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

Effect of homemade peanut oil consumption during pregnancy on low birth weight and preterm birth outcomes: a cohort study in Southwestern China

BACKGROUND

Homemade peanut oil is widely consumed in rural areas of Southwestern China, which is easily contaminated by aflatoxins (AFs) and associated with adverse birth outcomes.

OBJECTIVE

To identify the effect of exposure to homemade peanut oil consumption on low birth weight (LBW), preterm birth (PB) and other associated factors.

METHODS

A prospective cohort study was conducted among pregnant women in Guangxi province, Southwestern China. Information of all eligible women on homemade peanut oil consumption and potential factors associated with LBW and PB was collected, and all were followed up until delivery. The effect of homemade peanut oil exposure was analyzed using multiple logistic regression models using the directed acyclic graph (DAG) approach.

RESULTS

Of 1611 pregnant women, 1316 (81.7%) had consumed homemade peanut oil, and the rates of LBW and PB were 9.7% and 10.0%, respectively. Increased risks of LBW and PB in women with homemade peanut oil consumption were found with aORs of 1.9 (95% CI 1.1-3.2) and 1.8 (95% CI 1.1-3.0), respectively. Women with a history of PB or LBW were 3-5 times more likely to have higher rates of LBW or PB compared with those without this type of history. The odds of PB were approximately double in those taking medicine during pregnancy. Advanced maternal age, lack of physical exercise during pregnancy, passive smoking, or pregnancy complications were also more likely to have a higher risk of LBW.

CONCLUSIONS

Homemade peanut oil consumption was a potential risk factor for both LBW and PB, of which health authorities who are responsible for food safety of the country should pay more attention to providing recommendation for oil consumption during pregnancy.

Co-exposure to parabens, bisphenol A, and triclosan and the associations with dyslipidemia in Chinese older adults: The mediation effect of oxidative stress

Dyslipidemia is a prevalent metabolic disorder in older adults and has negative effects on cardiovascular health. However, the combined effect of paraben, bisphenol A (BPA), and triclosan (TCS) exposure on dyslipidemia and the underlying mechanisms remain unclear. This cross-sectional study recruited 486 individuals ≥60 years in Shenzhen, China. Morning spot urine samples were collected and analyzed for four parabens, BPA, TCS, and 8-hydroxy-2'-deoxyguanosine (8-OHdG), a typical biomarker for oxidative stress, using mass spectrometry. Blood samples were tested for lipid levels using an automated biochemical analyzer. Quantile-based g-computation (QGC) was used to assess the combined effects of exposures on dyslipidemia. Mediation analysis was applied to investigate the mediating role of 8-OHdG between exposure and dyslipidemia. QGC showed that co-exposure to parabens, BPA, and TCS was positively linked with hypercholesterolemia (OR: 1.17, 95%CI: 1.10-1.24, P < 0.001) and hyper-LDL-cholesterolemia (OR: 1.35, 95%CI: 1.05-1.75, P = 0.019). Methylparaben (MeP), n-propyl paraben (PrP), and butylparaben (BtP) were the major contributors. 8-OHdG mediated 6.5% and 13.0% of the overall effect of the examined chemicals on hypercholesterolemia and hyper-LDL-cholesterolemia, respectively (all P < 0.05). Our study indicated that co-exposure to parabens, BPA, and TCS is associated with dyslipidemia and oxidative stress partially mediate the association. Future research is needed to explore additional mechanisms underlying these relationships.

Do maternal haemodynamics have a causal influence on treatment for gestational diabetes?

BACKGROUND

Arterial stiffening is believed to contribute to the worsening of insulin resistance, and factors which are associated with needing pharmacological treatment of gestational diabetes (GDM), such as maternal obesity or advanced age, are associated with impaired cardiovascular adaptation to pregnancy. In this observational study, we aimed to investigate causal relationships between maternal haemodynamics and treatment requirement amongst women with GDM.

METHODS

We assessed maternal haemodynamics in women with GDM, comparing those who remained on dietary treatment with those who required pharmacological management. Maternal haemodynamics were assessed using the Arteriograph® (TensioMed Ltd, Budapest, Hungary) and the NICOM® non-invasive bio-reactance method (Cheetah Medical, Portland, Oregon, USA). A graphical causal inference technique was used for statistical analysis.

RESULTS

120 women with GDM were included in the analysis. Maternal booking BMI was identified as having a causative influence on treatment requirement, with each unit increase in BMI increasing the odds of needing metformin and/or insulin therapy by 12% [OR 1.12 (1.02 - 1.22)]. The raw values of maternal heart rate (87.6 ± 11.7 vs. 92.9 ± 11.90 bpm, p = 0.014) and PWV (7.8 ± 1.04 vs. 8.4 ± 1.61 m/s, p = 0.029) were both significantly higher amongst the women requiring pharmacological management, though these relationships did not remain significant in causal logistic regression.

CONCLUSIONS

Maternal BMI at booking has a causal, rather than simply associational, relationship on the need for pharmacological treatment of GDM. No significant causal relationships were found between maternal haemodynamics and the need for pharmacological treatment.

Using staged tree models for health data: Investigating invasive fungal infections by aspergillus and other filamentous fungi

Machine learning models are increasingly used in the medical domain to study the association between risk factors and diseases to support practitioners in understanding health outcomes. In this paper, we showcase the use of machine-learned staged tree models for investigating complex asymmetric dependence structures in health data. Staged trees are a specific class of generative, probabilistic graphical models that formally model asymmetric conditional independence and non-regular sample spaces. An investigation of the risk factors in invasive fungal infections demonstrates the insights staged trees provide to support medical decision-making.

Paediatric cancer survivors: lean mass attenuates negative impact of watching television on bone

OBJECTIVE

To investigate the associations of television (TV) watching time with bone parameters and to examine whether high lean mass attenuates the negative impact of watching TV more than one hour per day on bone parameters.

METHODS

This cross-sectional study comprised 116 young paediatric cancer survivors. Dual-energy X-ray Absorptiometry was used to obtain total body and regional areal bone mineral density (g/cm2), and lean mass (kg) outcomes. Hip Structural Analysis was performed at the narrowest point of the femoral neck. Trabecular Bone Score was obtained in the lumbar spine. TV watching time was obtained using the "Youth Activity Profile" questionnaire.

RESULTS

Multiple linear regression models showed negative associations of watching TV more than one hour with bone parameters in peri/post pubertal survivors (β = -0.359 to -0.614, P < 0.001 to 0.047). Those survivors watching TV more than one hour per day and with high lean mass presented higher bone parameter Z-score than those with low lean mass.

CONCLUSION

These findings underline the necessity of identifying strategies that promote musculoskeletal development while reducing TV watching time in young paediatric cancer survivors to maximise bone regeneration.

IMPACT

The results indicate that watching television (TV) more than one hour (compared to not watching TV) is negatively associated with bone parameters in peri/post pubertal survivors. Survivors with high lean mass counteract these negative associations of watching TV with bone parameters. It is important to promote musculoskeletal development in this vulnerable population to maximise bone regeneration.

Sex-Specific Impact of Serum Calcium Levels on Acute Coronary Syndrome Risk: A 19-Year Cohort Study in Korea

Background: This study aims to investigate the association between serum calcium levels and acute coronary syndrome (ACS) risk, examining whether this relationship differs by sex, given the known differences in calcium metabolism and hormonal influences between males and females. Methods: Utilizing the Korean Genome Epidemiology Study (KoGES) prospective cohort data, our primary exposure variables were serum calcium level and sex. The incidence of ACS served as the main outcome of interest. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using Cox regression analysis. An interaction analysis was conducted to assess the interaction effect of calcium level and sex on ACS incidence. Results: After adjusting for confounding variables, high calcium intake did not significantly increase ACS incidence, with a hazard ratio (HR) of 1.07 (95% CI: 0.90-1.26). There was also no significant difference in ACS risk between females and males (HR: 0.81, 95% CI: 0.61-1.04). However, interaction effect analysis revealed that higher calcium levels were associated with an increased risk of ACS only in females (HR: 1.24, 95% CI: 1.07-1.58), whereas the association in males was not statistically significant (HR: 0.90, 95% CI: 0.71-1.15). Conclusion: Our study results indicate that elevated serum calcium levels alone did not independently increase the risk of ACS; however, high serum calcium levels were associated with an increased risk of ACS in females but not in males, underscoring the importance of sex-specific factors in assessing and managing ACS risk and highlighting the necessity for personalized medical approaches to improve cardiovascular health outcomes for women.

Potential causal links of long-term exposure to PM2.5 and its chemical components with the risk of nasopharyngeal carcinoma recurrence: A 10-year cohort study in South China

There is a lack of evidence from cohort studies on the causal association of long-term exposure to ambient fine particulate matter (PM2.5) and its chemical components with the risk of nasopharyngeal carcinoma (NPC) recurrence. Based on a 10-year prospective cohort of 1184 newly diagnosed NPC patients, we comprehensively evaluated the potential causal links of ambient PM2.5 and its chemical components including black carbon (BC), organic matter (OM), sulfate (SO4 2-), nitrate (NO3 -), and ammonium (NH4 +) with the recurrence risk of NPC using a marginal structural Cox model adjusted with inverse probability weighting. We observed 291 NPC patients experiencing recurrence during the 10-year follow-up and estimated a 33% increased risk of NPC recurrence (hazard ratio [HR]: 1.33, 95% confidence interval [CI]: 1.02-1.74) following each interquartile range (IQR) increase in PM2.5 exposure. Each IQR increment in BC, NH4 +, OM, NO3 -, and SO4 2- was associated with HRs of 1.36 (95%CI: 1.13-1.65), 1.35 (95%CI: 1.07-1.70), 1.33 (95%CI: 1.11-1.59), 1.32 (95%CI: 1.06-1.64), 1.31 (95%CI: 1.08-1.57). The elderly, patients with no family history of cancer, no smoking history, no drinking history, and those with severe conditions may exhibit a greater likelihood of NPC recurrence following exposure to PM2.5 and its chemical components. Additionally, the effect estimates of the five components are greater among patients who were exposed to high concentration than in the full cohort of patients. Our study provides solid evidence for a potential relationship between long-term exposure to PM2.5 and its components and the risk of NPC recurrence.

Considering the role of speech processing in cleft-related speech sound disorders: Implications for causal pathways and classification systems

BACKGROUND

Classification systems in healthcare support shared understanding of conditions for clinical communication, service monitoring and development, and research. Children born with cleft palate with or without cleft lip (CP+/-L) are at high risk of developing cleft-related speech sound disorder (SSD). The way cleft-related SSD is represented and described in SSD classification systems varies. Reflecting on the potential causal pathways for different cleft-related speech features, including the role of speech processing skills, may inform how cleft-related SSD is represented in classification systems.

AIM & APPROACH

To explore and reflect on how cleft-related SSD is represented in current SSD classification systems in the context of considering how speech processing skills and other factors may be involved in causal pathways of cleft speech characteristics (CSCs).

MAIN CONTRIBUTION

Variation in the representation of cleft-related SSD in classification systems is described. Potential causal pathways for passive cleft- related speech features and different active CSCs are explored. The factors involved in the development and/or persistence of different active CSCs may vary. Some factors may be specific to children born with CP+/-L, but if speech processing skills are also involved, this is an overlap with other SSD subtypes. Current evidence regarding relationships between different speech processing skills and active CSCs is limited. Implications for the representation of cleft-related SSD in SSD classification systems are discussed.

CONCLUSION

There are different categories of cleft-related speech features which are essential to understand and identify in children with cleft-related SSD to ensure appropriate management. Representation of these feature categories in classification systems could support understanding of speech in this population. Speech processing skills could be involved in the development and/or persistence of different active CSCs in individual children. Reflection and discussion on how cleft-related SSD is represented in classification systems in relation to other SSD subtypes may inform future iterations of these systems. Further work is needed to understand factors influencing the development and/or persistence of active CSCs, including speech processing skills.

WHAT THIS PAPER ADDS

What is already known on the subject Cleft-related speech sound disorder (SSD) is commonly described as being of known origin. The features of cleft-related SSD have been described extensively and several authors have also examined factors which may contribute to speech development and outcomes in children born with cleft palate +/- lip. There is limited evidence regarding the role of speech processing in the development and persistence of cleft-related SSD. What this study adds This paper reflects on how cleft-related SSD is represented in SSD classification systems in relation to key feature categories of cleft-related SSD and possible causal pathways for passive features and active cleft speech characteristics (CSCs). The role of cognitive speech processing skills is specifically considered alongside other factors that may contribute to the development of active CSCs. What are the clinical implications of this work? Causal pathways for different features of cleft-related SSD may vary, particularly between passive and active features, abut also between different active CSCs. Speech and language therapists (SLTs) need to differentially diagnose passive speech features and active CSCs. Consideration of the role of different speech processing skills and interactions with other potentially influencing factors in relation to active CSCs may inform clinical hypotheses and speech and language therapy (SLT) intervention. Representing key features of cleft-related SSD in classification systems may support understanding of cleft-related SSD in relation to other SSD subtypes.

Greater ischiatic notch plane versus parasacral block in dogs undergoing pelvic limb surgery: a retrospective noninferiority cohort study on perioperative opioid consumption

OBJECTIVE

To compare perioperative opioid consumption in dogs undergoing pelvic limb surgeries and receiving a lumbosacral plexus block by combining a lumbar [lateral pre-iliac (LPI)] block with a lumbosacral trunk [i.e. parasacral (PS group) or greater ischiatic notch plane (GIN group)] block. Study design Retrospective, descriptive, exploratory, noninferiority cohort study.

METHODS

Medical records of 37 client-owned dogs were analyzed. Intraoperative (primary outcome) and postoperative (24 hours) opioid use, 24 hour cumulative pain scores and prevalence of complications were compared between the two groups. Opioid use was quantified in morphine equivalents (ME, mg kg-1 ). The noninferiority limit for intraoperative opioid consumption was set at 0.05 ME kg-1 hour-1. Demographic data, procedure duration, surgery type and perioperative dexmedetomidine and ketamine use were also collected. A t-test or Wilcoxon rank-sum test, a Fisher's exact test and multivariable linear regression were used. Significance was set at p < 0.05.

RESULTS

The GIN and PS groups comprised 17 and 20 dogs, respectively. Total intraoperative ME consumption was 0.17 (0.11-0.21) and 0.22 (0.16-0.30) mg kg-1 hour-1 for the GIN and PS groups, respectively (p = 0.077). The noninferiority analysis adjusted by surgery type and body mass revealed that the mean difference between the groups (GIN - PS) was -0.039 (95% CI -0.11-0.03, p = 0.247) ME mg kg-1 hour-1, indicating that the GIN group was not inferior to the PS group regarding intraoperative ME consumption. Dexmedetomidine and ketamine use, postoperative ME consumption and pain scores were similar between groups.

CONCLUSIONS AND CLINICAL RELEVANCE

This retrospective study and preliminary observations support the clinical use of the GIN plane block, demonstrating its noninferiority to the PS block when combined with the LPI block for multimodal perioperative analgesia in dogs undergoing pelvic limb surgeries.

Nucleoside/nucleotide reverse transcriptase inhibitor-associated weight gain in people living with HIV: data from the Copenhagen Comorbidity in HIV Infection (COCOMO) study

Weight gain effects of Nucleoside/Nucleotide Reverse Transcriptase Inhibitors in people with HIV (PWH) have been sparsely studied.Participants were enrolled in the Copenhagen Comorbidity in HIV Infection (COCOMO) study. PWH receiving a backbone of emtricitabine, or lamivudine combined with abacavir, tenofovir disoproxil, or tenofovir alafenamide were analysed. Weight gain according to ART backbone and to the third drug was analysed using a multiple linear regression model. Non-ART risk factors were also determined using multiple linear regression.A total of 591 participants were included in the analysis. The majority were middle-aged, virally suppressed males with a mean BMI just above the normal range. Both tenofovir disoproxil/emtricitabine or lamivudine and abacavir /emtricitabine or lamivudine, but not tenofovir alafenamide /emtricitabine or lamivudine were associated with weight gain over two years (0.6 kg, p = 0.025; 1.0 kg, p = 0.005). The third drugs associated with weight increase were non-nucleoside reverse transcriptase inhibitors (NNRTI) (p = 0.035), dolutegravir (p = 0.008) and atazanavir (p = 0.040). Non-ART risk factors for gaining weight were low or normal BMI, age <40 years, underweight, inactivity or highly active at baseline.Tenofovir disoproxil and abacavir-based ART regimens were associated with a small weight gain. Third drug NNRTI, dolutegravir and atazanavir were associated with an increase in weight.

Dependent censoring bias assessment using inverse probability of censoring weights: Type 2 diabetes mellitus risk in patients initiating bisoprolol versus other antihypertensives in a Clinical Practice Research Datalink cohort study

Dependent censoring involves a preferential attrition of a subgroup of interest; occurring in survival analysis, it may impact interpretation by introducing a selection bias. To assess the potential bias in a comparison of bisoprolol to other antihypertensives in terms of Type 2 diabetes mellitus (T2DM) incidence, inverse probability of censoring weights (IPCW) was used. It was further used to contextualize results obtained through competing risks analysis. Two estimands were considered to assess T2DM incidence while accounting for deviations from the initial antihypertensive monotherapy (DFM). A hypothetical estimand using IPCW, treating DFM as censoring, was interpreted together with a 'while-on-treatment' estimand, treating DFM as a competing risk. We illustrated our application with a cohort study based on Clinical Practice Research Datalink (CPRD) including 267,352 patients with newly diagnosed arterial hypertension between 2000 and 2017, initiating antihypertensive monotherapy among bisoprolol, other beta-blockers, renin-angiotensin system drugs (ACEi/ARB), diuretics and calcium-channel blockers. A mild dependent censoring process was hypothesized, leading to slight overestimation of T2DM incidence. Although subject to some limitations, a nonsignificant trend toward an excess of risk associated with ACEi/ARB was yielded consistently by IPCW and competing risks analyses. Conversely, in comparisons of bisoprolol versus either diuretics, other beta-blockers or calcium channel blockers, no significant differences or critical dependent censoring impact were found. Concurrent use of complementary estimands allowed formulating a refined interpretation of our findings: though not significant, the trend toward an excess of T2DM risk associated with a ACEi/ARB monotherapy compared with bisoprolol is likely not originating only from the minor dependent censoring. Reassessing identical estimands in other cohorts would provide insights to corroborate or refute this result.

Urinary heavy metals and overall survival of advanced high-grade serous ovarian cancer: A nested case-control study in China

BACKGROUND

Environmental pollution has emerged as a significant determinant in ovarian cancer prognosis. However, limited evidence exists regarding the correlations between heavy metals and ovarian cancer prognosis.

OBJECTIVE

To elucidate the relationship between urinary heavy metals and their mixtures with overall survival (OS) of advanced high-grade serous ovarian cancer (HGSOC).

METHODS

Within the Ovarian Cancer Follow-Up Study, we conducted a nested case-control study. A sum of 159 deceased patients and an equal number of alive patients were included, matched by sample date, body mass index, and age at diagnosis. Urinary concentrations of five heavy metals were quantified: arsenic (As), cadmium (Cd), chromium (Cr), mercury (Hg), and lead (Pb). Conditional logistic regression models were employed to calculate odds ratios (ORs) and their 95 % confidence intervals (CIs). To elucidate joint effects, we utilized quantile g-computation and Bayesian kernel machine regression models.

RESULTS

For the multivariable adjusted conditional logistic regression model, significant associations were found between high urinary levels of As (OR=1.99, 95 %CI: 1.05-3.79), Cd (OR=2.56, 95 %CI: 1.29-5.05), Hg (OR=2.24, 95 %CI: 1.09-4.62), and Pb (OR=3.80, 95 %CI: 1.75-8.27) and worse OS of HGSOC, comparing the highest tertile to the lowest. Analysis of joint effects showed that elevated concentrations of heavy metal mixtures were related to poor OS of HGSOC. Pb exhibited the highest contribution to the overall association within the metal mixtures.

CONCLUSIONS

High urinary heavy metal concentrations were linked to worse OS of HGSOC. Future research is necessary to validate our findings.

Association of triglyceride-glucose index and its combination with obesity indicators in predicting the risk of aortic aneurysm and dissection

Background

The association between the triglyceride-glucose (TyG) index and its combination with obesity indictors in aortic aneurysm and dissection (AAD) remains unclear. We aimed to investigate the association between TyG and TyG-body mass index (TyG-BMI), TyG-waist circumference (TyG-WC), TyG-waist height ratio (TyG-WHtR) and AAD risk.

Methods

This study included 387,483 baseline participants from the UK Biobank with complete data on TyG, TyG-BMI, TyG-WC and TyG-WHtR. Cox proportional hazard models evaluated the relationship between these four indicators and the risk of AAD occurrence. Restricted cubic spline (RCS) examined the non-linear relationship between these indicators and AAD risk, while receiver operating characteristic (ROC) curves assessed the predictive value of these four indicators for AAD risk.

Results

Over a median follow-up of 13.7 years, 3,041 AAD events were recorded. Multivariate Cox regression analysis indicated that for each standard deviation increase, the risk of AAD occurrence increased by 33% (HR: 1.33, 95%CI: 1.29-1.38), 25% (HR: 1.25, 95%CI: 1.21-1.29), 61% (HR: 1.61, 95%CI: 1.56-1.66) and 44% (HR: 1.44, 95%CI: 1.39-1.49) for TyG, TyG-BMI, TyG-WC and TyG-WHtR, respectively. RCS demonstrated a linear relationship between these indicators and AAD risk, with TyG-WC demonstrating the best performance in predicting AAD occurrence based on ROC curves.

Conclusion

The present study, based on a large prospective cohort design, showed that higher TyG index and its combination with obesity indices were significantly associated with the risk of AAD. Moreover, AFT models further showed that elevation of these indicators significantly advanced the onset of AAD. In addition, RCS analyses demonstrated a linear association between these indicators and the risk of AAD, and the TyG-WC showed higher predictive ability for AAD. These findings emphasize the potential application of the TyG index and its combination with obesity indicators in the early identification of AAD.

Building a better model: abandon kitchen sink regression

This paper critically examines 'kitchen sink regression', a practice characterised by the manual or automated selection of variables for a multivariable regression model based on p values or model-based information criteria. We highlight the pitfalls of this method, using examples from perinatal/neonatal medicine, and propose more robust alternatives. The concept of directed acyclic graphs (DAGs) is introduced as a tool for describing and analysing causal relationships. We highlight five key issues with 'kitchen sink regression': (1) the disregard for the directionality of variable relationships, (2) the lack of a meaningful causal interpretation of effect estimates from these models, (3) the inflated alpha error rate due to multiple testing, (4) the risk of overfitting and model instability and (5) the disregard for content expertise in model building. We advocate for the use of DAGs to guide variable selection for models that aim to examine associations between a putative risk factor and an outcome and emphasise the need for a more thoughtful and informed use of regression models in medical research.

Comparing Conservative and Early Surgical Treatments for Pyogenic Spondylodiskitis: An International Propensity Score-Matched Retrospective Outcome Analysis

BACKGROUND AND OBJECTIVES

Pyogenic spinal infections pose therapeutic challenges, with the optimal treatment approach remaining contentious. This study aimed to compare outcomes of conservative vs early surgical treatment (SuT) modalities in primary pyogenic spondylodiskitis through an international cohort analysis.

METHODS

A retrospective outcome analysis was conducted of 392 patients from the United Kingdom and Germany, treated between 2017 and 2022 with primary pyogenic spondylodiskitis. Patients were stratified by treatment modality. Propensity score matching, facilitated by a directed acyclic graph and principal component analysis, was used to balance risk factors between the compared patient cohorts. Primary outcome was mortality rate, while secondary outcomes included hospitalization duration, infection relapse, and neurological status at discharge.

RESULTS

The study included 95 patients undergoing conservative treatment, primarily treated in the United Kingdom, and 297 receiving SuT, predominantly in Germany. The mortality rate was notably higher in the conservative group (P < .001), while the surgical cohort experienced a shorter hospitalization duration (P < .01). After propensity score matching, 2 comparable cohorts of 95 patients each emerged. Subsequent analysis revealed a markedly increased mortality in the conservative group (24.2% for conservative treatment vs 4.2% for SuT, P < .001). Neither relapse nor neurological status at discharge showed significant differences between the groups.

CONCLUSION

The study indicates that early surgical intervention may be more effective than conservative management in reducing mortality and hospital stay for patients with primary pyogenic spondylodiskitis. These findings highlight the need for prospective trials and more definitive treatment guidelines.

Measurement error and information bias in causal diagrams: mapping epidemiological concepts and graphical structures

Measurement error and information bias are ubiquitous in epidemiology, yet directed acyclic graphs (DAGs) are infrequently used to represent them, in contrast with confounding and selection bias. This represents a missed opportunity to leverage the full utility of DAGs to depict associations between the variables we actually analyse in practice: empirically measured variables, which are necessarily measured with error. In this article, we focus on applying causal diagrams to depict the data-generating mechanisms that give rise to the data we analyse, including measurement error. We begin by considering empirical data considerations using a general example, and then build up to a specific worked example from the clinical epidemiology of hearing health. Throughout, our goal is to highlight both the challenges and the benefits of using DAGs to depict measurement error. In addition to the application of DAGs to conceptual causal questions (which pertain to unmeasured constructs free from measurement error), which is common, we highlight the advantages associated with applying DAGs to also include empirically measured variables and-potentially-information bias. We also highlight the implications implied by this use of DAGs, particularly regarding the unblocked backdoor path causal structure. Ultimately, we seek to help increase the clarity with which epidemiologists can map traditional epidemiological concepts (such as information bias and confounding) onto causal graphical structures.

Harm effects in non-registered versus registered randomized controlled trials of medications: a retrospective cohort study of clinical trials

BACKGROUND

Trial registration aims to address potential bias from selective or non-reporting of findings, and therefore has a vital role in promoting transparency and accountability of clinical research. In this study, we aim to investigate the influence of trial registration on estimated harm effects in randomized controlled trials of medication interventions.

METHODS

We searched PubMed for systematic reviews and meta-analyses of randomized trials on medication harms indexed between January 1, 2015, and January 1, 2020. To be included in the analyses, eligible meta-analyses should have at least five randomized trials with distinct registration statuses (i.e., prospectively registered, retrospectively registered, and non-registered) and 2 by 2 table data for adverse events for each trial. To control for potential confounding, trials in each meta-analysis were analyzed within confounder-harmonized groups (e.g., dosage) identified using the Directed Acyclic Graph method. The harm estimates arising from the trials with different registration statuses were compared within the confounder-harmonized groups using hierarchical linear regression. Results are shown as ratio of odds ratio (OR) and 95% confidence interval (CI).

RESULTS

The dataset consists of 629 meta-analyses of harms with 10,069 trials. Of these trials, 74.3% were registered, and 23.9% were not registered, and for those registered, 70.6% were prospectively registered, while 26.3% were retrospectively registered. In comparison to prospectively registered trials, both non-registered trials (ratio of OR = 0.82, 95%CI 0.68 to 0.98, P = 0.03) and retrospectively registered trials (ratio of OR = 0.75, 95%CI 0.66 to 0.86, P < 0.01) had lower OR for harms based on 69 and 126 confounders-harmonized groups. The OR of harms did not differ between retrospectively registered and non-registered trials (ratio of OR = 1.02, 95%CI 0.85 to 1.23, P = 0.83) based on 76 confounders-harmonized groups.

CONCLUSIONS

Medication-related harms may be understated in non-registered trials, and there was no obvious evidence that retrospective registration had a demonstrable benefit in reducing such selective or absent reporting. Prospective registration is highly recommended for future trials.

Objective and subjective accounts of urban exposures for epidemiological research on mental health. Measurement and analysis

The interest in the impact of urban environmental exposures (UrbEEs) on mental health has greatly increased in the last two decades. Researchers have tended to measure said exposures either via objective measurement procedures (eg, air pollution campaigns and geographic information systems computations) or by self-reported techniques such as the use of scales and questionnaires. It has been suggested that studying both the objective features of the environments and people's perceptions are key to understand environmental determinants of health and might be needed to tailor effective interventions. However, there is little guidance on how to approach this matter, the comparability between objective and subjective accounts of UrbEEs and, more importantly, suitable statistical procedures to deal with the practicalities of this kind of data. In this essay, we aim to build the case for the joint use of both sets of variables in epidemiological studies and propose socioecological models as a valid theoretical framework to accommodate these. In the methodological sphere, we will also review current literature to select examples of (un)appropriate subjective accounts of urban exposures and propose a series of statistical procedures to estimate the total, direct and indirect effects of UrbEEs on mental health and the potential associations between objective and subjective UrbEEs accounts.

Diabetes and all-cause mortality among middle-aged and older adults in China, England, Mexico, rural South Africa, and the United States: A population-based study of longitudinal aging cohorts

Objective

There is a need for comparable worldwide data on the impact of diabetes on mortality. This study assessed diabetes and all-cause mortality among middle-aged and older adults in five countries.

Research Design and Methods

We analyzed adults aged 51 years or older followed between 2010 and 2020 from population-based cohorts in China, England, Mexico, rural South Africa, and the United States. Diabetes was defined by self-report or an elevated diabetes blood-based biomarker meeting the clinical criteria for diabetes. All-cause mortality was assessed through linkages or informant interviews. We used Poisson regression models to estimate mortality rate ratios and mortality rate differences, comparing people with diabetes to those without diabetes. Models were adjusted for age, gender, education, smoking status, body mass index, and economic status.

Results

We included 29,397 individuals, of whom 4,916 (16.7%) died during the study period. The median follow-up time ranged from 4.6 years in South Africa to 8.3 years in China. The adjusted all-cause mortality rate ratios for people with diabetes versus those without diabetes ranged from 1.53 (95% CI: 1.39-1.68) in the United States to 2.02 (95% CI: 1.34-3.06) in Mexico. The adjusted mortality rate differences (per 1,000 person-years) for people with diabetes versus those without diabetes ranged from 11.9 (95% CI: 4.8-18.9) in England to 24.6 (95% CI: 12.2-37.0) in South Africa.

Conclusions

Diabetes was associated with increased all-cause mortality in population-based cohorts across five diverse countries. There is an urgent need to implement clinical and public health interventions to improve diabetes outcomes globally.

Accounting for time-varying exposures and covariates in the relationship between obesity and diabetes: analysis using parametric g-formula

BACKGROUND

Previous studies investigating the association between obesity and diabetes often did not consider the role of time-varying covariates affected by previous obesity status. This study quantified the association between obesity and diabetes using parametric g-formula.

METHODS

We included 8924 participants without diabetes from the Korean Genome and Epidemiology Study-Ansan and Ansung study(2001-2002)-with up to the seventh biennial follow-up data from 2015 to 2016. Obesity status was categorised as normal (body mass index (BMI) <23.5 kg/m2), overweight (23.5-24.9 kg/m2), obese 1 (25.0-27.4 kg/m2) and obese 2 (≥27.5 kg/m2). Hazard ratios (HRs) comparing baseline or time-varying obesity status were estimated using Cox models, whereas risk ratio (RR) was estimated using g-formula.

RESULTS

The Cox model for baseline obesity status demonstrated an increased risk of diabetes in overweight (HR 1.85; 95% CI=1.48-2.31), obese 1 (2.40; 1.97-2.93) and obese 2 (3.65; 2.98-4.47) statuses than that in normal weight status. Obesity as a time-varying exposure with time-varying covariates had HRs of 1.31 (1.07-1.60), 1.55 (1.29-1.86) and 2.58 (2.14-3.12) for overweight, obese 1 and obese 2 statuses. Parametric g-formula comparing if everyone had been in each obesity category versus normal over 15 years showed increased associations of RRs of 1.37 (1.34-1.40), 1.78 (1.76-1.80) and 2.42 (2.34-2.50).

CONCLUSIONS

Higher BMI classification category was associated with increased risk of diabetes after accounting for time-varying covariates using g-formula. The results from g-formula were smaller than when considering baseline obesity status only but comparable with the results from time-varying Cox model.

Pre-pregnancy BMI, gestational weight gain, and telomere length in amniotic fluid: a causal graph analysis

Previous investigations have suggested a potential association between pre-pregnancy body mass index (BMI) and gestational weight gain (GWG) with telomere length (TL) in various tissues of pregnant women and newborns. Nonetheless, as association does not imply causation, our objective was to investigate the causal connections among pre-pregnancy BMI, GWG, and TL in amniotic fluid. The analysis included 136 mother-child pairs from the Mamma & Bambino cohort, and three causal graph models were developed to depict the interconnections between pre-pregnancy BMI, GWG, and TL. Causal graph analysis was conducted utilizing the do-operator to estimate the causal effect of GWG and the controlled direct effect of pregestational BMI. We revealed that transitioning from non-adequate to adequate GWG had a positive impact on the probability of having "long" TL (i.e., a value greater than the population median) in all three models. When considering the effect of pre-pregnancy BMI, the highest probability of "long" TL was observed in normal weight women with adequate GWG. In contrast, the effect of adequate GWG became minimal among overweight women. These results shed light on the potential causality between pre-pregnancy BMI, GWG, and TL in amniotic fluid, emphasizing the importance of appropriate weight management before and during pregnancy for optimal TL outcomes.

Effects and interaction of single nucleotide polymorphisms at the pharmacokinetic/pharmacodynamic site: insights from the Rotterdam study into metformin clinical response and dose titration

Our study investigated the impact of genetic variations on metformin glycemic response in a cohort from the Rotterdam Study, comprising 14,926 individuals followed for up to 27 years. Among 1285 metformin users of European ancestry, using linear mixed models, we analyzed the association of single nucleotide polymorphisms (SNPs) and a Polygenic Risk Score (PRS) with glycemic response, measured by changes in metformin dosage or HbA1c levels. While individual genetic variants showed no significant association, rs622342 on SLC2A1 correlated with increased glycemic response only in metformin monotherapy patients (β = -2.09, P-value < 0.001). The collective effect of variants, as represented by PRS, weakly correlated with changes in metformin dosage (β = 0.023, P-value = 0.027). Synergistic interaction was observed between rs7124355 and rs8192675. Our findings suggest that while higher PRS correlates with increased metformin dosage, its modest effect size limits clinical utility, emphasizing the need for future research in diverse populations to refine genetic risk models.

Even worse for Black girls: the longitudinal association of racial bullying with the initiation of alcohol and tobacco use

We used Poisson's linear regression to examine the association between racial bullying (RB) and the initiation of alcohol and tobacco uses after 9 months. Two cluster-randomized controlled trials were conducted in 2019 with children in grades five (girls: 50.0%; 10 years old: 82.0%; White: 36.8%; Black: 58.7%; others: 4.5%) and seven (girls: 49.5%; 12 years old: 78.1%; White: 33.2%; Black: 60.4%; others: 6.4%) from 30 public schools in the municipality of São Paulo, Brazil. We restricted our analyses to 2 subsets of students in each grade: those who reported no lifetime alcohol use at baseline and those who reported no lifetime baseline tobacco use. At baseline, 16.2% of fifth and 10.7% of seventh graders reported suffering from RB in the 30 days before data collection. After 9 months, 14.9% of fifth graders started using alcohol and 2.5%, tobacco. Among seventh graders, the figures were 31.2% and 7.7%, respectively. RB predicted the initiation of use of alcohol (risk ratio [RR] = 1.36; 95% CI, 1.07-1.70) and tobacco (RR = 1.81; 95% CI, 1.14-2.76) among seventh graders, with race-gender differences, particularly in Black girls (alcohol: RR = 1.45; 95% CI, 1.07-1.93; tobacco: RR = 2.34; 95% CI, 1.31-3.99). School-based programs and policies must explicitly address issues related to racism and gender in alcohol and tobacco prevention strategies.

Exploring the pathways linking prenatal and early childhood greenness exposure to attention-deficit/hyperactivity disorder symptoms during childhood: An approach based on robust causal inference

BACKGROUND

Epidemiological studies suggest that exposure to greenness during childhood may protect children from developing attention-deficit hyperactivity disorder (ADHD).

OBJECTIVE

We analyzed the effect of both prenatal (pregnancy) and early childhood (4-5-year follow-up) residential greenness exposure and green space availability on ADHD symptoms during childhood (up to the age of 12 years) and further explored the potential mediating role of PM2.5 and physical activity in the association.

METHODS

The study population included participants from the INfancia y Medio Ambiente (INMA) prospective birth cohort (Gipuzkoa, Sabadell, and Valencia). Average Normalized Difference Vegetation Index (NDVI) in buffers of 100-, 300- and 500-m around the residential addresses was used as an indicator of greenness, while green space availability was determined based on the presence of a major green space within 150-m from the residence. Childhood ADHD symptoms were assessed at the 6-8- and 10-12-year follow-ups using Conners Parents Rating Scale-Revised: Short Form.

RESULTS

Although no association was found for the prenatal exposure period, increased early childhood NDVI inversely associated with the OR of clinically significant ADHD symptoms during the 6-8-year follow-up at the 100-m (OR 0.03, 95% CI: 0.003 to 0.44), 300-m (OR 0.04, 95% CI: 0.003 to 0.42) and 500-m (OR 0.08, 95% CI: 0.01 to 0.76) buffers, but exclusively in the context of direct effects. Additionally, the 10-12-year follow-up analysis found moderate to weak evidence of potential total and direct effects of NDVI at both 100- and 300-m buffers on inattention scores, as well as for NDVI at the 300-m buffer on ADHD index scores. The analysis did not reveal evidence of mediation through PM2.5 or physical activity.

CONCLUSIONS

The evidence suggests that early childhood greenness exposure may reduce the risk of developing ADHD symptoms later in childhood, and that this association is not mediated through PM2.5 and physical activity.

Plant-based diet indices and their interaction with ambient air pollution on the ovarian cancer survival: A prospective cohort study

BACKGROUND

Ambient air pollution might serve as a prognostic factor for ovarian cancer (OC) survival, yet the relationships between plant-based diet indices (PDIs) and OC survival remain unclear. We aimed to investigate the associations of comprehensive air pollution and PDIs with OC survival and explored the effects of air pollution-diet interactions.

METHODS

The present study encompassed 658 patients diagnosed with OC. The overall plant-based diet index (PDI), the healthful PDI (hPDI), and the unhealthful PDI (uPDI) were evaluated by a self-reported validated food frequency questionnaire. In addition, an air pollution score (APS) was formulated by summing the concentrations of particulate matter with a diameter of 2.5 microns or less, ozone, and nitrogen dioxide. Cox proportional hazard models were applied to calculate hazard ratios (HRs) and 95 % confidence intervals (CIs). The potential interactions of APS with PDIs in relation to overall survival (OS) were assessed on both multiplicative and additive scales.

RESULTS

Throughout a median follow-up of 37.60 (interquartile: 24.77-50.70) months, 123 deaths were confirmed. Comparing to the lowest tertiles, highest uPDI was associated with lower OS of OC (HR = 2.06, 95 % CI = 1.30, 3.28; P-trend < 0.01), whereas no significant associations were found between either overall PDI or hPDI and OC survival. Higher APS (HR for per interquartile range = 1.27, 95 % CI = 1.01, 1.60) was significantly associated with worse OC survival, and the association was exacerbated by adherence to uPDI. Notably, an additive interaction was identified between combined air pollution and uPDI (P < 0.005 for high APS and high uPDI). We also found that adherence to overall PDI aggravated associations of air pollution with OC survival (P-interaction = 0.006).

CONCLUSIONS

Joint exposure to various ambient air pollutants was significantly associated with lower survival among patients with OC, particularly for those who predominantly consumed unhealthy plant-based foods.

Prehospital interventions and outcomes in traumatic cardiac arrest: a population-based cohort study using the Danish Helicopter Emergency Medical Services data

BACKGROUND AND IMPORTANCE

Traumatic cardiac arrest is associated with poor prognosis, and timely evidence-based treatment is paramount for increasing survival rates. Physician-staffed helicopter emergency medical service use in major trauma has demonstrated improved outcomes. However, the sparsity of data highlights the necessity for a comprehensive understanding of the epidemiology of traumatic cardiac arrest.

OBJECTIVES

The primary objective of the present study was to evaluate survival and return of spontaneous circulation (ROSC) and to investigate the characteristics of patients with traumatic cardiac arrest assessed by the Danish HEMS.

DESIGN

This was a population-based cohort study based on data from the Danish helicopter emergency medical service database.

SETTINGS AND PARTICIPANTS

The study included all patients assessed by the Danish helicopter emergency medical services between 2016 and 2021.

OUTCOME MEASURES AND ANALYSIS

Data were analysed using descriptive statistics, non-parametric testing and logistic regression analyses. Descriptive analysis of prehospital interventions included cardiopulmonary resuscitation, defibrillation, airway management, administration of blood products, and thoracic decompression. The primary outcome was 30-day survival, and the key secondary outcome was prehospital ROSC.

MAIN RESULTS

A total of 223 patients with TCA were included. The median age was 54 years (IQR 34-68), and the majority were males. Overall, 23% of patients achieved prehospital ROSC, and the 30-day survival rate was 4%. Factors associated with an increased likelihood of ROSC were an initial shockable cardiac rhythm, odds ratio (OR) of 3.78 (95% CI 1.33-11.00) and endotracheal intubation, OR 7.10 (95% CI 2.55-22.85).

CONCLUSION

This study highlights the low survival rates observed among patients with traumatic cardiac arrest assessed by helicopter emergency medical services. The findings support the positive impact of an initial shockable cardiac rhythm and endotracheal intubation in improving the likelihood of ROSC. The study contributes to the limited literature on traumatic cardiac arrests assessed by physician-staffed helicopter emergency services. Finally, the findings emphasise the need for further research to understand and improve outcomes in this subgroup of cardiac arrest.

Anti-IL-4R versus anti-IL-5/5R after anti-IL-5/5R failure in asthma: An emulated target trial

BACKGROUND

Switching biologics is now common practice in severe eosinophilic asthma. After insufficient response to anti-IL-5 or 5 receptor (anti-IL-5/5R), the optimal switch between an anti-IL-4R mAb (interclass) or another anti-IL-5/5R drug (intraclass) remains unknown.

OBJECTIVE

We sought to compare the effectiveness of these 2 strategies in asthma control in patients with severe eosinophilic asthma and insufficient response to an anti-IL-5/5R mAb.

METHODS

We emulated a target randomized trial using observational data from the Recherche sur les AsthMes SEvèreS (RAMSES) cohort. Eligible patients were switched to an anti-IL-4R mAb or another anti-IL-5/5R drug after insufficient response to an anti-IL-5/5R mAb. The primary outcome was the change in Asthma Control Test score at 6 months.

RESULTS

Among the 2046 patients in the cohort, 151 were included in the study: 103 switched to an anti-IL-4R mAb and 48 to another anti-IL-5/5R. At 6 months, the difference in Asthma Control Test score improvement was not statistically significant (mean difference groups, 0.82 [-0.47 to 2.10], P = .213). The interclass group exhibited greater cumulative reduction in oral corticosteroid dose (Pinter-intra, -1.05 g [-1.76 to -0.34], P = .041). The interclass group had a better effect, although not significantly, on reducing exacerbations (Δinter-intra, -0.37 [-0.77 to 0.02], P = .124) and increasing lung function (FEV1) (126.8 mL [-12.7 to 266.4], P = .124).

CONCLUSIONS

After anti-IL-5/5R mAb insufficient response, switching to dupilumab demonstrated similar improvement in Asthma Control Test scores compared with intraclass switching. However, it appeared more effective in reducing oral corticosteroid use. Larger studies are warranted to confirm these results.

Co-morbid sarcopenia and low bone mineral density in young paediatric cancer survivors

BACKGROUND

Sarcopenia and low areal bone mineral density (aBMD) are prevalent musculoskeletal complications after paediatric cancer treatment. However, their relationship has not been examined in young paediatric cancers survivors. This study aimed to evaluate aBMD differences according to sarcopenia status and the risk of low aBMD Z-score in young paediatric cancer survivors with sarcopenia confirmed/probable.

METHODS

This cross-sectional study included 116 paediatric cancer survivors (12.1 ± 3.3 years old; 42.2% female). Handgrip strength was used to assessed muscle strength. Dual-energy X-ray absorptiometry estimated aBMD (g/cm2) and appendicular lean mass index (ALMI, kg/m2). 'No sarcopenia' was defined when muscle strength was >decile 2. 'Sarcopenia probable' was defined when muscle strength was ≤ decile 2 and ALMI Z-score was > -1.5 standard deviation (SD). 'Sarcopenia confirmed' was defined when muscle strength was ≤ decile 2 and ALMI Z-score ≤ -1.5 SD. Analysis of covariance and logistic regression, adjusted for time from treatment completion, radiotherapy exposure, calcium intake, and physical activity, was used to evaluate aBMD and estimate the odds ratios (ORs) of low aBMD (aBMD Z-score < -1.0).

RESULTS

Survivors with sarcopenia confirmed had significantly lower aBMD than those without sarcopenia at total body (-1.2 [95% CI: -1.5 to -0.8] vs. 0.2 [-0.2 to 0.6], P < 0.001), lumbar spine (-0.7 [-1.1 to -0.3] vs. 0.4 [0.0 to 0.8], P < 0.001), total hip (-0.5 [-0.9 to -0.2] vs. 0.4 [0.1 to 0.8], P < 0.001), and femoral neck (-1.0 [-1.4 to -0.6] vs. 0.1 [-0.3 to 0.4], P = 0.001). Compared with survivors with sarcopenia probable, survivors with sarcopenia confirmed had significantly lower aBMD Z-score at total body (-1.2 [-1.5 to -0.8] vs. -0.2 [-0.7 to 0.4], P = 0.009), total hip (-0.5 [-0.9 to -0.2] vs. 0.5 [-0.1 to 1.0], P = 0.010), and femoral neck (-1.0 [-1.4 to -0.6] vs. 0.1 [-0.5 to 0.7], P = 0.014). Survivors with sarcopenia confirmed were at higher risk of low aBMD Z-score at the total body (OR: 6.91, 95% CI: 2.31-24.15), total hip (OR: 2.98, 1.02-9.54), and femoral neck (OR: 4.72, 1.72-14.19), than those without sarcopenia. Survivors with sarcopenia probable were at higher risk of low aBMD Z-score at the total body (OR: 4.13, 1.04-17.60) than those without sarcopenia.

CONCLUSIONS

Young paediatric cancer survivors with sarcopenia present higher risk of low aBMD. Resistance training-based interventions designed to mitigate osteosarcopenia in this population should be implemented at early stages.

Sex-specific association between maternal mild anemia and children's behavioral development: a birth cohort study

There has been limited research on maternal anemia affecting children's behavioral development, with a lack of studies focusing on sex differences in this association. Based on the Ma'anshan Birth Cohort, 2132 mother-child pairs were included. Maternal anemia was evaluated based on the hemoglobin concentration and children's behavioral development was assessed by Achenbach Child Behavior Checklist 1.5-5. Binary logistic regression models indicated that compared with children born of mothers without anemia throughout pregnancy, maternal mild anemia during pregnancy or only anemia in the 3rd trimester was associated with increased risks of aggressive behaviors in boys. Maternal mild anemia only in the 2nd trimester was associated with increased risks of attention problems in boys. In girls, maternal mild anemia during pregnancy was associated with increased risks of withdrawn, internalizing problems and total problems. Girls born of mothers with mild anemia only in the 2nd trimester had higher risks of total problems. Maternal mild anemia in both 2nd and 3rd trimesters was associated with increased risks of internalizing problems in girls. Our study identified sex-specific effects of maternal mild anemia during pregnancy on children's behavioral development problems.

Association between pre-diagnosis and post-diagnosis Alternate Mediterranean Diet and ovarian cancer survival: evidence from a prospective cohort study

BACKGROUND

There is currently a lack of comprehensive evidence regarding the correlation between Alternate Mediterranean Diet (AMED) and the survival of patients with ovarian cancer (OC). This prospective cohort study first assessed the association of AMED, not only pre-diagnosis and post-diagnosis but also the change from pre-diagnosis to post-diagnosis with OC survival.

METHODS

A total of 560 OC patients were included in the study, and their dietary intake was assessed using a reliable 111-item food frequency questionnaire. The overall survival (OS) of the patients was monitored through active follow-up and review of medical records until February 16th, 2023. Cox proportional hazard regression models were utilized to compute the hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs).

RESULTS

Out of the total 560 patients with OC, 211 (37.68%) succumbed during a median follow-up period of 44.40 months (interquartile range: 26.97-61.37). Comparative analysis indicated a significant association between the highest tertiles of pre-diagnosis (HR = 0.59; 95% CI 0.38-0.90; Ptrend < 0.05) and post-diagnosis (HR = 0.61; 95% CI 0.41-0.91; Ptrend < 0.05) AMED intake and improved OS as opposed to the lowest tertile. Additionally, a significant linear trend was observed for AMED and OC survival. Notably, decreased intake (more than 5% change) and significantly increased intake (more than 15% change) of AMED from pre-diagnosis to post-diagnosis were linked to worse and better OS, respectively, when compared to the stable intake group (change within 5%). Furthermore, patients displaying consistently higher AMED intake both before and after diagnosis experienced enhanced OS in comparison to those with consistently low AMED intake (HRHigh-High vs. Low-Low = 0.47; 95% CI 0.31-0.70).

CONCLUSION

High pre-diagnosis and post-diagnosis AMED was associated with an improved OS in patients with OC, suggesting that maintaining a consistently high intake of AMED could potentially benefit the prognosis of OC.

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.

Associations of traditional healthy lifestyle and sleep quality with metabolic dysfunction-associated fatty liver disease: two population-based studies

AIMS

To examine whether an extended lifestyle metrics incorporating sleep quality improves risk stratification for metabolic dysfunction-associated fatty liver disease (MAFLD), at-risk metabolic dysfunction-associated steatohepatitis (MASH) and significant fibrosis.

METHODS

A total of 5011 participants with abdominal ultrasound from Imaging sub-cohort of South China Cohort (ISSCC) and 3672 participants underwent vibration controlled transient elastography from US National Health and Nutrition Examination Survey (US NHANES) were included. Liver Essential 5 was constructed by incorporating sleep quality into traditional healthy lifestyles (HLS).

RESULTS

A total of 4.66-17.72% of the association between traditional HLS and MAFLD was mediated by sleep quality regardless of the detection techniques, and their joint associations on MAFLD were significant in both cohorts. ORs for individuals with poor sleep and unfavorable HLS were 1.72 (1.29-2.30) in ISSCC and 2.25 (1.55-3.26) in US NHANES, respectively. Around half of the participants previously considered as following a favorable HLS were re-classified by Liver Essential 5 with significantly higher prevalences of MAFLD in both cohorts (P < 0.001). Similar results were also found on at-risk MASH and significant fibrosis in US NHANES. ORs of participants with per one increment increase in Liver Essential 5 were 0.82 (0.77-0.89) and 0.79 (0.70-0.88) for MAFLD in ISSCC and US NHANES, 0.62 (0.48-0.78) for at-risk MASH and 0.78 (0.65-0.93) for significant fibrosis.

CONCLUSIONS

Liver Essential 5, which incorporates sleep quality and traditional lifestyle factors, provides additional risk stratification for MAFLD-related outcomes.

The association of dietary total antioxidant capacity and gestational diabetes: a prospective cohort study from the Mothers and their children's health (MATCH)

BACKGROUND/OBJECTIVES

There is evidence to support the hypothesis that a diet rich in antioxidants can help safeguard against the development of gestational diabetes mellitus (GDM). This study aimed to investigate the association between dietary total antioxidant capacity (DTAC) during early pregnancy and the risk of GDM.

SUBJECTS/METHODS

We included 1856 pregnant women in their first trimester from the Mothers and their Children's Health (MATCH) prospective cohort study. Prepregnancy dietary intake was assessed using a validated food frequency questionnaire (FFQ) and was used to calculate the DTAC score. Incident GDM was diagnosed based on the American Diabetes Association criteria. We estimated the association between DTAC and GDM using propensity score-based inverse probability weighting (IPW).

RESULTS

Overall, 369 (14.6%) of the pregnant women were identified with GDM. The mean DTAC score and the corresponding standard deviation (SD) was 2.82± (2.56) mmol/100 g, with a range of 0.01 to 18.55. The adjusted risk of GDM decreased by 34% (95% CI = 10%, 52%, p = 0.023) for each DTAC score increase. The results showed that women in the highest quartile of DTAC had a lower risk of developing GDM compared to those in the lowest quartile (adjusted RR: 0.29, 95% CI: 0.12, 0.68, p = 0.005).

CONCLUSION

DTAC in early pregnancy is significantly associated with a lower risk of GDM. Additional larger cohort studies are needed to validate these findings.

Relationship between paternal excessive weight and neonatal anthropometry in a clinical trial of nutritional counseling for pregnant women with overweight

BACKGROUND/OBJECTIVES

Human studies suggest that fathers with obesity influence infant growth and development. This study aimed to evaluate the relationship between paternal body mass index (BMI) and waist circumference (WC) with neonatal anthropometry and adiposity.

METHODS

This study is a cohort nested in a randomized controlled clinical trial of nutritional counseling for pregnant women with overweight. In total, 89 partner-pregnant woman-neonate triads were included. Paternal anthropometric measurements were taken at the time of the interview. Secondary data related to birth were obtained through access to the health information systems. Neonatal skinfold thickness was assessed and the adiposity was estimated using a predictive anthropometric model. Pearson's correlation and adjusted multivariate linear regression models were employed to evaluate the relationship between paternal BMI and WC with neonatal anthropometric measurements and adiposity.

RESULTS

In total, 57.0% of the fathers presented a BMI ≥ 25 kg/m² and 14.6% a waist circumference ≥102 cm. The mean ± SD birth weight of the newborns (g) was 3357 ± 538. Paternal BMI and WC were inversely correlated with head circumference at birth [r = -0.31 (p = 0.004), r = -0.23 (p = 0.03), respectively]. Paternal BMI was also inversely correlated with the birth weight standardized by gestational age (z-score) [r = -0.23 (p = 0.03)]. In adjusted multivariate linear regression models, the paternal BMI (kg/m²) was inversely associated with the head circumference at birth (cm) [β = -0.07 (95% CI -0.15; -0.001) p = 0.04].

CONCLUSION

The data suggest that paternal excessive weight have a negative effect on fetal development, as assessed by anthropometric measurements. The inverse association between paternal BMI and the head circumference at birth was independent of confounders. Future studies with larger sample sizes are necessary to confirm or refute such hypotheses.

Effect of polypharmacy and potentially inappropriate medications on physical functional decline among older adults with advanced cancer receiving systemic treatment

BACKGROUND

Polypharmacy and potentially inappropriate medications (PIM) are common among older adults with advanced cancer, but their association with physical functional outcomes is understudied. This study aimed to estimate the risk of physical functional decline associated with medication measures in older adults with advanced cancer starting a new line of systemic treatment.

METHODS

This secondary analysis of GAP 70+ Trial (PI: Mohile) enrolled patients aged 70+ with advanced cancer, had ≥ 1 geriatric assessment domain impairment and planned to start a new antineoplastic regimen with a high risk of toxicity. Polypharmacy (concurrent use of ≥ 8 medications (meds)) was assessed before initiation of treatment. PIM were categorized using Screening Tool of Older Person's Prescriptions (STOPP) criteria and 2019 Beers criteria. Physical functional outcomes were assessed within 3 months of treatment initiation: (1) Activity of Daily Living (ADL) decline: 1-point decrease in ADL score between baseline and 3 months; (2) Instrumental ADL (IADL) decline: 1-point decrease in IADL score between baseline and 3 months; (3) Short physical performance battery (SPPB) decline, defined as 1-point decrease on SPPB; (4) ≥ 1 falls within 3 months of treatment. Separate multivariable, cluster-weighted Generalized Estimating Equations models adjusted for relevant covariates (e.g., age, baseline function/comorbidities).

RESULTS

Among 616 participants, mean number of meds was 6 (range 0-24); 28% received ≥ 8 meds. Polypharmacy was associated with increased risk of ADL decline (adjusted risk ratio [aRR], 1.31; 95% CI, 1.00-1.71). Taking ≥ 1 PIM per STOPP was associated with increased risk of IADL decline (aRR, 1.21; 95% CI, 1.04-1.40) and falls (aRR, 1.93; 95% CI, 1.49-2.51).

CONCLUSIONS

In a large cohort of vulnerable older adults with advanced cancer receiving systemic treatment, polypharmacy and PIM were independently associated with an increased risk of physical functional decline. This emphasizes the need to develop interventions to optimize medication use, intending to improve outcomes in these patients.

TRIAL REGISTRATION

ClinicalTrials.gov Identifier: NCT02054741. Registered 01-31-2014.

Adherence, Adverse Events and Viral Control among Children and Adolescents with HIV in Zambia Switched to an Integrase Inhibitor Regimen

Background

Based on recent World Health Organization recommendations, there has been a large-scale transition in Sub-Saharan Africa to integrase inhibitor (II)-based antiretroviral therapy (ART) regimens.

Setting

This study was conducted at an urban referral center in Lusaka, Zambia.

Methods

This study included 297 children and adolescents with HIV (CAWH) on ART for one year prior to enrollment and followed for 1-4 years after enrollment. ART adherence, ART regimen, and viral load were assessed periodically. Structured interviews were conducted with a subset of 95 children to assess adherence barriers and side effects.

Results

Children on protease inhibitor (PI)-based regimens were more likely to report adherence problems than children taking II- or Efavirenz-based regimens (10% vs. 28%, p=0.03) and noted more days with missed doses (median 1 vs. 0, p=0.02). In interviews, the most common reasons given for poor adherence included bad medication taste, not being home when medications were due, and perceived side effects. The PI group was more likely to report that taste was a problem affecting adherence (22% vs. 4%, p=0.05) and headache as an ART side effect (17% vs. 4%, p=0.05). Switching from a PI- to an II-based regimen was associated with improved adherence (72% vs. 92%, p=0.01) and an undetectable viral load (67% vs. 78%).

Conclusions

Switching CAWH from PI-based to II-based regimens has many advantages including superior side effect profiles, adherence, and viral suppression. PI taste aversion may be a significant contributor to pediatric adherence issues. Palatability should be considered in pediatric HIV drug development.

Canadian beach cohort study: protocol of a prospective study to assess the burden of recreational water illness

BACKGROUND

Recreational water activities at beaches are popular among Canadians. However, these activities can increase the risk of recreational water illnesses (RWI) among beachgoers. Few studies have been conducted in Canada to determine the risk of these illnesses. This protocol describes the methodology for a study to determine the risk and burden of RWI due to exposure to fecal pollution at beaches in Canada.

METHODS

This study will use a mixed-methods approach, consisting of a prospective cohort study of beachgoers with embedded qualitative research. The cohort study involves recruiting and enrolling participants at public beaches across Canada, ascertaining their water and sand contact exposure status, then following-up after seven days to determine the incidence of acute RWI outcomes. We will test beach water samples each recruitment day for culture-based E. coli, enterococci using rapid molecular methods, and microbial source tracking biomarkers. The study started in 2023 and will continue to 2025 at beaches in British Columbia, Manitoba, Ontario, and Nova Scotia. The target enrollment is 5000 beachgoers. Multilevel logistic regression models will be fitted to examine the relationships between water and sand contact and RWI among beachgoers. We will also examine differences in risks by beachgoer age, gender, and beach location and the influence of fecal indicator bacteria and other water quality parameters on these relationships. Sensitivity analyses will be conducted to examine the impact of various alternative exposure and outcome definitions on these associations. The qualitative research phase will include focus groups with beachgoers and key informant interviews to provide additional contextual insights into the study findings. The study will use an integrated knowledge translation approach.

DISCUSSION

Initial implementation of the study at two Toronto, Ontario, beaches in 2023 confirmed that recruitment is feasible and that a high completion rate (80%) can be achieved for the follow-up survey. While recall bias could be a concern for the self-reported RWI outcomes, we will examine the impact of this bias in a negative control analysis. Study findings will inform future recreational water quality guidelines, policies, and risk communication strategies in Canada.

Low haemoglobin in arduous seasons is associated with reduced chance of ovulation among women living in the Bolivian altiplano

Background and objectives

Female reproductive function flexibly responds to ecological variation in energy availability, but the roles of other ecologically limited resources, such as iron, remain poorly understood. This analysis investigates whether haemoglobin associates with investment in reproductive function in a rural natural fertility population living in the Bolivian altiplano.

Methodology

We conducted a cross-sectional secondary analysis of prospectively collected biomarker and sociodemographic data, comprising 152 menstrual cycles from 96 non-contracepting women living at 3800 m altitude. Multivariable multilevel models were used to investigate (i) whether haemoglobin concentration is associated with ecological variation in subsistence strategy and seasonal conditions, and (ii) whether haemoglobin concentration is associated with the occurrence of ovulation and/or the concentration of luteal progesterone, two biomarkers of current investment in reproduction.

Results

Haemoglobin concentrations were lower in arduous seasons among those women more dependent on traditional agropastoral subsistence strategies (β = -0.42, 95% CI: -0.80 to -0.04, P = 0.032). During more arduous seasons, a 1 standard deviation increase in haemoglobin was associated with an over 3-fold increase in the odds of ovulation after adjusting for body fat, breastfeeding status, and age (adjusted odds ratio = 3.27, 95% CI: 1.10 to 9.27, P = 0.033).

Conclusions and implications

When conditions are relatively harsh and may be expected to improve, low haemoglobin levels are associated with lower current investment in reproduction and reduced fecundity. These results support the role of iron, independent of energy stores, as a limiting resource in modulating reproductive trade-offs.

Intrafamilial associations of sleep multitrajectory groups between ages of 3 and 60 months in the SEPAGES cohort

OBJECTIVES

We investigated intrafamilial sleep evolution by identifying children's sleep multitrajectory groups between 3- and 60-month of age and their association with parental sleep multitrajectory groups.

METHODS

We included 180 children from the SEPAGES cohort (Grenoble, France) whose parents belonged to previously identified sleep multitrajectory groups, through group-based multitrajectory modeling, between 3 and 36months postpartum, using nighttime (NSD) and weekend daytime (DSD) sleep durations and subjective sleep loss, comprising "No," "Subjective," and "Global" sleep problems groups. Child sleep information (NSD, DSD, subjective sleep loss, night waking, and sleep onset difficulties) was collected by parental questionnaires at 3-, 12-, 36-, and 60-month. We identified sleep multitrajectory groups using group-based multitrajectory modeling in children and examined their associations with parental sleep multitrajectory groups using multinomial logistic regressions.

RESULTS

We identified three sleep multitrajectory groups in children: the "No/few" group (29.4%) had moderate NSD, long DSD, low subjective sleep loss/night waking/sleep onset difficulties prevalence, the "Moderate" group (60.0%) had long NSD and moderate DSD, and medium subjective sleep loss/night waking/sleep onset difficulties prevalence, and the "Global" group (10.6%) had the shortest NSD and DSD, and the highest subjective sleep loss/night waking/sleep onset difficulties prevalence. After adjusting for covariates, mothers in the "Global" group were more likely to have children in the same group, and mothers in "Subjective" and "Global" groups were less likely to have children in the "Moderate" group than in the "No/few" group. No association was identified with paternal or couple sleep multitrajectory groups.

CONCLUSIONS

The observed associations between parent-child sleep multitrajectory groups suggest greater maternal sensitivity to or involvement in the child's sleep than the fathers. Early preventive sleep actions could improve sleep in children and mothers.

Applying causal inference and Bayesian statistics to understanding vaccine safety signals using a simulation study

Community perception of vaccine safety influences vaccine uptake. Our objective was to assess current vaccine safety monitoring by examining factors that may influence the availability of post-vaccination survey data, and thereby the specificity and sensitivity of existing signal detection methods. We used causal directed acyclic graphs (DAGs) and a Bayesian posterior predictive analysis (PPA) signal detection method to understand biological and behavioural factors which may influence signal detection. The DAGs informed the data simulated for scenarios in which these factors were varied. The influence of biological factors such as severity of adverse reactions and behavioural factors such as healthcare-seeking behaviour upon survey participation was found to drive signal detection. Where there was a low prevalence of moderate to severe reactions, false signals were detected when there was a strong influence of reaction severity on both survey participation and seeking medical attention. These findings provide implications for future vaccine safety monitoring.

Geographic Access to Fertility Counseling among Adolescent and Young Adult Women with Cancer in North Carolina

BACKGROUND

Fertility counseling is recommended for adolescent and young adult women facing gonadotoxic cancer therapy. However, fertility care is subspecialized medical care offered at a limited number of institutions, making geographic access a potential barrier to guideline-concordant care. We assessed the relationship between geographic access and receipt of fertility counseling among adolescent and young adult women with cancer.

METHODS

Using data from the North Carolina Central Cancer Registry, we identified women diagnosed with lymphoma, gynecologic cancer, or breast cancer at ages 15 to 39 years during 2004 to 2015. Eligible women were invited to complete an online survey on various topics, including fertility counseling. Geographic access was measured, using geocoded addresses, as vehicular travel time from residence to the nearest fertility clinic available at diagnosis. Multivariable regression models were used to examine the association between travel time and receipt of fertility counseling by provider type: health care provider versus fertility specialist.

RESULTS

Analyses included 380 women. The median travel time to a fertility clinic was 31 (IQR: 17-71) minutes. Overall, 75% received fertility counseling from a health care provider and 16% by a fertility specialist. Women who lived ≥30 minutes from a clinic were 13% less likely to receive fertility counseling by a health care provider (prevalence ratio: 0.87; 95% confidence interval, 0.75-1.00) and 49% less likely to receive counseling by a fertility specialist (prevalence ratio: 0.51; 95% confidence interval, 0.28-0.93).

CONCLUSIONS

Women who lived further away from fertility clinics were less likely to receive fertility counseling.

IMPACT

Interventions to improve access to fertility counseling should include strategies to alleviate the burden of geographic access.

Occupational and Health-Related Risk Factors for Incident and Recurrent Back Disorders in Danish Fishers-A Register-Based Study

BACKGROUND

Fishers are at risk of back disorders due to their physically demanding work. The aim was to investigate risk factors for back disorders in fishers in Denmark.

METHODS

All male Danish registered fishers between 1994 and 2017 were included. ICD-10 codes classified back disorders (M40-M54* and DM99.1-4*). A multistate model on a cause-specific cox regression model was conducted.

RESULTS

Of 13,165 fishers included, 16% had a hospital contact with an incident back disorder, and 52% at least had 1 recurrent episode. Having worked in another occupation (HR 1.14; 95% CI: 1.02, 1.27) and another musculoskeletal disorder (HR 1.84; 95% CI: 1.69, 2.01) were significant risk factors for the incident back disorder. No risk factors were seen for recurrent episodes.

CONCLUSIONS

Risk factors for incident and recurrent back disorders were different; thus, episode-specific initiatives are needed to reduce back disorders among fishers.

Spatial Variability in Relationships between Early Childhood Lead Exposure and Standardized Test Scores in Fourth Grade North Carolina Public School Students (2013-2016)

BACKGROUND

Exposure to lead during childhood is detrimental to children's health. The extent to which the association between lead exposure and elementary school academic outcomes varies across geography is not known.

OBJECTIVE

Estimate associations between blood lead levels (BLLs) and fourth grade standardized test scores in reading and mathematics in North Carolina using models that allow associations between BLL and test scores to vary spatially across communities.

METHODS

We link geocoded, individual-level, standardized test score data for North Carolina public school students in fourth grade (2013-2016) with detailed birth records and blood lead testing data retrieved from the North Carolina childhood blood lead state registry on samples typically collected at 1-6 y of age. BLLs were categorized as: 1 μ g / dL (reference), 2 μ g / dL , 3 - 4 μ g / dL and ≥ 5 μ g / dL . We then fit spatially varying coefficient models that incorporate information sharing (smoothness), across neighboring communities via a Gaussian Markov random field to provide a global estimate of the association between BLL and test scores, as well as census tract-specific estimates (i.e., spatial coefficients). Models adjusted for maternal- and child-level covariates and were fit separately for reading and math.

RESULTS

The average BLL across the 91,706 individuals in the analysis dataset was 2.84 μ g / dL . Individuals were distributed across 2,002 (out of 2,195) census tracts in North Carolina. In models adjusting for child sex, birth weight percentile for gestational age, and Medicaid participation as well as maternal race/ethnicity, educational attainment, marital status, and tobacco use, BLLs of 2 μ g / dL , 3 - 4 μ g / dL and ≥ 5 μ g / dL were associated with overall lower reading test scores of - 0.28 [95% confidence interval (CI): - 0.43 , - 0.12 ], - 0.53 (- 0.69 , - 0.38 ), and - 0.79 (- 0.99 , - 0.604 ), respectively. For BLLs of 1 μ g / dL , 2 μ g / dL , 3 - 4 μ g / dL and ≥ 5 μ g / dL , spatial coefficients-that is, tract-specific adjustments in reading test score relative to the "global" coefficient-ranged from - 9.70 to 2.52, - 3.19 to 3.90, - 11.14 to 7.85, and - 4.73 to 4.33, respectively. Results for mathematics were similar to those for reading.

CONCLUSION

The association between lead exposure and reading and mathematics test scores exhibits considerable heterogeneity across North Carolina communities. These results emphasize the need for prevention and mitigation efforts with respect to lead exposures everywhere, with special attention to locations where the cognitive impact is elevated. https://doi.org/10.1289/EHP13898.

Short term effectiveness of ustekinumab versus vedolizumab in Crohn's disease after failure of anti-TNF agents: An observational comparative study design with a Bayesian analysis

BACKGROUND

Crohn's disease (CD) is a debilitating gastrointestinal disease with complex etiology. Although effective, recipients of anti-tumor necrosis factor (TNF) agents may experience primary or secondary nonresponse, necessitating alternative treatments. This study is intended to compare the short-term effectiveness of ustekinumab and vedolizumab in treating CD after failure of multiple lines of anti-TNF therapy using real-world data.

METHODS

A retrospective study was conducted at a tertiary hospital in Dammam, Saudi Arabia, including adults (≥18 years old) with CD who did not respond to anti-TNF therapy. Primary endpoints were clinical improvement per the Harvey-Bradshaw Index (HBI) scores and remission at 12 weeks on an ordinal outcome scale. Secondary endpoints included clinical, biochemical, and endoscopic remission; clinical response; corticosteroid-free days; and cumulative steroid dose. Proportional odds and logistic regression Bayesian models were used to analyze outcomes, and the probability of treatment effectiveness was calculated from the posterior distribution.

RESULTS

The study included 101 patients (ustekinumab, n = 71 and vedolizumab, n = 30) with a median age of 32 years (IQR: 26.0-38.0); 54.4% were male. At 12 weeks, the HBI endpoint showed an adjusted odds ratio (aOR) = 0.60 (95% confidence interval [CI]: 0.25-1.31), favoring ustekinumab, with a 75% probability of treatment effectiveness over vedolizumab. The clinical ordinal scale had an aOR = 0.61 (95% CI: 0.26-1.35) with a 73% probability of effectiveness for ustekinumab. Ustekinumab was also associated with favorable outcomes in secondary endpoints, reaching up to a 90% probability of effectiveness.

CONCLUSION

In CD patients with anti-TNF failure, ustekinumab was more effective than vedolizumab in the short term. These real-world insights contribute to understanding CD management but require validation in larger prospective studies and randomized controlled trials.

The association of body composition phenotypes before chemotherapy with epithelial ovarian cancer mortality

BACKGROUND

The association of body composition with epithelial ovarian carcinoma (EOC) mortality is poorly understood. To date, evidence suggests that high adiposity is associated with decreased mortality (an obesity paradox), but the impact of muscle on this association has not been investigated. Herein, we define associations of muscle and adiposity joint-exposure body composition phenotypes with EOC mortality.

METHODS

Body composition from 500 women in the Body Composition and Epithelial Ovarian Cancer Survival Study was dichotomized as normal or low skeletal muscle index (SMI), a proxy for sarcopenia, and high or low adiposity. Four phenotypes were classified as fit (normal SMI and low adiposity; reference; 16.2%), overweight or obese (normal SMI and high adiposity; 51.2%), sarcopenia and overweight or obese (low SMI and high adiposity; 15.6%), and sarcopenia or cachexia (low SMI and low adiposity; 17%). We used multivariable Cox models to estimate associations of each phenotype with mortality for EOC overall and high-grade serous ovarian carcinoma (HGSOC).

RESULTS

Overweight or obesity was associated with up to 51% and 104% increased mortality in EOC and HGSOC [Hazard Ratio (HR)] = 1.51, 95% CI = 1.05 to 2.19 and HR = 2.04, 95% CI = 1.29 to 3.21). Sarcopenia and overweight or obesity was associated with up to 66% and 67% increased mortality in EOC and HGSOC (HR = 1.66, 95% CI = 1.13 to 2.45 and HR = 1.67, 95% CI = 1.05 to 2.68). Sarcopenia or cachexia was associated with up to 73% and 109% increased mortality in EOC and HGSOC (HR = 1.73, 95% CI = 1.14 to 2.63 and HR = 2.09, 95% CI = 1.25 to 3.50).

CONCLUSIONS

Overweight or obesity, sarcopenia and overweight or obesity, and sarcopenia or cachexia phenotypes were each associated with increased mortality in EOC and HGSOC. Exercise and dietary interventions could be leveraged as ancillary treatment strategies for improving outcomes in the most fatal gynecological malignancy with no previously established modifiable prognostic factors.

Synthesizing Subject-matter Expertise for Variable Selection in Causal Effect Estimation: A Case Study

BACKGROUND

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

METHODS

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

RESULTS

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

CONCLUSION

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

Associations of Placental Inflammation and Oxidative Stress Biomarkers with Glucolipid Metabolism in Children: A Birth Cohort Study in China

BACKGROUND

The maternal intrauterine immune environment may affect offspring long-term health. We aimed to investigate the association between the intrauterine placental immunological milieu and glycolipid metabolic health in children.

METHODS AND RESULTS

This study enrolled 1803 mother-child pairs from the Ma'anshan birth cohort (2013-2014). Placental mRNA expression of inflammatory cytokines (interleukin-1β [IL-1β], IL-10, monocyte chemoattractant protein-1, tumor necrosis factor-α, IL-4, IL-6, IL-8, C-reactive protein, and interferon-γ) and oxidative stress biomarkers (heme oxygenase-1, hypoxia-inducible factor-1alpha, and glucose-related protein 78) was quantified using real-time quantitative polymerase chain reaction. Fasting blood glucose, insulin, triglycerides, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and total cholesterol were assessed at 5 to 6 years old. Statistical analyses included multiple linear regression, binary logistic regression, restricted cubic spline model, and the Bayesian kernel machine regression model. Placental inflammatory cytokines (IL-1β, monocyte chemoattractant protein-1, C-reactive protein, IL-6, IL-8, IL-10) and oxidative stress biomarkers (heme oxygenase-1, hypoxia-inducible factor-1alpha, glucose-related protein 78) showed positive associations with children's fasting blood glucose levels. Heme oxygenase-1 and glucose-related protein 78 exhibited negative correlations with children's fasting insulin levels. Elevated IL-6, heme oxygenase-1, hypoxia-inducible factor-1alpha, and glucose-related protein 78 were associated with increased risk of prediabetes in children. Overall upregulation of placental proinflammatory cytokines and oxidative stress factors mRNA expression correlated with higher prediabetes risk in children. Bayesian kernel machine regression analysis indicated a joint positive effect of the 12 placental inflammation and oxidative stress mixtures on children's risk of high fasting blood glucose.

CONCLUSIONS

This exploratory study underscores significant correlations between maternal intrauterine placental inflammation, oxidative stress markers, and offspring fasting blood glucose and insulin levels. These findings highlight the potential role of intrauterine holistic immunity in shaping offspring glucose metabolism health.

Older adults' compliance with mobile ecological momentary assessments in behavioral nutrition and physical activity research: pooled results of four intensive longitudinal studies and recommendations for future research

BACKGROUND

Mobile Ecological Momentary Assessment (EMA) is increasingly used to gather intensive, longitudinal data on behavioral nutrition, physical activity and sedentary behavior and their underlying determinants. However, a relevant concern is the risk of non-random non-compliance with mobile EMA protocols, especially in older adults. This study aimed to examine older adults' compliance with mobile EMA in health behavior studies according to participant characteristics, and prompt timing, and to provide recommendations for future EMA research.

METHODS

Data of four intensive longitudinal observational studies employing mobile EMA to understand health behavior, involving 271 community-dwelling older adults (M = 71.8 years, SD = 6.8; 52% female) in Flanders, were pooled. EMA questionnaires were prompted by a smartphone application during specific time slots or events. Data on compliance (i.e. information whether a participant answered at least one item following the prompt), time slot (morning, afternoon or evening) and day (week or weekend day) of each prompt were extracted from the EMA applications. Participant characteristics, including demographics, body mass index, and smartphone ownership, were collected via self-report. Descriptive statistics of compliance were computed, and logistic mixed models were run to examine inter- and intrapersonal variability in compliance.

RESULTS

EMA compliance averaged 77.5%, varying from 70.0 to 86.1% across studies. Compliance differed among subgroups and throughout the day. Age was associated with lower compliance (OR = 0.96, 95%CI = 0.93-0.99), while marital/cohabiting status and smartphone ownership were associated with higher compliance (OR = 1.83, 95%CI = 1.21-2.77, and OR = 4.43, 95%CI = 2.22-8.83, respectively). Compliance was lower in the evening than in the morning (OR = 0.82, 95%CI = 0.69-0.97), indicating non-random patterns that could impact study validity.

CONCLUSIONS

The findings of this study shed light on the complexities surrounding compliance with mobile EMA protocols among older adults in health behavior studies. Our analysis revealed that non-compliance within our pooled dataset was not completely random. This non-randomness could introduce bias into study findings, potentially compromising the validity of research findings. To address these challenges, we recommend adopting tailored approaches that take into account individual characteristics and temporal dynamics. Additionally, the utilization of Directed Acyclic Graphs, and advanced statistical techniques can help mitigate the impact of non-compliance on study validity.

Association between pre- and post-diagnosis healthy eating index 2020 and ovarian cancer survival: evidence from a prospective cohort study

Background: Previous studies on the association between diet quality and ovarian cancer (OC) survival are limited and inconsistent. We evaluated the relationship between pre- and post-diagnosis diet quality based on the Healthy Eating Index-2020 (HEI-2020), as well as their changes and OC survival. Methods: This prospective cohort study involved 1082 patients with OC aged 18-79 years, enrolled between 2015 and 2022. Detailed dietary intake before and after diagnosis was recorded using a validated food frequency questionnaire. Deaths were ascertained until February 16th, 2023 via medical records and active follow-up. Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CI). Results: We included 549 OC cases with a median follow-up of 44.9 months, representing 206 total deaths. Higher HEI scores were associated with better OS (pre-diagnosis: HRT3 vs. T1 0.66, 95%CI: 0.46-0.93, HR1-SD 0.84, 95%CI: 0.73-0.96; post-diagnosis: HRT3 vs. T1 0.68, 95%CI: 0.49-0.96, HR1-SD 0.80, 95%CI: 0.69-0.92). Compared to the stable group, the group with decreased HEI scores (>3%) from pre- to post-diagnosis had worse OS (HR 1.93, 95%CI: 1.26-2.97). Conclusion: High pre- and post-diagnosis diet quality was associated with improved OC survival, whereas deterioration in diet quality after diagnosis was associated with decreased OC survival.