation sharply decreased each month within the first 3 months postintervention (b = −0.050, z = −2.80, p = 0.005 for k = 14). Conclusions: Although BMIs provide motivational content and normative feedback and are assumed to motivate behavior change, the results do not wholly support the hypothesis that BMIs improve motivation for change. Changing motivation is difficult to assess during and following interventions, but it is still a theoretically important clinical endpoint. Further, the evidence cautiously suggests that changing motivation may be achievable, especially if motivation‐targeted content components are provided. [ABSTRACT FROM AUTHOR]

Summary: Background: The association between cannabis use and positive symptoms in schizophrenia spectrum disorders is well documented, especially via meta-analyses. Yet, findings are inconsistent regarding negative symptoms, while other dimensions such as disorganization, depression, and excitement, have not been investigated. In addition, meta-analyses use aggregated data discarding important confounding variables which is a source of bias. Methods: PubMed, ScienceDirect and PsycINFO were used to search for publications from inception to September 27, 2022. We contacted the authors of relevant studies to extract raw datasets and perform an Individual Participant Data meta-analysis (IPDMA). Inclusion criteria were: psychopathology of individuals with schizophrenia spectrum disorders assessed by the Positive and Negative Syndrome Scale (PANSS); cannabis-users had to either have a diagnosis of cannabis use disorder or use cannabis at least twice a week. The main outcomes were the PANSS subscores extracted via the 3-factor (positive, negative and general) and 5-factor (positive, negative, disorganization, depression, excitement) structures. Preregistration is accessible via Prospero: ID CRD42022329172. Findings: Among the 1149 identified studies, 65 were eligible and 21 datasets were shared, totaling 3677 IPD and 3053 complete cases. The adjusted multivariate analysis revealed that relative to non-use, cannabis use was associated with higher severity of positive dimension (3-factor: Adjusted Mean Difference, aMD = 0.34, 95% Confidence Interval, CI = [0.03; 0.66]; 5-factor: aMD = 0.38, 95% CI = [0.08; 0.63]), lower severity of negative dimension (3-factor: aMD = −0.49, 95% CI [−0.90; −0.09]; 5-factor: aMD = −0.50, 95% CI = [−0.91; −0.08]), higher severity of excitement dimension (aMD = 0.16, 95% CI = [0.03; 0.28]). No association was found between cannabis use and disorganization (aMD = −0.13, 95% CI = [−0.42; 0.17]) or depression (aMD = −0.14, 95% CI = [−0.34; 0.06]). Interpretation: No causal relationship can be inferred from the current results. The findings could be in favor of both a detrimental and beneficial effect of cannabis on positive and negative symptoms, respectively. Longitudinal designs are needed to understand the role of cannabis is this association. The reported effect sizes are small and CIs are wide, the interpretation of findings should be taken with caution. Funding: This research did not receive any specific grant or funding. Primary financial support for authors was provided by Le Vinatier Psychiatric Hospital.

Summary: Background: Vaccination of infants with pneumococcal conjugate vaccines (PCV) is recommended by the World Health Organization. Evidence is mixed regarding the differences in immunogenicity and efficacy of the different pneumococcal vaccines. Methods: In this systematic-review and network meta-analysis, we searched the Cochrane Library, Embase, Global Health, Medline, clinicaltrials.gov and trialsearch.who.int up to February 17, 2023 with no language restrictions. Studies were eligible if they presented data comparing the immunogenicity of either PCV7, PCV10 or PCV13 in head-to-head randomised trials of young children under 2 years of age, and provided immunogenicity data for at least one time point after the primary vaccination series or the booster dose. Publication bias was assessed via Cochrane's Risk Of Bias due to Missing Evidence tool and comparison-adjusted funnel plots with Egger's test. Individual participant level data were requested from publication authors and/or relevant vaccine manufacturers. Outcomes included the geometric mean ratio (GMR) of serotype-specific IgG and the relative risk (RR) of seroinfection. Seroinfection was defined for each individual as a rise in antibody between the post-primary vaccination series time point and the booster dose, evidence of presumed subclinical infection. Seroefficacy was defined as the RR of seroinfection. We also estimated the relationship between the GMR of IgG one month after priming and the RR of seroinfection by the time of the booster dose. The protocol is registered with PROSPERO, ID CRD42019124580. Findings: 47 studies were eligible from 38 countries across six continents. 28 and 12 studies with data available were included in immunogenicity and seroefficacy analyses, respectively. GMRs comparing PCV13 vs PCV10 favoured PCV13 for serotypes 4, 9V, and 23F at 1 month after primary vaccination series, with 1.14- to 1.54- fold significantly higher IgG responses with PCV13. Risk of seroinfection prior to the time of booster dose was lower for PCV13 for serotype 4, 6B, 9V, 18C and 23F than for PCV10. Significant heterogeneity and inconsistency were present for most serotypes and for both outcomes. Two-fold higher antibody after primary vaccination was associated with a 54% decrease in risk of seroinfection (RR 0.46, 95% CI 0.23–0.96). Interpretation: Serotype-specific differences were found in immunogenicity and seroefficacy between PCV13 and PCV10. Higher antibody response after vaccination was associated with a lower risk of subsequent infection. These findings could be used to compare PCVs and optimise vaccination strategies. Funding: The NIHR Health Technology Assessment Programme.

Background: Bacterial vaginosis (BV) increases preterm delivery (PTD) risk, but treatment trials showed mixed results in preventing PTD. Objectives: Determine, using individual participant data (IPD), whether BV treatment during pregnancy reduced PTD or prolonged time‐to‐delivery. Data Sources: Cochrane Systematic Review (2013), MEDLINE, EMBASE, journal searches, and searches (January 2013–September 2022) ("bacterial vaginosis AND pregnancy") of (i) clinicaltrials.gov; (ii) Cochrane Central Register of Controlled Trials; (iii) World Health Organization International Clinical Trials Registry Platform Portal; and (iv) Web of Science ("bacterial vaginosis"). Study Selection and Data Extraction: Studies randomising asymptomatic pregnant individuals with BV to antibiotics or control, measuring delivery gestation. Extraction was from original data files. Bias risk was assessed using the Cochrane tool. Analysis used "one‐step" logistic and Cox random effect models, adjusting gestation at randomisation and PTD history; heterogeneity by I2. Subgroup analysis tested interactions with treatment. In sensitivity analyses, studies not providing IPD were incorporated by "multiple random‐donor hot‐deck" imputation, using IPD studies as donors. Results: There were 121 references (96 studies) with 23 eligible trials (11,979 participants); 13 studies (6915 participants) provided IPD; 12 (6115) were incorporated. Results from 9 (4887 participants) not providing IPD were imputed. Odds ratios for PTD for metronidazole and clindamycin versus placebo were 1.00 (95% CI 0.84, 1.17), I2 = 62%, and 0.59 (95% CI 0.42, 0.82), I2 = 0 before; and 0.95 (95% CI 0.81, 1.11), I2 = 59%, and 0.90 (95% CI: 0.72, 1.12), I2 = 0, after imputation. Time‐to‐delivery did not differ from null with either treatment. Including imputed IPD, there was no evidence that either drug was more effective when administered earlier, or among those with a PTD history. Conclusions: Clindamycin, but not metronidazole, was beneficial in studies providing IPD, but after imputing data from missing IPD studies, treatment of BV during pregnancy did not reduce PTD, nor prolong pregnancy, in any subgroup or when started earlier in gestation. [ABSTRACT FROM AUTHOR]

A common problem in the analysis of multiple data sources, including individual participant data meta‐analysis (IPD‐MA), is the misclassification of binary variables. Misclassification may lead to biased estimators of model parameters, even when the misclassification is entirely random. We aimed to develop statistical methods that facilitate unbiased estimation of adjusted and unadjusted exposure‐outcome associations and between‐study heterogeneity in IPD‐MA, where the extent and nature of exposure misclassification may vary across studies. We present Bayesian methods that allow misclassification of binary exposure variables to depend on study‐ and participant‐level characteristics. In an example of the differential diagnosis of dengue using two variables, where the gold standard measurement for the exposure variable was unavailable for some studies which only measured a surrogate prone to misclassification, our methods yielded more accurate estimates than analyses naive with regard to misclassification or based on gold standard measurements alone. In a simulation study, the evaluated misclassification model yielded valid estimates of the exposure‐outcome association, and was more accurate than analyses restricted to gold standard measurements. Our proposed framework can appropriately account for the presence of binary exposure misclassification in IPD‐MA. It requires that some studies supply IPD for the surrogate and gold standard exposure, and allows misclassification to follow a random effects distribution across studies conditional on observed covariates (and outcome). The proposed methods are most beneficial when few large studies that measured the gold standard are available, and when misclassification is frequent. [ABSTRACT FROM AUTHOR]

Abstract Background Increasingly in network meta-analysis (NMA), there is a need to incorporate non-randomised evidence to estimate relative treatment effects, and in particular in cases with limited randomised evidence, sometimes resulting in disconnected networks of treatments. When combining different sources of data, complex NMA methods are required to address issues associated with participant selection bias, incorporating single-arm trials (SATs), and synthesising a mixture of individual participant data (IPD) and aggregate data (AD). We develop NMA methods which synthesise data from SATs and randomised controlled trials (RCTs), using a mixture of IPD and AD, for a dichotomous outcome. Methods We propose methods under both contrast-based (CB) and arm-based (AB) parametrisations, and extend the methods to allow for both within- and across-trial adjustments for covariate effects. To illustrate the methods, we use an applied example investigating the effectiveness of biologic disease-modifying anti-rheumatic drugs for rheumatoid arthritis (RA). We applied the methods to a dataset obtained from a literature review consisting of 14 RCTs and an artificial dataset consisting of IPD from two SATs and AD from 12 RCTs, where the artificial dataset was created by removing the control arms from the only two trials assessing tocilizumab in the original dataset. Results Without adjustment for covariates, the CB method with independent baseline response parameters (CBunadjInd) underestimated the effectiveness of tocilizumab when applied to the artificial dataset compared to the original dataset, albeit with significant overlap in posterior distributions for treatment effect parameters. The CB method with exchangeable baseline response parameters produced effectiveness estimates in agreement with CBunadjInd, when the predicted baseline response estimates were similar to the observed baseline response. After adjustment for RA duration, there was a reduction in across-trial heterogeneity in baseline response but little change in treatment effect estimates. Conclusions Our findings suggest incorporating SATs in NMA may be useful in some situations where a treatment is disconnected from a network of comparator treatments, due to a lack of comparative evidence, to estimate relative treatment effects. The reliability of effect estimates based on data from SATs may depend on adjustment for covariate effects, although further research is required to understand this in more detail.

Descriptive analyses of socially important or theoretically interesting phenomena and trends are a vital component of research in the behavioral, social, economic, and health sciences. Such analyses yield reliable results when using representative individual participant data (IPD) from studies with complex survey designs, including educational large‐scale assessments (ELSAs) or social, health, and economic survey and panel studies. The meta‐analytic integration of these results offers unique and novel research opportunities to provide strong empirical evidence of the consistency and generalizability of important phenomena and trends. Using ELSAs as an example, this tutorial offers methodological guidance on how to use the two‐stage approach to IPD meta‐analysis to account for the statistical challenges of complex survey designs (e.g., sampling weights, clustered and missing IPD), first, to conduct descriptive analyses (Stage 1), and second, to integrate results with three‐level meta‐analytic and meta‐regression models to take into account dependencies among effect sizes (Stage 2). The two‐stage approach is illustrated with IPD on reading achievement from the Programme for International Student Assessment (PISA). We demonstrate how to analyze and integrate standardized mean differences (e.g., gender differences), correlations (e.g., with students' socioeconomic status [SES]), and interactions between individual characteristics at the participant level (e.g., the interaction between gender and SES) across several PISA cycles. All the datafiles and R scripts we used are available online. Because complex social, health, or economic survey and panel studies share many methodological features with ELSAs, the guidance offered in this tutorial is also helpful for synthesizing research evidence from these studies. [ABSTRACT FROM AUTHOR]

Background: Stroke rehabilitation interventions are routinely personalized to address individuals' needs, goals, and challenges based on evidence from aggregated randomized controlled trials (RCT) data and meta-syntheses. Individual participant data (IPD) meta-analyses may better inform the development of precision rehabilitation approaches, quantifying treatment responses while adjusting for confounders and reducing ecological bias. Aim: We explored associations between speech and language therapy (SLT) interventions frequency (days/week), intensity (h/week), and dosage (total SLT-hours) and language outcomes for different age, sex, aphasia severity, and chronicity subgroups by undertaking prespecified subgroup network meta-analyses of the RELEASE database. Methods: MEDLINE, EMBASE, and trial registrations were systematically searched (inception-Sept2015) for RCTs, including ⩾ 10 IPD on stroke-related aphasia. We extracted demographic, stroke, aphasia, SLT, and risk of bias data. Overall-language ability, auditory comprehension, and functional communication outcomes were standardized. A one-stage, random effects, network meta-analysis approach filtered IPD into a single optimal model, examining SLT regimen and language recovery from baseline to first post-intervention follow-up, adjusting for covariates identified a-priori. Data were dichotomized by age (⩽/> 65 years), aphasia severity (mild–moderate/ moderate–severe based on language outcomes' median value), chronicity (⩽/> 3 months), and sex subgroups. We reported estimates of means and 95% confidence intervals. Where relative variance was high (> 50%), results were reported for completeness. Results: 959 IPD (25 RCTs) were analyzed. For working-age participants, greatest language gains from baseline occurred alongside moderate to high-intensity SLT (functional communication 3-to-4 h/week; overall-language and comprehension > 9 h/week); older participants' greatest gains occurred alongside low-intensity SLT (⩽ 2 h/week) except for auditory comprehension (> 9 h/week). For both age-groups, SLT-frequency and dosage associated with best language gains were similar. Participants ⩽ 3 months post-onset demonstrated greatest overall-language gains for SLT at low intensity/moderate dosage (⩽ 2 SLT-h/week; 20-to-50 h); for those > 3 months, post-stroke greatest gains were associated with moderate-intensity/high-dosage SLT (3–4 SLT-h/week; ⩾ 50 hours). For moderate–severe participants, 4 SLT-days/week conferred the greatest language gains across outcomes, with auditory comprehension gains only observed for ⩾ 4 SLT-days/week; mild–moderate participants' greatest functional communication gains were associated with similar frequency (⩾ 4 SLT-days/week) and greatest overall-language gains with higher frequency SLT (⩾ 6 days/weekly). Males' greatest gains were associated with SLT of moderate (functional communication; 3-to-4 h/weekly) or high intensity (overall-language and auditory comprehension; (> 9 h/weekly) compared to females for whom the greatest gains were associated with lower-intensity SLT (< 2 SLT-h/weekly). Consistencies across subgroups were also evident; greatest overall-language gains were associated with 20-to-50 SLT-h in total; auditory comprehension gains were generally observed when SLT > 9 h over ⩾ 4 days/week. Conclusions: We observed a treatment response in most subgroups' overall-language, auditory comprehension, and functional communication language gains. For some, the maximum treatment response varied in association with different SLT-frequency, intensity, and dosage. Where differences were observed, working-aged, chronic, mild–moderate, and male subgroups experienced their greatest language gains alongside high-frequency/intensity SLT. In contrast, older, moderate–severely impaired, and female subgroups within 3 months of aphasia onset made their greatest gains for lower-intensity SLT. The acceptability, clinical, and cost effectiveness of precision aphasia rehabilitation approaches based on age, sex, aphasia severity, and chronicity should be evaluated in future clinical RCTs. [ABSTRACT FROM AUTHOR]

Numerous arguments have been advanced for broadly sharing de-identified, participant-level clinical trials data, and trial sponsors and journals are increasingly requiring it. However, data sharing in pragmatic clinical trials presents ethical challenges related to the use of waivers or alterations of informed consent for some pragmatic clinical trials and corresponding limitations of informed consent to guide sharing decisions; the potential for data sharing in pragmatic clinical trials to present risks not only for individual patient-subjects, but also for health systems and the clinicians within them; sharing of data from electronic health records instead of data newly collected for research purposes; and researchers' limited capacity to control sensitive data within an electronic health record and potential implications of such limits for meeting obligations inherent to Certificates of Confidentiality. These challenges raise questions about the extent to which traditional research ethics governance structures are capable of guiding decisions about pragmatic clinical trial data sharing. This article identifies and examines these ethical challenges for pragmatic clinical trial data sharing. We suggest several areas for future empirical scholarship, including the need to identify patient and public attitudes regarding pragmatic clinical trial data sharing as well as to assess the demand for pragmatic clinical trial data and the correspondingly likely benefit of such sharing. Further conceptual work is also needed to explore how requirements to respect patient-subjects about whom data are shared in the context of pragmatic clinical trials should be understood, particularly in the absence of informed consent for initial research activities, and the appropriate balance between promoting the generation of socially valuable knowledge and respecting autonomy. [ABSTRACT FROM AUTHOR]

The coronavirus disease-2019 (COVID-19) pandemic has led to the irrational use of drugs in the absence of clinical management guidelines. Access to individual participant data (IPD) from clinical trials aids the evidence synthesis approaches. We undertook a rapid review to infer IPD sharing intentions based on data availability statements by the principal investigators (PIs) of drug and vaccine trials in the context of COVID-19.