Early-onset vasomotor symptoms and development of depressive symptoms among premenopausal women.

BACKGROUND: We investigated the association between vasomotor symptoms (VMSs) and the onset of depressive symptoms among premenopausal women. METHODS: This cross-sectional study included 4376 premenopausal women aged 42-52 years, and the cohort study included 2832 women without clinically relevant depressive symptoms at baseline. VMSs included the symptoms of hot flashes and night sweats. Depressive symptoms were evaluated using the Center for Epidemiological Studies Depression Scale; a score of ≥16 was considered to define clinically relevant depressive symptoms. RESULTS: Premenopausal Women with VMSs at baseline exhibited a higher prevalence of depressive symptoms compared with women without VMSs at baseline (multivariable-adjusted prevalence ratio 1.76, 95 % confidence interval [CI] 1.47-2.11). Among the 2832 women followed up (median, 6.1 years), 406 developed clinically relevant depressive symptoms. Women with versus without VMSs had a significantly higher risk of developing clinically relevant depressive symptoms (multivariable-adjusted hazard ratio, 1.72; 95 % CI 1.39-2.14). VMS severity exhibited a dose-response relationship with depressive symptoms (P for trend <0.05). LIMITATIONS: Self-reported questionnaires were only used to obtain VMSs and depressive symptoms, which could have led to misclassification. We also could not directly measure sex hormone levels. CONCLUSIONS: Even in the premenopausal stage, women who experience hot flashes or night sweats have an increased risk of present and developed clinically relevant depressive symptoms. It is important to conduct mental health screenings and provide appropriate support to middle-aged women who experience early-onset VMSs.

Multivariate probit linear mixed models for multivariate longitudinal binary data.

When analyzing multivariate longitudinal binary data, we estimate the effects on the responses of the covariates while accounting for three types of complex correlations present in the data. These include the correlations within separate responses over time, cross-correlations between different responses at different times, and correlations between different responses at each time point. The number of parameters thus increases quadratically with the dimension of the correlation matrix, making parameter estimation difficult; the estimated correlation matrix must also meet the positive definiteness constraint. The correlation matrix may additionally be heteroscedastic; however, the matrix structure is commonly considered to be homoscedastic and constrained, such as exchangeable or autoregressive with order one. These assumptions are overly strong, resulting in skewed estimates of the covariate effects on the responses. Hence, we propose probit linear mixed models for multivariate longitudinal binary data, where the correlation matrix is estimated using hypersphere decomposition instead of the strong assumptions noted above. Simulations and real examples are used to demonstrate the proposed methods. An open source R package, BayesMGLM, is made available on GitHub at https://github.com/kuojunglee/BayesMGLM/ with full documentation to produce the results.

Utilizing local likelihood in regression discontinuity design: Investigating the impact of antiretroviral therapy eligibility on retention in clinical HIV care in South Africa.

The regression discontinuity (RD) design is a widely utilized approach for assessing treatment effects. It involves assigning treatment based on the value of an observed covariate in relation to a fixed threshold. Although the RD design has been widely employed across various problems, its application to specific data types has received limited attention. For instance, there has been little research on utilizing the RD design when the outcome variable exhibits zero-inflation. This study introduces a novel RD estimator using local likelihood, which overcomes the limitations of the local linear regression model, a popular approach for estimating treatment effects in RD design, by considering the data type of the outcome variable. To determine the optimal bandwidth, we propose a modified Ludwig-Miller cross validation method. A set of simulations is carried out, involving binary, count, and zero-inflated outcome variables, to showcase the superior performance of the suggested method over local linear regression models. Subsequently, the proposed local likelihood model is employed on HIV care data, where antiretroviral therapy eligibility is determined by a CD4 count threshold. A comparison is made between the results obtained using the local likelihood model and those obtained using local linear regression.

Human papillomavirus infection and cardiovascular mortality: a cohort study.

BACKGROUND AND AIMS: High-risk human papillomavirus (HR-HPV) infection-a well-established risk factor for cervical cancer-has associations with cardiovascular disease (CVD). However, its relationship with CVD mortality remains uncertain. This study examined the associations between HR-HPV infection and CVD mortality. METHODS: As part of a health examination, 163 250 CVD-free Korean women (mean age: 40.2 years) underwent HR-HPV screening and were tracked for up to 17 years (median: 8.6 years). National death records identified the CVD mortality cases. Hazard ratios (HRs) and 95% confidence intervals (CIs) for CVD mortality were estimated using Cox proportional hazard regression analyses. RESULTS: During 1 380 953 person-years of follow-up, 134 CVD deaths occurred, with a mortality rate of 9.1 per 105 person-years for HR-HPV(-) women and 14.9 per 105 person-years for HR-HPV(+) women. After adjustment for traditional CVD risk factors and confounders, the HRs (95% CI) for atherosclerotic CVD (ASCVD), ischaemic heart disease (IHD), and stroke mortality in women with HR-HPV infection compared with those without infection were 3.91 (1.85-8.26), 3.74 (1.53-9.14), and 5.86 (0.86-40.11), respectively. The association between HR-HPV infection and ASCVD mortality was stronger in women with obesity than in those without (P for interaction = .006), with corresponding HRs (95% CI) of 4.81 (1.55-14.93) for obese women and 2.86 (1.04-7.88) for non-obese women. CONCLUSIONS: In this cohort study of young and middle-aged Korean women, at low risks for CVD mortality, those with HR-HPV infection had higher death rates from CVD, specifically ASCVD and IHD, with a more pronounced trend in obese individuals.

Characterizing temporal trends in populations exposed to aircraft noise around U.S. airports: 1995-2015.

BACKGROUND: Aircraft noise is a key concern for communities surrounding airports, with increasing evidence for health effects and inequitable distributions of exposure. However, there have been limited national-scale assessments of aircraft noise exposure over time and across noise metrics, limiting evaluation of population exposure patterns. OBJECTIVE: We evaluated national-scale temporal trends in aviation noise exposure by airport characteristics and across racial/ethnic populations in the U.S. METHODS: Noise contours were modeled for 90 U.S. airports in 5-year intervals between 1995 and 2015 using the Federal Aviation Administration's Aviation Environmental Design Tool. We utilized linear fixed effects models to estimate changes in noise exposure areas for day-night average sound levels (DNL) of 45, 65, and a nighttime equivalent sound level (Lnight) of 45 A-weighted decibels (dB[A]). We used group-based trajectory modeling to identify distinct groups of airports sharing underlying characteristics. We overlaid noise contours and Census tract data from the U.S. Census Bureau and American Community Surveys for 2000 to 2015 to estimate exposure changes overall and by race/ethnicity. RESULTS: National-scale analyses showed non-monotonic trends in mean exposed areas that peaked in 2000, followed by a 37% decrease from 2005 to 2010 and a subsequent increase in 2015. We identified four distinct trajectory groups of airports sharing latent characteristics related to size and activity patterns. Those populations identifying as minority (e.g., Hispanic/Latino, Black/African American, Asian) experienced higher proportions of exposure relative to their subgroup populations compared to non-Hispanic or White populations across all years, indicating ethnic and racial disparities in airport noise exposure that persist over time. SIGNIFICANCE: Overall, these data identified differential exposure trends across airports and subpopulations, helping to identify vulnerable communities for aviation noise in the U.S. IMPACT STATEMENT: We conducted a descriptive analysis of temporal trends in aviation noise exposure in the U.S. at a national level. Using data from 90 U.S. airports over a span of two decades, we characterized the noise exposure trends overall and by airport characteristics, while estimating the numbers of exposed by population demographics to help identify the impact on vulnerable communities who may bear the burden of aircraft noise exposure.

Serum 25-Hydroxyvitamin D Levels and Risk of Colorectal Cancer: An Age-Stratified Analysis.

BACKGROUND & AIMS: The role of circulating 25-hydroxyvitamin D (25(OH)D) in the prevention of early-onset colorectal cancer (CRC) in young adults aged <50 years is uncertain. We evaluated the age-stratified associations (<50 vs ≥50 years) between circulating 25(OH)D levels and the risk of CRC in a large sample of Korean adults. METHODS: Our cohort study included 236,382 participants (mean age, 38.0 [standard deviation, 9.0] years) who underwent a comprehensive health examination, including measurement of serum 25(OH)D levels. Serum 25(OH)D levels were categorized as <10, 10 to 20, and ≥20 ng/mL. CRC, along with the histologic subtype, site, and invasiveness, was ascertained through linkage with the national cancer registry. Cox proportional hazard models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for incident CRC according to the serum 25(OH)D status, with adjustment for potential confounders. RESULTS: During the 1,393,741 person-years of follow-up (median, 6.5 years; interquartile range, 4.5-7.5 years), 341 participants developed CRC (incidence rate, 19.2 per 105 person-years). Among young individuals aged <50 years, serum 25(OH)D levels were inversely associated with the risk of incident CRC with HRs (95% CIs) of 0.61 (0.43-0.86) and 0.41 (0.27-0.63) for 25(OH)D 10 to 19 ng/mL and ≥20 ng/mL, respectively, with respect to the reference (<10 ng/mL) (P for trend <.001, time-dependent model). Significant associations were evident for adenocarcinoma, colon cancer, and invasive cancers. For those aged ≥50 years, associations were similar, although slightly attenuated compared with younger individuals. CONCLUSIONS: Serum 25(OH)D levels may have beneficial associations with the risk of developing CRC for both early-onset and late-onset disease.

History of Gestational Diabetes and Incident Nonalcoholic Fatty Liver Disease: The Kangbuk Samsung Health Study.

INTRODUCTION: We examined the relationship between a previous history of gestational diabetes mellitus (pGDM) and risk of incident nonalcoholic fatty liver disease (NAFLD) and investigated the effect of insulin resistance or development of diabetes as mediators of any association. METHODS: We performed a retrospective cohort study of 64,397 Korean parous women without NAFLD. The presence of and the severity of NAFLD at baseline and follow-up were assessed using liver ultrasonography. Cox proportional hazards models were used to determine adjusted hazard ratios for incident NAFLD according to a self-reported GDM history, adjusting for confounders as time-dependent variables. Mediation analyses were performed to examine whether diabetes or insulin resistance may mediate the association between pGDM and incident NAFLD. RESULTS: During a median follow-up of 3.7 years, 6,032 women developed incident NAFLD (of whom 343 had moderate-to-severe NAFLD). Multivariable adjusted hazard ratios (95% confidence intervals) comparing women with time-dependent pGDM with the reference group (no pGDM) were 1.46 (1.33-1.59) and 1.75 (1.25-2.44) for incident overall NAFLD and moderate-to-severe NAFLD, respectively. These associations remained significant in analyses restricted to women with normal fasting glucose <100 mg/dL or that excluded women with prevalent diabetes at baseline or incident diabetes during follow-up. Diabetes and insulin resistance (Homeostatic Model Assessment for Insulin Resistance) each mediated <10% of the association between pGDM and overall NAFLD development. DISCUSSION: A previous history of GDM is an independent risk factor for NAFLD development. Insulin resistance, measured by the Homeostatic Model Assessment for Insulin Resistance, and development of diabetes each explained only <10% of the association between GDM and incident NAFLD.

Simulation-Based Cybersecurity Testing and Evaluation Method for Connected Car V2X Application Using Virtual Machine.

In a connected car, the vehicle's internal network is connected to the outside through communication technology. However, this can cause new security vulnerabilities. In particular, V2X communication, to provide the safety of connected cars, can directly threaten the lives of passengers if a security attack occurs. For V2X communication security, standards such as IEEE 1609.2 define the technical functions that digital signature and encryption to provide security of V2X messages. However, it is difficult to verify the security technology by applying it to the environment with real roads because it can be made up of other safety accidents. In addition, vehicle simulation R&D is steadily being carried out, but there is no simulation that evaluates security for the V2X application level. Therefore, in this paper, a virtual machine was used to implement a V2X communication simulation environment that satisfies the requirements for the security evaluation of connected cars. Then, we proposed scenarios for cybersecurity testing and evaluation, implemented and verified through CANoe Option.Car2X. Through this, it is possible to perform sufficient preliminary verification to minimize the variables before verifying security technology in a real road environment.

Bayesian nonparametric adjustment of confounding.

Analysis of observational studies increasingly confronts the challenge of determining which of a possibly high-dimensional set of available covariates are required to satisfy the assumption of ignorable treatment assignment for estimation of causal effects. We propose a Bayesian nonparametric approach that simultaneously (1) prioritizes inclusion of adjustment variables in accordance with existing principles of confounder selection; (2) estimates causal effects in a manner that permits complex relationships among confounders, exposures, and outcomes; and (3) provides causal estimates that account for uncertainty in the nature of confounding. The proposal relies on specification of multiple Bayesian additive regression trees models, linked together with a common prior distribution that accrues posterior selection probability to covariates on the basis of association with both the exposure and the outcome of interest. A set of extensive simulation studies demonstrates that the proposed method performs well relative to similarly-motivated methodologies in a variety of scenarios. We deploy the method to investigate the causal effect of emissions from coal-fired power plants on ambient air pollution concentrations, where the prospect of confounding due to local and regional meteorological factors introduces uncertainty around the confounding role of a high-dimensional set of measured variables. Ultimately, we show that the proposed method produces more efficient and more consistent results across adjacent years than alternative methods, lending strength to the evidence of the causal relationship between SO2 emissions and ambient particulate pollution.

Promoting Survivor Safety in Immigrant Communities: Online Simulation Training for Korean American Faith Leaders.

This study examined the efficacy of a short virtual case simulation for Korean American (KA) faith leaders, "Religious Leaders for Healthy Families." The goal of the program is to increase knowledge about intimate partner violence (IPV) and healthy intimate partner relationships, enhance self-efficacy in IPV prevention and intervention, strengthen attitudes that support their roles on IPV prevention and intervention, increase positive outcome expectations of their actions, and increase behavioral intentions and behaviors on IPV prevention and intervention. KA faith leaders from two large metropolitan areas with a high concentration of KA immigrants were invited to participate in the study (N=102). Participants completed three online assessments: baseline, a 3-month, and a 6-month follow-up. After the baseline assessment, participants were randomized to either intervention (n = 53) or control (n = 49). The intervention consisted of four online simulation modules, each taking approximately 15-20 min to complete. At the 6-month follow-up, faith leaders in the intervention group significantly increased their knowledge and self-efficacy in IPV prevention and intervention compared to the control group. Mean scores for attitudes against IPV and prevention behaviors increased from baseline to the 6-month follow-up for the intervention group more than the control group, but the differences were not statistically significant. "Religious Leaders for Healthy Families" has the potential to reduce disparities in accessing resources and services for immigrant survivors of IPV. With its ease of use, this short, free online intervention has a high potential for uptake among faith leaders. Results are promising, but the COVID-19 pandemic negatively affected the study, with participants having scarce opportunities to practice the skills learned from the intervention. A larger follow-up study that combines "Religious Leaders for Healthy Families" with a community-wide intervention that targets all community members is warranted to reach more faith leaders and community members.

Assessing treatment benefit in the presence of placebo response using the sequential parallel comparison design.

In clinical trials, placebo response is considered a beneficial effect arising from multiple factors, including the patient's expectations for the treatment. Its presence makes the classical parallel study design suboptimal and can bias the inference. The sequential parallel comparison design (SPCD), a two-stage design where the first stage is a classical parallel study design, followed by another parallel design among placebo subjects from the first stage, was proposed to address the shortcomings of the classical design. In SPCD, in lieu of treatment effect, a weighted average of the mean treatment difference in Stage I among all randomized subjects and the mean treatment difference in Stage II among placebo non-responders was proposed as the efficacy measure. However, by linking two possibly different populations, this weighted average lacks interpretability, and the choice of weight remains controversial. In this work, under the principal stratification framework, we propose a causal estimand for the treatment effect under each of three clinically important principal strata: Always Responders, Never Responders, and Drug-only Responders. To make the stratum treatment effect identifiable, we introduce a set of assumptions and two sensitivity parameters. By further considering the strata as latent characteristics, the sensitivity parameters can be estimated. An extensive simulation study is conducted to evaluate the operating characteristics of the proposed method. Finally, we apply our method on the ADAPT-A study data to assess the benefit of low-dose aripiprazole adjunctive to antidepressant therapy treatment.

Prevention of partner violence: virtual case simulation for religious leaders of Korean American immigrant communities.

The project 'Religious Leaders for Healthy Families' aims to prevent intimate partner violence (IPV) and increase access to resources for immigrant victims by enhancing the capacity of religious leaders. Due to cultural and psychosocial barriers, immigrant women prefer to exhaust informal resources, including religious leaders, before seeking help from professional service providers. This study reports the development of a virtual case simulation in which Korean American (KA) religious leaders living in the USA practice how to prevent and address IPV in their congregation. Intervention mapping (IM) guided the development of the simulation. This process was informed by theory, research and expertise in prevention and virtual case simulation. We partnered with victim service organizations and received feedback from religious leaders. The simulation was pilot tested with nine KA religious leaders and three community leaders with IPV expertise in the KA community. The resulting intervention consists of 4 behavioral outcomes and 24 performance objectives. We identified the knowledge, attitudes, outcome expectations and self-efficacy needed to accomplish each objective. The most creative phase was the development of four modules, as interactive virtual case simulations, that address all performance objectives. IM and social cognitive theory provided a useful framework for developing this virtual case simulation. With culturally responsive modifications, the intervention has the potential to be adapted for religious leaders from other immigrant communities.

Religion can play a pivotal role in helping immigrant women. Religious institutions help immigrants maintain their ethnic identity and language and support their integration into the new culture. This study explains the development of an online intervention for Korean American religious leaders to prevent partner violence. Religious leaders in immigrant communities can play a significant role in perpetuating cultural norms that contribute to partner violence or, conversely, create norms that promote equity between partners and value seeking help. Partner violence is high in cultures that adhere to traditional gender roles, with male-dominated interactions, and perceive seeking help as a failure that would bring shame to the whole family. The intervention consists of four modules that depict interactions between a religious leader and female parishioners: (i) a deacon worried about a parishioner who might be a victim, (ii) an immigrant woman married to an abusive husband, (iii) a woman in the hospital because of multiple bruises and a broken arm and (iv) a graduate student concerned about her fiancé's behavior. Experts in various fields, community leaders in the prevention of partner violence, and religious leaders provided their feedback and expertise for program development.

Measuring rater bias in diagnostic tests with ordinal ratings.

Diagnostic tests are frequently reliant upon the interpretation of images by skilled raters. In many clinical settings, however, the variability observed between experts' ratings plays a detrimental role in the degree of confidence in these interpretations, leading to uncertainty in the diagnostic process. For example, in breast cancer testing, radiologists interpret mammographic images, while breast biopsy results are examined by pathologists. Each of these procedures involves elements of subjectivity. We propose here a flexible two-stage Bayesian latent variable model to investigate how the skills of individual raters impact the diagnostic accuracy of image-related testing in large-scale medical testing studies. A strength of the proposed model is that the true disease status of a patient within a reasonable time frame may or may not be known. In these studies, many raters each contribute classifications on a large sample of patients using a defined ordinal grading scale, leading to a complex correlation structure between ratings. Our modeling approach considers the different sources of variability contributed by experts and patients while accounting for correlations present between ratings and patients, in contrast to currently available methods. We propose a novel measure of a rater's ability (magnifier) that, in contrast to conventional measures of sensitivity and specificity, is robust to the underlying prevalence of disease in the population, providing an alternative measure of diagnostic accuracy across patient populations. Extensive simulation studies demonstrate lower bias in estimation of parameters and measures of accuracy, and illustrate outperformance of the proposed model when compared with existing models. Receiver operator characteristic curves are derived to assess the diagnostic accuracy of individual experts and their overall performance. Our proposed modeling approach is applied to a large breast imaging study for known disease status and a uterine cancer dataset for unknown disease status.

Developing and evaluating a pediatric asthma severity computable phenotype derived from electronic health records.

BACKGROUND: Extensive data available in electronic health records (EHRs) have the potential to improve asthma care and understanding of factors influencing asthma outcomes. However, this work can be accomplished only when the EHR data allow for accurate measures of severity, which at present are complex and inconsistent. OBJECTIVE: Our aims were to create and evaluate a standardized pediatric asthma severity phenotype based in clinical asthma guidelines for use in EHR-based health initiatives and studies and also to examine the presence and absence of these data in relation to patient characteristics. METHODS: We developed an asthma severity computable phenotype and compared the concordance of different severity components contributing to the phenotype to trends in the literature. We used multivariable logistic regression to assess the presence of EHR data relevant to asthma severity. RESULTS: The asthma severity computable phenotype performs as expected in comparison with national statistics and the literature. Severity classification for a child is maximized when based on the long-term medication regimen component and minimized when based only on the symptom data component. Use of the severity phenotype results in better, clinically grounded classification. Children for whom severity could be ascertained from these EHR data were more likely to be seen for asthma in the outpatient setting and less likely to be older or Hispanic. Black children were less likely to have lung function testing data present. CONCLUSION: We developed a pragmatic computable phenotype for pediatric asthma severity that is transportable to other EHRs.

Patient-Level Factors Influencing Palliative Care Consultation at a Safety-Net Urban Hospital.

The influence of patient-level factors on palliative and hospice care is unclear. We conducted a retrospective review of 2321 patients aged ≥18 that died within 6 months of admission to our institution between 2012 and 2017. Patients were included for analysis if their chart was complete, their length of stay was ≥48 hours, and if based on their diagnoses, they would have benefited from palliative care consultation (PCC). Bayesian regression with a weakly informative prior was used to find the odds ratio (OR) and 99% credible interval (CrI) of receiving PCC based on race/ethnicity, education, language, insurance status, and income. 730 patients fit our inclusion criteria and 30% (n = 211) received PCC. The OR of receiving PCC was 1.26 (99% CrI, 0.73-2.12) for Blacks, 0.81 (99% CrI, 0.31-1.86) for Hispanics, and 0.69 (99% CrI, 0.19-2.46) for other minorities. Less than high school education was associated with greater odds of PCC (OR 2.28, 99% CrI, 1.09-4.93) compared to no schooling. Compared to English speakers, non-English speakers had higher odds of receiving PCC when cared for by medical services (OR 3.01 [99% CrI, 1.44-5.32]) but lower odds of PCC when cared for by surgical services (0.22 [99% CrI, <0.01-3.42]). Insurance status and income were not associated with differences in PCC. At our institution, we found no evidence of racial/ethnic, insurance, or income status affecting PCC while primary language spoken and educational status did. Further investigation is warranted to examine the system and provider-level factors influencing PCC's low utilization by medical and surgical specialties.

Differential Utilization of Palliative Care Consultation Between Medical and Surgical Services.

There is a paucity of data regarding the utilization of palliative care consultation (PCC) in surgical specialties. We conducted a retrospective review of 2321 adult patients (age ≥18) who died within 6 months of admission to Boston Medical Center from 2012 to 2017. Patients were included for analysis if their length of stay was more than 48 hours and if, based on their diagnoses as determined by literature review and expert consensus, they would have benefited from PCC. Bayesian regression was used to estimate the odds ratio (OR) and 99% credible intervals (CrI) of receiving PCC adjusted for age, sex, race, insurance status, median income, and comorbidity status. Among the 739 patients who fit the inclusion criteria, only 30% (n = 222) received PCC even though 664 (90%) and 75 (10%) of these patients were identified as warranting PCC on medical and surgical services, respectively. Of the 222 patients who received PCC, 214 (96%) were cared for by medical services and 8 (4%) were cared for by surgical services. Patients cared for primarily by surgical were significantly less likely to receive PCC than primary patients of medical service providers (OR, 0.19, 99% CrI, 0.056-0.48). At our institution, many surgical patients appropriate for PCC are unable to benefit from this service due to low consultation numbers. Further investigation is warranted to examine if this phenomenon is observed at other institutions, elucidate the reasons for this disparity, and develop interventions to increase the appropriate use of PCC throughout all medical specialties.

BAYESIAN METHODS FOR MULTIPLE MEDIATORS: RELATING PRINCIPAL STRATIFICATION AND CAUSAL MEDIATION IN THE ANALYSIS OF POWER PLANT EMISSION CONTROLS.

Emission control technologies installed on power plants are a key feature of many air pollution regulations in the US. While such regulations are predicated on the presumed relationships between emissions, ambient air pollution, and human health, many of these relationships have never been empirically verified. The goal of this paper is to develop new statistical methods to quantify these relationships. We frame this problem as one of mediation analysis to evaluate the extent to which the effect of a particular control technology on ambient pollution is mediated through causal effects on power plant emissions. Since power plants emit various compounds that contribute to ambient pollution, we develop new methods for multiple intermediate variables that are measured contemporaneously, may interact with one another, and may exhibit joint mediating effects. Specifically, we propose new methods leveraging two related frameworks for causal inference in the presence of mediating variables: principal stratification and causal mediation analysis. We define principal effects based on multiple mediators, and also introduce a new decomposition of the total effect of an intervention on ambient pollution into the natural direct effect and natural indirect effects for all combinations of mediators. Both approaches are anchored to the same observed-data models, which we specify with Bayesian nonparametric techniques. We provide assumptions for estimating principal causal effects, then augment these with an additional assumption required for causal mediation analysis. The two analyses, interpreted in tandem, provide the first empirical investigation of the presumed causal pathways that motivate important air quality regulatory policies.

A Bayesian semiparametric latent variable approach to causal mediation.

In assessing causal mediation effects in randomized studies, a challenge is that the direct and indirect effects can vary across participants due to different measured and unmeasured characteristics. In that case, the population effect estimated from standard approaches implicitly averages over and does not estimate the heterogeneous direct and indirect effects. We propose a Bayesian semiparametric method to estimate heterogeneous direct and indirect effects via clusters, where the clusters are formed by both individual covariate profiles and individual effects due to unmeasured characteristics. These cluster-specific direct and indirect effects can be estimated through a set of regression models where specific coefficients are clustered by a stick-breaking prior. To let clustering be appropriately informed by individual direct and indirect effects, we specify a data-dependent prior. We conduct simulation studies to assess performance of the proposed method compared to other methods. We use this approach to estimate heterogeneous causal direct and indirect effects of an expressive writing intervention for patients with renal cell carcinoma.

A framework for Bayesian nonparametric inference for causal effects of mediation.

We propose a Bayesian non-parametric (BNP) framework for estimating causal effects of mediation, the natural direct, and indirect, effects. The strategy is to do this in two parts. Part 1 is a flexible model (using BNP) for the observed data distribution. Part 2 is a set of uncheckable assumptions with sensitivity parameters that in conjunction with Part 1 allows identification and estimation of the causal parameters and allows for uncertainty about these assumptions via priors on the sensitivity parameters. For Part 1, we specify a Dirichlet process mixture of multivariate normals as a prior on the joint distribution of the outcome, mediator, and covariates. This approach allows us to obtain a (simple) closed form of each marginal distribution. For Part 2, we consider two sets of assumptions: (a) the standard sequential ignorability (Imai et al., 2010) and (b) weakened set of the conditional independence type assumptions introduced in Daniels et al. (2012) and propose sensitivity analyses for both. We use this approach to assess mediation in a physical activity promotion trial.

Causal mediation analysis in the context of clinical research.

Clinical researches usually collected numerous intermediate variables besides treatment and outcome. These variables are often incorrectly treated as confounding factors and are thus controlled using a variety of multivariable regression models depending on the types of outcome variable. However, these methods fail to disentangle underlying mediating processes. Causal mediation analysis (CMA) is a method to dissect total effect of a treatment into direct and indirect effect. The indirect effect is transmitted via mediator to the outcome. The mediation package is designed to perform CMA under the assumption of sequential ignorability. It reports average causal mediation effect (ACME), average direct effect (ADE) and total effect. Also, the package provides visualization tool for these estimated effects. Sensitivity analysis is designed to examine whether the results are robust to the violation of the sequential ignorability assumption since the assumption has been criticized to be too strong to be satisfied in research practice.