Census Block Group Research Articles

High-risk neighbourhoods and behaviour outcomes in children born extremely preterm.

Study the association between neighbourhood risk and behaviour in extreme preterm (EPT) children. We hypothesised that EPT children living in high-risk neighbourhoods have increased risk of clinical range behaviour problems at age 30-36 months. Retrospective analyses of Child Behavior Checklist (CBCL)scores for 739 EPTs born 2005-2016. Addresses were geocoded to identify census block groups and create high versus low-risk groups. Regression analyses assessed the impact of neighbourhood risk on behaviour. Children from high-risk (N = 272, 39%) and low-risk (N = 417, 61%) neighbourhoods were compared. In adjusted analyses, odds of clinical range scores remained greater in high-risk neighbourhoods for Emotionally Reactive (OR: 4.32, CI: 1.13, 16.51), Somatic Complaints (2.30, CI 1.11,4.79), Withdrawn (OR: 2.56, CI: 1.21, 5,42), Aggressive Behaviour (OR: 4.12, CI: 1.45, 11.68), Internalising (OR: 1.96, CI: 1.17, 3.28), and Total score (OR: 1.86, OR: CI: 1.13, 3.07). Cognitive delay was higher in high-risk neighbourhoods and a risk factor for Attention Problems (2.10,1.08, 4.09). Breast milk was protective for Emotionally Reactive (OR: 0.22, CI: 0.06, 0.85) and Sleep Problems (OR: 0.47, CI:0.24, 0.94). Neighbourhood risk provided an independent contribution to preterm adverse behaviour outcomes with cognitive delay an additional independent risk factor. Breast milk at discharge was protective.

Uncovering Representation Bias in Large‐scale Cellular Phone‐based Data: A Case Study in North Carolina

Large cellular phone‐based mobility datasets are an important new data source for research on human movement. We investigate and illustrate bias in representation in a large mobility data set at the census block group, tract, and county levels. We paired American Community Survey (ACS) 2019 data with SafeGraph (SG) cell phone mobility data to elucidate potential bias in SG data by examining ACS estimated population against the number of devices in the SG data, stratifying by key sociodemographic variables such as income, percent Black population, percent of population over 55 years, percent of population 18–65 years, percent of people living in crowded living conditions, and urbanization level. We evaluated whether the bias varied over time by examining a 10‐month period. This bias changes with key demographic characteristics and changes over time. Specifically, we see underrepresentation in areas that have the highest percentage of Black population at all aggregation levels. We also see underrepresentation at all levels in areas with the highest percentage of working age residents as well as areas with the lowest median incomes. Researchers should be cautious when using mobility datasets because of bias differential on key sociodemographic factors and collection time.

Peaked too soon? Analyzing the shifting patterns of PM peak period travel in Southern California

Daily vehicle travel collapsed with the onset of the COVID-19 pandemic in early 2020 but largely bounced back by late 2021. The pandemic caused dramatic changes to working, schooling, shopping, and leisure activities, and to the travel associated with them. Several of these changes have so far proven enduring. So, while overall vehicle travel had largely returned to pre-pandemic levels by late 2021, the underlying drivers of this travel have likely changed.To examine one element of this issue, we analyzed whether patterns of daily trip-making shifted temporally between the fall of 2019 and 2021 in the Greater Los Angeles megaregion. We used location-based service data to examine vehicle trip originations for each hour of the day at the U.S. census block group level in October 2019 and October 2021. We observed notable shifts in the timing of post-pandemic PM peak travel, so we examined changes in the ratio of mid-week trips originating in the early afternoon (12–3:59 PM) and the late afternoon/early evening (4–7:59 PM).We found a clear shift in the temporal distribution of PM trip-making, with relatively more late PM peak period trip-making prior to the pandemic, and more early PM peak trip-making in 2021. The peak afternoon/evening trip-making hour shifted from 5–5:59 PM to 3–3:59 PM. We also found that afternoon/evening trip-making in each year is largely explained by three workplace-area/school-area factors: (1) the number of schoolchildren in a block group (earlier); (2) block groups with large shares of potential remote workers (earlier), and (3) block groups with large shares of low-wage jobs and workers of color (later, except for Black workers in 2021). We found the earlier shift in PM peak travel between pre- and late-pandemic periods to be explained most by (1) higher shares of potential remote workers and (2) higher shares of low-wage jobs and workers of color. These findings suggest that the rise of working from home has likely led to a shift in PM peak travel earlier in the afternoon when school chauffeuring trips are most common. This is especially true for low-income workers and workers of color.

Investigating neighborhood adaptability using mobility networks: a case study of the COVID-19 pandemic

What predicts a neighborhood’s adaptability to essential public health policies and shelter-in-place regulations that prevent the harmful spread of COVID-19? To answer this question, we present a novel application of human mobility patterns and human behavior in a network setting. We analyze 2 years of mobility data (January 2019 to December 2020) from New York City and construct weekly mobility networks between census block groups based on aggregated point-of-interest visit patterns. Our results indicate that neighborhoods’ socioeconomic and geographic characteristics play a significant role in predicting their adaptability to active shelter-in-place policies. Our simulation outcomes reveal that, alongside factors such as race, education, and income, the geographical attributes of neighborhoods, such as access to amenities that satisfy community needs are equally important factors in predicting neighborhood adaptability to public health policies. These findings offer valuable insights that can enhance urban planning strategies, thereby aiding pandemic mitigation efforts and fostering increased adaptability of urban areas in the face of exogenous shocks like the COVID-19 pandemic.

Development of a community severance index for urban areas in the United States: A case study in New York City

Background and aimsTraffic-related exposures, such as air pollution and noise, have a detrimental impact on human health, especially in urban areas. However, there remains a critical research and knowledge gap in understanding the impact of community severance, a measure of the physical separation imposed by road infrastructure and motorized road traffic, limiting access to goods, services, or social connections, breaking down the social fabric and potentially also adversely impacting health. We aimed to robustly quantify a community severance metric in urban settings exemplified by its characterization in New York City (NYC). MethodsWe used geospatial location data and dimensionality reduction techniques to capture NYC community severance variation. We employed principal component pursuit, a pattern recognition algorithm, combined with factor analysis as a novel method to estimate the Community Severance Index. We used public data for the year 2019 at census block group (CBG) level on road infrastructure, road traffic activity, and pedestrian infrastructure. As a demonstrative application of the Community Severance Index, we investigated the association between community severance and traffic collisions, as a proxy for road safety, in 2019 in NYC at CBG level. ResultsOur data revealed one multidimensional factor related to community severance explaining 74% of the data variation. In adjusted analyses, traffic collisions in general, and specifically those involving pedestrians or cyclists, were nonlinearly associated with an increasing level of Community Severance Index in NYC. ConclusionWe developed a high spatial-resolution Community Severance Index for NYC using data available nationwide, making it feasible for replication in other cities across the United States. Our findings suggest that increases in the Community Severance Index across CBG may be linked to increases in traffic collisions in NYC. The Community Severance Index, which provides a novel traffic-related exposure, may be used to inform equitable urban policies that mitigate health risks and enhance well-being.

Party affiliation and the legacy of HOLC redlining maps

The Home Owners' Loan Corporation (HOLC) redlining maps were formalized in the 1930s to assess lending risks in certain neighborhoods based on their racial or ethnic composition and systematically denied services such as mortgage loans and insurance. Despite the abolishment in the 1960s, many consequences of redlining persist in many American cities today. In this paper, we combine historical HOLC redlining maps and modern-day voter registration data to examine the relationships between redlining and party affiliation in Miami-Dade County, Florida. We use probit regressions to estimate the likelihood of residents’ affiliation with different parties in relation to redlining, controlling for resident characteristics and census block group factors. We find that, after accounting for socio-demographics, residents in historically redlined neighborhoods are more likely to be Democratic voters and less likely to be Republican voters. Additionally, we show that certain resident characteristics contribute to their party affiliation. Specifically, female residents are more likely to be Democratic voters but less likely to be Republican voters. There is also a clear pattern in racial disparities regarding party affiliation. Black residents are more likely to be Democratic voters but less likely to be Republican voters when compared to White residents. Our results highlight future urban planning efforts to prioritize the integration of redlined and non-redlined neighborhoods to avoid political polarization and isolation induced by historical urban policies such as redlining.

Crossing the line: access to trauma care across state borders

ObjectiveThis study investigates the challenge posed by state borders by identifying the population, injury, and geographic scope of areas of the country where the closest trauma center is out-of-state, and...

Abstract TMP20: Mapping the Expansion of the Minnesota Stroke System of Care

Objectives: Stroke is the fifth leading cause of death in Minnesota (MN) and in the United States. The state of MN a has a unique geographical layout with many healthcare facilities located in rural areas. In 2013 legislation was passed authorizing the Minnesota Department of Health to designate hospitals as Acute Stroke Ready Hospitals, Primary Stroke Centers, and Comprehensive Stroke Centers. When a hospital joins the MN Stroke System they maintain vital infrastructure, expertise, and care processes to provide high quality stroke care in accordance with evidence-based guidelines. We used geographic information systems (GIS) methods to visualize the growth and expansion of the Minnesota Stroke System and quantify its reach to Minnesotans across the state over a 10-year period. Methods: Using the Network Analyst extension of ESRI ArcGIS Desktop 10.8.2, 30- and 60- minute drive time service areas were calculated around each designated hospital in 2013, 2017 and 2023. Census block groups were identified where the mean center of population fell within the 30-minute drive time areas and the populations of those block groups were summed. A percentage was calculated using the total Minnesota population as the denominator. Results: In 2013, only 66% of Minnesota’s population lived within 30 minutes of a designated stroke center and 80% lived within a 60-minute drive time. Stroke designated hospitals (19) were largely concentrated in metropolitan areas, leaving many rural Minnesotans without ready access to stroke centers. After the first cycle in 2017, or three-year period post statute, drive time status increased to 90% of the population living within 30 minutes of a designated stroke system hospital and 98% living within a 60- minute drive time. As of 2023, 94% of Minnesotans now live within a 30-minute drive of a designated Stroke System Hospital and 99% live within a 60-minute drive time. Conclusions: Stroke statute was the catalyst for improving stroke care access in MN. The voluntary system of care expanded rapidly from 19 initial hospitals to 119 over a 10-year period. This progress means more than 1.9 million additional Minnesotans now have better access to emergency stroke care.

Effects of the Sacramento Neighborhood Alcohol Prevention Project on rates of child abuse and neglect 7 years post-implementation (1999-2010).

Evaluations of alcohol environmental prevention efforts examine short-term effects of these interventions on alcohol-related problems. We examine whether the effects of the Sacramento Neighborhood Alcohol Prevention Project (SNAPP), an alcohol environmental intervention aimed to reduce alcohol-related problems in two neighbourhoods, on child abuse and neglect remained 7 years post-implementation. SNAPP used a quasi-experimental non-equivalent control group design, where intervention activities occurred in the South area, followed by those in the North area 2 years later. Our sample size is 3912 space-time units (326 census block groups × 12 years [1999-2010]). Outcomes were measured at the household level and included: (i) all foster care entries total; and (ii) the subset of foster care entries that were alcohol related. Data were analysed using Bayesian conditionally autoregressive space-time models. We find that the decreases in total (relative rate [RR] = 0.882, 95% credible interval [CrI] 0.795, 0.980) and alcohol-related (RR = 0.888, 95% CrI 0.791, 0.997) foster care entries remain in the North intervention area although the magnitude of those changes are smaller than immediately post-intervention. Increases found in alcohol-related foster care entries in the South area immediately post-intervention were not significant 7 years later (RR = 1.128, 95% CrI 0.975, 1.307). Reductions in child abuse and neglect due to an alcohol environmental intervention can be maintained. Environmental interventions that provide community-level primary prevention strategies could be more easily sustained and more cost effective than individual-level interventions, although more research is needed to identify why interventions may be successful in specific contexts and not others.

AI-generated CT body composition biomarkers associated with increased mortality risk in socioeconomically disadvantaged individuals.

To evaluate the relationship between socioeconomic disadvantage using national area deprivation index (ADI) and CT-based body composition measures derived from fully automated artificial intelligence (AI) tools to identify body composition measures associated with increased risk for all-cause mortality and adverse cardiovascular events. Fully automated AI body composition tools quantifying abdominal aortic calcium, abdominal fat (visceral [VAT], visceral-to-subcutaneous ratio [VSR]), and muscle attenuation (muscle HU) were applied to non-contrast CT examinations in adults undergoing screening CT colonography (CTC). Patients were partitioned into 5 socioeconomic groups based on the national ADI rank at the census block group level. Pearson correlation analysis was performed to determine the association between national ADI and body composition measures. One-way analysis of variance was used to compare means across groups. Odds ratios (ORs) were generated using high-risk, high specificity (90% specificity) body composition thresholds with the most disadvantaged groups being compared to the least disadvantaged group (ADI < 20). 7785 asymptomatic adults (mean age, 57years; 4361:3424 F:M) underwent screening CTC from April 2004-December 2016. ADI rank data were available in 7644 patients. Median ADI was 31 (IQR 22-43). Aortic calcium, VAT, and VSR had positive correlation with ADI and muscle attenuation had a negative correlation with ADI (all p < .001). Compared with the least disadvantaged group, mean differences for the most disadvantaged group (ADI > 80) were: Aortic calcium (Agatston) = 567, VAT = 27 cm2, VSR = 0.1, and muscle HU = -6 HU (all p < .05). Compared with the least disadvantaged group, the most disadvantaged group had significantly higher odds of having high-risk body composition measures: Aortic calcium OR = 3.8, VAT OR = 2.5, VSR OR = 2.0, and muscle HU OR = 3.1(all p < .001). Fully automated CT body composition tools show that socioeconomic disadvantage is associated with high-risk body composition measures and can be used to identify individuals at increased risk for all-cause mortality and adverse cardiovascular events.

Long-term exposure to air pollution and risk of stroke by ecoregions: The REGARDS study

Several cohort studies have found associations between long-term exposure to air pollution and stroke risk. However, it is unclear whether the surrounding ecology may modify these associations. This study evaluates associations of air pollution with stroke risk by ecoregions, which are areas of similar type, quality, and quantity of environmental resources in the REasons for Geographic and Racial Differences in Stroke (REGARDS) study. We assessed the incidence of stroke in 26,792 participants (45+ yrs) from the REGARDS study, a prospective cohort recruited across the contiguous United States. One-yr and 3-yr means of PM2.5, PM10, O3, NO2, SO2, and CO were estimated at baseline using data from the Center for Air, Climate, & Energy Solution, and assigned to participants at the census block group level. Incident stroke was ascertained through September 30, 2020. Relations of air pollutants with the risk of incident stroke were estimated using Cox proportional hazards models, adjusting for relevant demographics, behavioral risk factors, and neighborhood urbanicity. Models were stratified by EPA designated ecoregions. A 5.4 μg/m3 (interquartile range) increase in 1-yr PM10 was associated with a hazard ratio (95 %CI) for incident stroke of 1.07 (1.003, 1.15) in the overall study population. We did not find evidence of positive associations for PM2.5, O3, NO2, SO2, and CO in the fully adjusted models. In our ecoregion-specific analysis, associations of PM2.5 with stroke were stronger in the Great Plains ecoregion (HR = 1.44) than other ecoregions, while associations for PM10 were strongest in the Eastern Temperate Forests region (HR = 1.15). The associations between long-term exposure to air pollution and risk of stroke varied by ecoregion. Our results suggests that the type, quality, and quantity of the surrounding ecology can modify the effects of air pollution on risk of stroke.

Did the association between alcohol outlet density and crime change during COVID shelter-in-place orders?

We investigated whether greater concentrations of on- and off-sale alcohol outlets were associated with crime and whether this association was moderated by COVID-19 shelter-in-place orders (SIP) that restricted on-premises consumption of alcohol. Crimes (2019-2020) and addresses of licenced alcohol outlets in a medium-sized California city were geocoded within census block groups (N = 61). On- and off-sale alcohol outlet density was calculated as licenced outlets/2.59 km2 (1 square mile). Multilevel negative binomial regression analyses were conducted to examine associations between alcohol outlet density and crime, and possible moderating effects of SIP, controlling for block group demographic characteristics and density of other retail businesses. On-sale outlet density was positively associated with total crimes and Part 2 crimes, while off-sale outlet density was inversely associated with total crime and Part 2 crimes. Overall, SIP was not significantly associated with crime, but moderated the associations of on-sale density with total crime and Part 1 crimes such that reductions in crime during SIP were observed in higher density areas. The association of off-sale outlets with crime was not moderated by SIP policies. On-sale outlet density, but not off-sale density, appears to be associated with increased crime. The results further indicate that restrictions in hours and service imposed by SIP policies reduced crime in high on-sale outlet density areas. These findings reinforce the importance of regulating alcohol outlet density and hours of service, especially for on-sale outlets, as a crime reduction strategy.

Potential for climate equity: Integrating nocturnal drainage winds into heat mitigation strategies in Portland

Limited adaptation options and local-scale heat mitigation measures to counter elevated nighttime temperatures have increased the need for effective and equitable city-scale cooling measures. I proposed that cold and dense air from surrounding high landscapes, called drainage winds, could be a potential city-scale nocturnal heat mitigation measure. I further suggest that this measure can moderate spatial disparities in heat mitigation capacity due to the unequal distribution of urban trees in Portland, Oregon. First, I conducted KLAM_21 wind simulations using grid cells of 30 m with elevation and historical and recent land use types. Then, I used census block groups including the proportion of tree canopy area, socioeconomic variables, and simulated wind attributes in statistical analysis. I found: (1) KLAM_21 simulation results generally align with actual wind observations, (2) urban development over the last two decades lowered the cold air layer, and (3) marginalized neighborhoods with less tree canopy coverage experience greater nocturnal cooling effects from drainage winds. These results emphasize the potential of drainage winds as an equitable heat mitigation option in Portland. Urban planners should integrate scenario-based simulated wind data into heat mitigation strategies to foster a more equitable and comfortable nighttime thermal environment in the long term.

Understanding the bias of mobile location data across spatial scales and over time: A comprehensive analysis of SafeGraph data in the United States.

Mobile location data has emerged as a valuable data source for studying human mobility patterns in various contexts, including virus spreading, urban planning, and hazard evacuation. However, these data are often anonymized overviews derived from a panel of traced mobile devices, and the representativeness of these panels is not well documented. Without a clear understanding of the data representativeness, the interpretations of research based on mobile location data may be questionable. This article presents a comprehensive examination of the potential biases associated with mobile location data using SafeGraph Patterns data in the United States as a case study. The research rigorously scrutinizes and documents the bias from multiple dimensions, including spatial, temporal, urbanization, demographic, and socioeconomic, over a five-year period from 2018 to 2022 across diverse geographic levels, including state, county, census tract, and census block group. Our analysis of the SafeGraph Patterns dataset revealed an average sampling rate of 7.5% with notable temporal dynamics, geographic disparities, and urban-rural differences. The number of sampled devices was strongly correlated with the census population at the county level over the five years for both urban (r > 0.97) and rural counties (r > 0.91), but less so at the census tract and block group levels. We observed minor sampling biases among groups such as gender, age, and moderate-income, with biases typically ranging from -0.05 to +0.05. However, minority groups such as Hispanic populations, low-income households, and individuals with low levels of education generally exhibited higher levels of underrepresentation bias that varied over space, time, urbanization, and across geographic levels. These findings provide important insights for future studies that utilize SafeGraph data or other mobile location datasets, highlighting the need to thoroughly evaluate the spatiotemporal dynamics of the bias across spatial scales when employing such data sources.

Climate risk and commercial mortgage delinquency

AbstractNatural disasters such as hurricanes, floods, heatwaves, and wildfires are projected to become more prevalent in the foreseeable future. Climate risk is, therefore, increasingly recognized as an important factor by policy makers, the investment community, and financial markets. Due to the immobility of assets, the commercial real estate industry is especially vulnerable to climate risk, and there is an increasing interest to understand the impact of climate risk on the value of commercial real estate. For commercial real estate lenders, changes in collateral value are only of partial importance. The ability of borrowers to meet their payment obligations is equally, if not more important. By combining historical data on two major climate‐related disasters—Hurricanes Harvey and Sandy—with longitudinal information on commercial mortgage performance, this paper identifies the impact of climate risks on mortgage delinquency rates for commercial real estate mortgages. The results show that both Harvey and Sandy led to elevated levels of commercial mortgage delinquency, with significant heterogeneity based on the extent of damage in the Census block group. Information provided through FEMA 100‐year floodplain maps partially mitigates the effects, an indication that lenders incorporate flood risk information in the underwriting process. An analysis of potential mechanisms indicates a decrease in property income during the 2‐year period following the event for Hurricane Harvey, but no evidence of income effects for Sandy.

There goes the neighborhood? The public safety enhancing effects of a mobile harm reduction intervention

BackgroundBuprenorphine is a gold-standard treatment for opioid use disorders, but most people with these disorders do not access it. Barriers to treatment access may be diminished by low-threshold mobile treatment programs but concern regarding their impact on local public safety challenges their adoption. MethodsThis quasi-experimental study uses difference-in-differences analyses to measure the impact of four mobile buprenorphine clinics in Pittsburgh on neighborhood arrest rates. The study period spans 2018 to 2022, with a pre-intervention period of 11 to 12 quarters and a post-intervention period of 7 to 8 quarters (dependent on neighborhood). A treatment group of 84 census block groups in the areas surrounding clinics during the time period after their establishment were compared to a control group of city census blocks not within one mile of a clinic plus treated block groups in the two years prior to clinic establishment. Outcome variables include drug, non-drug, and total arrests, measured quarterly per 100 in population. ResultsCompared to block groups further than 1 mile from a clinic, arrests fell by 34.13 % (b = −0.358, 95 % CI = −0.557, −0.158), drug arrests by 33.85 % (b = −0.087, 95 % CI = −0.151, −0.023), and non-drug related arrests by 22.29 % (b = −0.179, 95 % CI = −0.302, −0.057). Drug arrests declined significantly on days when the clinics were not present (b = −0.015, 95 % CI = −0.025, −0.006), with no significant change on clinic operational days (b = −0.002, 95 % CI = −0.016, −0.013). Total arrests declined significantly on days when clinics were and were not present (b = −0.045, 95 % CI = −0.078, −0.012; and b = −0.052, CI = −0.082, −0.023, respectively). ConclusionsMobile clinics providing medication for opioid use disorders were associated with reduced neighborhood arrest rates. Expansion of mobile services could promote health equity and public safety.

Modeling the relationship between urban tree canopy, landscape heterogeneity, and land surface temperature: A machine learning approach

Cities across the United States and around the globe are embracing urban greening as a strategy for mitigating the effects of rising temperatures on human health and quality-of-life. Better understanding how the spatial configuration of tree canopy influences land surface temperature should help to increase the positive impacts of urban greening. This study applies a machine learning approach for modeling the relationship between urban tree canopy, landscape heterogeneity, and land surface temperature (LST) using data from nine cities located in nine different climate zones of the United States. We collected summer LST data from the U.S. Geological Survey (USGS) Analysis Ready Data series and processed them to derive mean, minimum, and maximum LST in degrees Fahrenheit for each Census block group within the cities considered. We also calculated the percentage of each block group comprised by the land cover designations in the 2016 or 2019 National Land Cover Database (NLCD) maintained by the USGS, depending on the vintage of the available LST data. High resolution tree canopy data were purchased for all the study cities and the spatial configuration of tree canopy was measured at the block group level using established landscape metrics. Landscape metrics of the waterbodies were also calculated to incorporate the cooling effects of waterbodies. We used a Generalized Boosted Regression Model (GBM) algorithm to predict LST from the collected data. Our results show that tree canopy exerts a consistent and significant influence on predicted land surface temperatures across all study cities, but that the configuration of tree canopy and water patches matters more in some locations than in others. The findings underscore the importance of considering the local climate and existing landscape features when planning for urban greening.

Telehealth Expansion, Internet Speed, and Primary Care Access Before and During COVID-19

Primary care (PC) receipt is associated with better health outcomes. How telehealth expansion and internet speed are associated with PC use is unclear. To examine the association of telehealth and internet speed with PC use across sociodemographic determinants of health. This cohort study performed difference-in-differences regression of the change in in-person and telehealth PC visits between pre-COVID-19 public health emergency (PHE) (June 1, 2019, to February 29, 2020) and an initial (March 1, 2020, to May 31, 2020) and prolonged (March 1, 2020, to December 31, 2021) PHE period among continuously enrolled nonpregnant, nondisabled Wisconsin Medicaid beneficiaries aged 18 to 64 years. Data were analyzed from March 2022 to March 2023. PHE-induced telehealth expansion. Change in PC telehealth (using Current Procedural Terminology codes) visits: (1) count; (2) visit share completed by telehealth; (3) percentage of PHE-induced visit decline offset by telehealth. High-speed internet (HSI) defined as living in a census block group with a median block maximum download speed of 940 megabits per second or greater (June 2020 Federal Communications Commission broadband data); other census block groups classified as low-speed internet (LSI). In the total cohort of 172 387 participants, 102 989 (59.7%) were female, 103 848 (60.2%) were non-Hispanic White, 34 258 (19.9%) were non-Hispanic Black, 15 020 (8.7%) were Hispanic, 104 239 (60.5%) were aged 26 to 45 years, and 112 355 (66.0%) lived in urban counties. A total of 142 433 (82.6%) had access to HSI; 72 524 (42.1%) had a chronic condition. There was a mean (SD) of 0.138 (0.261) pre-PHE PC visits per month. In the pre-PHE period, visit rates were significantly higher for female than male participants, non-Hispanic White than non-Hispanic Black individuals, urban than rural residents, those with HSI than LSI, and patients with chronic disease than patients without. In the initial PHE period, female participants had a greater increase in telehealth visits than male participants (43.1%; 95% CI, 37.02%-49.18%; P < .001), share (2.20 percentage point difference [PPD]; 95% CI, 1.06-3.33 PPD; P < .001) and offset (6.81 PPD; 95% CI, 3.74-9.87 PPD; P < .001). Non-Hispanic Black participants had a greater increase in share than non-Hispanic White participants (5.44 PPD; 95% CI, 4.07-6.81 PPD; P < .001) and offset (15.22 PPD; 95% CI, 10.69-19.75 PPD; P < .001). Hispanic participants had a greater increase in telehealth visits than Non-Hispanic White participants (35.60%; 95% CI, 25.55%-45.64%; P < .001), share (8.50 PPD; 95% CI, 6.75-10.26 PPD; P < .001) and offset (12.93 PPD; 95% CI, 6.25-19.60 PPD; P < .001). Urban participants had a greater increase in telehealth visits than rural participants (63.87%; 95% CI, 52.62%-75.11%; P < .001), share (9.13 PPD; 95% CI, 7.84-10.42 PPD; P < .001), and offset (13.31 PPD; 95% CI; 9.62-16.99 PPD; P < .001). Participants with HSI had a greater increase in telehealth visits than those with LSI (55.23%; 95% CI, 42.26%-68.20%; P < .001), share (6.61 PPD; 95% CI, 5.00-8.23 PPD; P < .001), and offset (6.82 PPD; 95% CI, 2.15-11.49 PPD; P = .004). Participants with chronic disease had a greater increase in telehealth visits than those with none (188.07%; 95% CI, 175.27%-200.86%; P < .001), share (4.50 PPD; 95% CI, 3.58-5.42 PPD; P < .001), and offset (9.03 PPD; 95% CI, 6.01-12.04 PPD; P < .001). Prolonged PHE differences were similar. Differences persisted among those with HSI. In this cohort study of Wisconsin Medicaid beneficiaries, greater telehealth uptake occurred in groups with higher pre-PHE utilization, except for high uptake among Hispanic and non-Hispanic Black individuals despite low pre-PHE utilization. HSI did not moderate disparities. These findings suggest telehealth and HSI may boost PC receipt, but will generally not close utilization gaps.

A Comparison of Three Area-Level Indices of Neighborhood Deprivation and Socioeconomic Status and their Applicability to Breast Cancer Mortality.

Neighborhood deprivation indices are widely used in research, but the performance of these indices has rarely been directly compared in the same analysis. We examined the Area Deprivation Index, Neighborhood Deprivation Index, and Yost index, and compared their associations with breast cancer mortality. Indices were constructed for Georgia census block groups using 2011-2015 American Community Survey data. Pearson correlation coefficients and percent agreement were calculated. Associations between each index and breast cancer mortality were estimated among 36,795 women diagnosed with breast cancer using Cox proportional hazards regression. The indices were strongly correlated (absolute value of correlation coefficients > 0.77), exhibited moderate (41.4%) agreement, and were similarly associated with a 36% increase in breast cancer mortality. The similar associations with breast cancer mortality suggest the indices measure the same underlying construct, despite only moderate agreement. By understanding their correlations, agreement, and associations with health outcomes, researchers can choose the most appropriate index for analysis.

Local exposure misclassification in national models: relationships with urban infrastructure and demographics.

National-scale linear regression-based modeling may mischaracterize localized patterns, including hyperlocal peaks and neighborhood- to regional-scale gradients. For studies focused on within-city differences, this mischaracterization poses a risk of exposure misclassification, affecting epidemiological and environmental justice conclusions. Characterize the difference between intraurban pollution patterns predicted by national-scale land use regression modeling and observation-based estimates within a localized domain and examine the relationship between that difference and urban infrastructure and demographics. We compare highly resolved (0.01 km2) observations of NO2 mixing ratio and ultrafine particle (UFP) count obtained via mobile monitoring with national model predictions in thirteen neighborhoods in the San Francisco Bay Area. Grid cell-level divergence between modeled and observed concentrations is termed "localized difference." We use a flexible machine learning modeling technique, Bayesian Additive Regression Trees, to investigate potentially nonlinear relationships between discrepancy between localized difference and known local emission sources as well as census block group racial/ethnic composition. We find that observed local pollution extremes are not represented by land use regression predictions and that observed UFP count significantly exceeds regression predictions. Machine learning models show significant nonlinear relationships among localized differences between predictions and observations and the density of several types of pollution-related infrastructure (roadways, commercial and industrial operations). In addition, localized difference was greater in areas with higher population density and a lower share of white non-Hispanic residents, indicating that exposure misclassification by national models differs among subpopulations. Comparing national-scale pollution predictions with hyperlocal observations in the San Francisco Bay Area, we find greater discrepancies near major roadways and food service locations and systematic underestimation of concentrations in neighborhoods with a lower share of non-Hispanic white residents. These findings carry implications for using national-scale models in intraurban epidemiological and environmental justice applications and establish the potential utility of supplementing large-scale estimates with publicly available urban infrastructure and pollution source information.