An assessment of overdose mortality risk across the urban–rural continuum: Integrating satellite-derived and socioeconomic indicators

An assessment of overdose mortality risk across the urban–rural continuum: Integrating satellite-derived and socioeconomic indicators | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article An assessment of overdose mortality risk across the urban–rural continuum: Integrating satellite-derived and socioeconomic indicators Gia Barboza, Taylor Harrington This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7222499/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 5 You are reading this latest preprint version Abstract Geographic differences in drug overdose patterns across the rural-urban continuum are well documented. Few studies utilize remote sensing data to quantify the environmental and structural features that shape overdose risk at varying geographic scales. We conducted a retrospective ecological study to examine fatal opioid overdoses across census block groups in Cook County, Illinois, from 2020 to 2023. Urbanicity was determined based on the Global Human Settlement Layer Model (GHSL-MOD), with satellite data used to derive metrics of built-up intensity, vegetative greenness, and nighttime light emissions. Environmental indicators were combined with census-based measures of neighborhood deprivation to characterize spatial variation in physical and social conditions. A Bayesian spatiotemporal model estimated neighborhood-level overdose risk, accounting for spatial dependence, temporal trends, and environmental exposures. Overdose risk exhibited significant spatial clustering and strong associations with both social and environmental factors. Neighborhood disadvantage had a dose-response relationship, with fatal overdose risk in areas with the most deprivation, experiencing over seven times the risk, compared to the least deprived. Nighttime light intensity was strongly associated with increased overdose risk, while vegetative greenness and park access showed no significant protective effects. Increasing trends were detected in rural and transitional zones despite a higher risk in urban centers. Demographic characteristics of overdose victims varied across the county, suggesting potential geographic disparities in risk. The physical and social features of neighborhoods underscore the need for early surveillance and intervention within and outside urban centers. These factors should be incorporated into targeted, place-based strategies to lower opioid-related deaths. Figures Figure 1 Figure 2 Figure 3 Introduction Opioid-related overdose is a public health crisis, driven by synthetic opioids like fentanyl, causing record-high overdose rates. Between 1999 and 2020, opioid-related mortality increased by 158% in large metro counties and by 740% in rural areas, with rural Midwestern counties experiencing a staggering 1,600% increase. 1 By 2016, overdose mortality in these rural Midwestern areas was 16 times higher than in urban counterparts, 1 underscoring the crisis’s spread beyond urban cores and the growing importance of geographic context. 2 However, standard rural–urban classifications often obscure the overlapping social and environmental conditions that shape opioid-related risk in suburban, peri-urban, and rural communities. Research has shown that areas classified as rural often face unique risks, including earlier onset of misuse, higher prevalence, and barriers to treatment. 3 Yet structural factors like poverty and education have been shown to predict overdose risk to a larger extent than geography alone. For instance, Pear et al. found that prescription overdose rates were higher in socioeconomically vulnerable communities regardless of setting, while the impact of other risk factors, including low education or unemployment, varied across rural or urban contexts. 4 Similarly, a study in Georgia found that rural counties had higher overdose reversals but that rurality itself was not a consistent predictor once socioeconomic factors were considered. 5 In predominantly White, rural counties, risk is also shaped by isolation and limited healthcare access, 6 whereas urban areas are often affected by complex opioid syndemics linked to illicit markets, concentrated disadvantage, 7 and heightened law enforcement activity. 8 In these urban settings, vulnerable individuals like younger Black males or unsheltered individuals navigate public drug use in heavily surveilled environments, increasing the risk of arrest and overdose under conditions of racialized policing. 9 Two primary considerations limit existing literature examining opioid-related overdose across different geographical contexts. First, there is no consistent definition of what constitutes “urban” versus “rural.” Most studies rely on classifications from the Census Bureau, OMB, or USDA (e.g., RUCA codes), which focus on population size or commuting patterns. As Palombi et al. note, these definitions often oversimplify heterogeneous areas and fail to capture conditions in transitional zones like peri-urban or exurban communities. 10–12 Roth et al. similarly argue that using population-based definitions to measure rurality obscures meaningful variation in these areas that shape harm-related risk 5 Second, few studies investigate how these factors interact with built environments across diverse settings. Tempalski et al. call for a socio-built environment framework that integrates infrastructure (e.g., housing density, vacancy, accessibility) with social indicators like poverty and segregation. 13 However, when such frameworks are used, they are usually applied to urban settings, with limited use in rural or transitional areas. Rather than treating areas as binary — rural or urban —satellite indicators capture continuous variation across the landscape, including exurbs, peri-urban zones, and inner-ring suburbs, based on observed land use and development intensity. Measures such as built-up area and nighttime light emissions reflect infrastructure density and human activity, providing more precise insights into spatial gradients of risk than conventional classifications, like RUCA codes. Further, remote sensing data offers key proxies for measuring environmental exposure and land use, revealing built environmental features often overlooked in opioid-related mortality studies. Nighttime light exposure captures human activity, infrastructure, surveillance, and artificial illumination, linking to both social conditions and health. 14–17 . Whereas indicators like impervious surface and NDVI are associated with lower stress and overdose risk, dense built environments with limited vegetation often reflect stressful, resource-limited environments. 18–21 Combining satellite data with indices of social deprivation enables a richer understanding of how structural disadvantage and environmental factors jointly influence opioid overdose vulnerability. 13,22 This study uses satellite-derived measures of the built environment to examine fatal opioid overdoses in Cook County, Illinois. In 2022, the county reported nearly 2,000 opioid-related deaths, with fentanyl involved in over 90% of cases, 23 and an opioid fatality rate that far exceeds the national average 24,25 . Although commonly classified as urban, Cook County spans a diverse set of geographies, including dense urban cores, suburban and peri-urban areas, and rural-like settings. Our approach draws on Tempalski et al.’s socio-built environment framework 13 and Neely and Samura’s spatial theory of inequality, 26 which conceptualize space as relational, materially produced, and shaped by systems of structural power. Neely and Samura highlight how the built environment can simultaneously foster safety and care while also reinforcing exclusion and surveillance. Their work illustrates how spatial context in areas with different levels of urbanization plays a critical role in shaping overdose risk and access to care for people who use drugs. 26 Accordingly, we ask two central questions: (1) To what extent are satellite-derived features of the built environment associated with fatal opioid overdose risk, after adjusting for social deprivation and urbanicity?; and (2) How do patterns of deprivation and built environment intensity co-locate to form geographic clusters of high overdose mortality, and do these patterns persist over time? Methods Data This study employed an ecological approach to examine fatal opioid overdoses across all 2,347 census block groups (CBGs) in Cook County, Illinois, from 2020 to 2023. The dataset was organized as a four-year panel, with annual mortality counts linked to population estimates and covariate data for each CBG. Geographic boundaries for CBGs were obtained from the 2020 TIGER/Line shapefiles and transformed to WGS84 (EPSG:4326) using the tigris 27 package. Annual population estimates were retrieved from the American Community Survey 5-year data using the tidycensus package 28 and merged with CBG shapefiles via unique geographic identifiers. Dependent Variable. We downloaded fatal opioid overdose records from the Cook County Medical Examiner's Case Archive. We filtered the data to retain only those with valid geographic coordinates and death dates from January 1, 2020, to December 31, 2023. We converted locations of death into spatial features and spatially joined to CBG polygons using intersection methods. Annual death counts were then aggregated for each CBG. For block groups with a non-zero population but no observed deaths, zero values were imputed. Built Environment Different classifications of urbanization were derived from the 2020 Global Human Settlement Layer Settlement Model (GHSL-SMOD), which categorizes land based on settlement structure and density at a 1 km resolution. Two adjacent raster tiles covering Cook County were mosaicked, and the modal value (i.e., most frequent urban code) within each CBG was extracted using zonal statistics. We collapsed the GHSL-SMOD classes into four broader typologies for interpretation: Urban (dense urban cluster or urban center), Suburban (suburban or pre-urban), Transitional (semi-dense urban cluster), and Rural (low-density or very low-density rural zones). Water-classified areas were excluded due to their small number. Figure 1 shows the urban-rural continuum overlaid onto fatal opioid-related overdoses across the county. Built Environment Exposure (BEE) was assessed using a normalized raster derived from the GHSL-BUILT dataset, which quantifies the proportion of impervious surfaces (e.g., buildings, pavement) in each 10-meter grid cell. The median built-up value within each CBG was extracted using the exactextractr package 29 . The index ranges from 0 (entirely undeveloped) to 1 (fully developed), capturing the density of constructed surfaces and serving as a proxy for urbanization, infrastructure, and environmental stressors such as heat exposure and population density. Nighttime Light Intensity (NLI) was captured from the VIIRS Day/Night Band (DNB) product, which provides stable, cloud-free nighttime radiance measurements from satellite imagery. NLI captured via satellite using the VIIRS reflects both the concentration of human activity and the extent of artificial lighting. These data were rescaled to a 0–1 range and filtered to exclude transient light sources (e.g., fires, aurorae). The median normalized light intensity was calculated for each CBG using zonal extraction. Natural Environment Vegetative Greenness was measured by calculating the Normalized Difference Vegetation Index (NDVI) from Landsat 9 Surface Reflectance bands: the red (Band 4) and near-infrared (Band 5) bands were scaled and combined using the standard NDVI formula:​ $$\:NDVI=\frac{NIR-Red}{NIR+Red}$$ where NIR is near-infrared reflectance, and Red is visible red reflectance. The imagery was filtered to include only cloud-free scenes with less than 10% cloud cover from summer months (June through August). The raster was masked to remove water bodies and cropped to Cook County boundaries. Median NDVI values for each CBG were then extracted and standardized as z -scores. Higher NDVI values indicate more abundant and healthier vegetation cover, which has been linked to improved mental health and environmental resilience. Park Accessibility was measured using the 15-minute cumulative park access indicator from TransitCenter's Transit Equity Dashboard. Park accessibility reflects the total acreage of parks reachable within a 15-minute public transit trip from each CBG, accounting for both spatial proximity and transportation infrastructure. Higher values denote greater accessible green space within equitable travel time, making this a more meaningful exposure measure than static distance to the nearest park. Social Environment The ADI, developed by the University of Wisconsin’s Neighborhood Atlas, combines metrics such as income, education, employment, and housing conditions into a composite index. 30 ADI scores for 2020 were merged by GEOID using the sociome package. 31 Raw scores were recoded into quintile-based groupings, with Quintile 5 indicating the most deprived neighborhoods. Individual-level incidents were spatially linked to the CBG measures, allowing for comparisons of age, gender, and race/ethnicity across various neighborhood characteristics, including built environment, deprivation, and vegetation exposure. See Supplementary Tables 1 and 2 for the details information regarding data sources and variable construction. Statistical Analysis We computed expected counts for each CBG and year by multiplying their population by the countywide overdose mortality rate for the corresponding year. Standardized mortality ratios (SMRs) were then calculated as the ratio of observed to expected deaths. Expected counts were calculated as: $$\:{E}_{it}={P}_{it}\times\:\left(\frac{\sum\:_{i}{O}_{it}}{\sum\:_{i}{P}_{it}}\right)$$ Where: \(\:{E}_{it}\) = expected overdoses in CBG \(\:i\) at year \(\:t\) \(\:{P}_{it}\) = population of CBG \(\:i\) at year \(\:t\) \(\:{O}_{it}\) = observed overdose count in CBG \(\:i\) at year \(\:t\) Before spatial modeling, fixed-effect Poisson regression models were estimated to explore associations between overdose mortality and neighborhood-level exposures for each year. Covariates were standardized where applicable and entered as main effects. We included a mean-centered year variable defined as the year minus the base year (e.g., \(\:{\text{Year}}_{ct}=t-2020\) ), ADI quintiles, standardized NDVI, park access, recoded urban classification, built environment intensity, and nighttime light emissions. All models used a log link and incorporated the log of the population or expected count as an offset. To evaluate spatial dependence, a spatial contiguity matrix was constructed. Using the spdep package, 32 a queen contiguity neighbors list was converted into a binary spatial weights matrix where each CBG’s neighbors were assigned equal weights. Moran’s I was computed to assess global spatial autocorrelation in observed overdose rates, and this spatial structure was retained and incorporated into all subsequent Bayesian models. To account for spatial dependence and temporal trends, a Bayesian hierarchical spatiotemporal Poisson model with autoregressive temporal structure was implemented using the CARBayesST package. 33 An intrinsic conditional autoregressive (ICAR) prior was specified for the spatial component and temporal autocorrelation was modeled using a first-order autoregressive (AR(1)) process. Letting \(\:{Y}_{it}\) denote the observed number of overdose deaths in CBG i at time t , and \(\:{E}_{it},\) the expected count, the model assumed that: $$\:{Y}_{it}\sim\:\text{Poisson}\left({\mu\:}_{it}\right)\:\text{with}\:{\mu\:}_{it}={E}_{it}\cdot\:{\theta\:}_{it}$$ The log relative risk was modeled as: $$\:\text{l}\text{o}\text{g}\left({\theta\:}_{it}\right)=\alpha\:+{\beta\:}_{1}\cdot\:{\text{Year}}_{ct}+{\beta\:}_{2}\cdot\:{\text{ADI}}_{qi}+{\beta\:}_{3}\cdot\:{\text{NDVI}}_{i}+{\beta\:}_{4}\cdot\:{\text{ParkAcc}}_{i}+{\beta\:}_{5}\cdot\:{\text{Urbanicity}}_{i}+{\beta\:}_{6}\cdot\:{\text{BuiltEnv}}_{i}+{\beta\:}_{7}\cdot\:{\text{Lights}}_{i}+{\varphi\:}_{i}+{\delta\:}_{it}$$ Here, \(\:\alpha\:\) is the global intercept, \(\:{\beta\:}_{k}\) are fixed effects, \(\:{\varphi\:}_{i}\) is the spatially structured random effect capturing spatial dependence via a conditional autoregressive prior, and \(\:{\delta\:}_{it}\:\) is an unstructured spatiotemporal random effect. Fixed effects were given diffuse Gaussian priors: \(\:{\beta\:}_{k}\sim\:\mathcal{N}\left(0,1000\right)\) . Spatial random effects ( _i ) followed a CAR prior with a precision parameter: \(\:{\tau\:}_{\varphi\:}\sim\:\text{Gamma}\left(0.5,0.0005\right).\) Spatiotemporal noise followed an exchangeable normal prior with variance: \(\:{\tau\:}_{\delta\:}^{-1}\sim\:\text{Gamma}\left(0.5,0.0005\right).\) The temporal autocorrelation parameter was assigned a uniform prior on the interval [0, 1]: \(\:\rho\:\sim\:\text{Uniform}\left(0,1\right).\) Posterior estimation sampling was conducted using three parallel chains, each running for 20,000 iterations. This included a 5,000 iteration burn-in period and thinning every 10th sample, resulting in 4,500 posterior samples per chain. Model diagnostics included checking for multicollinearity using variance inflation factors (VIF). Model convergence was assessed using trace plots and posterior summaries. Relative risk estimates were extracted for each CBG-year combination and mapped using the ggplot2 package. 34 We descriptively examined the intersection of structural and environmental clustering by mapping the bivariate distribution using the biscale package. 35 ADI and nighttime light intensity were jointly classified using quantile binning (4x4 grid), and a custom legend was constructed using cowplot. 36 Model-derived relative risk (RR) defined high-risk areas as having a RR > 1.5 and visualized using both single-year and faceted maps. To examine trends in risk over time, both modeled mean relative risk and observed death counts were plotted by year. We used the model-smoothed fitted values from the Bayesian spatiotemporal model to classify CBGs into one of four trend types: sharply increasing, increasing, decreasing, or no change. These trend types were derived by fitting linear and quadratic regressions to the fitted overdose risk over time for each block group and evaluating both the direction and significance of the slope and curvature. Results Decedent Characteristics. Table 1 displays the characteristics of opioid-related overdose by victim's sex. Male decedents were slightly older than female decedents on average (47.2 vs. 45.7 years; p < 0.001), and racial/ethnic composition varied by gender ( p < 0.001), with males more likely to be Latino (16.7% vs. 8.8%) and females more likely to be non-Hispanic White (36.1% vs. 28.6%). Manner of death also differed significantly ( p < 0.001), with a greater proportion of suicides among female decedents (2.2% vs. 0.5%) and a slightly higher proportion of accidental deaths among males (98.8% vs. 96.7%). Environmental exposures varied modestly but significantly by gender. Females died in areas with slightly higher deprivation (mean ADI = 114.6 vs. 113.1; p = 0.006), lower built environment intensity and nighttime light exposure ( p < 0.001 for both), and marginally lower greenness as measured by NDVI ( p = 0.001). While expected overdose counts—adjusted for population and risk—were consistently lower among female decedents across all years (2020–2023; p = 0.003–0.019), observed overdose counts did not significantly differ by gender. This suggests that although modeled risk differed by gender-linked place characteristics, actual mortality patterns were similar. Figure 2 displays the spatial intersection of socioeconomic deprivation (ADI) and nighttime light intensity. High-deprivation, high-light areas were concentrated in Chicago’s South and West Sides—historically disinvested yet densely developed neighborhoods. We examined correlations between environmental and social predictors (see Supplementary Table 3). Built environment intensity was significantly and positively correlated with nighttime light ( r = 0.496) and negatively correlated with NDVI (r = − 0.454). Nighttime light and NDVI were strongly inversely related ( r = − 0.577). Park access correlated positively with NDVI ( r = 0.140) and negatively with built-up area ( r = − 0.280) and light ( r = − 0.101). ADI was modestly correlated with built environment intensity ( r = 0.162, p < .001) and nighttime light ( r = 0.218, p < .001), and negatively correlated with NDVI (r = − 0.128, p < .001) and park access (r = − 0.062, p < .001). Bayesian spatiotemporal Poisson models adjusted for spatial and temporal autocorrelation and included environmental and sociodemographic covariates. VIFs and GVIFs were all well below conventional thresholds (maximum \(\:{G}_{VIF}^{.5df}=1.20\) , indicating minimal multicollinearity and stable model estimation; Supplementary Table 4). The number of high-risk CBGs (relative risk [RR] > 1.5) increased from 164 in 2020 to 177 in 2021, 207 in 2022, then declined slightly to 192 in 2023, suggesting persistence and spatial spread of overdose risk during the COVID-19 pandemic. Table 2 shows the results from the Bayesian spatiotemporal model. Neighborhood deprivation was strongly and monotonically associated with fatal overdose risk. Compared to the least deprived quintile, CBGs in the second through fifth quintiles had IRRs of 1.98, 3.08, 4.20, and 7.56, respectively, with all estimates showing narrow credible intervals and strong evidence of dose-response. In contrast, neither NDVI (IRR = 0.99; 95% CrI: 0.94–1.04) nor park access (IRR = 0.99; 95% CrI: 0.94–1.04) were significantly associated with overdose risk in adjusted models. Urbanicity demonstrated mixed effects. CBGs classified as Transitional exhibited lower risk than Urban areas (IRR = 0.74; 95% CrI: 0.56–0.95), while Rural areas showed a non-significant increase (IRR = 1.05; 95% CrI: 0.74–1.61). Built environment intensity had a non-significant inverse association (IRR = 0.71; 95% CrI: 0.38–1.34). In contrast, nighttime light intensity was a strong positive predictor of risk: a one-unit increase in standardized light intensity was associated with a 8.24-fold increase in overdose mortality (95% CrI: 6.47–14.75). Temporally, overdose risk remained relatively stable across years (IRR = 0.99; 95% CrI: 0.97–1.02). The spatial variance (τ² = 10.64; 95% CrI: 8.33–13.72) and strong residual spatial autocorrelation (ρS = 2.54; 95% CrI: 2.45–2.62) indicated persistent geographic clustering of risk beyond measured covariates. Temporal autocorrelation was modest (ρT = 1.07; 95% CrI: 1.01–1.17). Most census block groups across all urban–rural categories showed no significant change in opioid-related overdose trends over time. Increasing trends were observed in suburban (0.6%), transitional (12.5%), and rural CBGs (5.4%). Decreasing trends were found in suburban (3.1%) and rural CBGs (3.6%), while no decreasing trends occurred in Transitional areas (Fig. 3 ). Discussion To our knowledge, this is the first study to integrate detailed urban typologies, nighttime light data from the VIIRS sensor, and satellite-derived measures of built environment intensity, alongside ADI-based measures of social disadvantage, to examine how overdose-related deaths vary across diverse socio-environmental contexts. Our findings align with a small but growing body of research emphasizing the importance of considering both structural disadvantage and physical environmental characteristics in understanding the opioid overdose crisis. 22 Similar to Williams et al. 22 , who analyzed 565 municipalities across New Jersey to capture meaningful variation in infrastructure, social conditions, and service environments, our study demonstrated that environmental conditions shape overdose risk across diverse neighborhood contexts. Rather than constructing a composite index, however, we focused on isolating the independent and categorical effects of specific environmental exposures and examined how overdose risk varies across discrete urban typologies. We found persistent geographic and temporal clustering of overdose risk, underscoring the importance of spatially explicit models for identifying trends and informing prevention efforts. The number of CBGs with overdose risk exceeding 150% of the county average increased, mainly in highly urbanized Chicago areas on South and West Sides, regions historically affected by disinvestment, racialized poverty, and organized abandonment. 37 Urban centers still bear the highest overdose burden, but transitional areas saw the most marked increases. These spatial trends carry important policy implications. In urban areas with rising rates, new interventions may be needed, while stable regions could strengthen current strategies. Future research would benefit by incorporating urban, suburban, peri-urban, and rural distinctions to better understand how environmental and social factors interact across the urban–rural continuum. Our findings support a growing body of research identifying NLI as a chronic environmental stressor associated with a wide range of adverse health outcomes 16,38 , including sleep disorders 39 , preterm birth 14 , gestational diabetes, 40 colorectal cancer, 41 and mood disorders. NLI has also been linked to depressive symptoms, anxiety, and suicidal ideation or attempt. 39,42 These effects are attributed to NLI’s disruption of sleep cycles and circadian rhythms, which reduces melatonin production and impairs the functioning of the immune and hormonal systems. Such circadian disruptions have, in turn, been associated with increased vulnerability to substance use and addiction. 43 Importantly, NLI also reflects the presence of other environmental stressors that contribute to poor health. For example, areas with high NLI exposure tend to have less green space, more air pollution, concentrated poverty, and greater area deprivation, 38 conditions that are independently linked to poor mental health 38 and increased risk for substance misuse. 44 These overlapping exposures suggest that NLI contributes to overdose vulnerability both through direct biological mechanisms and as an indicator of broader structural disadvantage. Regarding area deprivation, we found a dose-response relationship between opioid-related overdose and ADI consistent with past studies. 45 NLI contributed to geographic disparities in overdose deaths in urban areas of Chicago experiencing concentrated structural disadvantage. These findings align with prior research showing that NLI exposure is significantly higher in the most socially vulnerable neighborhoods, disproportionately affecting racially and ethnically minoritized communities. 42 We also note that the least deprived neighborhoods exhibited lower levels of NLI, despite their dense population and high economic activity. At a minimum, our results contribute to a growing body of evidence that identifies nighttime light exposure not only as a public health concern but also as a potential marker of environmental injustice. 42 We also observed that opioid-related fatality rates were significantly lower in transitional areas, which do not fit neatly into entirely urban or rural categories. However, these areas experienced the sharpest increases in overdose risk for the period we considered, which overlaps with the COVID-19 pandemic. This suggests that areas associated with lower risk may represent emerging hotspots with trends that increase over time. By contrast, suburban or peri-urban areas exhibited the most stable overdose trajectories, with over 96% of block groups showing no significant change across the study period. This relative stability may reflect a more substantial presence of protective factors in these areas 46 . Consistent with prior research, 47 we found no significant association between opioid-related mortality and measures of vegetative greenness (e.g., NDVI) or park access. This suggests that, at least at the spatial resolution and form captured in this study, the availability of green space may not confer a measurable protective effect against overdose. Additionally, our satellite-derived measure of Built Environment Intensity, which reflects infrastructure density and land-use patterns, was not significantly associated with opioid mortality after adjusting for other covariates. Although the examination of sex differences and environmental exposures was not the primary aim of this study, it yielded some key findings worth mentioning. Across all four years examined, the observed and expected overdose counts were consistently lower in the areas where female decedents overdosed. Yet, fatal overdoses among women were significantly more likely to occur in neighborhoods with higher social deprivation, lower built environment intensity, reduced nighttime light exposure, and diminished vegetative greenness compared to locations associated with male decedents. This pattern aligns with research on gender and substance use, which highlights the disproportionate stigma, trauma exposure, and structural barriers that women face in accessing care. 48 There is also evidence that females may be more biologically sensitive to circadian disruption and socially vulnerable to environmental neglect, both of which may increase overdose risk in underrecognized ways. While studies have generally not found significant sex interactions with NLI in outcomes such as preterm birth or gestational diabetes 14,15 , underlying physiological and contextual sex differences remain highly relevant. These findings suggest a dimension to overdose vulnerability among males and females that warrants future research. This study has several limitations that should be acknowledged. First, the analysis is limited to Cook County, Illinois, which includes the city of Chicago, an area characterized by distinct patterns of segregation, infrastructure, and access to services. As such, the findings may not be generalizable to regions with different geographic, demographic, or policy contexts. Second, although we modeled time-varying estimates of overdose risk, we did not incorporate time-varying covariates. Environmental exposures are shaped by both natural and human factors that vary across time and space. 49 Therefore, future research should investigate how temporal changes in social and ecological conditions may influence overdose patterns across levels of urbanization. Third, the study period was relatively short and coincided with the COVID-19 pandemic. Despite studies finding no association between social distancing and opioid-related overdose in Cook County, 45 the unique context may limit the detection of long-term or typical trends, as the pandemic introduced widespread disruptions in healthcare access, social services, and substance use patterns. 50–53 Finally, Cook County lacks a balanced representation of rural, suburban, and urban areas, restricting our ability to assess overdose risk across the full urban–rural continuum. Future studies should apply these methods in regions with greater spatial diversity to more fully examine how social and environmental conditions interact across varying geographic contexts to shape overdose vulnerability. Conclusion Proactive public health prevention should consider both aspects of the physical and social environment, as well as the demographic characteristics of the population residing there. Our results contribute to a more nuanced understanding of how social and built environmental conditions shape overdose vulnerability at fine geographic scales. The integration of remote sensing data provides new insight into how features of urban form and human activity influence health outcomes beyond what is captured by traditional socioeconomic indicators. References Monnat SM, Rigg KK. The opioid crisis in rural and small town America. Published online 2018. Accessed July 23, 2025. https://scholars.unh.edu/carsey/343/ Peters DJ, Monnat SM, Hochstetler AL, Berg MT. The Opioid Hydra: Understanding Overdose Mortality Epidemics and Syndemics Across the Rural-Urban Continuum. Rural Sociology . 2020;85(3):589-622. doi:10.1111/ruso.12307 Palombi L, Hawthorne AN, Irish A, Becher E, Bowen E. “One Out of Ten Ain’t Going to Make It”: An Analysis of Recovery Capital in the Rural Upper Midwest. Journal of Drug Issues . 2019;49(4):680-702. doi:10.1177/0022042619859309 Pear VA, Ponicki WR, Gaidus A, et al. Urban-rural variation in the socioeconomic determinants of opioid overdose. Drug and alcohol dependence . 2019;195:66-73. Roth KB, Goplerud DK, Babjak JS, Nguyen JL, Gilbert LR. Examining the association of rurality with opioid-related morbidity and mortality in Georgia: A geospatial analysis. Journal of Substance Use and Addiction Treatment . 2024;162:209336. Latimore A, Ortega Hinojose A, Kestner L, Patel. Exploring Urban-Rural Disparities in Accessing Treatment for Opioid Use Disorder. American Institutes for Research. November 19, 2021. Accessed July 10, 2025. http://www.air.org/resource/field/exploring-urban-rural-disparities-accessing-treatment-opioid-use-disorder Patino JE, Duque JC, Pardo-Pascual JE, Ruiz LA. Using remote sensing to assess the relationship between crime and the urban layout. Applied Geography . 2014;55:48-60. doi:10.1016/j.apgeog.2014.08.016 Wynveen CJ, Bixler RD, Hammitt WE. Law Enforcement Perceptions and Changes in the United States Park Service: Urban proximity and level of enforcement practices. Annals of Leisure Research . 2007;10(3/4):532-549. doi:10.1080/11745398.2007.9686780 Brunson RK, Miller J. YOUNG BLACK MEN AND URBAN POLICING IN THE UNITED STATES. British Journal of Criminology . 2006;46(4):613-640. doi:10.1093/bjc/azi093 Palombi LC, St Hill CA, Lipsky MS, Swanoski MT, Lutfiyya MN. A scoping review of opioid misuse in the rural United States. Annals of Epidemiology . 2018;28(9):641-652. doi:10.1016/j.annepidem.2018.05.008 Stuck in the middle with you: Peri-urban areas and the food system. Center for a Livable Future. Accessed July 25, 2025. https://clf.jhsph.edu/stories/stuck-middle-you-peri-urban-areas-and-food-system Baumgartner J, Rodriguez J, Berkhout F, et al. Synthesizing the links between secure housing tenure and health for more equitable cities. Wellcome open research . 2023;7:18. Tempalski B, Williams LD, Kolak M, Ompad DC, Koschinsky J, McLafferty SL. Conceptualizing the Socio-Built Environment: An Expanded Theoretical Framework to Promote a Better Understanding of Risk for Nonmedical Opioid Overdose Outcomes in Urban and Non-Urban Settings. J Urban Health . 2022;99(4):701-716. doi:10.1007/s11524-022-00645-3 Sun Q, Yang Y, Liu J, et al. Association between exposure to outdoor artificial light at night and the risk of preterm birth. Frontiers in Public Health . 2023;11:1280790. Zou HX, Hu LW, Zhang Z, et al. Outdoor light at night exposure was associated with hypothyroidism in pregnant women: A national study in China. Science of The Total Environment . 2025;958:178017. Badpa M, Schneider A, Ziegler AG, et al. Outdoor Light at Night and Children’s Body Mass: A Cross-Sectional Analysis in the Fr1da Study. Environmental Research . 2023;232:116325. doi:10.1016/j.envres.2023.116325 Filippini T, Costanzini S, Chiari A, et al. Light at night exposure and risk of dementia conversion from mild cognitive impairment in a Northern Italy population. Int J Health Geogr . 2024;23(1). doi:10.1186/s12942-024-00384-5 Hartley K, Perazzo J, Brokamp C, et al. Residential surrounding greenness and self-reported symptoms of anxiety and depression in adolescents. Environmental research . 2021;194:110628. Lee HY, Chang HT, Herianto S, et al. Do greenness and landscape indices for greenspace correlate with suicide ratio? Landscape and Urban Planning . 2024;242:104935. doi:10.1016/j.landurbplan.2023.104935 Zhang C, Wang C, Chen C, et al. Effects of tree canopy on psychological distress: A repeated cross-sectional study before and during the COVID-19 epidemic. Environmental Research . 2022;203:111795. doi:10.1016/j.envres.2021.111795 Becker DA, Browning MHEM, McAnirlin O, Yuan S, Helbich M. Is green space associated with opioid-related mortality? An ecological study at the U.S. county level. Urban Forestry & Urban Greening . 2022;70:127529. doi:10.1016/j.ufug.2022.127529 Williams LD, Kolak M, Villanueva C, Ompad DC, Tempalski B. Creation and Validation of a New Socio-built Environment Index Measure of Opioid Overdose Risk for Use in Both Non-urban and Urban Settings. J Urban Health . 2023;100(5):1048-1061. doi:10.1007/s11524-023-00754-7 Cook County Medical Examiners Office. Medical Examiner Case Archive. Cook County Government Open Data. April 25, 2022. https://datacatalog.cookcountyil.gov/Public-Safety/Medical-Examiner-Case-Archive/cjeq-bs86 Galea S. Income distribution and risk of fatal drug overdose in New York City neighborhoods. Drug and Alcohol Dependence . 2003;70(2):139-148. doi:10.1016/S0376-8716(02)00342-3 Galea S, Rudenstine S, Vlahov D. Drug use, misuse, and the urban environment. Drug and Alcohol Review . 2005;24(2):127-136. doi:10.1080/09595230500102509 Neely B, Samura M. Social geographies of race: connecting race and space. Ethnic and Racial Studies . 2011;34(11):1933-1952. doi:10.1080/01419870.2011.559262 Walker K. Tigris: Load Census TIGER/Line Shapefiles .; 2023. https://CRAN.R-project.org/package=tigris Walker K, Herman M. Tidycensus: Load US Census Boundary and Attribute Data as “tidyverse” and ’Sf’-Ready Data Frames .; 2022. https://CRAN.R-project.org/package=tidycensus Daniel Baston. Exactextractr: Fast Extraction from Raster Datasets Using Polygons .; 2023. https://CRAN.R-project.org/package=exactextractr Singh GK. Area Deprivation and Widening Inequalities in US Mortality, 1969–1998. Am J Public Health . 2003;93(7):1137-1143. Krieger N, Dalton J, Wang C, Perzynski A. Sociome: Operationalizing Social Determinants of Health Data for Researchers .; 2023. https://CRAN.R-project.org/package=sociome Pebesma E, Bivand RS. Spatial Data Science with Applications in R . Chapman & Hall; 2023. https://r-spatial.org/book/ Lee D, Rushworth A, Napier G. Spatio-Temporal Areal Unit Modeling in R with Conditional Autoregressive Priors Using the CARBayesST Package. Journal of Statistical Software . 2018;84:1-39. doi:10.18637/jss.v084.i09 H Wickham, Chang W, Pedersen T, Takahashi K, Wilke C, Woo K. ggplot2: Create Elegant Data Visualisations Using the Grammar of Graphics. Published online 2018. https://CRAN.R-project.org/package=ggplot2 Prener C, Grossenbacher T, Zehr A. Biscale: Tools and Palettes for Bivariate Thematic Mapping .; 2022. https://CRAN.R-project.org/package=biscale Wilke CO. Cowplot: Streamlined Plot Theme and Plot Annotations for “Ggplot2.” ; 2023. https://CRAN.R-project.org/package=cowplot Barboza-Salerno G, Liebhard B, Duhaney S, Harrington T. Redlining, Reinvestment, and Racial Segregation: A Bayesian Spatial Analysis of Mortgage Lending T. Liao YA. Mental Health Problems in Urban Neighbourhoods: The Role of Physical and Social Environment . PhD Thesis. Ludwig-Maximilians-Universität München München; 2022. Accessed July 24, 2025. https://pure.mpg.de/pubman/faces/ViewItemOverviewPage.jsp?itemId=item_3608538 Hu K, Li W, Zhang Y, et al. Association between outdoor artificial light at night and sleep duration among older adults in China: a cross-sectional study. Environmental research . 2022;212:113343. Sun Q, Ye F, Liu J, et al. Outdoor artificial light at night exposure and gestational diabetes mellitus: a case–control study. Frontiers in Public Health . 2024;12:1396198. Harding BN, Palomar-Cros A, Valentín A, et al. Comparing Data from Three Satellites on Artificial Light at Night (ALAN): Focusing on Blue Light’s Influence on Colorectal Cancer in a Case–Control Study in Spain. Environ Health Perspect . 2024;132(5). doi:10.1289/ehp14414 Xiao Q, Lyu Y, Zhou M, et al. Artificial light at night and social vulnerability: An environmental justice analysis in the U.S. 2012–2019. Environment International . 2023;178:108096. doi:10.1016/j.envint.2023.108096 Hasler BP, Smith LJ, Cousins JC, Bootzin RR. Circadian rhythms, sleep, and substance abuse. Sleep medicine reviews . 2012;16(1):67-81. Van Draanen J, Tsang C, Mitra S, et al. Mental disorder and opioid overdose: a systematic review. Soc Psychiatry Psychiatr Epidemiol . 2022;57(4):647-671. doi:10.1007/s00127-021-02199-2 Barboza G, Angulski K, Hines L, Brown P. Variability in Opioid-Related Drug Overdoses, Social Distancing, and Area-Level Deprivation during the COVID-19 Pandemic: a Bayesian Spatiotemporal Analysis. Journal of Urban Health . 2022;99(5):873-886. Srinivasan S, Pustz J, Marsh E, Young LD, Stopka TJ. Risk factors for persistent fatal opioid-involved overdose clusters in Massachusetts 2011–2021: a spatial statistical analysis with socio-economic, accessibility, and prescription factors. BMC Public Health . 2024;24(1). doi:10.1186/s12889-024-19399-5 Becker DA, Browning MH, McAnirlin O, Yuan S, Helbich M. Is green space associated with opioid-related mortality? An ecological study at the US county level. Urban Forestry & Urban Greening . 2022;70:127529. Lamonica AK, Boeri M, Turner J. Circumstances of overdose among suburban women who use opioids: extending an urban analysis informed by drug, set, and setting. International Journal of Drug Policy . 2021;90:103082. Wang H, Lin C, Ou S, et al. Multilevel Change of Urban Green Space and Spatiotemporal Heterogeneity Analysis of Driving Factors. Sustainability . 2024;16(11):4762. doi:10.3390/su16114762 Bartholomew TS, Nakamura N, Metsch LR, Tookes HE. Syringe services program (SSP) operational changes during the COVID-19 global outbreak. International Journal of Drug Policy . 2020;83:102821. doi:10.1016/j.drugpo.2020.102821 Ghose R, Forati AM, Mantsch JR. Impact of the COVID-19 Pandemic on Opioid Overdose Deaths: a Spatiotemporal Analysis. J Urban Health . 2022;99(2):316-327. doi:10.1007/s11524-022-00610-0 Khatri UG, Pizzicato LN, Viner K, et al. Racial/ethnic disparities in unintentional fatal and nonfatal emergency medical services–attended opioid overdoses during the COVID-19 pandemic in Philadelphia. JAMA network open . 2021;4(1):e2034878-e2034878. Barboza GE, Schiamberg LB, Pachl L. A spatiotemporal analysis of the impact of COVID-19 on child abuse and neglect in the city of Los Angeles, California. Child abuse & neglect . 2021;116:104740. Tables Table 1. Characteristics of individuals who died from opioid-related overdose, stratified by sex Variable Female (N = 1628, 22.7%) Male (N = 5530, 77.3%) Total (N = 7158) p -value Age (Mean, SD) 45.7 (13.9) 47.2 (13.7) 46.8 (13.8) <0.001 Race/Ethnicity (%) Latino 8.8 16.7 14.9 <0.001 Non-Hispanic Asian 0.7 0.7 0.7 Non-Hispanic Black 53.7% 53.5 53.5 Non-Hispanic Indigenous 0.1% 0.1 0.1 Non-Hispanic White 36.1% 28.6 30.3 Other 0.2% 0.3 0.3 Unknown 0.1% 0.1 0.1 Manner of Death (%) Accident 96.7 98.8 98.3 <0.001 Homicide 0.2 0.1 0.1 Natural 0.4 0.3 0.3 Suicide 2.2 0.5 0.9 Undetermined 0.6 0.4 0.4 Year of Death (%) 0.189 2020 24.5 24.1 24.2 2021 27.1 25.0 25.4 2022 25.8 26.1 26.1 2023 22.6 24.8 24.3 Environmental Indicators (Mean, SD) ADI 114.6 (19.9) 113.1 (19.6) 113.4 (19.7) 0.006 Built Environment 0.3 (0.1) 0.3 (0.1) 0.3 (0.1) <0.001 Nighttime Light 0.2 (0.1) 0.2 (0.1) 0.2 (0.1) <0.001 NDVI 0.1 (0.1) 0.1 (0.1) 0.1 (0.1) 0.001 Park Access (acres) 59.8 (159.4) 59.8 (254.8) 59.8 (236.5) 0.989 Expected Overdose Counts (Mean, SD) 2020 0.4 (0.2) 0.5 (0.3) 0.5 (0.3) 0.003 2021 0.5 (0.2) 0.5 (0.3) 0.5 (0.3) 0.006 2022 0.5 (0.2) 0.5 (0.3) 0.5 (0.3) 0.006 2023 0.4 (0.2) 0.5 (0.3) 0.5 (0.3) 0.019 Observed Overdose Counts (Mean, SD) 2020 1.6 (2.0) 1.6 (2.0) 1.6 (2.0) 0.391 2021 1.8 (2.2) 1.8 (2.3) 1.8 (2.3) 0.743 2022 1.7 (2.1) 1.8 (2.1) 1.8 (2.1) 0.350 2023 1.6 (1.9) 1.6 (1.9) 1.6 (1.9) 0.504 Notes. Means and standard deviations are reported for continuous variables; percentages are reported for categorical variables. P -values reflect results from two-sample t-tests (for continuous variables) or chi-square tests (for categorical variables) comparing female and male decedents. ADI = Area Deprivation Index; NDVI = Normalized Difference Vegetation Index. Environmental indicators reflect the census block group where the decedent’s death occurred. Expected overdose counts were derived using countywide rates adjusted for population size. Table 2 . Bayesian Spatiotemporal Poisson Model Predictor IRR (95% CrI) Percent Change (%) Intercept 0.12 (0.10, 0.15) -87.7 Year (centered) 0.99 (0.97, 1.02) -0.85 Area Deprivation Index (ADI) (ref = most deprived) Quintile 2 1.98 (1.72, 2.28) +98.0 Quintile 3 3.08 (2.66, 3.54) +208.0 Quintile 4 4.20 (3.63, 4.83) +320.3 Quintile 5 7.56 (6.57, 8.68) +656.0 NDVI (Vegetative Greenness) 0.99 (0.94, 1.04) -0.81 Park Access (Acres within CBG) 0.99 (0.94, 1.04) -0.80 Urbanization Category (ref = Urban) Suburban 0.75 (0.58, 0.95) –25.1% Rural 1.15 (0.80, 1.64) +14.9% Transitional 0.69 (0.26, 1.72) –30.3% Built Environment Intensity 0.71 (0.38, 1.34) -28.6 Nighttime Light Intensity 8.24 (6.47, 14.75) +724.1 Spatial Variance (τ²) 10.64 (8.33, 13.72) -- Spatial Autocorrelation (ρS) 2.54 (2.45, 2.62) -- Temporal Autocorrelation (ρT) 1.07 (1.01, 1.17) -- Notes. Incidence Rate Ratios (IRRs) and 95% Credible Intervals (CrIs) are reported for fixed effects from a Bayesian hierarchical Poisson model with spatial and temporal structure. IRR = exp(Beta); Percent change is calculated as (IRR − 1) × 100. Supplementary Files SupplementTables7.26.25.docx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revise and resubmit 04 Feb, 2026 Reviewers agreed at journal 19 Sep, 2025 Reviewers invited by journal 25 Aug, 2025 Editor assigned by journal 28 Jul, 2025 First submitted to journal 28 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7222499","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":492186735,"identity":"b9a3b7e4-db68-4d9f-8445-08d3ed4b5354","order_by":0,"name":"Gia Barboza","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABAklEQVRIiWNgGAWjYBACAwYGNiiTuQFMsbFDaMYGwlqgavh4DpCqRU4iAb8Wc/azzx58+GPHIN9+sPFx5Q4bBjbJx4c/8zDYyG44gF2LZU+6ueEMnmQGgzOJzYZnz6QxsEmnpUnzMKQZ49JicCCNTZpHgrl+A0Nim2Rj22GglhwzZh6Gw4k4tZx/BtRiUM8g3/+w/SdYi+QZY6DD/uPWcgNkS8JhBoYbiW2MYC0SPAZAhx3AqcVyxjN2wxkHjgP1PmwGOiyNh40nLU1yjkGy8UwcWsz509iAIVYNdFjywY+NbTZy8u2HD394U2En24dDCwbggTqYSOWjYBSMglEwCrACANlRVk4LsZ8WAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-4953-4148","institution":"The Ohio State University College of Public Health","correspondingAuthor":true,"prefix":"","firstName":"Gia","middleName":"","lastName":"Barboza","suffix":""},{"id":492186736,"identity":"6487c419-4888-4f0c-8dac-7fcf2934fb21","order_by":1,"name":"Taylor Harrington","email":"","orcid":"","institution":"The Ohio State University","correspondingAuthor":false,"prefix":"","firstName":"Taylor","middleName":"","lastName":"Harrington","suffix":""}],"badges":[],"createdAt":"2025-07-26 17:04:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7222499/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7222499/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":87893302,"identity":"24bafd11-9060-4635-bf87-19bba96e11dc","added_by":"auto","created_at":"2025-07-30 07:07:36","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":445933,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eUrban–Rural Continuum and Opioid-Related Overdose Deaths in Cook County, IL (2020–2023). \u003c/strong\u003eSpatial distribution of opioid-related overdose deaths (black dots) across Cook County census block groups, overlaid with classifications from the Global Human Settlement Layer – Settlement Model (GHSL-SMOD). The urban–rural continuum comprises seven typologies based on built density and settlement structure: Dense Urban Cluster, Semi-Dense Urban Cluster, Urban Centre, Suburban or Peri-Urban, Rural Cluster, Low-Density Rural, and Very Low-Density Rural. Municipal community boundaries (labeled) are shown for reference.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7222499/v1/930109bedc0b11c6d447a92f.png"},{"id":87893300,"identity":"ef59f3d4-467a-47bf-9850-473f6c966fe3","added_by":"auto","created_at":"2025-07-30 07:07:36","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":140649,"visible":true,"origin":"","legend":"\u003cp\u003eBivariate choropleth map of Area Deprivation Index (ADI) and Nighttime Light Intensity for Census Block Groups in Cook County, Illinois. Notes. Bivariate map based on the joint distribution of area level deprivation (ADI) and light intensity using quantile-based breaks. Areas are classified as: 1) high deprivation and low light intensity (orange); 2) high deprivation but high light intensity (green); 3) low deprivation and high light intensity (blue), and 4) low deprivation and low light intensity (grey).\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7222499/v1/03a2ecfb596388a76f215766.png"},{"id":87893301,"identity":"615f522a-ebe4-435b-bb68-1689a5e450bf","added_by":"auto","created_at":"2025-07-30 07:07:36","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":50873,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eModel-based trend analysis within urban classification. \u003c/strong\u003eThe plot shows the distribution of modeled overdose risk trends across different types of urban environments. The \u003cem\u003ex\u003c/em\u003e-axis of the plot represents four categories of urbanicity. The \u003cem\u003ey\u003c/em\u003e-axis shows the proportion of block groups within each category that fall into each of the four trend classifications. Bars are stacked and color-coded by trend type to allow for a direct comparison of the composition of overdose risk trajectories across different geographic contexts.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7222499/v1/7e9d0fbc5eeec38e14eb002c.png"},{"id":87894246,"identity":"724358c3-0d9c-4ef5-8a44-650cc14d824a","added_by":"auto","created_at":"2025-07-30 07:15:42","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1341295,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7222499/v1/57a42340-ba47-49d8-9a29-7104d600c15d.pdf"},{"id":87894220,"identity":"ed1ee364-71fa-4e57-a9c6-5ab95dfbfee4","added_by":"auto","created_at":"2025-07-30 07:15:36","extension":"docx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":26956,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTables7.26.25.docx","url":"https://assets-eu.researchsquare.com/files/rs-7222499/v1/d758f15dfaff3757d4470cd6.docx"}],"financialInterests":"","formattedTitle":"An assessment of overdose mortality risk across the urban–rural continuum: Integrating satellite-derived and socioeconomic indicators","fulltext":[{"header":"Introduction","content":"\u003cp\u003eOpioid-related overdose is a public health crisis, driven by synthetic opioids like fentanyl, causing record-high overdose rates. Between 1999 and 2020, opioid-related mortality increased by 158% in large metro counties and by 740% in rural areas, with rural Midwestern counties experiencing a staggering 1,600% increase.\u003csup\u003e1\u003c/sup\u003e By 2016, overdose mortality in these rural Midwestern areas was 16 times higher than in urban counterparts,\u003csup\u003e1\u003c/sup\u003e underscoring the crisis’s spread beyond urban cores and the growing importance of geographic context.\u003csup\u003e2\u003c/sup\u003e However, standard rural–urban classifications often obscure the overlapping social and environmental conditions that shape opioid-related risk in suburban, peri-urban, and rural communities.\u003c/p\u003e\u003cp\u003eResearch has shown that areas classified as rural often face unique risks, including earlier onset of misuse, higher prevalence, and barriers to treatment.\u003csup\u003e3\u003c/sup\u003e Yet structural factors like poverty and education have been shown to predict overdose risk to a larger extent than geography alone. For instance, Pear et al. found that prescription overdose rates were higher in socioeconomically vulnerable communities regardless of setting, while the impact of other risk factors, including low education or unemployment, varied across rural or urban contexts.\u003csup\u003e4\u003c/sup\u003e Similarly, a study in Georgia found that rural counties had higher overdose reversals but that rurality itself was not a consistent predictor once socioeconomic factors were considered.\u003csup\u003e5\u003c/sup\u003e In predominantly White, rural counties, risk is also shaped by isolation and limited healthcare access,\u003csup\u003e6\u003c/sup\u003e whereas urban areas are often affected by complex opioid syndemics linked to illicit markets, concentrated disadvantage,\u003csup\u003e7\u003c/sup\u003e and heightened law enforcement activity.\u003csup\u003e8\u003c/sup\u003e In these urban settings, vulnerable individuals like younger Black males or unsheltered individuals navigate public drug use in heavily surveilled environments, increasing the risk of arrest and overdose under conditions of racialized policing.\u003csup\u003e9\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eTwo primary considerations limit existing literature examining opioid-related overdose across different geographical contexts. First, there is no consistent definition of what constitutes “urban” versus “rural.” Most studies rely on classifications from the Census Bureau, OMB, or USDA (e.g., RUCA codes), which focus on population size or commuting patterns. As Palombi et al. note, these definitions often oversimplify heterogeneous areas and fail to capture conditions in transitional zones like peri-urban or exurban communities.\u003csup\u003e10–12\u003c/sup\u003e Roth et al. similarly argue that using population-based definitions to measure rurality obscures meaningful variation in these areas that shape harm-related risk\u003csup\u003e5\u003c/sup\u003e Second, few studies investigate how these factors interact with built environments across diverse settings. Tempalski et al. call for a socio-built environment framework that integrates infrastructure (e.g., housing density, vacancy, accessibility) with social indicators like poverty and segregation.\u003csup\u003e13\u003c/sup\u003e However, when such frameworks are used, they are usually applied to urban settings, with limited use in rural or transitional areas.\u003c/p\u003e\u003cp\u003eRather than treating areas as binary — rural or urban —satellite indicators capture continuous variation across the landscape, including exurbs, peri-urban zones, and inner-ring suburbs, based on observed land use and development intensity. Measures such as built-up area and nighttime light emissions reflect infrastructure density and human activity, providing more precise insights into spatial gradients of risk than conventional classifications, like RUCA codes. Further, remote sensing data offers key proxies for measuring environmental exposure and land use, revealing built environmental features often overlooked in opioid-related mortality studies. Nighttime light exposure captures human activity, infrastructure, surveillance, and artificial illumination, linking to both social conditions and health.\u003csup\u003e14–17\u003c/sup\u003e. Whereas indicators like impervious surface and NDVI are associated with lower stress and overdose risk, dense built environments with limited vegetation often reflect stressful, resource-limited environments.\u003csup\u003e18–21\u003c/sup\u003e Combining satellite data with indices of social deprivation enables a richer understanding of how structural disadvantage and environmental factors jointly influence opioid overdose vulnerability.\u003csup\u003e13,22\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eThis study uses satellite-derived measures of the built environment to examine fatal opioid overdoses in Cook County, Illinois. In 2022, the county reported nearly 2,000 opioid-related deaths, with fentanyl involved in over 90% of cases,\u003csup\u003e23\u003c/sup\u003e and an opioid fatality rate that far exceeds the national average \u003csup\u003e24,25\u003c/sup\u003e. Although commonly classified as urban, Cook County spans a diverse set of geographies, including dense urban cores, suburban and peri-urban areas, and rural-like settings. Our approach draws on Tempalski et al.’s socio-built environment framework\u003csup\u003e13\u003c/sup\u003e and Neely and Samura’s spatial theory of inequality,\u003csup\u003e26\u003c/sup\u003e which conceptualize space as relational, materially produced, and shaped by systems of structural power. Neely and Samura highlight how the built environment can simultaneously foster safety and care while also reinforcing exclusion and surveillance. Their work illustrates how spatial context in areas with different levels of urbanization plays a critical role in shaping overdose risk and access to care for people who use drugs.\u003csup\u003e26\u003c/sup\u003e Accordingly, we ask two central questions: (1) To what extent are satellite-derived features of the built environment associated with fatal opioid overdose risk, after adjusting for social deprivation and urbanicity?; and (2) How do patterns of deprivation and built environment intensity co-locate to form geographic clusters of high overdose mortality, and do these patterns persist over time?\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cb\u003eData\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThis study employed an ecological approach to examine fatal opioid overdoses across all 2,347 census block groups (CBGs) in Cook County, Illinois, from 2020 to 2023. The dataset was organized as a four-year panel, with annual mortality counts linked to population estimates and covariate data for each CBG. Geographic boundaries for CBGs were obtained from the 2020 TIGER/Line shapefiles and transformed to WGS84 (EPSG:4326) using the \u003cem\u003etigris\u003c/em\u003e \u003csup\u003e27\u003c/sup\u003e package. Annual population estimates were retrieved from the American Community Survey 5-year data using the \u003cem\u003etidycensus\u003c/em\u003e package \u003csup\u003e28\u003c/sup\u003e and merged with CBG shapefiles via unique geographic identifiers.\u003c/p\u003e\u003cp\u003e\u003cb\u003eDependent Variable.\u003c/b\u003e We downloaded fatal opioid overdose records from the Cook County Medical Examiner's Case Archive. We filtered the data to retain only those with valid geographic coordinates and death dates from January 1, 2020, to December 31, 2023. We converted locations of death into spatial features and spatially joined to CBG polygons using intersection methods. Annual death counts were then aggregated for each CBG. For block groups with a non-zero population but no observed deaths, zero values were imputed.\u003c/p\u003e\u003cp\u003e\u003cb\u003eBuilt Environment\u003c/b\u003e\u003c/p\u003e\u003cp\u003eDifferent classifications of urbanization were derived from the 2020 Global Human Settlement Layer Settlement Model (GHSL-SMOD), which categorizes land based on settlement structure and density at a 1 km resolution. Two adjacent raster tiles covering Cook County were mosaicked, and the modal value (i.e., most frequent urban code) within each CBG was extracted using zonal statistics. We collapsed the GHSL-SMOD classes into four broader typologies for interpretation: Urban (dense urban cluster or urban center), Suburban (suburban or pre-urban), Transitional (semi-dense urban cluster), and Rural (low-density or very low-density rural zones). Water-classified areas were excluded due to their small number. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e shows the urban-rural continuum overlaid onto fatal opioid-related overdoses across the county.\u003c/p\u003e\u003cp\u003eBuilt Environment Exposure (BEE) was assessed using a normalized raster derived from the GHSL-BUILT dataset, which quantifies the proportion of impervious surfaces (e.g., buildings, pavement) in each 10-meter grid cell. The median built-up value within each CBG was extracted using the \u003cem\u003eexactextractr\u003c/em\u003e package \u003csup\u003e29\u003c/sup\u003e. The index ranges from 0 (entirely undeveloped) to 1 (fully developed), capturing the density of constructed surfaces and serving as a proxy for urbanization, infrastructure, and environmental stressors such as heat exposure and population density.\u003c/p\u003e\u003cp\u003eNighttime Light Intensity (NLI) was captured from the VIIRS Day/Night Band (DNB) product, which provides stable, cloud-free nighttime radiance measurements from satellite imagery. NLI captured via satellite using the VIIRS reflects both the concentration of human activity and the extent of artificial lighting. These data were rescaled to a 0–1 range and filtered to exclude transient light sources (e.g., fires, aurorae). The median normalized light intensity was calculated for each CBG using zonal extraction.\u003c/p\u003e\u003cp\u003e\u003cb\u003eNatural Environment\u003c/b\u003e\u003c/p\u003e\u003cp\u003eVegetative Greenness was measured by calculating the Normalized Difference Vegetation Index (NDVI) from Landsat 9 Surface Reflectance bands: the red (Band 4) and near-infrared (Band 5) bands were scaled and combined using the standard NDVI formula:​\u003c/p\u003e\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:NDVI=\\frac{NIR-Red}{NIR+Red}$$\u003c/div\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003ewhere NIR is near-infrared reflectance, and Red is visible red reflectance. The imagery was filtered to include only cloud-free scenes with less than 10% cloud cover from summer months (June through August). The raster was masked to remove water bodies and cropped to Cook County boundaries. Median NDVI values for each CBG were then extracted and standardized as \u003cem\u003ez\u003c/em\u003e-scores. Higher NDVI values indicate more abundant and healthier vegetation cover, which has been linked to improved mental health and environmental resilience.\u003c/p\u003e\u003cp\u003ePark Accessibility was measured using the 15-minute cumulative park access indicator from TransitCenter's Transit Equity Dashboard. Park accessibility reflects the total acreage of parks reachable within a 15-minute public transit trip from each CBG, accounting for both spatial proximity and transportation infrastructure. Higher values denote greater accessible green space within equitable travel time, making this a more meaningful exposure measure than static distance to the nearest park.\u003c/p\u003e\u003cp\u003e\u003cb\u003eSocial Environment\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe ADI, developed by the University of Wisconsin’s Neighborhood Atlas, combines metrics such as income, education, employment, and housing conditions into a composite index.\u003csup\u003e30\u003c/sup\u003e ADI scores for 2020 were merged by GEOID using the \u003cem\u003esociome\u003c/em\u003e package.\u003csup\u003e31\u003c/sup\u003e Raw scores were recoded into quintile-based groupings, with Quintile 5 indicating the most deprived neighborhoods. Individual-level incidents were spatially linked to the CBG measures, allowing for comparisons of age, gender, and race/ethnicity across various neighborhood characteristics, including built environment, deprivation, and vegetation exposure. See Supplementary Tables\u0026nbsp;1 and 2 for the details information regarding data sources and variable construction.\u003c/p\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003eWe computed expected counts for each CBG and year by multiplying their population by the countywide overdose mortality rate for the corresponding year. Standardized mortality ratios (SMRs) were then calculated as the ratio of observed to expected deaths. Expected counts were calculated as:\u003c/p\u003e\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:{E}_{it}={P}_{it}\\times\\:\\left(\\frac{\\sum\\:_{i}{O}_{it}}{\\sum\\:_{i}{P}_{it}}\\right)$$\u003c/div\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eWhere:\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{it}\\)\u003c/span\u003e\u003c/span\u003e = expected overdoses in CBG \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:i\\)\u003c/span\u003e\u003c/span\u003e at year \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:t\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{P}_{it}\\)\u003c/span\u003e\u003c/span\u003e= population of CBG \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:i\\)\u003c/span\u003e\u003c/span\u003e at year \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:t\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{O}_{it}\\)\u003c/span\u003e\u003c/span\u003e = observed overdose count in CBG \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:i\\)\u003c/span\u003e\u003c/span\u003e at year \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:t\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eBefore spatial modeling, fixed-effect Poisson regression models were estimated to explore associations between overdose mortality and neighborhood-level exposures for each year. Covariates were standardized where applicable and entered as main effects. We included a mean-centered year variable defined as the year minus the base year (e.g., \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\text{Year}}_{ct}=t-2020\\)\u003c/span\u003e\u003c/span\u003e), ADI quintiles, standardized NDVI, park access, recoded urban classification, built environment intensity, and nighttime light emissions. All models used a log link and incorporated the log of the population or expected count as an offset.\u003c/p\u003e\u003cp\u003eTo evaluate spatial dependence, a spatial contiguity matrix was constructed. Using the \u003cem\u003espdep\u003c/em\u003e package,\u003csup\u003e32\u003c/sup\u003e a queen contiguity neighbors list was converted into a binary spatial weights matrix where each CBG’s neighbors were assigned equal weights. Moran’s I was computed to assess global spatial autocorrelation in observed overdose rates, and this spatial structure was retained and incorporated into all subsequent Bayesian models. To account for spatial dependence and temporal trends, a Bayesian hierarchical spatiotemporal Poisson model with autoregressive temporal structure was implemented using the \u003cem\u003eCARBayesST\u003c/em\u003e package.\u003csup\u003e33\u003c/sup\u003e An intrinsic conditional autoregressive (ICAR) prior was specified for the spatial component and temporal autocorrelation was modeled using a first-order autoregressive (AR(1)) process. Letting \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{Y}_{it}\\)\u003c/span\u003e\u003c/span\u003e denote the observed number of overdose deaths in CBG\u003csub\u003e\u003cem\u003ei\u003c/em\u003e\u003c/sub\u003e at time \u003cem\u003et\u003c/em\u003e, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{it},\\)\u003c/span\u003e\u003c/span\u003e the expected count, the model assumed that:\u003c/p\u003e\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:{Y}_{it}\\sim\\:\\text{Poisson}\\left({\\mu\\:}_{it}\\right)\\:\\text{with}\\:{\\mu\\:}_{it}={E}_{it}\\cdot\\:{\\theta\\:}_{it}$$\u003c/div\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe log relative risk was modeled as:\u003c/p\u003e\u003cdiv id=\"Equd\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e\n$$\\:\\text{l}\\text{o}\\text{g}\\left({\\theta\\:}_{it}\\right)=\\alpha\\:+{\\beta\\:}_{1}\\cdot\\:{\\text{Year}}_{ct}+{\\beta\\:}_{2}\\cdot\\:{\\text{ADI}}_{qi}+{\\beta\\:}_{3}\\cdot\\:{\\text{NDVI}}_{i}+{\\beta\\:}_{4}\\cdot\\:{\\text{ParkAcc}}_{i}+{\\beta\\:}_{5}\\cdot\\:{\\text{Urbanicity}}_{i}+{\\beta\\:}_{6}\\cdot\\:{\\text{BuiltEnv}}_{i}+{\\beta\\:}_{7}\\cdot\\:{\\text{Lights}}_{i}+{\\varphi\\:}_{i}+{\\delta\\:}_{it}$$\u003c/div\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eHere, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\alpha\\:\\)\u003c/span\u003e\u003c/span\u003e is the global intercept, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\beta\\:}_{k}\\)\u003c/span\u003e\u003c/span\u003e are fixed effects, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varphi\\:}_{i}\\)\u003c/span\u003e\u003c/span\u003e is the spatially structured random effect capturing spatial dependence via a conditional autoregressive prior, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\delta\\:}_{it}\\:\\)\u003c/span\u003e\u003c/span\u003eis an unstructured spatiotemporal random effect. Fixed effects were given diffuse Gaussian priors: \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\beta\\:}_{k}\\sim\\:\\mathcal{N}\\left(0,1000\\right)\\)\u003c/span\u003e\u003c/span\u003e. Spatial random effects ( _i ) followed a CAR prior with a precision parameter: \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\tau\\:}_{\\varphi\\:}\\sim\\:\\text{Gamma}\\left(0.5,0.0005\\right).\\)\u003c/span\u003e\u003c/span\u003e Spatiotemporal noise followed an exchangeable normal prior with variance: \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\tau\\:}_{\\delta\\:}^{-1}\\sim\\:\\text{Gamma}\\left(0.5,0.0005\\right).\\)\u003c/span\u003e\u003c/span\u003e The temporal autocorrelation parameter was assigned a uniform prior on the interval [0, 1]: \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\rho\\:\\sim\\:\\text{Uniform}\\left(0,1\\right).\\)\u003c/span\u003e\u003c/span\u003e Posterior estimation sampling was conducted using three parallel chains, each running for 20,000 iterations. This included a 5,000 iteration burn-in period and thinning every 10th sample, resulting in 4,500 posterior samples per chain. Model diagnostics included checking for multicollinearity using variance inflation factors (VIF). Model convergence was assessed using trace plots and posterior summaries.\u003c/p\u003e\u003cp\u003eRelative risk estimates were extracted for each CBG-year combination and mapped using the \u003cem\u003eggplot2\u003c/em\u003e package.\u003csup\u003e34\u003c/sup\u003e We descriptively examined the intersection of structural and environmental clustering by mapping the bivariate distribution using the \u003cem\u003ebiscale\u003c/em\u003e package.\u003csup\u003e35\u003c/sup\u003e ADI and nighttime light intensity were jointly classified using quantile binning (4x4 grid), and a custom legend was constructed using \u003cem\u003ecowplot.\u003c/em\u003e\u003csup\u003e36\u003c/sup\u003e Model-derived relative risk (RR) defined high-risk areas as having a RR \u0026gt; 1.5 and visualized using both single-year and faceted maps. To examine trends in risk over time, both modeled mean relative risk and observed death counts were plotted by year. We used the model-smoothed fitted values from the Bayesian spatiotemporal model to classify CBGs into one of four trend types: sharply increasing, increasing, decreasing, or no change. These trend types were derived by fitting linear and quadratic regressions to the fitted overdose risk over time for each block group and evaluating both the direction and significance of the slope and curvature.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cem\u003eDecedent Characteristics.\u003c/em\u003e Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e displays the characteristics of opioid-related overdose by victim's sex. Male decedents were slightly older than female decedents on average (47.2 vs. 45.7 years; \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and racial/ethnic composition varied by gender (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with males more likely to be Latino (16.7% vs. 8.8%) and females more likely to be non-Hispanic White (36.1% vs. 28.6%). Manner of death also differed significantly (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with a greater proportion of suicides among female decedents (2.2% vs. 0.5%) and a slightly higher proportion of accidental deaths among males (98.8% vs. 96.7%).\u003c/p\u003e\u003cp\u003eEnvironmental exposures varied modestly but significantly by gender. Females died in areas with slightly higher deprivation (mean ADI\u0026thinsp;=\u0026thinsp;114.6 vs. 113.1; \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.006), lower built environment intensity and nighttime light exposure (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001 for both), and marginally lower greenness as measured by NDVI (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.001). While expected overdose counts\u0026mdash;adjusted for population and risk\u0026mdash;were consistently lower among female decedents across all years (2020\u0026ndash;2023; \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.003\u0026ndash;0.019), observed overdose counts did not significantly differ by gender. This suggests that although modeled risk differed by gender-linked place characteristics, actual mortality patterns were similar.\u003c/p\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e displays the spatial intersection of socioeconomic deprivation (ADI) and nighttime light intensity. High-deprivation, high-light areas were concentrated in Chicago\u0026rsquo;s South and West Sides\u0026mdash;historically disinvested yet densely developed neighborhoods.\u003c/p\u003e\u003cp\u003eWe examined correlations between environmental and social predictors (see Supplementary Table\u0026nbsp;3). Built environment intensity was significantly and positively correlated with nighttime light (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.496) and negatively correlated with NDVI (r = \u0026minus;\u0026thinsp;0.454). Nighttime light and NDVI were strongly inversely related (\u003cem\u003er\u003c/em\u003e = \u0026minus;\u0026thinsp;0.577). Park access correlated positively with NDVI (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.140) and negatively with built-up area (\u003cem\u003er\u003c/em\u003e = \u0026minus;\u0026thinsp;0.280) and light (\u003cem\u003er\u003c/em\u003e = \u0026minus;\u0026thinsp;0.101). ADI was modestly correlated with built environment intensity (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.162, p\u0026thinsp;\u0026lt;\u0026thinsp;.001) and nighttime light (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.218, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001), and negatively correlated with NDVI (r = \u0026minus;\u0026thinsp;0.128, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001) and park access (r = \u0026minus;\u0026thinsp;0.062, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001).\u003c/p\u003e\u003cp\u003eBayesian spatiotemporal Poisson models adjusted for spatial and temporal autocorrelation and included environmental and sociodemographic covariates. VIFs and GVIFs were all well below conventional thresholds (maximum \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{G}_{VIF}^{.5df}=1.20\\)\u003c/span\u003e\u003c/span\u003e, indicating minimal multicollinearity and stable model estimation; Supplementary Table\u0026nbsp;4).\u003c/p\u003e\u003cp\u003eThe number of high-risk CBGs (relative risk [RR]\u0026thinsp;\u0026gt;\u0026thinsp;1.5) increased from 164 in 2020 to 177 in 2021, 207 in 2022, then declined slightly to 192 in 2023, suggesting persistence and spatial spread of overdose risk during the COVID-19 pandemic. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the results from the Bayesian spatiotemporal model. Neighborhood deprivation was strongly and monotonically associated with fatal overdose risk. Compared to the least deprived quintile, CBGs in the second through fifth quintiles had IRRs of 1.98, 3.08, 4.20, and 7.56, respectively, with all estimates showing narrow credible intervals and strong evidence of dose-response. In contrast, neither NDVI (IRR\u0026thinsp;=\u0026thinsp;0.99; 95% CrI: 0.94\u0026ndash;1.04) nor park access (IRR\u0026thinsp;=\u0026thinsp;0.99; 95% CrI: 0.94\u0026ndash;1.04) were significantly associated with overdose risk in adjusted models.\u003c/p\u003e\u003cp\u003eUrbanicity demonstrated mixed effects. CBGs classified as Transitional exhibited lower risk than Urban areas (IRR\u0026thinsp;=\u0026thinsp;0.74; 95% CrI: 0.56\u0026ndash;0.95), while Rural areas showed a non-significant increase (IRR\u0026thinsp;=\u0026thinsp;1.05; 95% CrI: 0.74\u0026ndash;1.61). Built environment intensity had a non-significant inverse association (IRR\u0026thinsp;=\u0026thinsp;0.71; 95% CrI: 0.38\u0026ndash;1.34). In contrast, nighttime light intensity was a strong positive predictor of risk: a one-unit increase in standardized light intensity was associated with a 8.24-fold increase in overdose mortality (95% CrI: 6.47\u0026ndash;14.75). Temporally, overdose risk remained relatively stable across years (IRR\u0026thinsp;=\u0026thinsp;0.99; 95% CrI: 0.97\u0026ndash;1.02). The spatial variance (τ\u0026sup2; = 10.64; 95% CrI: 8.33\u0026ndash;13.72) and strong residual spatial autocorrelation (ρS\u0026thinsp;=\u0026thinsp;2.54; 95% CrI: 2.45\u0026ndash;2.62) indicated persistent geographic clustering of risk beyond measured covariates. Temporal autocorrelation was modest (ρT\u0026thinsp;=\u0026thinsp;1.07; 95% CrI: 1.01\u0026ndash;1.17).\u003c/p\u003e\u003cp\u003eMost census block groups across all urban\u0026ndash;rural categories showed no significant change in opioid-related overdose trends over time. Increasing trends were observed in suburban (0.6%), transitional (12.5%), and rural CBGs (5.4%). Decreasing trends were found in suburban (3.1%) and rural CBGs (3.6%), while no decreasing trends occurred in Transitional areas (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eTo our knowledge, this is the first study to integrate detailed urban typologies, nighttime light data from the VIIRS sensor, and satellite-derived measures of built environment intensity, alongside ADI-based measures of social disadvantage, to examine how overdose-related deaths vary across diverse socio-environmental contexts. Our findings align with a small but growing body of research emphasizing the importance of considering both structural disadvantage and physical environmental characteristics in understanding the opioid overdose crisis.\u003csup\u003e22\u003c/sup\u003e Similar to Williams et al. \u003csup\u003e22\u003c/sup\u003e, who analyzed 565 municipalities across New Jersey to capture meaningful variation in infrastructure, social conditions, and service environments, our study demonstrated that environmental conditions shape overdose risk across diverse neighborhood contexts. Rather than constructing a composite index, however, we focused on isolating the independent and categorical effects of specific environmental exposures and examined how overdose risk varies across discrete urban typologies.\u003c/p\u003e\u003cp\u003eWe found persistent geographic and temporal clustering of overdose risk, underscoring the importance of spatially explicit models for identifying trends and informing prevention efforts. The number of CBGs with overdose risk exceeding 150% of the county average increased, mainly in highly urbanized Chicago areas on South and West Sides, regions historically affected by disinvestment, racialized poverty, and organized abandonment.\u003csup\u003e37\u003c/sup\u003e Urban centers still bear the highest overdose burden, but transitional areas saw the most marked increases. These spatial trends carry important policy implications. In urban areas with rising rates, new interventions may be needed, while stable regions could strengthen current strategies. Future research would benefit by incorporating urban, suburban, peri-urban, and rural distinctions to better understand how environmental and social factors interact across the urban\u0026ndash;rural continuum.\u003c/p\u003e\u003cp\u003eOur findings support a growing body of research identifying NLI as a chronic environmental stressor associated with a wide range of adverse health outcomes \u003csup\u003e16,38\u003c/sup\u003e, including sleep disorders \u003csup\u003e39\u003c/sup\u003e, preterm birth \u003csup\u003e14\u003c/sup\u003e, gestational diabetes,\u003csup\u003e40\u003c/sup\u003e colorectal cancer,\u003csup\u003e41\u003c/sup\u003e and mood disorders. NLI has also been linked to depressive symptoms, anxiety, and suicidal ideation or attempt.\u003csup\u003e39,42\u003c/sup\u003e These effects are attributed to NLI\u0026rsquo;s disruption of sleep cycles and circadian rhythms, which reduces melatonin production and impairs the functioning of the immune and hormonal systems. Such circadian disruptions have, in turn, been associated with increased vulnerability to substance use and addiction.\u003csup\u003e43\u003c/sup\u003e Importantly, NLI also reflects the presence of other environmental stressors that contribute to poor health. For example, areas with high NLI exposure tend to have less green space, more air pollution, concentrated poverty, and greater area deprivation,\u003csup\u003e38\u003c/sup\u003e conditions that are independently linked to poor mental health\u003csup\u003e38\u003c/sup\u003e and increased risk for substance misuse.\u003csup\u003e44\u003c/sup\u003e These overlapping exposures suggest that NLI contributes to overdose vulnerability both through direct biological mechanisms and as an indicator of broader structural disadvantage.\u003c/p\u003e\u003cp\u003eRegarding area deprivation, we found a dose-response relationship between opioid-related overdose and ADI consistent with past studies.\u003csup\u003e45\u003c/sup\u003e NLI contributed to geographic disparities in overdose deaths in urban areas of Chicago experiencing concentrated structural disadvantage. These findings align with prior research showing that NLI exposure is significantly higher in the most socially vulnerable neighborhoods, disproportionately affecting racially and ethnically minoritized communities.\u003csup\u003e42\u003c/sup\u003e We also note that the least deprived neighborhoods exhibited lower levels of NLI, despite their dense population and high economic activity. At a minimum, our results contribute to a growing body of evidence that identifies nighttime light exposure not only as a public health concern but also as a potential marker of environmental injustice.\u003csup\u003e42\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eWe also observed that opioid-related fatality rates were significantly lower in transitional areas, which do not fit neatly into entirely urban or rural categories. However, these areas experienced the sharpest increases in overdose risk for the period we considered, which overlaps with the COVID-19 pandemic. This suggests that areas associated with lower risk may represent emerging hotspots with trends that increase over time. By contrast, suburban or peri-urban areas exhibited the most stable overdose trajectories, with over 96% of block groups showing no significant change across the study period. This relative stability may reflect a more substantial presence of protective factors in these areas \u003csup\u003e46\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eConsistent with prior research,\u003csup\u003e47\u003c/sup\u003e we found no significant association between opioid-related mortality and measures of vegetative greenness (e.g., NDVI) or park access. This suggests that, at least at the spatial resolution and form captured in this study, the availability of green space may not confer a measurable protective effect against overdose. Additionally, our satellite-derived measure of Built Environment Intensity, which reflects infrastructure density and land-use patterns, was not significantly associated with opioid mortality after adjusting for other covariates.\u003c/p\u003e\u003cp\u003eAlthough the examination of sex differences and environmental exposures was not the primary aim of this study, it yielded some key findings worth mentioning. Across all four years examined, the observed and expected overdose counts were consistently lower in the areas where female decedents overdosed. Yet, fatal overdoses among women were significantly more likely to occur in neighborhoods with higher social deprivation, lower built environment intensity, reduced nighttime light exposure, and diminished vegetative greenness compared to locations associated with male decedents. This pattern aligns with research on gender and substance use, which highlights the disproportionate stigma, trauma exposure, and structural barriers that women face in accessing care.\u003csup\u003e48\u003c/sup\u003e There is also evidence that females may be more biologically sensitive to circadian disruption and socially vulnerable to environmental neglect, both of which may increase overdose risk in underrecognized ways. While studies have generally not found significant sex interactions with NLI in outcomes such as preterm birth or gestational diabetes \u003csup\u003e14,15\u003c/sup\u003e, underlying physiological and contextual sex differences remain highly relevant. These findings suggest a dimension to overdose vulnerability among males and females that warrants future research.\u003c/p\u003e\u003cp\u003eThis study has several limitations that should be acknowledged. First, the analysis is limited to Cook County, Illinois, which includes the city of Chicago, an area characterized by distinct patterns of segregation, infrastructure, and access to services. As such, the findings may not be generalizable to regions with different geographic, demographic, or policy contexts. Second, although we modeled time-varying estimates of overdose risk, we did not incorporate time-varying covariates. Environmental exposures are shaped by both natural and human factors that vary across time and space.\u003csup\u003e49\u003c/sup\u003e Therefore, future research should investigate how temporal changes in social and ecological conditions may influence overdose patterns across levels of urbanization. Third, the study period was relatively short and coincided with the COVID-19 pandemic. Despite studies finding no association between social distancing and opioid-related overdose in Cook County,\u003csup\u003e45\u003c/sup\u003e the unique context may limit the detection of long-term or typical trends, as the pandemic introduced widespread disruptions in healthcare access, social services, and substance use patterns.\u003csup\u003e50\u0026ndash;53\u003c/sup\u003e Finally, Cook County lacks a balanced representation of rural, suburban, and urban areas, restricting our ability to assess overdose risk across the full urban\u0026ndash;rural continuum. Future studies should apply these methods in regions with greater spatial diversity to more fully examine how social and environmental conditions interact across varying geographic contexts to shape overdose vulnerability.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eProactive public health prevention should consider both aspects of the physical and social environment, as well as the demographic characteristics of the population residing there. Our results contribute to a more nuanced understanding of how social and built environmental conditions shape overdose vulnerability at fine geographic scales. The integration of remote sensing data provides new insight into how features of urban form and human activity influence health outcomes beyond what is captured by traditional socioeconomic indicators.\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMonnat SM, Rigg KK. The opioid crisis in rural and small town America. Published online 2018. Accessed July 23, 2025. https://scholars.unh.edu/carsey/343/\u003c/li\u003e\n\u003cli\u003ePeters DJ, Monnat SM, Hochstetler AL, Berg MT. The Opioid Hydra: Understanding Overdose Mortality Epidemics and Syndemics Across the Rural-Urban Continuum. \u003cem\u003eRural Sociology\u003c/em\u003e. 2020;85(3):589-622. doi:10.1111/ruso.12307\u003c/li\u003e\n\u003cli\u003ePalombi L, Hawthorne AN, Irish A, Becher E, Bowen E. \u0026ldquo;One Out of Ten Ain\u0026rsquo;t Going to Make It\u0026rdquo;: An Analysis of Recovery Capital in the Rural Upper Midwest. \u003cem\u003eJournal of Drug Issues\u003c/em\u003e. 2019;49(4):680-702. doi:10.1177/0022042619859309\u003c/li\u003e\n\u003cli\u003ePear VA, Ponicki WR, Gaidus A, et al. Urban-rural variation in the socioeconomic determinants of opioid overdose. \u003cem\u003eDrug and alcohol dependence\u003c/em\u003e. 2019;195:66-73.\u003c/li\u003e\n\u003cli\u003eRoth KB, Goplerud DK, Babjak JS, Nguyen JL, Gilbert LR. Examining the association of rurality with opioid-related morbidity and mortality in Georgia: A geospatial analysis. \u003cem\u003eJournal of Substance Use and Addiction Treatment\u003c/em\u003e. 2024;162:209336.\u003c/li\u003e\n\u003cli\u003eLatimore A, Ortega Hinojose A, Kestner L, Patel. Exploring Urban-Rural Disparities in Accessing Treatment for Opioid Use Disorder. American Institutes for Research. November 19, 2021. Accessed July 10, 2025. http://www.air.org/resource/field/exploring-urban-rural-disparities-accessing-treatment-opioid-use-disorder\u003c/li\u003e\n\u003cli\u003ePatino JE, Duque JC, Pardo-Pascual JE, Ruiz LA. Using remote sensing to assess the relationship between crime and the urban layout. \u003cem\u003eApplied Geography\u003c/em\u003e. 2014;55:48-60. doi:10.1016/j.apgeog.2014.08.016\u003c/li\u003e\n\u003cli\u003eWynveen CJ, Bixler RD, Hammitt WE. Law Enforcement Perceptions and Changes in the United States Park Service: Urban proximity and level of enforcement practices. \u003cem\u003eAnnals of Leisure Research\u003c/em\u003e. 2007;10(3/4):532-549. doi:10.1080/11745398.2007.9686780\u003c/li\u003e\n\u003cli\u003eBrunson RK, Miller J. YOUNG BLACK MEN AND URBAN POLICING IN THE UNITED STATES. \u003cem\u003eBritish Journal of Criminology\u003c/em\u003e. 2006;46(4):613-640. doi:10.1093/bjc/azi093\u003c/li\u003e\n\u003cli\u003ePalombi LC, St Hill CA, Lipsky MS, Swanoski MT, Lutfiyya MN. A scoping review of opioid misuse in the rural United States. \u003cem\u003eAnnals of Epidemiology\u003c/em\u003e. 2018;28(9):641-652. doi:10.1016/j.annepidem.2018.05.008\u003c/li\u003e\n\u003cli\u003eStuck in the middle with you: Peri-urban areas and the food system. Center for a Livable Future. Accessed July 25, 2025. https://clf.jhsph.edu/stories/stuck-middle-you-peri-urban-areas-and-food-system\u003c/li\u003e\n\u003cli\u003eBaumgartner J, Rodriguez J, Berkhout F, et al. Synthesizing the links between secure housing tenure and health for more equitable cities. \u003cem\u003eWellcome open research\u003c/em\u003e. 2023;7:18.\u003c/li\u003e\n\u003cli\u003eTempalski B, Williams LD, Kolak M, Ompad DC, Koschinsky J, McLafferty SL. Conceptualizing the Socio-Built Environment: An Expanded Theoretical Framework to Promote a Better Understanding of Risk for Nonmedical Opioid Overdose Outcomes in Urban and Non-Urban Settings. \u003cem\u003eJ Urban Health\u003c/em\u003e. 2022;99(4):701-716. doi:10.1007/s11524-022-00645-3\u003c/li\u003e\n\u003cli\u003eSun Q, Yang Y, Liu J, et al. Association between exposure to outdoor artificial light at night and the risk of preterm birth. \u003cem\u003eFrontiers in Public Health\u003c/em\u003e. 2023;11:1280790.\u003c/li\u003e\n\u003cli\u003eZou HX, Hu LW, Zhang Z, et al. Outdoor light at night exposure was associated with hypothyroidism in pregnant women: A national study in China. \u003cem\u003eScience of The Total Environment\u003c/em\u003e. 2025;958:178017.\u003c/li\u003e\n\u003cli\u003eBadpa M, Schneider A, Ziegler AG, et al. Outdoor Light at Night and Children\u0026rsquo;s Body Mass: A Cross-Sectional Analysis in the Fr1da Study. \u003cem\u003eEnvironmental Research\u003c/em\u003e. 2023;232:116325. doi:10.1016/j.envres.2023.116325\u003c/li\u003e\n\u003cli\u003eFilippini T, Costanzini S, Chiari A, et al. Light at night exposure and risk of dementia conversion from mild cognitive impairment in a Northern Italy population. \u003cem\u003eInt J Health Geogr\u003c/em\u003e. 2024;23(1). doi:10.1186/s12942-024-00384-5\u003c/li\u003e\n\u003cli\u003eHartley K, Perazzo J, Brokamp C, et al. Residential surrounding greenness and self-reported symptoms of anxiety and depression in adolescents. \u003cem\u003eEnvironmental research\u003c/em\u003e. 2021;194:110628.\u003c/li\u003e\n\u003cli\u003eLee HY, Chang HT, Herianto S, et al. Do greenness and landscape indices for greenspace correlate with suicide ratio? \u003cem\u003eLandscape and Urban Planning\u003c/em\u003e. 2024;242:104935. doi:10.1016/j.landurbplan.2023.104935\u003c/li\u003e\n\u003cli\u003eZhang C, Wang C, Chen C, et al. Effects of tree canopy on psychological distress: A repeated cross-sectional study before and during the COVID-19 epidemic. \u003cem\u003eEnvironmental Research\u003c/em\u003e. 2022;203:111795. doi:10.1016/j.envres.2021.111795\u003c/li\u003e\n\u003cli\u003eBecker DA, Browning MHEM, McAnirlin O, Yuan S, Helbich M. Is green space associated with opioid-related mortality? An ecological study at the U.S. county level. \u003cem\u003eUrban Forestry \u0026amp; Urban Greening\u003c/em\u003e. 2022;70:127529. doi:10.1016/j.ufug.2022.127529\u003c/li\u003e\n\u003cli\u003eWilliams LD, Kolak M, Villanueva C, Ompad DC, Tempalski B. Creation and Validation of a New Socio-built Environment Index Measure of Opioid Overdose Risk for Use in Both Non-urban and Urban Settings. \u003cem\u003eJ Urban Health\u003c/em\u003e. 2023;100(5):1048-1061. doi:10.1007/s11524-023-00754-7\u003c/li\u003e\n\u003cli\u003eCook County Medical Examiners Office. Medical Examiner Case Archive. Cook County Government Open Data. April 25, 2022. https://datacatalog.cookcountyil.gov/Public-Safety/Medical-Examiner-Case-Archive/cjeq-bs86\u003c/li\u003e\n\u003cli\u003eGalea S. Income distribution and risk of fatal drug overdose in New York City neighborhoods. \u003cem\u003eDrug and Alcohol Dependence\u003c/em\u003e. 2003;70(2):139-148. doi:10.1016/S0376-8716(02)00342-3\u003c/li\u003e\n\u003cli\u003eGalea S, Rudenstine S, Vlahov D. Drug use, misuse, and the urban environment. \u003cem\u003eDrug and Alcohol Review\u003c/em\u003e. 2005;24(2):127-136. doi:10.1080/09595230500102509\u003c/li\u003e\n\u003cli\u003eNeely B, Samura M. Social geographies of race: connecting race and space. \u003cem\u003eEthnic and Racial Studies\u003c/em\u003e. 2011;34(11):1933-1952. doi:10.1080/01419870.2011.559262\u003c/li\u003e\n\u003cli\u003eWalker K. \u003cem\u003eTigris: Load Census TIGER/Line Shapefiles\u003c/em\u003e.; 2023. https://CRAN.R-project.org/package=tigris\u003c/li\u003e\n\u003cli\u003eWalker K, Herman M. \u003cem\u003eTidycensus: Load US Census Boundary and Attribute Data as \u0026ldquo;tidyverse\u0026rdquo; and \u0026rsquo;Sf\u0026rsquo;-Ready Data Frames\u003c/em\u003e.; 2022. https://CRAN.R-project.org/package=tidycensus\u003c/li\u003e\n\u003cli\u003eDaniel Baston. \u003cem\u003eExactextractr: Fast Extraction from Raster Datasets Using Polygons\u003c/em\u003e.; 2023. https://CRAN.R-project.org/package=exactextractr\u003c/li\u003e\n\u003cli\u003eSingh GK. Area Deprivation and Widening Inequalities in US Mortality, 1969\u0026ndash;1998. \u003cem\u003eAm J Public Health\u003c/em\u003e. 2003;93(7):1137-1143.\u003c/li\u003e\n\u003cli\u003eKrieger N, Dalton J, Wang C, Perzynski A. \u003cem\u003eSociome: Operationalizing Social Determinants of Health Data for Researchers\u003c/em\u003e.; 2023. https://CRAN.R-project.org/package=sociome\u003c/li\u003e\n\u003cli\u003ePebesma E, Bivand RS. \u003cem\u003eSpatial Data Science with Applications in R\u003c/em\u003e. Chapman \u0026amp; Hall; 2023. https://r-spatial.org/book/\u003c/li\u003e\n\u003cli\u003eLee D, Rushworth A, Napier G. Spatio-Temporal Areal Unit Modeling in R with Conditional Autoregressive Priors Using the CARBayesST Package. \u003cem\u003eJournal of Statistical Software\u003c/em\u003e. 2018;84:1-39. doi:10.18637/jss.v084.i09\u003c/li\u003e\n\u003cli\u003eH Wickham, Chang W, Pedersen T, Takahashi K, Wilke C, Woo K. ggplot2: Create Elegant Data Visualisations Using the Grammar of Graphics. Published online 2018. https://CRAN.R-project.org/package=ggplot2\u003c/li\u003e\n\u003cli\u003ePrener C, Grossenbacher T, Zehr A. \u003cem\u003eBiscale: Tools and Palettes for Bivariate Thematic Mapping\u003c/em\u003e.; 2022. https://CRAN.R-project.org/package=biscale\u003c/li\u003e\n\u003cli\u003eWilke CO. \u003cem\u003eCowplot: Streamlined Plot Theme and Plot Annotations for \u0026ldquo;Ggplot2.\u0026rdquo;\u003c/em\u003e; 2023. https://CRAN.R-project.org/package=cowplot\u003c/li\u003e\n\u003cli\u003eBarboza-Salerno G, Liebhard B, Duhaney S, Harrington T. Redlining, Reinvestment, and Racial Segregation: A Bayesian Spatial Analysis of Mortgage Lending T.\u003c/li\u003e\n\u003cli\u003eLiao YA. \u003cem\u003eMental Health Problems in Urban Neighbourhoods: The Role of Physical and Social Environment\u003c/em\u003e. PhD Thesis. Ludwig-Maximilians-Universit\u0026auml;t M\u0026uuml;nchen M\u0026uuml;nchen; 2022. Accessed July 24, 2025. https://pure.mpg.de/pubman/faces/ViewItemOverviewPage.jsp?itemId=item_3608538\u003c/li\u003e\n\u003cli\u003eHu K, Li W, Zhang Y, et al. Association between outdoor artificial light at night and sleep duration among older adults in China: a cross-sectional study. \u003cem\u003eEnvironmental research\u003c/em\u003e. 2022;212:113343.\u003c/li\u003e\n\u003cli\u003eSun Q, Ye F, Liu J, et al. Outdoor artificial light at night exposure and gestational diabetes mellitus: a case\u0026ndash;control study. \u003cem\u003eFrontiers in Public Health\u003c/em\u003e. 2024;12:1396198.\u003c/li\u003e\n\u003cli\u003eHarding BN, Palomar-Cros A, Valent\u0026iacute;n A, et al. Comparing Data from Three Satellites on Artificial Light at Night (ALAN): Focusing on Blue Light\u0026rsquo;s Influence on Colorectal Cancer in a Case\u0026ndash;Control Study in Spain. \u003cem\u003eEnviron Health Perspect\u003c/em\u003e. 2024;132(5). doi:10.1289/ehp14414\u003c/li\u003e\n\u003cli\u003eXiao Q, Lyu Y, Zhou M, et al. Artificial light at night and social vulnerability: An environmental justice analysis in the U.S. 2012\u0026ndash;2019. \u003cem\u003eEnvironment International\u003c/em\u003e. 2023;178:108096. doi:10.1016/j.envint.2023.108096\u003c/li\u003e\n\u003cli\u003eHasler BP, Smith LJ, Cousins JC, Bootzin RR. Circadian rhythms, sleep, and substance abuse. \u003cem\u003eSleep medicine reviews\u003c/em\u003e. 2012;16(1):67-81.\u003c/li\u003e\n\u003cli\u003eVan Draanen J, Tsang C, Mitra S, et al. Mental disorder and opioid overdose: a systematic review. \u003cem\u003eSoc Psychiatry Psychiatr Epidemiol\u003c/em\u003e. 2022;57(4):647-671. doi:10.1007/s00127-021-02199-2\u003c/li\u003e\n\u003cli\u003eBarboza G, Angulski K, Hines L, Brown P. Variability in Opioid-Related Drug Overdoses, Social Distancing, and Area-Level Deprivation during the COVID-19 Pandemic: a Bayesian Spatiotemporal Analysis. \u003cem\u003eJournal of Urban Health\u003c/em\u003e. 2022;99(5):873-886.\u003c/li\u003e\n\u003cli\u003eSrinivasan S, Pustz J, Marsh E, Young LD, Stopka TJ. Risk factors for persistent fatal opioid-involved overdose clusters in Massachusetts 2011\u0026ndash;2021: a spatial statistical analysis with socio-economic, accessibility, and prescription factors. \u003cem\u003eBMC Public Health\u003c/em\u003e. 2024;24(1). doi:10.1186/s12889-024-19399-5\u003c/li\u003e\n\u003cli\u003eBecker DA, Browning MH, McAnirlin O, Yuan S, Helbich M. Is green space associated with opioid-related mortality? An ecological study at the US county level. \u003cem\u003eUrban Forestry \u0026amp; Urban Greening\u003c/em\u003e. 2022;70:127529.\u003c/li\u003e\n\u003cli\u003eLamonica AK, Boeri M, Turner J. Circumstances of overdose among suburban women who use opioids: extending an urban analysis informed by drug, set, and setting. \u003cem\u003eInternational Journal of Drug Policy\u003c/em\u003e. 2021;90:103082.\u003c/li\u003e\n\u003cli\u003eWang H, Lin C, Ou S, et al. Multilevel Change of Urban Green Space and Spatiotemporal Heterogeneity Analysis of Driving Factors. \u003cem\u003eSustainability\u003c/em\u003e. 2024;16(11):4762. doi:10.3390/su16114762\u003c/li\u003e\n\u003cli\u003eBartholomew TS, Nakamura N, Metsch LR, Tookes HE. Syringe services program (SSP) operational changes during the COVID-19 global outbreak. \u003cem\u003eInternational Journal of Drug Policy\u003c/em\u003e. 2020;83:102821. doi:10.1016/j.drugpo.2020.102821\u003c/li\u003e\n\u003cli\u003eGhose R, Forati AM, Mantsch JR. Impact of the COVID-19 Pandemic on Opioid Overdose Deaths: a Spatiotemporal Analysis. \u003cem\u003eJ Urban Health\u003c/em\u003e. 2022;99(2):316-327. doi:10.1007/s11524-022-00610-0\u003c/li\u003e\n\u003cli\u003eKhatri UG, Pizzicato LN, Viner K, et al. Racial/ethnic disparities in unintentional fatal and nonfatal emergency medical services\u0026ndash;attended opioid overdoses during the COVID-19 pandemic in Philadelphia. \u003cem\u003eJAMA network open\u003c/em\u003e. 2021;4(1):e2034878-e2034878.\u003c/li\u003e\n\u003cli\u003eBarboza GE, Schiamberg LB, Pachl L. A spatiotemporal analysis of the impact of COVID-19 on child abuse and neglect in the city of Los Angeles, California. \u003cem\u003eChild abuse \u0026amp; neglect\u003c/em\u003e. 2021;116:104740.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1.\u003c/strong\u003e Characteristics of individuals who died from opioid-related overdose, stratified by sex\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"104%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 192px;\"\u003e\n \u003cp\u003eVariable\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 131px;\"\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003cp\u003e(N = 1628, 22.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 142px;\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003cp\u003e(N = 5530, 77.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 111px;\"\u003e\n \u003cp\u003eTotal\u003c/p\u003e\n \u003cp\u003e(N = 7158)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e-value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge (Mean, SD)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e45.7 (13.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e47.2 (13.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e46.8 (13.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRace/Ethnicity (%)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Latino\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e8.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e16.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e14.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Non-Hispanic Asian\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Non-Hispanic Black\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e53.7%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e53.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e53.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Non-Hispanic Indigenous\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Non-Hispanic White\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e36.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e28.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e30.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Other\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Unknown\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eManner of Death (%)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Accident\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e96.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e98.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e98.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Homicide\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Natural\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Suicide\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e2.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Undetermined\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eYear of Death (%)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.189\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2020\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e24.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e24.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e24.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2021\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e27.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e25.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e25.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2022\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e25.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e26.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e26.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2023\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e22.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e24.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e24.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEnvironmental Indicators\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(Mean, SD)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; ADI\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e114.6 (19.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e113.1 (19.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e113.4 (19.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.006\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Built Environment\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.3 (0.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.3 (0.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.3 (0.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Nighttime Light\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.2 (0.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.2 (0.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.2 (0.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; NDVI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.1 (0.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.1 (0.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.1 (0.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; Park Access (acres)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e59.8 (159.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e59.8 (254.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e59.8 (236.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.989\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eExpected Overdose Counts (Mean, SD)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2020\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.4 (0.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.5 (0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.5 (0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2021\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.5 (0.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.5 (0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.5 (0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.006\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2022\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.5 (0.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.5 (0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.5 (0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.006\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2023\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e0.4 (0.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e0.5 (0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e0.5 (0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.019\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eObserved Overdose Counts (Mean, SD)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2020\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e1.6 (2.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e1.6 (2.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e1.6 (2.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.391\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2021\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e1.8 (2.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e1.8 (2.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e1.8 (2.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.743\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2022\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e1.7 (2.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e1.8 (2.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e1.8 (2.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.350\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 192px;\"\u003e\n \u003cp\u003e\u0026nbsp; 2023\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e1.6 (1.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003e1.6 (1.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e1.6 (1.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 73px;\"\u003e\n \u003cp\u003e0.504\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eNotes. Means and standard deviations are reported for continuous variables; percentages are reported for categorical variables.\u0026nbsp;\u003cem\u003eP\u003c/em\u003e-values reflect results from two-sample t-tests (for continuous variables) or chi-square tests (for categorical variables) comparing female and male decedents.\u003cbr\u003e\u0026nbsp;ADI = Area Deprivation Index; NDVI = Normalized Difference Vegetation Index.\u003cbr\u003e\u0026nbsp;Environmental indicators reflect the census block group where the decedent\u0026rsquo;s death occurred.\u003cbr\u003e\u0026nbsp;Expected overdose counts were derived using countywide rates adjusted for population size.\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"3\" valign=\"bottom\" style=\"width: 542px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTable 2\u003c/strong\u003e. Bayesian Spatiotemporal Poisson Model\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u003cstrong\u003ePredictor\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u003cstrong\u003eIRR (95% CrI)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u003cstrong\u003ePercent Change (%)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eIntercept\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.12 (0.10, 0.15)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e-87.7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eYear (centered)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.99 (0.97, 1.02)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e-0.85\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eArea Deprivation Index (ADI) (ref = most deprived)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Quintile 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1.98 (1.72, 2.28)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e+98.0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Quintile 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3.08 (2.66, 3.54)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e+208.0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Quintile 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4.20 (3.63, 4.83)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e+320.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Quintile 5\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e7.56 (6.57, 8.68)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e+656.0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eNDVI (Vegetative Greenness)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.99 (0.94, 1.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e-0.81\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003ePark Access (Acres within CBG)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.99 (0.94, 1.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e-0.80\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eUrbanization Category (ref = Urban)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Suburban\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.75 (0.58, 0.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026ndash;25.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Rural\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1.15 (0.80, 1.64)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e+14.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Transitional\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.69 (0.26, 1.72)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u0026ndash;30.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eBuilt Environment Intensity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.71 (0.38, 1.34)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e-28.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eNighttime Light Intensity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e8.24 (6.47, 14.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e+724.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eSpatial Variance (\u0026tau;\u0026sup2;)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e10.64 (8.33, 13.72)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e--\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eSpatial Autocorrelation (\u0026rho;S)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e2.54 (2.45, 2.62)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e--\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eTemporal Autocorrelation (\u0026rho;T)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1.07 (1.01, 1.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e--\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eNotes. Incidence Rate Ratios (IRRs) and 95% Credible Intervals (CrIs) are reported for fixed effects from a Bayesian hierarchical Poisson model with spatial and temporal structure. IRR = exp(Beta); Percent change is calculated as (IRR \u0026minus; 1) \u0026times; 100.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-urban-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jurh","sideBox":"Learn more about [Journal of Urban Health](https://www.springer.com/journal/11524)","snPcode":"11524","submissionUrl":"https://www.editorialmanager.com/jurh","title":"Journal of Urban Health","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7222499/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7222499/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eGeographic differences in drug overdose patterns across the rural-urban continuum are well documented. Few studies utilize remote sensing data to quantify the environmental and structural features that shape overdose risk at varying geographic scales. We conducted a retrospective ecological study to examine fatal opioid overdoses across census block groups in Cook County, Illinois, from 2020 to 2023. Urbanicity was determined based on the Global Human Settlement Layer Model (GHSL-MOD), with satellite data used to derive metrics of built-up intensity, vegetative greenness, and nighttime light emissions. Environmental indicators were combined with census-based measures of neighborhood deprivation to characterize spatial variation in physical and social conditions. A Bayesian spatiotemporal model estimated neighborhood-level overdose risk, accounting for spatial dependence, temporal trends, and environmental exposures. Overdose risk exhibited significant spatial clustering and strong associations with both social and environmental factors. Neighborhood disadvantage had a dose-response relationship, with fatal overdose risk in areas with the most deprivation, experiencing over seven times the risk, compared to the least deprived. Nighttime light intensity was strongly associated with increased overdose risk, while vegetative greenness and park access showed no significant protective effects. Increasing trends were detected in rural and transitional zones despite a higher risk in urban centers. Demographic characteristics of overdose victims varied across the county, suggesting potential geographic disparities in risk. The physical and social features of neighborhoods underscore the need for early surveillance and intervention within and outside urban centers. These factors should be incorporated into targeted, place-based strategies to lower opioid-related deaths.\u003c/p\u003e","manuscriptTitle":"An assessment of overdose mortality risk across the urban–rural continuum: Integrating satellite-derived and socioeconomic indicators","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-30 07:07:31","doi":"10.21203/rs.3.rs-7222499/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revise and resubmit","date":"2026-02-04T20:41:24+00:00","index":"","fulltext":""},{"type":"reviewerAgreed","content":"","date":"2025-09-19T22:20:42+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-25T16:00:05+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-28T19:32:03+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of Urban Health","date":"2025-07-28T13:21:04+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-urban-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jurh","sideBox":"Learn more about [Journal of Urban Health](https://www.springer.com/journal/11524)","snPcode":"11524","submissionUrl":"https://www.editorialmanager.com/jurh","title":"Journal of Urban Health","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"79931cbf-a771-4eaf-9d7a-c377bb8fad9d","owner":[],"postedDate":"July 30th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-03-04T18:00:10+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-30 07:07:31","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7222499","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7222499","identity":"rs-7222499","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}