publications

2025

1. Large floods drive changes in cause-specific mortality in the United States

   Victoria D. Lynch, Jonathan A. Sullivan, Aaron B. Flores, and 6 more authors

   Nature Medicine, Jan 2025

   Publisher: Nature Publishing Group

   Abs DOI

   Flooding greatly endangers public health and is an urgent concern as rapid population growth in flood-prone regions and more extreme weather events will increase the number of people at risk. However, an exhaustive analysis of mortality following floods has not been conducted. Here we used 35.6 million complete death records over 18 years (2001–2018) from the National Center for Health Statistics in the United States, highly resolved flood exposure data and a Bayesian conditional quasi-Poisson model to estimate the association of flooding with monthly county-level death rates for cancers, cardiovascular diseases, infectious and parasitic diseases, injuries, neuropsychiatric conditions and respiratory diseases up to 3 months after the flood. During the month of flooding, very severe heavy rain-related floods were associated with increased infectious disease (3.2%; 95% credible interval (CrI): 0.1%, 6.2%) and cardiovascular disease (2.1%; 95% CrI: 1.3%, 3.0%) death rates and tropical cyclone-related floods were associated with increased injury death rates (15.3%; 95% CrI: 12.4%, 18.1%). During the month of very severe tropical cyclone-related flooding, increases in injury death rate were higher for those ≥65 years old (24.9; 95% CrI: 20.0%, 29.8%) than for those aged \textless65 years (10.2%; 95% CrI: 6.6%, 13.8%) and for females (21.2%; 95% CrI: 16.3%, 26.1%) than for males (12.6%; 95% CrI: 9.1%,16.1%). Effective public health responses are critical now and with projected increased flood severity driven by climate change.

2. A Bayesian spatial measurement error approach to incorporate heterogeneous population-at-risk uncertainty in estimating small-area opioid mortality rates

   Emily N. Peterson, Rachel C. Nethery, Jarvis T. Chen, and 4 more authors

   Spatial and Spatio-temporal Epidemiology, Jun 2025

   Abs DOI

   Monitoring small-area geographical population trends in opioid mortality has significant implications for informing preventative resource allocation. A common approach to estimating small-area opioid mortality uses a standard disease mapping method where population-at-risk estimates (denominators) are treated as fixed. This assumption ignores the uncertainty in small-area population estimates, potentially biasing risk estimates and underestimating their uncertainties. We compare a Bayesian Spatial Berkson Error model and a Bayesian Spatial Classical Error model to a naive approach that treats denominators as fixed. Using simulations, we illustrate potential bias from ignored population-at-risk uncertainty. We apply these methods to obtain 2020 opioid mortality risk estimates for 159 counties in Georgia. Assessing differences in bias and uncertainty across approaches can improve the accuracy of small-area opioid risk estimates, guiding public health interventions, policies, and resource allocation.

3. Optimizing Heat Alert Issuance with Reinforcement Learning

   Ellen M. Considine, Rachel C. Nethery, Gregory A. Wellenius, and 2 more authors

   Proceedings of the AAAI Conference on Artificial Intelligence, Apr 2025

   Number: 27

   Abs DOI

   A key strategy in societal adaptation to climate change is using alert systems to prompt preventative action and reduce the adverse health impacts of extreme heat events. This paper implements and evaluates reinforcement learning (RL) as a tool to optimize the effectiveness of such systems. Our contributions are threefold. First, we introduce a new publicly available RL environment enabling the evaluation of the effectiveness of heat alert policies to reduce heat-related hospitalizations. The rewards model is trained from a comprehensive dataset of historical weather, Medicare health records, and socioeconomic/geographic features. We use scalable Bayesian techniques tailored to the low-signal effects and spatial heterogeneity present in the data. The transition model uses real historical weather patterns enriched by a data augmentation mechanism based on climate region similarity. Second, we use this environment to evaluate standard RL algorithms in the context of heat alert issuance. Our analysis shows that policy constraints are needed to improve RL’s initially poor performance. Third, a post-hoc contrastive analysis provides insight into scenarios where our modified heat alert-RL policies yield significant gains/losses over the current National Weather Service alert policy in the United States.

4. Wildfire Smoke Exposure and Cause-Specific Hospitalization in Older Adults

   Sofia L. Vega, Marissa L. Childs, Sarika Aggarwal, and 1 more author

   JAMA Network Open, Apr 2025

   Abs DOI

   The escalating intensity of wildfires in the western US is increasing exposure to smoke pollution. Previous studies of wildfire smoke and health have primarily focused on mortality and respiratory and cardiovascular events, with limited research on broader health impacts or on the shape of concentration-response curves.To characterize the associations between exposure to smoke-specific fine particulate matter (PM2.5) and cause-specific hospitalizations among older adults in the western US.This retrospective cohort study used Medicare inpatient claims data from 2006 to 2016 linked with machine learning–derived smoke-specific PM2.5 to assess associations between smoke PM2.5 and hospitalization rates. Participants included Medicare beneficiaries aged 65 years or older who lived in a western US state (ie, Arizona, California, Colorado, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, or Wyoming). Analyses were conducted from October 2023 to February 2025.Daily county-level smoke-specific PM2.5 concentrations were estimated from machine learning models trained on monitor and satellite data.Daily county-level rates of unscheduled hospitalization for each of 13 broad cause categories. To characterize the association between each cause of hospitalization and smoke PM2.5, distributed lag models were fitted with hospitalization rates modeled as a function of same-day smoke PM2.5 exposure and exposures on each day of the preceding week, using splines on exposure to allow for nonlinearity.The study included 10 369 361 individuals (mean [SD] age, 74.7 [7.9] years; 4 862 826 male [46.9%]; 5 506 535 female [53.1%]; 373 252 Black [3.6%]; 420 577 Hispanic [4.1%]; and 8 365 607 White [80.7%]), 57 million person-months of follow-up, and 4.7 million unscheduled hospitalizations. Smoke PM2.5 concentration-response curves for respiratory hospitalizations and cardiovascular hospitalizations were flat at lower concentrations but showed increasing trends at concentrations above 25 μg/m3. On average, daily hospitalizations (per 100 000) increased by 2.40 (95% CI, 0.17 to 4.63) for respiratory concerns when increasing same-day and preceding week smoke PM2.5 concentrations from 0 to 40 μg/m3; hospitalizations for cardiovascular concerns increased by 2.61 (95% CI, –0.09 to 5.30), a difference that was not statistically significant. No associations were observed for other causes of hospitalization.In this cohort study, exposure to high levels of smoke pollution was associated with an increase in hospitalizations for respiratory diseases. These findings underscore the need for interventions to mitigate the health impacts of wildfire smoke exposure.

5. Long-term Exposure to Air Pollution and Incidence of Type 2 Diabetes in the Nurses’ Health Study and Nurses’ Health Study II

   Melissa R Fiffer, Jie Chen, Emily L Silva, and 8 more authors

   Environmental Health Perspectives, Apr 2025

   Publisher: Environmental Health Perspectives

   Abs DOI

   Background: Research has detected associations between air pollution exposure and type 2 diabetes (T2DM), but findings from large cohort studies are needed to ascertain the most influential pollutants, susceptible subpopulations, and low-level exposure associations. Our aim was to prospectively evaluate the association between long-term exposure to fine particulate matter (PM2.5) and nitrogen dioxide (NO2) and T2DM incidence in the Nurses’ Health Study (NHS) and Nurses’ Health Study II (NHSII) cohorts of U.S. women. Methods: Monthly PM2.5 and NO2 exposures were predicted from spatiotemporal models and linked to participants’ residential addresses. We used Cox proportional hazards models to assess the association between 24-month moving average PM2.5 and NO2 exposure and self-reported, clinician diagnosed T2DM from 1992-2019. We adjusted for time-varying lifestyle factors, reproductive hormonal factors, and individual and neighborhood socioeconomic status (SES). Results were meta-analyzed. We evaluated whether relationships persisted at levels below the current U.S. EPA National Ambient Air Quality Standards (NAAQS). Lastly, we examined multiplicative and additive interactions by body mass index (BMI), smoking status, physical activity, neighborhood SES, and region. Results: Over follow-up, there were 19,083 incident T2DM cases among the 208,733 women in NHS and NHSII. In fully-adjusted single pollutant models, the HR for an interquartile range (IQR=4.9 µg/m3) higher 24-month average PM2.5 exposure was 1.05 (95% CI: 1.02, 1.08) for incident T2DM. The HR for an IQR (7.3 ppb) higher NO2 exposure was 1.05 (95% CI: 1.01, 1.09). Both associations were robust to co-adjustment. Associations remained stable when restricting to PM2.5 levels below the NAAQS as compared to the full dataset. Stronger associations were observed in individuals who had a BMI ≥30, were physically active, and resided in the Northeast. Conclusions: Our results showed a positive association between T2DM and long-term exposure to PM2.5 and NO2, persisting even at levels below the current EPA NAAQS. https://doi.org/10.1289/EHP15673

6. The effect of air pollution exposure on menstrual cycle health using self-reported data from a mobile health app: a prospective, observational study

   Priyanka N deSouza, Amanda A Shea, Virginia J Vitzthum, and 8 more authors

   The Lancet Planetary Health, May 2025

   Abs DOI

   Background Toxicological evidence suggests that ambient air pollution has endocrine-disrupting properties that can affect menstrual cycle functioning, which represents an important marker of women’s reproductive health. We aimed to estimate the effect of short-term and long-term PM2·5 exposure on menstrual cycle outcomes across the USA, Brazil, and Mexico using self-reported data from a mobile health app. Methods For this prospective observational study, we collected de-identified self-reported data from the Clue mobile health app, in which users self-tracked menstruation cycles. For the current study, eligible participants were aged 18–44 years, were not using hormonal birth control, and lived in one of 230 cities in the USA, Mexico, or Brazil. The primary outcome of interest at the city level was the proportion of menstrual cycles with abnormally short length (\textless24 days) and long length (\textgreater38 days) of all cycles recorded. The primary outcome at the cycle level was a binary indicator: abnormal cycle length (\textless24 days or \textgreater38 days) or not (normal cycle length). We used regression analyses to evaluate associations between long-term PM2·5 concentrations (mean concentration between 2016 and 2020) and the city-level outcomes after controlling for potential confounders. Conditional logistic regression models were used to evaluate associations between cycle-specific PM2·5 and if a cycle was of abnormal length within an individual in the dataset, after controlling for time-varying factors. Findings Between Jan 1, 2016 and Dec 31, 2020, 92 550 app users residing in 230 cities across the USA, Brazil, and Mexico provided data corresponding to 2 220 281 menstrual cycles, and were included in our main cohort. A significant association was observed between long-term PM2·5 exposure and the proportion of menstrual cycles of abnormally long or short duration (odds ratio [OR] 1·023 [95% CI 1·013–1·033]) and the proportion of cycles that were specifically abnormally long (OR 1·036 [1·023–1·049]) for every 10 μg/m3 increase in PM2·5. No associations were identified between short-term PM2·5 concentrations and abnormal cycle length. Interpretation These findings suggest that PM2·5 exposure affects menstrual cycle outcomes. More research is needed to better elucidate the biological mechanisms through which PM2·5 affects the menstrual cycle. Funding None.

7. Residential greenness and diabetes incidence in two prospective cohorts of US women

   Melissa R. Fiffer, Peter James, Jie Chen, and 8 more authors

   Environmental Epidemiology, Aug 2025

   Abs DOI

   Background:  Epidemiologic studies have associated higher neighborhood greenness with lower type 2 diabetes (T2D) risk. However, more work is needed to assess interrelationships between greenness, T2D risk factors, and T2D. Our aim was to prospectively evaluate the association between greenness and T2D incidence, and investigate effect modifiers, in the Nurses’ Health Study (NHS) and Nurses’ Health Study II (NHSII) cohorts of US women. Methods:  Greenness exposure was defined using the normalized difference vegetation index (NDVI), a quantitative indicator of photosynthetic vegetation. We obtained 30m2 resolution Landsat satellite data and calculated average NDVI within 270 m and 1230 m radial buffers to represent residential exposure and exposure within a short walk/drive using addresses from 1992 to 2017. We used time-varying Cox proportional hazards models to assess summer average NDVI in the 2 years before diagnosis and self-reported, validated clinician T2D diagnosis through 2019. We adjusted for time-varying covariates including lifestyle factors, hormone use, individual and neighborhood socioeconomic status (nSES), population density, particulate matter (PM)2.5 and nitrogen dioxide (NO2) exposure, and baseline body mass index (BMI). Results from the two cohorts (n = 212,548) were meta-analyzed. We examined effect modification by time-varying BMI, physical activity, smoking, region, air pollution, population density, and nSES. Supplemental analyses explored mediation by physical activity and air pollution. Results:  During the 27 years of follow-up, there were 18,527 incident T2D cases. In fully adjusted models, the meta-analyzed hazard ratio was 0.96 (95% confidence interval = 0.95, 0.97) for a 0.1 unit increase in 2-year summer average NDVI. In NHS, stronger associations were found among participants in the lowest PM2.5 tertile, and in NHSII, among those with BMI \textless30 and those in neighborhoods above the lowest nSES quartile. Conclusions:  In one of the first US nationwide prospective analyses of greenness and T2D, we found a protective association robust to air pollution co-exposure adjustment and persistent across subpopulations.

8. Severe flooding and cause-specific hospitalisation among older adults in the USA: a retrospective matched cohort analysis

   Sarika Aggarwal, Jie K Hu, Jonathan A Sullivan, and 2 more authors

   The Lancet Planetary Health, Jul 2025

   Abs DOI

   Background Floods are the most common climate-related disaster; yet previous studies have investigated the impact of floods on only a few health outcomes in narrow spatiotemporal settings. We aimed to assess the association between severe flood exposure and cause-specific hospitalisation rates in adults older than 65 years in the contiguous USA. Methods In this retrospective matched cohort analysis, we obtained inpatient claims data from Medicare fee-for-service beneficiaries older than 65 years living in the contiguous USA from Jan 1, 2000, to Dec 31, 2016. From each inpatient hospitalisation record, we extracted the admission date, primary International Classification of Diseases, 9th revision, clinical modification (ICD-9-CM) code (or 10th revision [ICD-10-CM] code on or after Oct 1, 2015), and self-reported residential ZIP code. Hospitalisation data were linked with satellite-based, high-resolution historical flood maps from the Global Flood Database by ZIP code. Days during and shortly after a flood exposure were matched to non-flood-affected control days by ZIP code and day-of-year. We estimated relative percentage changes in hospitalisation rates for 13 mutually exclusive, well-defined disease categories during and in the 4 weeks following flood exposure with conditional quasi-Poisson regression models. Findings This study captured 72 major flood events and included over 4·5 million hospitalisations occurring over a 17-year period. We observed elevated rates of hospitalisation on average during and following flood exposure for skin diseases (3·1% [95% CI 1·4 to 4·9]), nervous system diseases (2·5% [1·0 to 4·1]), musculoskeletal system diseases (1·3% [0·3 to 2·3]), and injuries or poisoning (1·1% [0·2 to 2·0]). Communities with lower proportions of Black residents experienced exacerbated flood effects for nervous system diseases (7·6% [95% CI 2·8 to 12·6]), whereas skin diseases (6·1% [1·9 to 10·5]) and mental health-related impacts (3·0% [–0·3 to 6·5]) were more pronounced for areas with larger percentages of Black residents during flood exposure. Interpretation Among adults older than 65 years, exposure to severe flood events was associated with increased hospitalisation rates for skin diseases, nervous system diseases, musculoskeletal system diseases, and injuries. Different patterns of hospital admission persisted for populations with higher versus lower proportions of Black residents. Our findings indicate a need for targeted flood-specific preparedness and adaptation strategies for socially vulnerable populations, including older individuals and racially minoritised communities. Funding National Institutes of Health, Harvard Data Science Initiative, and Alfred P Sloan Foundation.

9. Spatio-temporal methods to handle missing data in syndromic surveillance with applications to health management information system data

   Nicholas B. Link, Anuraag Gopaluni, Isabel Fulcher, and 3 more authors

   Spatial and Spatio-temporal Epidemiology, Aug 2025

   Abs DOI

   Syndromic surveillance monitors infectious diseases, especially in situations where direct disease monitoring is unavailable. However, conventional syndromic surveillance methods face challenges in handling missing data, particularly when the missing completely at random (MCAR) assumption is violated. Additionally, these methods often do not leverage spatio-temporal techniques that can reduce bias and improve their performance. This study addresses both of these limitations by comparing a baseline syndromic surveillance model with a frequentist spatio-temporal model used in infectious diseases and a Bayesian spatio-temporal conditional autoregressive (CAR) model. Drawing inspiration from COVID-19 symptom data collected via routine health systems in Liberia, we conduct simulations with various data generating processes, spatio-temporal correlation structures, and missing data mechanisms. Across the diverse simulations for outbreak detection, the baseline model and the Bayesian CAR model had high specificity, thus limiting outbreak false alarms. The findings underscore the importance of considering spatio-temporal models for syndromic surveillance.

2024

1. Air pollution and serious bleeding events in high-risk older adults

   Rindala Fayyad, Kevin Josey, Poonam Gandhi, and 5 more authors

   Environmental Research, Jun 2024

   Abs DOI

   Importance Despite biological plausibility, very few epidemiologic studies have investigated the risks of clinically significant bleeding events due to particulate air pollution. Objective To measure the independent and synergistic effects of PM2.5 exposure and anticoagulant use on serious bleeding events. Design Retrospective cohort study (2008–2016). Setting Nationwide Medicare population. Participants A 50% random sample of Medicare Part D-eligible Fee-for-Service beneficiaries at high risk for cardiovascular and thromboembolic events. Exposures Fine particulate matter (PM2.5) and anticoagulant drugs (apixaban, dabigatran, edoxaban, rivaroxaban, or warfarin). Main outcomes and measures The outcomes were acute hospitalizations for gastrointestinal bleeding, intracranial bleeding, or epistaxis. Hazard ratios and 95% CIs for PM2.5 exposure were estimated by fitting inverse probability weighted marginal structural Cox proportional hazards models. The relative excess risk due to interaction was used to assess additive-scale interaction between PM2.5 exposure and anticoagulant use. Results The study cohort included 1.86 million high-risk older adults (mean age 77, 60% male, 87% White, 8% Black, 30% anticoagulant users, mean PM2.5 exposure 8.81 μg/m3). A 10 μg/m3 increase in PM2.5 was associated with a 48% (95% CI: 45%–52%), 58% (95% CI: 49%–68%) and 55% (95% CI: 37%–76%) increased risk of gastrointestinal bleeding, intracranial bleeding, and epistaxis, respectively. Significant additive interaction between PM2.5 exposure and anticoagulant use was observed for gastrointestinal and intracranial bleeding. Conclusions Among older adults at high risk for cardiovascular and thromboembolic events, increasing PM2.5 exposure was significantly associated with increased risk of gastrointestinal bleeding, intracranial bleeding, and epistaxis. In addition, PM2.5 exposure and anticoagulant use may act together to increase risks of severe gastrointestinal and intracranial bleeding. Thus, clinicians may recommend that high-risk individuals limit their outdoor air pollution exposure during periods of increased PM2.5 concentrations. Our findings may inform environmental policies to protect the health of vulnerable populations.

2. Quantifying Multipollutant Health Impacts Using the Environmental Benefits Mapping and Analysis Program–Community Edition (BenMAP-CE): A Case Study in Atlanta, Georgia

   Evan Coffman, Ana G. Rappold, Rachel C. Nethery, and 7 more authors

   Environmental Health Perspectives, Mar 2024

   Publisher: Environmental Health Perspectives

   Abs DOI

   Background: Air pollution risk assessments do not generally quantify health impacts using multipollutant risk estimates, but instead use results from single-pollutant or copollutant models. Multipollutant epidemiological models account for pollutant interactions and joint effects but can be computationally complex and data intensive. Risk estimates from multipollutant studies are therefore challenging to implement in the quantification of health impacts. Objectives: Our objective was to conduct a case study using a developmental multipollutant version of the Environmental Benefits Mapping and Analysis Program—Community Edition (BenMAP-CE) to estimate the health impact associated with changes in multiple air pollutants using both a single and multipollutant approach. Methods: BenMAP-CE was used to estimate the change in the number of pediatric asthma emergency department (ED) visits attributable to simulated changes in air pollution between 2011 and 2025 in Atlanta, Georgia, applying risk estimates from an epidemiological study that examined short-term single-pollutant and multipollutant (with and without first-order interactions) exposures. Analyses examined individual pollutants (i.e., ozone, fine particulate matter, carbon monoxide, nitrogen dioxide ( NO2 ), sulfur dioxide, and particulate matter components) and combinations of these pollutants meant to represent shared properties or predefined sources (i.e., oxidant gases, secondary pollutants, traffic, power plant, and criteria pollutants). Comparisons were made between multipollutant health impact functions (HIF) and the sum of single-pollutant HIFs for the individual pollutants that constitute the respective pollutant groups. Results: Photochemical modeling predicted large decreases in most of the examined pollutant concentrations between 2011 and 2025 based on sector specific (i.e., source-based) estimates of growth and anticipated controls. Estimated number of avoided asthma ED visits attributable to any given multipollutant group were generally higher when using results from models that included interaction terms in comparison with those that did not. We estimated the greatest number of avoided pediatric asthma ED visits for pollutant groups that include NO2 (i. e., criteria pollutants, oxidants, and traffic pollutants). In models that accounted for interaction, year-round estimates for pollutant groups that included NO2 ranged from 27.1 [95% confidence interval (CI): 1.6, 52.7; traffic pollutants] to 55.4 (95% CI: 41.8, 69.0; oxidants) avoided pediatric asthma ED visits. Year-round results using multipollutant risk estimates with interaction were comparable to the sum of the single-pollutant results corresponding to most multipollutant groups [e.g., 52.9 (95% CI: 43.6, 62.2) for oxidants] but were notably lower than the sum of the single-pollutant results for some pollutant groups [e.g., 77.5 (95% CI: 66.0, 89.0) for traffic pollutants]. Discussion: Performing a multipollutant health impact assessment is technically feasible but computationally complex. It requires time, resources, and detailed input parameters not commonly reported in air pollution epidemiological studies. Results estimated using the sum of single-pollutant models are comparable to those quantified using a multipollutant model. Although limited to a single study and location, assessing the trade-offs between a multipollutant and single-pollutant approach is warranted. https://doi.org/10.1289/EHP12969

3. 1965 US Voting Rights Act Impact on Black and Black Versus White Infant Death Rates in Jim Crow States, 1959–1980 and 2017–2021

   Tamara Rushovich, Rachel C. Nethery, Ariel White, and 1 more author

   American Journal of Public Health, Feb 2024

   Publisher: American Public Health Association

   Abs DOI

   Objectives. To investigate the impact of the US Voting Rights Act (VRA) of 1965 on Black and Black versus White infant deaths in Jim Crow states. Methods. Using data from 1959 to 1980 and 2017 to 2021, we applied difference-in-differences methods to quantify differential pre–post VRA changes in infant deaths in VRA-exposed versus unexposed counties, controlling for population size and social, economic, and health system characteristics. VRA-exposed counties, identified by Section 4, were subject to government interventions to remove existing racist voter suppression policies. Results. Black infant deaths in VRA-exposed counties decreased by an average of 11.4 (95% confidence interval [CI] = 1.7, 21.0) additional deaths beyond the decrease experienced by unexposed counties between the pre-VRA period (1959–1965) and the post-VRA period (1966–1970). This translates to 6703 (95% CI = 999.6, 12 348) or 17.5% (95% CI = 3.1%, 28.1%) fewer deaths than would have been experienced in the absence of the VRA. The equivalent differential changes were not significant among the White or total population. Conclusions. Passage of the VRA led to pronounced reductions in Black infant deaths in Southern counties subject to government intervention because these counties had particularly egregious voter suppression practices. (Am J Public Health. Published online ahead of print February 1, 2024:e1–e9. https://doi-org.ezp-prod1.hul.harvard.edu/10.2105/AJPH.2023.307518)

4. Ambient heat exposure patterns and emergency department visits and hospitalizations among medicare beneficiaries 2008–2019

   Aayush Visaria, Euntaik Kang, Ashwaghosha Parthasarathi, and 9 more authors

   The American Journal of Emergency Medicine, Apr 2024

   Abs DOI

   Objective To assess the association between ambient heat and all-cause and cause-specific emergency department (ED) visits and acute hospitalizations among Medicare beneficiaries in the conterminous United States. Design Retrospective cohort study. Setting Conterminous US from 2008 and 2019. Participants 2% random sample of all Medicare fee-for-service beneficiaries eligible for Parts A, B, and D. Main outcome measures All-cause and cause-specific (cardiovascular, renal, and heat-related) ED visits and unplanned hospitalizations were identified using primary ICD-9 or ICD-10 diagnosis codes. We measured the association between ambient temperature – defined as daily mean temperature percentile of summer (June through September) – and the outcomes. Hazard ratios and their associated 95% confidence intervals were estimated using multivariable Cox proportional hazards regression, adjusting for individual level demographics, comorbidities, healthcare utilization factors and zip-code level social factors. Results Among 809,636 Medicare beneficiaries (58% female, 81% non-Hispanic White, 24% \textless65), older beneficiaries (aged ≥65) exposed to \textgreater95th percentile temperature had a 64% elevated adjusted risk of heat-related ED visits (HR [95% CI], 1.64 [1.46,1.85]) and a 4% higher risk of all-cause acute hospitalization (1.04 [1.01,1.06]) relative to \textless25th temperature percentile. Younger beneficiaries (aged \textless65) showed increased risk of heat-related ED visits (2.69 [2.23,3.23]) and all-cause ED visits (1.03 [1.01,1.05]). The associations with heat related events were stronger in males and individuals dually eligible for Medicare and Medicaid. No significant differences were observed by climatic region. We observed no significant relationship between temperature percentile and risk of CV-related ED visits or renal-related ED visits. Conclusions Among Medicare beneficiaries from 2008 to 2019, exposure to daily mean temperature ≥ 95th percentile was associated with increased risk of heat-related ED visits, with stronger associations seen among beneficiaries \textless65, males, and patients with low socioeconomic position. Further longitudinal studies are needed to understand the impact of heat duration, intensity, and frequency on cause-specific hospitalization outcomes.

5. A Bayesian hierarchical small area population model accounting for data source specific methodologies from American Community Survey, Population Estimates Program, and Decennial census data

   Emily N. Peterson, Rachel C. Nethery, Tullia Padellini, and 6 more authors

   The Annals of Applied Statistics, Jun 2024

   Publisher: Institute of Mathematical Statistics

   Abs DOI

   Small area population counts are necessary for many epidemiological studies, yet their quality and accuracy are often not assessed. In the United States, small area population counts are published by the United States Census Bureau (USCB) in the form of the decennial census counts, intercensal population projections (PEP), and American Community Survey (ACS) estimates. Although there are significant relationships between these three data sources, there are important contrasts in data collection, data availability, and processing methodologies such that each set of reported population counts may be subject to different sources and magnitudes of error. Additionally, these data sources do not report identical small area population counts due to post-survey adjustments specific to each data source. Consequently, in public health studies, small area disease/mortality rates may differ depending on which data source is used for denominator data. To accurately estimate annual small area population counts and their associated uncertainties, we present a Bayesian population (BPop) model, which fuses information from all three USCB sources, accounting for data source specific methodologies and associated errors. We produce comprehensive small area race-stratified estimates of the true population, and associated uncertainties, given the observed trends in all three USCB population estimates. The main features of our framework are: (1) a single model integrating multiple data sources, (2) accounting for data source specific data generating mechanisms and specifically accounting for data source specific errors, and (3) prediction of population counts for years without USCB reported data. We focus our study on the Black and White only populations for 159 counties of Georgia and produce estimates for years 2006–2023. We compare BPop population estimates to decennial census counts, PEP annual counts, and ACS multi-year estimates. Additionally, we illustrate and explain the different types of data source specific errors. Lastly, we compare model performance using simulations and validation exercises. Our Bayesian population model can be extended to other applications at smaller spatial granularity and for demographic subpopulations defined further by race, age, and sex, and/or for other geographical regions.

6. Algorithm-driven estimation of household cooking activity and its impact on indoor PM2.5 assessments

   Sanjana Bhaskar, Andrew Shapero, Futu Chen, and 4 more authors

   Indoor Environments, Oct 2024

   Abs DOI

   Background Household PM2.5 exposures have adverse health effects, and cooking behaviors are an important source of PM2.5 in the home. There is a need for accurate measures of cooking activity to better understand its associations with household PM2.5 since self-reported surveys are often subject to recall bias and misreporting of cooking events. Objective In this study, we aimed to address limitations associated with a self-reported cooking metric, by using temperature data to estimate cooking activity. Methods We developed an algorithm to identify cooking events at the 5-minute level using real-time temperature data measured near the stove and in the living room, across 148 households in Chelsea and Dorchester, MA. We compared the number of cooking events identified by this algorithm with cooking events self-reported by participants in daily activity logs and survey responses, and further assessed how these metrics differed with respect to their associations with occurrence of peak PM2.5, in mixed effects logistic regression models. Results We found that 65 % of the cooking events identified by the algorithm were not reported by participants. Furthermore, households classified as frequent vs infrequent cooking households using the algorithm had a larger difference in indoor PM2.5 levels, compared to households classified by self-report. In mixed effects logistic regression models for elevated household PM2.5 levels, we observed much stronger associations between household PM2.5 and algorithm-derived cooking activity (OR: 2.85 [95 % CI: 2.76, 2.95]) as compared to the association between household PM2.5 and self-reported cooking activity (OR: 1.22 [95 % CI: 1.17, 1.27] for stove use and OR: 1.67 [95 % CI: 1.58, 1.76] for grill use/frying/broiling/sauteing). Significance Overall, the algorithm developed in this study presents a data-driven approach to collecting cooking activity data in U.S. households, that may be more indicative of actual cooking events and also more predictive of household PM2.5 in indoor environmental models.

7. Neighborhood greenness and long-term physical and psychosocial quality of life among prostate cancer survivors in the Health Professionals Follow-up Study

   Naiyu Chen, Cindy R. Hu, Hari S. Iyer, and 4 more authors

   Environmental Research, Dec 2024

   Abs DOI

   Introduction Neighborhood greenness may benefit long-term prostate cancer survivorship by promoting physical activity and social integration, and reducing stress and exposure to air pollution, noise, and extreme temperatures. We examined associations of neighborhood greenness and long-term physical and psychosocial quality of life in prostate cancer survivors in the Health Professionals Follow-up Study. Methods We included 1437 individuals diagnosed with non-metastatic prostate cancer between 2008 and 2016 across the United States. Neighborhood greenness within a 1230m buffer of each individual’s mailing address was measured using the Landsat satellite image-based Normalized Difference Vegetation Index (NDVI). We fit generalized linear mixed effect models to assess associations of greenness (in quintiles) with longitudinal patient reported outcome measures on prostate cancer-specific physical and psychosocial quality of life, adjusting for time-varying individual- and neighborhood-level demographic factors and clinical factors. Results The greatest symptom burden was in the sexual domain. More than half of survivors reported good memory function and the lack of depressive signs at diagnosis. In fully adjusted models, cumulative average greenness since diagnosis was associated with fewer vitality/hormonal symptoms (highest quintile, Q5, vs lowest quintile, Q1: mean difference: 0.46, 95% confidence interval [CI]: 0.81, −0.12). Other domains of physical quality of life (bowel symptoms, urinary incontinence, urinary irritation, and sexual symptoms) did not differ by greenness overall. Psychosocial quality of life did not differ by greenness overall (Q5 vs Q1, odds ratio [95% CI]: memory function: 1.01 [0.61, 1.73]; lack of depressive signs: 1.10 [0.63, 1.95]; and wellbeing: 1.17 [0.71, 1.91]). Conclusion During long-term prostate cancer survivorship, cumulative average 1230m greenness since diagnosis was associated with fewer vitality/hormonal symptoms. Other domains of physical quality of life and psychosocial quality of life did not differ by greenness overall. Limitations included potential non-differential exposure measurement error and NDVI’s lack of time-activity pattern.

8. Gerrymandering and the Packing and Cracking of Medical Uninsurance Rates in the United States

   Tamara Rushovich, Rachel C. Nethery, Ariel White, and 1 more author

   Journal of Public Health Management and Practice, Dec 2024

   Abs DOI

   Context:  Technological innovation and access to big data have allowed partisan gerrymandering to increase dramatically in recent redistricting cycles. Objective:  To understand whether and how partisan gerrymandering, including “packing” and “cracking” (ie, respectively concentrating within or dividing specified social groups across political boundaries), distorts understanding of public health need when health statistics are calculated for congressional districts (CDs). Design:  Cross-sectional study using 2020 CDs and nonpartisan simulated districts. Setting:  United States, 2017-2021. Participants:  United States residents. Main Outcome Measure:  Percent with no medical insurance (uninsured), within-district variance of percent uninsured, and between-district variance of percent uninsured. Results:  At the state level, states where partisan redistricting plans showed greater evidence of partisan gerrymandering were more likely to contain CDs with more extreme values of uninsurance rates than districts in states with less evidence for gerrymandering (association between z-scores for gerrymandering and between-district variation in uninsurance = 0.25 (−0.04, 0.53), P = .10). Comparing variation in uninsurance rates for observed CDs vs nonpartisan simulated districts across all states with more than 1 CD, in analyses stratified by state gerrymander status (no gerrymander, Democratic gerrymander, and Republican gerrymander), we found evidence of particularly extreme distortion of rates in Republican gerrymandered states, whereby Republican-leaning districts tended to have lower uninsurance rates (the percentage of Republican-leaning districts that were significantly lower than nonpartisan simulated districts was 5.1 times that of Democratic-leaning districts) and Democrat-leaning districts had higher uninsurance rates (the percentage of Democrat-leaning districts that were significantly higher than nonpartisan simulated districts was 3.0 times that of Republican-leaning districts). Conclusions:  Partisan gerrymandering can affect determination of CD-level uninsurance rates and distort understanding of public health burdens.

9. Spatio-temporal quasi-experimental methods for rare disease outcomes: the impact of reformulated gasoline on childhood haematologic cancer

   Sofia L Vega and Rachel C Nethery

   Journal of the Royal Statistical Society Series A: Statistics in Society, Nov 2024

   Abs DOI

   Although some pollutants emitted in vehicle exhaust, such as benzene, are known to cause leukaemia in adults with high exposure levels, less is known about the relationship between traffic-related air pollution (TRAP) and childhood haematologic cancer. In the 1990s, the US EPA enacted the reformulated gasoline program in select areas of the U.S., which drastically reduced ambient TRAP in affected areas. This created an ideal quasi-experiment to study the effects of TRAP on childhood haematologic cancers. However, existing methods for quasi-experimental analyses can perform poorly when outcomes are rare and unstable, as with childhood cancer incidence. We develop Bayesian spatio-temporal matrix completion methods to conduct causal inference in quasi-experimental settings with rare outcomes. Selective information sharing across space and time enables stable estimation, and the Bayesian approach facilitates uncertainty quantification. We evaluate the methods through simulations and apply them to estimate the causal effects of TRAP on childhood leukaemia and lymphoma.

10. Increased Emergency Department Medical Imaging: Association with Short-Term Exposures to Ambient Heat and Particulate Air Pollution

    Kate Hanneman, Omar Taboun, Anish Kirpalani, and 12 more authors

    Radiology, Nov 2024

    Publisher: Radiological Society of North America

    Abs DOI

    Background Climate change adversely affects human health, resulting in higher demand for health care services. However, the impact of climate-related environmental exposures on medical imaging utilization is currently unknown. Purpose To determine associations of short-term exposures to ambient heat and particulate air pollution with utilization of emergency department medical imaging. Materials and Methods In this retrospective time-stratified case-crossover study, daily imaging utilization counts from four emergency departments were linked to local daily environmental data—including fine particulate matter with 2.5-µm or smaller aerodynamic diameter (PM2.5) and ambient temperature—over 10 years (January 2013 to December 2022). Conditional Poisson regression models were used to evaluate the associations between daily imaging utilization and environmental exposures on the same day and each of the 7 days preceding imaging, lag days 0–7, controlling for day of the week, month, and year. Moving averages of mean daily PM2.5 and temperature were calculated to account for lagged exposure effects. Imaging counts were also stratified by modality (CT, radiography, US, and MRI). Results In an analysis of 1 666 420 emergency department imaging studies, a rise of 10 °C in the 2-day moving average of mean daily temperature and a rise of 10 μg/m3 in the 3-day moving average of mean daily PM2.5 were associated with overall imaging utilization increases of 5.1% (incidence rate ratio [IRR], 1.051; 95% CI: 1.045, 1.056) and 4.0% (IRR, 1.040; 95% CI: 1.035, 1.046), respectively. Heat exposure days (mean temperature \textgreater20 °C) and air pollution exposure days (mean PM2.5 \textgreater12 μg/m3) were associated with same-day excess absolute risk of 5.5 and 6.4 imaging studies per 1 million people at risk per day, respectively. Heat exposure days and air pollution exposure days were associated with increased utilization of radiography (excess relative risk, 2.7% [P \textless .001] and 2.1% [P \textless .001], respectively) and CT (excess relative risk, 2.0% [P = .001] and 2.7% [P \textless .001]) but not US (P = .14 and P = .14) or MRI (P = .70 and P = .65). Conclusion Short-term exposures to ambient heat and particulate air pollution were associated with increased utilization of radiography and CT but not US or MRI. © RSNA, 2024 Supplemental material is available for this article. See also the editorial by Vosshenrich in this issue.

2023

1. COVID-19 and Beyond: COVID-19 Interventions and Power Plant Emissions in the United States

   Munshi Md Rasel, Kevin L. Chen, Rachel C. Nethery, and 1 more author

   ACS ES&T Engineering, Jun 2023

   Publisher: American Chemical Society

   Abs DOI

   Short- and long-term changes in electricity generating unit (EGU) emissions were observed during COVID-19 public health interventions in the United States. In a generalized synthetic control framework, we employ weekly EGU SO2, NOx, and CO2 emissions data from EPA’s Clean Air Markets Database and location-specific meteorology from 2010 to 2019 to estimate each EGU’s hypothetical business as usual (BAU) emissions throughout 2020. We find that over 60% (covering \textgreater50% of total electricity generation) of EGUs saw SO2, NOx, and CO2 emissions increases relative to BAU, with most of the increases occurring in the eastern U.S. We find increases relative to BAU in the March–April stringent lockdown period for SO2, NOx, and CO2 of 44% (4500 tons/week), 23% (2200 tons/week), and 14% (2.3 million tons/week), respectively, with similar results from March to December 2020. We find that EGUs using coal as primary fuels are the main driver of increased emissions due to increased operations, and SO2 emissions increases at coal EGUs led to a 28% increase in PM2.5 related to coal SO2 emissions relative to BAU across March–December. We find increases in SO2 and NOx emissions factors at coal EGUs in 2020 relative to 2019 that likely played a role in these increases, and we identify changes in coal fuel consumption and price that may have played a role.

2. Evaluation of Model-Based PM2.5 Estimates for Exposure Assessment during Wildfire Smoke Episodes in the Western U.S.

   Ellen M. Considine, Jiayuan Hao, Priyanka deSouza, and 3 more authors

   Environmental Science & Technology, Feb 2023

   Publisher: American Chemical Society

   DOI
3. Investigating Use of Low-Cost Sensors to Increase Accuracy and Equity of Real-Time Air Quality Information

   Ellen M. Considine, Danielle Braun, Leila Kamareddine, and 2 more authors

   Environmental Science & Technology, Jan 2023

   Publisher: American Chemical Society

   Abs DOI

   U.S. Environmental Protection Agency (EPA) air quality (AQ) monitors, the “gold standard” for measuring air pollutants, are sparsely positioned across the U.S. Low-cost sensors (LCS) are increasingly being used by the public to fill in the gaps in AQ monitoring; however, LCS are not as accurate as EPA monitors. In this work, we investigate factors impacting the differences between an individual’s true (unobserved) exposure to air pollution and the exposure reported by their nearest AQ instrument (which could be either an LCS or an EPA monitor). We use simulations based on California data to explore different combinations of hypothetical LCS placement strategies (e.g., at schools or near major roads), for different numbers of LCS, with varying plausible amounts of LCS device measurement errors. We illustrate how real-time AQ reporting could be improved (or, in some cases, worsened) by using LCS, both for the population overall and for marginalized communities specifically. This work has implications for the integration of LCS into real-time AQ reporting platforms.

4. Who benefits from shifting metal-to-pedal? Equity in the health tradeoffs of cycling

   Lindsay M. Braun, Huyen T. K. Le, Carole Turley Voulgaris, and 1 more author

   Transportation Research Part D: Transport and Environment, Feb 2023

   Abs DOI

   Health impact assessments (HIAs) have been used to evaluate the benefits and risks of cycling and other transportation interventions. Most HIAs use aggregate, city-level data rather than considering how impacts might vary across neighborhoods. To address this limitation, we developed a novel HIA framework for evaluating intra-city spatial variation and equity in the health tradeoffs of cycling. We illustrated the utility of this framework by applying it to Los Angeles, CA, estimating changes in mortality risk that might be expected from shifting a 2.5-mile daily car trip to cycling for five years. This shift was associated on average with a 12.4% net reduction in mortality risk, and a 50% increase in cycling could prevent approximately 600 deaths over five years. However, benefits were significantly lower in places with higher percentages of Black and Hispanic residents and lower socioeconomic status. To avoid widening health disparities, cycling promotion should be coupled with mitigation strategies in marginalized communities where risks are currently highest.

5. Estimating a Causal Exposure Response Function with a Continuous Error-Prone Exposure: A Study of Fine Particulate Matter and All-Cause Mortality

   Kevin P. Josey, Priyanka deSouza, Xiao Wu, and 2 more authors

   Journal of Agricultural, Biological and Environmental Statistics, Mar 2023

   Abs DOI

   Numerous studies have examined the associations between long-term exposure to fine particulate matter (PM\\_{2.5}\\) and adverse health outcomes. Recently, many of these studies have begun to employ high-resolution predicted PM\\_{2.5}\\concentrations, which are subject to measurement error. Previous approaches for exposure measurement error correction have either been applied in non-causal settings or have only considered a categorical exposure. Moreover, most procedures have failed to account for uncertainty induced by error correction when fitting an exposure response function (ERF). To remedy these deficiencies, we develop a multiple imputation framework that combines regression calibration and Bayesian techniques to estimate a causal ERF. We demonstrate how the output of the measurement error correction steps can be seamlessly integrated into a Bayesian additive regression trees (BART) estimator of the causal ERF. We also demonstrate how kernel-weighted smoothing of the posterior samples from BART can be used to create a more accurate ERF estimate. Our proposed approach also properly propagates the exposure measurement error uncertainty to yield accurate standard error estimates. We assess the robustness of our proposed approach in an extensive simulation study. We then apply our methodology to estimate the effects of PM\\_{2.5}\\on all-cause mortality among Medicare enrollees in New England from 2000 to 2012. Supplementary materials accompanying this paper appear on-line

6. Gestational thyroid hormone concentrations and risk of attention-deficit hyperactivity disorder in the Norwegian Mother, Father and Child Cohort Study

   Stephanie M. Engel, Gro D. Villanger, Amy Herring, and 9 more authors

   Paediatric and Perinatal Epidemiology, Mar 2023

   _eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/ppe.12941

   Abs DOI

   Background Maternal thyroid function plays an important role in foetal brain development; however, little consensus exists regarding the relationship between normal variability in thyroid hormones and common neurodevelopmental disorders, such as attention-deficit hyperactivity disorder (ADHD). Objective We sought to examine the association between mid-pregnancy maternal thyroid function and risk of clinically diagnosed ADHD in offspring. Methods We conducted a nested case–control study in the Norwegian Mother, Father and Child Cohort Study. Among children born 2003 or later, we randomly sampled singleton ADHD cases obtained through linkage with the Norwegian Patient Registry (n = 298) and 554 controls. Concentrations of maternal triiodothyronine (T3), thyroxine (T4), T3-Uptake, thyroid-stimulating hormone (TSH) and thyroid peroxidase antibody (TPO-Ab) were measured in maternal plasma, collected at approximately 17 weeks’ gestation. Indices of free T4 (FT4i) and free T3 (FT3i) were calculated. We used multivariable adjusted logistic regression to calculate odds ratios and accounted for missing covariate data using multiple imputation. We used restricted cubic splines to assess non-linear trends and provide flexible representations. We examined effect measure modification by dietary iodine and selenium intake. In sensitivity analyses, we excluded women with clinically significant thyroid disorders (n = 73). Results High maternal T3 was associated with increased risk of ADHD (5th vs 1st quintile odds ratio 2.27, 95% confidence interval 1.21, 4.26). For FT4i, both the lowest and highest quintiles were associated with an approximate 1.6-fold increase in risk of ADHD, with similar trends found for T4. The FT4i association was modified by dietary iodine intake such that the highest risk strata were confined to the low intake group. Conclusions Both high and low concentrations of maternal thyroid hormones, although within population reference ranges, increase the risk of ADHD in offspring. Increased susceptibility may be found among women with low dietary intake of iodine and selenium.

7. Mobile Source Benzene Regulations and Risk of Childhood and Young Adult Hematologic Cancers in Alaska: A Quasi-experimental Study

   Rachel C. Nethery, Sofia Vega, A. Lindsay Frazier, and 1 more author

   Epidemiology, May 2023

   DOI
8. Air Pollution and Mortality at the Intersection of Race and Social Class

   Kevin P. Josey, Scott W. Delaney, Xiao Wu, and 4 more authors

   New England Journal of Medicine, Apr 2023

   Publisher: Massachusetts Medical Society _eprint: https://doi.org/10.1056/NEJMsa2300523

   DOI
9. Air Pollution and Cardiovascular and Thromboembolic Events in Older Adults with High-Risk Conditions

   Rachel C Nethery, Kevin Josey, Poonam Gandhi, and 6 more authors

   American Journal of Epidemiology, Apr 2023

   Abs DOI

   Little epidemiologic research has focused on pollution-related risks in medically vulnerable or marginalized groups. Using a nationwide 50% random sample of 2008-2016 Medicare Part D-eligible Fee-for-Service participants in the US, we identified a cohort with high-risk conditions for cardiovascular and thromboembolic events (CTE) and linked individuals with seasonal average zip code level concentrations of fine particulate matter (PM2.5). We assessed the relationship between seasonal PM2.5 exposure and hospitalization for each of seven CTE-related causes using history-adjusted marginal structural models adjusted for individual demographic and neighborhood socio-economic variables as well as baseline comorbidities, health behaviors, and health service measures. We examined effect modification across geographically- and demographically-defined subgroups. The cohort included 1,934,453 individuals with high-risk conditions (mean age 77, 60% female, 87% white). A 1 μg/m3 increase in PM2.5 exposure was significantly associated with increased risk of six out of seven CTE hospitalization types. Strong increases were observed for transient ischemic attack (HR: 1.039 (1.034,1.044)), venous thromboembolism (HR: 1.031 (1.027,1.035)), and heart failure (HR: 1.019 (1.017,1.020)). Asian Americans were found to be particularly susceptible to thromboembolic effects of PM2.5 (venous thromboembolism HR: 1.063 (1.021,1.106)), while Native Americans were most vulnerable to cerebrovascular effects (transient ischemic attack HR: 1.093 (1.030,1.161)).

10. Integrated causal-predictive machine learning models for tropical cyclone epidemiology

    Rachel C Nethery, Nina Katz-Christy, Marianthi-Anna Kioumourtzoglou, and 3 more authors

    Biostatistics, Apr 2023

    Abs DOI

    Strategic preparedness reduces the adverse health impacts of hurricanes and tropical storms, referred to collectively as tropical cyclones (TCs), but its protective impact could be enhanced by a more comprehensive and rigorous characterization of TC epidemiology. To generate the insights and tools necessary for high-precision TC preparedness, we introduce a machine learning approach that standardizes estimation of historic TC health impacts, discovers common patterns and sources of heterogeneity in those health impacts, and enables identification of communities at highest health risk for future TCs. The model integrates (i) a causal inference component to quantify the immediate health impacts of recent historic TCs at high spatial resolution and (ii) a predictive component that captures how TC meteorological features and socioeconomic/demographic characteristics of impacted communities are associated with health impacts. We apply it to a rich data platform containing detailed historic TC exposure information and records of all-cause mortality and cardiovascular- and respiratory-related hospitalization among Medicare recipients. We report a high degree of heterogeneity in the acute health impacts of historic TCs, both within and across TCs, and, on average, substantial TC-attributable increases in respiratory hospitalizations. TC-sustained windspeeds are found to be the primary driver of mortality and respiratory risks.

11. Association between Outdoor Light at Night and Prostate Cancer in the Health Professionals Follow-up Study

    Ilkania M. Chowdhury-Paulino, Jaime E. Hart, Peter James, and 7 more authors

    Cancer Epidemiology, Biomarkers & Prevention, Oct 2023

    Abs DOI

    Circadian disruption is a potential risk factor for advanced prostate cancer, and light at night (LAN) exposure may disrupt circadian rhythms. We evaluated whether outdoor LAN increases the risk of prostate cancer.We prospectively followed 49,148 participants in the Health Professionals Follow-up Study from 1986 through 2016. We estimated baseline and cumulative time-varying outdoor LAN with ∼1 km2 resolution using data from the US Defense Meteorological Satellite Program’s Operational Linescan System, which was assigned to participants’ geocoded addresses. Participants reside in all 50 U.S. states and reported a work or home address. We used multivariable Cox models to estimate HRs and 95% confidence intervals (CI) for the association between outdoor LAN and risk of overall (7,175 cases) and fatal (915 cases) prostate cancer adjusting for individual and contextual factors.There was no association between the interquartile range increase in cumulative LAN and total (HR, 1.02; 95% CI, 0.98–1.06) or fatal (HR, 1.05; 95% CI, 0.96–1.15) prostate cancer in adjusted models. However, there was a positive association between baseline LAN and total prostate cancer among non-movers (HR, 1.06; 95% CI, 1.00–1.14) including among highly screened participants (HR, 1.11; 95% CI, 1.01–1.23).There was a suggestive positive association between baseline outdoor LAN and total prostate cancer. Additional studies with different measures of outdoor LAN and in more diverse populations are necessary.To our knowledge, this is the first longitudinal cohort study exploring the relationship between outdoor LAN and prostate cancer.

12. Associations between air pollution, residential greenness, and glycated hemoglobin (HbA1c) in three prospective cohorts of U.S. adults

    Melissa R. Fiffer, Huichu Li, Hari S. Iyer, and 7 more authors

    Environmental Research, Dec 2023

    Abs DOI

    Background While studies suggest impacts of individual environmental exposures on type 2 diabetes (T2D) risk, mechanisms remain poorly characterized. Glycated hemoglobin (HbA1c) is a biomarker of glycemia and diagnostic criterion for prediabetes and T2D. We explored associations between multiple environmental exposures and HbA1c in non-diabetic adults. Methods HbA1c was assessed once in 12,315 women and men in three U.S.-based prospective cohorts: the Nurses’ Health Study (NHS), Nurses’ Health Study II (NHSII), and Health Professionals Follow-up Study (HPFS). Residential greenness within 270 m and 1,230 m (normalized difference vegetation index, NDVI) was obtained from Landsat. Fine particulate matter (PM2.5) and nitrogen dioxide (NO2) were estimated from nationwide spatiotemporal models. Three-month and one-year averages prior to blood draw were assigned to participants’ addresses. We assessed associations between single exposure, multi-exposure, and component scores from Principal Components Analysis (PCA) and HbA1c. Fully-adjusted models built on basic models of age and year at blood draw, BMI, alcohol use, and neighborhood socioeconomic status (nSES) to include diet quality, race, family history, smoking status, postmenopausal hormone use, population density, and season. We assessed interactions between environmental exposures, and effect modification by population density, nSES, and sex. Results Based on HbA1c, 19% of participants had prediabetes. In single exposure fully-adjusted models, an IQR (0.14) higher 1-year 1,230 m NDVI was associated with a 0.27% (95% CI: 0.05%, 0.49%) lower HbA1c. In basic component score models, a SD increase in Component 1 (high loadings for 1-year NDVI) was associated with a 0.19% (95% CI: 0.04%, 0.34%) lower HbA1c. CI’s crossed the null in multi-exposure and fully-adjusted component score models. There was little evidence of associations between air pollution and HbA1c, and no evidence of effect modification. Conclusions Among non-diabetic adults, environmental exposures were not consistently associated with HbA1c. More work is needed to elucidate biological pathways between the environment and prediabetes.

13. Air Pollution and Temperature: a Systematic Review of Ubiquitous Environmental Exposures and Sudden Cardiac Death

    William Borchert, Stephanie T. Grady, Jie Chen, and 12 more authors

    Current Environmental Health Reports, Oct 2023

    Abs DOI

    Environmental exposures have been associated with increased risk of cardiovascular mortality and acute coronary events, but their relationship with out-of-hospital cardiac arrest (OHCA) and sudden cardiac death (SCD) remains unclear. SCD is an important contributor to the global burden of cardiovascular disease worldwide.

14. Impacts of census differential privacy for small-area disease mapping to monitor health inequities

    Yanran Li, Brent A. Coull, Nancy Krieger, and 4 more authors

    Science Advances, Aug 2023

    Publisher: American Association for the Advancement of Science

    Abs DOI

    The U.S. Census Bureau will implement a modernized privacy-preserving disclosure avoidance system (DAS), which includes application of differential privacy, on publicly released 2020 census data. There are concerns that the DAS may bias small-area and demographically stratified population counts, which play a critical role in public health research, serving as denominators in estimation of disease/mortality rates. Using three DAS demonstration products, we quantify errors attributable to reliance on DAS-protected denominators in standard small-area disease mapping models for characterizing health inequities. We conduct simulation studies and real data analyses of inequities in premature mortality at the census tract level in Massachusetts and Georgia. Results show that overall patterns of inequity by racialized group and economic deprivation level are not compromised by the DAS. While early versions of DAS induce errors in mortality rate estimation that are larger for Black than non-Hispanic white populations in Massachusetts, this issue is ameliorated in newer DAS versions.

15. Retrospective cohort study investigating synergism of air pollution and corticosteroid exposure in promoting cardiovascular and thromboembolic events in older adults

    Kevin Josey, Rachel Nethery, Aayush Visaria, and 6 more authors

    BMJ Open, Sep 2023

    Publisher: British Medical Journal Publishing Group Section: Epidemiology

    Abs DOI

    Objective To evaluate the synergistic effects created by fine particulate matter (PM2.5) and corticosteroid use on hospitalisation and mortality in older adults at high risk for cardiovascular thromboembolic events (CTEs). Design and setting A retrospective cohort study using a US nationwide administrative healthcare claims database. Participants A 50% random sample of participants with high-risk conditions for CTE from the 2008–2016 Medicare Fee-for-Service population. Exposures Corticosteroid therapy and seasonal-average PM2.5. Main outcome measures Incidences of myocardial infarction or acute coronary syndrome (MI/ACS), ischaemic stroke or transient ischaemic attack, heart failure (HF), venous thromboembolism, atrial fibrillation and all-cause mortality. We assessed additive interactions between PM2.5 and corticosteroids using estimates of the relative excess risk due to interaction (RERI) obtained using marginal structural models for causal inference. Results Among the 1 936 786 individuals in the high CTE risk cohort (mean age 76.8, 40.0% male, 87.4% white), the mean PM2.5 exposure level was 8.3±2.4 µg/m3 and 37.7% had at least one prescription for a systemic corticosteroid during follow-up. For all outcomes, we observed increases in risk associated with corticosteroid use and with increasing PM2.5 exposure. PM2.5 demonstrated a non-linear relationship with some outcomes. We also observed evidence of an interaction existing between corticosteroid use and PM2.5 for some CTEs. For an increase in PM2.5 from 8 μg/m3 to 12 μg/m3 (a policy-relevant change), the RERI of corticosteroid use and PM2.5 was significant for HF (15.6%, 95% CI 4.0%, 27.3%). Increasing PM2.5 from 5 μg/m3 to 10 μg/m3 yielded significant RERIs for incidences of HF (32.4; 95% CI 14.9%, 49.9%) and MI/ACSs (29.8%; 95% CI 5.5%, 54.0%). Conclusion PM2.5 and systemic corticosteroid use were independently associated with increases in CTE hospitalisations. We also found evidence of significant additive interactions between the two exposures for HF and MI/ACSs suggesting synergy between these two exposures.

16. Ambient Heat and Risk of Serious Hypoglycemia in Older Adults With Diabetes Using Insulin in the U.S. and Taiwan: A Cross-National Case-Crossover Study

    Aayush Visaria, Shu-Ping Huang, Chien-Chou Su, and 12 more authors

    Diabetes Care, Dec 2023

    Abs DOI

    To measure the association between ambient heat and hypoglycemia-related emergency department visit or hospitalization in insulin users.We identified cases of serious hypoglycemia among adults using insulin aged ≥65 in the U.S. (via Medicare Part A/B/D-eligible beneficiaries) and Taiwan (via National Health Insurance Database) from June to September, 2016–2019. We then estimated odds of hypoglycemia by heat index (HI) percentile categories using conditional logistic regression with a time-stratified case-crossover design.Among ∼2 million insulin users in the U.S. (32,461 hypoglycemia case subjects), odds ratios of hypoglycemia for HI >99th, 95–98th, 85–94th, and 75–84th percentiles compared with the 25–74th percentile were 1.38 (95% CI, 1.28–1.48), 1.14 (1.08–1.20), 1.12 (1.08–1.17), and 1.09 (1.04–1.13) respectively. Overall patterns of associations were similar for insulin users in the Taiwan sample (∼283,000 insulin users, 10,162 hypoglycemia case subjects).In two national samples of older insulin users, higher ambient temperature was associated with increased hypoglycemia risk.

2022

1. Statistical Implications of Endogeneity Induced by Residential Segregation in Small-Area Modeling of Health Inequities

   Rachel C. Nethery, Jarvis T. Chen, Nancy Krieger, and 4 more authors

   The American Statistician, Apr 2022

   Publisher: Taylor & Francis _eprint: https://doi.org/10.1080/00031305.2021.2003245

   Abs DOI

   Health inequities are assessed by health departments to identify social groups disproportionately burdened by disease and by academic researchers to understand how social, economic, and environmental inequities manifest as health inequities. To characterize inequities, group-specific small-area health data are often modeled using log-linear generalized linear models (GLM) or generalized linear mixed models (GLMM) with a random intercept. These approaches estimate the same marginal rate ratio comparing disease rates across groups under standard assumptions. Here we explore how residential segregation combined with social group differences in disease risk can lead to contradictory findings from the GLM and GLMM. We show that this occurs because small-area disease rate data collected under these conditions induce endogeneity in the GLMM due to correlation between the model’s offset and random effect. This results in GLMM estimates that represent conditional rather than marginal associations. We refer to endogeneity arising from the offset, which to our knowledge has not been noted previously, as “offset endogeneity.” We illustrate this phenomenon in simulated data and real premature mortality data, and we propose alternative modeling approaches to address it. We also introduce to a statistical audience the social epidemiologic terminology for framing health inequities, which enables responsible interpretation of results.

2. Ultraviolet radiation and age at natural menopause in a nationwide, prospective US cohort

   Huichu Li, Jaime E. Hart, Shruthi Mahalingaiah, and 4 more authors

   Environmental Research, Jan 2022

   Abs DOI

   Background Solar ultraviolet radiation (UV) is a critical environmental factor for dermal conversion of vitamin D, which is suggested to support reproductive health. However, current epidemiological studies have reported conflicting results on the associations between vitamin D levels and ovarian reserve. Further, few studies have considered UV exposure and reproductive aging, which is closely related to declined ovarian reserve. Objectives We sought to examine the associations of long-term UV exposure and age at natural menopause in a large, nationwide, prospective cohort. Methods Participants in the Nurses’ Health Study II (NHS II) who were premenopausal at age 40 were included and followed through 2015. Erythemal UV radiation from a high-resolution geospatial model was linked to the participants’ residential histories. Early-life UV was estimated using the reported state of residence at birth, age 15, and age 30. We used time-varying Cox proportional hazards models to estimate the hazard ratio (HR) and 95% confidence intervals (CIs) for natural menopause, adjusting for potential confounders and predictors of menopause. Results A total of 63,801 women reported natural menopause across the 1,051,185 person-years of follow-up among 105,631 eligible participants. We found very modest associations with delayed menopause for long-term UV exposure (adjusted HR comparing highest to lowest quartile of cumulative average UV: 0.96, 95% CI: 0.94, 0.99). There was a suggestive inverse association between UV at age 30 with menopause (adjusted HR comparing highest to lowest quartile: 0.97, 95% CI: 0.95, 1.00) but not with UV at birth and age 15. Conclusions Solar UV exposure in adulthood was modestly associated with later onset of menopause. Although consistent with previous findings on vitamin D intake and menopause in the same population, these weak associations found in this study may not be of clinical relevance.

3. Protective Behaviors Associated With Gender During the 2018-2020 Ebola Outbreak in Eastern Democratic Republic of the Congo

   Phuong N. Pham, Manasi Sharma, Kennedy Kihangi Bindu, and 4 more authors

   JAMA Network Open, Feb 2022

   Abs DOI

   In 2018 to 2020, the Democratic Republic of the Congo experienced the world’s second largest Ebola virus disease (EVD) outbreak, killing 2290 individuals; women were disproportionately infected (57% of all cases) despite no evidence of differential biological EVD risk. Understanding how gender norms may influence exposure to EVD, intensity, and prognosis as well as personal protective behaviors against the virus is important to disease risk reduction and control interventions.To assess whether men and women differ in personal protective behaviors (vaccine acceptance, health-seeking behaviors, physical distancing) and the mediating role of EVD information and knowledge, perceived disease risk, and social relations.This cross-sectional, multistage cluster survey study of 1395 randomly selected adults was conducted in the Ebola-affected regions of North Kivu from April 20, 2019, to May 10, 2019. Path analyses were conducted using structural equation modeling to examine associations among study variables. Statistical analysis was conducted from August 2019 to May 2020.The main behavioral outcomes of interest were (1) vaccine acceptance, (2) formal health care seeking, and (3) self-protective behaviors. The primary factor of interest was self-reported gender identity. We also assessed sociodemographic factors.Among the study’s 1395 participants, 1286 (93%) had Nande ethnicity and 698 (50%) were women; the mean (SD) age was 34.5 (13.1) years. Compared with female participants, male participants reported significantly higher levels of education, wealth, and mobile phone access. There were associations found between gender and all EVD preventive behavioral outcomes, with evidence for mediation through EVD knowledge and belief in rumors. Men reported greater EVD knowledge accuracy compared with women (mean [SE] score for men: 12.06 [0.13] vs women: 11.08 [0.16]; P < .001), and greater knowledge accuracy was associated with increases in vaccine acceptance (β = 0.37; P < .001), formal care seeking (β = 0.39; P < .001), and self-protective behaviors (β = 0.35; P < .001). Lower belief in rumors was associated with greater vaccine acceptance (β = −0.30; P < .001), and greater EVD information awareness was associated with increased adoption of self-protective behaviors (β = 0.23; P < .001).This survey study found gender differences in adopting preventive protective behaviors against EVD. These findings suggest that it is critical to design gender-sensitive communication and vaccination strategies, while engaging women and their community as a whole in any response to infectious disease outbreaks. Research on the potential link between gender and sociodemographics factors associated with disease risk and outcomes is needed.

4. Association of Tropical Cyclones With County-Level Mortality in the US

   Robbie M. Parks, Jaime Benavides, G. Brooke Anderson, and 5 more authors

   JAMA, Mar 2022

   Abs DOI

   Tropical cyclones have a devastating effect on society, but a comprehensive assessment of their association with cause-specific mortality over multiple years of study is lacking.To comprehensively evaluate the association of county-level tropical cyclone exposure and death rates from various causes in the US.A retrospective observational study using a Bayesian conditional quasi-Poisson model to examine how tropical cyclones were associated with monthly death rates. Data from 33.6 million deaths in the US were collected from the National Center for Health Statistics over 31 years (1988-2018), including residents of the 1206 counties in the US that experienced at least 1 tropical cyclone during the study period.Tropical cyclone days per county-month, defined as number of days in a month with a sustained maximal wind speed 34 knots or greater.Monthly cause-specific county-level death rates by 6 underlying causes of death: cancers, cardiovascular diseases, infectious and parasitic diseases, injuries, neuropsychiatric conditions, and respiratory diseases. The model yielded information about the association between each additional cyclone day per month and monthly county-level mortality compared with the same county-month in different years, up to 6 months after tropical cyclones, and how these estimated associations varied by age, sex, and social vulnerability. The unit of analysis was county-month.There were 33 619 393 deaths in total (16 691 681 females and 16 927 712 males; 8 587 033 aged 0-64 years and 25 032 360 aged 65 years or older) from the 6 causes recorded in 1206 US counties. There was a median of 2 tropical cyclone days experienced in total in included US counties. Each additional cyclone day was associated with increased death rates in the month following the cyclone for injuries (3.7% [95% credible interval {CrI}, 2.5%-4.9%]; 2.0 [95% CrI, 1.3-2.7] additional deaths per 1 000 000 for 2018 monthly age-standardized median rate [DPM]; 54.3 to 56.3 DPM), infectious and parasitic diseases (1.8% [95% CrI, 0.1%-3.6%]; 0.2 [95% CrI, 0.0-0.4] additional DPM; 11.7 to 11.9 DPM), respiratory diseases (1.3% [95% CrI, 0.2%-2.4%]; 0.6 [95% CrI, 0.1-1.1] additional DPM; 44.9 to 45.5 DPM), cardiovascular diseases (1.2% [95% CrI, 0.6%-1.7%]; 1.5 [95% CrI, 0.8-2.2] additional DPM; 129.6 to 131.1 DPM), neuropsychiatric conditions (1.2% [95% CrI, 0.1%-2.4%]; 0.6 [95% CrI, 0.1-1.2] additional DPM; 52.1 to 52.7 DPM), with no change for cancers (−0.3% [95% CrI, −0.9% to 0.3%]; −0.3 [95% CrI, −0.9 to 0.3] additional DPM; 100.4 to 100.1 DPM).Among US counties that experienced at least 1 tropical cyclone from 1988-2018, each additional cyclone day per month was associated with modestly higher death rates in the months following the cyclone for several causes of death, including injuries, infectious and parasitic diseases, cardiovascular diseases, neuropsychiatric conditions, and respiratory diseases.

5. Community solutions to food apartheid: A spatial analysis of community food-growing spaces and neighborhood demographics in Philadelphia

   Ashley B. Gripper, Rachel Nethery, Tori L. Cowger, and 3 more authors

   Social Science & Medicine, Oct 2022

   Abs DOI

   Black and low-income neighborhoods tend to have higher concentrations of fast-food restaurants and low produce supply stores. Limited access to and consumption of nutrient-rich foods is associated with poor health outcomes. Given the realities of food access, many members within the Black communities grow food as a strategy of resistance to food apartheid, and for the healing and self-determination that agriculture offers. In this paper, we unpack the history of Black people, agriculture, and land in the United States. In addition to our brief historical review, we conduct a descriptive epidemiologic study of community food-growing spaces, food access, and neighborhood racial composition in present day Philadelphia. We leverage one of the few existing datasets that systematically documents community food-growing locations throughout a major US city. By applying spatial regression techniques, we use conditional autoregressive models to determine if there are spatial associations between Black neighborhoods, poverty, food access, and urban agriculture in Philadelphia. Fully adjusted spatial models showed significant associations between Black neighborhoods and urban agriculture (RR: 1.28, 95% CI = 1.03, 1.59) and poverty and urban agriculture (RR: 1.27, 95% CI = 1.1, 1.46). The association between low food access and the presence of urban agriculture was generally increased across neighborhoods with a higher proportion of Black residents. These results show that Philadelphia neighborhoods with higher populations of Black people and neighborhoods with lower incomes, on average, tend to have more community gardens and urban farms. While the garden data is non-temporal and non-causal, one possible explanation for these findings, in alignment with what Philadelphia growers have claimed, is that urban agriculture may be a manifestation of collective agency and community resistance in Black and low-income communities, particularly in neighborhoods with low food access.

6. Environmental Exposures and Anti-Müllerian Hormone: A Mixture Analysis in the Nurses’ Health Study II

   Huichu Li, Jaime E. Hart, Shruthi Mahalingaiah, and 5 more authors

   Epidemiology, Nov 2022

   Abs DOI

   Background: Previous studies have linked environmental exposures with anti-Müllerian hormone (AMH), a marker of ovarian reserve. However, associations with multiple environment factors has to our knowledge not been addressed. Methods: We included a total of 2,447 premenopausal women in the Nurses’ Health Study II (NHSII) who provided blood samples during 1996–1999. We selected environmental exposures linked previously with reproductive outcomes that had measurement data available in NHSII, including greenness, particulate matter, noise, outdoor light at night, ultraviolet radiation, and six hazardous air pollutants (1,3-butadiene, benzene, diesel particulate matter, formaldehyde, methylene chloride, and tetrachloroethylene). For these, we calculated cumulative averages from enrollment (1989) to blood draw and estimated associations with AMH in adjusted single-exposure models, principal component analysis (PCA), and hierarchical Bayesian kernel machine regression (BKMR). Results: Single-exposure models showed negative associations of AMH with benzene (percentage reduction in AMH per interquartile range [IQR] increase = 5.5%, 95% confidence interval [CI] = 1.0, 9.8) and formaldehyde (6.1%, 95% CI = 1.6, 10). PCA identified four major exposure patterns but only one with high exposure to air pollutants and light at night was associated with lower AMH. Hierarchical BKMR pointed to benzene, formaldehyde, and greenness and suggested an inverse joint association with AMH (percentage reduction comparing all exposures at the 75th percentile to median = 8.2%, 95% CI = 0.7, 15.1). Observed associations were mainly among women above age 40. Conclusions: We found exposure to benzene and formaldehyde to be consistently associated with lower AMH levels. The associations among older women are consistent with the hypothesis that environmental exposures accelerate reproductive aging.

2021

1. Evaluation of the health impacts of the 1990 Clean Air Act Amendments using causal inference and machine learning

   Rachel C. Nethery, Fabrizia Mealli, Jason D. Sacks, and 1 more author

   Journal of the American Statistical Association, Jul 2021

   Publisher: Taylor & Francis _eprint: https://doi.org/10.1080/01621459.2020.1803883

   Abs DOI

   We develop a causal inference approach to estimate the number of adverse health events that were prevented due to changes in exposure to multiple pollutants attributable to a large-scale air quality intervention/regulation, with a focus on the 1990 Clean Air Act Amendments (CAAA). We introduce a causal estimand called the Total Events Avoided (TEA) by the regulation, defined as the difference in the number of health events expected under the no-regulation pollution exposures and the number observed with-regulation. We propose matching and machine learning methods that leverage population-level pollution and health data to estimate the TEA. Our approach improves upon traditional methods for regulation health impact analyses by formalizing causal identifying assumptions, utilizing population-level data, minimizing parametric assumptions, and collectively analyzing multiple pollutants. To reduce model-dependence, our approach estimates cumulative health impacts in the subset of regions with projected no-regulation features lying within the support of the observed with-regulation data, thereby providing a conservative but data-driven assessment to complement traditional parametric approaches. We analyze the health impacts of the CAAA in the US Medicare population in the year 2000, and our estimates suggest that large numbers of cardiovascular and dementia-related hospitalizations were avoided due to CAAA-attributable changes in pollution exposure.

2. Long-term exposure to particulate matter and roadway proximity with age at natural menopause in the Nurses’ Health Study II Cohort

   Huichu Li, Jaime E. Hart, Shruthi Mahalingaiah, and 3 more authors

   Environmental Pollution, Jan 2021

   Abs DOI

   Evidence has shown associations between air pollution and traffic-related exposure with accelerated aging, but no study to date has linked the exposure with age at natural menopause, an important indicator of reproductive aging. In this study, we sought to examine the associations of residential exposure to ambient particulate matter (PM) and distance to major roadways with age at natural menopause in the Nurses’ Health Study II (NHS II), a large, prospective female cohort in US. A total of 105,996 premenopausal participants in NHS II were included at age 40 and followed through 2015. Time-varying residential exposures to PM10, PM2.5-10, and PM2.5 and distance to roads was estimated. We calculated hazard ratios (HR) and 95% confidence intervals (CIs) for natural menopause using Cox proportional hazard models adjusting for potential confounders and predictors of age at menopause. We also examined effect modification by region, smoking, body mass, physical activity, menstrual cycle length, and population density. There were 64,340 reports of natural menopause throughout 1,059,229 person-years of follow-up. In fully adjusted models, a 10 μg/m3 increase in the cumulative average exposure to PM10 (HR: 1.02, 95% CI: 1.00, 1.04), PM2.5-10 (HR: 1.03, 95% CI: 1.00, 1.05), and PM2.5 (HR: 1.03, 95% CI: 1.00, 1.06) and living within 50 m to a major road at age 40 (HR: 1.03, 95%CI: 1.00, 1.06) were associated with slightly earlier menopause. No statistically significant effect modification was found, although the associations of PM were slightly stronger for women who lived in the West and for never smokers. To conclude, we found exposure to ambient PM and traffic in midlife was associated with slightly earlier onset of natural menopause. Our results support previous evidence that exposure to air pollution and traffic may accelerate reproductive aging.

3. Impact of Differential Privacy and Census Tract Data Source (Decennial Census Versus American Community Survey) for Monitoring Health Inequities

   Nancy Krieger, Rachel C. Nethery, Jarvis T. Chen, and 4 more authors

   American Journal of Public Health, Feb 2021

   Publisher: American Public Health Association

   Abs DOI

   Objectives. To investigate how census tract (CT) estimates of mortality rates and inequities are affected by (1) differential privacy (DP), whereby the public decennial census (DC) data are injected with statistical “noise” to protect individual privacy, and (2) uncertainty arising from the small number of different persons surveyed each year in a given CT for the American Community Survey (ACS). Methods. We compared estimates of the 2008–2012 average annual premature mortality rate (death before age 65 years) in Massachusetts using CT data from the 2010 DC, 2010 DC with DP, and 2008–2012 ACS 5-year estimate data. Results. For these 3 denominator sources, the age-standardized premature mortality rates (per 100 000) for the total population respectively equaled 166.4 (95% confidence interval [CI] = 162.2, 170.6), 166.4 (95% CI = 162.2, 170.6), and 166.3 (95% CI = 162.1, 170.5), and inequities in the range from best to worst quintile for CT racialized economic segregation were from 103.4 to 260.1, 102.9 to 258.7, and 102.8 to 262.4. Similarity of results across CT denominator sources held for analyses stratified by gender and race/ethnicity. Conclusions. Estimates of health inequities at the CT level may not be affected by use of 2020 DP data and uncertainty in the ACS data.

4. PM2.5 and hospital admissions among Medicare enrollees with chronic debilitating brain disorders

   Maayan Yitshak-Sade, Rachel Nethery, Joel D. Schwartz, and 6 more authors

   Science of The Total Environment, Feb 2021

   Abs DOI

   Background Although long-term exposure to particulate matter\textless2.5 μm (PM2.5) has been linked to chronic debilitating brain disorders (CDBD), the role of short-term exposure in health care demand, and increased susceptibility for PM2.5-related health conditions, among Medicare enrollees with CDBD has received little attention. We used a causal modeling approach to assess the effect of short-term high PM2.5 exposure on all-cause admissions, and prevalent cause-specific admissions among Medicare enrollees with CDBD (Parkinson’s disease-PD, Alzheimer’s disease-AD and other dementia). Methods We constructed daily zipcode counts of hospital admissions of Medicare beneficiaries older than 65 across the United-States (2000–2014). We obtained daily PM2.5 estimates from a satellite-based model. A propensity score matching approach was applied to match high-pollution (PM2.5 \textgreater 17.4 μg/m3) to low-pollution zip code-days with similar background characteristics. Then, we estimated the percent change in admissions attributable to high pollution. We repeated the models restricting the analysis to zipcode-days with PM2.5 below of 35 μg/m3. Results We observed significant increases in all-cause hospital admissions (2.53% in PD and 2.49% in AD/dementia) attributable to high PM2.5 exposure. The largest observed effect for common causes was for pneumonia and urinary tract infection. All the effects were larger in CDBD compared to the general Medicare population, and similarly strong at levels of exposure considered safe by the EPA. Conclusion We found Medicare beneficiaries with CDBD to be at higher risk of being admitted to the hospital following acute exposure to PM2.5 levels well below the National Ambient Air Quality Standard defined as safe by the EPA.

5. Tropical cyclone exposure is associated with increased hospitalization rates in older adults

   Robbie M. Parks, G. Brooke Anderson, Rachel C. Nethery, and 3 more authors

   Nature Communications, Mar 2021

   Number: 1 Publisher: Nature Publishing Group

   Abs DOI

   Hurricanes and other tropical cyclones have devastating effects on society. Previous case studies have quantified their impact on some health outcomes for particular tropical cyclones, but a comprehensive assessment over longer periods is currently missing. Here, we used data on 70 million Medicare hospitalizations and tropical cyclone exposures over 16 years (1999–2014). We formulated a conditional quasi-Poisson model to examine how tropical cyclone exposure (days greater than Beaufort scale gale-force wind speed; ≥34 knots) affect hospitalizations for 13 mutually-exclusive, clinically-meaningful causes. We found that tropical cyclone exposure was associated with average increases in hospitalizations from several causes over the week following exposure, including respiratory diseases (14.2%; 95% confidence interval [CI]: 10.9–17.9%); infectious and parasitic diseases (4.3%; 95%CI: 1.2–8.1%); and injuries (8.7%; 95%CI: 6.0–11.8%). Average decadal tropical cyclone exposure in all impacted counties would be associated with an estimated 16,772 (95%CI: 8,265–25,278) additional hospitalizations. Our findings demonstrate the need for targeted preparedness strategies for hospital personnel before, during, and after tropical cyclones.

6. Prenatal phthalate exposures and executive function in preschool children

   Giehae Choi, Gro D. Villanger, Samantha S. M. Drover, and 12 more authors

   Environment International, Apr 2021

   Abs DOI

   Background Prenatal phthalate exposure has been linked with altered neurodevelopment, including externalizing behaviors and attention-deficit hyperactivity disorder (ADHD). However, the implicated metabolite, neurobehavioral endpoint, and child sex have not always been consistent across studies, possibly due to heterogeneity in neurodevelopmental instruments. The complex set of findings may be synthesized using executive function (EF), a construct of complex cognitive processes that facilitate ongoing goal-directed behaviors. Impaired EF can be presented with various phenotypes of poor neurodevelopment, differently across structured conditions, home/community, or preschool/school. We evaluated the relationship between prenatal phthalate exposure and comprehensive assessment of preschool EF. Methods Our study comprised 262 children with clinically significant/subthreshold ADHD symptoms and 78 typically developing children who were born between 2003 and 2008 and participated in the Preschool ADHD Substudy, which is nested within a population-based prospective cohort study, the Norwegian Mother, Father, and Child Cohort (MoBa). Twelve phthalate metabolites were measured from urine samples that their mothers had provided during pregnancy, at 17 weeks’ gestation. All children, at approximately 3.5-years, took part in a detailed clinical assessment that included parent-and teacher-rated inventories and administered tests. We used instruments that measured constructs related to EF, which include a parent-and teacher-reported Behavior Rating Inventory of Executive Function-Preschool (BRIEF-P) and three performance-based tests: A Developmental NEuroPSYchological Assessment (NEPSY), Stanford-Binet intelligence test V (SB5), and the cookie delay task (CDT). The standard deviation change in test score per interquartile range (IQR) increase in phthalate metabolite was estimated with multivariable linear regression. We applied weighting in all models to account for the oversampling of children with clinically significant or subthreshold symptoms of ADHD. Additionally, we assessed modification by child sex and potential co-pollutant confounding. Results Elevated exposure to mono-benzyl phthalate (MBzP) during pregnancy was associated with poorer EF, across all domains and instruments, in both sex. For example, an IQR increase in MBzP was associated with poorer working memory rated by parent (1.23 [95% CI: 0.20, 2.26]) and teacher (1.13 [0.14, 2.13]) using BRIEF-P, and administered tests such as SB5 (no-verbal: 0.19 [0.09, 0.28]; verbal: 0.13 [0.01, 0.25]). Adverse associations were also observed for mono-n-butyl phthalate (MnBP) and mono-iso-butyl phthalate (MiBP), although results varied by instruments. EF domains reported by parents using BRIEF-P were most apparently implicated, with stronger associations among boys (e.g., MnBP and inhibition: 2.74 [1.77, 3.72]; MiBP and inhibition: 1.88 [0.84, 2.92]) than among girls (e.g., MnBP and inhibition: −0.63 [−2.08, 0.83], interaction p-value: 0.04; MiBP and inhibition: −0.15 [−1.04, 0.74], interaction p-value: 0.3). Differences by sex, however, were not found for the teacher-rated BRIEF-P or administered tests including NEPSY, SB5, and CDT. Conclusion and relevance Elevated mid-pregnancy MBzP, MiBP, and MnBP were associated with more adverse profiles of EF among preschool-aged children across a range of instruments and raters, with some associations found only among boys. Given our findings and accumulating evidence of the prenatal period as a critical window for phthalate exposure, there is a timely need to expand the current phthalate regulations focused on baby products to include pregnancy exposures.

7. Comparing denominator sources for real-time disease incidence modeling: American Community Survey and WorldPop

   Rachel C. Nethery, Tamara Rushovich, Emily Peterson, and 5 more authors

   SSM - Population Health, Jun 2021

   Abs DOI

   Across the United States public health community in 2020, in the midst of a pandemic and increased concern regarding racial/ethnic health disparities, there is widespread concern about our ability to accurately estimate small-area disease incidence rates due to the absence of a recent census to obtain reliable population denominators. 2010 decennial census data are likely outdated, and intercensal population estimates from the Census Bureau, which are less temporally misaligned with real-time disease incidence data, are not recommended for use with small areas. Machine learning-based population estimates are an attractive option but have not been validated for use in epidemiologic studies. Treating 2010 decennial census counts as a “ground truth”, we conduct a case study to compare the performance of alternative small-area population denominator estimates from surrounding years for modeling real-time disease incidence rates. Our case study focuses on modeling health disparities in census tract incidence rates in Massachusetts, using population size estimates from the American Community Survey (ACS), the most commonly-used intercensal small-area population data in epidemiology, and WorldPop, a machine learning model for high-resolution population size estimation. Through simulation studies and an analysis of real premature mortality data, we evaluate whether WorldPop denominators can provide improved performance relative to ACS for quantifying disparities using both census tract-aggregate and race-stratified modeling approaches. We find that biases induced in parameter estimates due to temporally incompatible incidence and denominator data tend to be larger for race-stratified models than for area-aggregate models. In most scenarios considered here, WorldPop denominators lead to greater bias in estimates of health disparities than ACS denominators. These insights will assist researchers in intercensal years to select appropriate population size estimates for modeling disparities in real-time disease incidence. We highlight implications for health disparity studies in the coming decade, as 2020 census counts may introduce new sources of error.

8. Associations of long-term exposure to environmental noise and outdoor light at night with age at natural menopause in a US women cohort

   Huichu Li, Jaime E. Hart, Shruthi Mahalingaiah, and 5 more authors

   Environmental Epidemiology, May 2021

   Abs DOI

   Supplemental Digital Content is available in the text., Previous studies have suggested noise, especially at night time, and light at night (LAN) could cause neuroendocrine disturbance and circadian disruption, which may lead to ovarian follicle atresia and earlier onset of menopause. However, no study to date has directly investigated the associations of exposure to these factors and menopausal age.

9. Effect of Availability of Transcatheter Aortic-Valve Implantation on Survival for all Patients With Severe Aortic Stenosis

   Ariel Chao, Michael H. Picard, Jonathan J. Passeri, and 3 more authors

   The American Journal of Cardiology, Jun 2021

   Abs DOI

   Clinical outcomes for the overall severe aortic stenosis (AS) patient population are not well described because those medically managed are not included in procedural registries, and AS severity is not identifiable from administrative data. We aim to assess whether transcatheter aortic valve implantation (TAVI) availability has been associated with overall changes in survival for the whole AS patient population. This is important because patients with AS in real-world practice may differ from those included in randomized controlled trials, potentially attenuating the purported treatment efficacy estimated in trials. Classic severe AS patients (mean gradient ≥40 mmHg) were identified from an echocardiography database. Survival was defined as time since severe AS diagnosis until death. We first compared survival among all patients before and after TAVI availability in 2008. To further understand mechanism, we then assessed whether any survival changes were attributable to TAVI with extended Cox regression models comparing survival among TAVI, surgical aortic valve replacement, and medically managed patients. 3663 classic severe AS patients were included in the study. Median survival years for all patients were greater during the TAVI-era than Pre-TAVI-era (\textgreater11.5 vs 6.8, 5-year-HR = 0.8, time-varying effect p \textless0.0001), and increased median survival was greatest for patients age 65 to74 (\textgreater11.5 vs 9.5, 5-year-HR = 0.7, time-varying effect p = 0.045). TAVI patients age 65 to 74 had the lowest risk of death compared to medically managed patients (HR = 0.2, 95% CI = [0.1, 0.3], p \textless0.0001). In conclusion, in the TAVI-era, overall survival for patients with severe AS has doubled. This improvement is most marked for patients 65 to 74 years of age.

10. Gestational Phthalate Exposure and Preschool Attention Deficit Hyperactivity Disorder in Norway

    Elizabeth M. Kamai, Gro D. Villanger, Rachel C. Nethery, and 11 more authors

    Environmental Epidemiology, Jul 2021

    Abs DOI

    Supplemental Digital Content is available in the text., Prenatal phthalate exposure has been linked to altered neurobehavioral development in both animal models and epidemiologic studies, but whether or not these associations translate to increased risk of neurodevelopmental disorders is unclear. We used a nested case-cohort study design to assess whether maternal urinary concentrations of 12 phthalate metabolites at 17 weeks gestation were associated with criteria for Attention Deficit Hyperactivity Disorder (ADHD) classified among 3-year-old children in the Norwegian Mother, Father and Child Cohort Study (MoBa). Between 2007 and 2011, 260 children in this substudy were classified with ADHD using a standardized, on-site clinical assessment; they were compared with 549 population-based controls. We modeled phthalate levels both linearly and by quintiles in logistic regression models adjusted for relevant covariates and tested for interaction by child sex. Children of mothers in the highest quintile of di-iso-nonyl phthalate (∑DiNP) metabolite levels had 1.70 times the odds of being classified with ADHD compared with those in the lowest quintile (95% confidence interval [CI] = 1.03 to 2.82). In linear models, there was a trend with the sum of di-2-ethylhexyl phthalate metabolites (∑DEHP); each natural log-unit increase in concentration was associated with 1.22 times the odds of ADHD (95% CI = 0.99 to 1.52). In boys, but not girls, mono-n-butyl phthalate exposure was associated with increased odds of ADHD (odds ratio [OR] 1.42; 95% CI = 1.07 to 1.88). Additional adjustment for correlated phthalate metabolites attenuated estimates. These results suggest gestational phthalate exposure may impact the behavior of children as young as 3 years.

11. Pregnancy exposure to common-detect organophosphate esters and phthalates and maternal thyroid function

    Giehae Choi, Alexander P. Keil, Gro D. Villanger, and 10 more authors

    Science of The Total Environment, Aug 2021

    Abs DOI

    Background Contemporary human populations are exposed to elevated concentrations of organophosphate esters (OPEs) and phthalates. Some metabolites have been linked with altered thyroid function, however, inconsistencies exist across thyroid function biomarkers. Research on OPEs is sparse, particularly during pregnancy, when maintaining normal thyroid function is critical to maternal and fetal health. In this paper, we aimed to characterize relationships between OPEs and phthalates exposure and maternal thyroid function during pregnancy, using a cross-sectional investigation of pregnant women nested within the Norwegian Mother, Father, and Child Cohort (MoBa). Methods We included 473 pregnant women, who were euthyroid and provided bio-samples at 17 weeks’ gestation (2004–2008). Four OPE and six phthalate metabolites were measured from urine; six thyroid function biomarkers were estimated from blood. Relationships between thyroid function biomarkers and log-transformed concentrations of OPE and phthalate metabolites were characterized using two approaches that both accounted for confounding by co-exposures: co-pollutant adjusted general linear model (GLM) and Bayesian Kernal Machine Regression (BKMR). Results We restricted our analysis to common-detect OPE and phthalate metabolites (\textgreater94%): diphenyl phosphate (DPHP), di-n-butyl phosphate (DNBP), and all phthalate metabolites. In GLM, pregnant women with summed di-isononyl phthalate metabolites (∑DiNP) concentrations in the 75th percentile had a 0.37 ng/μg lower total triiodothyronine (TT3): total thyroxine (TT4) ratio (95% credible interval: [−0.59, −0.15]) as compared to those in the 25th percentile, possibly due to small but diverging influences on TT3 (−1.99 ng/dL [−4.52, 0.53]) and TT4 (0.13 μg/dL [−0.01, 0.26]). Similar trends were observed for DNBP and inverse associations were observed for DPHP, monoethyl phthalate, mono-isobutyl phthalate, and mono-n-butyl phthalate. Most associations observed in co-pollutants adjusted GLMs were attenuated towards the null in BKMR, except for the case of ∑DiNP and TT3:TT4 ratio (−0.48 [−0.96, 0.003]). Conclusions Maternal thyroid function varied modestly with ∑DiNP, whereas results for DPHP varied by the type of statistical models.

12. Differential impacts of COVID-19 lockdowns on PM2.5 across the United States

    Kevin L. Chen, Lucas R. F. Henneman, and Rachel C. Nethery

    Environmental Advances, Dec 2021

    Abs DOI

    The COVID-19 pandemic has induced large-scale behavioral changes, presenting a unique opportunity to study how air pollution is affected by societal shifts. At 455 PM2.5 monitoring sites across the United States, we conduct a causal inference analysis to determine the impacts of COVID-19 lockdowns on PM2.5. Our approach allows for rigorous confounding adjustment with highly spatio-temporally resolved effect estimates. We find that, with the exception of the Southwest, most of the US experienced increases in PM2.5 compared to concentrations expected under business-as-usual. To investigate possible drivers of this phenomenon, we use a regression model to characterize the relationship of various factors with the observed impacts. Our findings have immense environmental policy relevance, suggesting that mobility reductions alone may be insufficient to substantially and uniformly reduce PM2.5.

13. Healthy for whom? Equity in the spatial distribution of cycling risks in Los Angeles, CA

    Lindsay M. Braun, Huyen T. K. Le, Carole Turley Voulgaris, and 1 more author

    Journal of Transport & Health, Dec 2021

    Abs DOI

    Introduction The health benefits of cycling have widely been recognized, but cycling is also associated with health risks (e.g., pollution exposure, crash risk). Past studies of these competing health impacts have been limited in their treatment of social equity, rarely considering spatial variations in risk that could be highly salient for marginalized populations. This study investigates the health risks of cycling through the lens of social equity by considering variations in PM2.5 concentrations and crash risk across space and across sociodemographic groups. Methods We conducted this analysis in Los Angeles County, CA, which has a relatively high bicycle fatality rate and consistent non-attainment status for traffic-related PM2.5. We used publicly available data, including PM2.5 concentrations, crash locations, bicycle counts, and street network data, to derive measures of pollution exposure and crash risk. We performed descriptive, visualization, and regression analyses to assess how pollution exposure and crash risk vary across census block groups and are associated with area-level sociodemographic characteristics. Results We found that the health risks of cycling are disproportionately high among marginalized populations (i.e. in block groups with lower income, lower educational attainment, and higher shares of racial/ethnic minority populations). Census block groups with worse outdoor air quality and higher crash risk are particularly likely to be home to low-income people of color. Even in these places, the health benefits of cycling could outweigh the risks; however, the net health benefits of cycling, accounting for these risks, are likely to be lower in marginalized communities. Conclusions Health impact assessments related to cycling should incorporate neighborhood-level data to better assess the distribution of benefits and risks across space and across population groups. Efforts to promote cycling should focus on making cycling safer and healthier, placing emphasis on communities where it is associated with the greatest health risks.

2020

1. Air pollution and COVID-19 mortality in the United States: Strengths and limitations of an ecological regression analysis

   X. Wu, R. C. Nethery, M. B. Sabath, and 2 more authors

   Science Advances, Nov 2020

   Publisher: American Association for the Advancement of Science

   Abs DOI

   Assessing whether long-term exposure to air pollution increases the severity of COVID-19 health outcomes, including death, is an important public health objective. Limitations in COVID-19 data availability and quality remain obstacles to conducting conclusive studies on this topic. At present, publicly available COVID-19 outcome data for representative populations are available only as area-level counts. Therefore, studies of long-term exposure to air pollution and COVID-19 outcomes using these data must use an ecological regression analysis, which precludes controlling for individual-level COVID-19 risk factors. We describe these challenges in the context of one of the first preliminary investigations of this question in the United States, where we found that higher historical PM2.5 exposures are positively associated with higher county-level COVID-19 mortality rates after accounting for many area-level confounders. Motivated by this study, we lay the groundwork for future research on this important topic, describe the challenges, and outline promising directions and opportunities.

2. Associations between urine phthalate metabolites and thyroid function in pregnant women and the influence of iodine status

   Gro D. Villanger, Samantha S. M. Drover, Rachel C. Nethery, and 8 more authors

   Environment International, Apr 2020

   Abs DOI

   Background Human populations, including susceptible subpopulations such as pregnant women and their fetuses, are continuously exposed to phthalates. Phthalates may affect the thyroid hormone system, causing concern for pregnancy health, birth outcomes and child development. Few studies have investigated the joint effect of phthalates on thyroid function in pregnant women, although they are present as a mixture with highly inter-correlated compounds. Additionally, no studies have investigated if the key nutrient for thyroid health, iodine, modifies these relationships. Methods In this study, we examined the cross-sectional relationships between concentrations of 12 urinary phthalate metabolites and 6 plasma thyroid function biomarkers measured mid-pregnancy (~17 week gestation) in pregnant women (N = 1072), that were selected from a population-based prospective birth cohort, The Norwegian Mother, Father and Child Cohort study (MoBa). We investigated if the phthalate metabolite-thyroid function biomarker associations differed by iodine status by using a validated estimate of habitual dietary iodine intake based on a food frequency questionnaire from the 22nd gestation week. We accounted for the phthalate metabolite mixture by factor analyses, ultimately reducing the exposure into two uncorrelated factors. These factors were used as predictors in multivariable adjusted linear regression models with thyroid function biomarkers as the outcomes. Results Factor 1, which included high loadings for mono-iso-butyl phthalate (MiBP), mono-n-butyl phthalate (MnBP), and monobenzyl phthalate (MBzP), was associated with increased total triiodothyronine (TT3) and free T3 index (fT3i). These associations appeared to be driven primarily by women with low iodine intake (\textless150 µg/day, ~70% of our sample). Iodine intake significantly modified (p-interaction \textless 0.05) the association of factor 1 with thyroid stimulating hormone (TSH), total thyroxine (TT4) and free T4 index (fT4i), such that only among women in the high iodine intake category (≥150 µg/day, i.e. sufficient) was this factor associated with increased TSH and decreased TT4 and FT4i, respectively. In contrast, factor 2, which included high loadings for di-2-ethylhexyl phthalate metabolites (∑DEHP) and di-iso-nonyl phthalate metabolites (∑DiNP), was associated with a decrease in TT3 and fT3i, which appeared fairly uniform across iodine intake categories. Conclusion We find that phthalate exposure is associated with thyroid function in mid-pregnancy among Norwegian women, and that iodine intake, which is essential for thyroid health, could influence some of these relationships.

3. Lowering Air Pollution Levels in Massachusetts May Prevent Cardiovascular Hospital Admissions

   -Sade Maayan Yitshak, Rachel Nethery, Awad Yara Abu, and 4 more authors

   Journal of the American College of Cardiology, May 2020

   Publisher: American College of Cardiology Foundation

   DOI
4. A causal inference framework for cancer cluster investigations using publicly available data

   Rachel C. Nethery, Yue Yang, Anna J. Brown, and 1 more author

   Journal of the Royal Statistical Society. Series A, (Statistics in Society), Jun 2020

   Abs DOI

   Often, a community becomes alarmed when high rates of cancer are noticed, and residents suspect that the cancer cases could be caused by a known source of hazard. In response, the US Centers for Disease Control and Prevention recommend that departments of health perform a standardized incidence ratio (SIR) analysis to determine whether the observed cancer incidence is higher than expected. This approach has several limitations that are well documented in the existing literature. In this paper we propose a novel causal inference framework for cancer cluster investigations, rooted in the potential outcomes framework. Assuming that a source of hazard representing a potential cause of increased cancer rates in the community is identified a priori, we focus our approach on a causal inference estimand which we call the causal SIR (cSIR). The cSIR is a ratio defined as the expected cancer incidence in the exposed population divided by the expected cancer incidence for the same population under the (counterfactual) scenario of no exposure. To estimate the cSIR we need to overcome two main challenges: 1) identify unexposed populations that are as similar as possible to the exposed one to inform estimation of the expected cancer incidence under the counterfactual scenario of no exposure, and 2) publicly available data on cancer incidence for these unexposed populations are often available at a much higher level of spatial aggregation (e.g. county) than what is desired (e.g. census block group). We overcome the first challenge by relying on matching. We overcome the second challenge by building a Bayesian hierarchical model that borrows information from other sources to impute cancer incidence at the desired level of spatial aggregation. In simulations, our statistical approach was shown to provide dramatically improved results, i.e., less bias and better coverage, than the current approach to SIR analyses. We apply our proposed approach to investigate whether trichloroethylene vapor exposure has caused increased cancer incidence in Endicott, New York.

2019

1. Estimating pollution-attributable mortality at the regional and global scales: challenges in uncertainty estimation and causal inference

   Rachel C. Nethery and Francesca Dominici

   European Heart Journal, May 2019

   Abs DOI

   This editorial refers to ‘Cardiovascular disease burden from ambient air pollution in Europe reassessed using novel hazard ratio functions’†, by J. Lelieveld et

2. Estimating population average causal effects in the presence of non-overlap: The effect of natural gas compressor station exposure on cancer mortality

   Rachel C. Nethery, Fabrizia Mealli, and Francesca Dominici

   The Annals of Applied Statistics, Jun 2019

   Abs DOI

   Most causal inference studies rely on the assumption of overlap to estimate population or sample average causal effects. When data suffer from non-overlap, estimation of these estimands requires reliance on model specifications due to poor data support. All existing methods to address non-overlap, such as trimming or down-weighting data in regions of poor data support, change the estimand so that inference cannot be made on the sample or the underlying population. In environmental health research settings where study results are often intended to influence policy, population-level inference may be critical and changes in the estimand can diminish the impact of the study results, because estimates may not be representative of effects in the population of interest to policymakers. Researchers may be willing to make additional, minimal modeling assumptions in order to preserve the ability to estimate population average causal effects. We seek to make two contributions on this topic. First, we propose a flexible, data-driven definition of propensity score overlap and non-overlap regions. Second, we develop a novel Bayesian framework to estimate population average causal effects with minor model dependence and appropriately large uncertainties in the presence of non-overlap and causal effect heterogeneity. In this approach the tasks of estimating causal effects in the overlap and non-overlap regions are delegated to two distinct models suited to the degree of data support in each region. Tree ensembles are used to nonparametrically estimate individual causal effects in the overlap region, where the data can speak for themselves. In the non-overlap region where insufficient data support means reliance on model specification is necessary, individual causal effects are estimated by extrapolating trends from the overlap region via a spline model. The promising performance of our method is demonstrated in simulations. Finally, we utilize our method to perform a novel investigation of the causal effect of natural gas compressor station exposure on cancer outcomes. Code and data to implement the method and reproduce all simulations and analyses is available on Github (https://github.com/rachelnethery/overlap).

3. A joint spatial factor analysis model to accommodate data from misaligned areal units with application to Louisiana social vulnerability

   Rachel C Nethery, Dale P Sandler, Shanshan Zhao, and 2 more authors

   Biostatistics, Jul 2019

   Abs DOI

   With the threat of climate change looming, the public health community has an interest in identifying communities at the highest risk of devastation based not only on geographic features but also on social characteristics. Indices of community social vulnerability can be created by applying a spatial factor analysis to a set of relevant social variables measured for each community; however, current spatial factor analysis methodology is ill-equipped to handle spatially misaligned data. We introduce a joint spatial factor analysis model that can accommodate spatial data from two distinct partitions of a geographic space and identify a common set of latent factors underlying them. By defining the latent factors over the intersection of the two partitions, the model minimizes loss of information. Using simulated data constructed to mimic the spatial structure of our real data, we confirm the reliability of the model and demonstrate its superiority over competing ad hoc methods for dealing with misaligned data in spatial factor analysis. Finally, we construct an index of community social vulnerability for each census tract in Louisiana, a state prone to environmental disasters, which could be exacerbated by climate change, by applying the joint spatial factor analysis model to a set of misaligned social indicator data from the state. To demonstrate the utility of this index, we integrate it with Louisiana flood insurance claims data to identify communities that may be at particularly high risk during natural disasters, based on both social and geographic features.

2018

1. Prenatal Phthalates, Maternal Thyroid Function, and Risk of Attention-Deficit Hyperactivity Disorder in the Norwegian Mother and Child Cohort

   Stephanie M. Engel, Gro D. Villanger, Rachel C. Nethery, and 9 more authors

   Environmental Health Perspectives, Jul 2018

   Publisher: Environmental Health Perspectives

   Abs DOI

   Background: There is growing concern that phthalate exposures may have an impact on child neurodevelopment. Prenatal exposure to phthalates has been linked with externalizing behaviors and executive functioning defects suggestive of an attention-deficit hyperactivity disorder (ADHD) phenotype. Objectives: We undertook an investigation into whether prenatal exposure to phthalates was associated with clinically confirmed ADHD in a population-based nested case–control study of the Norwegian Mother and Child Cohort (MoBa) between the years 2003 and 2008. Methods: Phthalate metabolites were measured in maternal urine collected at midpregnancy. Cases of ADHD ( N equals 297 n=297 n=297 ) were obtained through linkage between MoBa and the Norwegian National Patient Registry. A random sample of controls ( N equals 553 n=553 n=553 ) from the MoBa population was obtained. Results: In multivariable adjusted coexposure models, the sum of di-2-ethylhexyl phthalate metabolites ( summation of DEHP ∑DEHP ∑DEHP ) was associated with a monotonically increasing risk of ADHD. Children of mothers in the highest quintile of summation of DEHP ∑DEHP ∑DEHP had almost three times the odds of an ADHD diagnosis as those in the lowest [ O R equals 2.99 OR=2.99 OR=2.99 (95% CI: 1.47, 5.49)]. When summation of DEHP ∑DEHP ∑DEHP was modeled as a log-linear (natural log) term, for each log-unit increase in exposure, the odds of ADHD increased by 47% [ O R equals 1.47 OR=1.47 OR=1.47 (95% CI: 1.09, 1.94)]. We detected no significant modification by sex or mediation by prenatal maternal thyroid function or by preterm delivery. Conclusions: In this population-based case–control study of clinical ADHD, maternal urinary concentrations of DEHP were monotonically associated with increased risk of ADHD. Additional research is needed to evaluate potential mechanisms linking phthalates to ADHD. https://doi.org/10.1289/EHP2358

2. Lipoprotein particle concentration measured by nuclear magnetic resonance spectroscopy is associated with gestational age at delivery: a prospective cohort study

   Mr Grace, Cj Vladutiu, Rc Nethery, and 5 more authors

   BJOG: An International Journal of Obstetrics & Gynaecology, Jul 2018

   _eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/1471-0528.14927

   Abs DOI

   Objective To estimate the association between lipoprotein particle concentrations in pregnancy and gestational age at delivery. Design Prospective cohort study. Setting The study was conducted in the USA at the University of North Carolina. Population We assessed 715 women enrolled in the Pregnancy, Infection, and Nutrition study from 2001 to 2005. Methods Fasting blood was collected at two time points (\textless20 and 24–29 weeks of gestation). Nuclear magnetic resonance (NMR) quantified lipoprotein particle concentrations [low-density lipoprotein (LDL), high-density lipoprotein (HDL), very-low density lipoprotein (VLDL)] and 10 subclasses of lipoproteins. Concentrations were assessed as continuous measures, with the exception of medium HDL which was classified as any or no detectable level, given its distribution. Cox proportional hazards models estimated hazard ratios (HR) for gestational age at delivery adjusting for covariates. Main outcome measures Gestational age at delivery, preterm birth (\textless37 weeks of gestation), and spontaneous preterm birth. Results At \textless20 weeks of gestation, three lipoproteins were associated with later gestational ages at delivery [large LDLNMR (HR 0.78, 95% CI 0.64–0.96), total VLDLNMR (HR 0.77, 95% CI 0.61–0.98), and small VLDLNMR (HR 0.78, 95% CI 0.62–0.98], whereas large VLDLNMR (HR 1.19, 95% CI 1.01–1.41) was associated with a greater hazard of earlier delivery. At 24–28 weeks of gestation, average VLDLNMR (HR 1.25, 95% CI 1.03–1.51) and a detectable level of medium HDLNMR (HR 1.90, 95% CI 1.19–3.02) were associated with earlier gestational ages at delivery. Conclusion In this sample of pregnant women, particle concentrations of VLDLNMR, LDLNMR, IDLNMR, and HDLNMR were each independently associated with gestational age at delivery for all deliveries or spontaneous deliveries \textless37 weeks of gestation. These findings may help formulate hypotheses for future studies of the complex relationship between maternal lipoproteins and preterm birth. Tweetable abstract Nuclear magnetic resonance spectroscopy may identify lipoprotein particles associated with preterm delivery.

2017

1. Flavored little cigar smoke induces cytotoxicity and apoptosis in airway epithelia

   Arunava Ghosh, Rachel C. Nethery, Amy H. Herring, and 1 more author

   Cell Death Discovery, Apr 2017

   Number: 1 Publisher: Nature Publishing Group

   Abs DOI

   Addition of flavors reduces the harsh taste of tobacco, facilitating the initiation and maintenance of addiction among youths. Flavored cigarettes (except menthol) are now banned. However, the legislation on little cigars remains unclear and flavored little cigars are currently available for purchase. Since inhaled tobacco smoke directly exerts toxic effects on the lungs, we tested whether non-flavored and flavored little cigar smoke exposure had the potential for harm in cultured pulmonary epithelia. We cultured Calu-3 lung epithelia on both 96-well plates and at the air–liquid interface and exposed them to smoke from non-flavored Swisher Sweets and flavored (sweet cherry, grape, menthol, peach and strawberry) Swisher Sweets little cigars. Irrespective of flavor, acute little cigar smoke exposure (10×35 ml puffs) significantly increased cell death and decreased the percentage of live cells. Chronic exposure (10×35 ml puffs per day for 4 days) of smoke to Calu-3 cultures significantly increased lactate dehydrogenase release, further indicating toxicity. To determine whether this exposure was associated with increased cell death/apoptosis, a protein array was used. Chronic exposure to smoke from all types of little cigars induced the activation of the two major apoptosis pathways, namely the intrinsic (mitochondrial-mediated) and the extrinsic (death receptor-mediated) pathways. Both flavored and non-flavored little cigar smoke caused similar levels of toxicity and activation of apoptosis, suggesting that flavored and non-flavored little cigars are equally harmful. Hence, the manufacture, advertisement, sale and use of both non-flavored and flavored little cigars should be strictly controlled.

2. Flavored e-cigarette liquids reduce proliferation and viability in the CALU3 airway epithelial cell line

   Temperance R. Rowell, Steven L. Reeber, Shernita L. Lee, and 5 more authors

   American Journal of Physiology-Lung Cellular and Molecular Physiology, Jul 2017

   Publisher: American Physiological Society

   Abs DOI

   E-cigarettes are generally thought of as a safer smoking alternative to traditional cigarettes. However, little is known about the effects of e-cigarette liquids (e-liquids) on the lung. Since over 7,000 unique flavors have been identified for purchase in the United States, our goal was to conduct a screen that would test whether different flavored e-liquids exhibited different toxicant profiles. We tested the effects of 13 different flavored e-liquids [with nicotine and propylene glycol/vegetable glycerin (PG/VG) serving as controls] on a lung epithelial cell line (CALU3). Using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay as an indicator of cell proliferation/viability, we demonstrated a dose-dependent decrease of MTT metabolism by all flavors tested. However, a group of four flavors consistently showed significantly greater toxicity compared with the PG/VG control, indicating the potential for some flavors to elicit more harmful effects than others. We also tested the aerosolized “vapor” from select e-liquids on cells and found similar dose-dependent trends, suggesting that direct e-liquid exposures are a justifiable first-pass screening approach for determining relative e-liquid toxicity. We then identified individual chemical constituents for all 13 flavors using gas chromatography-mass spectrometry. These data revealed that beyond nicotine and PG/VG, the 13 flavored e-liquids have diverse chemical constituents. Since all of the flavors exhibited some degree of toxicity and a diverse array of chemical constituents with little inhalation toxicity available, we conclude that flavored e-liquids should be extensively tested on a case-by-case basis to determine the potential for toxicity in the lung and elsewhere.

2015

1. A common spatial factor analysis model for measured neighborhood-level characteristics: The Multi-Ethnic Study of Atherosclerosis

   Rachel C. Nethery, Joshua L. Warren, Amy H. Herring, and 3 more authors

   Health & Place, Nov 2015

   Abs DOI

   The purpose of this study was to reduce the dimensionality of a set of neighborhood-level variables collected on participants in the Multi-Ethnic Study of Atherosclerosis (MESA) while appropriately accounting for the spatial structure of the data. A common spatial factor analysis model in the Bayesian setting was utilized in order to properly characterize dependencies in the data. Results suggest that use of the spatial factor model can result in more precise estimation of factor scores, improved insight into the spatial patterns in the data, and the ability to more accurately assess associations between the neighborhood environment and health outcomes.