Gaps in the usage and reporting of multiple imputation for incomplete data: findings from a scoping review of observational studies addressing causal questions.

BACKGROUND: Missing data are common in observational studies and often occur in several of the variables required when estimating a causal effect, i.e. the exposure, outcome and/or variables used to control for confounding. Analyses involving multiple incomplete variables are not as straightforward as analyses with a single incomplete variable. For example, in the context of multivariable missingness, the standard missing data assumptions ("missing completely at random", "missing at random" [MAR], "missing not at random") are difficult to interpret and assess. It is not clear how the complexities that arise due to multivariable missingness are being addressed in practice. The aim of this study was to review how missing data are managed and reported in observational studies that use multiple imputation (MI) for causal effect estimation, with a particular focus on missing data summaries, missing data assumptions, primary and sensitivity analyses, and MI implementation. METHODS: We searched five top general epidemiology journals for observational studies that aimed to answer a causal research question and used MI, published between January 2019 and December 2021. Article screening and data extraction were performed systematically. RESULTS: Of the 130 studies included in this review, 108 (83%) derived an analysis sample by excluding individuals with missing data in specific variables (e.g., outcome) and 114 (88%) had multivariable missingness within the analysis sample. Forty-four (34%) studies provided a statement about missing data assumptions, 35 of which stated the MAR assumption, but only 11/44 (25%) studies provided a justification for these assumptions. The number of imputations, MI method and MI software were generally well-reported (71%, 75% and 88% of studies, respectively), while aspects of the imputation model specification were not clear for more than half of the studies. A secondary analysis that used a different approach to handle the missing data was conducted in 69/130 (53%) studies. Of these 69 studies, 68 (99%) lacked a clear justification for the secondary analysis. CONCLUSION: Effort is needed to clarify the rationale for and improve the reporting of MI for estimation of causal effects from observational data. We encourage greater transparency in making and reporting analytical decisions related to missing data.

Genetics causal analysis of oral microbiome on type 2 diabetes in East Asian populations: a bidirectional two-sample Mendelian randomized study.

Introduction: The dysbiosis of the oral microbiome is associated with the progression of various systemic diseases, including diabetes. However, the precise causal relationships remain elusive. This study aims to investigate the potential causal associations between oral microbiome and type 2 diabetes (T2D) using Mendelian randomization (MR) analyses. Methods: We conducted bidirectional two-sample MR analyses to investigate the impact of oral microbiome from saliva and the tongue T2D. This analysis was based on metagenome-genome-wide association studies (mgGWAS) summary statistics of the oral microbiome and a large meta-analysis of GWAS of T2D in East Asian populations. Additionally, we utilized the T2D GWAS summary statistics from the Biobank Japan (BBJ) project for replication. The MR methods employed included Wald ratio, inverse variance weighting (IVW), weighted median, MR-Egger, contamination mixture (ConMix), and robust adjusted profile score (RAPS). Results: Our MR analyses revealed genetic associations between specific bacterial species in the oral microbiome of saliva and tongue with T2D in East Asian populations. The MR results indicated that nine genera were shared by both saliva and tongue. Among these, the genera Aggregatibacter, Pauljensenia, and Prevotella were identified as risk factors for T2D. Conversely, the genera Granulicatella and Haemophilus D were found to be protective elements against T2D. However, different species within the genera Catonella, Lachnoanaerobaculum, Streptococcus, and Saccharimonadaceae TM7x exhibited multifaceted influences; some species were positively correlated with the risk of developing T2D, while others were negatively correlated. Discussion: This study utilized genetic variation tools to confirm the causal effect of specific oral microbiomes on T2D in East Asian populations. These findings provide valuable insights for the treatment and early screening of T2D, potentially informing more targeted and effective therapeutic strategies.

Inpatient Hospitalizations for COVID-19 Among Patients with Prader-Willi Syndrome: a National Inpatient Sample Analysis.

Background: Prader-Willi syndrome (PWS) is a genetic disorder associated with baseline respiratory impairment caused by multiple contributing etiologies. While this may be expected to increase the risk of severe COVID-19 infections in PWS patients, survey studies have suggested paradoxically low disease severity. To better characterize the course of COVID-19 infection in patients with PWS, this study analyzes the outcomes of hospitalizations for COVID-19 among patients with and without PWS. Methods: The National Inpatient Sample, an all-payors administrative claims database of hospitalizations in the United States, was queried for patients with a coded diagnosis COVID-19 in 2020 and 2021. Hospitalizations for patients with PWS compared to those for patients without PWS using Augmented Inverse Propensity Weighting (AIPW). Results: There were 295 (95% CI: 228 to 362) COVID-19 hospitalizations for individuals with PWS and 4,112,400 (95% CI: 4,051,497 to 4,173,303) for individuals without PWS. PWS patients had a median age of 33 years compared to 63 for those without PWS. Individuals with PWS had higher baseline rates of obesity (47.5% vs. 28.4%). AIPW models show that PWS diagnosis is associated with increased hospital length of stay by 7.43 days, hospital charges by $80,126, and the odds of mechanical ventilation and in-hospital death (odds ratios of 1.79 and 1.67, respectively). Conclusions: PWS patients hospitalized with COVID-19 experienced longer hospital stays, higher charges, and increased risk of mechanical ventilation and death. PWS should be considered a risk factor for severe COVID-19, warranting continued protective measures and vaccination efforts. Further research is needed to validate coding for PWS and assess the impact of evolving COVID-19 variants and population immunity on this vulnerable population.

Species interactions drive continuous assembly of freshwater communities in stochastic environments.

Understanding the factors driving the maintenance of long-term biodiversity in changing environments is essential for improving restoration and sustainability strategies in the face of global environmental change. Biodiversity is shaped by both niche and stochastic processes, however the strength of deterministic processes in unpredictable environmental regimes is highly debated. Since communities continuously change over time and space-species persist, disappear or (re)appear-understanding the drivers of species gains and losses from communities should inform us about whether niche or stochastic processes dominate community dynamics. Applying a nonparametric causal discovery approach to a 30-year time series containing annual abundances of benthic invertebrates across 66 locations in New Zealand rivers, we found a strong negative causal relationship between species gains and losses directly driven by predation indicating that niche processes dominate community dynamics. Despite the unpredictable nature of these system, environmental noise was only indirectly related to species gains and losses through altering life history trait distribution. Using a stochastic birth-death framework, we demonstrate that the negative relationship between species gains and losses can not emerge without strong niche processes. Our results showed that even in systems that are dominated by unpredictable environmental variability, species interactions drive continuous community assembly.

Exploring genetic associations between metabolites and atopic dermatitis: insights from bidirectional Mendelian randomization analysis in European population.

Introduction: There is growing evidence indicating a complex interaction between blood metabolites and atopic dermatitis (AD). The objective of this study was to investigate and quantify the potential influence of plasma metabolites on AD through Mendelian randomization (MR) analysis. Methods: Our procedures followed these steps: instrument variable selection, primary analysis, replication analysis, Meta-analysis of results, reverse MR analysis, and multivariate MR (MVMR) analysis. In our study, the exposure factors were derived from the Canadian Longitudinal Study on Aging (CLSA), encompassing 8,299 individuals of European descent and identifying 1,091 plasma metabolites and 309 metabolite ratios. In primary analysis, AD data, was sourced from the GWAS catalog (Accession ID: GCST90244787), comprising 60,653 cases and 804,329 controls. For replication, AD data from the Finnish R10 database included 15,208 cases and 367,046 controls. We primarily utilized the inverse variance weighting method to assess the causal relationship between blood metabolites and AD. Results: Our study identified significant causal relationships between nine genetically predicted blood metabolites and AD. Specifically, 1-palmitoyl-2-stearoyl-GPC (16:0/18:0) (OR = 0.92, 95% CI 0.89-0.94), 1-methylnicotinamide (OR = 0.93, 95% CI 0.89-0.98), linoleoyl-arachidonoyl-glycerol (18:2/20:4) [1] (OR = 0.94, 95% CI 0.92-0.96), and 1-arachidonoyl-GPC (20:4n6) (OR = 0.94, 95% CI 0.92-0.96) were associated with a reduced risk of AD. Conversely, phosphate / linoleoyl-arachidonoyl-glycerol (18:2/20:4) [2] (OR = 1.07, 95% CI 1.04-1.10), docosatrienoate (22:3n3) (OR = 1.07, 95% CI 1.04-1.10), retinol (Vitamin A) / linoleoyl-arachidonoyl-glycerol (18:2/20:4) [2] (OR = 1.08, 95% CI 1.05-1.11), retinol (Vitamin A) / linoleoyl-arachidonoylglycerol (18:2/20:4) [1] (OR = 1.08, 95% CI 1.05-1.12), and phosphate / linoleoyl-arachidonoyl-glycerol (18:2/20:4) [1] (OR = 1.09, 95% CI 1.07-1.12 were associated with an increased risk of AD. No evidence of reverse causality was found in the previously significant results. MVMR analysis further confirmed that 1-palmitoyl-2-stearoyl-GPC (16:0/18:0) and 1-methylnicotinamide are independent and dominant contributors to the development of AD. Conclusion: Our study revealed a causal relationship between genetically predicted blood metabolites and AD. This discovery offers specific targets for drug development in the treatment of AD patients and provides valuable insights for investigating the underlying mechanisms of AD in future research.

Causal roles of skin and gut microbiota in skin appendage disorders suggested by genetic study.

Objectives: There is evidence from observational studies that human microbiota is linked to skin appendage Disorders (SADs). Nevertheless, the causal association between microbiota and SADs is yet to be fully clarified. Methods: A comprehensive two-sample Mendelian randomization (MR) was first performed to determine the causal effect of skin and gut microbiota on SADs. A total of 294 skin taxa and 211 gut taxa based on phylum, class, order, family, genus, and ASV level information were identified. Summary data of SADs and eight subtypes (acne vulgaris, hidradenitis suppurativa, alopecia areata, rogenic alopecia, rosacea, rhinophyma, seborrhoeic dermatitis, and pilonidal cyst) were obtained from the FinnGen consortium. We performed bidirectional MR to determine whether the skin and gut microbiota are causally associated with multiple SADs. Furthermore, sensitivity analysis was conducted to examine horizontal pleiotropy and heterogeneity. Results: A total of 65 and 161 causal relationships between genetic liability in the skin and gut microbiota with SADs were identified, respectively. Among these, we separately found 5 and 11 strong causal associations that passed Bonferroni correction in the skin and gut microbiota with SADs. Several skin bacteria, such as Staphylococcus, Streptococcus, and Propionibacterium, were considered associated with multiple SADs. As gut probiotics, Bifidobacteria and Lactobacilli were associated with a protective effect on SAD risk. There was no significant heterogeneity in instrumental variables or horizontal pleiotropy. Conclusions: Our MR analysis unveiled bidirectional causal relationships between SADs and the gut and skin microbiota, and had the potential to offer novel perspectives on the mechanistic of microbiota-facilitated dermatosis.

Causal Inference of the Effect of Vaccination on COVID-19 Disease Severity and Need for Intensive Care Unit Admission Among Hospitalized Patients in an African Setting.

Background Coronavirus disease 2019 (COVID-19) is a novel, primarily respiratory, coronavirus that became a pandemic when it spread to over 210 countries and led to the death of over six million people. There is no definitive treatment for COVID-19, but vaccines have been developed that can help prevent severe illness and death. Studies have investigated the effect of vaccination on disease severity and outcome, and the findings indicate that vaccination is linked to a significant reduction in the risk of hospitalization, intensive care unit (ICU) admission, and disease mortality. However, there is a scarcity of evidence in Africa in general, and no similar study has been conducted in Ethiopia yet. Therefore, the study aimed to assess the effect of vaccination on COVID-19 disease severity and the need for ICU admission among hospitalized patients at a private specialty clinic in Ethiopia. Methods A retrospective cohort study was conducted among 126 patients with COVID-19, 41 vaccinated and 85 unvaccinated, who were hospitalized between September 2021 and May 2022. Data were summarized using frequency (percentage) and median (interquartile range (IQR)). To compare the characteristics of the two groups, Chi-square/Fisher's exact and Mann-Whitney U tests at p-values of ≤ 0.05 were used. To identify the effect of vaccination on COVID-19 disease severity, a marginal structural model (MSM) with an inverse probability weighting (IPW) approach using a robust Poisson regression model was fitted. Adjusted relative risk (ARR) and 95% confidence interval (CI) for ARR were used for interpreting the result. Results The cohort included groups that were comparable in terms of their sociodemographic and clinical characteristics. More than half of the participants were older than 60 years (n = 66, 52.4%), were males (n = 71, 56.3%), and had one or more comorbid illnesses (n = 66, 52.4%). At admission, 85 (67.5%) had severe disease, and 11 (8.7%) progressed after hospitalization and required ICU admission, of which three unvaccinated cases died. From the final model, vaccination was found to be associated with a 62% decreased risk of developing severe COVID-19 disease if infected, compared to not getting vaccinated (ARR = 0.38, 95% CI = 0.23-0.65, p < 0.0001). Conclusions The study's findings support previous reports that vaccinated people are less likely to develop severe COVID-19 disease if later infected with the virus, emphasizing the importance of continuing efforts to promote COVID-19 vaccination not only to safeguard individuals but also to confer community-level immunity.

Causal Associations of Sleep Apnea With Alzheimer Disease and Cardiovascular Disease: A Bidirectional Mendelian Randomization Analysis.

BACKGROUND: Sleep apnea (SA) has been linked to an increased risk of dementia in numerous observational studies; whether this is driven by neurodegenerative, vascular, or other mechanisms is not clear. We sought to examine the bidirectional causal relationships between SA, Alzheimer disease (AD), coronary artery disease (CAD), and ischemic stroke using Mendelian randomization. METHODS AND RESULTS: Using summary statistics from 4 recent, large genome-wide association studies of SA (n=523 366), AD (n=94 437), CAD (n=1 165 690), and stroke (n=1 308 460), we conducted bidirectional 2-sample Mendelian randomization analyses. Our primary analytic method was fixed-effects inverse variance-weighted (IVW) Mendelian randomization; diagnostics tests and sensitivity analyses were conducted to verify the robustness of the results. We identified a significant causal effect of SA on the risk of CAD (odds ratio [ORIVW]=1.35 per log-odds increase in SA liability [95% CI=1.25-1.47]) and stroke (ORIVW=1.13 [95% CI=1.01-1.25]). These associations were somewhat attenuated after excluding single-nucleotide polymorphisms associated with body mass index (ORIVW=1.26 [95% CI=1.15-1.39] for CAD risk; ORIVW=1.08 [95% CI=0.96-1.22] for stroke risk). SA was not causally associated with a higher risk of AD (ORIVW=1.14 [95% CI=0.91-1.43]). We did not find causal effects of AD, CAD, or stroke on risk of SA. CONCLUSIONS: These results suggest that SA increased the risk of CAD, and the identified causal association with stroke risk may be confounded by body mass index. Moreover, no causal effect of SA on AD risk was found. Future studies are warranted to investigate cardiovascular pathways between sleep disorders, including SA, and dementia.

Causal inference using multivariate generalized linear mixed-effects models.

Dynamic prediction of causal effects under different treatment regimens is an essential problem in precision medicine. It is challenging because the actual mechanisms of treatment assignment and effects are unknown in observational studies. We propose a multivariate generalized linear mixed-effects model and a Bayesian g-computation algorithm to calculate the posterior distribution of subgroup-specific intervention benefits of dynamic treatment regimes. Unmeasured time-invariant factors are included as subject-specific random effects in the assumed joint distribution of outcomes, time-varying confounders, and treatment assignments. We identify a sequential ignorability assumption conditional on treatment assignment heterogeneity, that is, analogous to balancing the latent treatment preference due to unmeasured time-invariant factors. We present a simulation study to assess the proposed method's performance. The method is applied to observational clinical data to investigate the efficacy of continuously using mycophenolate in different subgroups of scleroderma patients.

Causally interpretable meta-analysis combining aggregate and individual participant data.

Recent work in causally-interpretable meta-analysis (CIMA) has bridged the gap between traditional meta-analysis and causal inference. While traditional meta-analysis results generally do not apply to any well-defined population, CIMA approaches specify a target population to which meta-analytic treatment effect estimates are transported. While theoretically attractive, these approaches currently have some practical limitations. Most assume that all studies in the meta-analysis have individual participant data (IPD), which is rare in practice because most trials share only aggregate data. We propose a method to perform CIMA using a combination of aggregate data and IPD. This method borrows information from studies with IPD to augment the aggregate data and create aggregate-matched synthetic IPD (AMSIPD), which can be used readily in the existing CIMA framework. By allowing use of both aggregate data and IPD, the method opens CIMA to more applications and can avoid biases arising from using only studies with IPD. We present a case study and simulations showing the AMSIPD approach is promising and merits further investigation as an advancement of CIMA.

The metabolic signature of blood lipids: a causal inference study using twins.

Dyslipidemia is one of the cardiometabolic risk factors that influences mortality globally. Unraveling the causality between blood lipids and metabolites and the complex networks connecting lipids, metabolites, and other cardiometabolic traits can help to more accurately reflect the body's metabolic disorders and even cardiometabolic diseases. We conducted targeted metabolomics of 248 metabolites in 437 twins from the Chinese National Twin Registry. Inference about Causation through Examination of FAmiliaL CONfounding (ICE FALCON) analysis was used for causal inference between metabolites and lipid parameters. Bidirectional mediation analysis was performed to explore the linkages between blood lipids, metabolites, and other seven cardiometabolic traits. We identified 44, 1, and 31 metabolites associated with triglyceride (TG), total cholesterol (TC), and high-density lipoprotein-cholesterol (HDL-C), most of which were gut microbiota-derived metabolites. There were 9, 1, and 14 metabolites that showed novel associations with TG, TC, and HDL-C, respectively. ICE FALCON analysis found that TG and HDL-C may have a predicted causal effect on 23 and six metabolites, respectively, and one metabolite may have a predicted causal effect on TG. Mediation analysis discovered 14 linkages connecting blood lipids, metabolites, and other cardiometabolic traits. Our study highlights the significance of gut microbiota-derived metabolites in lipid metabolism. Most of the identified cross-sectional associations may be due to the lipids having a predicted causal effect on metabolites, but not vice versa, nor are they due to family confounding. These findings shed new light on lipid metabolism and personalized management of cardiometabolic diseases.

GENIUS-MAWII: for robust Mendelian randomization with many weak invalid instruments.

Mendelian randomization (MR) addresses causal questions using genetic variants as instrumental variables. We propose a new MR method, G-Estimation under No Interaction with Unmeasured Selection (GENIUS)-MAny Weak Invalid IV, which simultaneously addresses the 2 salient challenges in MR: many weak instruments and widespread horizontal pleiotropy. Similar to MR-GENIUS, we use heteroscedasticity of the exposure to identify the treatment effect. We derive influence functions of the treatment effect, and then we construct a continuous updating estimator and establish its asymptotic properties under a many weak invalid instruments asymptotic regime by developing novel semiparametric theory. We also provide a measure of weak identification, an overidentification test, and a graphical diagnostic tool.

Association and causality between diabetes and activin A: a two-sample Mendelian randomization study.

Activin A, a cytokine belonging to the transforming growth factor-beta (TGF-ß) superfamily, mediates a multifunctional signaling pathway that is essential for embryonic development, cell differentiation, metabolic regulation, and physiological equilibrium. Biomedical research using diabetes-based model organisms and cellular cultures reports evidence of different activin A levels between diabetic and control groups. Activin A is highly conserved across species and universally expressed among disparate tissues. A systematic review of published literatures on human populations reveals association of plasma activin A levels with diabetic patients in some (7) but not in others (5) of the studies. With summarized data from publicly available genome-wide association studies (GWASs), a two-sample Mendelian randomization (TSMR) analysis is conducted on the causality between the exposure and the outcome. Wald ratio estimates from single instruments are predominantly non-significant. In contrast to positive controls between diabetes and plasma cholesterol levels, inverse-variance-weighted (IVW), Egger, weighted median, and weighted mode MR methods all lead to no observed causal link between diabetes (type 1 and type 2) and plasma activin A levels. Unavailability of strong instruments prevents the reversal MR analysis of activin A on diabetes. In summary, further research is needed to confirm or deny the potential association between diabetes and plasma activin A, and to elucidate the temporal incidence of these traits in human populations. At this stage, no causality has been found between diabetes and plasma activin A based on TSMR analysis.

How to develop causal directed acyclic graphs for observational health research: a scoping review.

Causal directed acyclic graphs (DAGs) serve as intuitive tools to visually represent causal relationships between variables. While they find widespread use in guiding study design, data collection and statistical analysis, their adoption remains relatively rare in the domain of psychology. In this paper we describe the relevance of DAGs for health psychology, review guidelines for developing causal DAGs, and offer recommendations for their development. A scoping review searching for papers and resources describing guidelines for DAG development was conducted. Information extracted from the eligible papers and resources (n = 11) was categorised, and results were used to formulate recommendations. Most records focused on DAG development for data analysis, with similar steps outlined. However, we found notable variations on how to implement confounding variables (i.e., sequential inclusion versus exclusion). Also, how domain knowledge should be integrated in the development process was scarcely addressed. Only one paper described how to perform a literature search for DAG development. Key recommendations for causal DAG development are provided and discussed using an illustrative example.

A Bayesian nonparametric approach for causal mediation with a post-treatment confounder.

We propose a new Bayesian nonparametric method for estimating the causal effects of mediation in the presence of a post-treatment confounder. The methodology is motivated by the Rural Lifestyle Intervention Treatment Effectiveness Trial (Rural LITE) for which there is interest in estimating causal mediation effects but is complicated by the presence of a post-treatment confounder. We specify an enriched Dirichlet process mixture (EDPM) to model the joint distribution of the observed data (outcome, mediator, post-treatment confounder, treatment, and baseline confounders). For identifiability, we use the extended version of the standard sequential ignorability (SI) as introduced in Hong et al. along with a Gaussian copula model assumption. The observed data model and causal identification assumptions enable us to estimate and identify the causal effects of mediation, that is, the natural direct effects (NDE) and natural indirect effects (NIE). Our method enables easy computation of NIE and NDE for a subset of confounding variables and addresses missing data through data augmentation under the assumption of ignorable missingness. We conduct simulation studies to assess the performance of our proposed method. Furthermore, we apply this approach to evaluate the causal mediation effect in the Rural LITE trial, finding that there was not strong evidence for the potential mediator.

The Effect of Long-Term Particulate Matter Exposure on Respiratory Mortality: Cohort Study in China.

Background: Particulate matter (PM), which affects respiratory health, has been well documented; however, substantial evidence from large cohorts is still limited, particularly in highly polluted countries and for PM1. Objective: Our objective was to examine the potential causal links between long-term exposure to PMs (PM2.5, PM10, and more importantly, PM1) and respiratory mortality. Methods: A total of 580,757 participants from the Guangzhou area, China, were recruited from 2009 to 2015 and followed up through 2020. The annual average concentrations of PMs at a 1-km spatial resolution around the residential addresses were estimated using validated spatiotemporal models. The marginal structural Cox model was used to estimate the associations of PM exposure with respiratory mortality, accounting for time-varying PM exposure. Results were stratified by demographics and lifestyle behaviors factors. Results: Among the participants, the mean age was 48.33 (SD 17.55) years, and 275,676 (47.47%) of them were men. During the follow-up period, 7260 deaths occurred due to respiratory diseases. The annual average concentrations of PM1, PM2.5, and PM10 showed a declining trend during the follow-up period. After adjusting for confounders, a 6.6% (95% CI 5.6%-7.6%), 4.2% (95% CI 3.6%-4.7%), and 4.0% (95% CI 3.6%-4.5%) increase in the risk of respiratory mortality was observed following each 1-µg/m3 increase in concentrations of PM1, PM2.5, and PM10, respectively. In addition, older participants, nonsmokers, participants with higher exercise frequency, and those exposed to a lower normalized difference vegetation index tended to be more susceptible to the effects of PMs. Furthermore, participants in the low-exposure group tended to be at a 7.6% and 2.7% greater risk of respiratory mortality following PM1 and PM10 exposure, respectively, compared to the entire cohort. Conclusions: This cohort study provides causal clues of the respiratory impact of long-term ambient PM exposure, indicating that PM reduction efforts may continuously benefit the population's respiratory health.

Integrating lymph node ratio into personalized radiotherapy for oral cavity squamous cell carcinoma.

PURPOSE: The use of postoperative radiotherapy (PORT) in patients with oral squamous cell carcinoma (OCSCC) lacks clear boundaries due to the non-negligible toxicity accompanying its remarkable cancer-killing effect. This study aims at validating the ability of deep learning models to develop individualized PORT recommendations for patients with OCSCC and quantifying the impact of patient characteristics on treatment selection. METHODS: Participants were categorized into two groups based on alignment between model-recommended and actual treatment regimens, with their overall survival compared. Inverse probability treatment weighting was used to reduce bias, and a mixed-effects multivariate linear regression illustrated how baseline characteristics influenced PORT selection. RESULTS: 4990 patients with OCSCC met the inclusion criteria. Deep Survival regression with Mixture Effects (DSME) demonstrated the best performance among all the models and National Comprehensive Cancer Network guidelines. The efficacy of PORT is enhanced as the lymph node ratio (LNR) increases. Similar enhancements in efficacy are observed in patients with advanced age, large tumors, multiple positive lymph nodes, tongue involvement, and stage IVA. Early-stage (stage 0-II) OCSCC may safely omit PORT. CONCLUSIONS: This is the first study to incorporate LNR as a tumor character to make personalized recommendations for patients. DSME can effectively identify potential beneficiaries of PORT and provide quantifiable survival benefits.

On the compatibility of established methods with emerging artificial intelligence and machine learning methods for disaster risk analysis.

There is growing interest in leveraging advanced analytics, including artificial intelligence (AI) and machine learning (ML), for disaster risk analysis (RA) applications. These emerging methods offer unprecedented abilities to assess risk in settings where threats can emerge and transform quickly by relying on "learning" through datasets. There is a need to understand these emerging methods in comparison to the more established set of risk assessment methods commonly used in practice. These existing methods are generally accepted by the risk community and are grounded in use across various risk application areas. The next frontier in RA with emerging methods is to develop insights for evaluating the compatibility of those risk methods with more recent advancements in AI/ML, particularly with consideration of usefulness, trust, explainability, and other factors. This article leverages inputs from RA and AI experts to investigate the compatibility of various risk assessment methods, including both established methods and an example of a commonly used AI-based method for disaster RA applications. This article utilizes empirical evidence from expert perspectives to support key insights on those methods and the compatibility of those methods. This article will be of interest to researchers and practitioners in risk-analytics disciplines who leverage AI/ML methods.

Association between iron status, preeclampsia and gestational hypertension: A bidirectional two-sample Mendelian randomization study.

BACKGROUND: Several recent observational studies have reported that iron overload during pregnancy is associated with preeclampsia (PE) and gestational hypertension (GH). However, the causal association between iron status, PE, and GH is still not clear. METHODS: We performed a two-sample Mendelian randomization (MR) study using the genome-wide association study (GWAS) summary statistics of iron status, included serum iron, ferritin, total iron-binding capacity (TIBC), and transferrin saturation (TSAT) from the largest available GWAS meta-analysis, and the summary statistics of PE and GH were obtained from the FinnGen consortium. Fixed-effect inverse variance weighted (IVW), random-effect IVW, maximum likelihood (ML), MR-Egger regression, weighted median, and MR-PRESSO methods were used. RESULTS: A total of 21, 58, 28, and 22 SNPs were used as IVs for serum iron, ferritin, TIBC, and TSAT, respectively. The F-statistics of IVs ranged from 95.23 to 421.36. The results of the fixed effects IVW method suggested that for per SD unit increase in serum iron, the risk of PE increases by 24 % (OR = 1.24, 95 % CI: 1.03-1.50, P = 0.02). No significant heterogeneity or horizontal pleiotropy was found. The association between ferritin, TIBC, TSAT and PE were statistically insignificant (P>0.05). Furthermore, the results of each MR methods do not support a causal association between iron status and GH, nor a reverse causal association between PE and GH and iron status. CONCLUSION: This two-sample MR study provides evidence supporting a causal association between serum iron level and PE.

Exploring and Accounting for Genetically Driven Effect Heterogeneity in Mendelian Randomization.

Mendelian randomization (MR) is a framework to estimate the causal effect of a modifiable health exposure, drug target or pharmaceutical intervention on a downstream outcome by using genetic variants as instrumental variables. A crucial assumption allowing estimation of the average causal effect in MR, termed homogeneity, is that the causal effect does not vary across levels of any instrument used in the analysis. In contrast, the science of pharmacogenetics seeks to actively uncover and exploit genetically driven effect heterogeneity for the purposes of precision medicine. In this study, we consider a recently proposed method for performing pharmacogenetic analysis on observational data-the Triangulation WIthin a STudy (TWIST) framework-and explore how it can be combined with traditional MR approaches to properly characterise average causal effects and genetically driven effect heterogeneity. We propose two new methods which not only estimate the genetically driven effect heterogeneity but also enable the estimation of a causal effect in the genetic group with and without the risk allele separately. Both methods utilise homogeneity-respecting and homogeneity-violating genetic variants and rely on a different set of assumptions. Using data from the ALSPAC study, we apply our new methods to estimate the causal effect of smoking before and during pregnancy on offspring birth weight in mothers whose genetics mean they find it (relatively) easier or harder to quit smoking.