On the use of ZIP codes and ZIP code tabulation areas (ZCTAs) for the spatial analysis of epidemiological data

Assessing the Impact of Neighborhood and Built Environment on Pediatric Perioperative Care: A Systematic Review of the Literature

CONTEXT

Neighborhood and built environment encompass one key area of the Social Determinants of Health (SDOH) and is frequently assessed using area-level indices.

OBJECTIVE

We sought to systematically review the pediatric surgery literature for use of commonly applied area-level indices and to compare their utility for prediction of outcomes.

DATA SOURCES

A literature search was conducted using PubMed, Ovid MEDLINE, Ovid MEDLINE Epub Ahead of Print, PsycInfo, and an artificial intelligence search tool (1/2013-2/2023).

STUDY SELECTION

Inclusion required pediatric surgical patients in the US, surgical intervention performed, and use of an area-level metric.

DATA EXTRACTION

Extraction domains included study, patient, and procedure characteristics.

RESULTS

Area Deprivation Index is the most consistent and commonly accepted index. It is also the most granular, as it uses Census Block Groups. Child Opportunity Index is less granular (Census Tract), but incorporates pediatric-specific predictors of risk. Results with Social Vulnerability Index, Neighborhood Deprivation Index, and Neighborhood Socioeconomic Status were less consistent.

LIMITATIONS

All studies were retrospective and quality varied from good to fair.

CONCLUSIONS

While each index has strengths and limitations, standardization on ideal metric(s) for the pediatric surgical population will help build the inferential power needed to move from understanding the role of SDOH to building meaningful interventions towards equity in care.

TYPE OF STUDY

Systematic Review.

LEVEL OF EVIDENCE

Level III.

Neighborhood built environment, obesity, and diabetes: A Utah siblings study

Background

This study utilizes innovative computer vision methods alongside Google Street View images to characterize neighborhood built environments across Utah.

Methods

Convolutional Neural Networks were used to create indicators of street greenness, crosswalks, and building type on 1.4 million Google Street View images. The demographic and medical profiles of Utah residents came from the Utah Population Database (UPDB). We implemented hierarchical linear models with individuals nested within zip codes to estimate associations between neighborhood built environment features and individual-level obesity and diabetes, controlling for individual- and zip code-level characteristics (n = 1,899,175 adults living in Utah in 2015). Sibling random effects models were implemented to account for shared family attributes among siblings (n = 972,150) and twins (n = 14,122).

Results

Consistent with prior neighborhood research, the variance partition coefficients (VPC) of our unadjusted models nesting individuals within zip codes were relatively small (0.5%-5.3%), except for HbA1c (VPC = 23%), suggesting a small percentage of the outcome variance is at the zip code-level. However, proportional change in variance (PCV) attributable to zip codes after the inclusion of neighborhood built environment variables and covariates ranged between 11% and 67%, suggesting that these characteristics account for a substantial portion of the zip code-level effects. Non-single-family homes (indicator of mixed land use), sidewalks (indicator of walkability), and green streets (indicator of neighborhood aesthetics) were associated with reduced diabetes and obesity. Zip codes in the third tertile for non-single-family homes were associated with a 15% reduction (PR: 0.85; 95% CI: 0.79, 0.91) in obesity and a 20% reduction (PR: 0.80; 95% CI: 0.70, 0.91) in diabetes. This tertile was also associated with a BMI reduction of -0.68 kg/m2 (95% CI: -0.95, -0.40).

Conclusion

We observe associations between neighborhood characteristics and chronic diseases, accounting for biological, social, and cultural factors shared among siblings in this large population-based study.

A Narrative Review Investigating Practices and Disparities in Child Abuse Amongst United States Pediatric Trauma Patients & Associated Outcomes

INTRODUCTION

Although non-accidental trauma continues to be a leading cause of morbidity and mortality among children in the United States, the underlying factors leading to NAT are not well characterized. We aim to review reporting practices, clinical outcomes, and associated disparities among pediatric trauma patients experiencing NAT.

METHODS

A literature search utilizing PubMed, Google Scholar, EMBASE, ProQuest, and Cochrane was conducted from database inception until April 6, 2023. This review includes studies that assessed pediatric (age <18) trauma patients treated for NAT in the United States emergency departments. The evaluated outcome was in-hospital mortality rates stratified by race, age, sex, insurance status, and socioeconomic advantage.

RESULTS

The literature search yielded 2641 initial articles, and after screening and applying inclusion and exclusion criteria, 15 articles remained. African American pediatric trauma patients diagnosed with NAT had higher mortality odds than white patients, even when adjusting for comparable injury severity. Children older than 12 mo experienced higher mortality rates compared to those younger than 12 mo, although some studies did not find a significant association between age and mortality. Uninsured insurance status was associated with the highest mortality rate, followed by Medicaid and private insurance. No significant association between sex and mortality or socioeconomic advantage and mortality was observed.

CONCLUSIONS

Findings showed higher in-hospital mortality among African American pediatric trauma patients experiencing child abuse, and in patients 12 mo or older. Medicaid and uninsured pediatric patients faced higher mortality odds from their abuse compared to privately insured patients.

Neighbourhood-level policing as a racialised gendered stressor: multilevel analysis of police stops and preterm birth in Seattle, Washington

BACKGROUND

Most studies capturing the health effects of police violence focus on directly impacted individuals, but a burgeoning field of study is capturing the indirect, community-level health effects of policing. Few empirical studies have examined neighbourhood-level policing, a contextual and racialised gendered stressor, in relation to preterm birth risk among Black and other racially minoritised people.

METHODS

We spatially linked individual birth records (2017-2019) in Seattle, Washington (n=25 909) with geocoded data on police stops for three exposure windows: year before pregnancy, first and second trimester. We fit race-stratified multilevel modified Poisson regression models predicting preterm birth (<37 gestational weeks) across tertiles of neighbourhood stop rates controlling for individual and neighbourhood-level covariates. For the second trimester exposure window, birth was operationalised as a time-to-event outcome using multilevel Cox proportional hazard models.

RESULTS

Neighbourhood stop rates of Black residents was higher compared with White residents, and Black and Asian pregnant people were exposed to the highest median neighbourhood-level stop rates. Black birthing people living in neighbourhoods with more frequent police stops had increased risk of preterm birth across all exposure windows including the year before pregnancy (adjusted risk ratio (aRR): 1.38, 95% CI 1.02 to 1.85), first trimester (aRR:1.74, 95% CI 1.17 to 2.57) and second trimester (aHR: 1.66, 95% CI 1.14 to 2.42). We found null or inverse associations among Asian, Hispanic and White people.

CONCLUSION

Our study adds to the growing evidence documenting associations of higher risk of preterm birth with neighbourhood police stops among Black birthing people. These findings suggest that routine police practices are one aspect of structural racism contributing to racialised perinatal health inequities.

Food insecurity screening and local food access: Contributions to nutritional outcomes among children and adults with cystic fibrosis in the United States

BACKGROUND

As the nutritional status of people with CF (PwCF) is associated with their socioeconomic status, it is important to understand factors related to food security and food access that play a role in the nutritional outcomes of this population. We assessed the contributions of CF program-level food insecurity screening practices and area-level food access for nutritional outcomes among PwCF.

METHODS

We conducted a cross-sectional analysis of 2019 data from the U.S. CF Patient Registry (CFFPR), linked to survey data on CF program-level food insecurity screening and 2019 patient zip code-level food access. Pediatric and adult populations were analyzed separately. Nutritional outcomes were assessed with annualized BMI percentiles (CDC charts) for children and BMI (kg/m2) for adults, with underweight status defined as BMIp <10% for children and BMI <18.5 kg/m2 for adults, and overweight or obese status defined as BMIp >85% for children and BMI >25 kg/m2 for adults. Analyses were adjusted for patient sociodemographic and clinical characteristics.

RESULTS

The study population included 11,971 pediatric and 14,817 adult PwCF. A total of 137 CF programs responded to the survey, representing 71% of the pediatric sample and 45% of the CFFPR adult sample. The joint models of nutritional status as a function of both program-level food insecurity screening and area-level food access produced the following findings. Among children with CF, screening at every visit vs less frequently was associated with 39% lower odds of being underweight (OR 0.61, p = 0.019), and the effect remained the same and statistically significant after adjusting for all covariates (aOR 0.61, p = 0.047). Residence in a food desert was associated both with higher odds of being underweight (OR 1.66, p = 0.036; aOR 1.58, p = 0.008) and with lower BMIp (-4.81%, p = 0.004; adjusted -3.73%, p = 0.014). Among adults with CF, screening in writing vs verbally was associated with higher odds of being overweight (OR 1.22, p = 0.028; aOR 1.36, p = 0.002) and higher BMI (adjusted 0.43 kg/m2, p = 0.032). Residence in a food desert was associated with higher odds of being underweight (OR 1.48, p = 0.025).

CONCLUSIONS

Food insecurity screening and local food access are independent predictors of nutritional status among PwCF. More frequent screening is associated with less underweight among children with CF, whereas screening in writing (vs verbally) is associated with higher BMI among adults. Limited food access is associated with higher odds of being underweight in both children and adults with CF, and additionally with lower BMI among children with CF. Study results highlight the need for standardized, evidence-based food insecurity screening across CF care programs and for equitable food access to optimize the nutritional outcomes of PwCF.

Insights Into Eye Care Accessibility: Geospatial Distribution of Eye Care Providers and Socioeconomic Factors by ZIP Code

Purpose

In the United States, the ZIP Code has long been used to collect geospatial data revealing disparities in social determinants of health. This cross-sectional study examines the distribution of eye care access in association with local socioeconomic factors at a ZIP Code level.

Methods

Data from the 2020 Centers of Medicare and Medicaid Services and American Community Survey were used to examine locations of 47,949 providers (17,631 ophthalmologists and 30,318 optometrists) and corresponding local socioeconomic variables (education, employment, and income). Multivariable zero-inflated negative binomial regression was used to model eye care provider count per capita in each ZIP Code area with socioeconomic factors as independent covariates.

Results

For every 1% increase in percentage of population over 25 years with a bachelor's degree or higher, the expected number of providers increases by 4.4% (incidence rate ratio [IRR] = 1.044; 95% confidence interval [CI], 1.041-1.046; P < 0.001). For every 1% increase in percentage unemployment, the expected number of providers decreases by 2.7% (IRR = 0.973; 95% CI, 0.964-0.983; P < 0.001). However, for every $1000 increase in median household income, the expected number of providers decreases by 1.6% (IRR = 0.984; 95% CI, 0.983-0.986; P < 0.001).

Conclusions

Disparities in access exist in areas of lower employment and educational attainment, as both have positive correlations with eye care provider access. Conversely, areas of greater median household income have lower access to providers.

Translational Relevance

This research contributes to a greater field studying social determinants of health and may inform public health strategies on allocation of providers to improve equitable access to vision care.

Short-Term Increases in NO2 and O3 Concentrations during Pregnancy and Stillbirth Risk in the U.S.: A Time-Stratified Case-Crossover Study

Associations between gaseous pollutant exposure and stillbirth have focused on exposures averaged over trimesters or gestation. We investigated the association between short-term increases in nitrogen dioxide (NO2) and ozone (O3) concentrations and stillbirth risk among a national sample of 116 788 Medicaid enrollees from 2000 to 2014. A time-stratified case-crossover design was used to estimate distributed (lag 0-lag 6) and cumulative lag effects, which were adjusted for PM2.5 concentration and temperature. Effect modification by race/ethnicity and proximity to hydraulic fracturing (fracking) wells was assessed. Short-term increases in the NO2 and O3 concentrations were not associated with stillbirth in the overall sample. Among American Indian individuals (n = 1694), a 10 ppb increase in NO2 concentrations was associated with increased stillbirth odds at lag 0 (5.66%, 95%CI: [0.57%, 11.01%], p = 0.03) and lag 1 (4.08%, 95%CI: [0.22%, 8.09%], p = 0.04) but not lag 0-6 (7.12%, 95%CI: [-9.83%, 27.27%], p = 0.43). Among participants living in zip codes within 15 km of active fracking wells (n = 9486), a 10 ppb increase in NO2 concentration was associated with increased stillbirth odds in single-day lags (2.42%, 95%CI: [0.37%, 4.52%], p = 0.02 for lag 0 and 1.83%, 95%CI: [0.25%, 3.43%], p = 0.03 for lag 1) but not the cumulative lag (lag 0-6) (4.62%, 95%CI: [-2.75%, 12.55%], p = 0.22). Odds ratios were close to the null in zip codes distant from fracking wells. Future studies should investigate the role of air pollutants emitted from fracking and potential racial disparities in the relationship between short-term increases in NO2 concentrations and stillbirth.

Increased disparities in waitlist and post-heart transplantation outcomes according to socioeconomic status with the new heart transplant allocation system

BACKGROUND

The study objective was to assess disparities in outcomes in the waitlist and post-heart transplantation (HT) according to socioeconomic status (SES) in the old and new U.S. HT allocation systems.

METHODS

Adult HT candidates in the United Network for Organ Sharing database from 2014 through 2021 were included. Old or new system classification was according to listing before or after October 18, 2018. SES was stratified by patient ZIP code and median household income via U.S. Census Bureau and classified into terciles. Competing waitlist outcomes and post-transplantation survival were compared between systems.

RESULTS

In total, 26,450 patients were included. Waitlisted candidates with low SES were more frequently younger, female, African American, and with higher body mass index. Reduced cumulative incidence (CI) of HT in the old system occurred in low SES (53.5%) compared to middle (55.7%, p = 0.046), and high (57.9%, p < 0.001). In the new system, the CI of HT was 65.3% in the low SES vs middle (67.6%, p = 0.002) and high (70.2%, p < 0.001), and SES remained significant in the adjusted analysis. In the old system, CI of death/delisting was similar across SES. In the new system, low SES had increased CI of death/delisting (7.4%) vs middle (6%, p = 0.012) and high (5.4%, p = 0.002). The old system showed similar 1-year survival across SES. In the new system, recipients with low SES had decreased 1-year survival (p = 0.041).

CONCLUSIONS

SES affects waitlist and post-transplant outcomes. In the new system, all SES had increased access to HT; however, low SES had increased death/delisting due to worsening clinical status and decreased post-transplant survival.

Neighborhood disadvantage, household chaos, and personal stressors: exploring early-life contextual factors and current mental health symptoms in college students

OBJECTIVE

Using Bronfenbrenner's socio-ecological model as a frame, we explored the impact of neighborhood disadvantage, household chaos, and personal stressors on current mental health symptoms in college students.

PARTICIPANTS

144 students at a large, public university in the southern U.S.

METHODS

Participants completed measures of demographics, family-of-origin household chaos, stressors, anxiety, and depression, and provided their childhood home ZIP code. Using U.S. Census Data, four structural indicators of neighborhood disadvantage were extracted and appended to each participant's ZIP code.

RESULTS

Hierarchical regression revealed that all three variables predicted anxiety symptoms. However, only household chaos and personal stressors predicted current depressive symptoms. Unexpectedly, greater neighborhood disadvantage predicted lower levels of current anxiety. Mediation analyses demonstrated that personal stressors partially mediated the relationships between household chaos and mental health symptoms.

CONCLUSIONS

College administration and counseling centers may wish to consider pre-college factors that influence college students' current anxious and depressive symptoms.

The Geography of Injuries in Trauma Systems: Using Home as a Proxy for Incident Location

INTRODUCTION

Effective trauma system organization is crucial to timely access to care and requires accurate understanding of injury and resource locations. Many systems rely on home zip codes to evaluate geographic distribution of injury; however, few studies have evaluated the reliability of home as a proxy for incident location after injury.

METHODS

We analyzed data from a multicenter prospective cohort collected from 2017 to 2021. Injured patients with both home and incident zip codes were included. Outcomes included discordance and differential distance between home and incident zip code. Associations of discordance with patient characteristics were determined by logistic regression. We also assessed trauma center catchment areas based on home versus incident zip codes and variation regionally at each center.

RESULTS

Fifty thousand one hundred seventy-five patients were included in the analysis. Home and incident zip codes were discordant in 21,635 patients (43.1%). Injuries related to motor vehicles (aOR: 4.76 [95% CI 4.50-5.04]) and younger adults 16-64 (aOR: 2.46 [95% CI 2.28-2.65]) were most likely to be discordant. Additionally, as injury severity score increased, discordance increased. Trauma center catchment area differed up to two-thirds of zip codes when using home versus incident location. Discordance rate, discordant distance, and catchment area overlap between home and incident zip codes all varied significantly by geographic region.

CONCLUSIONS

Home location as proxy for injury location should be used with caution and may impact trauma system planning and policy, especially in certain populations. More accurate geolocation data are warranted to further optimize trauma system design.

Improving data capture of race and ethnicity for the Food and Drug Administration Sentinel database: a narrative review

PURPOSE

The U.S. Food and Drug Administration's Sentinel System is a national medical product safety surveillance system consisting of a large multisite distributed database of administrative claims supplemented by electronic health-care record data. The program seeks to improve data capture of race and ethnicity for pharmacoepidemiology studies.

METHODS

We conducted a narrative literature review of published research on data augmentation and imputation methods to improve race and ethnicity capture in U.S. health-care systems databases. We focused on methods with limited (five-digit ZIP codes only) or full patient identifiers available to link to external sources of self-reported data. We organized the literature by themes: (1) variation in data capture of self-reported data, (2) data augmentation from external sources of self-reported data, and (3) imputation methods, including Bayesian analysis and multiple regression.

RESULTS

Researchers reduced data missingness with high validity for Asian, Black, White, and Pacific Islander racial groups and Hispanic ethnicity. Native American and multiracial groups were difficult to validate due to relatively small sample sizes.

CONCLUSIONS

Limitations on accessible self-reported data for validation will dictate methods to improve race and ethnicity data capture. We recommend methods leveraging multiple sources that account for variations in geography, age, and sex.

Assessing performance of ZCTA-level and Census Tract-level social and environmental risk factors in a model predicting hospital events

Predictive analytics are used in primary care to efficiently direct health care resources to high-risk patients to prevent unnecessary health care utilization and improve health. Social determinants of health (SDOH) are important features in these models, but they are poorly measured in administrative claims data. Area-level SDOH can be proxies for unavailable individual-level indicators, but the extent to which the granularity of risk factors impacts predictive models is unclear. We examined whether increasing the granularity of area-based SDOH features from ZIP code tabulation area (ZCTA) to Census Tract strengthened an existing clinical prediction model for avoidable hospitalizations (AH events) in Maryland Medicare fee-for-service beneficiaries. We created a person-month dataset for 465,749 beneficiaries (59.4% female; 69.8% White; 22.7% Black) with 144 features indexing medical history and demographics using Medicare claims (September 2018 through July 2021). Claims data were linked with 37 SDOH features associated with AH events from 11 publicly-available sources (e.g., American Community Survey) based on the beneficiaries' ZCTA and Census Tract of residence. Individual AH risk was estimated using six discrete time survival models with different combinations of demographic, condition/utilization, and SDOH features. Each model used stepwise variable selection to retain only meaningful predictors. We compared model fit, predictive performance, and interpretation across models. Results showed that increasing the granularity of area-based risk factors did not dramatically improve model fit or predictive performance. However, it did affect model interpretation by altering which SDOH features were retained during variable selection. Further, the inclusion of SDOH at either granularity level meaningfully reduced the risk that was attributed to demographic predictors (e.g., race, dual-eligibility for Medicaid). Differences in interpretation are critical given that this model is used by primary care staff to inform the allocation of care management resources, including those available to address drivers of health beyond the bounds of traditional health care.

No Escape: Mass Incarceration and the Social Ecology of Intimate Partner Violence Against Women

Women in heavily policed and incarcerated communities face extremely high rates of intimate partner violence (IPV)-but how criminal legal system contact affects such violence remains poorly understood. This study explores the social ecology of IPV by fitting structural equation models to longitudinal, dyadic data from households in contact with the criminal legal system (N = 2,224) and their local communities. Results suggest that a complex of factors at multiple social-ecological levels-including adverse local conditions, dysfunctional couple conflict, and men's behavioral health and perceptions of their neighborhoods-may put women at heightened risk of IPV victimization in a time of mass incarceration.

Investigation of Geographical Disparities: The Use of An Interpolation Method For Cancer Registry Data

The American Cancer Society estimated 1.9 million diagnosed cancer cases and 608,570 cancer deaths in 2021 in the US; for Oklahoma, they estimated 22,820 cases and 8,610 deaths. This project aimed to demonstrate a method to systematically describe cancer in an accurate and visually attractive, yet simple to make, interpolated map using ZIP Code level registry data, as it is the smallest area unit with high accuracy using inverse distance weighting. We describe a process of creating smoothed maps with an appropriate, well-described, simple, replicable method. These smoothed maps display low (cold) or high (hot) areas of incidence rates of: (a) all cancer combined, (b) colorectal cancer and lung cancer rates by gender, (c) female breast cancer, and (d) prostate cancer, by ZIP Codes for Oklahoma from 2013-2017. The methods we present in this paper provide an effective visualization to pinpoint low (cold) or high (hot) areas of cancer incidence.

Telemedicine Use Among People With HIV in 2021: The Hybrid-Care Environment

BACKGROUND

Telemedicine use for the care of people with HIV (PWH) significantly expanded during the COVID-19 pandemic. During 2021, vaccine uptake increased and patients were encouraged to resume in-person care, resulting in a mixture of in-person and telemedicine visits. We studied how different patient populations used telemedicine in this hybrid-care environment.

METHODS

Using observational data from patients enrolled in the Johns Hopkins HIV Clinical Cohort, we analyzed all in-person and telemedicine HIV primary care visits completed in an HIV clinic from January 1st, 2021, to December 31st, 2021. We used log-binomial regression to investigate the association between patient characteristics and the probability of completing a telemedicine versus in-person visit and the probability of completing a video versus telephone visit.

RESULTS

A total of 5518 visits were completed by 1884 patients; 4282 (77.6%) visits were in-person, 800 (14.5%) by phone, and 436 (7.9%) by video. The relative risk (RR) of completing telemedicine vs. in-person visits was 0.65 (95% Confidence Interval (CI): 0.47, 0.91) for patients age 65 years or older vs. age 20-39 years; 0.84 (95% CI: 0.72, 0.98) for male patients vs. female patients; 0.81 (95% CI: 0.66, 0.99) for Black vs. White patients; 0.62 (95% CI: 0.49, 0.79) for patients in the highest vs. lowest quartile of Area Deprivation Index; and 1.52 (95% CI: 1.26, 1.84) for patients >15 miles vs. <5 miles from clinic.

CONCLUSIONS

In the second year of the pandemic, overall in-person care was used more than telemedicine and significant differences persist across subgroups in telemedicine uptake.

Treatment Engagement Among Children Exposed to Violence: A Systems Perspective

Childhood exposure to violence is a major public health issue. Effective treatment can reduce the impact of violence exposure on child outcomes. However, disparities in treatment engagement can interfere with effective treatment. This study reviews data collected from 2,546 children referred to community-based mental health services from 2001 to 2015 after exposure to violence. Children were categorized into three groups: those who attended intake but never started treatment, referred to as the Nonengager group; those who started but discontinued treatment prior to meeting treatment goals, referred to as the Attriter group; and those who completed treatment as rated by the treating therapist, referred to as the Completer group. The three groups were analyzed for differences in behavioral and emotional problems, racial identity, child social support, household income, number of people living in the home, parent stress, parent social support, community violence exposure, and neighborhood-level child opportunity. Analyses demonstrated that the Completer group were more likely to: live in neighborhoods with higher levels of childhood opportunity, identify as White, have an annual household income of $40,000 or greater, have significantly fewer people living in the home, report lower levels of parental stress, report higher levels of parental social support, report higher levels of child social support, and have significantly lower scores of emotional and behavioral problems after treatment. Overall, our study supports the importance of considering multiple ecological levels when targeting treatment engagement for children after exposure to violence. Results indicate that children from more advantaged environments are more likely to complete treatment, which leads to better outcomes. This can exacerbate existing disparities. Findings highlight the need for systems change and advocacy for children in less advantaged environments and meeting families in their specific context, to combat treatment disparities.

'Unmasking' masked address data: A medoid geocoding solution

In recent years, there has been a consistent push for more open data initiatives, particularly for datasets collected by public agencies or groups that receive public funding. However, there is a tension between the release of open data and the preservation of individual and household privacy, whose balance shifts due to increased data availability, the sophistication of analysis techniques, and the computational power available to users. As a result, data masking is a standard tool used to preserve privacy. This is a process in which the data publishers obfuscate some identifying features in the dataset while attempting to maintain as much accuracy and precision as possible. For spatial datasets, the geocoding of administratively-masked data has been a consistent problem. Here, we present a medoid-based technique that geocodes masked data while minimizing the spatial uncertainty associated with the masking approach. Unfortunately, many commercial geocoding software packages either fail to geocode administratively-masked data or provide false positives by assigning points to city or street centroids. We demonstrate the results of our medoid-based geocoding approach by comparing it to commercial geocoding software. The results suggest that a medoid geocoding approach is mechanically simple to deploy and maximizes the spatial accuracy of the resulting geocodes.•Administratively-masked data are difficult to geocode•A medoid geocoding method maximizes geocoding accuracy•This method outperforms commercial geocoding software.

Characterizing the neighborhood risk environment in multisite clinic-based cohort studies: A practical geocoding and data linkages protocol for protected health information

BACKGROUND

Maintaining patient privacy when geocoding and linking residential address information with neighborhood-level data can create challenges during research. Challenges may arise when study staff have limited training in geocoding and linking data, or when non-study staff with appropriate expertise have limited availability, are unfamiliar with a study's population or objectives, or are not affordable for the study team. Opportunities for data breaches may also arise when working with non-study staff who are not on-site. We detail a free, user-friendly protocol for constructing indices of the neighborhood risk environment during multisite, clinic-based cohort studies that rely on participants' protected health information. This protocol can be implemented by study staff who do not have prior training in Geographic Information Systems (GIS) and can help minimize the operational costs of integrating geographic data into public health projects.

METHODS

This protocol demonstrates how to: (1) securely geocode patients' residential addresses in a clinic setting and match geocoded addresses to census tracts using Geographic Information System software (Esri, Redlands, CA); (2) ascertain contextual variables of the risk environment from the American Community Survey and ArcGIS Business Analyst (Esri, Redlands, CA); (3) use geoidentifiers to link neighborhood risk data to census tracts containing geocoded addresses; and (4) assign randomly generated identifiers to census tracts and strip census tracts of their geoidentifiers to maintain patient confidentiality.

RESULTS

Completion of this protocol generates three neighborhood risk indices (i.e., Neighborhood Disadvantage Index, Murder Rate Index, and Assault Rate Index) for patients' coded census tract locations.

CONCLUSIONS

This protocol can be used by research personnel without prior GIS experience to easily create objective indices of the neighborhood risk environment while upholding patient confidentiality. Future studies can adapt this protocol to fit their specific patient populations and analytic objectives.

Exposure to industrial hog and poultry operations and urinary tract infections in North Carolina, USA

An increasing share of urinary tract infections (UTIs) are caused by extraintestinal pathogenic Escherichia coli (ExPEC) lineages that have also been identified in poultry and hogs with high genetic similarity to human clinical isolates. We investigated industrial food animal production as a source of uropathogen transmission by examining relationships of hog and poultry density with emergency department (ED) visits for UTIs in North Carolina (NC). ED visits for UTI in 2016-2019 were identified by ICD-10 code from NC's ZIP code-level syndromic surveillance system and livestock counts were obtained from permit data and aerial imagery. We calculated separate hog and poultry spatial densities (animals/km²) by Census block with a 5 km buffer on the block perimeter and weighted by block population to estimate mean ZIP code densities. Associations between livestock density and UTI incidence were estimated using a reparameterized Besag-York-Mollié (BYM2) model with ZIP code population offsets to account for spatial autocorrelation. We excluded metropolitan and offshore ZIP codes and assessed effect measure modification by calendar year, ZIP code rurality, and patient sex, age, race/ethnicity, and health insurance status. In single-animal models, hog exposure was associated with increased UTI incidence (rate ratio [RR]: 1.21, 95 % CI: 1.07-1.37 in the highest hog-density tertile), but poultry exposure was associated with reduced UTI rates (RR: 0.86, 95 % CI: 0.81-0.91). However, the reference group for single-animal poultry models included ZIP codes with only hogs, which had some of the highest UTI rates; when compared with ZIP codes without any hogs or poultry, there was no association between poultry exposure and UTI incidence. Hog exposure was associated with increased UTI incidence in areas that also had medium to high poultry density, but not in areas with low poultry density, suggesting that intense hog production may contribute to increased UTI incidence in neighboring communities.

Assessment of Population-Level Disadvantage Indices to Inform Equitable Health Policy

Policy Points The rapid uptake of disadvantage indices during the pandemic highlights investment in implementing tools that address health equity to inform policy. Existing indices differ in their design, including data elements, social determinants of health domains, and geographic unit of analysis. These differences can lead to stark discrepancies in place-based social risk scores depending on the index utilized. Disadvantage indices are useful tools for identifying geographic patterns of social risk; however, indiscriminate use of indices can have varied policy implications and unintentionally worsen equity. Implementers should consider which indices are suitable for specific communities, objectives, potential interventions, and outcomes of interest.

CONTEXT

There has been unprecedented uptake of disadvantage indices such as the Centers for Disease Control and Prevention Social Vulnerability Index (SVI) to identify place-based patterns of social risk and guide equitable health policy during the COVID-19 pandemic. However, limited evidence around data elements, interoperability, and implementation leaves unanswered questions regarding the utility of indices to prioritize health equity.

METHODS

We identified disadvantage indices that were (a) used three or more times from 2018 to 2021, (b) designed using national-level data, and (c) available at the census-tract or block-group level. We used a network visualization to compare social determinants of health (SDOH) domains across indices. We then used geospatial analyses to compare disadvantage profiles across indices and geographic areas.

FINDINGS

We identified 14 indices. All incorporated data from public sources, with half using only American Community Survey data (n = 7) and the other half combining multiple sources (n = 7). Indices differed in geographic granularity, with county level (n = 5) and census-tract level (n = 5) being the most common. Most states used the SVI during the pandemic. The SVI, the Area Deprivation Index (ADI), the COVID-19 Community Vulnerability Index (CCVI), and the Child Opportunity Index (COI) met criteria for further analysis. Selected indices shared five indicators (income, poverty, English proficiency, no high school diploma, unemployment) but varied in other metrics and construction method. While mapping of social risk scores in Durham County, North Carolina; Cook County, Illinois; and Orleans Parish, Louisiana, showed differing patterns within the same locations depending on choice of disadvantage index, risk scores across indices showed moderate to high correlation (rs 0.7-1). However, spatial autocorrelation analyses revealed clustering, with discrepant distributions of social risk scores between different indices.

CONCLUSIONS

Existing disadvantage indices use varied metrics to represent place-based social risk. Within the same geographic area, different indices can provide differences in social risk values and interpretations, potentially leading to varied public health or policy responses.

Historical redlining and the epidemiology of present-day firearm violence in the United States: A multi-city analysis

Firearm violence is a major cause of morbidity, mortality, and racial health disparities in the United States. Previous studies have identified associations between historically racist housing discrimination (i.e., redlining practices) and firearm violence; however, these studies generally have been limited to a single city and have yet to provide sufficient evidence through which to determine the extent and dynamics of the impact of this relationship across the country. The aim of our study was (1) to estimate the association of historical redlining on both violent and firearm death across the country in nested models; and (2) to examine spatial non-stationarity to determine whether the impact of historical redlining on violent and firearm death was the same across the U.S. We used multilevel Bayesian conditional autoregressive Poisson models to determine the relationship between redlining as illustrated through Home Owners' Loan Corporation maps and 2019 violent and firearm deaths at the ZIP code-level nested within 21 cities across the U.S. We found that at the ZIP code level, there was a dose-responsive relationship between HOLC grading and the incidence of present-day firearm deaths. In general, redlined ZIP codes had higher relative incidence of firearm deaths. Associations were not stable across cities. For example, associations were relatively stronger in Baltimore, MD and weaker in Los Angeles, CA. This research reinforces the findings of previous studies examining the impact of redlining on firearm death across the extent of the entire country in 21 cities and claim that HOLC grades are associated with present-day violence.

Associations Between Surface Mining Airsheds and Birth Outcomes in Central Appalachia at Multiple Spatial Scales

A considerable body of research exists outlining ecological impacts of surface coal mining, but less work has explicitly focused on human health, and few studies have examined potential links between health and surface coal mining at fine spatial scales. In particular, relationships between individual birth outcomes and exposure to air contaminants from coal mining activities has received little attention. Central Appalachia (portions of Virginia, West Virginia, Kentucky, and Tennessee, USA), our study area, has a history of resource extraction, and epidemiologic research notes that the region experiences a greater level of adverse health outcomes compared to the rest of the country that are not fully explained by socioeconomic and behavioral factors. The purpose of this study is to examine associations between surface mining and birth outcomes at four spatial scales: individual, Census tract, county, and across county-sized grid cells. Notably, this study is among the first to examine these associations at the individual scale, providing a more direct measure of exposure and outcome. Airsheds were constructed for surface mines using an atmospheric trajectory model. We then implemented linear (birthweight) and logistic (preterm birth [PTB]) regression models to examine associations between airsheds and birth outcomes, which were geocoded to home address for individual analyses and then aggregated for areal unit analyses, while controlling for a number of demographic variables. This study found that surface mining airsheds are significantly associated with PTB and decreased birthweight at all four spatial scales, suggesting that surface coal mining activities impact birth outcomes via airborne contaminants.

Association between the geographic accessibility of PrEP and PrEP use among MSM in nonurban areas

PURPOSE

The US HIV epidemic has become a public health issue that increasingly affects men who have sex with men (MSM), including those residing in nonurban areas. Increasing access to pre-exposure prophylaxis (PrEP) in nonurban areas will prevent HIV acquisition and could address the growing HIV epidemic. No studies have quantified the associations between PrEP access and PrEP use among nonurban MSM.

METHODS

Using 2020 PrEP Locator data and American Men's Internet Survey data, we conducted multilevel log-binomial regression to examine the association between area-level geographic accessibility of PrEP-providing clinics and individual-level PrEP use among MSM residing in nonurban areas in the United States.

FINDINGS

Of 4,792 PrEP-eligible nonurban MSM, 20.1% resided in a PrEP desert (defined as more than a 30-minute drive to access PrEP), and 15.2% used PrEP in the past 12 months. In adjusted models, suburban MSM residing in PrEP deserts were less likely to use PrEP in the past year (adjusted prevalence ratio [aPR] = 0.35; 95% confidence interval [CI] = 0.15, 0.80) than suburban MSM not residing in PrEP deserts, and other nonurban MSM residing in PrEP deserts were less likely to use PrEP in the past year (aPR = 0.75; 95% CI = 0.60, 0.95) than other nonurban MSM not residing in PrEP deserts.

CONCLUSIONS

Structural interventions designed to decrease barriers to PrEP access that are unique to nonurban areas in the United States are needed to address the growing HIV epidemic in these communities.

Geospatial Analysis of Organ Transplant Referral Regions

IMPORTANCE

System and center-level interventions to improve health equity in organ transplantation benefit from robust characterization of the referral population served by each transplant center. Transplant referral regions (TRRs) define geographic catchment areas for transplant centers in the US, but accurately characterizing the demographics of populations within TRRs using US Census data poses a challenge.

OBJECTIVE

To compare 2 methods of linking US Census data with TRRs-a geospatial intersection method and a zip code cross-reference method.

DESIGN, SETTING, AND PARTICIPANTS

This cohort study compared spatial congruence of spatial intersection and zip code cross-reference methods of characterizing TRRs at the census block level. Data included adults aged 18 years and older on the waiting list for kidney transplant from 2008 through 2018.

EXPOSURES

End-stage kidney disease.

MAIN OUTCOMES AND MEASURES

Multiple assignments, where a census tract or block group crossed the boundary between 2 hospital referral regions and was assigned to multiple different TRRs; misassigned area, the portion of census tracts or block groups assigned to a TRR using either method but fall outside of the TRR boundary.

RESULTS

In total, 102 TRRs were defined for 238 transplant centers. The zip code cross-reference method resulted in 4627 multiple-assigned census block groups (representing 18% of US land area assigned to TRRs), while the spatial intersection method eliminated this problem. Furthermore, the spatial method resulted in a mean and median reduction in misassigned area of 65% and 83% across all TRRs, respectively, compared with the zip code cross-reference method.

CONCLUSIONS AND RELEVANCE

In this study, characterizing populations within TRRs with census block groups provided high spatial resolution, complete coverage of the country, and balanced population counts. A spatial intersection approach avoided errors due to duplicative and incorrect assignments, and allowed more detailed and accurate characterization of the sociodemographics of populations within TRRs; this approach can enrich transplant center knowledge of local referral populations, assist researchers in understanding how social determinants of health may factor into access to transplant, and inform interventions to improve heath equity.

Disaggregating disparities: A case study of heterogenous COVID-19 disparities across waves, geographies, social vulnerability, and political lean in Louisiana

While the first wave of COVID-19 primarily impacted urban areas, subsequent waves were more widespread. Most analysis of Covid-19 rates examine state or metropolitan areas, ignoring potential heterogeneity within states and metro areas, over time, and between populations with differing contextual and compositional features. In this study, we compare spatial and temporal trends in Covid-19 cases and deaths in Louisiana, USA, over time and across populations and geographies (New Orleans, other urban areas, suburban, rural) and parish-level political lean. We employ publicly available longitudinal census tract and parish-level Covid-19 data reported from February 27th, 2020 to October 27th, 2021. We find that incidence and mortality rates were initially highest in New Orleans and Democratic areas and higher in other geographies and more conservative areas during subsequent waves. We also find wide relative disparities during the first wave, where increased social vulnerability was associated with increased positivity and incidence across geographies and political contexts. However, relative disparities diverged by geography and political lean and outcome across the remaining waves. This work draws attention to the differential rates of Covid-19 cases and deaths by geography, time, and population throughout the pandemic, and importance of political and geographic boundaries for rates of Covid-19.

Spatial and temporal pattern of cannabis use disorder in California 2010-2019

As cannabis use is being legalized in an increasing number of states, it is important to understand the changing dynamic of the risk in cannabis use disorder (CUD). Shape-based time-series clustering was used to identify ZIP Code Tabulation Areas (ZCTAs) with similar changing pattern in CUD over time. We conducted a cross-sectional logistic regression analysis to investigate the most recent ZCTA socio-demographic characteristics in relation to the changing CUD rates. The emergency discharge rates generally increased during 2010-2016. Increase during 2017-2019 was found in Sacramento and Santa Barbara County. Approximately 13% of ZCTAs showed an increasing trend of hospitalization discharge during 2017-2019. Males and non-Hispanic Black had larger increase than other groups during 2017-2019. The recent growing trend was found associated with greater racial diversity and rural ZCTAs. The findings from this study hold promise for local public health officials to adjust the cannabis intervention strategies in target districts and improve overall health outcomes.

Spatial Syndromic Surveillance and COVID-19 in the U.S.: Local Cluster Mapping for Pandemic Preparedness

Maps have become the de facto primary mode of visualizing the COVID-19 pandemic, from identifying local disease and vaccination patterns to understanding global trends. In addition to their widespread utilization for public communication, there have been a variety of advances in spatial methods created for localized operational needs. While broader dissemination of this more granular work is not commonplace due to the protections under Health Insurance Portability and Accountability Act (HIPAA), its role has been foundational to pandemic response for health systems, hospitals, and government agencies. In contrast to the retrospective views provided by the aggregated geographies found in the public domain, or those often utilized for academic research, operational response requires near real-time mapping based on continuously flowing address level data. This paper describes the opportunities and challenges presented in emergent disease mapping using dynamic patient data in the response to COVID-19 for northeast Ohio for the period 2020 to 2022. More specifically it shows how a new clustering tool developed by geographers in the initial phases of the pandemic to handle operational mapping continues to evolve with shifting pandemic needs, including new variant surges, vaccine targeting, and most recently, testing data shortfalls. This paper also demonstrates how the geographic approach applied provides the framework needed for future pandemic preparedness.

Exposure to industrial hog operations and gastrointestinal illness in North Carolina, USA

With 9 million hogs, North Carolina (NC) is the second leading hog producer in the United States. Most hogs are housed at concentrated animal feeding operations (CAFOs), where millions of tons of hog waste can pollute air and water with fecal pathogens that can cause diarrhea, vomiting, and/or nausea (known as acute gastrointestinal illness (AGI)). We used NC's ZIP code-level emergency department (ED) data to calculate rates of AGI ED visits (2016-2019) and swine permit data to estimate hog exposure. Case exposure was estimated as the inverse distances from each hog CAFO to census block centroids, weighting with Gaussian decay and by manure amount per CAFO, then aggregated to ZIP code using population weights. We compared ZIP codes in the upper quartile of hog exposure ("high hog exposed") to those without hog exposure. Using inverse probability of treatment weighting, we created a control with similar demographics to the high hog exposed population and calculated rate ratios using quasi-Poisson models. We examined effect measure modification of rurality and race using adjusted models. In high hog exposed areas compared to areas without hog exposure, we observed a 11% increase (95% CI: 1.06, 1.17) in AGI rate and 21% increase specifically in rural areas (95% CI: 0.98, 1.43). When restricted to rural areas, we found an increased AGI rate among American Indian (RR = 4.29, 95% CI: 3.69, 4.88) and Black (RR = 1.45, 95% CI: 0.98, 1.91) residents. The association was stronger during the week after heavy rain (RR = 1.41, 95% CI: 1.19, 1.62) and in areas with both poultry and swine CAFOs (RR = 1.52, 95% CI: 1.48, 1.57). Residing near CAFOs may increase rates of AGI ED visits. Hog CAFOs are disproportionally built near rural Black and American Indian communities in NC and are associated with increased AGI most strongly in these populations.

Understanding Racial and Ethnic Inequities in Uptake and Outcomes Following Multigene Prognostic Testing in Early Breast Cancer: The Promise of Real-World Data

In the past decades, multigene prognostic testing, such as Oncotype DX (ODX), has been increasingly used to inform treatment decisions for patients with early-stage breast cancer. This advance in precision oncology has increased existing concerns about differential access to genomic testing across racial and ethnic groups. The investigation by Moore and colleagues, analyzing real-world data from the National Cancer Database, shows that patients of color with breast cancer were less likely to receive ODX testing and Black patients were more likely to have a high risk Recurrence Score (RS) compared with White patients. This study emphasizes that the appropriate adoption of ODX testing is critical to promote equitable cancer care for patients with breast cancer. The reported associations on overall survival across specific racial and ethnic groups provided here give additional insight to the known associations between the ODX RS and outcomes of distant recurrence and cancer-specific mortality. Analyses of contemporary, real-world data from diverse populations with long-term follow-up should continue to keep pace with the expansion of precision breast cancer care to better understand and mitigate potentially widening inequities in genomic testing. See related article by Moore et al., p. 821.

Evaluating How Safety-Net Hospitals Are Identified: Systematic Review and Recommendations

Objective:

To systematically review how safety-net hospitals' status is identified and defined, discuss current definitions' limitations, and provide recommendations for a new classification and evaluation framework.

Data Sources:

Safety-net hospital-related studies in the MEDLINE database published before May 16, 2019.

Study Design:

Systematic review of the literature that adheres to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.

Data Collection/Extraction Methods:

We followed standard selection protocol, whereby studies went through an abstract review followed by a full-text screening for eligibility. For each included study, we extracted information about the identification method itself, including the operational definition, the dimension(s) of disadvantage reflected, study objective, and how safety-net status was evaluated.

Principal Findings:

Our review identified 132 studies investigating safety-net hospitals. Analysis of identification methodologies revealed substantial heterogeneity in the ways disadvantage is defined, measured, and summarized at the hospital level, despite a 4.5-fold increase in studies investigating safety-net hospitals for the past decade. Definitions often exclusively used low-income proxies captured within existing health system data, rarely incorporated external social risk factor measures, and were commonly separated into distinct safety-net status categories when analyzed.

Conclusions:

Consistency in research and improvement in policy both require a standard definition for identifying safety-net hospitals. Yet no standardized definition of safety-net hospitals is endorsed and existing definitions have key limitations. Moving forward, approaches rooted in health equity theory can provide a more holistic framework for evaluating disadvantage at the hospital level. Furthermore, advancements in precision public health technologies make it easier to incorporate detailed neighborhood-level social determinants of health metrics into multidimensional definitions. Other countries, including the United Kingdom and New Zealand, have used similar methods of identifying social need to determine more accurate assessments of hospital performance and the development of policies and targeted programs for improving outcomes.

Association of Socioeconomic Status and Infarct Volume With Functional Outcome in Patients With Ischemic Stroke

IMPORTANCE

Long-term disability after stroke is associated with socioeconomic status (SES). However, the reasons for such disparities in outcomes remain unclear.

OBJECTIVE

To assess whether lower SES is associated with larger admission infarct volume and whether initial infarct volume accounts for the association between SES and long-term disability.

DESIGN, SETTING, AND PARTICIPANTS

This cohort study was conducted in a prospective, consecutive population (n = 1256) presenting with acute ischemic stroke who underwent magnetic resonance imaging (MRI) within 24 hours of admission. Patients were recruited in Massachusetts General Hospital, Boston, from May 31, 2009, to December 31, 2011. Data were analyzed from May 1, 2019, until June 30, 2020.

MAIN OUTCOMES AND MEASURES

Initial stroke severity (within 24 hours of presentation) was determined using clinical (National Institutes of Health Stroke Scale [NIHSS]) and imaging (infarct volume by diffusion-weighted MRI) measures. Stroke etiologic subtypes were determined using the Causative Classification of Ischemic Stroke algorithm. Long-term stroke disability was measured using the modified Rankin Scale. Socioeconomic status was estimated using zip code-derived median household income and census block group-derived area deprivation index (ADI). Regression and mediation analyses were performed.

RESULTS

A total of 1098 patients had imaging and SES data available (mean [SD] age, 68.1 [15.7] years; 607 men [55.3%]). Income was inversely associated with initial infarct volume (standardized β, -0.074 [95% CI, -0.127 to -0.020]; P = .007), initial NIHSS (standardized β, -0.113 [95% CI, -0.171 to -0.054]; P < .001), and long-term disability (standardized β, -0.092 [95% CI, -0.149 to -0.035]; P = .001), which remained significant after multivariable adjustments. Initial stroke severity accounted for 64% of the association between SES and long-term disability (standardized β, -0.063 [95% CI, -0.095 to -0.029]; P < .05). Findings were similar when SES was alternatively assessed using ADI.

CONCLUSIONS AND RELEVANCE

The findings of this cohort study suggest that lower SES is associated with larger infarct volumes on presentation. These SES-associated differences in initial stroke severity accounted for most of the subsequent disparities in long-term disability in this study. These findings shift the culpability for SES-associated disparities in poststroke disability from poststroke factors to those that precede presentation.

COVID-19 Vaccine Rollouts and the Reproduction of Urban Spatial Inequality: Disparities Within Large US Cities in March and April 2021 by Racial/Ethnic and Socioeconomic Composition

Rollouts of COVID-19 vaccines in the USA were opportunities to redress disparities that surfaced during the pandemic. Initial eligibility criteria, however, neglected geographic, racial/ethnic, and socioeconomic considerations. Marginalized populations may have faced barriers to then-scarce vaccines, reinforcing disparities. Inequalities may have subsided as eligibility expanded. Using spatial modeling, we investigate how strongly local vaccination levels were associated with socioeconomic and racial/ethnic composition as authorities first extended vaccine eligibility to all adults. We harmonize administrative, demographic, and geospatial data across postal codes in eight large US cities over 3 weeks in Spring 2021. We find that, although vaccines were free regardless of health insurance coverage, local vaccination levels in March and April were negatively associated with poverty, enrollment in means-tested public health insurance (e.g., Medicaid), and the uninsured population. By April, vaccination levels in Black and Hispanic communities were only beginning to reach those of Asian and White communities in March. Increases in vaccination were smaller in socioeconomically disadvantaged Black and Hispanic communities than in more affluent, Asian, and White communities. Our findings suggest vaccine rollouts contributed to cumulative disadvantage. Populations that were left most vulnerable to COVID-19 benefited least from early expansions in vaccine availability in large US cities.

Geographic variation in and contextual factors related to biguanide adherence amongst medicaid enrolees with type 2 Diabetes Mellitus

Much is known about the adverse impacts on diabetes outcomes of non-adherence to diabetes medication. Less is known about how adherence to diabetes medication varies geographically, and the correspondence of this variation to social and contextual factors. Using pharmacy claims data over a two-year period, this study analysed non-adherence to biguanide medication for N=24,387 adult Medicaid enrolees diagnosed with Type 2 Diabetes Mellitus (T2DM) and residing in Ohio. Spatial analysis was used to detect clusters of census tract level rates of non-adherence, defined as the proportion of patients below the Proportion Days Covered (PDC) threshold of 80%, the level at which patients have a reasonable likelihood of achieving most clinical benefit from their medication. Multilevel models were used to understand associations between medication non-adherence and contextual factors including social vulnerability, urbanicity and distance to utilised pharmacy, with adjustment for individual-level covariates. These findings indicate that contextual factors are associated with medication non-adherence in Medicaid clients with T2DM. They suggest a need for spatially specific, multifaceted intervention programmes that target and/or account for the features of residential settings beyond individual and health system-level factors alone. While “environmental” considerations are often acknowledged, few intervention initiatives are predicated on explicit knowledge of spatially variable influences that can be targeted to enable and support medication adherence.

•

Medication adherence is a problem amongst Type 2 Diabetes patients on Medicaid.

•

Social and contextual factors' influence on medication adherence is underexplored.

•

Higher social vulnerability is associated with non-adherence to biguanides.

•

Adopting spatial analysis techniques enables geographic targeting of health risk.

Hurricane flooding and acute gastrointestinal illness in North Carolina

Hurricanes often flood homes and industries, spreading pathogens. Contact with pathogen-contaminated water can result in diarrhea, vomiting, and/or nausea, known collectively as acute gastrointestinal illness (AGI). Hurricanes Matthew and Florence caused record-breaking flooding in North Carolina (NC) in October 2016 and September 2018, respectively. To examine the relationship between hurricane flooding and AGI in NC, we first calculated the percent of each ZIP code flooded after Hurricanes Matthew and Florence. Rates of all-cause AGI emergency department (ED) visits were calculated from NC's ED surveillance system data. Using controlled interrupted time series, we compared AGI ED visit rates during the three weeks after each hurricane in ZIP codes with a third or more of their area flooded to the predicted rates had these hurricanes not occurred, based on AGI 2016-2019 ED trends, and controlling for AGI ED visit rates in unflooded areas. We examined alternative case definitions (bacterial AGI) and effect measure modification by race and age. We observed an 11% increase (rate ratio (RR): 1.11, 95% CI: 1.00, 1.23) in AGI ED visit rates after Hurricanes Matthew and Florence. This effect was particularly strong among American Indian patients and patients aged 65 years and older after Florence and elevated among Black patients for both hurricanes. Florence's effect was more consistent than Matthew's effect, possibly because little rain preceded Florence and heavy rain preceded Matthew. When restricted to bacterial AGI, we found an 85% (RR: 1.85, 95% CI: 1.37, 2.34) increase in AGI ED visit rate after Florence, but no increase after Matthew. Hurricane flooding is associated with an increase in AGI ED visit rate, although the strength of effect may depend on total storm rainfall or antecedent rainfall. American Indians and Black people-historically pushed to less desirable, flood-prone land-may be at higher risk for AGI after storms.

The Impact of Socioeconomic Deprivation on Clinical Outcomes for Pancreatic Adenocarcinoma at a High-volume Cancer Center: A Retrospective Cohort Analysis

OBJECTIVE

To assess the impact of a granular measure of SED on pancreatic surgical and cancer-related outcomes at a high-volume cancer center that employs a standardized clinic pathway.

SUMMARY OF BACKGROUND DATA

Prior research has shown that low socioeconomic status leads to less treatment and worse outcomes for PDAC. However, these studies employed inconsistent definitions and categorizations of socioeconomic status, aggregated individual socioeconomic data using large geographic areas, and lacked detailed clinicopathologic variables.

METHODS

We conducted a retrospective cohort study of 1552 PDAC patients between 2008 and 2015. Patients were stratified using the area deprivation index, a validated dataset that ranks census block groups based on SED. Multivariable models were used in the curative surgery cohort to predict the impact of SED on (1) grade 3/4 Clavien-Dindo complications, (2) initiation of adjuvant therapy, (3) completion of adjuvant therapy, and (4) overall survival.

RESULTS

Patients from high SED neighborhoods constituted 29.9% of the cohort. Median overall survival was 28 months. The rate of Clavien-Dindo grade 3/4 complications was 14.2% and completion of adjuvant therapy was 65.6%. There was no evidence that SED impacted surgical evaluation, receipt of curative-intent surgery, postoperative complications, receipt of adjuvant therapy or overall survival.

CONCLUSIONS

Although nearly one-quarter of curative-intent surgery patients were from high SED neighborhoods, this factor was not associated with measures of treatment quality or survival. These observations suggest that treatment at a high-volume cancer center employing a standardized clinical pathway may in part address socioeconomic disparities in pancreatic cancer.

The leading neighborhood-level predictors of drug overdose: A mixed machine learning and spatial approach

BACKGROUND

Drug overdose is a leading cause of unintentional death in the United States and has contributed significantly to a decline in life expectancy during recent years. To combat this health issue, this study aims to identify the leading neighborhood-level predictors of drug overdose and develop a model to predict areas at the highest risk of drug overdose using geographic information systems and machine learning (ML) techniques.

METHOD

Neighborhood-level (block group) predictors were grouped into three domains: socio-demographic factors, drug use variables, and protective resources. We explored different ML algorithms, accounting for spatial dependency, to identify leading predictors in each domain. Using geographically weighted regression and the best-performing ML algorithm, we combined the output prediction of three domains to produce a final ensemble model. The model performance was validated using classification evaluation metrics, spatial cross-validation, and spatial autocorrelation testing.

RESULTS

The variables contributing most to the predictive model included the proportion of households with food stamps, households with an annual income below $35,000, opioid prescription rate, smoking accessories expenditures, and accessibility to opioid treatment programs and hospitals. Compared to the error estimated from normal cross-validation, the generalized error of the model did not increase considerably in spatial cross-validation. The ensemble model using ML outperformed the GWR method.

CONCLUSION

This study identified strong neighborhood-level predictors that place a community at risk of experiencing drug overdoses, as well as protective factors. Our findings may shed light on several specific avenues for targeted intervention in neighborhoods at risk for high drug overdose burdens.

Community versus individual risk of SARS-CoV-2 infection in two municipalities of Louisiana, USA: An assessment of Area Deprivation Index (ADI) paired with seroprevalence data over time

OBJECTIVE

Determine whether an individual is at greater risk of severe acute respiratory distress syndrome coronavirus 2 (SARS-CoV-2) infection because of their community or their individual risk factors.

STUDY DESIGN AND SETTING

4,752 records from two large prevalence studies in New Orleans and Baton Rouge, Louisiana were used to assess whether zip code tabulation areas (ZCTA)-level area deprivation index (ADI) or individual factors accounted for risk of infection. Logistic regression models assessed associations of individual-level demographic and socioeconomic factors and the zip code-level ADI with SARS-CoV-2 infection.

RESULTS

In the unadjusted model, there were increased odds of infection among participants residing in high versus low ADI (both cities) and high versus mid-level ADI (Baton Rouge only) zip codes. When individual-level covariates were included, the odds of infection remained higher only among Baton Rouge participants who resided in high versus mid-level ADI ZCTAs. Several individual factors contributed to infection risk. After adjustment for ADI, race and age (Baton Rouge) and race, marital status, household size, and comorbidities (New Orleans) were significant.

CONCLUSIONS

While higher ADI was associated with higher risk of SARS-CoV-2 infection, individual-level participant characteristics accounted for a significant proportion of this association. Additionally, stage of the pandemic may affect individual risk factors for infection.

Crowdsourced Tick Image-Informed Updates to U.S. County Records of Three Medically Important Tick Species

Burgeoning cases of tick-borne disease present a significant public health problem in the United States. Passive tick surveillance gained traction as an effective way to collect epidemiologic data, and in particular, photograph-based tick surveillance can complement in-hand tick specimen identification to amass distribution data and related encounter demographics. We compared the Federal Information Processing Standards (FIPS) code of tick photos submitted to a free public identification service (TickSpotters) from 2014 to 2019 to published nationwide county reports for three tick species of medical concern: Ixodes scapularis Say (Ixodida: Ixodidae), Ixodes pacificus Cooley and Kohls (Ixodida: Ixodidae), and Amblyomma americanum Linneaus (Ixodida: Ixodidae). We tallied the number of TickSpotters submissions for each tick species according to "Reported" or "Established" criteria per county, and found that TickSpotters submissions represented more than half of the reported counties of documented occurrence, and potentially identified hundreds of new counties with the occurrence of these species. We detected the largest number of new county reports of I. scapularis presence in Michigan, North Carolina, and Texas. Tick image submissions revealed potentially nine new counties of occurrence for I. pacificus, and we documented the largest increase in new county reports of A. americanum in Kentucky, Illinois, Indiana, and Ohio. These findings demonstrate the utility of crowdsourced photograph-based tick surveillance as a complement to other tick surveillance strategies in documenting tick distributions on a nationwide scale, its potential for identifying new foci, and its ability to highlight at-risk localities that might benefit from tick-bite prevention education.

Sociodemographic and clinical features predictive of SARS-CoV-2 test positivity across healthcare visit-types

Background

Despite increased testing efforts and the deployment of vaccines, COVID-19 cases and death toll continue to rise at record rates. Health systems routinely collect clinical and non-clinical information in electronic health records (EHR), yet little is known about how the minimal or intermediate spectra of EHR data can be leveraged to characterize patient SARS-CoV-2 pretest probability in support of interventional strategies.

Methods and findings

We modeled patient pretest probability for SARS-CoV-2 test positivity and determined which features were contributing to the prediction and relative to patients triaged in inpatient, outpatient, and telehealth/drive-up visit-types. Data from the University of Washington (UW) Medicine Health System, which excluded UW Medicine care providers, included patients predominately residing in the Seattle Puget Sound area, were used to develop a gradient-boosting decision tree (GBDT) model. Patients were included if they had at least one visit prior to initial SARS-CoV-2 RT-PCR testing between January 01, 2020 through August 7, 2020. Model performance assessments used area-under-the-receiver-operating-characteristic (AUROC) and area-under-the-precision-recall (AUPR) curves. Feature performance assessments used SHapley Additive exPlanations (SHAP) values. The generalized pretest probability model using all available features achieved high overall discriminative performance (AUROC, 0.82). Performance among inpatients (AUROC, 0.86) was higher than telehealth/drive-up testing (AUROC, 0.81) or outpatient testing (AUROC, 0.76). The two-week test positivity rate in patient ZIP code was the most informative feature towards test positivity across visit-types. Geographic and sociodemographic factors were more important predictors of SARS-CoV-2 positivity than individual clinical characteristics.

Conclusions

Recent geographic and sociodemographic factors, routinely collected in EHR though not routinely considered in clinical care, are the strongest predictors of initial SARS-CoV-2 test result. These findings were consistent across visit types, informing our understanding of individual SARS-CoV-2 risk factors with implications for deployment of testing, outreach, and population-level prevention efforts.

Geographically targeted COVID-19 vaccination is more equitable and averts more deaths than age-based thresholds alone

COVID-19 mortality increases markedly with age and is also substantially higher among Black, Indigenous, and People of Color (BIPOC) populations in the United States. These two facts can have conflicting implications because BIPOC populations are younger than white populations. In analyses of California and Minnesota—demographically divergent states—we show that COVID vaccination schedules based solely on age benefit the older white populations at the expense of younger BIPOC populations with higher risk of death from COVID-19. We find that strategies that prioritize high-risk geographic areas for vaccination at all ages better target mortality risk than age-based strategies alone, although they do not always perform as well as direct prioritization of high-risk racial/ethnic groups. Vaccination schemas directly implicate equitability of access, both domestically and globally.

Associations of Race, Insurance, and Zip Code-Level Income with Nonadherence Diagnoses in Primary and Specialty Diabetes Care

INTRODUCTION

Evidence suggests that clinicians may view or label patients as nonadherent in a biased manner. Therefore, we performed a retrospective cohort analysis exploring associations between patient demographics and zip code-level income with the International Classification of Diseases, Tenth Version (ICD-10) diagnoses for nonadherence among type 2 diabetes mellitus (T2DM) patients, comparing primary and specialty care settings. Providers in the primary care group included internal medicine and family medicine physicians. In the specialty care group, providers included endocrinologists and diabetologists only.

METHODS

Participants were identified from 5 primary care and 4 endocrinology sites in the University of Pennsylvania Health System between January 1, 2015, and January 1, 2019. Demographics, hemoglobin A1c (HbA1c), and ICD-10 codes for T2DM and nonadherence were extracted from the electronic health record and analyzed in October 2019. Log-binomial regression models were used to estimate patients' risk of nonadherence labeling by race, insurance, and zip code-level median household income, controlling for patient characteristics and HbA1c as a proxy for diabetes self-management. Results were compared between primary and specialty care sites.

RESULTS

A total of 6072 patients aged 18-70 years were included in this study. Black race, Medicare, and Medicaid were associated with increased nonadherence labeling while controlling for patient characteristics ([ARR = 2.48, 95% CI: 2.01, 3.04], [ARR = 1.82, 95% CI: 1.50, 2.18], [ARR = 1.61, 95% CI: 1.32, 1.93], respectively). The results remained significant on adjustment with zip code-level income and showed no differences between primary and specialty sites. Lower-income zip codes showed a significant association with increased rates of nonadherence labeling.

CONCLUSIONS

Black race, non-private insurance, and lower-income zip codes were associated with disproportionately high rates of nonadherence labeling in both primary and specialty management of T2DM, possibly suggestive of racial or class bias.

Spatial Inequities in COVID-19 Testing, Positivity, Confirmed Cases, and Mortality in 3 U.S. Cities : An Ecological Study

BACKGROUND

Preliminary evidence has shown inequities in coronavirus disease 2019 (COVID-19)-related cases and deaths in the United States.

OBJECTIVE

To explore the emergence of spatial inequities in COVID-19 testing, positivity, confirmed cases, and mortality in New York, Philadelphia, and Chicago during the first 6 months of the pandemic.

DESIGN

Ecological, observational study at the ZIP code tabulation area (ZCTA) level from March to September 2020.

SETTING

Chicago, New York, and Philadelphia.

PARTICIPANTS

All populated ZCTAs in the 3 cities.

MEASUREMENTS

Outcomes were ZCTA-level COVID-19 testing, positivity, confirmed cases, and mortality cumulatively through the end of September 2020. Predictors were the Centers for Disease Control and Prevention Social Vulnerability Index and its 4 domains, obtained from the 2014-2018 American Community Survey. The spatial autocorrelation of COVID-19 outcomes was examined by using global and local Moran I statistics, and estimated associations were examined by using spatial conditional autoregressive negative binomial models.

RESULTS

Spatial clusters of high and low positivity, confirmed cases, and mortality were found, co-located with clusters of low and high social vulnerability in the 3 cities. Evidence was also found for spatial inequities in testing, positivity, confirmed cases, and mortality. Specifically, neighborhoods with higher social vulnerability had lower testing rates and higher positivity ratios, confirmed case rates, and mortality rates.

LIMITATIONS

The ZCTAs are imperfect and heterogeneous geographic units of analysis. Surveillance data were used, which may be incomplete.

CONCLUSION

Spatial inequities exist in COVID-19 testing, positivity, confirmed cases, and mortality in 3 large U.S. cities.

PRIMARY FUNDING SOURCE

National Institutes of Health.

Spatial Inequities in COVID-19 Testing, Positivity, Confirmed Cases, and Mortality in 3 U.S. Cities : An Ecological Study

BACKGROUND

Preliminary evidence has shown inequities in coronavirus disease 2019 (COVID-19)-related cases and deaths in the United States.

OBJECTIVE

To explore the emergence of spatial inequities in COVID-19 testing, positivity, confirmed cases, and mortality in New York, Philadelphia, and Chicago during the first 6 months of the pandemic.

DESIGN

Ecological, observational study at the ZIP code tabulation area (ZCTA) level from March to September 2020.

SETTING

Chicago, New York, and Philadelphia.

PARTICIPANTS

All populated ZCTAs in the 3 cities.

MEASUREMENTS

Outcomes were ZCTA-level COVID-19 testing, positivity, confirmed cases, and mortality cumulatively through the end of September 2020. Predictors were the Centers for Disease Control and Prevention Social Vulnerability Index and its 4 domains, obtained from the 2014-2018 American Community Survey. The spatial autocorrelation of COVID-19 outcomes was examined by using global and local Moran I statistics, and estimated associations were examined by using spatial conditional autoregressive negative binomial models.

RESULTS

Spatial clusters of high and low positivity, confirmed cases, and mortality were found, co-located with clusters of low and high social vulnerability in the 3 cities. Evidence was also found for spatial inequities in testing, positivity, confirmed cases, and mortality. Specifically, neighborhoods with higher social vulnerability had lower testing rates and higher positivity ratios, confirmed case rates, and mortality rates.

LIMITATIONS

The ZCTAs are imperfect and heterogeneous geographic units of analysis. Surveillance data were used, which may be incomplete.

CONCLUSION

Spatial inequities exist in COVID-19 testing, positivity, confirmed cases, and mortality in 3 large U.S. cities.

PRIMARY FUNDING SOURCE

National Institutes of Health.

Poverty and Excess Length of Hospital Stay in Neonatal Opioid Withdrawal Syndrome

OBJECTIVES

To study the impact of sociodemographic factors on length of stay (LOS) for infants with neonatal opioid withdrawal syndrome (NOWS) secondary to fetal opioid exposure.

METHODS

In this retrospective cohort study, we included term infants with NOWS, excluding those with other significant medical issues. Comprehensive clinical and sociodemographic data were collected. Multivariate regression modeling was used to identify factors which contributed to excess LOS, which was defined as the number of days beyond the standard monitoring and/or treatment protocol.

RESULTS

In all, 129 infants were identified; mean gestational age of 37.9 ± 1.3 weeks and mean body weight of 2880 ± 496 g. Among them, 68% of infants were exposed to opioids; 27% were exposed to methadone; and 67% required pharmacologic treatment. The degree of poverty was assessed using the Area Deprivation Index (ADI) based on the mother's address at the time of birth. Median LOS for treated infants was 23 days versus 8 days for those who did not need pharmacologic treatment. The median excess LOS was 4 days (range 0-24).Excess hospital days were strongly correlated with degree of deprivation in the mother's community (r = 0.55, P < 0.01). ADI remained a strong predictor of excess LOS, even when controlling for pharmacologic treatment, placement in state's custody, race, and gestational age at birth.

CONCLUSIONS

These results suggest poverty is associated with excess LOS and that early allocation of resources for at-risk families may help to reduce overall length of hospital stay.

Natural gas development, flaring practices and paediatric asthma hospitalizations in Texas

BACKGROUND

Recent advancements in drilling technology led to a rapid increase in natural gas development (NGD). Air pollution may be elevated in these areas and may vary by drilling type (conventional and unconventional), production volume and gas flaring. Impacts of NGD on paediatric asthma are largely unknown. This study quantifies associations between specific NGD activities and paediatric asthma hospitalizations in Texas.

METHODS

We leveraged a database of Texas inpatient hospitalizations between 2000 and 2010 at the zip code level by quarter to examine associations between NGD and paediatric asthma hospitalizations, where our primary outcome is 0 vs ≥1 hospitalization. We used quarterly production reports to assess additional drilling-specific exposures at the zip code-level including drilling type, production and gas flaring. We developed logistic regression models to assess paediatric asthma hospitalizations by zip code-quarter-year observations, thus capturing spatiotemporal exposure patterns.

RESULTS

We observed increased odds of ≥1 paediatric asthma hospitalization in a zip code per quarter associated with increasing tertiles of NGD exposure and show that spatiotemporal variation impacts results. Conventional drilling, compared with no drilling, is associated with odds ratios up to 1.23 [95% confidence interval (CI): 1.13, 1.34], whereas unconventional drilling is associated with odds ratios up to 1.59 (95% CI: 1.46, 1.73). Increasing production volumes are associated with increased paediatric asthma hospitalizations in an exposure-response relationship, whereas associations with flaring volumes are inconsistent.

CONCLUSIONS

We found evidence of associations between paediatric asthma hospitalizations and NGD, regardless of drilling type. Practices related to production volume may be driving these positive associations.

Geodemographies of Breastfeeding Support

BACKGROUND

Mother-to-mother breastfeeding support organizations provide important information and guidance for helping mothers initiate and maintain breastfeeding, postpartum. However, the availability of this support is limited by a constellation of barriers, including race, culture, socioeconomic status, and geography.

RESEARCH AIMS

To identify the geodemographic composition of communities where breastfeeding support was available from the mother-to-mother support organizations Breastfeeding USA and La Leche League, identify underlying issues of equity, and highlight locations where more support resources may be needed.

METHODS

The locations of mother-to-mother support meetings were collected by ZIP code (N = 180) and were combined with a geodemographic database and exploratory spatial data analysis to explore the compositional characteristics of communities served (N = 1,173).

RESULTS

Significant gaps in the geographic distribution of breastfeeding support existed. While many metropolitan areas benefited from numerous mother-to-mother support groups and peer counselors, the geographic footprint of this support favored communities that were white, affluent, and suburban.

CONCLUSION

Spatial analytics combined with geodemographic analysis provide a unique perspective into the diverse landscape of mother-to-mother breastfeeding support groups at a local level. Our results highlighted inequities in the distribution of support provided and prescriptive guidance regarding where more resources may be needed.

The association of area deprivation and state child health with respiratory outcomes of pediatric patients with cystic fibrosis in the United States

BACKGROUND

Differences in socioenvironmental exposures influence overall child health, but their association with pediatric cystic fibrosis (CF) outcomes is less clear. This observational study investigated the relationship between area-level socioeconomic deprivation, state child health, and CF respiratory outcomes in a national cohort.

METHODS

We assessed relationships between the 2015 area deprivation index, a composite measure of socioeconomic disadvantage; the 2016 child health index, a state-specific measure of overall child health; and CF respiratory outcomes in the 2016 CF Foundation Patient Registry.

RESULTS

The sample included 9934 individuals with CF, aged 6-18 years. In multiple regression analysis adjusted for demographic and clinical covariates, those residing in the worst tertile for area deprivation had 2.8% lower percent predicted forced expiratory volume in 1 s (ppFEV₁ ; 95% confidence interval [CI]: -4.1 to -1.5), 1.2 more intravenous (IV) treatment nights (CI: 0.1-2.4), and 20% higher odds of ≥2 pulmonary exacerbations (odds ratio [OR]: 1.2, CI: 1.0-1.5) than best-tertile counterparts. Children with CF in states at the worst tertile for child health had 2.3% lower ppFEV₁ (CI: -4.5 to -0.2), 2.2 more IV treatment nights (CI: 0.5-3.6), and 40% higher odds of ≥2 exacerbations (OR: 1.4, CI: 1.1-1.8) than best-tertile counterparts. State child health accounted for the association between area deprivation and multiple exacerbations and more IV treatment nights.

CONCLUSIONS

Both area socioeconomic characteristics and state child health play a role in pediatric CF outcomes. The residual association of the state child health with CF outcomes after controlling for area deprivation reflects the ability of state programs to mitigate the effect of poverty.

Neighborhood characteristics associated with COVID-19 burden-the modifying effect of age

BACKGROUND

Neighborhood characteristics have been linked to community incidence of COVID-19, but the modifying effect of age has not been examined.

OBJECTIVE

We adapted a neighborhood-wide analysis study (NWAS) design to systematically examine associations between neighborhood characteristics and COVID-19 incidence among different age groups.

METHODS

The number of daily cumulative cases of COVID-19 by zip code area in Illinois has been made publicly available by the Illinois Department of Public Health. The number of COVID-19 cases was reported for eight age groups (under 20, 20-29, 30-39, 40-49, 50-59, 60-69, 70-79, and 80+). We reviewed this data published from May 23 through June 17, 2020 with complete data for all eight age groups and linked the data to neighborhood characteristics measured by the American Community Survey (ACS). Geographic age-specific cumulative incidence (cases per 1000 people) of COVID-19 was calculated by dividing the number of daily cumulative cases by the population of the same age group at each zip code area. The association between individual characteristics and COVID-19 incidence was examined using Poisson regression models.

RESULTS

At the zip code level, neighborhood socioeconomic status was a more important risk factor of COVID-19 incidence in children and working-age adults than in seniors. Social demographics and housing conditions were important risk factors of COVID-19 incidence in older age groups. We additionally observed significant associations between transportation-related variables and COVID-19 incidences in multiple age groups.

SIGNIFICANCE

We concluded that age modified the association between neighborhood characteristics and COVID-19 incidence.

Social determinants of health and coronavirus disease 2019 in pregnancy

Background

The social and physical environments in which people live affect the emergence, prevalence, and severity of both infectious and noninfectious diseases. There are limited data on how such social determinants of health, including neighborhood socioeconomic conditions, affect the risk of severe acute respiratory syndrome coronavirus 2 infection and severity of coronavirus disease 2019 during pregnancy.

Objective

Our objective was to determine how social determinants of health are associated with severe acute respiratory syndrome coronavirus 2 infection and the severity of coronavirus disease 2019 illness in hospitalized pregnant patients in New York during the global coronavirus disease 2019 pandemic.

Study Design

This cross-sectional study evaluated all pregnant patients who delivered and had polymerase chain reaction testing for severe acute respiratory syndrome coronavirus 2 between March 15, 2020, and June 15, 2020, at 7 hospitals within Northwell Health, the largest academic health system in New York. During the study period, universal severe acute respiratory syndrome coronavirus 2 testing protocols were implemented at all sites. Polymerase chain reaction testing was performed using nasopharyngeal swabs. Patients were excluded if the following variables were not available: polymerase chain reaction results, race, ethnicity, or zone improvement plan (ZIP) code of residence. Clinical data were obtained from the enterprise electronic health record system. For each patient, ZIP code was used as a proxy for neighborhood. Socioeconomic characteristics were determined by linking to ZIP code data from the United States Census Bureau's American Community Survey and the Internal Revenue Service's Statistics of Income Division. Specific variables of interest included mean persons per household, median household income, percent unemployment, and percent with less than high school education. Medical records were manually reviewed for all subjects with positive polymerase chain reaction test results to correctly identify symptomatic patients and then classify those subjects using the National Institutes of Health severity of illness categories. Classification was based on the highest severity of illness throughout gestation and not necessarily at the time of presentation for delivery.

Results

A total of 4873 patients were included in the study. The polymerase chain reaction test positivity rate was 11% (n=544). Among this group, 359 patients (66%) were asymptomatic or presymptomatic, 115 (21%) had mild or moderate coronavirus disease 2019, and 70 (13%) had severe or critical coronavirus disease 2019. On multiple logistic regression modeling, pregnant patients who had a positive test result for severe acute respiratory syndrome coronavirus 2 were more likely to be younger or of higher parity, belong to minoritized racial and ethnic groups, have public health insurance, have limited English proficiency, and reside in low-income neighborhoods with less educational attainment. On ordinal logit regression modeling, obesity, income and education were associated with coronavirus disease 2019 severity.

Conclusion

Social and physical determinants of health play a role in determining the risk of infection. The severity of coronavirus disease 2019 illness was not associated with race or ethnicity but was associated with maternal obesity and neighborhood level characteristics such as educational attainment and household income.