Defining Spatial Epidemiology: A Systematic Review and Re-orientation.

BACKGROUND: Spatial epidemiology has emerged as an important subfield of epidemiology over the past quarter century. We trace the origins of spatial epidemiology and note that its emergence coincided with technological developments in spatial statistics and geography. We hypothesize that spatial epidemiology makes important contributions to descriptive epidemiology and analytic risk-factor studies but is not yet aligned with epidemiology's current focus on causal inference and intervention. METHODS: We conducted a systematic review of studies indexed in PubMed that used the term "spatial epidemiolog*" in the title, abstract, or keywords. Excluded articles were not written in English, examined disease in animals, or reported biologic pathogen distribution only. We coded the included papers into five categories (review, demonstration of method, descriptive, analytic, and intervention) and recorded the unit of analysis (i.e., individual vs. ecological). We additionally examined articles coded as analytic ecologic studies using scales for lexical content. RESULTS: A total of 482 articles met the inclusion criteria, including 76 reviews, 117 demonstrations of methods, 122 descriptive studies, 167 analytic studies, and 0 intervention studies. Demonstration studies were most common from 2006 to 2014, and analytic studies were most common after 2015. Among the analytic ecologic studies, those published in later years used more terms relevant to spatial statistics (incidence rate ratio =1.3; 95% confidence interval [CI] = 1.1, 1.5) and causal inference (incidence rate ratio =1.1; 95% CI = 1.1, 1.2). CONCLUSIONS: Spatial epidemiology is an important and growing subfield of epidemiology. We suggest a re-orientation to help align its practice with the goals of contemporary epidemiology.

Examining racial and ethnic heat exposure disparities in New York City (NYC) across different spatial and political scales through geographic effect measure modification.

Structural racism in the United States has resulted in neighborhoods with higher proportions of non-Hispanic Black (Black) or Hispanic/Latine residents having more features that intensify, and less that cool, the local-heat environment. This study identifies areas of New York City (NYC) where racial/ethnic heat exposure disparities are concentrated. We analyzed data from the 2013-2017 American Community Survey, U.S Landsat-8 Analysis Ready Data on summer surface temperatures, and NYC Land Cover Dataset at the census tract-level (n = 2098). Four cross-sectional regression modeling strategies were used to estimate the overall City-wide association, and associations across smaller intra-city areas, between tract-level percent of Black and percent Hispanic/Latine residents and summer day surface temperature, adjusting for altitude, shoreline, and nature-cover: overall NYC linear, borough-specific linear, Community District-specific linear, and geographically weighted regression models. All three linear regressions identified associations between neighborhood racial and ethnic composition and summer day surface temperatures. The geographically weighted regression models, which address the issue of spatial autocorrelation, identified specific locations (such as northwest Bronx, central Brooklyn, and uptown Manhattan) within which racial and ethnic disparities for heat exposures are concentrated. Through examining the overall effects and geographic effect measure modification across spatial scales, the results of this study identify specific geographic areas for intervention to mitigate heat exposure disparities experienced by Black and Hispanic/Latine NYC residents.

The CHANGE (Climate Health ANalysis Grading Evaluation) tool for weight of evidence reviews on climate change and health research.

BACKGROUND: Climate change has been identified as one of the biggest threats to human health. Despite this claim, there are no standardized tools that assess the rigor of published literature for use in weight of evidence (WOE) reviews. Standardized assessment tools are essential for creating clear and comparable WOE reviews. As such, we developed a standardized tool for evaluating the quality of climate change and health studies focused on evaluating studies that quantify exposure-response relationships and studies that implement and/or evaluate adaptation interventions. METHODS: The authors explored systematic-review methodology to enhance transparency and increase efficiency in summarizing and synthesizing findings from studies on climate change and health research. The authors adapted and extended existing WOE methods to develop the CHANGE (Climate Health ANalysis Grading Evaluation) tool. The resulting assessment tool has been refined through application and subsequent team input. RESULTS: The CHANGE tool is a two-step standardized tool for systematic review of climate change and health studies of exposure-response relationships and adaptation intervention studies. Step one of the CHANGE tool aims to classify studies included in weight-of-evidence reviews and step two assesses the quality and presence of bias in the climate change and health studies. CONCLUSION: The application of the CHANGE tool in WOE reviews of climate change and health will lead to increased comparability, objectivity, and transparency within this research area.

The Creation of a Multidomain Neighborhood Environmental Vulnerability Index Across New York City.

Compared to previous studies commonly using a single summary score, we aimed to construct a multidomain neighborhood environmental vulnerability index (NEVI) to characterize the magnitude and variability of area-level factors with the potential to modify the association between environmental pollutants and health effects. Using the Toxicological Prioritization Index framework and data from the 2015-2019 U.S. Census American Community Survey and the 2020 CDC PLACES Project, we quantified census tract-level vulnerability overall and in 4 primary domains (demographic, economic, residential, and health status), 24 subdomains, and 54 distinct area-level features for New York City (NYC). Overall and domain-specific indices were calculated by summing standardized feature values within the subdomains and then aggregating and weighting based on the number of features within each subdomain within equally-weighted primary domains. In citywide comparisons, NEVI was correlated with multiple existing indices, including the Neighborhood Deprivation Index (r = 0.91) and Social Vulnerability Index (r = 0.87) but provided additional information on features contributing to vulnerability. Vulnerability varied spatially across NYC, and hierarchical cluster analysis using subdomain scores revealed six patterns of vulnerability across domains: 1) low in all, 2) primarily low except residential, 3) medium in all, 4) high demographic, economic, and residential 5) high economic, residential, and health status, and 6) high demographic, economic and health status. Created using methods that offer flexibility for theory-based construction, NEVI provided detailed vulnerability metrics across domains that can inform targeted research and public health interventions aimed at reducing the health impacts from environmental exposures across urban centers.

Urinary metabolites of polycyclic aromatic hydrocarbons and short-acting beta agonist or systemic corticosteroid asthma medication use within NHANES.

BACKGROUND: Within cross-sectional studies like the U.S. National Health and Nutritional Examination Survey (NHANES), researchers have observed positive associations between polycyclic aromatic hydrocarbon (PAH) exposure and asthma diagnosis. It is unclear whether similar relationships exist for measures of acute asthma outcomes, including short-term asthma medication use to alleviate symptoms. We examined the relationship between markers of recent PAH exposure and 30-day short-acting beta agonist (SABA) or systemic corticosteroid use, an indicator for recent asthma symptoms. MATERIALS AND METHODS: For 16,550 children and adults across multiple waves of NHANES (2005-2016), we fit quasi-Poisson multivariable regression models to describe the association between urinary 1-hydroxypyrene (a metabolite of PAH) and SABA or systemic corticosteroid use. We assessed for effect modification by age group and asthma controller medication use. All models were adjusted for urinary creatinine, age, female/male designation, race/ethnicity, poverty, insurance coverage, and serum cotinine. RESULTS: After controlling for confounding, an increase of one standard deviation of 1-hydroxypyrene was associated with greater prevalence of recent SABA or systemic corticosteroid use (PR: 1.06, 95% CI: 1.03-1.10). The results were similar among those with ever asthma diagnosis and across urine creatinine dilution methods. We did not observe effect modification by age group (p-interaction = 0.22) or asthma controller medication use (p-interaction = 0.73). CONCLUSION: Markers of recent PAH exposure was positively associated with SABA or systemic corticosteroid use, across various urine dilution adjustment methods. It is important to ensure appropriate temporality between exposures and outcomes in cross-sectional studies.

Screening for Excessive Alcohol Consumption in Emergency Departments: A Nationwide Assessment of Emergency Department Physicians.

OBJECTIVE: To assess current screening practices for excessive alcohol consumption, as well as perceived barriers, perceptions, and attitudes toward performing this screening among emergency department (ED) physicians. DESIGN: A brief online assessment of screening practices for excessive drinking was disseminated electronically to a representative panel of ED physicians from November 2016 to January 2017. Descriptive statistics were calculated on the frequency of alcohol screening, factors affecting screening, and attitudes toward screening. SETTING: An online assessment was sent to a national panel of ED physicians. PARTICIPANTS: A panel of ED physicians who volunteered to be part of the American College of Emergency Physicians Emergency Medicine Practice Research Network survey panel. MAIN OUTCOME MEASURE: The primary outcome measures were the percentage of respondents who reported screening for excessive alcohol consumption and the percentage of respondents using a validated excessive alcohol consumption screening tool. RESULTS: Of the 347 ED physicians evaluated (38.6% response rate), approximately 16% reported "always/usually," 70% "sometimes," and 14% "never" screening adult patients (≥18 years) for excessive alcohol use. Less than 20% of the respondents who screened for excessive drinking used a recommended screening tool. Only 10.5% of all respondents (15.4% "always," 9.5% "sometimes" screened) received an electronic health record (EHR) reminder to screen for excessive alcohol use. Key barriers to screening included limited time (66.2%) and treatment options for patients with drinking problems (43.1%). CONCLUSIONS: Only 1 in 6 ED physicians consistently screened their patients for excessive drinking. Increased use of EHR reminders and other systems interventions (eg, electronic screening and brief intervention) could help improve the delivery of screening and follow-up services for excessive drinkers in EDs.

Individual- and neighborhood-level socioeconomic status and risk of aggressive breast cancer subtypes in a pooled cohort of women from Kaiser Permanente Northern California.

BACKGROUND: Low socioeconomic status (SES) has been associated with a higher risk of aggressive breast cancer (BC) subtypes, but few studies have examined the independent effects of both neighborhood-level socioeconomic status (nSES) and individual-level SES measures. METHODS: This study included 5547 women from the Pathways and Life After Cancer Epidemiology cohorts who were diagnosed with invasive BC. Generalized estimating equation models were used to examine associations of nSES (a composite score based on income, poverty, education, occupation, employment, rent, and house value) and individual-level SES (income and education) with BC subtypes: luminal B (LumB), Her2-enriched (Her2-e), and triple-negative breast cancer (TNBC) relative to luminal A (LumA). Models controlled for age, race, nativity, stage, days from diagnosis to survey, and study cohort and simultaneously for nSES and individual-level SES. RESULTS: In fully adjusted models, low nSES was significantly associated with the LumB (odds ratio for quartile 1 vs quartile 4 [ORQ1vQ4 ], 1.31; 95% confidence interval [CI], 1.11-1.54; P for trend = .005) and TNBC subtypes (ORQ1vQ4 , 1.32; 95% CI, 1.02-1.71; P for trend = .037) relative to LumA. Conversely, individual education was significantly associated with only the Her2-e subtype (odds ratio for high school degree or less vs postgraduate, 1.68; 95% CI, 1.03-2.75; P for trend = .030) relative to LumA. Individual income was not significantly associated with any BC subtype. CONCLUSIONS: nSES and individual-level SES are independently associated with different BC subtypes; specifically, low nSES and individual-level education are independent predictors of more aggressive BC subtypes relative to LumA.

Risk of Distal Embolization From tPA (Tissue-Type Plasminogen Activator) Administration Prior to Endovascular Stroke Treatment.

BACKGROUND AND PURPOSE: In large artery occlusion stroke, both intravenous (IV) tPA (tissue-type plasminogen activator) and endovascular stroke treatment (EST) are standard-of-care. It is unknown how often tPA causes distal embolization, in which a procedurally accessible large artery occlusion is converted to a more distal and potentially inaccessible occlusion. METHODS: We analyzed data from a decentralized stroke telemedicine program in an integrated healthcare delivery system covering 21 hospitals, with 2 high-volume EST centers. We captured all cases sent for EST and examined the relationship between IV tPA administration and the rate of distal embolization, the rate of target recanalization (modified Treatment in Cerebral Infarction scale 2b/3), clinical improvement before EST, and short-term and long-term clinical outcomes. RESULTS: Distal embolization before EST was quite common (63/314 [20.1%]) and occurred more often after IV tPA before EST (57/229 [24.9%]) than among those not receiving IV tPA (6/85 [7.1%]; P<0.001). Distal embolization was associated with an inability to attempt EST: after distal embolization, 26/63 (41.3%) could not have attempted EST because of the new clot location, while in cases without distal embolization, only 8/249 (3.2%) were unable to have attempted EST (P<0.001). Among patients who received IV tPA, 13/242 (5.4%) had sufficient symptom improvement that a catheter angiogram was not performed; 6/342 (2.5%) had improvement to within 2 points of their baseline NIHSS. At catheter angiogram, 2/229 (0.9%) of patients who had received tPA had complete recanalization without distal embolization. Both IV tPA and EST recanalization were associated with improved long-term outcome. CONCLUSIONS: IV tPA administration before EST for large artery occlusion is associated with distal embolization, which in turn may reduce the chance that EST can be attempted and recanalization achieved. At the same time, some IV tPA-treated patients show symptomatic improvement and complete recanalization. Because IV tPA is associated with both distal embolization and improved long-term clinical outcome, there is a need for prospective clinical trials testing the net benefit or harm of IV tPA before EST.

Assessment of Bias in Estimates of Sexual Network Degree using Prospective Cohort Data.

BACKGROUND: Sexual network degree, a count of ongoing partnerships, plays a critical role in the transmission dynamics of human immunodeficiency virus and other sexually transmitted infections. Researchers often quantify degree using self-reported cross-sectional data on the day of survey, which may result in bias because of uncertainty about future sexual activity. METHODS: We evaluated the bias of a cross-sectional degree measure with a prospective cohort study of men who have sex with men (MSM). At baseline, we asked men about whether recent sexual partnerships were ongoing. We confirmed the true, ongoing status of those partnerships at baseline at follow-up. With logistic regression, we estimated the partnership-level predictors of baseline measure accuracy. With Poisson regression, we estimated the longitudinally confirmed degree as a function of baseline predicted degree. RESULTS: Across partnership types, the baseline ongoing status measure was 70% accurate, with higher negative predictive value (91%) than positive predictive value (39%). Partnership exclusivity and racial pairing were associated with higher accuracy. Baseline degree generally overestimated confirmed degree. Bias, or number of ongoing partners different than predicted at baseline, was -0.28 overall, ranging from -1.91 to -0.41 for MSM with any ongoing partnerships at baseline. Comparing MSM of the same baseline degree, the level of bias was stronger for black compared with white MSM, and for younger compared with older MSM. CONCLUSIONS: Research studies may overestimate degree when it is quantified cross-sectionally. Adjustment and structured sensitivity analyses may account for bias in studies of human immunodeficiency virus or sexually transmitted infection prevention interventions.

Differences in Smoking Behavior by Nativity, Race/Ethnicity, and Education among Women Diagnosed with Breast Cancer.

BACKGROUND: We evaluated smoking differences across nativity and race/ethnicity among women diagnosed with breast cancer. METHODS: In our Northern Californian pooled population of 5,653 [670 Asian, 690 Hispanic, and 4,300 non-Hispanic White (White)] women diagnosed with breast cancer, we evaluated smoking differences across nativity, race/ethnicity, and acculturation and effect modification of nativity by race/ethnicity and education. RESULTS: Foreign-born women currently smoked less than US-born women [odds ratio (OR) = 0.46, 95% confidence limit (CL): 0.29-0.72]. Hispanic (OR = 0.50; 95% CL: 0.32-0.78) women currently smoked less than White women. Among those who ever smoked (n = 2,557), foreign-born women smoked 5.23 fewer pack-years (PY) than US-born women (95% CL: -2.75 to -7.70). Furthermore, Asian (-4.60, 95% CL: -0.81 to -8.39) and Hispanic (-6.79, 95% CL: -4.14 to -9.43) women smoked fewer PY than White women. Associations were generally suggestive of greater smoking with greater acculturation (immigration age, US years, survey language). Finally, associations for nativity differed by education but not race/ethnicity, with a higher likelihood of smoking in US-born women only among those with less than a bachelor's degree (OR = 2.84, 95% CL: 2.15-3.77; current smoking: P = 0.01, PY: P = 0.05). CONCLUSIONS: Asian and Hispanic (vs. White) and foreign-born (vs. US-born) breast cancer survivors reported fewer smoking behaviors. Smoking differences across nativity and education were driven by higher rates of smoking in US-born women with lower educational attainment. IMPACT: Smoking behavioral patterns were similar among breast cancer survivors and the general population, informing potential smoking interventions.

Incidence and Medical Hospitalization Rates of Patients With Pediatric Eating Disorders.

Objective Eating disorders typically onset in preadolescence and adolescence and cause negative mental and physical health sequelae over the life span. This study examined the incidence and medical hospitalization rates of pediatric eating disorders in an integrated health system in the United States. Methods This retrospective cohort study examined 4883 Kaiser Permanente Northern California members 8-18 years of age with an eating disorder diagnosis from January 2015 to June 2019. Medical hospitalizations include admissions at any of the 13 Kaiser Permanente Northern California hospitals with a primary or secondary eating disorder diagnosis. Results Incidence rates ranged between 177 and 205 per 100,000 adolescents per year. More than half the adolescents were non-White: 10.8% Asian, 4.3% Black, 26.7% Hispanic/Latinx, 8.4% multiracial, 0.3% Native American/Alaskan Native, and 0.5% Native Hawaiian/Pacific Islander. Thirteen percent had a body mass index (BMI) below the 5th percentile, 61.8% had a BMI between the 5th and the 84th percentiles, 19.7% had a BMI above the 85th percentile, and 5.6% had an unknown BMI. During the 12-month follow-up period, 5.4% of adolescents had medical hospitalizations. Conclusions This study adds to the evidence that eating disorders affect children/adolescents across all weight/BMI ranges and racial/ethnic backgrounds. Future studies call for exploration on treatment strategies that tailor to the diverse populations.

Joint associations between neighborhood walkability, greenness, and particulate air pollution on cardiovascular mortality among adults with a history of stroke or acute myocardial infarction.

Fine particulate matter (PM2.5) is a known risk factor for cardiovascular disease (CVD). Neighborhood walkability and greenness may also be associated with CVD, but there is limited evidence on their joint or interacting effects with PM2.5. Methods: Cox proportional hazard models were used to estimate the risk of CVD mortality among adults with a history of acute myocardial infarction and/or stroke living in Northern California. We assessed the independent and joint effects of walkability, greenness (Normalized Differentiated Vegetation Index [NDVI]), and PM2.5 at residential addresses, controlling for age, sex, race/ethnicity, comorbidities, BMI, smoking, revascularization, medications, and socioeconomic status. Results: Greenness had a nonlinear association with CVD mortality (P = 0.038), with notably protective effects (HR = 0.87 [95% confidence interval {CI} = 0.78, 0.97]) at higher greenness levels (NDVI ≥ 0.3) and moderate attenuation after adjusting for PM2.5 (HR = 0.92 [95% CI = 0.82, 1.03]) per 0.1 increase in NDVI. Walkability had no independent effect on CVD mortality. PM2.5 had a strong independent effect in models adjusted for greenness and walkability (HR = 1.20 [95% CI = 1.08, 1.33)) per 10 µg/m3 increase in PM2.5. There was an interaction between walkability and PM2.5 (P = 0.037), where PM2.5 had slightly stronger associations in more walkable than less walkable neighborhoods (HR = 1.23 [95% CI = 1.06, 1.42] vs. 1.17 [95% CI = 1.04, 1.32]) per 10 µg/m3 increase in PM2.5. Greenness had no interaction with PM2.5 (P = 0.768) nor walkability (P = 0.385). Conclusions: High greenness may be protective of CVD mortality among adults with CVD history. PM2.5 associated CVD mortality risk varies slightly by level of neighborhood walkability, though these small differences may not be clinically meaningful.

Clinical Outcomes Following Regionalization of Gastric Cancer Care in a US Integrated Health Care System.

PURPOSE: In 2016, Kaiser Permanente Northern California regionalized gastric cancer care, introducing a regional comprehensive multidisciplinary care team, standardizing staging and chemotherapy, and implementing laparoscopic gastrectomy and D2 lymphadenectomy for patients eligible for curative-intent surgery. This study evaluated the effect of regionalization on outcomes. METHODS: The retrospective cohort study included gastric cancer cases diagnosed from January 2010 to May 2018. Information was obtained from the electronic medical record, cancer registry, state vital statistics, and chart review. Overall survival was compared in patients with all stages of disease, stage I-III disease, and curative-intent gastrectomy patients using annual inception cohorts. For the latter, the surgical approach and surgical outcomes were also compared. RESULTS: Among 1,429 eligible patients with gastric cancer with all stages of disease, one third were treated after regionalization, 650 had stage I-III disease, and 394 underwent curative-intent surgery. Among surgical patients, neoadjuvant chemotherapy utilization increased from 35% to 66% (P < .0001), laparoscopic gastrectomy increased from 18% to 92% (P < .0001), and D2 lymphadenectomy increased from 2% to 80% (P < .0001). Dissection of ≥ 15 lymph nodes increased from 61% to 95% (P < .0001). Surgical complication rates did not appear to increase after regionalization. Length of hospitalization decreased from 7 to 3 days (P < .001). Overall survival at 2 years was as follows: all stages, 32.8% pre and 37.3% post (P = .20); stage I-III cases with or without surgery, 55.6% and 61.1%, respectively (P = .25); and among surgery patients, 72.7% and 85.5%, respectively (P < .03). CONCLUSION: Regionalization of gastric cancer care within an integrated system allowed comprehensive multidisciplinary care, conversion to laparoscopic gastrectomy and D2 lymphadenectomy, increased overall survival among surgery patients, and no increase in surgical complications.

Regionalization of Acute Myeloid Leukemia Treatment in a Community-Based Population: Implementation and Early Results.

INTRODUCTION: Regionalization of care for acute myeloid leukemia (AML) has not been described for community-based settings. In 2015, we shifted AML induction from 21 local centers to 3 regional centers. METHODS: Using time-specific inception cohorts, we assessed whether regionalization was associated with the frequency of use of induction therapy, receipt of bone marrow transplantation, 60-day mortality (treatment toxicity), and 180-day mortality (treatment effectiveness). Information for all adult patients diagnosed with AML from 2013 to 2017 was obtained from the electronic health record. Multivariable methods were used to estimate the adjusted associations of induction, bone marrow transplantation, and death in relation to year of diagnosis before and after regionalization. RESULTS: Of 661 patients diagnosed during 2013 to 2017, 53% were ≥ 70 years, 22% were ≥ 80 years, and 10% died within the week following diagnosis. Comparing 2017 with 2013, the proportion of patients who received induction therapy increased 2.88 times (95% confidence interval [CI] = 1.55-5.35), and the proportion of non-acute promyelocytic leukemia patients receiving bone marrow transplantation increased 2.00 times (95% CI = 0.89-4.50). Regionalization was associated with lower 180-day mortality (hazard ratio [HR] = 0.64; 95% CI = 0.44-0.92), whereas change in 60-day mortality was not statistically significant (HR = 0.67; 95%CI = 0.43-1.04). CONCLUSION: In this community-based population, many patients were of advanced age yet benefitted from AML induction therapy delivered at a regionally specialized center. These early results suggest the benefit of regionalizing subspecialty leukemia care.

Development and Validation of Machine Learning Models: Electronic Health Record Data To Predict Visual Acuity After Cataract Surgery.

BACKGROUND: To develop predictive models of final corrected distance visual acuity (CDVA) following cataract surgery using machine learning algorithms and electronic health record data. METHODS: In this predictive modeling study we used decision tree, random forest, and gradient boosting. We included the first surgical eye of 64,768 members of Kaiser Permanente Northern California who underwent cataract surgery from June 1, 2010 through May 31, 2015. We measured discrimination and calibration of machine learning models for predicting postoperative CDVA 20/50 or worse vs 20/40 or better. RESULTS: The training set included 51,712 patients, and the validation set included 13,056 patients. We compared 3 machine learning models and found that the gradient boosting model provided the best discrimination ability for CDVA. The most important variables for predicting final CDVA 20/50 or worse were preoperative CDVA, age, and age-related macular degeneration, which together accounted for 41% of the gain in optimization of the gradient boosting model. Other important variables in the model included dispensed glaucoma medication, epiretinal membrane, cornea disorder, cataract surgery operating time, surgeon experience, and census block neighborhood characteristics (household income, family income, family poverty, college education, and home residence by owner). CONCLUSION: For predicting CDVA after cataract surgery, gradient boosting had the best ability to discriminate patients with postoperative CDVA 20/50 or worse from patients with postoperative CDVA 20/40 or better. Machine learning has the potential to improve prognosis and can improve patient information when making decisions to undergo cataract surgery.