Prior research on the mental well-being of adolescents has shown conflicting findings regarding the impact of only-child status. This study uses Propensity Score Matching (PSM) to control for confounding variables and investigate the impact of only-child status on the psychological health of adolescents.

A multi-center cluster sampling approach included 7,359 students from 33 middle schools in Western China. The Depression, Anxiety, and Stress Scale-21 (DASS-21), the Childhood Psychological Abuse and Neglect Scale (CPANS), the Pittsburgh Sleep Quality Index (PSQI), the Chinese iteration of the Barratt Impulsiveness Scale (BIS-11), and the Chinese version of the Positive and Negative Affect Scale for Children (PANAS-C) were utilized to assess the adolescents' emotions, sleep, psychological abuse, and neglect. PSM was employed to address confounding variables. Univariate analysis used t-tests, chi-square tests, and Wilcoxon rank-sum tests, while multivariate analysis used GLM to compare DASS-21, PSQI, and PANAS-C scores.

After PSM, 980 only-child and 980 non-only-child adolescents were included. Only-child adolescents showed significantly lower levels of depression, stress, emotional neglect, and negative affect. Further scrutiny of the mean ranks of these dimensions indicated that the only-child group yielded lower scores. In the GLM analysis, after adjusting for neglect, no significant associations were observed (all p > 0.05). However, post adjustments for psychological abuse, the only-child group also recorded lower scores in depression, stress, and negative affect.

This study reveals that the only-child adolescents possibly showing better psychological well-being overall in western China. This study's findings suggest that, during adolescent development, families and society should pay greater attention to the mental health of non-only children.

Poor oral health is a public health issue in the United States, disproportionately affecting people in poverty. This cross-sectional study investigates the impact of reducing absolute and relative poverty on the prevalence of periodontitis, caries, and dental pain among US adults. Data from 13,139 adults aged 30 to 70 y who completed dental examinations in the 2011-2018 National Health and Nutrition Examination Surveys were used. Periodontitis and dental caries outcomes were assessed with the 2011-2014 surveys (n = 6,563). Assessment of absolute and relative poverty was based on the poverty income ratio established by the US Census Bureau. Hypothetical counterfactual scenarios were emulated to assess the impact of poverty reductions (10%, 25%, and 50%) on periodontitis, dental caries, and dental pain. A targeted minimum loss-based estimator was used to estimate the outcomes under each scenario adjusted for age, sex, race, comorbidity, and marital status. Reductions in absolute and relative poverty were associated with a lower prevalence of oral disease. A 50% reduction in absolute poverty would avert 1.1 million cases of periodontitis, 0.4 million individuals with dental caries, and 0.6 million dental pain cases. A similar reduction in relative poverty would avert 5.4 million cases of periodontitis, 3.8 million individuals with caries, and 2 million cases of dental pain. The greatest impact was seen with a 50% relative poverty reduction: 12% reduction in periodontitis (prevalence ratio [PR], 0.88; 95% CI, 0.85 to 0.92), 13% reduction in caries (PR, 0.87; 95% CI, 0.81 to 0.92), and 18% reduction in frequent dental pain (PR, 0.82; 95% CI, 0.73 to 0.91). These findings highlight the potential of poverty reduction, especially relative poverty, to significantly lower the US oral disease burden and emphasize policy importance for reducing income inequality to achieve equitable oral health.

Hashimoto's thyroiditis (HT) is a prevalent autoimmune disorder and thyroid cancer (TC) is the most prevalent endocrine malignancy. Recent debates have focused on whether HT increases the risk of developing TC. This study combined Mendelian randomization (MR) and observational methods to investigate the potential causal relationship between HT and TC risk.

First, we performed two-sample MR and multivariable MR (MVMR) analysis using the genome-wide association studies (GWAS) data from multiple databases, including European and East Asian populations, to estimate the effect of HT and thyroid-stimulating hormone (TSH) levels on TC risk. Second, we conducted an observational study using data from the National Health and Nutrition Examination Survey (NHANES) database and evaluated the association between HT, TSH, and TC prevalence through logistic regression model and restricted cubic spline model.

Our MR findings revealed no significant association between HT and TC risk in both populations. However, elevated TSH levels significantly increased TC and papillary thyroid carcinoma (PTC) risk, while lower TSH levels were associated with reduced TC risk. Further MVMR analysis and an observational study confirmed this. Additionally, our observational study also indicated no significant relationship between HT and TC prevalence and abnormal TSH levels correlated with higher TC risk.

HT was not a TC risk factor, but high TSH levels increased TC risk. Controlling TSH within normal ranges through thyroid hormone replacement was recommended to reduce TC risk in HT patients with elevated TSH levels, even those without symptoms.

Confounder adjustment is critical for accurate causal inference in observational studies. However, the appropriateness of methods for confounder adjustment in studies investigating multiple risk factors, where the factors are not simply mutually confounded, is often overlooked. This study aims to summarise the methods for confounder adjustment and the related issues in studies investigating multiple risk factors.

A methodological study was performed. We searched PubMed from January 2018 to March 2023 to identify cohort and case-control studies investigating multiple risk factors for three chronic diseases (cardiovascular disease, diabetes and dementia). Study selection and data extraction were conducted independently by two reviewers. The study objectives were grouped into two categories: widely exploring potential risk factors and examining specific risk factors. The methods for confounder adjustment were classified based on a summarisation of the included studies, identifying six categories: (1) each risk factor was adjusted for potential confounders separately (the recommended method); (2) all risk factors were mutually adjusted (i.e. including all factors in a multivariable model); (3) all risk factors were adjusted for the same confounders separately; (4) all risk factors were adjusted for the same confounders with some factors being mutually adjusted; (5) all risk factors were adjusted for the same confounders with mutual adjustment among them being unclear; and (6) unable to judge. All data were descriptively analysed.

A total of 162 studies were included, with 88 (54.3%) exploring potential risk factors and 74 (45.7%) examining specific risk factors. The current status of confounder adjustment was unsatisfactory: only ten studies (6.2%) used the recommended method, all of which aimed at examining several specific risk factors; in contrast, mutual adjustment was adopted in over 70% of the studies. The remaining studies either adjusted for the same confounders across all risk factors, or unable to judge.

There is substantial variation in the methods for confounder adjustment among studies investigating multiple risk factors. Mutual adjustment was the most commonly adopted method, which might lead to overadjustment bias and misleading effect estimates. Future research should avoid indiscriminately including all risk factors in a multivariable model to prevent inappropriate adjustment.

Methods for matching in longitudinal cohort studies, such as sequential stratification and time-varying propensity scores, facilitate causal inferences in the context of time-dependent treatments that are not randomized where patient eligibility or treatment status changes over time. The tradeoffs in available approaches have not been compared previously, so we compare two methods using simulations based on a retrospective cohort of patients eligible for weight loss surgery, some of whom received it.

This study compares matching completeness, bias, coverage, and precision among three approaches to longitudinal matching: (1) time-varying propensity scores (tvPS), (2) sequential stratification that matches exactly on all covariates used in tvPS (SS-Full) and (3) sequential stratification that exact matches on a subset of covariates (SS-Selected). These comparisons are made in the context of a deep sampling frame (50:1) and a shallow sampling frame (5:1) of eligible comparators. A simulation study was employed to estimate the relative performance of these approaches.

In 1,000 simulations each, tvPS retained more than 99.9% of treated patients in both the deep and shallow sampling frames, while a smaller proportion of treated patients were retained for SS-Full (91.6%) and SS-Selected (98.2%) in the deep sampling frame. In the shallow sampling frame, sequential stratification retained many fewer treated patients (73.9% SS-Full, 92.0% SS-Selected) than tvPS yet coverage, precision and bias were comparable for tvPS, SS-Full and SS-Selected in the deep and shallow sampling frames.

Time-varying propensity scores have comparable performance to sequential stratification in terms of coverage, bias, and precision, with superior match completeness. While performance was generally comparable across methods, greater match completeness makes tvPS an attractive option for longitudinal matching studies where external validity is highly valued.

Segmenting peripancreatic vessels in CT, including the superior mesenteric artery (SMA), the coeliac artery (CA), and the partial portal venous system (PPVS), is crucial for preoperative resectability analysis in pancreatic cancer. However, the clinical applicability of vessel segmentation methods is impeded by the low generalizability on multi-center data, mainly attributed to the wide variations in image appearance, namely the spurious correlation factor. Therefore, we propose a causal-invariance-driven generalizable segmentation model for peripancreatic vessels. It incorporates interventions at both image and feature levels to guide the model to capture causal information by enforcing consistency across datasets, thus enhancing the generalization performance. Specifically, firstly, a contrast-driven image intervention strategy is proposed to construct image-level interventions by generating images with various contrast-related appearances and seeking invariant causal features. Secondly, the feature intervention strategy is designed, where various patterns of feature bias across different centers are simulated to pursue invariant prediction. The proposed model achieved high DSC scores (79.69%, 82.62%, and 83.10%) for the three vessels on a cross-validation set containing 134 cases. Its generalizability was further confirmed on three independent test sets of 233 cases. Overall, the proposed method provides an accurate and generalizable segmentation model for peripancreatic vessels and offers a promising paradigm for increasing the generalizability of segmentation models from a causality perspective. Our source codes will be released at https://github.com/ SJTUBME-QianLab/PC_VesselSeg.

Resilience, the capacity to maintain or regain functionality in the face of adversity, is a dynamic process influenced by individual, familial, and community factors. Despite its variability, distinct resilience trajectories can be identified within populations, yet the predictors defining these distinct groups remains largely unclear. Here, using data from the Avon Longitudinal Study of Parents and Children (ages 0-18), we quantify resilience as the remaining variance in psychosocial functioning after taking into account the exposure to adversity. Growth mixture modeling identified seven distinct resilience trajectories, with over half of the study population maintaining resilience throughout early life. Factors increasing the likelihood of resilient trajectory membership included a less emotional temperament, high cognitive abilities, high self-esteem, low levels of autistic social traits, strong sibling relationships, high maternal care, and positive school experiences. Among the socioeconomic factors considered, maternal education - a significant indicator of socioeconomic status - and birth-order were associated with resilient trajectories. Our findings underscore the importance of fostering cognitive abilities, self-esteem, social relationships, positive school experiences, and extracurricular engagement to bolster resilience in adversity-exposed individuals and communities. This research informs resilience-focused interventions in mental health, education, and social policy sectors, and prompts further exploration of socioeconomic influences on resilience trajectories.

Causal graphs are an important tool for covariate selection but there is limited applied research on how best to create them. Here, we used data from the Coronary Drug Project trial to assess a range of approaches to directed acyclic graph (DAG) creation. We focused on the effect of adherence on mortality in the placebo arm, since the true causal effect is believed with a high degree of certainty.

We created DAGs for the effect of placebo adherence on mortality using different approaches for identifying variables and links to include or exclude. For each DAG, we identified minimal adjustment sets of covariates for estimating our causal effect of interest and applied these to analyses of the Coronary Drug Project data.

When we used only baseline covariate values to estimate the cumulative effect of placebo adherence on mortality, all adjustment sets performed similarly. The specific choice of covariates had minimal effect on these (biased) point estimates, but including nonconfounding prognostic factors resulted in smaller variance estimates. When we additionally adjusted for time-varying covariates of adherence using inverse probability weighting, covariates identified from the DAG created by focusing on prognostic factors performed best.

Theoretical advice on covariate selection suggests that including prognostic factors that are not exposure predictors can reduce variance without increasing bias. In contrast, for exposure predictors that are not prognostic factors, inclusion may result in less bias control. Our results empirically confirm this advice. We recommend that hand-creating DAGs begin with the identification of all potential outcome prognostic factors.

Randomized controlled trials (RCTs) are considered the gold standard for causal inference. With a sufficient sample size, randomization removes confounding up to the time of randomization and allows the treatment effect to be isolated. However, RCTs may have limited generalizability and transportability and are often not feasible in addiction research due to ethical or logistical constraints. The importance of observational studies from real-world settings has been increasingly recognized in research on health. This paper provides an overview of modern approaches to designing observational studies that enable causal inference. It illustrates three key techniques, Directed Acyclic Graphs (DAGs), modified Disjunctive Cause Criterion and Target Trial Emulation, and discusses the strengths and limitations of their applications.

The Toronto Transit Commission (TTC) operates the public transit system in Toronto, Canada. From 1954 to 1980, there were 430 suicide deaths/attempts on the TTC subway system. In 2011, TTC implemented Crisis Link, a suicide helpline to connect subway passengers with counsellors. Upstream factors such as media reporting about suicide incidents may also influence suicidal behaviour. Our objectives were to investigate how Crisis Link and media reports about TTC suicide incidents influenced suicide rates.

Suicide data were obtained from the TTC and Coroner, with Crisis Link data provided by Distress Centres of Greater Toronto (1998-2021). Media articles were identified through a database search of Toronto media publications. Interrupted time-series analysis investigated the association between Crisis Link calls, media articles, and quarterly suicide rates on the subway system.

There were 302 suicides on TTC's subway system from 1998 to 2021. The introduction of Crisis Link was associated with a large but non-significant decrease in TTC-related suicide rate in the same quarter (IRR = 0.64, 95% CI = 0.36-1.12). Each subsequent post-Crisis-Link quarter experienced an average 2% increase in suicide rate (IRR = 1.02, 95% CI = 1.004-1.04). Furthermore, for each TTC-related media article in the previous quarter, the suicide rate on the TTC increased by 2% (IRR = 1.02, 95% CI = 1.004-1.04).

The Crisis Link helpline was associated with a large but non-significant short-term decrease in suicide rates. However, this outcome was not sustained; this may, in part, be attributable to media reporting which was associated with increased suicides. This should inform suicide prevention policies in Canada and worldwide.

No funding.

Per- and polyfluoroalkyl substances (PFAS) are a broad class of synthetic chemicals; some are present in most humans in developed countries. Some studies suggest that certain PFAS may have immunotoxic effects in humans, which could put individuals with high levels of exposure at increased risk for infectious diseases such as COVID-19. We conducted a case-control study to examine the association between COVID-19 diagnosis and PFAS serum concentrations among employees and retirees from two 3 M facilities, one of which historically generated perfluorooctanoic acid (PFOA), perfluorooctane sulfonic acid (PFOS), and perfluorohexane sulfonic acid (PFHxS). Participants completed enrollment and follow-up study visits in the Spring of 2021. Participants were categorized as cases if they reported a COVID-19 diagnosis or became sick with at least one symptom of COVID-19 when someone else in their household was diagnosed, otherwise they were categorized as a control. COVID-19 diagnosis was modeled in relation to concentration of serum PFAS measured at enrollment after adjusting for covariates. The analytic sample comprised 573 individuals, 111 cases (19.4%) and 462 controls (80.6%). In adjusted models, the odds ratio of COVID-19 was 0.94 per interquartile range (14.3 ng/mL) increase in PFOS (95% confidence interval 0.85, 1.04). Results for PFOA, PFHxS, and perfluorononanoic acid (PFNA) were similar. Other PFAS present at lower concentrations were examined as categorical variables (above the limit of quantification [LOQ], yes vs. no [referent category]), and also showed no positive associations. In our study, which used individual-level data and included people with high occupational exposure, the serum concentrations of all PFAS examined were not associated with an increased odds ratio for COVID-19. At this point, the epidemiologic data supporting no association of COVID-19 occurrence with PFAS exposure are stronger than those suggesting a positive association.

We discuss 12 misperceptions, misstatements, or mistakes concerning the use of covariates in observational or nonrandomized research. Additionally, we offer advice to help investigators, editors, reviewers, and readers make more informed decisions about conducting and interpreting research where the influence of covariates may be at issue. We primarily address misperceptions in the context of statistical management of the covariates through various forms of modeling, although we also emphasize design and model or variable selection. Other approaches to addressing the effects of covariates, including matching, have logical extensions from what we discuss here but are not dwelled upon heavily. The misperceptions, misstatements, or mistakes we discuss include accurate representation of covariates, effects of measurement error, overreliance on covariate categorization, underestimation of power loss when controlling for covariates, misinterpretation of significance in statistical models, and misconceptions about confounding variables, selecting on a collider, and p value interpretations in covariate-inclusive analyses. This condensed overview serves to correct common errors and improve research quality in general and in nutrition research specifically.

Several epidemiological studies have investigated the possible role that living in areas with greater amounts of greenspace has on the incidence of childhood asthma. These findings have been inconsistent, and few studies explored the relevance of timing of exposure. We investigated the role of residential surrounding greenness on the risk of incident asthma using a population-based retrospective cohort study. We included 982,131 singleton births in Ontario, Canada between 2006 and 2013. Two measures of greenness, the Normalized Difference Vegetation Index (NDVI) and the Green View Index (GVI), were assigned to the residential histories of these infants from pregnancy through to 12 years of age. Longitudinally-based diagnoses of asthma were determined by using provincial administrative health data. The extended Cox hazards model was used to characterize associations between greenness measures and asthma (up to age 12 years) while adjusting for several risk factors. In a fully adjusted model, that included a term for traffic-related air pollution (NO2), we found no association between an interquartile range increase (0.08) of the NDVI during childhood and asthma incidence (HR = 0.99; 95 % CI = 0.99-1.01). In contrast, we found that an 0.08 increase in NDVI during childhood reduced the risk of asthma in children 7-12 years of age by 14 % (HR = 0.86, 95 % CI:0.79-0.95). Seasonal differences in the association between greenness and asthma were noted. Our findings suggest that residential proximity to greenness reduces the risk of asthma in children aged 7-12.

The use of secondary databases has become popular for evaluating the effectiveness and safety of interventions in real-life settings. However, the absence of important confounders in these databases is challenging. To address this issue, the high-dimensional propensity score (hdPS) algorithm was developed in 2009. This algorithm uses proxy variables for mitigating confounding by combining information available across several healthcare dimensions. This study assessed the methodology and reporting of the hdPS in comparative effectiveness and safety research.

In this methodological review, we searched PubMed and Google Scholar from July 2009 to May 2022 for studies that used the hdPS for evaluating the effectiveness or safety of healthcare interventions. Two reviewers independently extracted study characteristics and assessed how the hdPS was applied and reported. Risk of bias was evaluated with the Risk Of Bias In Non-randomised Studies - of Interventions (ROBINS-I) tool.

In total, 136 studies met the inclusion criteria; the median publication year was 2018 (Q1-Q3 2016-2020). The studies included 192 datasets, mostly North American databases (n = 132, 69%). The hdPS was used in primary analysis in 120 studies (88%). Dimensions were defined in 101 studies (74%), with a median of 5 (Q1-Q3 4-6) dimensions included. A median of 500 (Q1-Q3 200-500) empirically identified covariates were selected. Regarding hdPS reporting, only 11 studies (8%) reported all recommended items. Most studies (n = 81, 60%) had a moderate overall risk of bias.

There is room for improvement in the reporting of hdPS studies, especially regarding the transparency of methodological choices that underpin the construction of the hdPS.

Scientists aim to remediate artifacts in their experimental datasets. However, the remediation of one artifact can result in another. Why might this happen, and what does this consequence tell us about how we should account for artifacts and their control? In this paper, I explore a case in functional neuroimaging where remediation appears to have caused this problem. I argue that remediation amounts to a change to an experimental arrangement. These changes need not be surgical, and the arrangement need not satisfy the criterion of causal modularity. Thus, remediation can affect more than just the factor responsible for the artifact. However, if researchers can determine the consequences of their remediation, they can make adjustments that control for the present artifact as well as for previously controlled ones. Current philosophical accounts of artifacts and the factors responsible for them cannot adequately address this issue, as they do not account for what is needed for artifact remediation (and specifically correction). I support my argument by paralleling it with ongoing concerns regarding the transparency of complex computational systems, as near future remediation across the experimental life sciences will likely make greater use of AI tools to correct for artifacts.

Recent advancements in single-cell technologies allow characterization of experimental perturbations at single-cell resolution. While methods have been developed to analyze such experiments, the application of a strict causal framework has not yet been explored for the inference of treatment effects at the single-cell level. Here we present a causal-inference-based approach to single-cell perturbation analysis, termed CINEMA-OT (causal independent effect module attribution + optimal transport). CINEMA-OT separates confounding sources of variation from perturbation effects to obtain an optimal transport matching that reflects counterfactual cell pairs. These cell pairs represent causal perturbation responses permitting a number of novel analyses, such as individual treatment-effect analysis, response clustering, attribution analysis, and synergy analysis. We benchmark CINEMA-OT on an array of treatment-effect estimation tasks for several simulated and real datasets and show that it outperforms other single-cell perturbation analysis methods. Finally, we perform CINEMA-OT analysis of two newly generated datasets: (1) rhinovirus and cigarette-smoke-exposed airway organoids, and (2) combinatorial cytokine stimulation of immune cells. In these experiments, CINEMA-OT reveals potential mechanisms by which cigarette-smoke exposure dulls the airway antiviral response, as well as the logic that governs chemokine secretion and peripheral immune cell recruitment.

We present a simulation study and application that shows inclusion of binary proxy variables related to binary unmeasured confounders improves the estimate of a related treatment effect in binary logistic regression. The simulation study included 60,000 randomly generated parameter scenarios of sample size 10,000 across six different simulation structures. We assessed bias by comparing the probability of finding the expected treatment effect relative to the modeled treatment effect with and without the proxy variable. Inclusion of a proxy variable in the logistic regression model significantly reduced the bias of the treatment or exposure effect when compared to logistic regression without the proxy variable. Including proxy variables in the logistic regression model improves the estimation of the treatment effect at weak, moderate, and strong association with unmeasured confounders and the outcome, treatment, or proxy variables. Comparative advantages held for weakly and strongly collapsible situations, as the number of unmeasured confounders increased, and as the number of proxy variables adjusted for increased.

Resting-state functional magnetic resonance imaging (rs-fMRI) is a non-invasive imaging technique widely used in neuroscience to understand the functional connectivity of the human brain. While rs-fMRI multi-site data can help to understand the inner working of the brain, the data acquisition and processing of this data has many challenges. One of the challenges is the variability of the data associated with different acquisitions sites, and different MRI machines vendors. Other factors such as population heterogeneity among different sites, with variables such as age and gender of the subjects, must also be considered. Given that most of the machine-learning models are developed using these rs-fMRI multi-site data sets, the intrinsic confounding effects can adversely affect the generalizability and reliability of these computational methods, as well as the imposition of upper limits on the classification scores. This work aims to identify the phenotypic and imaging variables producing the confounding effects, as well as to control these effects. Our goal is to maximize the classification scores obtained from the machine learning analysis of the Autism Brain Imaging Data Exchange (ABIDE) rs-fMRI multi-site data. To achieve this goal, we propose novel methods of stratification to produce homogeneous sub-samples of the 17 ABIDE sites, as well as the generation of new features from the static functional connectivity values, using multiple linear regression models, ComBat harmonization models, and normalization methods. The main results obtained with our statistical models and methods are an accuracy of 76.4%, sensitivity of 82.9%, and specificity of 77.0%, which are 8.8%, 20.5%, and 7.5% above the baseline classification scores obtained from the machine learning analysis of the static functional connectivity computed from the ABIDE rs-fMRI multi-site data.

Infection prevention and control (IPC) is essential to prevent nosocomial infections. This manuscript aims at investigating the current use and role of robots and smart environments on IPC systems in nosocomial settings METHODS: The systematic literature review was performed following the PRISMA statement. Literature was searched for articles published in the period January 2016 to October 2022. Two authors determined the eligibility of the papers, with conflicting decisions being mitigated by a third. Relevant data was then extracted using an ad-hoc extraction table to facilitate the analysis and narrative synthesis.

The search strategy returned 1520 citations and 17 papers were included. This review identified 3 main areas of interest: hand hygiene and personal protective equipment compliance, automatic infection cluster detection and environments cleaning (ie, air quality control, sterilization). This review demonstrates that IPC practices within hospitals mostly do not rely on automation and robotic technology, and few advancements have been made in this field.

Increasing the awareness of healthcare workers on these technologies, through training and involving them in the design process, is essential to accomplish the Health 4.0 transformation. Research priorities should also be considering how to implement similar or more contextualized alternatives for low-income countries.

Causal feature selection is essential for estimating effects from observational data. Identifying confounders is a crucial step in this process. Traditionally, researchers employ content-matter expertise and literature review to identify confounders. Uncontrolled confounding from unidentified confounders threatens validity, conditioning on intermediate variables (mediators) weakens estimates, and conditioning on common effects (colliders) induces bias. Additionally, without special treatment, erroneous conditioning on variables combining roles introduces bias. However, the vast literature is growing exponentially, making it infeasible to assimilate this knowledge. To address these challenges, we introduce a novel knowledge graph (KG) application enabling causal feature selection by combining computable literature-derived knowledge with biomedical ontologies. We present a use case of our approach specifying a causal model for estimating the total causal effect of depression on the risk of developing Alzheimer's disease (AD) from observational data.

We extracted computable knowledge from a literature corpus using three machine reading systems and inferred missing knowledge using logical closure operations. Using a KG framework, we mapped the output to target terminologies and combined it with ontology-grounded resources. We translated epidemiological definitions of confounder, collider, and mediator into queries for searching the KG and summarized the roles played by the identified variables. We compared the results with output from a complementary method and published observational studies and examined a selection of confounding and combined role variables in-depth.

Our search identified 128 confounders, including 58 phenotypes, 47 drugs, 35 genes, 23 collider, and 16 mediator phenotypes. However, only 31 of the 58 confounder phenotypes were found to behave exclusively as confounders, while the remaining 27 phenotypes played other roles. Obstructive sleep apnea emerged as a potential novel confounder for depression and AD. Anemia exemplified a variable playing combined roles.

Our findings suggest combining machine reading and KG could augment human expertise for causal feature selection. However, the complexity of causal feature selection for depression with AD highlights the need for standardized field-specific databases of causal variables. Further work is needed to optimize KG search and transform the output for human consumption.

Depression frequently coexists with chronic pain. Contemporary models suggest that these conditions share pathobiological mechanisms, prompting a need to investigate their temporal association. This investigation aimed to explore two distinctly different chronic pain conditions, and their cross-sectional and prospective associations with depression.

Self-reported information was available on chronic widespread pain (CWP), chronic low back pain (CLBP) (45 years), and depression symptoms (45 and 50 years) from up to 9,377 participants in the 1958 British cohort. Depression symptom outcomes were derived by "Clinical Interview Schedule-Revised" (45 years) and "Short Form-36" (50 years). Relationships between both chronic pain conditions and depression symptoms were investigated by fitting four separate logistic regression models, each with varying levels of covariate adjustment, including depression at baseline.

CWP was associated with depression symptoms cross-sectionally (odds ratio [OR] = 2.04, 95% confidence interval [CI] 1.65, 2.52; P < 0.001, n = 7,629), and prospectively when fully adjusted for baseline, sociodemographic, lifestyle, and health covariates (OR = 1.45, 95% CI 1.17, 1.80; P = < 0.001, n = 6,275). CLBP was associated with depression symptoms prospectively (full model: OR = 1.28, 95% CI 1.01, 1.61; P = 0.04, n = 6,288). In fully adjusted models the prospective association of CWP with depression symptoms was more heavily influenced by our covariates than CLBP with depression symptoms.

Pain may be a stressor from which depression can arise. Development of depression may be differentially dependant upon the type of pain experienced. Screening for depression symptoms among individuals with both chronic pain conditions is indicated and should be repeated over time.

To estimate causal effects, analysts performing observational studies in health settings utilize several strategies to mitigate bias due to confounding by indication. There are two broad classes of approaches for these purposes: use of confounders and instrumental variables (IVs). Because such approaches are largely characterized by untestable assumptions, analysts must operate under an indefinite paradigm that these methods will work imperfectly. In this tutorial, we formalize a set of general principles and heuristics for estimating causal effects in the two approaches when the assumptions are potentially violated. This crucially requires reframing the process of observational studies as hypothesizing potential scenarios where the estimates from one approach are less inconsistent than the other. While most of our discussion of methodology centers around the linear setting, we touch upon complexities in non-linear settings and flexible procedures such as target minimum loss-based estimation and double machine learning. To demonstrate the application of our principles, we investigate the use of donepezil off-label for mild cognitive impairment. We compare and contrast results from confounder and IV methods, traditional and flexible, within our analysis and to a similar observational study and clinical trial.

Little is still known of how multiple urban exposures interact as health determinants. This study investigated various ways in which greenspace, traffic-related air pollution, and noise could operate together, influencing general health status.

In 2022, a cross-sectional population-based survey was conducted in Sofia, Bulgaria. Included were 917 long-term adult residents who completed questionnaires on poor self-rated health (PSRH), total time spent in physical activity (PA), home garden presence, time spent in urban greenspace and nature, and sociodemographics. Residential greenspace was operationalized using the normalized difference vegetation index (NDVI), tree cover density, number of trees, and access to local greenspace and parks. Nitrogen dioxide (NO₂) was modeled for the study area. Road traffic, railway, and aircraft day-evening-night sound levels (L_den) were extracted from EU noise maps. Area-level income and urbanicity were considered. Analyses included multivariate ordinal regressions, interactions, and structural equation modeling (SEM).

Associations with PSRH were per 0.10 NDVI _300 m: OR = 0.65 (0.42-1.01), home garden: OR = 0.72 (0.49-1.07), per 5 μg/m³ NO₂: OR = 1.57 (1.00-2.48), per 5 dB(A) L_den road traffic: OR = 1.06 (0.91-1.23), railway: OR = 1.11 (1.03-1.20), and aircraft: OR = 1.22 (1.11-1.34). Spending >30 min/week in nature related to better health. In multi-exposure models, only associations with aircraft and railway L_den persisted. People with lower education and financial difficulties or living in poorer districts experienced some exposures stronger. In SEM, time spent in nature and PA mediated the effect of greenspace.

Greenspace was associated with better general health, with time spent in nature and PA emerging as intermediate pathways. NO₂, railway, and aircraft noise were associated with poorer general health. These results could inform decision-makers, urban planners, and civil society organizations facing urban development problems. Mitigation and abatement policies and measures should target socioeconomically disadvantaged citizens.

To study whether endometriosis is associated with pregnancy loss and recurrent pregnancy loss (RPL).

Nationwide historical cohort study with a nested case-control analysis.

National health registers.

A total of 29,563 women born between 1957 and 1997 were identified in the national health registers, diagnosed with endometriosis between 1977 and 2017, and age-matched 1:10 with 295,630 women without endometriosis. The number of pregnancy losses was assessed, and data were analyzed with conditional logistic regression.

Endometriosis (International Classification of Diseases, 8th Revision, 62530-62539, and International Classification of Diseases, 10th Revision, DN80.0-9).

The primary outcomes of interest were the numbers of pregnancy losses categorized as 0, 1, 2, and ≥ 3 losses, unadjusted and adjusted for gravidity, and RPL. The secondary outcome measures were the predefined types of pregnancy losses. Pregnancy loss was defined as the spontaneous demise of a pregnancy until 22 weeks of gestation. Primary RPL was defined as 3 or more consecutive pregnancy losses with no prior live birth or stillbirth, and secondary RPL was defined as 1 or more births followed by 3 or more consecutive losses.

A total of 18.9%, 3.9%, and 2.1% of ever-pregnant women with endometriosis had 1, 2, and ≥ 3 pregnancy losses compared with 17.3%, 3.5%, and 1.5% of the women without endometriosis, corresponding to the odds ratios of 1.13 (95% confidence interval, 1.09-1.17), 1.18 (1.10-1.26), and 1.44 (1.31-1.59), respectively. When adjusted also for gravidity, the corresponding results were 1.37 (95% confidence interval, 1.32-1.42), 1.75 (1.62-1.89), and 2.57 (2.31-2.85), respectively. The following predefined subgroups of RPL were positively associated with endometriosis: primary; secondary; secondary after giving birth to a boy; after a complicated delivery; and ≥ 3 pregnancy losses before the age of 30 years. Six endometriosis subgroup analyses found an association between endometriosis and pregnancy loss. These analyses were women diagnosed in the 4 decades between 1977 and 2017, women with adenomyosis, and women with adenomyosis only.

This nationwide cohort study found endometriosis to be associated with pregnancy loss and RPL, and the association strengthened with an increasing number of losses.

Observational studies have shown an association between statin or aspirin use and a decreased risk of HCC, but the effects of a well-defined treatment strategy remain unknown. We emulated trials of the effects of continuous statin or aspirin use on HCC risk in patients with cirrhosis due to alcohol-related liver disease (ALD cirrhosis).

We specified target trials for statins and, separately, aspirin and emulated them using Danish health care registries. All eligible patients with ALD cirrhosis diagnosed in 2000-2018 were included in either an exposed or an unexposed arm. Patients were followed until HCC or death without HCC. The 5-year risk of HCC was estimated using marginal structural models with inverse probability weighting. Using statins continuously for 5 years compared with not using statins resulted in a relative risk (RR) of HCC of 0.67 (95% CI: 0.45-0.91). The RR of death without HCC was 0.69 (95% CI: 0.65-0.77). For aspirin, the RR was 1.05 (95% CI: 0.60-1.42) for HCC and 1.02 (95% CI: 0.95-1.09) for death without HCC.

In patients with ALD cirrhosis, 5 years of continuous statin use resulted in a 33% RR reduction of HCC (number needed to treat = 94) and a 31% RR reduction of death without HCC (number needed to treat = 7). Such strong causal effects are implausible and best explained by uncontrollable confounding, highlighting the need for randomized trials. Aspirin use likely does not affect the risk of HCC or death without HCC.

Causal diagrams are used in biomedical research to develop and portray conceptual models that accurately and concisely convey assumptions about putative causal relations. Specifically, causal diagrams can be used for both observational studies and clinical trials to provide a scientific basis for some aspects of multivariable model selection. This methodology also provides an explicit framework for classifying potential sources of bias and enabling the identification of confounder, collider, and mediator variables for statistical analyses. We illustrate the potential serious miscalculation of effect estimates resulting from incorrect selection of variables for multivariable modeling without regard to their type and causal ordering as demonstrated by causal diagrams. Our objective is to improve researchers' understanding of the critical variable selection process to enhance their communication with collaborating statisticians regarding the scientific basis for intended study designs and multivariable statistical analyses. We introduce the concept of causal diagrams and their development as directed acyclic graphs to illustrate the usefulness of this methodology. We present numeric examples of effect estimate calculations and miscalculations based on analyses of the well-known Framingham Heart Study. Clinical researchers can use causal diagrams to improve their understanding of complex causation relations to determine accurate and valid multivariable models for statistical analyses. The Framingham Heart Study dataset and software codes for our analyses are provided in Appendix E1 (available online at http://www.annemergmed.com) to allow readers the opportunity to conduct their analyses.

Two important considerations in clinical research studies are proper evaluations of internal and external validity. While randomized clinical trials can overcome several threats to internal validity, they may be prone to poor external validity. Conversely, large prospective observational studies sampled from a broadly generalizable population may be externally valid, yet susceptible to threats to internal validity, particularly confounding. Thus, methods that address confounding and enhance transportability of study results across populations are essential for internally and externally valid causal inference, respectively. These issues persist for another problem closely related to transportability known as data-fusion. We develop a calibration method to generate balancing weights that address confounding and sampling bias, thereby enabling valid estimation of the target population average treatment effect. We compare the calibration approach to two additional doubly robust methods that estimate the effect of an intervention on an outcome within a second, possibly unrelated target population. The proposed methodologies can be extended to resolve data-fusion problems that seek to evaluate the effects of an intervention using data from two related studies sampled from different populations. A simulation study is conducted to demonstrate the advantages and similarities of the different techniques. We also test the performance of the calibration approach in a motivating real data example comparing whether the effect of biguanides vs sulfonylureas-the two most common oral diabetes medication classes for initial treatment-on all-cause mortality described in a historical cohort applies to a contemporary cohort of US Veterans with diabetes.

The extent and severity of post-COVID-19 mental health symptoms among frontline clinicians are not clear. This study compared mental health symptoms (i.e., depression, anxiety, and insomnia symptoms) and global quality of life (QOL) after the first COVID-19 outbreak between the COVID-19 treating and non-COVID-19 treating frontline clinicians.

This cross-sectional, comparative, convenient-sampling study was conducted between October 13 and 22, 2020, which was five months after the first COVID-19 outbreak in China was brought under control. The severity of depression, anxiety, insomnia symptoms, and global QOL of the clinicians were assessed using the Patient Health Questionnaire-9 items (PHQ-9), Generalized Anxiety Disorder Scale-7 items (GAD-7), Insomnia Severity Index (ISI), and the World Health Organization Quality of Life Questionnaire-brief version (WHOQOL-BREF), respectively. The propensity score matching (PSM) method was used to identify comparable COVID-19 treating and non-COVID-19 treating frontline clinicians. A generalized linear model (GLM) was used to assess the differences in PHQ-9, GAD-7, ISI, and QOL scores between the COVID-19 treating and non-COVID-19 treating frontline clinicians.

In total, 260 COVID-19 treating frontline clinicians and 260 matched non- COVID-19 treating frontline clinicians were included. Non-COVID-19 treating frontline clinicians experienced more frequent workplace violence (WPV) than the COVID-19 treating frontline clinicians (χ² = 7.6, p = 0.006). COVID-19 treating frontline clinicians reported higher QOL compared to their non-COVID-19 treating frontline counterparts (b = 0.3, p = 0.042), after adjusting for WPV experience. COVID-19 treating and non- COVID-19 treating frontline clinicians reported similar PHQ-9, GAD-7, and ISI total scores (all p values > 0.05).

This study did not reveal more severe post-COVID-19 mental health symptoms in COVID-19 treating frontline clinicians compared to non-COVID-19 treating frontline clinicians. It is possible that the implementation of timely and appropriate mental health, social and financial supports could have prevented the worsening of mental health symptoms among the COVID-19 treating frontline clinicians after the first COVID-19 outbreak in China.

The pharmacoepidemiology of the long-term benefits and harms of medicines in pregnancy and breastfeeding has received little attention. The impact of maternal medicines on children is increasingly recognised as a source of avoidable harm. The focus of attention has expanded from congenital anomalies to include less visible, but equally important, outcomes, including cognition, neurodevelopmental disorders, educational performance, and childhood ill-health. Breastfeeding, whether as a source of medicine exposure, a mitigator of adverse effects or as an outcome, has been all but ignored in pharmacoepidemiology and pharmacovigilance: a significant 'blind spot'.

WHY WE NEED THEM: Optimal child development and maternal health necessitate breastfeeding, yet little information exists to guide families regarding the safety of medicine use during lactation. Breastfeeding initiation or success may be altered by medicine use, and breastfeeding may obscure the true relationship between medicine exposure during pregnancy and developmental outcomes. Absent or poorly standardised recording of breastfeeding in most population databases hampers analysis and understanding of the complex relationships between medicine, pregnancy, breastfeeding and infant and maternal health. The purpose of this paper is to present the arguments for breastfeeding to be included alongside medicine use and neurodevelopmental outcomes in whole-population database investigations of the harms and benefits of medicines during pregnancy, the puerperium and postnatal period. We review: 1) the current situation, 2) how these complexities might be accommodated in pharmacoepidemiological models, using antidepressants and antiepileptics as examples; 3) the challenges in obtaining comprehensive data.

The scarcity of whole-population data and the complexities of the inter-relationships between breastfeeding, medicines, co-exposures and infant outcomes are significant barriers to full characterisation of the benefits and harms of medicines during pregnancy and breastfeeding. This makes it difficult to answer the questions: 'is it safe to breastfeed whilst taking this medicine', and 'will this medicine interfere with breastfeeding and/ or infants' development'?

Studies have shown that COVID-19 has had a worldwide psychological impact. Confinement due to COVID-19 has had important repercussions on the mental health of the general population, with high levels of stress, anxiety, depressive symptoms, post-traumatic stress disorder, and so forth. Similarly, important labor, economic and social changes taking place are affecting people's well-being. The objective of this study was to analyze the repercussions of perceived threat from COVID-19 on the mental health of the population, and to evaluate the mediating role of perceived economic impact. The participants were 1160 adult residents of Spain aged 18 to 82, 69.9% of whom were women. A sociodemographic questionnaire, the Questionnaire on Perception of Threat from COVID-19 and the General Health Questionnaire were administered. Perceived threat had a positive direct effect on all four health dimensions analyzed. Among the indirect effects, the perceived economic impact of COVID-19 mediated in the relationship between perceived threat and each of the health dimensions. The results of this study have demonstrated the need to promote joint action promoting public mental health to minimize the psychological repercussions of new outbreaks.

To combat the opioid epidemic in the USA, unprecedented federal funding has been directed to states and territories to expand access to prevention, overdose rescue, and medications for opioid use disorder (MOUD). Similar to other states, California rapidly allocated these funds to increase reach and adoption of MOUD in safety-net, primary care settings such as Federally Qualified Health Centers. Typical of current real-world implementation endeavors, a package of four implementation strategies was offered to all clinics. The present study examines (i) the pre-post effect of the package of strategies, (ii) whether/how this effect differed between new (start-up) versus more established (scale-up) MOUD practices, and (iii) the effect of clinic engagement with each of the four implementation strategies.

Forty-one primary care clinics were offered access to four implementation strategies: (1) Enhanced Monitoring and Feedback, (2) Learning Collaboratives, (3) External Facilitation, and (4) Didactic Webinars. Using linear mixed effects models, RE-AIM guided outcomes of reach, adoption, and implementation quality were assessed at baseline and at 9 months follow-up.

Of the 41 clinics, 25 (61%) were at MOUD start-up and 16 (39%) were at scale-up phases. Pre-post difference was observed for the primary outcome of percent of patient prescribed MOUD (reach) (βtime = 3.99; 0.73 to 7.26; p = 0.02). The largest magnitude of change occurred in implementation quality (ES = 0.68; 95% CI = 0.66 to 0.70). Baseline MOUD capability moderated the change in reach (start-ups 22.60%, 95% CI = 16.05 to 29.15; scale-ups -4.63%, 95% CI = -7.87 to -1.38). Improvement in adoption and implementation quality were moderately associated with early prescriber engagement in Learning Collaboratives (adoption: ES = 0.61; 95% CI = 0.25 to 0.96; implementation quality: ES = 0.55; 95% CI = 0.41 to 0.69). Improvement in adoption was also associated with early prescriber engagement in Didactic Webinars (adoption: ES = 0.61; 95% CI = 0.20 to 1.05).

Rather than providing an all-clinics-get-all-components package of implementation strategies, these data suggest that it may be more efficient and effective to tailor the provision of implementation strategies based on the needs of clinic. Future implementation endeavors could benefit from (i) greater precision in the provision of implementation strategies based on contextual determinants, and (ii) the inclusion of strategies targeting engagement.

Regular dental visits are essential for the prevention, early detection and treatment of worldwide highly prevalent oral diseases. Personality traits were previously associated with treatment compliance, medication adherence and regular doctor visits, however, the link between personality traits and regular dental visit attendance remains largely unexplored. Thus, the objective of this study is to clarify this link.

Data (wave 7) of the Survey of Health, Ageing and Retirement in Europe (SHARE) were used, focusing on Germany (n = 2822). Personality was assessed using the 10-item Big Five Inventory (BFI-10). Regular dental visits were assessed. Multiple logistic regressions were used, adjusting for various covariates.

Majority of the participants (84%) reported to attend regular dental visits during lifetime. Regularity of lifetime dental visit attendance was positively and significantly associated with increased extraversion [OR 1.13, 95% CI (1.01-1.26)], increased conscientiousness [OR 1.26, 95% CI (1.10-1.44)], and increased openness to experience [OR 1.12, 95% CI (1.01-1.26)]. However, there was a lack of association with agreeableness and neuroticism. Moreover, the outcome measure was positively associated with younger age, being female, born in Germany, being married, higher education, being retired (compared to being homemaker), whereas it was not associated with obesity or chronic diseases.

Identification of personality traits that are associated with regular dental visits can support prevention, screening and clinical management of oral diseases. Further research in this field may facilitate the development and increase the incorporation of individualized concepts to enhance patient compliance and attendance, and thus the provision of oral and dental care services.

In the last decade, Bayesian networks (BNs) have been widely used in engineering risk assessment due to the benefits that they provide over other methods. Among these, the most significant is the ability to model systems, causal factors, and their dependencies in a probabilistic manner. This capability has enabled the community to do causal reasoning through associations, which answers questions such as: "How does new evidence x ' $x^{\prime }$ about the occurrence of event X $X$ change my belief about the occurrence of event Y $Y$ ?" Associative reasoning has helped risk analysts to identify relevant risk-contributing factors and perform scenario analysis by evidence propagation. However, engineering risk assessment has yet to explore other features of BNs, such as the ability to reason through interventions, which enables the BN model to support answering questions of the form "How does doing X = x ' $X=x^{\prime }$ change my belief about the occurrence of event Y $Y$ ?" In this article, we propose to expand the scope of use of BN models in engineering risk assessment to support intervention reasoning. This will provide more robust risk-informed decision support by enabling the modeling of policies and actions before being implemented. To do this, we provide the formal mathematical background and tools to model interventions in BNs and propose a framework that enables its use in engineering risk assessment. This is demonstrated in an illustrative case study on third-party damage of natural gas pipelines, showing how BNs can be used to inform decision-makers about the effect that new actions/policies can have on a system.

The scale of genetic, epigenomic, transcriptomic, cheminformatic and proteomic data available today, coupled with easy-to-use machine learning (ML) toolkits, has propelled the application of supervised learning in genomics research. However, the assumptions behind the statistical models and performance evaluations in ML software frequently are not met in biological systems. In this Review, we illustrate the impact of several common pitfalls encountered when applying supervised ML in genomics. We explore how the structure of genomics data can bias performance evaluations and predictions. To address the challenges associated with applying cutting-edge ML methods to genomics, we describe solutions and appropriate use cases where ML modelling shows great potential.

To evaluate class suspension and mass vaccination implemented among Taipei schoolchildren during the 2009 influenza pandemic and investigate factors affecting antibody responses.

We conducted 2 cohort studies on: (1) 972 schoolchildren from November 2009-March 2010 to evaluate pandemic policies and (2) 935 schoolchildren from November 2011-March 2012 to verify factors in antibody waning. Anti-influenza H1N1pdm09 hemagglutination inhibition antibodies (HI-Ab) were measured from serum samples collected before vaccination, and at 1 and 4 months after vaccination. Factors affecting HI-Ab responses were investigated through logistic regression and generalized estimating equation.

Seroprevalence of H1N1pdm09 before vaccination was significantly higher among schoolchildren who experienced class suspensions than those who did not (59.6% vs 47.5%, p<0.05). Participating in after-school activities (adjusted odds ratio [aOR]=2.47, p=0.047) and having ≥3 hours per week of exercise (aOR=2.86, p=0.019) were significantly correlated with H1N1pdm09 infection. Two doses of the H1N1pdm09 vaccine demonstrated significantly better antibody persistence than 1 dose (HI-Ab geometric mean titer: 132.5 vs 88.6, p=0.047). Vaccine effectiveness after controlling for preexisting immunity was 86% (32%-97%). Exercise ≥3 hours per week and preexisting immunity were significantly associated with antibody waning/maintenance.

This study is the first to show that exercise and preexisting immunity may affect antibody waning. Further investigation is needed to identify immune correlates of protection.

Frequency matching is commonly used in epidemiological case control studies to balance the distributions of the matching factors between the case and control groups and to improve the efficiency of case-control designs. Applied researchers have held a common opinion that unconditional logistic regression should be used to analyze frequency matched designs and conditional logistic regression is unnecessary. However, the justification of this view is unclear. To compare the performances of ULR and CLR in terms of simplicity, unbiasedness, and efficiency in a more intuitive way, we viewed frequency matching from the perspective of weighted sampling and derived the outcome models describing how the exposure and matching factors are associated with the outcome in the matched data separately in two scenarios: (1) only categorical variables are used for matching; (2) continuous variables are categorized for matching. In either scenario the derived outcome model is a logit model with stratum-specific intercepts. Correctly specified unconditional logistic regression can be more efficient than conditional logistic regression, particularly when continuous matching factors are used, whereas conditional logistic regression is a more practical approach because it is less dependent on modeling choices.

Context dependence is widely invoked to explain disparate results in ecology. It arises when the magnitude or sign of a relationship varies due to the conditions under which it is observed. Such variation, especially when unexplained, can lead to spurious or seemingly contradictory conclusions, which can limit understanding and our ability to transfer findings across studies, space, and time. Using examples from biological invasions, we identify two types of context dependence resulting from four sources: mechanistic context dependence arises from interaction effects; and apparent context dependence can arise from the presence of confounding factors, problems of statistical inference, and methodological differences among studies. Addressing context dependence is a critical challenge in ecology, essential for increased understanding and prediction.

The Adolescent Brain Cognitive Development (ABCD) Study is the largest single-cohort prospective longitudinal study of neurodevelopment and children's health in the United States. A cohort of n = 11,880 children aged 9-10 years (and their parents/guardians) were recruited across 22 sites and are being followed with in-person visits on an annual basis for at least 10 years. The study approximates the US population on several key sociodemographic variables, including sex, race, ethnicity, household income, and parental education. Data collected include assessments of health, mental health, substance use, culture and environment and neurocognition, as well as geocoded exposures, structural and functional magnetic resonance imaging (MRI), and whole-genome genotyping. Here, we describe the ABCD Study aims and design, as well as issues surrounding estimation of meaningful associations using its data, including population inferences, hypothesis testing, power and precision, control of covariates, interpretation of associations, and recommended best practices for reproducible research, analytical procedures and reporting of results.

Integration of external control data, with patient-level information, in clinical trials has the potential to accelerate the development of new treatments in neuro-oncology by contextualising single-arm studies and improving decision making (eg, early stopping decisions). Based on a series of presentations at the 2020 Clinical Trials Think Tank hosted by the Society of Neuro-Oncology, we provide an overview on the use of external control data representative of the standard of care in the design and analysis of clinical trials. High-quality patient-level records, rigorous methods, and validation analyses are necessary to effectively leverage external data. We review study designs, statistical methods, risks, and potential distortions in using external data from completed trials and real-world data, as well as data sources, data sharing models, ongoing work, and applications in glioblastoma.