Use of directed acyclic graphs (DAGs) to identify confounders in applied health research: review and recommendations

Key predictors of psychological distress and wellbeing in Australian frontline healthcare workers during COVID-19 (Omicron wave)

Introduction

The COVID-19 pandemic has led to significant challenges for frontline healthcare workers' (FHW), raising many mental health and wellbeing concerns for this cohort. To facilitate identification of risk and protective factors to inform treatment and interventions, this study investigated key predictors of psychological distress and subjective wellbeing in FHWs.

Methods

During the Omicron wave of the COVID-19 pandemic (January 2022), Victorian (Australia) doctors, nurses, allied health and non-medical staff from Emergency Departments, Intensive Care units, Aged Care, Hospital In The Home, and COVID Wards completed a cross-sectional survey consisting of the Kessler 6 item (Psychological Distress), Personal Wellbeing Index (Subjective Wellbeing), Coronavirus Health Impact Survey tool (COVID-19 related factors) and occupational factors. Multivariable linear regressions were used to evaluate unadjusted and adjusted associations. Relative weight analysis was used to compare and identify key predictors.

Results

Out of 167 participants, 18.1% screened positive for a probable mental illness and a further 15.3% screened positive for low wellbeing. Key risk factors for greater psychological distress included COVID infection worries, relationship stress and younger age. For both psychological distress and lower wellbeing, health status and supervisor support were key protective factors, while infection risks were key risk factors. Only positive changes in relationship quality was protective of lower wellbeing.

Conclusion

This study highlights the significance of social determinants and individual level factors alongside work related factors, in influencing FHWs' mental health and wellbeing during public health crises, such as the COVID-19 pandemic. Findings suggest that future interventions and supports should take a more holistic approach that considers work, social and individual level factors when supporting FHWs' mental health and wellbeing.

Association of heavy metal complex exposure and neurobehavioral function of children

Background

Exposure to heavy metals is a public health concern worldwide. Previous studies on the association between heavy metal exposure and neurobehavioral functions in children have focused on single exposures and clinical manifestations. However, the present study evaluated the effects of heavy metal complex exposure on subclinical neurobehavioral function using a Korean Computerized Neurobehavior Test (KCNT).

Methods

Urinary mercury, lead, cadmium analyses as well as symbol digit substitution (SDS) and choice reaction time (CRT) tests of the KCNT were conducted in children aged between 10 and 12 years. Reaction time and urinary heavy metal levels were analyzed using partial correlation, linear regression, Bayesian kernel machine regression (BKMR), the weighted quantile sum (WQS) regression and quantile G-computation analysis.

Results

Participants of 203 SDS tests and 198 CRT tests were analyzed, excluding poor cooperation and inappropriate urine sample. Partial correlation analysis revealed no association between neurobehavioral function and exposure to individual heavy metals. The result of multiple linear regression shows significant positive association between urinary lead, mercury, and CRT. BMKR, WQS regression and quantile G-computation analysis showed a statistically significant positive association between complex urinary heavy metal concentrations, especially lead and mercury, and reaction time.

Conclusions

Assuming complex exposures, urinary heavy metal concentrations showed a statistically significant positive association with CRT. These results suggest that heavy metal complex exposure during childhood should be evaluated and managed strictly.

High Serum Folate Concentration, Vitamin B12 Deficiency, and Anthropometric Nutritional Status in Brazilian Children Aged 6-59 Months: Data from the Brazilian National Survey of Child Nutrition

BACKGROUND

Vitamin B12 and folate are key nutrients that help children reach their full potential in growth and development; however, little is known about the status of these vitamins in Brazilian children.

OBJECTIVES

To describe the serum concentrations of vitamin B12 and folate, to investigate the association between high folate concentration (HFC) and vitamin B12 deficiency, and to evaluate the association between vitamin B12 and stunting/underweight in Brazilian children aged 6-59 mo.

METHODS

Data from 7417 children aged 6-59 mo collected during the Brazilian National Survey on Child Nutrition were used. Serum concentrations of vitamin B12 of <150 pmol/L and folate of <10 nmol/L were classified as deficient, and folate concentrations of >45.3 nmol/L were classified as HFC. Children with length/height-for-age z-score of less than -2 were considered stunted, and those with weight-for-age z-score of less than -2 were underweight. Logistic regression models were performed.

RESULTS

In Brazil, 14.2% (95% CI: 12.2, 16.1) of children aged 6-59 mo had vitamin B12 deficiency, 1.1% (95% CI: 0.5, 1.6) had folate deficiency, and 36.9% (95% CI: 33.4, 40.3) had HFC. Vitamin B12 deficiency was higher in children from the northern region of Brazil (28.5%), between 6 and 24 mo (25.3%), whose mothers had lower formal education (0-7 y; 18.7%). Children with HFC had 62% lower odds (OR: 0.38; 95% CI: 0.27, 0.54) of vitamin B12 deficiency than those with normal/deficient folate. Children with vitamin B12 deficiency and normal/deficient folate had higher odds of stunting (OR: 1.58; 95% CI: 1.02, 2.43) than children without vitamin B12 deficiency and normal/deficient folate.

CONCLUSIONS

Vitamin B12 deficiency is a public health problem among Brazilian children aged <2 y with vulnerable socioeconomic status. HFC was inversely associated with vitamin B12 deficiency, and lower odds of stunting were observed in children with HFC and vitamin B12 deficiency than in those with vitamin B12 deficiency and normal/deficient folate.

The dawn of directed acyclic graphs in primary care research and education

Dramatical increase in articles mentioning "directed acyclic graph."

The effect of internal displacement due to armed conflict on tuberculosis treatment outcomes in northwest Syria, 2019-2020

Background

Northwest Syria accounts for over 4 million people of whom more than half are internally displaced persons (IDPs). More than 1 million IDPs reside in camps and many more live in settings which are overcrowded and poorly ventilated. Suboptimal social living conditions leave these populations susceptible to tuberculosis (TB) morbidity and mortality. This study aimed to assess the effect of internal displacement due to armed conflict on the risk of unsuccessful treatment outcomes among TB patients in northwest Syria.

Methods

All patients registered to start TB treatment at three centres in northwest Syria between the 1st June 2019 to the 31st December 2020 were included. Unsuccessful TB treatment outcome was defined as a composite outcome combining the WHO TB treatment outcomes of treatment failure, loss to follow-up, and death. We assessed the association between internal displacement and unsuccessful TB treatment outcomes using multivariable logistic regression. We also explored the risk of an unsuccessful treatment outcome by internal displacement setting (camp, village or city).

Results

Of the total 737 registered patients included in the analysis, 400 (54%) were documented as internally displaced. The median age of IDPs was 30 (IQR: 21.5-48) and the median age of residents was 34 (IQR:22-50). A significantly higher percentage of those who were IDPs had an unsuccessful treatment outcome compared to residents (40% vs 18%, p<0.001). After adjustment for confounders, the relative risk of having unsuccessful TB treatment was two-fold higher in internally displaced TB patients compared to residents (95% CI: 1.5-2.6). IDPs living in villages had a 30% lower risk of an unsuccessful treatment outcome compared to IDPs living in camps (RR 95% CI: 0.50-0.91), and IDPs living in cities had a 13% lower risk of an unsuccessful treatment outcome (RR 95% CI: 0.57-1.18).

Conclusion

This is one of few studies which aims to quantify the effect of internal displacement on TB treatment outcomes during times of conflict. Our findings starkly highlight how social determinants contribute to poor TB outcomes and act as a starting point for much needed research on how best to manage TB in humanitarian crisis settings.

Healthy Bus Drivers, Sustainable Public Transport: A Three-Time Repeated Cross-Sectional Study in Switzerland

Objectives: To study the change in the prevalence of bus drivers' health outcomes between 2010 and 2022 and their relationship with working conditions. Method: Unionized bus drivers completed a self-administered questionnaire in 2010, 2018, and 2022 on 13 health outcomes, sick leaves, and accidents and working conditions and their change during SARS-CoV-2 crisis. For outcomes which prevalence increased since 2010, we performed logistic regression models adjusted for covariates. Results: The study sample included 772 participants in 2010, 393 in 2018, and 916 in 2022. The most prevalent health problem (≥50%) was shoulder or neck muscle pain. The most tedious working conditions were working days over 10 h. Shoulder or neck pain, sleep disorders, sick leaves, and accidents increased since 2010 and were associated with working conditions, and co-morbidity. The SARS-CoV-2 pandemic had additional negative consequences. Conclusion: Most bus drivers' working and health conditions worsened in the last 12 years. Given the study design, the results deserve a cautious interpretation and generalization. Cohort studies should confirm these results and inform interventions targeting the most tedious and harmful working conditions.

Causal feature selection using a knowledge graph combining structured knowledge from the biomedical literature and ontologies: A use case studying depression as a risk factor for Alzheimer's disease

BACKGROUND

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.

METHODS

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.

RESULTS

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.

CONCLUSION

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.

Risk Factors for Multisystem Inflammatory Syndrome in Children: A Case-control Investigation

BACKGROUND

In a 2020 pilot case-control study using medical records, we reported that non-Hispanic Black children were more likely to develop multisystem inflammatory syndrome in children (MIS-C) after adjustment for sociodemographic factors and underlying medical conditions. Using structured interviews, we investigated patient, household, and community factors underlying MIS-C likelihood.

METHODS

MIS-C case patients hospitalized in 2021 across 14 US pediatric hospitals were matched by age and site to outpatient controls testing positive for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) within 3 months of the admission date. Caregiver interviews queried race/ethnicity, medical history, and household and potential community exposures 1 month before MIS-C hospitalization (case-patients) or after SARS-CoV-2 infection (controls). We calculated adjusted odds ratios (aOR) using mixed-effects multivariable logistic regression.

RESULTS

Among 275 case patients and 496 controls, race/ethnicity, social vulnerability and patient or family history of autoimmune/rheumatologic disease were not associated with MIS-C. In previously healthy children, MIS-C was associated with a history of hospitalization for an infection [aOR: 4.8; 95% confidence interval (CI): 2.1-11.0]. Household crowding (aOR: 1.7; 95% CI: 1.2-2.6), large event attendance (aOR: 1.7; 95% CI: 1.3-2.1), school attendance with limited masking (aOR: 2.6; 95% CI: 1.1-6.6), public transit use (aOR: 1.8; 95% CI: 1.4-2.4) and co-resident testing positive for SARS-CoV-2 (aOR: 2.2; 95% CI: 1.3-3.7) were associated with increased MIS-C likelihood, with risk increasing with the number of these factors.

CONCLUSIONS

From caregiver interviews, we clarify household and community exposures associated with MIS-C; however, we did not confirm prior associations between sociodemographic factors and MIS-C.

Rural, Large Town, and Urban Differences in Optimal Subspecialty Follow-up and Survivorship Care Plan Documentation among Childhood Cancer Survivors

BACKGROUND

Children with cancer from rural and nonurban areas face unique challenges. Health equity for this population requires attention to geographic disparities in optimal survivorship-focused care.

METHODS

The Oklahoma Childhood Cancer Survivor Cohort was based on all patients reported to the institutional cancer registry and ≤ 18 years old at diagnosis between January 1, 2005, and September 24, 2014. Suboptimal follow-up was defined as no completed oncology-related clinic visit five to 7 years after their initial diagnosis (survivors were 7-25 years old at end of the follow-up period). The primary predictor of interest was rurality.

RESULTS

Ninety-four (21%) of the 449 eligible survivors received suboptimal follow-up. There were significant differences (P = 0.01) as 36% of survivors from large towns (n = 28/78) compared with 21% (n = 20/95) and 17% (n = 46/276) of survivors from small town/isolated rural and urban areas received suboptimal follow-up, respectively. Forty-five percent of adolescents at diagnosis were not seen in the clinic compared with 17% of non-adolescents (P < 0.01). An adjusted risk ratio of 2.2 (95% confidence interval, 1.5, 3.2) was observed for suboptimal follow-up among survivors from large towns, compared with survivors from urban areas. Seventy-three percent of survivors (n = 271/369) had a documented survivorship care plan with similar trends by rurality.

CONCLUSIONS

Survivors from large towns and those who were adolescents at the time of diagnosis were more likely to receive suboptimal follow-up care compared with survivors from urban areas and those diagnosed younger than thirteen.

IMPACT

Observed geographic disparities in survivorship care will inform interventions to promote equitable care for survivors from nonurban areas.

A Cross-sectional Study of Perceived Stress and Racial Discrimination Among a National Sample of Young Men Who Have Sex With Men

ABSTRACT

Research regarding the impact of racism on stress among young men who have sex with men (YMSM) is sparse. Secondary data were assessed from a 2018-2020 national mHealth prevention trial for YMSM aged 13-18 years (N = 542). Linear regression models examined associations between perceived stress and interpersonal and vicarious racism, adjusting for covariates. Stratified models by race/ethnicity were included. A subanalysis (n = 288) examined associations between nine interpersonal racial discriminatory events and perceived stress. Over 50% of participants experienced racial discrimination. In the multivariable models, exposure to interpersonal (β = 1.43, p-value: .038) and vicarious (β = 1.77, p-value: .008) racism was associated with perceived stress because there were four interpersonal racial discriminatory events. Stratified analysis by race/ethnicity found significant associations between interpersonal and vicarious racism and perceived stress among some racial/ethnic groups. Racial discrimination was common among YMSM, making them susceptible to the possible effects of vicarious and interpersonal racism on stress.

Monitoring multidimensional aspects of quality of life after cancer immunotherapy: protocol for the international multicentre, observational QUALITOP cohort study

INTRODUCTION

Immunotherapies, such as immune checkpoint inhibitors and chimeric antigen receptor T-cell therapy, have significantly improved the clinical outcomes of various malignancies. However, they also cause immune-related adverse events (irAEs) that can be challenging to predict, prevent and treat. Although they likely interact with health-related quality of life (HRQoL), most existing evidence on this topic has come from clinical trials with eligibility criteria that may not accurately reflect real-world settings. The QUALITOP project will study HRQoL in relation to irAEs and its determinants in a real-world study of patients treated with immunotherapy.

METHODS AND ANALYSIS

This international, observational, multicentre study takes place in France, the Netherlands, Portugal and Spain. We aim to include about 1800 adult patients with cancer treated with immunotherapy in a specifically recruited prospective cohort, and to additionally obtain data from historical real-world databases (ie, databiobanks) and medical administrative registries (ie, national cancer registries) in which relevant data regarding other adult patients with cancer treated with immunotherapy has already been stored. In the prospective cohort, clinical health status, HRQoL and psychosocial well-being will be monitored until 18 months after treatment initiation through questionnaires (at baseline and 3, 6, 12 and 18 months thereafter), and by data extraction from electronic patient files. Using advanced statistical methods, including causal inference methods, artificial intelligence algorithms and simulation modelling, we will use data from the QUALITOP cohort to improve the understanding of the complex relationships among treatment regimens, patient characteristics, irAEs and HRQoL.

ETHICS AND DISSEMINATION

All aspects of the QUALITOP project will be conducted in accordance with the Declaration of Helsinki and with ethical approval from a suitable local ethics committee, and all patients will provide signed informed consent. In addition to standard dissemination efforts in the scientific literature, the data and outcomes will contribute to a smart digital platform and medical data lake. These will (1) help increase knowledge about the impact of immunotherapy, (2) facilitate improved interactions between patients, clinicians and the general population and (3) contribute to personalised medicine.

TRIAL REGISTRATION NUMBER

NCT05626764.

Association of development quotient with nutritional status of vitamins B6, B12, and folate in 6-59-month-old children: Results from the Brazilian National Survey on Child Nutrition (ENANI-2019)

BACKGROUND

Vitamins B6, B12, and folate are essential for the formation and maintenance of the human brain, but studies evaluating these vitamins with early childhood development (ECD) in children under five are limited and controversial.

OBJECTIVE

To evaluate the association between vitamins B6, B12, and folate concentrations/status and ECD.

METHODS

Data regarding 6,520 children aged 6-59 months (from the Brazilian National Survey on Child Nutrition [ENANI-2019]) were analyzed. ECD was assessed using the Survey of Well-being of Young Children's milestones questionnaire. Vitamin B6 concentration (nmol/L) was classified according to the tertile of the distribution and with the cutoff <20 nmol/L. Folate concentrations >45.3 nmol/L were classified as high, and vitamin B12 <150 pmol/L as deficient. The graded response model was used to estimate developmental age, and the developmental quotient (DQ) was calculated as the developmental age divided by chronological age. Multiple linear regression models were adjusted for confounders.

RESULTS

The DQ mean (95% confidence interval) for Brazilian children was 0.99 (0.97-1.01). Children aged 6-23 months (1.13 [1.10-1.16]) had a higher DQ mean than those aged 24-35 (0.99 [0.95-1.03]) and 36-59 months (0.89 [0.86-0.92]). Child age was inversely associated with DQ (β=-0.007; p<0.001). An interaction between child age and vitamin B12 deficiency in the DQ (β=-0.005; p<0.001) indicated that, in children aged 36-59 months, the DQ was markedly lower in children with B12 deficiency than in those without B12 deficiency. Vitamin B6 concentrations were directly associated with the DQ (β=0.0004; p=0.031) among children aged 24-59 months in the adjusted model. No association was observed between folate status and DQ.

CONCLUSIONS

In Brazil, the DQ is lower among older children, and those with vitamin B12 deficiency. Vitamin B6 status was directly associated with the DQ in children aged 24-59 months.

Spillover Effects of Prenatal Care Coordination on Older Siblings Beyond the Mother-Infant Dyad

BACKGROUND

Pregnancy care coordination increases preventive care receipt for mothers and infants. Whether such services affect other family members' health care is unknown.

OBJECTIVE

To estimate the spillover effect of maternal exposure to Wisconsin Medicaid's Prenatal Care Coordination (PNCC) program during pregnancy with a younger sibling on the preventive care receipt for an older child.

RESEARCH DESIGN

Gain-score regressions-a sibling fixed effects strategy-estimated spillover effects while controlling for unobserved family-level confounders.

SUBJECTS

Data came from a longitudinal cohort of linked Wisconsin birth records and Medicaid claims. We sampled 21,332 sibling pairs (one older; one younger) who were born during 2008-2015, who were <4 years apart in age, and whose births were Medicaid-covered. In all, 4773 (22.4%) mothers received PNCC during pregnancy with the younger sibling.

MEASURES

The exposure was maternal PNCC receipt during pregnancy with the younger sibling (none; any). The outcome was the older sibling's number of preventive care visits or preventive care services in the younger sibling's first year of life.

RESULTS

Overall, maternal exposure to PNCC during pregnancy with the younger sibling did not affect older siblings' preventive care. However, among siblings who were 3 to <4 years apart in age, there was a positive spillover on the older sibling's receipt of care by 0.26 visits (95% CI: 0.11, 0.40 visits) and by 0.34 services (95% CI: 0.12, 0.55 services).

CONCLUSION

PNCC may only have spillover effects on siblings' preventive care in selected subpopulations but not in the broader population of Wisconsin families.

Maternal depression or anxiety during pregnancy and offspring type 1 diabetes: a population-based family-design cohort study

INTRODUCTION

To investigate the association between maternal depression/anxiety during pregnancy and offspring type 1 diabetes, to assess the specific importance of exposure during pregnancy by comparing across different exposure periods before and/or after pregnancy, and to explore potential unmeasured familial confounding.

RESEARCH DESIGN AND METHODS

This was a population-based cohort including 1 807 809 offspring born in Sweden 2002-2019. From national registers, data were available on diagnosis or medication prescription for depression/anxiety in and around pregnancy, as well as incident cases of type 1 diabetes defined through diagnosis or insulin treatment. Associations were examined using flexible parametric and Cox regression models. Familial confounding was explored using paternal exposure as a negative control and by comparing offspring exposed to maternal depression/anxiety with their unexposed siblings.

RESULTS

For exposure during pregnancy, maternal depression/anxiety was associated with an increased risk of offspring type 1 diabetes onset after, but not before, 8 years of age (adjusted HR (aHR) 1.21 (95% CI 1.03 to 1.42]). Exposure occurring only during pregnancy was similarly associated to type 1 diabetes (aHR 1.24 (0.96 to 1.60)), whereas exposure occurring only before pregnancy was not (aHR 0.91 (0.64 to 1.30)). Associations were close to the null for paternal depression/anxiety (aHR 0.95 (0.72 to 1.25)), and point estimates were above 1 in sibling comparisons, although with wide CIs (aHR 1.36 (0.82 to 2.26)).

CONCLUSIONS

Maternal depression/anxiety specifically during pregnancy seems to be associated with offspring type 1 diabetes. Paternal negative control and sibling comparisons indicate that the results cannot entirely be explained by familial confounding.

Association of polypharmacy and potential drug-drug interactions with adverse treatment outcomes in older adults with advanced cancer

BACKGROUND

Polypharmacy is common in older adults who are starting cancer treatment and is associated with an increased risk of potentially inappropriate medications (PIMs) and potential drug-drug interactions (PDIs). The authors evaluated the association of medication measures with adverse outcomes in older adults with advanced cancer who were receiving systemic therapy.

METHODS

This secondary analysis from GAP 70+ Trial (ClinicalTrials.gov identifier NCT02054741; principal investigator, Supriya G. Mohile) enrolled patients aged 70 years and older with advanced cancer who planned to start a new treatment regimen (n = 718). Polypharmacy was assessed before the initiation of treatment and was defined as the concurrent use of eight or more medications. PIMs were categorized using 2019 Beers Criteria and the Screening Tool of Older Persons' Prescriptions. PDIs were evaluated using Lexi-Interact Online. Study outcomes were assessed within 3 months of treatment and included: (1) the number of grade ≥2 and ≥3 toxicities according to the National Cancer Institute Common Toxicity Criteria, (2) treatment-related unplanned hospitalization, and (3) early treatment discontinuation. Multivariable regression models examined the association of medication measures with outcomes.

RESULTS

The mean patient age was 77 years, and 57% had lung or gastrointestinal cancers. The median number of medications was five (range, 0-24 medications), 28% of patients received eight or more medications, 67% received one or more PIM, and 25% had one or more major PDI. The mean number of grade ≥2 toxicities in patients with polypharmacy was 9.8 versus 7.7 in those without polypharmacy (adjusted β = 1.87; standard error, 0.71; p <.01). The mean number of grade ≥3 toxicities in patients with polypharmacy was 2.9 versus 2.2 in patients without polypharmacy (adjusted β = 0.59; standard error, 0.29; p = .04). Patients with who had one or more major PDI had 59% higher odds of early treatment discontinuation (odds ratio, 1.59; 95% confidence interval, 1.03-2.46; p = .03).

CONCLUSIONS

In a cohort of older adults with advanced cancer, polypharmacy and PDIs were associated with an increased risk of adverse treatment outcomes. Providing meaningful screening and interventional tools to optimize medication use may improve treatment-related outcomes in these patients.

Long-term Exposure to Multiple Ambient Air Pollutants and Association With Incident Depression and Anxiety

Importance

Air pollution is increasingly recognized as an important environmental risk factor for mental health. However, epidemiologic evidence on long-term exposure to low levels of air pollutants with incident depression and anxiety is still very limited.

Objectives

To investigate the association of long-term joint exposure to multiple air pollutants with incident depression and anxiety.

Design, Setting, and Participants

This prospective, population-based cohort study used data from the UK Biobank. The participants were recruited between March 13, 2006, and October 1, 2010, and included individuals who had never been diagnosed with depression or anxiety at baseline and had full information on exposure and covariates. Data were analyzed from May 1 to October 10, 2022.

Exposures

Annual mean air pollution concentrations of particulate matter (PM) with aerodynamic diameter of 2.5 μm or less (PM2.5) and PM with aerodynamic diameter between 2.5 μm and 10 μm (PM2.5-10). Nitrogen dioxide (NO2) and nitric oxide (NO) were estimated for each participant's residential address using the land use regression model, and joint exposure to air pollution reflected by air pollution score was calculated by principal components analysis.

Main Outcomes and Measures

Incidence of diagnosed depression (F32-F33) and anxiety (F40-F48) were ascertained with International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes.

Results

During a median (IQR) follow-up of 10.9 (10.1-11.6) years, among 389 185 participants (mean [SD] age, 56.7 [8.1] years, 205 855 female individuals [52.9%]), a total of 13 131 and 15 835 patients were diagnosed with depression and anxiety, respectively. The median (IQR) concentration of pollutants was as follows: PM2.5, 9.9 (9.3-10.6) μg/m3; PM2.5-10, 6.1 (5.8-6.6) μg/m3; NO2, 26.0 (21.3-31.1) μg/m3; and NO, 15.9 (11.6-20.6) μg/m3. Long-term estimated exposure to multiple air pollutants was associated with increased risk of depression and anxiety, and the exposure-response curves were nonlinear, with steeper slopes at lower concentrations and plateauing trends at higher exposure. The hazard ratios (HRs) and 95% CIs for depression and anxiety were 1.16 (95% CI, 1.09-1.23; P < .001) and 1.11 (95% CI, 1.05-1.17; P < .001) in the highest quartile compared with the lowest quartile of air pollution score, respectively. Similar trends were shown for PM2.5, NO2, and NO. Subgroup analysis showed the association between PM2.5 and anxiety tended to be higher in male individuals than in female individuals (quartile 4: male individuals, 1.18; 95% CI, 1.08-1.29; female individuals, 1.07; 95% CI, 1.00-1.14; P = .009).

Conclusions and Relevance

Study results suggest that estimates of long-term exposure to multiple air pollutants was associated with increased risk of depression and anxiety. The nonlinear associations may have important implications for policy making in air pollution control. Reductions in joint exposure to multiple air pollutants may alleviate the disease burden of depression and anxiety.

Inferring the impact of humanitarian responses on population mortality: methodological problems and proposals

Reducing excess population mortality caused by crises due to armed conflict and natural disasters is an existential aim of humanitarian assistance, but the extent to which these deaths are averted in different humanitarian responses is mostly unknown. This information gap arguably weakens governance and accountability. This paper considers methodological challenges involved in making inferences about humanitarian assistance's effect on excess mortality, and outlines proposed approaches. Three possible measurement questions, each of which contributes some inferential evidence, are presented: (1) whether mortality has remained within an acceptable range during the crisis (for which different direct estimation options are presented); (2) whether the humanitarian response is sufficiently appropriate and performant to avert excess mortality (a type of contribution analysis requiring in-depth audits of the design of humanitarian services and of their actual availability, coverage and quality); and (3) the actual extent to which humanitarian assistance has reduced excess deaths (potentially the most complex question to answer, requiring application of causal thinking and careful specification of the exposure, and for which either quasi-experimental statistical modelling approaches or a combination of verbal and social autopsy methods are proposed). The paper concludes by considering possible 'packages' of the above methods that could be implemented at different stages of a humanitarian response, and calls for investment in improved methods and actual measurement.

Modeling COVID-19 disease processes by remote elicitation of causal Bayesian networks from medical experts

BACKGROUND

COVID-19 is a new multi-organ disease causing considerable worldwide morbidity and mortality. While many recognized pathophysiological mechanisms are involved, their exact causal relationships remain opaque. Better understanding is needed for predicting their progression, targeting therapeutic approaches, and improving patient outcomes. While many mathematical causal models describe COVID-19 epidemiology, none have described its pathophysiology.

METHODS

In early 2020, we began developing such causal models. The SARS-CoV-2 virus's rapid and extensive spread made this particularly difficult: no large patient datasets were publicly available; the medical literature was flooded with sometimes conflicting pre-review reports; and clinicians in many countries had little time for academic consultations. We used Bayesian network (BN) models, which provide powerful calculation tools and directed acyclic graphs (DAGs) as comprehensible causal maps. Hence, they can incorporate both expert opinion and numerical data, and produce explainable, updatable results. To obtain the DAGs, we used extensive expert elicitation (exploiting Australia's exceptionally low COVID-19 burden) in structured online sessions. Groups of clinical and other specialists were enlisted to filter, interpret and discuss the literature and develop a current consensus. We encouraged inclusion of theoretically salient latent (unobservable) variables, likely mechanisms by extrapolation from other diseases, and documented supporting literature while noting controversies. Our method was iterative and incremental: systematically refining and validating the group output using one-on-one follow-up meetings with original and new experts. 35 experts contributed 126 hours face-to-face, and could review our products.

RESULTS

We present two key models, for the initial infection of the respiratory tract and the possible progression to complications, as causal DAGs and BNs with corresponding verbal descriptions, dictionaries and sources. These are the first published causal models of COVID-19 pathophysiology.

CONCLUSIONS

Our method demonstrates an improved procedure for developing BNs via expert elicitation, which other teams can implement to model emergent complex phenomena. Our results have three anticipated applications: (i) freely disseminating updatable expert knowledge; (ii) guiding design and analysis of observational and clinical studies; (iii) developing and validating automated tools for causal reasoning and decision support. We are developing such tools for the initial diagnosis, resource management, and prognosis of COVID-19, parameterized using the ISARIC and LEOSS databases.

Associations of residential green space with incident type 2 diabetes and the role of air pollution: A prospective analysis in UK Biobank

BACKGROUND

Green space is a newly modifiable environmental factor which would bring health benefits, and identifying potential pathways is important to better promote public health. Nowadays, limited evidence is available on residential green space and risk of incident type 2 diabetes (T2D). To evaluate the longitudinal association between residential green space and incident T2D, and further illustrate the role of air pollution, we conducted a prospective analysis in UK Biobank.

METHODS

Incident cases of T2D were ascertained through medical linkage of hospital admissions. Residential green space indicated by percentage of green space at 300 m buffer was estimated using land use data. Annual average air pollution was modelled using Land Use Regression model. Cox proportional hazard regression models were used to determine the association between the exposure and incident T2D. Effect modification of air pollution was assessed using log-likelihood tests. Casual mediation analysis including interaction of green space and air pollution was used to quantity pure natural indirect effect of air pollution.

RESULTS

Of 379,238 participants at baseline, 15564 incident T2D cases were identified with 12.4 years of follow-up. In main models, individuals who exposed to residential green space at 300 m buffer in high level (≥75 %) had 14.4 % (95 % CI: 8.0 %, 20.3 %) lower risk of incident T2D compared with those in low level (<25 %). This association was modified by NO₂, with green space indicating higher protective effect in low NO₂ level (P_interaction = 0.098). PM_2.5 had a high mediation effect of 37.0 % in the association.

CONCLUSION

Elevated residential green space level was associated with lower risk of incident T2D, and air pollution played an important role in this association. These findings would contribute to policy making and healthy city construction to take air pollution into consideration while planning green space, which would maximize public health benefits.

Bayesian network modelling to identify on-ramps to childhood obesity

BACKGROUND

When tackling complex public health challenges such as childhood obesity, interventions focused on immediate causes, such as poor diet and physical inactivity, have had limited success, largely because upstream root causes remain unresolved. A priority is to develop new modelling frameworks to infer the causal structure of complex chronic disease networks, allowing disease "on-ramps" to be identified and targeted.

METHODS

The system surrounding childhood obesity was modelled as a Bayesian network, using data from The Longitudinal Study of Australian Children. The existence and directions of the dependencies between factors represent possible causal pathways for childhood obesity and were encoded in directed acyclic graphs (DAGs). The posterior distribution of the DAGs was estimated using the Partition Markov chain Monte Carlo.

RESULTS

We have implemented structure learning for each dataset at a single time point. For each wave and cohort, socio-economic status was central to the DAGs, implying that socio-economic status drives the system regarding childhood obesity. Furthermore, the causal pathway socio-economic status and/or parental high school levels → parental body mass index (BMI) → child's BMI existed in over 99.99% of posterior DAG samples across all waves and cohorts. For children under the age of 8 years, the most influential proximate causal factors explaining child BMI were birth weight and parents' BMI. After age 8 years, free time activity became an important driver of obesity, while the upstream factors influencing free time activity for boys compared with girls were different.

CONCLUSIONS

Childhood obesity is largely a function of socio-economic status, which is manifest through numerous downstream factors. Parental high school levels entangle with socio-economic status, and hence, are on-ramp to childhood obesity. The strong and independent causal relationship between birth weight and childhood BMI suggests a biological link. Our study implies that interventions that improve the socio-economic status, including through increasing high school completion rates, may be effective in reducing childhood obesity prevalence.

Unravelling the potential of social prescribing in individual-level type 2 diabetes prevention: a mixed-methods realist evaluation

BACKGROUND

Social prescribing (SP) usually involves linking patients in primary care with services provided by the voluntary and community sector. Preliminary evidence suggests that SP may offer a means of connecting patients with community-based health promotion activities, potentially contributing to the prevention of long-term conditions, such as type 2 diabetes (T2D).

METHODS

Using mixed-methods realist evaluation, we explored the possible contribution of SP to individual-level prevention of T2D in a multi-ethnic, socio-economically deprived population in London, UK. We made comparisons with an existing prevention programme (NHS Diabetes Prevention Programme (NDPP)) where relevant and possible. Anonymised primary care electronic health record data of 447,360 people 18+ with an active GP registration between December 2016 and February 2022 were analysed using quantitative methods. Qualitative data (interviews with 11 primary care clinicians, 11 social prescribers, 13 community organisations and 8 SP users at high risk of T2D; 36 hours of ethnographic observations of SP and NDPP sessions; and relevant documents) were analysed thematically. Data were integrated using visual means and realist methods.

RESULTS

People at high risk of T2D were four times more likely to be referred into SP than the eligible general population (RR 4.31 (95% CI 4.17-4.46)), with adjustment for socio-demographic variables resulting in attenuation (RR 1.33 (95% CI 1.27-1.39)). More people at risk of T2D were referred to SP than to NDPP, which could be explained by the broad referral criteria for SP and highly supportive (proactive, welcoming) environments. Holistic and sustained SP allowed acknowledgement of patients' wider socio-economic constraints and provision of long-term personalised care. The fact that SP was embedded within the local community and primary care infrastructure facilitated the timely exchange of information and cross-referrals across providers, resulting in enhanced service responsiveness.

CONCLUSIONS

Our study suggests that SP may offer an opportunity for individual-level T2D prevention to shift away from standardised, targeted and short-term strategies to approaches that are increasingly personalised, inclusive and long-term. Primary care-based SP seems most ideally placed to deliver such approaches where practitioners, providers and commissioners work collectively to achieve holistic, accessible, sustained and integrated services.

The risk of an incident hospital contact with a musculoskeletal disorder in Danish occupational fishers: a register-based study

BACKGROUND

The prevalence of musculoskeletal disorders (MSDs) among occupational fishers is high, yet knowledge of the risk factors is scarce and inconsistent. The aim of this study was to investigate the risk from various work-related characteristics on incident hospital contact due to a musculoskeletal disorders and other pain disorders among Danish occupational fishers.

METHODS

This register-based study comprised data from the Danish Occupational Cohort with eXposure (DOC*X) for all persons registered as occupational fishers between 1994 and 2017. Time-to-event analysis with Cox regression model was used with age as the time scale.

RESULTS

Among the 15,739 fishers, 40% (n = 5,669 cases) had an incident hospital contact with an MSD during follow-up. Back disorders were the dominant complaint. Male fishers working less than 5 years or more than 15 years had higher risks of MSD (HR 2.40 (95% CI: 2.06, 2.80), HR: 2.04 (95% CI: 1.76, 2.35), respectively, than those working for over 20 years. Period effects confounded and reduced the risk from occupational seniority.

CONCLUSION

Fishers occupational seniority vary in risk of MSDs across working life. Results showed a nonlinear relationship between the highest risk for fishers working less than 5 years and the lowest risk working more than 20 years as occupational fisher. More years in the workforce, a captain education, and primarily working part time significantly reduced the risk of experiencing a first MSDs for men. Healthy worker effect was documented.

The association between meat consumption and muscle strength index in young adults: the mediating role of total protein intake and lean mass percentage

PURPOSE

The aim of this study was to analyse the associations between the consumption of different types of meat and the muscle strength index (MSI) and to examine whether this relationship is mediated by total protein intake (TPI) and lean mass percentage (LM%) in young adults.

METHODS

We conducted a cross-sectional study with first-year university students from Castilla-La Mancha, Spain. Different types of meat consumption (total, red, processed, and white and fish) were separately evaluated using a Food-Frequency Questionnaire. MSI was determined from the handgrip and standing long jump tests. ANCOVA models were used to test the mean differences in MSI by categories of meat consumption. Serial multiple mediation models were used to explore the mediating role of TPI and LM% in the relationship between meat consumption and MSI. All analyses were adjusted for age, sex, and socioeconomic level, identified through a directed acyclic graph. Additional analyses were performed with a small subsample including alcohol intake, tobacco smoking, physical activity, cardiorespiratory fitness, and total energy intake as covariates in the multiple mediation models.

RESULTS

A total of 230 students (mean age 21.1 ± 2.1 years, 66.5% women) were included in the analysis. Young adults with higher meat consumption (total, red, and white and fish) had higher MSI adjusted means than their peers with lower meat consumption (p < 0.05). These associations did not remain after controlling for TPI and LM%. In adjusted mediation analyses, a significant indirect effect was observed through TPI and LM% in the associations between each of the types of meat consumption and MSI. In the additional analyses, a greater effect of white and fish meat consumption on muscle strength through mediation of TPI and LM% was reported compared to red or processed meat consumption, and no significant effects were observed between processed meat consumption and MSI.

CONCLUSION

Higher consumption of total, red, and white and fish meat was associated with increased MSI in young adults. TPI and LM% mediated this relationship.

Low-dose aspirin use in pregnancy and the risk of preterm birth: a Swedish register-based cohort study

BACKGROUND

Preterm birth is the leading cause of neonatal mortality and morbidity. Women who have had a previous preterm birth are at increased risk for preterm birth in their subsequent pregnancies. Low-dose aspirin use reduces the risk for preterm birth among women at risk of developing preeclampsia, however, it is unclear whether low-dose aspirin may reduce the risk of recurrent preterm birth.

OBJECTIVE

This study aimed to investigate the association between low-dose aspirin use and preterm birth among women with a previous preterm birth.

STUDY DESIGN

We conducted a Swedish register-based cohort study and included women who had a first and second pregnancy between 2006 and 2019, with the first pregnancy ending in preterm birth (medically indicated or with spontaneous onset <37 weeks of gestation). The association between low-dose aspirin use and preterm birth in the second pregnancy was estimated via logistic regression via standardization and expressed as marginal relative risks with the 95% confidence interval.

RESULTS

Among the study cohort (N=22,127), 3057 women (14%) were prescribed low-dose aspirin during their second pregnancy and 3703 women (17%) gave birth prematurely. Low-dose aspirin use was associated with a reduced risk for preterm birth, (marginal relative risk, 0.87; 95% confidence interval, 0.77-0.99). There were no statistically significant associations between low-dose aspirin use and an altered risk for moderate preterm birth, defined as birth between 32 and 36 weeks' gestation (marginal relative risk, 0.90; 95% confidence interval, 0.78-1.03), or very preterm birth, defined as birth <32 weeks' gestation (marginal relative risk, 0.75; 95% confidence interval, 0.54-1.04). Regarding the onset of preterm birth, low-dose aspirin use was associated with a reduced risk for spontaneous preterm birth (marginal relative risk, 0.70; 95% confidence interval, 0.57-0.86) but no reduction in the risk for medically indicated preterm birth (marginal relative risk, 1.09; 95% confidence interval, 0.91-1.30) was observed.

CONCLUSION

Among women with a previous preterm birth, low-dose aspirin use was associated with a reduced risk for preterm birth. When investigating preterm birth by onset in the second pregnancy, low-dose aspirin use was associated with a reduced risk for spontaneous preterm birth. Our results suggest that low-dose aspirin may be an effective prophylaxis for recurrent preterm birth.

The age-related contribution of cognitive function to dual-task gait in middle-aged adults in Spain: observations from a population-based study

BACKGROUND

Poor dual-task gait performance is associated with a risk of falls and cognitive decline in adults aged 65 years or older. When and why dual-task gait performance begins to deteriorate is unknown. This study aimed to characterise the relationships between age, dual-task gait, and cognitive function in middle age (ie, aged 40-64 years).

METHODS

We conducted a secondary analysis of data from community-dwelling adults aged 40-64 years that took part in the Barcelona Brain Health Initiative (BBHI) study, an ongoing longitudinal cohort study in Barcelona, Spain. Participants were eligible for inclusion if they were able to walk independently without assistance and had completed assessments of both gait and cognition at the time of analysis and ineligble if they could not understand the study protocol, had any clinically diagnosed neurological or psychiatric diseases, were cognitively impaired, or had lower-extremity pain, osteoarthritis, or rheumatoid arthritis that could cause abnormal gait. Stride time and stride time variability were measured under single-task (ie, walking only) and dual-task (ie, walking while performing serial subtractions) conditions. Dual-task cost (DTC; the percentage increase in the gait outcomes from single-task to dual-task conditions) to each gait outcome was calculated and used as the primary measure in analyses. Global cognitive function and composite scores of five cognitive domains were derived from neuropsychological testing. We used locally estimated scatterplot smoothing to characterise the relationship between age and dual-task gait, and structural equation modelling to establish whether cognitive function mediated the association between observed biological age and dual tasks.

FINDINGS

996 people were recruited to the BBHI study between May 5, 2018, and July 7, 2020, of which 640 participants completed gait and cognitive assessments during this time (mean 24 days [SD 34] between first and second visit) and were included in our analysis (342 men and 298 women). Non-linear associations were observed between age and dual-task performance. Starting at 54 years, the DTC to stride time (β=0·27 [95% CI 0·11 to 0·36]; p<0·0001) and stride time variability (0·24 [0·08 to 0·32]; p=0·0006) increased with advancing age. In individuals aged 54 years or older, decreased global cognitive function correlated with increased DTC to stride time (β=-0·27 [-0·38 to -0·11]; p=0·0006) and increased DTC to stride time variability (β=-0·19 [-0·28 to -0·08]; p=0·0002).

INTERPRETATION

Dual-task gait performance begins to deteriorate in the sixth decade of life and, after this point, interindividual variance in cognition explains a substantial portion of dual-task performance.

FUNDING

La Caixa Foundation, Institut Guttmann, and Fundació Abertis.

Effects of extended oral antibiotic prophylaxis on surgical site infections after instrumented spinal fusion: a cohort study of 901 patients with a minimum follow-up of 1 year

BACKGROUND AND PURPOSE

We aimed to determine whether an extended oral antibiotic prophylaxis protocol may reduce the rate of surgical site infection (SSI) in patients undergoing instrumented spinal fusion.

PATIENTS AND METHODS

This retrospective cohort study comprise 901 consecutive patients subjected to spinal fusion between September 2011 and December 2018 with a minimum 1-year follow-up. 368 patients operated on between September 2011 and August 2014 were administered standard intravenous prophylaxis. 533 patients operated on between September 2014 and December 2018 were administered an extended protocol with 500 mg of oral cefuroxime axetil every 12 hours (clindamycin or levofloxacin in allergic individuals) until the removal of sutures. SSI was defined following the Centers for Disease Control and Prevention criteria. The association between risk factors and the incidence of SSI was evaluated by odds ratio (OR) with a multiple logistic regression model.

RESULTS

The bivariate analysis showed a statistically significant association between SSI and the type of prophylaxis used ("extended"' = 1.7% vs. "standard" = 6.2%, p= 0.001), with a lower proportion of superficial SSIs with the extended regimen (0.8% vs. 4.1%, p = 0.001). The multiple logistic regression model showed an OR = 0.25 (95% confidence interval [CI] 0.10-0.53) for extended prophylaxis and an OR = 3.5 (CI 1.3-8.1) for non-beta-lactams antibiotics.

CONCLUSION

Extended antibiotic prophylaxis seems to be associated with a reduction in the incidence of superficial SSI in instrumented spine surgery.

Associations between screen use, outdoor time/daylight exposure and sleep changes during the first COVID-19 lockdown in French children from the ELFE and EPIPAGE2 birth cohorts

AIMS

To investigate associations between outdoor and screen time and changes in sleep patterns in children from two nationwide birth-cohorts in the SAPRIS project.

METHODS

During the first French COVID-19 pandemic lockdown, volunteer parents of children enrolled in the ELFE and EPIPAGE2 birth-cohorts completed online questions about their child's outdoor time, screen time, and changes in sleep duration and quality compared with the pre-lockdown situation. In 5700 children (aged 8-9 years, 52% boys) with available data, we assessed associations between outdoor time, screen time, and sleep changes using multinomial logistic regression models adjusted for confounders.

RESULTS

Children spent on average 3 h08 outdoors and 4 h34 using screens/day (3 h27 for leisure, 1 h07 for class-work). Sleep duration increased in 36% of children and decreased in 13.4%; sleep difficulties appeared/increased in 22.5% and decreased/disappeared/remained stable in 18.3%. After adjustment, increased screen time, especially for leisure, was associated with increased and decreased sleep duration (OR(95%CI) = 1.03(1.00-1.06) and OR = 1.06(1.02-1.10), respectively). No association was observed between outdoor time and sleep changes after adjustment.

CONCLUSIONS

Our study adds evidence for the association between high leisure-time screen time and shorter sleep time. It supports current screen guidelines for children, especially during leisure time and for those whose sleep duration is short.

Directed acyclic graphs in perioperative observational research-A systematic review and critique against best practice recommendations

The Directed Acyclic Graph (DAG) is a graph representing causal pathways for informing the conduct of an observational study. The use of DAGs allows transparent communication of a causal model between researchers and can prevent over-adjustment biases when conducting causal inference, permitting greater confidence and transparency in reported causal estimates. In the era of 'big data' and increasing number of observational studies, the role of the DAG is becoming more important. Recent best-practice guidance for constructing a DAG with reference to the literature has been published in the 'Evidence synthesis for constructing DAGs' (ESC-DAG) protocol. We aimed to assess adherence to these principles for DAGs constructed within perioperative literature. Following registration on the International Prospective Register of Systematic Reviews (PROSPERO) and with adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting framework for systematic reviews, we searched the Excerpta Medica dataBASE (Embase), the Medical Literature Analysis and Retrieval System Online (MEDLINE) and Cochrane databases for perioperative observational research incorporating a DAG. Nineteen studies were included in the final synthesis. No studies demonstrated any evidence of following the mapping stage of the protocol. Fifteen (79%) fulfilled over half of the translation and integration one stages of the protocol. Adherence with one stage did not guarantee fulfilment of the other. Two studies (11%) undertook the integration two stage. Unmeasured variables were handled inconsistently between studies. Only three (16%) studies included unmeasured variables within their DAG and acknowledged their implication within the main text. Overall, DAGs that were constructed for use in perioperative observational literature did not consistently adhere to best practice, potentially limiting the benefits of subsequent causal inference. Further work should focus on exploring reasons for this deviation and increasing methodological transparency around DAG construction.

Mixed messages: most spinal pain and osteoarthritis observational research is unclear or misaligned

OBJECTIVES

We assessed authors' language and methods to determine alignment between reported aims, methods, intent, and interpretations in observational studies in spinal pain or osteoarthritis.

STUDY DESIGN AND SETTING

We searched five databases for observational studies that included people with spinal pain or osteoarthritis published in the last 5 years. We randomized 100 eligible studies, and classified study intent (aims and methods) and interpretations as causal, non-causal, unclear, or misaligned.

RESULTS

Overall, 38% of studies were aligned regarding their intent and interpretation (either causally (22%) or non-causally (16%)). 29% of studies' aims and 29% of study methods were unclear. Intent was misaligned in 16% of studies (where aim differed to method) and 23% of studies had misaligned interpretations (where there were multiple conflicting claims). The most common kind of aim was non-causal (38%), and the most common type of method (39%), intent (38%), and interpretations (35%) was causal.

CONCLUSIONS

Misalignment and mixed messages are common in observational research of spinal pain and osteoarthritis. More than 6 in 10 observational studies may be uninterpretable, because study intent and interpretations do not align. While causal methods and intent are most common in observational research, authors commonly shroud causal intent in non-causal terminology.

Facility Volume Thresholds for Optimization of Short- and Long-Term Outcomes in Patients Undergoing Hepatectomy for Primary Liver Tumors

BACKGROUND

Volume-outcome relationships have been described for a variety of surgical procedures. We aimed to define the facility volume threshold at which postoperative mortality after hepatectomy was optimal.

METHODS

We determined volume percentiles for institutions performing hepatectomy for any primary liver tumor within the National Cancer Database (2004-2017). Marginal structural logistic regression defined the volume percentile (Vmin) at which the odds of 90-day mortality were optimally reduced in patients with hepatocellular carcinoma (HCC) or intrahepatic cholangiocarcinoma (ICC). Short-term postoperative and survival outcomes were compared between patients treated at facilities above and below Vmin.

RESULTS

Thresholds for the 10^th/25^th/50^th/75^th/90^th percentiles were 2/7/26/46/59 hepatectomies/year. A total of 17,833 patients underwent resection of HCC or ICC. The 90-day postoperative mortality was optimized at the 75^th percentile for all hepatectomies (IP-weighted OR = 0.67, 95% CI = 0.52-0.87) and major hepatectomy (IP-weighted OR = 0.62, 95% CI = 0.49-0.80). Seven of 446 facilities met the Vmin threshold. The odds of 30-day mortality were also reduced for all hepatectomies (IP-weighted OR = 0.55, 95% CI = 0.42-0.73) and major hepatectomy (IP-weighted OR = 0.58, 95% CI = 0.41-0.75). There were no differences in length of stay or 30-day readmission rate. Patients with HCC or ICC treated at facilities ≥ 10^th percentile had an associated improvement in overall survival.

CONCLUSIONS

Resection of HCC and ICC is performed at a large number of facilities. Postoperative mortality is optimally reduced at facilities performing at least 46 liver operations annually. Regionalization of surgical care among patients with primary liver malignancies to high-volume centers may result in improved outcomes.

Average treatment effect of facility hepatopancreatobiliary malignancy case volume on survival of patients with nonoperatively managed hepatobiliary malignancies

BACKGROUND

Surgical volume-outcome relationships have been described for a variety of procedures. There is scant literature on total institutional volume and outcomes in patients who are nonoperatively managed. We examined the average treatment effect of total hepatopancreatobiliary malignancy case volume on survival outcomes of patients with nonresected hepatobiliary malignancies.

METHODS

We identified patients with hepatopancreatobiliary malignancies [pancreatic adenocarcinoma, pancreatic neuroendocrine neoplasms, hepatocellular carcinoma, biliary tract cancers] within the National Cancer Database (2004-2018). We determined percentile thresholds based on the total annual hepatopancreatobiliary malignancy case volume. We then identified nonoperatively managed patients with hepatocellular carcinoma or biliary tract cancers. We used inverse probability-weighted Cox regression to estimate the effect of facility volume on overall survival.

RESULTS

We identified 710,988 patients with hepatopancreatobiliary malignancies. Total annual hepatopancreatobiliary malignancy case volume of 32, 71, and 177 cases/year corresponded to the 25th, 50th, and 75th percentiles. A total of 96,420 with hepatocellular carcinoma and 52,627 patients with biliary tract cancers were managed nonoperatively. In patients with hepatocellular carcinoma or biliary tract cancer, treatment at ≥25th, ≥50th, and ≥75th percentile facilities was associated with improved median, 1-, 2-, and 3-year overall survival compared with treatment at lower-percentile facilities. On inverse probability-weighted Cox analysis, treatment at higher-percentile facilities resulted in a lower hazard of death. Consistent findings were observed in patients with early or intermediate/advanced hepatocellular carcinoma or metastatic biliary tract cancers.

CONCLUSION

Patients with nonoperatively managed hepatocellular carcinoma or biliary tract cancer who receive treatment at higher-volume facilities have improved survival outcomes. These data suggest regionalization of care for patients with hepatocellular carcinoma or biliary tract cancer to high-volume centers may improve survival.

Preconception sleep duration, sleep timing, and shift work in association with fecundability and live birth among women with a history of pregnancy loss

OBJECTIVE

To evaluate the associations between preconception sleep characteristics and shift work with fecundability and live birth.

DESIGN

Secondary analysis of the Effects of Aspirin in Gestation and Reproduction study, a preconception cohort.

SETTING

Four US academic medical centers.

PATIENT(S)

Women aged 18-40 with a history of 1-2 pregnancy losses who were attempting to conceive again.

INTERVENTION(S)

Not applicable.

MAIN OUTCOME MEASURES(S)

We evaluated baseline, self-reported sleep duration, sleep midpoint, social jetlag, and shift work among 1,228 women who were observed for ≤6 cycles of pregnancy attempts to ascertain fecundability. We ascertained live birth at the end of follow up via chart abstraction. We estimated fecundability odds ratios (FORs) using discrete, Cox proportional hazards models and risk ratios (RRs) for live birth using log-Poisson models.

RESULT(S)

Sleep duration ≥9 vs. 7 to <8 hours (FOR: 0.81, 95% confidence interval [CI], 0.61; 1.08), later sleep midpoints (3rd tertile vs. 2nd tertile: FOR: 0.85; 95% CI, 0.69, 1.04) and social jetlag (continuous per hour; FOR: 0.93, 95% CI: 0.86, 1.00) were not associated with reduced fecundability. In sensitivity analyses, excluding shift workers, sleep duration ≥9 vs. 7 to <8 hours (FOR: 0.62; 95% CI, 0.42; 0.93) was associated with low fecundability. Night shift work was not associated with fecundability (vs. non-night shift work FOR: 1.17, 95% CI, 0.96; 1.42). Preconception sleep was not associated with live birth.

CONCLUSION(S)

Overall, there does not appear to be a strong association between sleep characteristics, fecundability, and live birth. Although these findings may suggest weak and imprecise associations with some sleep characteristics, our findings should be evaluated in larger cohorts of women with extremes of sleep characteristics.

CLINICAL TRIAL REGISTRATION NUMBER

Clinicaltrials.gov NCT00467363.

Trends and disparities in the utilization of systemic chemotherapy in patients with metastatic hepato-pancreato-biliary cancers

BACKGROUND

We described trends and disparities in utilization of systemic chemotherapy in metastatic hepato-pancreato-biliary (HPB) cancers.

METHODS

We queried the National Cancer Database for metastatic HPB cancers [hepatocellular carcinoma (HCC), biliary tract cancers (BTC), pancreatic adenocarcinoma (PDAC)]. We used multivariable analysis to examine the factors associated with utilization of systemic chemotherapy. We utilized marginal structural logistic models to estimate the effect of health insurance, facility type, or facility volume on utilization of systemic chemotherapy.

RESULTS

We identified 162,283 patients with metastatic HPB cancers: 23,923 (14.7%) had HCC, 26,766 (16.5%) had BTC, and 111,594 (68.8%) had PDAC. A total of 37.2% patients with HCC, 55.6% with BTC, and 56.4% with PDAC received chemotherapy. Age ≥70 years and Charlson-Deyo score ≥2 were associated with lower likelihood of receiving chemotherapy across all cancers. Patients with private health insurance had higher receipt of chemotherapy. Receiving treatment at academic facilities had no effect on the receipt of chemotherapy. Treatment of patients with HCC or PDAC at high-volume facilities resulted in higher receipt of chemotherapy.

CONCLUSION

A significant proportion of patients with metastatic HPB cancers do not receive systemic chemotherapy. Several disparities in administration of chemotherapy for metastatic HPB cancers exist.

Association of service use with subjective oral health indicators in a freedom of choice pilot

OBJECTIVES

A freedom of choice pilot provided access to private oral health care services without queuing and with fixed public service-fees for participants in Tampere region, Finland in 2018-2019. The aim of this study was to investigate how use of oral health care services differed by demographics, socioeconomic status, dental fear, and self-reported oral health in this pilot.

MATERIAL AND METHODS

SMS-messages including a link to online questionnaire were sent to participants who had booked an appointment, and to those who had not booked an appointment despite registering to pilot. We categorized participants to (1) those who had booked their first appointment before receiving SMS (visitors), (2) those who booked an appointment after receiving the SMS-message (late-visitors), and (3) those who had not booked an appointment during pilot (nonvisitors). We used regression analysis to estimate the association of age, gender, dental fear, economic situation, Oral Health Impact Profile-14-severity (oral health-related quality of life [OHRQoL]), self-reported oral health and need for oral health care (exposures) with oral health care service use during the pilot (outcome).

RESULTS

Out of 2300 participants, 636 (28%) responded. Late-visitors were more likely older and reported more likely need for oral health care, poorer oral health and OHRQoL than visitors or nonvisitors. Nonvisitors were younger and had better OHRQoL than the others. The differences in the service use by gender, economic situation, and dental fear were small.

CONCLUSIONS

Service use during the pilot depended on the subjective oral health. Our findings highlight the potential of reminders in increasing care use among those with perceived need for services.

Directed Acyclic Graphs in Surgical Research

Surgical research often utilizes multivariable regression to evaluate causal relationships between variables, but there is usually little explanation of the decision-making regarding which variables were controlled for. We propose that directed acyclic graphs (DAGs)-a formal logic tool that illustrates connections between variables-should be used to define and communicate variable relationships to readers and other audiences. While literature in epidemiology and other medical fields has recently started to incorporate DAGs more, they are still seldom seen in surgical publications. In this review, we describe the background and need for DAGs and argue for their use. Next, we explain how bias can be introduced without a thoughtful approach to control variable selection. Finally, we recommend that researchers communicate their choices and rationale when selecting control variables in published surgical research.

Strong instrumental variables biased propensity scores in comparative effectiveness research: A case study in oncology

BACKGROUND AND OBJECTIVES

Some medications require specific medical procedures in the weeks before their start. Such procedures may meet the definition of instrumental variables (IVs). We examined how they may influence treatment effect estimation in propensity score (PS)-adjusted comparative studies, and how to remedy.

STUDY DESIGN AND SETTING

Different covariate assessment periods (CAPs) did and did not include the month preceding treatment start were used to compute PS in the French claims database (Sytème National des Données de Santé-SNDS), and 1:1 match patients with metastatic castration resistant prostate cancer initiating abiraterone acetate or docetaxel. The 36-month survival was assessed.

RESULTS

Among 1, 213 docetaxel and 2, 442 abiraterone initiators, the PS distribution resulting from the CAP [-12; 0 months] distinctly separated populations (c = 0.93; 273 matched pairs). The CAPs [-12;-1 months] identified 765 pairs (c = 0.81). Strong docetaxel treatment predictors during the month before treatment start were implantable delivery systems (1% vs. 59%), which fulfilled IV conditions. The 36-month survival was not meaningfully different under the [-12; 0 months] CAP but differed by 10% points (38% vs. 28%) after excluding month -1.

CONCLUSION

In the setting of highly predictive pretreatment procedures, excluding the immediate pre-exposure time from the CAP will reduce the risk of including potential IVs in PS models and may reduce bias.

Cystic fibrosis related diabetes is not associated with maximal aerobic exercise capacity in cystic fibrosis: a cross-sectional analysis of an international multicenter trial

BACKGROUND

Previous studies have reported differences in aerobic exercise capacity, expressed as peak oxygen uptake (VO_2peak), between people with and without cystic fibrosis (CF) related diabetes (CFRD). However, none of the studies controlled for the potential influence of physical activity on VO_2peak. We investigated associations between CFRD and VO_2peak following rigorous control for confounders including objectively measured physical activity.

METHODS

Baseline data from the international multicenter trial ACTIVATE-CF with participants ≥12 years performing up to 4 h per week of vigorous physical activity were used for this project. Multivariable models were computed to study associations between CFRD and VO_2peak (mL.min^-1) adjusting for a set of pre-defined covariates: age, sex, weight, forced expiratory volume in 1 s (FEV₁), breathing reserve index, Pseudomonas aeruginosa infection, and physical activity (aerobic step counts from pedometry). Variables were selected based on their potential confounding effect on the association between VO_2peak and CFRD.

RESULTS

Among 117 randomized individuals, 103 (52% female) had a maximal exercise test and were included in the analysis. Participants with (n = 19) and without (n = 84) CFRD did not differ in FEV₁, physical activity, nutritional status, and other clinical characteristics. There were also no differences in VO_2peak (mL.min^-1 or mL.kg^-1.min^-1 or% predicted). In the final multivariable model, all pre-defined covariates were significant predictors of VO_2peak (mL.min^-1), however CFRD [coefficient 82.1, 95% CI -69.5 to 233.8, p = 0.28] was not.

CONCLUSIONS

This study suggests no meaningful differences in VO_2peak between people with and without CFRD given comparable levels of physical activity.

Estimating causal links of long-term exposure to particulate matters with all-cause mortality in South China

BACKGROUND

The association between long-term particulate matter (PM) exposure and all-cause mortality has been well-documented. However, evidence is still limited from high-exposed cohorts, especially for PM₁ which is smaller while more toxic than other commonly investigated particles. We aimed to examine the potential casual links of long-term PMs exposure with all-cause mortality in high-exposed areas.

METHODS

A total of 580,757 participants in southern China were enrolled during 2009-2015 and followed up to 2020. The annual average concentration of PM₁, PM_2.5, and PM₁₀ at 1 km² spatial resolution was assessed for each residential address through validated spatiotemporal models. We used marginal structural Cox models to estimate the PM-mortality associations which were further stratified by sociodemographic, lifestyle factors and general exposure levels.

RESULTS

37,578 deaths were totally identified during averagely 8.0 years of follow-up. Increased exposure to all 3 PM size fractions were significantly associated with increased risk of all-cause mortality, with hazard ratios (HRs) of 1.042 (95 % confidence interval (CI): 1.037-1.046), 1.031 (95 % CI: 1.028-1.033), and 1.029 (95 % CI: 1.027-1.031) per 1 μg/m³ increase in PM₁, PM_2.5, and PM₁₀ concentrations, respectively. We observed greater effect estimates among the elderly (age ≥ 65 years), unmarried participants, and those with low education attainment. Additionally, the effect of PM₁, PM_2.5, and PM₁₀ tend to be higher in the low-exposure group than in the general population.

CONCLUSIONS

We provided comprehensive evidence for the potential causal links betweenlong-term PM exposureand all-cause mortality, and suggested stronger links for PM₁compared to large particles and among certain vulnerable subgroups.

Associations between total dairy, high-fat dairy and low-fat dairy intake, and depressive symptoms: findings from a population-based cross-sectional study

PURPOSE

Evidence on the association between dairy intake and depression is conflicting. Given numerous dietary guidelines recommend the consumption of low-fat dairy products, this study examined associations between total dairy, high-fat dairy, and low-fat dairy intake and the prevalence of elevated depressive symptoms. Associations between dairy products, which differed in both fat content and fermentation status, and depressive symptoms were also explored.

METHODS

This cross-sectional study included 1600 Finnish adults (mean age 63 ± 6 years; 51% female) recruited as part of the Kuopio Ischaemic Heart Disease Risk Factor Study. Dairy intake was assessed using 4-day food records. Elevated depressive symptoms were defined as having a score ≥ 5 on the Diagnostic and Statistical Manual of Mental Disorders-III Depression Scale, and/or regularly using one or more prescription drugs for depressive symptoms.

RESULTS

In total, 166 participants (10.4%) reported having elevated depressive symptoms. Using multivariate logistic regression models, intake in the highest tertile of high-fat dairy products (OR 0.64, 95% CI 0.41-0.998, p trend = 0.04) and high-fat non-fermented dairy products (OR 0.60, 95% CI 0.39-0.92, p trend = 0.02) were associated with reduced odds for having elevated depressive symptoms. Whereas no significant association was observed between intake of total dairy, low-fat dairy, or other dairy products, and depressive symptoms.

CONCLUSION

Higher intake of high-fat dairy and high-fat non-fermented dairy products were associated with reduced odds for having elevated depressive symptoms in middle-aged and older Finnish adults. Given the high global consumption of dairy products, and widespread burden of depression, longitudinal studies that seek to corroborate these findings are required.

Association between chemical mixtures and female fertility in women undergoing assisted reproduction in Sweden and Estonia

OBJECTIVE

Women of reproductive age are exposed to ubiquitous chemicals such as phthalates, parabens, and per- and polyfluoroalkyl substances (PFAS), which have potential endocrine disrupting properties and might affect fertility. Our objective was to investigate associations between potential endocrine-disrupting chemicals (EDCs) and female fertility in two cohorts of women attending fertility clinics.

METHODS

In a total population of 333 women in Sweden and Estonia, we studied the associations between chemicals and female fertility, evaluating ovarian sensitivity index (OSI) as an indicator of ovarian response, as well as clinical pregnancy and live birth from fresh and frozen embryo transfers. We measured 59 chemicals in follicular fluid samples and detected 3 phthalate metabolites, di-2-ethylhexyl phthalate (DEHP) metabolites, 1 paraben, and 6 PFAS in >90% of the women. Associations were evaluated using multivariable-adjusted linear or logistic regression, categorizing EDCs into quartiles of their distributions, as well as with Bayesian Kernel Machine Regression.

RESULTS

We observed statistically significant lower OSI at higher concentrations of the sum of DEHP metabolites in the Swedish cohort (Q4 vs Q1, β = -0.21, 95% CI: -0.38, -0.05) and methylparaben in the Estonian cohort (Q3 vs Q1, β = -0.22, 95% CI: -0.44, -0.01). Signals of potential associations were also observed at higher concentrations of PFUnDA in both the combined population (Q2 vs. Q1, β = -0.16, 95% CI -0.31, -0.02) and the Estonian population (Q2 vs. Q1, β = -0.27, 95% CI -0.45, -0.08), and for PFOA in the Estonian population (Q4 vs. Q1, β = -0.31, 95% CI -0.61, -0.01). Associations of chemicals with clinical pregnancy and live birth presented wide confidence intervals.

CONCLUSIONS

Within a large chemical mixture, we observed significant inverse associations levels of DEHP metabolites and methylparaben, and possibly PFUnDA and PFOA, with OSI, suggesting that these chemicals may contribute to altered ovarian function and infertility in women.

School-age outcomes among IVF-conceived children: A population-wide cohort study

BACKGROUND

In vitro fertilisation (IVF) is a common mode of conception. Understanding the long-term implications for these children is important. The aim of this study was to determine the causal effect of IVF conception on primary school-age childhood developmental and educational outcomes, compared with outcomes following spontaneous conception.

METHODS AND FINDINGS

Causal inference methods were used to analyse observational data in a way that emulates a target randomised clinical trial. The study cohort comprised statewide linked maternal and childhood administrative data. Participants included singleton infants conceived spontaneously or via IVF, born in Victoria, Australia between 2005 and 2014 and who had school-age developmental and educational outcomes assessed. The exposure examined was conception via IVF, with spontaneous conception the control condition. Two outcome measures were assessed. The first, childhood developmental vulnerability at school entry (age 4 to 6), was assessed using the Australian Early Developmental Census (AEDC) (n = 173,200) and defined as scoring <10th percentile in ≥2/5 developmental domains (physical health and wellbeing, social competence, emotional maturity, language and cognitive skills, communication skills, and general knowledge). The second, educational outcome at age 7 to 9, was assessed using National Assessment Program-Literacy and Numeracy (NAPLAN) data (n = 342,311) and defined by overall z-score across 5 domains (grammar and punctuation, reading, writing, spelling, and numeracy). Inverse probability weighting with regression adjustment was used to estimate population average causal effects. The study included 412,713 children across the 2 outcome cohorts. Linked records were available for 4,697 IVF-conceived cases and 168,503 controls for AEDC, and 8,976 cases and 333,335 controls for NAPLAN. There was no causal effect of IVF-conception on the risk of developmental vulnerability at school-entry compared with spontaneously conceived children (AEDC metrics), with an adjusted risk difference of -0.3% (95% CI -3.7% to 3.1%) and an adjusted risk ratio of 0.97 (95% CI 0.77 to 1.25). At age 7 to 9 years, there was no causal effect of IVF-conception on the NAPLAN overall z-score, with an adjusted mean difference of 0.030 (95% CI -0.018 to 0.077) between IVF- and spontaneously conceived children. The models were adjusted for sex at birth, age at assessment, language background other than English, socioeconomic status, maternal age, parity, and education. Study limitations included the use of observational data, the potential for unmeasured confounding, the presence of missing data, and the necessary restriction of the cohort to children attending school.

CONCLUSIONS

In this analysis, under the given causal assumptions, the school-age developmental and educational outcomes for children conceived by IVF are equivalent to those of spontaneously conceived children. These findings provide important reassurance for current and prospective parents and for clinicians.

The effect of face-to-face versus online learning on student performance in anatomy: an observational study using a causal inference approach

Introduction

This study aimed to estimate the causal effect of face-to-face learning on student performance in anatomy, compared to online learning, by analysing examination marks under a causal structure.

Methods

We specified a causal graph to indicate how the mode of learning affected student performance. We sampled purposively to obtain end-semester examination marks of undergraduate and postgraduate students who learned using face-to-face (pre-COVID, 2019) or online modes (post-COVID, 2020). The analysis was informed by the causal graph. Marks were compared using linear regression, and sensitivity analyses were conducted to assess if effects were robust to unmeasured confounding.

Results

On average, face-to-face learning improved student performance in the end-semester examination in undergraduate students (gain of mean 8.3%, 95% CI 3.3 to 13.4%; E-value 2.77, lower limit of 95% CI 1.80) but lowered performance in postgraduate students (loss of 8.1%, 95% CI 3.6 to 12.6%; E-value 2.89, lower limit of 95% CI 1.88), compared to online learning.

Discussion

Under the assumed causal graph, we found that compared to online learning, face-to-face learning improved student performance in the end-semester examination in undergraduate students, but worsened student performance in postgraduate students. These findings suggest that different modes of learning may suit different types of students. Importantly, this is the first attempt to estimate causal effects of the mode of learning on student performance under a causal structure. This approach makes our assumptions transparent, informs data analysis, and is recommended when using observational data to make causal inferences.

Preliminary Evidence for the Sequentially Mediated Effect of Racism-Related Stress on Pain Sensitivity Through Sleep Disturbance and Corticolimbic Opioid Receptor Function

Sleep disturbance predicts worse pain outcomes. Because sleep disturbance inequitably impacts Black adults - with racism as the upstream cause - understanding how racism-related stress impacts pain through sleep might help minimize racialized pain inequities. This preliminary study examined sequential mediation of the effect of racism-related stress on experimental pain through sleep disturbance and corticolimbic μOR function in pain-free non-Hispanic Black (NHB) and White (NHW) adults. Participants completed questionnaires, actigraphy, positron emission tomography, and sensory testing. We reproduced findings showing greater sleep disturbance and pain sensitivity among NHB participants; greater sleep disturbance (r = .35) and lower pain tolerance (r=-.37) were significantly associated with greater racism-related stress. In a sequential mediation model, the total effect of racism-related stress on pain tolerance (β=-.38, P = .005) weakened after adding sleep disturbance and ventromedial prefrontal cortex (vmPFC) μOR binding potential (BPND) as mediators (β = -.18, P = .16). The indirect effect was statistically significant [point estimate = -.003, (-.007, -.0003). Findings showed a potential sequentially mediated effect of racism-related stress on pain sensitivity through sleep disturbance and vmPFC μOR BPND. As policy efforts are enacted to eliminate the upstream cause of systemic racism, these results cautiously suggest that sleep interventions within racism-based trauma informed therapy might help prevent downstream effects on pain. PERSPECTIVE: This preliminary study identified the effect of racism-related stress on pain through sleep disturbance and mu-opioid receptor binding potential in the ventromedial prefrontal cortex. Findings cautiously support the application of sleep interventions within racism-based trauma-informed therapy to prevent pain inequities as policy changes function to eliminate all levels of racism.

The executive disruption model of tinnitus distress: Model validation in two independent datasets using factor score regression

This study presents the executive disruption model (EDM) of tinnitus distress and subsequently validates it statistically using two independent datasets (the Construction Dataset: n = 96 and the Validation Dataset: n = 200). The conceptual EDM was first operationalised as a structural causal model (construction phase). Then multiple regression was used to examine the effect of executive functioning on tinnitus-related distress (validation phase), adjusting for the additional contributions of hearing threshold and psychological distress. For both datasets, executive functioning negatively predicted tinnitus distress score by a similar amount (the Construction Dataset: β = -3.50, p = 0.13 and the Validation Dataset: β = -3.71, p = 0.02). Theoretical implications and applications of the EDM are subsequently discussed; these include the predictive nature of executive functioning in the development of distressing tinnitus, and the clinical utility of the EDM.

Development and application of an evidence-based directed acyclic graph to evaluate the associations between metal mixtures and cardiometabolic outcomes

Objectives

Specifying causal models to assess relationships among metal mixtures and cardiometabolic outcomes requires evidence-based models of the causal structures; however, such models have not been previously published. The objective of this study was to develop and evaluate a directed acyclic graph (DAG) diagraming metal mixture exposure and cardiometabolic outcomes.

Methods

We conducted a literature search to develop the DAG of metal mixtures and cardiometabolic outcomes. To evaluate consistency of the DAG, we tested the suggested conditional independence statements using linear and logistic regression analyses with data from the San Luis Valley Diabetes Study (SLVDS; n=1795). We calculated the proportion of statements supported by the data and compared this to the proportion of conditional independence statements supported by 1,000 DAGs with the same structure but randomly permuted nodes. Next, we used our DAG to identify minimally sufficient adjustment sets needed to estimate the association between metal mixtures and cardiometabolic outcomes (i.e., cardiovascular disease, fasting glucose, and systolic blood pressure). We applied them to the SLVDS using Bayesian kernel machine regression, linear mixed effects, and Cox proportional hazards models.

Results

From the 42 articles included in the review, we developed an evidence-based DAG with 74 testable conditional independence statements (43 % supported by SLVDS data). We observed evidence for an association between As and Mn and fasting glucose.

Conclusions

We developed, tested, and applied an evidence-based approach to analyze associations between metal mixtures and cardiometabolic health.