Applications of Bayesian Methods to Epidemiologic Research

Exogenous female sex steroid hormones and new-onset asthma in women: a matched case-control study

BACKGROUND

Evidence on the role of exogenous female sex steroid hormones in asthma development in women remains conflicting. We sought to quantify the potential causal role of hormonal contraceptives and menopausal hormone therapy (MHT) in the development of asthma in women.

METHODS

We conducted a matched case-control study based on the West Sweden Asthma Study, nested in a representative cohort of 15,003 women aged 16-75 years, with 8-year follow-up (2008-2016). Data were analyzed using Frequentist and Bayesian conditional logistic regression models.

RESULTS

We included 114 cases and 717 controls. In Frequentist analysis, the odds ratio (OR) for new-onset asthma with ever use of hormonal contraceptives was 2.13 (95% confidence interval [CI] 1.03-4.38). Subgroup analyses showed that the OR increased consistently with older baseline age. The OR for new-onset asthma with ever MHT use among menopausal women was 1.17 (95% CI 0.49-2.82). In Bayesian analysis, the ORs for ever use of hormonal contraceptives and MHT were, respectively, 1.11 (95% posterior interval [PI] 0.79-1.55) and 1.18 (95% PI 0.92-1.52). The respective probability of each OR being larger than 1 was 72.3% and 90.6%.

CONCLUSIONS

Although use of hormonal contraceptives was associated with an increased risk of asthma, this may be explained by selection of women by baseline asthma status, given the upward trend in the effect estimate with older age. This indicates that use of hormonal contraceptives may in fact decrease asthma risk in women. Use of MHT may increase asthma risk in menopausal women.

Associations of Gestational Perfluoroalkyl Substances Exposure with Early Childhood BMI z-Scores and Risk of Overweight/Obesity: Results from the ECHO Cohorts

BACKGROUND

Gestational per- and polyfluoroalkyl substances (PFAS) exposure may be associated with adiposity and increased risk of obesity among children and adolescents. However, results from epidemiological studies evaluating these associations are inconsistent.

OBJECTIVES

We estimated the associations of pregnancy PFAS concentrations with child body mass index (BMI) z -scores and risk of overweight/obesity in eight U.S. cohorts.

METHODS

We used data from 1,391 mother-child pairs who enrolled in eight Environmental influences on Child Health Outcomes (ECHO) cohorts (enrolled: 1999-2019). We quantified concentrations of seven PFAS in maternal plasma or serum in pregnancy. We measured child weight and height between the ages of 2 and 5 y and calculated age- and sex-specific BMI z -scores; 19.6% children had more than one BMI measurement. We estimated covariate-adjusted associations of individual PFAS and their mixture with child BMI z -scores and risk of overweight/obesity using linear mixed models, modified Poisson regression models, and Bayesian approaches for mixtures. We explored whether child sex modified these associations.

RESULTS

We observed a pattern of subtle positive associations of PFAS concentrations in pregnancy with BMI z -scores and risk of overweight/obesity. For instance, each doubling in perfluorohexane sulfonic acid concentrations was associated with higher BMI z -scores (β = 0.07 ; 95% CI: 0.01, 0.12). Each doubling in perfluroundecanoic acid [relative risk  ( RR ) = 1.10 ; 95% CI: 1.04, 1.16] and N -methyl perfluorooctane sulfonamido acetic acid (RR = 1.06 ; 95% CI: 1.00, 1.12) was associated with increased risk of overweight/obesity, with some evidence of a monotonic dose-response relation. We observed weaker and more imprecise associations of the PFAS mixture with BMI or risk of overweight/obesity. Associations did not differ by child sex.

DISCUSSION

In eight U.S.-based prospective cohorts, gestational exposure to higher levels of PFAS were associated with slightly higher childhood BMI z -score and risk of overweight or obesity. Future studies should examine associations of gestational exposure to PFAS with adiposity and related cardiometabolic consequences in older children. https://doi.org/10.1289/EHP11545.

Prenatal Exposure to Per- and Polyfluoroalkyl Substances and Childhood Autism-related Outcomes

BACKGROUND

Epidemiologic evidence linking prenatal exposure to per- and polyfluoroalkyl substances (PFAS) with altered neurodevelopment is inconclusive, and few large studies have focused on autism-related outcomes. We investigated whether blood concentrations of PFAS in pregnancy are associated with child autism-related outcomes.

METHODS

We included 10 cohorts from the National Institutes of Health (NIH)-funded Environmental influences on Child Health Outcomes (ECHO) program (n = 1,429). We measured 14 PFAS analytes in maternal blood collected during pregnancy; eight analytes met detection criteria for analysis. We assessed quantitative autism-related traits in children via parent report on the Social Responsiveness Scale (SRS). In multivariable linear models, we examined relationships of each PFAS (natural log-transformed) with SRS scores. We further modeled PFAS as a complex mixture using Bayesian methods and examined modification of these relationships by child sex.

RESULTS

Most PFAS in maternal blood were not associated with child SRS T-scores. Perfluorononanoic acid (PFNA) showed the strongest and most consistent association: each 1-unit increase in ln-transformed PFNA was associated with greater autism-related traits (adjusted β [95% confidence interval (CI)] = 1.5 [-0.1, 3.0]). The summed mixture, which included six PFAS detected in >70% of participants, was not associated with SRS T-scores (adjusted β [95% highest posterior density interval] = 0.7 [-1.4, 3.0]). We did not observe consistent evidence of sex differences.

CONCLUSIONS

Prenatal blood concentrations of PFNA may be associated with modest increases in child autism-related traits. Future work should continue to examine the relationship between exposures to both legacy and emerging PFAS and additional dimensional, quantitative measures of childhood autism-related outcomes.

Interactions of agricultural pesticide use near home during pregnancy and adverse childhood experiences on adolescent neurobehavioral development in the CHAMACOS study

BACKGROUND

Studies have documented independent adverse associations between prenatal and early-life exposure to environmental chemicals and social adversity with child neurodevelopment; however, few have considered these exposures jointly. The objective of this analysis is to examine whether associations of pesticide mixtures and adolescent neurobehavioral development are modified by early-life adversity in the Center for the Health Assessment of Mothers and Children of Salinas (CHAMACOS) cohort.

METHODS

We used linear mixed effects Bayesian Hierarchical Models (BHM) to examine the joint effect of applications of 11 agricultural pesticides within 1 km of maternal homes during pregnancy and youth-reported Adverse Childhood Experiences (ACEs) with maternal and youth-reported internalizing behaviors, hyperactivity, and attention problems assessed via the Behavior Assessment for Children (BASC) (mean = 50, standard deviation = 10) at ages 16 and 18 years (n = 458).

RESULTS

The median (25th-75th percentiles) of ACEs was 1 (0-3); 72.3% of participants had low ACEs (0-2 events) and 27.7% had ACEs (3+ events). Overall, there was little evidence of modification of exposure-outcome associations by ACEs. A two-fold increase in malathion use was associated with increased internalizing behaviors among those with high ACEs from both maternal- (β = 1.9; 95% Credible Interval (CrI): 0.2, 3.7 for high ACEs vs. β = -0.1; 95% CrI: 1.2, 0.9 for low ACEs) and youth-report (β = 2.1; 95% CrI: 0.4, 3.8 for high ACEs vs. β = 0.2; 95% CrI: 0.8, 1.2 for low ACEs). Applications of malathion and dimethoate were also associated with higher youth-reported hyperactivity and/or inattention among those with high ACEs.

CONCLUSION

We observed little evidence of effect modification of agricultural pesticide use near the home during pregnancy and adolescent behavioral problems by child ACEs. Future studies should examine critical windows of susceptibility of exposure to chemical and non-chemical stressors and should consider biomarker-based exposure assessment methods.

A Bayesian approach to comparing common models of life-course epidemiology

BACKGROUND

Life-course epidemiology studies people's health over long periods, treating repeated measures of their experiences (usually risk factors) as predictors or causes of subsequent morbidity and mortality. Three hypotheses or models often guide the analyst in assessing these sequential risks: the accumulation model (all measurement occasions are equally important for predicting the outcome), the critical period model (only one occasion is important) and the sensitive periods model (a catch-all model for any other pattern of temporal dependence).

METHODS

We propose a Bayesian omnibus test of these three composite models, as well as post hoc decompositions that identify their best respective sub-models. We test the approach via simulations, before presenting an empirical example that relates five sequential measurements of body weight to an RNAseq measure of colorectal-cancer disposition.

RESULTS

The approach correctly identifies the life-course model under which the data were simulated. Our empirical cohort study indicated with >90% probability that colorectal-cancer disposition reflected a sensitive process, with current weight being most important but prior body weight also playing a role.

CONCLUSIONS

The Bayesian methods we present allow precise inferences about the probability of life-course models given the data and are applicable in realistic scenarios involving causal analysis and missing data.

Associations between pesticide mixtures applied near home during pregnancy and early childhood with adolescent behavioral and emotional problems in the CHAMACOS study

Supplemental Digital Content is available in the text.

Studies suggest that exposure to pesticides during pregnancy and early childhood is associated with adverse child neurodevelopment. Research to date has focused primarily on exposure to single pesticides or pesticide classes in isolation; there are little data on the effect of exposure to pesticide mixtures on child and adolescent neurodevelopment.

Prenatal Exposure to Mixtures of Phthalates, Parabens, and Other Phenols and Obesity in Five-Year-Olds in the CHAMACOS Cohort

Exposures to phthalates, parabens, and other phenols are often correlated due to their ubiquitous use in personal care products and plastics. Examining these compounds as a complex mixture may clarify inconsistent relationships between individual chemicals and childhood adiposity. Using data from the Center for the Health Assessment of Mothers and Children of Salinas (CHAMACOS) study, a longitudinal cohort of children in Salinas Valley, California (n = 309), we examined biomarkers of 11 phthalate metabolites and 9 phenols, including several parabens and bisphenol A, measured in maternal urine at two time points during pregnancy. We measured child height and weight at age five to calculate the body mass index (BMI) z-scores and overweight/obesity status. The association between prenatal urinary concentrations of biomarkers with the childhood BMI z-score and overweight/obesity status was analyzed using single-pollutant models and two mixture methods: Bayesian hierarchical modeling (BMH) and Bayesian kernel machine regression (BKMR). Urinary concentrations of monoethyl phthalate, monocarboxy-isononly phthalate (metabolites of diethyl phthalate and di-isodecyl phthalate, respectively), and propylparaben were consistently associated with an increased BMI z-score and overweight/obesity status across all modeling approaches. Higher prenatal exposures to the cumulative biomarker mixture also trended with greater childhood adiposity. These results, robust across two methods that control for co-pollutant confounding, suggest that prenatal exposure to certain phthalates and parabens may increase the risk for obesity in early childhood.

Myoid gonadal tumor. Case series, systematic review, and Bayesian analysis

Myoid gonadal stromal tumor represents a rare testicular neoplasm displaying smooth muscular and gonadal stromal differentiation. This entity has very few cases reported in the literature that describe heterogeneous clinical and pathological characteristics. Bayesian statistics provides a useful framework to combine information from diverse sources. We here presented a case series-the largest so far reported-of myoid gonadal stromal tumor (4 cases) with extensive morphologic, immunohistochemical, and molecular characterization, performed a systematic review of the literature (that identified 9 papers), and used a Bayesian data analysis to understand the characteristics of this disease. Our study collectively described 16 cases. This neoplasm is mainly found in adults (mean age about 40 years) and often has a size of about 3 cm. By morphology, the tumor can infiltrate testicular tubules and is composed of spindle cells; few mitoses can be seen (usually 2/10 HPF). Neoplastic cells are diffusely positive with α-smooth muscle actin with a tram-track staining pattern. S100 protein, FOXL2, and SF1 are also characteristically positive. Moreover, this neoplasm can display epithelial differentiation, in about half of the cases. In conclusion, we foresee the use of this statistical approach in pathology: our analysis allowed a more precise description of this rare entity.

Software Application Profile: The Anchored Multiplier calculator-a Bayesian tool to synthesize population size estimates

Estimating the number of people in hidden populations is needed for public health research, yet available methods produce highly variable and uncertain results. The Anchored Multiplier calculator uses a Bayesian framework to synthesize multiple population size estimates to generate a consensus estimate. Users submit point estimates and lower/upper bounds which are converted to beta probability distributions and combined to form a single posterior probability distribution. The Anchored Multiplier calculator is available as a web browser-based application. The software allows for unlimited empirical population size estimates to be submitted and combined according to Bayes Theorem to form a single estimate. The software returns output as a forest plot (to visually compare data inputs and the final Anchored Multiplier estimate) and a table that displays results as population percentages and counts. The web application 'Anchored Multiplier Calculator' is free software and is available at [http://globalhealthsciences.ucsf.edu/resources/tools] or directly at [http://anchoredmultiplier.ucsf.edu/].

Computational Health Engineering Applied to Model Infectious Diseases and Antimicrobial Resistance Spread

Infectious diseases are the primary cause of mortality worldwide. The dangers of infectious disease are compounded with antimicrobial resistance, which remains the greatest concern for human health. Although novel approaches are under investigation, the World Health Organization predicts that by 2050, septicaemia caused by antimicrobial resistant bacteria could result in 10 million deaths per year. One of the main challenges in medical microbiology is to develop novel experimental approaches, which enable a better understanding of bacterial infections and antimicrobial resistance. After the introduction of whole genome sequencing, there was a great improvement in bacterial detection and identification, which also enabled the characterization of virulence factors and antimicrobial resistance genes. Today, the use of in silico experiments jointly with computational and machine learning offer an in depth understanding of systems biology, allowing us to use this knowledge for the prevention, prediction, and control of infectious disease. Herein, the aim of this review is to discuss the latest advances in human health engineering and their applicability in the control of infectious diseases. An in-depth knowledge of host–pathogen–protein interactions, combined with a better understanding of a host’s immune response and bacterial fitness, are key determinants for halting infectious diseases and antimicrobial resistance dissemination.

Complex Mixtures, Complex Analyses: an Emphasis on Interpretable Results

PURPOSE OF REVIEW

The purpose of this review is to outline the main questions in environmental mixtures research and provide a non-technical explanation of novel or advanced methods to answer these questions.

RECENT FINDINGS

Machine learning techniques are now being incorporated into environmental mixture research to overcome issues with traditional methods. Though some methods perform well on specific tasks, no method consistently outperforms all others in complex mixture analyses, largely because different methods were developed to answer different research questions. We discuss four main questions in environmental mixtures research: (1) Are there specific exposure patterns in the study population? (2) Which are the toxic agents in the mixture? (3) Are mixture members acting synergistically? And, (4) what is the overall effect of the mixture? We emphasize the importance of robust methods and interpretable results over predictive accuracy. We encourage collaboration with computer scientists, data scientists, and biostatisticians in future mixture method development.

Environmental exposure mixtures: questions and methods to address them

PURPOSE OF THIS REVIEW

This review provides a summary of statistical approaches that researchers can use to study environmental exposure mixtures. Two primary considerations are the form of the research question and the statistical tools best suited to address that question. Because the choice of statistical tools is not rigid, we make recommendations about when each tool may be most useful.

RECENT FINDINGS

When dimensionality is relatively low, some statistical tools yield easily interpretable estimates of effect (e.g., risk ratio, odds ratio) or intervention impacts. When dimensionality increases, it is often necessary to compromise this interpretablity in favor of identifying interesting statistical signals from noise; this requires applying statistical tools that are oriented more heavily towards dimension reduction via shrinkage and/or variable selection.

SUMMARY

The study of complex exposure mixtures has prompted development of novel statistical methods. We suggest that further validation work would aid practicing researchers in choosing among existing and emerging statistical tools for studying exposure mixtures.

The Harm Done to Reproducibility by the Culture of Null Hypothesis Significance Testing

In the last few years, stakeholders in the scientific community have raised alarms about a perceived lack of reproducibility of scientific results. In reaction, guidelines for journals have been promulgated and grant applicants have been asked to address the rigor and reproducibility of their proposed projects. Neither solution addresses a primary culprit, which is the culture of null hypothesis significance testing that dominates statistical analysis and inference. In an innovative research enterprise, selection of results for further evaluation based on null hypothesis significance testing is doomed to yield a low proportion of reproducible results and a high proportion of effects that are initially overestimated. In addition, the culture of null hypothesis significance testing discourages quantitative adjustments to account for systematic errors and quantitative incorporation of prior information. These strategies would otherwise improve reproducibility and have not been previously proposed in the widely cited literature on this topic. Without discarding the culture of null hypothesis significance testing and implementing these alternative methods for statistical analysis and inference, all other strategies for improving reproducibility will yield marginal gains at best.

Adverse events associated with incretin-based drugs in Japanese spontaneous reports: a mixed effects logistic regression model

BACKGROUND

Spontaneous Reporting Systems (SRSs) are passive systems composed of reports of suspected Adverse Drug Events (ADEs), and are used for Pharmacovigilance (PhV), namely, drug safety surveillance. Exploration of analytical methodologies to enhance SRS-based discovery will contribute to more effective PhV. In this study, we proposed a statistical modeling approach for SRS data to address heterogeneity by a reporting time point. Furthermore, we applied this approach to analyze ADEs of incretin-based drugs such as DPP-4 inhibitors and GLP-1 receptor agonists, which are widely used to treat type 2 diabetes.

METHODS

SRS data were obtained from the Japanese Adverse Drug Event Report (JADER) database. Reported adverse events were classified according to the MedDRA High Level Terms (HLTs). A mixed effects logistic regression model was used to analyze the occurrence of each HLT. The model treated DPP-4 inhibitors, GLP-1 receptor agonists, hypoglycemic drugs, concomitant suspected drugs, age, and sex as fixed effects, while the quarterly period of reporting was treated as a random effect. Before application of the model, Fisher's exact tests were performed for all drug-HLT combinations. Mixed effects logistic regressions were performed for the HLTs that were found to be associated with incretin-based drugs. Statistical significance was determined by a two-sided p-value <0.01 or a 99% two-sided confidence interval. Finally, the models with and without the random effect were compared based on Akaike's Information Criteria (AIC), in which a model with a smaller AIC was considered satisfactory.

RESULTS

The analysis included 187,181 cases reported from January 2010 to March 2015. It showed that 33 HLTs, including pancreatic, gastrointestinal, and cholecystic events, were significantly associated with DPP-4 inhibitors or GLP-1 receptor agonists. In the AIC comparison, half of the HLTs reported with incretin-based drugs favored the random effect, whereas HLTs reported frequently tended to favor the mixed model.

CONCLUSION

The model with the random effect was appropriate for analyzing frequently reported ADEs; however, further exploration is required to improve the model. The core concept of the model is to introduce a random effect of time. Modeling the random effect of time is widely applicable to various SRS data and will improve future SRS data analyses.

The researcher and the consultant: from testing to probability statements

In the first instalment of this series, Stang and Poole provided an overview of Fisher significance testing (ST), Neyman-Pearson null hypothesis testing (NHT), and their unfortunate and unintended offspring, null hypothesis significance testing. In addition to elucidating the distinction between the first two and the evolution of the third, the authors alluded to alternative models of statistical inference; namely, Bayesian statistics. Bayesian inference has experienced a revival in recent decades, with many researchers advocating for its use as both a complement and an alternative to NHT and ST. This article will continue in the direction of the first instalment, providing practicing researchers with an introduction to Bayesian inference. Our work will draw on the examples and discussion of the previous dialogue.