External validation of the Dat'AIDS score: A risk score for predicting 5-year overall mortality in people living with HIV aged 60 years or older.

OBJECTIVE: To perform an external validation of the Dat'AIDS score for predicting 5-year overall mortality among people with HIV (PWH) aged 60 years or older. METHODS: This was a multi-centre prospective cohort study at all sites participating in the Swiss HIV Cohort Study (SHCS). We calculated the Dat'AIDS score in PWH aged 60 years or older at their first visit between 1 January 2015 and 1 January 2020. People living with HIV-2 and those whose Dat'AIDS score could not be calculated were excluded. Patients were followed until 1 January 2020. The primary endpoint was all-cause mortality. Vital status was collected throughout the study period. We obtained population and score descriptive statistics and assessed the score's discrimination and calibration. RESULTS: We included 2205 participants (82% male) of median [interquartile range (IQR)] age 62.0 (60.3-67.0) years, mostly with viraemia <50 copies/mL (92.7%). Median follow-up time was 15.9 years and median (IQR) CD4 cell count at enrolment was 586 (420-782) cells/µL. In all, 152 deaths were recorded during a total follow-up period of 7147 patient-years. The median (IQR) observed Dat'AIDS score was 3 (0-8). Discriminative capacities were good as the C-statistic was 0.73 (95% CI: 0.69-0.77) and consistent across all subgroups. Comparison of observed and expected survival probabilities showed good calibration. CONCLUSIONS: External validation of the Dat'AIDS score in patients aged 60 years or older showed that it could be a useful tool not only for research purposes, but also to identify older patients at a higher mortality risk and to tailor the most appropriate interventions.

Multiple imputation of incomplete multilevel data using Heckman selection models.

Missing data is a common problem in medical research, and is commonly addressed using multiple imputation. Although traditional imputation methods allow for valid statistical inference when data are missing at random (MAR), their implementation is problematic when the presence of missingness depends on unobserved variables, that is, the data are missing not at random (MNAR). Unfortunately, this MNAR situation is rather common, in observational studies, registries and other sources of real-world data. While several imputation methods have been proposed for addressing individual studies when data are MNAR, their application and validity in large datasets with multilevel structure remains unclear. We therefore explored the consequence of MNAR data in hierarchical data in-depth, and proposed a novel multilevel imputation method for common missing patterns in clustered datasets. This method is based on the principles of Heckman selection models and adopts a two-stage meta-analysis approach to impute binary and continuous variables that may be outcomes or predictors and that are systematically or sporadically missing. After evaluating the proposed imputation model in simulated scenarios, we illustrate it use in a cross-sectional community survey to estimate the prevalence of malaria parasitemia in children aged 2-10 years in five regions in Uganda.

Treatment methods for cervical intraepithelial neoplasia in England: A cost-effectiveness analysis.

OBJECTIVE: To compare the cost-effectiveness of different treatments for cervical intraepithelial neoplasia (CIN). DESIGN: A cost-effectiveness analysis based on data available in the literature and expert opinion. SETTING: England. POPULATION: Women treated for CIN. METHODS: We developed a decision-analytic model to simulate the clinical course of 1000 women who received local treatment for CIN and were followed up for 10 years after treatment. In the model we considered surgical complications as well as oncological and reproductive outcomes over the 10-year period. The costs calculated were those incurred by the National Health Service (NHS) of England. MAIN OUTCOME MEASURES: Cost per one CIN2+ recurrence averted (oncological outcome); cost per one preterm birth averted (reproductive outcome); overall cost per one adverse oncological or reproductive outcome averted. RESULTS: For young women of reproductive age, large loop excision of the transformation zone (LLETZ) was the most cost-effective treatment overall at all willingness-to-pay thresholds. For postmenopausal women, LLETZ remained the most cost-effective treatment up to a threshold of £31,500, but laser conisation became the most cost-effective treatment above that threshold. CONCLUSIONS: LLETZ is the most cost-effective treatment for both younger and older women. However, for older women, more radical excision with laser conisation could also be considered if the NHS is willing to spend more than £31,500 to avert one CIN2+ recurrence.

Risk factors for falls among hospitalized medical patients - A systematic review and meta-analysis.

OBJECTIVE: To identify and quantify risk factors for in-hospital falls in medical patients. DATA SOURCES: Six databases (MEDLINE, Embase, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, CINAHL, and Google Scholar) were systematically screened until April 11, 2023, to identify relevant articles. STUDY SELECTION: All titles and abstracts of the retrieved articles were independently screened by two researchers who also read the full texts of the remaining articles. Quantitative studies that assessed risk factors for falls among adult patients acutely hospitalized were included in the review. Publications that did not capture internal medicine patients or focused on other specific populations were excluded. DATA EXTRACTION: Information on study characteristics and potential risk factors were systematically extracted. Risk of bias was assessed using the Quality in Prognosis Studies (QUIPS) tool. PRISMA and MOOSE guidelines were followed for reporting. DATA SYNTHESIS: The main outcome was any in-hospital falls. Using a random-effects meta-analysis model, association measures for each risk factor reported in five or more studies were pooled. Separate analyses according to effect measure and studies adjusted for sex and age at least were performed. Of 5,067 records retrieved, 119 original publications from 25 countries were included. In conclusion, 23 potential risk factors were meta-analyzed. Strong evidence with large effect sizes was found for a history of falls (OR 2.54; 95% CI 1.63- 3.96; I2 91%), antidepressants (pooled OR 2.25; 95% confidence interval [95% CI] 1.92-2.65; I2 0%), benzodiazepines (OR 1.97; 95% CI 1.68-2.31; I2 0%), hypnotics-sedatives (OR 1.90; 95% CI 1.53-2.36; I2 46%), and antipsychotics (OR 1.61; 95% CI 1.33-1.95; I2 0%). Furthermore, evidence of associations with male sex (OR 1.22, 95% CI 0.99-1.50, I2 65%) and age (OR 1.17, 95% CI 1.02-1.35, I2 72%) were found, but effect sizes were small. CONCLUSIONS: The comprehensive list of risk factors, which specifies the strength of evidence and effect sizes, could assist in the prioritization of preventive measures and interventions.

Meta-analysis of the prevalence of attention-deficit hyperactivity disorder in prison: A comment on Fazel and Favril (2024) and reanalysis of the data.

BACKGROUND: Fazel and Favril presented a reanalysis of our previously published systematic review and meta-analysis on the prevalence of attention deficit hyperactivity disorder (ADHD) in prison. AIMS: The current paper addresses some of the criticisms of Fazel and Favril on our meta-analysis and presents a reanalysis of the data, focusing on adult detained persons. METHODS: We conducted a meta-regression on 28 studies (n = 7710) to estimae the pooled prevalence of ADHD. RESULTS: This reanalysis yielded a pooled estimate of 22.2% for the prevalence of ADHD (95% confidence interval [CI]: 15.7; 28.6), which disagrees with the estimate given by Fazel and Favril (8.3%, 95% CI: 3.8; 12.8). CONCLUSION: We argue that the ADHD prevalence provided by Fazel and Favril was an underestimate due to their use of too restrictive exclusion criteria and suboptimal analysis methods. Our reanalysis on detained adults suggests a higher ADHD prevalence, which highlights the need to diagnose and treat ADHD in prison.

Comparative effects of pharmacological interventions for the acute and long-term management of insomnia disorder in adults: a systematic review and network meta-analysis.

BACKGROUND: Behavioural, cognitive, and pharmacological interventions can all be effective for insomnia. However, because of inadequate resources, medications are more frequently used worldwide. We aimed to estimate the comparative effectiveness of pharmacological treatments for the acute and long-term treatment of adults with insomnia disorder. METHODS: In this systematic review and network meta-analysis, we searched the Cochrane Central Register of Controlled Trials, MEDLINE, PubMed, Embase, PsycINFO, WHO International Clinical Trials Registry Platform, ClinicalTrials.gov, and websites of regulatory agencies from database inception to Nov 25, 2021, to identify published and unpublished randomised controlled trials. We included studies comparing pharmacological treatments or placebo as monotherapy for the treatment of adults (≥18 year) with insomnia disorder. We assessed the certainty of evidence using the confidence in network meta-analysis (CINeMA) framework. Primary outcomes were efficacy (ie, quality of sleep measured by any self-rated scale), treatment discontinuation for any reason and due to side-effects specifically, and safety (ie, number of patients with at least one adverse event) both for acute and long-term treatment. We estimated summary standardised mean differences (SMDs) and odds ratios (ORs) using pairwise and network meta-analysis with random effects. This study is registered with Open Science Framework, https://doi.org/10.17605/OSF.IO/PU4QJ. FINDINGS: We included 170 trials (36 interventions and 47 950 participants) in the systematic review and 154 double-blind, randomised controlled trials (30 interventions and 44 089 participants) were eligible for the network meta-analysis. In terms of acute treatment, benzodiazepines, doxylamine, eszopiclone, lemborexant, seltorexant, zolpidem, and zopiclone were more efficacious than placebo (SMD range: 0·36-0·83 [CINeMA estimates of certainty: high to moderate]). Benzodiazepines, eszopiclone, zolpidem, and zopiclone were more efficacious than melatonin, ramelteon, and zaleplon (SMD 0·27-0·71 [moderate to very low]). Intermediate-acting benzodiazepines, long-acting benzodiazepines, and eszopiclone had fewer discontinuations due to any cause than ramelteon (OR 0·72 [95% CI 0·52-0·99; moderate], 0·70 [0·51-0·95; moderate] and 0·71 [0·52-0·98; moderate], respectively). Zopiclone and zolpidem caused more dropouts due to adverse events than did placebo (zopiclone: OR 2·00 [95% CI 1·28-3·13; very low]; zolpidem: 1·79 [1·25-2·50; moderate]); and zopiclone caused more dropouts than did eszopiclone (OR 1·82 [95% CI 1·01-3·33; low]), daridorexant (3·45 [1·41-8·33; low), and suvorexant (3·13 [1·47-6·67; low]). For the number of individuals with side-effects at study endpoint, benzodiazepines, eszopiclone, zolpidem, and zopiclone were worse than placebo, doxepin, seltorexant, and zaleplon (OR range 1·27-2·78 [high to very low]). For long-term treatment, eszopiclone and lemborexant were more effective than placebo (eszopiclone: SMD 0·63 [95% CI 0·36-0·90; very low]; lemborexant: 0·41 [0·04-0·78; very low]) and eszopiclone was more effective than ramelteon (0.63 [0·16-1·10; very low]) and zolpidem (0·60 [0·00-1·20; very low]). Compared with ramelteon, eszopiclone and zolpidem had a lower rate of all-cause discontinuations (eszopiclone: OR 0·43 [95% CI 0·20-0·93; very low]; zolpidem: 0·43 [0·19-0·95; very low]); however, zolpidem was associated with a higher number of dropouts due to side-effects than placebo (OR 2·00 [95% CI 1·11-3·70; very low]). INTERPRETATION: Overall, eszopiclone and lemborexant had a favorable profile, but eszopiclone might cause substantial adverse events and safety data on lemborexant were inconclusive. Doxepin, seltorexant, and zaleplon were well tolerated, but data on efficacy and other important outcomes were scarce and do not allow firm conclusions. Many licensed drugs (including benzodiazepines, daridorexant, suvorexant, and trazodone) can be effective in the acute treatment of insomnia but are associated with poor tolerability, or information about long-term effects is not available. Melatonin, ramelteon, and non-licensed drugs did not show overall material benefits. These results should serve evidence-based clinical practice. FUNDING: UK National Institute for Health Research Oxford Health Biomedical Research Centre.

Measuring the performance of prediction models to personalize treatment choice.

When data are available from individual patients receiving either a treatment or a control intervention in a randomized trial, various statistical and machine learning methods can be used to develop models for predicting future outcomes under the two conditions, and thus to predict treatment effect at the patient level. These predictions can subsequently guide personalized treatment choices. Although several methods for validating prediction models are available, little attention has been given to measuring the performance of predictions of personalized treatment effect. In this article, we propose a range of measures that can be used to this end. We start by defining two dimensions of model accuracy for treatment effects, for a single outcome: discrimination for benefit and calibration for benefit. We then amalgamate these two dimensions into an additional concept, decision accuracy, which quantifies the model's ability to identify patients for whom the benefit from treatment exceeds a given threshold. Subsequently, we propose a series of performance measures related to these dimensions and discuss estimating procedures, focusing on randomized data. Our methods are applicable for continuous or binary outcomes, for any type of prediction model, as long as it uses baseline covariates to predict outcomes under treatment and control. We illustrate all methods using two simulated datasets and a real dataset from a trial in depression. We implement all methods in the R package predieval. Results suggest that the proposed measures can be useful in evaluating and comparing the performance of competing models in predicting individualized treatment effect.

Comparative effectiveness and risk of preterm birth of local treatments for cervical intraepithelial neoplasia and stage IA1 cervical cancer: a systematic review and network meta-analysis.

BACKGROUND: The trade-off between comparative effectiveness and reproductive morbidity of different treatment methods for cervical intraepithelial neoplasia (CIN) remains unclear. We aimed to determine the risks of treatment failure and preterm birth associated with various treatment techniques. METHODS: In this systematic review and network meta-analysis, we searched MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials database for randomised and non-randomised studies reporting on oncological or reproductive outcomes after CIN treatments from database inception until March 9, 2022, without language restrictions. We included studies of women with CIN, glandular intraepithelial neoplasia, or stage IA1 cervical cancer treated with excision (cold knife conisation [CKC], laser conisation, and large loop excision of the transformation zone [LLETZ]) or ablation (radical diathermy, laser ablation, cold coagulation, and cryotherapy). We excluded women treated with hysterectomy. The primary outcomes were any treatment failure (defined as any abnormal histology or cytology) and preterm birth (<37 weeks of gestation). The network for preterm birth also included women with untreated CIN (untreated colposcopy group). The main reference group was LLETZ for treatment failure and the untreated colposcopy group for preterm birth. For randomised controlled trials, we extracted group-level summary data, and for observational studies, we extracted relative treatment effect estimates adjusted for potential confounders, when available, and we did random-effects network meta-analyses to obtain odds ratios (ORs) with 95% CIs. We assessed within-study and across-study risk of bias using Cochrane tools. This systematic review is registered with PROSPERO, CRD42018115495 and CRD42018115508. FINDINGS: 7880 potential citations were identified for the outcome of treatment failure and 4107 for the outcome of preterm birth. After screening and removal of duplicates, the network for treatment failure included 19 240 participants across 71 studies (25 randomised) and the network for preterm birth included 68 817 participants across 29 studies (two randomised). Compared with LLETZ, risk of treatment failure was reduced for other excisional methods (laser conisation: OR 0·59 [95% CI 0·44-0·79] and CKC: 0·63 [0·50-0·81]) and increased for laser ablation (1·69 [1·27-2·24]) and cryotherapy (1·84 [1·33-2·56]). No differences were found for the comparison of cold coagulation versus LLETZ (1·09 [0·68-1·74]) but direct data were based on two small studies only. Compared with the untreated colposcopy group, risk of preterm birth was increased for all excisional techniques (CKC: 2·27 [1·70-3·02]; laser conisation: 1·77 [1·29-2·43]; and LLETZ: 1·37 [1·16-1·62]), whereas no differences were found for ablative methods (laser ablation: 1·05 [0·78-1·41]; cryotherapy: 1·01 [0·35-2·92]; and cold coagulation: 0·67 [0·02-29·15]). The evidence was based mostly on observational studies with their inherent risks of bias, and the credibility of many comparisons was low. INTERPRETATION: More radical excisional techniques reduce the risk of treatment failure but increase the risk of subsequent preterm birth. Although there is uncertainty, ablative treatments probably do not increase risk of preterm birth, but are associated with higher failure rates than excisional techniques. Although we found LLETZ to have balanced effectiveness and reproductive morbidity, treatment choice should rely on a woman's age, size and location of lesion, and future family planning. FUNDING: National Institute for Health and Care Research: Research for Patient Benefit.

Introducing the Treatment Hierarchy Question in Network Meta-Analysis.

Comparative effectiveness research using network meta-analysis can present a hierarchy of competing treatments, from the most to the least preferable option. However, in published reviews, the research question associated with the hierarchy of multiple interventions is typically not clearly defined. Here we introduce the novel notion of a treatment hierarchy question that describes the criterion for choosing a specific treatment over one or more competing alternatives. For example, stakeholders might ask which treatment is most likely to improve mean survival by at least 2 years, or which treatment is associated with the longest mean survival. We discuss the most commonly used ranking metrics (quantities that compare the estimated treatment-specific effects), how the ranking metrics produce a treatment hierarchy, and the type of treatment hierarchy question that each ranking metric can answer. We show that the ranking metrics encompass the uncertainty in the estimation of the treatment effects in different ways, which results in different treatment hierarchies. When using network meta-analyses that aim to rank treatments, investigators should state the treatment hierarchy question they aim to address and employ the appropriate ranking metric to answer it. Following this new proposal will avoid some controversies that have arisen in comparative effectiveness research.

Bayesian models for aggregate and individual patient data component network meta-analysis.

Network meta-analysis can synthesize evidence from studies comparing multiple treatments for the same disease. Sometimes the treatments of a network are complex interventions, comprising several independent components in different combinations. A component network meta-analysis (CNMA) can be used to analyze such data and can in principle disentangle the individual effect of each component. However, components may interact with each other, either synergistically or antagonistically. Deciding which interactions, if any, to include in a CNMA model may be difficult, especially for large networks with many components. In this article, we present two Bayesian CNMA models that can be used to identify prominent interactions between components. Our models utilize Bayesian variable selection methods, namely the stochastic search variable selection and the Bayesian LASSO, and can benefit from the inclusion of prior information about important interactions. Moreover, we extend these models to combine data from studies providing aggregate information and studies providing individual patient data (IPD). We illustrate our models in practice using three real datasets, from studies in panic disorder, depression, and multiple myeloma. Finally, we describe methods for developing web-applications that can utilize results from an IPD-CNMA, to allow for personalized estimates of relative treatment effects given a patient's characteristics.

Development and validation of a meta-learner for combining statistical and machine learning prediction models in individuals with depression.

BACKGROUND: The debate of whether machine learning models offer advantages over standard statistical methods when making predictions is ongoing. We discuss the use of a meta-learner model combining both approaches as an alternative. METHODS: To illustrate the development of a meta-learner, we used a dataset of 187,757 people with depression. Using 31 variables, we aimed to predict two outcomes measured 60 days after initiation of antidepressant treatment: severity of depressive symptoms (continuous) and all-cause dropouts (binary). We fitted a ridge regression and a multi-layer perceptron (MLP) deep neural network as two separate prediction models ("base-learners"). We then developed two "meta-learners", combining predictions from the two base-learners. To compare the performance across the different methods, we calculated mean absolute error (MAE, for continuous outcome) and the area under the receiver operating characteristic curve (AUC, for binary outcome) using bootstrapping. RESULTS: Compared to the best performing base-learner (MLP base-learner, MAE at 4.63, AUC at 0.59), the best performing meta-learner showed a 2.49% decrease in MAE at 4.52 for the continuous outcome and a 6.47% increase in AUC at 0.60 for the binary outcome. CONCLUSIONS: A meta-learner approach may effectively combine multiple prediction models. Choosing between statistical and machine learning models may not be necessary in practice.

Development and validation of prediction model for incident overactive bladder: The Nagahama study.

OBJECTIVES: We aimed to develop models to predict new-onset overactive bladder in 5 years using a large prospective cohort of the general population. METHODS: This is a secondary analysis of a longitudinal cohort study in Japan. The baseline characteristics were measured between 2008 and 2010, with follow-ups every 5 years. We included subjects without overactive bladder at baseline and with follow-up data 5 years later. Overactive bladder was assessed using the overactive bladder symptom score. Baseline characteristics (demographics, health behaviors, comorbidities, and overactive bladder symptom scores) and blood test data were included as predictors. We developed two competing prediction models for each sex based on logistic regression with penalized likelihood (LASSO). We chose the best model separately for men and women after evaluating models' performance in terms of discrimination and calibration using an internal validation via 200 bootstrap resamples and a temporal validation. RESULTS: We analyzed 7218 participants (male: 2238, female: 4980). The median age was 60 and 55 years, and the number of new-onset overactive bladder was 223 (10.0%) and 288 (5.8%) per 5 years in males and females, respectively. The in-sample estimates for C-statistic, calibration intercept, and slope for the best performing models were 0.77 (95% confidence interval 0.74-0.80), 0.28 and 1.15 for males, and 0.77 (95% confidence interval 0.74-0.80), 0.20 and 1.08 for females. Internal and temporal validation gave broadly similar estimates of performance, indicating low optimism. CONCLUSION: We developed risk prediction models for new-onset overactive bladder among men and women with good predictive ability.

Immunodeficiency and Cancer in 3.5 Million People Living With Human Immunodeficiency Virus (HIV): The South African HIV Cancer Match Study.

BACKGROUND: We analyzed associations between immunodeficiency and cancer incidence in a nationwide cohort of people living with human immunodeficiency virus (HIV; PLWH) in South Africa. METHODS: We used data from the South African HIV Cancer Match Study built on HIV-related laboratory measurements from the National Health Laboratory Services and cancer records from the National Cancer Registry. We evaluated associations between time-updated CD4 cell count and cancer incidence rates using Cox proportional hazards models. We reported adjusted hazard ratios (aHRs) over a grid of CD4 values and estimated the aHR per 100 CD4 cells/µL decrease. RESULTS: Of 3 532 266 PLWH, 15 078 developed cancer. The most common cancers were cervical cancer (4150 cases), Kaposi sarcoma (2262 cases), and non-Hodgkin lymphoma (1060 cases). The association between lower CD4 cell count and higher cancer incidence rates was strongest for conjunctival cancer (aHR per 100 CD4 cells/µL decrease: 1.46; 95% confidence interval [CI], 1.38-1.54), Kaposi sarcoma (aHR, 1.23; 95% CI, 1.20-1.26), and non-Hodgkin lymphoma (aHR, 1.18; 95% CI, 1.14-1.22). Among infection-unrelated cancers, lower CD4 cell counts were associated with higher incidence rates of esophageal cancer (aHR, 1.06; 95% CI, 1.00-1.11) but not breast, lung, or prostate cancer. CONCLUSIONS: Lower CD4 cell counts were associated with an increased risk of developing various infection-related cancers among PLWH. Reducing HIV-induced immunodeficiency may be a potent cancer-prevention strategy among PLWH in sub-Saharan Africa, a region heavily burdened by cancers attributable to infections.

The Comparative Efficacy of Chlorhexidine Gluconate and Povidone-iodine Antiseptics for the Prevention of Infection in Clean Surgery: A Systematic Review and Network Meta-analysis.

OBJECTIVE: There is uncertainty around preoperative skin antisepsis in clean surgery. Network meta-analysis provides more precise estimates than standard pairwise meta-analysis and can rank interventions by efficacy, to better inform clinical decisions. BACKGROUND: Infection is the most common and costly complication of surgery. The relative efficacy of CHG and PVI based skin antiseptics in clean surgery remains unclear. METHODS: We searched for randomized or nonrandomized studies comparing the effect of different preparations of CHG and PVI on the dichotomous outcome of surgical site infection. We included studies of adults undergoing clean surgery. We excluded studies concerning indwelling vascular catheters, blood sampling, combination antiseptics or sequential applications of different antiseptics. We performed a network meta-analysis to estimate the relative efficacy of interventions using relative risks (RR). RESULTS: We included 17 studies comparing 5 antiseptics in 14,593 individuals. The overall rate of surgical site infection was 3%. Alcoholic CHG 4%-5% was ranked as the most effective antiseptic as it halved the risk of surgical site infection when compared to aqueous PVI [RR 0.49 (95% confidence interval 0.24, 1.02)] and also to alcoholic PVI, although uncertainty was larger [RR 0.51 (95% confidence interval 0.21, 1.27)]. Adverse events related to antiseptic application were only observed with patients exposed to PVI. CONCLUSIONS: Alcoholic formulations of 4%-5% CHG seem to be safe and twice as effective as PVI (alcoholic or aqueous solutions) in preventing infection after clean surgery in adults. Our findings concur with the literature on contaminated and clean-contaminated surgery, and endorse guidelines worldwide which advocate the use of alcoholic CHG for preoperative skin antisepsis. REGISTRATION: PROSPERO ID CRD42018113001.

Comparing methods for estimating patient-specific treatment effects in individual patient data meta-analysis.

Meta-analysis of individual patient data (IPD) is increasingly used to synthesize data from multiple trials. IPD meta-analysis offers several advantages over meta-analyzing aggregate data, including the capacity to individualize treatment recommendations. Trials usually collect information on many patient characteristics. Some of these covariates may strongly interact with treatment (and thus be associated with treatment effect modification) while others may have little effect. It is currently unclear whether a systematic approach to the selection of treatment-covariate interactions in an IPD meta-analysis can lead to better estimates of patient-specific treatment effects. We aimed to answer this question by comparing in simulations the standard approach to IPD meta-analysis (no variable selection, all treatment-covariate interactions included in the model) with six alternative methods: stepwise regression, and five regression methods that perform shrinkage on treatment-covariate interactions, that is, least absolute shrinkage and selection operator (LASSO), ridge, adaptive LASSO, Bayesian LASSO, and stochastic search variable selection. Exploring a range of scenarios, we found that shrinkage methods performed well for both continuous and dichotomous outcomes, for a variety of settings. In most scenarios, these methods gave lower mean squared error of the patient-specific treatment effect as compared with the standard approach and stepwise regression. We illustrate the application of these methods in two datasets from cardiology and psychiatry. We recommend that future IPD meta-analysis that aim to estimate patient-specific treatment effects using multiple effect modifiers should use shrinkage methods, whereas stepwise regression should be avoided.

Protocol for development and validation of a prediction model for 5-year risk of incident overactive bladder in the general population: the Nagahama study.

BACKGROUND: An accurate prediction model could identify high-risk subjects of incident Overactive bladder (OAB) among the general population and enable early prevention which may save on the related medical costs. However, no efficient model has been developed for predicting incident OAB. In this study, we will develop a model for predicting the onset of OAB at 5-year in the general population setting. METHODS: Data will be obtained from the Nagahama Cohort Project, a longitudinal, general population cohort study. The baseline characteristics were measured between Nov 28, 2008 and Nov 28, 2010, and follow-up was performed every 5 years. From the total of 9,764 participants (male: 3,208, female: 6,556) at baseline, we will exclude participants who could not attend the follow-up assessment and those who were defined as having OAB at baseline. The outcome will be incident OAB defined using the Overactive Bladder Symptom Score (OABSS) at follow-up assessment. Baseline questionnaires (demographic, health behavior, comorbidities and OABSS) and blood test data will be included as predictors. We will develop a logistic regression model utilizing shrinkage methods (LASSO penalization method). Model performance will be evaluated by discrimination and calibration. Net benefit will be evaluated by decision curve analysis. We will perform an internal validation and a temporal validation of the model. We will develop a web-based application to visualize the prediction model and facilitate its use in clinical practice. DISCUSSION: This will be the first study to develop a model to predict the incidence of OAB.

Drug-eluting or bare-metal stents for percutaneous coronary intervention: a systematic review and individual patient data meta-analysis of randomised clinical trials.

BACKGROUND: New-generation drug-eluting stents (DES) have mostly been investigated in head-to-head non-inferiority trials against early-generation DES and have typically shown similar efficacy and superior safety. How the safety profile of new-generation DES compares with that of bare-metal stents (BMS) is less clear. METHODS: We did an individual patient data meta-analysis of randomised clinical trials to compare outcomes after implantation of new-generation DES or BMS among patients undergoing percutaneous coronary intervention. The primary outcome was the composite of cardiac death or myocardial infarction. Data were pooled in a one-stage random-effects meta-analysis and examined at maximum follow-up and a 1-year landmark. Risk estimates are reported as hazard ratios (HRs) with 95% CIs. This study is registered in PROSPERO, number CRD42017060520. FINDINGS: We obtained individual data for 26 616 patients in 20 randomised trials. Mean follow-up was 3·2 (SD 1·8) years. The risk of the primary outcome was reduced in DES recipients compared with BMS recipients (HR 0·84, 95% CI 0·78-0·90, p<0·001) owing to a reduced risk of myocardial infarction (0·79, 0·71-0·88, p<0·001) and a possible slight but non-significant cardiac mortality benefit (0·89, 0·78-1·01, p=0·075). All-cause death was unaffected (HR with DES 0·96, 95% CI 0·88-1·05, p=0·358), but risk was lowered for definite stent thrombosis (0·63, 0·50-0·80, p<0·001) and target-vessel revascularisation (0·55, 0·50-0·60, p<0·001). We saw a time-dependent treatment effect, with DES being associated with lower risk of the primary outcome than BMS up to 1 year after placement. While the effect was maintained in the longer term, there was no further divergence from BMS after 1 year. INTERPRETATION: The performance of new-generation DES in the first year after implantation means that BMS should no longer be considered the gold standard for safety. Further development of DES technology should target improvements in clinical outcomes beyond 1 year. FUNDING: Bern University Hospital.

Network meta-analysis of rare events using the Mantel-Haenszel method.

The Mantel-Haenszel (MH) method has been used for decades to synthesize data obtained from studies that compare two interventions with respect to a binary outcome. It has been shown to perform better than the inverse-variance method or Peto's odds ratio when data is sparse. Network meta-analysis (NMA) is increasingly used to compare the safety of medical interventions, synthesizing, eg, data on mortality or serious adverse events. In this setting, sparse data occur often and yet there is to-date, no extension of the MH method for the case of NMA. In this paper, we fill this gap by presenting a MH-NMA method for odds ratios. Similarly to the pairwise MH method, we assume common treatment effects. We implement our approach in R, and we provide freely available easy-to-use routines. We illustrate our approach using data from two previously published networks. We compare our results to those obtained from three other approaches to NMA, namely, NMA with noncentral hypergeometric likelihood, an inverse-variance NMA, and a Bayesian NMA with a binomial likelihood. We also perform simulations to assess the performance of our method and compare it with alternative methods. We conclude that our MH-NMA method offers a reliable approach to the NMA of binary outcomes, especially in the case or sparse data, and when the assumption of methodological and clinical homogeneity is justifiable.

Dismantling cognitive-behaviour therapy for panic disorder: a systematic review and component network meta-analysis.

Cognitive-behaviour therapy (CBT) for panic disorder may consist of different combinations of several therapeutic components such as relaxation, breathing retraining, cognitive restructuring, interoceptive exposure and/or in vivo exposure. It is therefore important both theoretically and clinically to examine whether specific components of CBT or their combinations are superior to others in the treatment of panic disorder. Component network meta-analysis (NMA) is an extension of standard NMA that can be used to disentangle the treatment effects of different components included in composite interventions. We searched MEDLINE, EMBASE, PsycINFO and Cochrane Central, with supplementary searches of reference lists and clinical trial registries, for all randomized controlled trials comparing different CBT-based psychological therapies for panic disorder with each other or with control interventions. We applied component NMA to disentangle the treatment effects of different components included in these interventions. After reviewing 2526 references, we included 72 studies with 4064 participants. Interoceptive exposure and face-to-face setting were associated with better treatment efficacy and acceptability. Muscle relaxation and virtual-reality exposure were associated with significantly lower efficacy. Components such as breathing retraining and in vivo exposure appeared to improve treatment acceptability while having small effects on efficacy. The comparison of the most v. the least efficacious combination, both of which may be provided as 'evidence-based CBT,' yielded an odds ratio for the remission of 7.69 (95% credible interval: 1.75 to 33.33). Effective CBT packages for panic disorder would include face-to-face and interoceptive exposure components, while excluding muscle relaxation and virtual-reality exposure.

Cognitive-Behavioral Analysis System of Psychotherapy, Drug, or Their Combination for Persistent Depressive Disorder: Personalizing the Treatment Choice Using Individual Participant Data Network Metaregression.

BACKGROUND: Persistent depressive disorder is prevalent, disabling, and often difficult to treat. The cognitive-behavioral analysis system of psychotherapy (CBASP) is the only psychotherapy specifically developed for its treatment. However, we do not know which of CBASP, antidepressant pharmacotherapy, or their combination is the most efficacious and for which types of patients. This study aims to present personalized prediction models to facilitate shared decision-making in treatment choices to match patients' characteristics and preferences based on individual participant data network metaregression. METHODS: We conducted a comprehensive search for randomized controlled trials comparing any two of CBASP, pharmacotherapy, or their combination and sought individual participant data from identified trials. The primary outcomes were reduction in depressive symptom severity for efficacy and dropouts due to any reason for treatment acceptability. RESULTS: All 3 identified studies (1,036 participants) were included in the present analyses. On average, the combination therapy showed significant superiority over both monotherapies in terms of efficacy and acceptability, while the latter 2 treatments showed essentially similar results. Baseline depression, anxiety, prior pharmacotherapy, age, and depression subtypes moderated their relative efficacy, which indicated that for certain subgroups of patients either drug therapy or CBASP alone was a recommendable treatment option that is less costly, may have fewer adverse effects and match an individual patient's preferences. An interactive web app (https://kokoro.med.kyoto-u.ac.jp/CBASP/prediction/) shows the predicted disease course for all possible combinations of patient characteristics. CONCLUSIONS: Individual participant data network metaregression enables treatment recommendations based on individual patient characteristics.