Development and validation of statewide survey-based measures of livability in Connecticut.

Livability, or how a place and its systems (e.g., housing, transportation) supports the ability to lead a livable life, is a determinant of health. There is a lack of standard, validated measures to assess livability in the US. This study employed factor analytic methods to create measures of livability in Connecticut using data from the DataHaven Community Wellbeing Survey (DCWS) (n = 32,262). Results identified a 3-factor model (safety, opportunity, and infrastructure) as the best fit, explaining 69% of the variance in survey items. Newly created livability measures had high internal consistency, in addition to high convergent validity with other area-level measures.

Area-Level Social Vulnerability and Severe COVID-19: A Case-Control Study Using Electronic Health Records from Multiple Health Systems in the Southeastern Pennsylvania Region.

Knowledge about neighborhood characteristics that predict disease burden can be used to guide equity-based public health interventions or targeted social services. We used a case-control design to examine the association between area-level social vulnerability and severe COVID-19 using electronic health records (EHR) from a regional health information hub in the greater Philadelphia region. Severe COVID-19 cases (n = 15,464 unique patients) were defined as those with an inpatient admission and a diagnosis of COVID-19 in 2020. Controls (n = 78,600; 5:1 control-case ratio) were a random sample of individuals who did not have a COVID-19 diagnosis from the same geographic area. Retrospective data on comorbidities and demographic variables were extracted from EHR and linked to area-level social vulnerability index (SVI) data using ZIP codes. Models adjusted for different sets of covariates showed incidence rate ratios (IRR) ranging from 1.15 (95% CI, 1.13-1.17) in the model adjusted for individual-level age, sex, and marital status to 1.09 (95% CI, 1.08-1.11) in the fully adjusted model, which included individual-level comorbidities and race/ethnicity. The fully adjusted model indicates that a 10% higher area-level SVI was associated with a 9% higher risk of severe COVID-19. Individuals in neighborhoods with high social vulnerability were more likely to have severe COVID-19 after accounting for comorbidities and demographic characteristics. Our findings support initiatives incorporating neighborhood-level social determinants of health when planning interventions and allocating resources to mitigate epidemic respiratory diseases, including other coronavirus or influenza viruses.

Spatial Heterogeneity in Fatal Overdose Rate Trends in Mexican Cities: 2005-2021.

Objectives. To describe national and city-level fatal drug overdose trends between 2005 and 2021 in Mexico. Methods. We calculated fatal overdose rates at the city level in 3-year periods from 2005 to 2021 and annually at the national level for people aged 15 to 64 years in Mexico. We calculated rate differences and rate ratios for each city between periods. Results. The national fatal overdose rate was 0.53 overdose deaths per 100 000 population and was almost twice as high in urban than in nonurban areas. The national fatal overdose rate was stable over the period 2005 to 2014 and increased monotonically to a peak in 2021. Fatal overdose rates varied across cities. Cities with the 8 highest fatal overdose rates in the period were all in states along the US-Mexico border. Conclusions. Fatal overdoses have doubled over the past 15 years in Mexico. Overdose rates are particularly high and increasing in cities close to the US-Mexico border. Public Health Implications. There is a need for enhanced overdose surveillance data and coordinated harm reduction strategies, particularly in the northern border region of Mexico. (Am J Public Health. 2024;114(7):705-713. https://doi.org/10.2105/AJPH.2024.307650).

Association Between Household Deprivation and Living in Informal Settlements and Incidence of Diarrhea in Children Under 5 in Eleven Latin American Cities.

Diarrhea is a leading cause of death in children globally, mostly due to inadequate sanitary conditions and overcrowding. Poor housing quality and lack of tenure security that characterize informal settlements are key underlying contributors to these risk factors for childhood diarrhea deaths. The objective of this study is to better understand the physical attributes of informal settlement households in Latin American cities that are associated with childhood diarrhea. We used data from a household survey (Encuesta CAF) conducted by the Corporación Andina de Fomento (CAF), using responses from sampled individuals in eleven cities. We created a household deprivation score based on household water and sewage infrastructure, overcrowding, flooring and wall material, and security of tenure. We fitted a multivariable logistic regression model to estimate odds ratios (OR) and 95% confidence intervals (95% CI) to test the association between the deprivation score and its individual components and childhood diarrhea during the prior 2 weeks. We included a total of 4732 households with children, out of which 12.2% had diarrhea in the 2-week period prior to completing the survey. After adjusting for respondent age, gender, and city, we found a higher risk of diarrhea associated with higher household deprivation scores. Specifically, we found that the odds of diarrhea for children living in a mild and severe deprived household were 1.04 (95% CI 0.84-1.28) and 3.19 times (95% CI 1.80-5.63) higher, respectively, in comparison to households with no deprivation. These results highlight the connections between childhood health and deprived living conditions common in informal settlements.

Modifying effect of urban parks on socioeconomic inequalities in diabetes prevalence: a cross-sectional population study of Madrid City, Spain.

BACKGROUND: Evidence has shown contradicting results on how the density of urban green spaces may reduce socioeconomic inequalities in type 2 diabetes (equigenic hypothesis). The aim of this study is to test whether socioeconomic inequalities in diabetes prevalence are modified by park density. METHODS: We designed a population-wide cross-sectional study of all adults registered in the primary healthcare centres in the city of Madrid, Spain (n=1 305 050). We obtained georeferenced individual-level data from the Primary Care Electronic Health Records, and census-tract level data on socioeconomic status (SES) and park density. We modelled diabetes prevalence using robust Poisson regression models adjusted by age, country of origin, population density and including an interaction term with park density, stratified by gender. We used this model to estimate the Relative Index of Inequality (RII) at different park density levels. FINDINGS: We found an overall RII of 2.90 (95% CI 2.78 to 3.02) and 4.50 (95% CI 4.28 to 4.74) in men and women, respectively, meaning that the prevalence of diabetes was three to four and a half times higher in low SES compared with high SES areas. These inequalities were wider in areas with higher park density for both men and women, with a significant interaction only for women (p=0.008). INTERPRETATION: We found an inverse association between SES and diabetes prevalence in both men and women, with wider inequalities in areas with more parks. Future works should study the mechanisms of these findings, to facilitate the understanding of contextual factors that may mitigate diabetes inequalities.

Measuring Variation in Infant Mortality and Deaths of Despair by U.S. Congressional Districts in Pennsylvania: A Methodological Case Study.

Many ecological studies examine health outcomes and disparities using administrative boundaries such as census tracts, counties, or states. These boundaries help us to understand the patterning of health by place along with impacts of policies implemented at these levels. However, additional geo-political units, or units with both geographic and political meaning, such as congressional districts, present further opportunities to connect research with public policy. We provide a step-by-step guide in how to conduct disparities-focused analysis at the congressional district level, and as an applied case study we use geocoded vital statistics data from 2010-2015 to examine levels and disparities of infant mortality (IM) and deaths of despair (DoD) in the 19 U.S. congressional districts of Pennsylvania for the 111th-112th (2009-2012) Congresses, and 18 districts for the 113th-114th (2013-2016) Congresses. We also provide recommendations for extending congressional district level analysis to other outcomes, states, and geopolitical boundaries such as state legislative districts. Increased surveillance of health outcomes at the congressional district level can help prompt policy action, advocacy, and hopefully, reduce rates and disparities in health.

Investigating the measurement properties of livability: a scoping review.

Connecting evidence-based public health recommendations to livability, a popular and relatable construct, can increase the policy relevance of research to improve community design. However, there are many different definitions and conceptualizations of livability and little consensus about its measurement. Improved measurement, including standardization, is needed to increase understanding of livability's influence on health and to facilitate comparisons across contexts. This study sought to review existing livability measures, how they were created, and evidence regarding their reliability and validity. A scoping review of three databases (PubMed, Google Scholar, and Web of Science) identified 744 eligible studies. After screening, 24 studies, 15 from the original search and 9 through backward citation searches, were included in the review. Most studies were carried out in an urban context. There was minimal consensus across studies on the conceptualization of livability. However, measure domains and indicators overlapped significantly. While the process used to validate the measures varied, most studies reported high levels of reliability and found that livability was correlated with similar measures (e.g. place satisfaction, neighborhood safety, and sense of place) and self-reported health and wellbeing. Further research is needed to develop parsimonious, standardized measures of livability in order to create and sustain livable communities worldwide.

Ambient nitrogen dioxide in 47 187 neighbourhoods across 326 cities in eight Latin American countries: population exposures and associations with urban features.

BACKGROUND: Health research on ambient nitrogen dioxide (NO2) is sparse in Latin America, despite the high prevalence of NO2-associated respiratory diseases in the region. This study describes within-city distributions of ambient NO2 concentrations at high spatial resolution and urban characteristics associated with neighbourhood ambient NO2 in 326 Latin American cities. METHODS: We aggregated estimates of annual surface NO2 at 1 km2 spatial resolution for 2019, population counts, and urban characteristics compiled by the SALURBAL project to the neighbourhood level (ie, census tracts). We described the percentage of the urban population living with ambient NO2 concentrations exceeding WHO air quality guidelines. We used multilevel models to describe associations of neighbourhood ambient NO2 concentrations with population and urban characteristics at the neighbourhood and city levels. FINDINGS: We examined 47 187 neighbourhoods in 326 cities from eight Latin American countries. Of the roughly 236 million urban residents observed, 85% lived in neighbourhoods with ambient annual NO2 above WHO guidelines. In adjusted models, higher neighbourhood-level educational attainment, closer proximity to the city centre, and lower neighbourhood-level greenness were associated with higher ambient NO2. At the city level, higher vehicle congestion, population size, and population density were associated with higher ambient NO2. INTERPRETATION: Almost nine out of every ten residents of Latin American cities live with ambient NO2 concentrations above WHO guidelines. Increasing neighbourhood greenness and reducing reliance on fossil fuel-powered vehicles warrant further attention as potential actionable urban environmental interventions to reduce population exposure to ambient NO2. FUNDING: Wellcome Trust, National Institutes of Health, Cotswold Foundation.

A scaling investigation of urban form features in Latin America cities.

This paper examines scaling behaviors of urban landscape and street design metrics with respect to city population in Latin America. We used data from the SALURBAL project, which has compiled and harmonized data on health, social, and built environment for 371 Latin American cities above 100,000 inhabitants. These metrics included total urbanized area, effective mesh size, area in km2 and number of streets. We obtained scaling relations by regressing log(metric) on log (city population). The results show an overall sub-linear scaling behavior of most variables, indicating a relatively lower value of each variable in larger cities. We also explored the potential influence of colonization on the current built environment, by analyzing cities colonized by Portuguese (Brazilian cities) or Spaniards (Other cities in Latin America) separately. We found that the scaling behaviors are similar for both sets of cities.

Social inequalities and COVID-19 mortality between neighborhoods of Bariloche city, Argentina.

BACKGROUND: The COVID-19 pandemic has shown how intraurban inequalities are likely to reinforce health and social inequalities. Studies at small area level help to visualize social inequialities hidden in large areas as cities or regions. AIM: To describe the spatial patterning of COVID-19 death rates in neighborhoods of the medium-sized city of Bariloche, Argentina, and to explore its relationship with the socioeconomic characteristics of neighborhoods. METHODS: We conducted an ecological study in Bariloche, Argentina. The outcome was counts of COVID-19 deaths between June 2020 and May 2022 obtained from the surveillance system and georeferenced to neighborhoods. We estimated crude- and age-adjusted death rates by neighborhood using a Bayesian approach through a Poisson regression that accounts for spatial-autocorrelation via Conditional Autoregressive (CAR) structure. We also analyzed associations of age-adjusted death rates with area-level socioeconomic indicators. RESULTS: Median COVID-19 death rate across neighborhoods was 17.9 (10th/90th percentile of 6.3/35.2) per 10,000 inhabitants. We found lower age-adjusted rates in the city core and western part of the city. The age-adjusted death rate in the most deprived areas was almost double than in the least deprived areas, with an education-related relative index of inequality (RII) of 2.14 (95% CI 1.55 to 2.96). CONCLUSION: We found spatial heterogeneity and intraurban variability in age-adjusted COVID-19 death rates, with a clear social gradient, and a higher burden in already deprived areas. This highlights the importance of studying inequalities in health outcomes across small areas to inform placed-based interventions.

Does Urban Greenspace Reduce Mortality Inequalities Based on Poverty, Race, or Both in Philadelphia, PA?

While past research suggests that urban greenspace is associated with weaker income-based mortality inequities, little is known about associations with racial inequities, which may be distinct owing to historical and contemporary forms of racism. We quantified the extent to which different measures of greenspace modified socioeconomic and racial/ethnic inequities in all-cause and cardiovascular disease mortality. For every residential census tract in Philadelphia, PA (N = 376), we linked counts of all-cause and cardiovascular mortality (years 2008-2015) with measures of greenspace (proportion tree canopy or grass/shrub cover, proportion residents reporting park access, and the normalized difference vegetation index measure of overall greenness) and American Community Survey-based measures of sociodemographic composition (proportion of residents living in poverty, proportion identifying as non-Hispanic Black, and the index of concentration at the extremes (ICE) representing racialized economic deprivation). We used age- and sex-adjusted negative binomial models, with the natural logarithm of age-specific population counts as an offset, to quantify the magnitude of inequities by each composition variable, overall and stratified by categories of each greenspace measure. Inequities in mortality were weaker among neighborhoods with higher proportion grass/shrub cover or overall greenness. The most substantially narrowed inequities were those by the ICE. Mortality inequities did not differ substantially by perceived park access, and tree canopy was associated with weaker ICE-based inequities only. In this ecologic analysis, neighborhood greenspace was associated with weaker mortality inequities. However, associations varied across greenspace type and sociodemographic composition metrics, with generally stronger associations with overall greenness and grass/shrub coverage, and for ICE-basedinequities.

Small area vulnerability, household food insecurity and child malnutrition in Medellin, Colombia: results from a repeated cross-sectional study.

Background: Malnutrition and food insecurity might be driven not only by individual factors but also by contextual conditions, such as area-level deprivation or vulnerability. This study aimed to analyze the association between area-level vulnerability and i) household food insecurity and ii) malnutrition in children in Medellin, Colombia, during the years 2017 and 2018. Methods: We obtained data from two different sources: the Living Standards Measurement Survey (LSMS) and the nutrition surveillance system of Medellin. The main outcomes were food insecurity in households with children and anthropometric indicators for children under five. The main predictor was area-level vulnerability. Mixed effects Poisson regression with robust standard errors models were conducted to test the association of quintiles of deprivation with each outcome. Findings: Households with children living in areas with the highest deprivation had 1.9 times the prevalence of food insecurity as compared to those living in areas with the lowest deprivation (PR 1.91, 95% CI 1.42-2.57). Similar results were observed for underweight/risk of underweight (PR 1.26, 95% CI 1.11-1.42), stunting/risk of stunting (PR 1.36, 95% CI 1.22-1.53) and stunting (PR 1.93 95% CI 1.55-2.39) among children under five. We found no consistent associations with wasting/risk of wasting or excess weight/risk of overweight across quintiles of deprivation. Interpretation: This study sheds light on the role of area-level vulnerability on malnutrition in children in Medellin, Colombia, showing a pattern of increasing prevalence of food insecurity, underweight and stunting by quintile of deprivation. Funding: Swiss School of Public Health (SSPH+) and Centre for Global Health Inequalities Research (CHAIN).

How Can We Measure Alcohol Outlet Density Around Schools? A Comparison Between Two Buffer-Based Methods.

Measuring the density of alcohol outlets around schools is a critical step towards understanding the drivers of drinking among adolescents. Different methodologies have been used in the literature for this purpose, but the implications of using one methodology or another have not been clearly assessed. Our aim was to compare different methods to measure alcohol outlet density and highlight under which characteristics of the environment might be best using each approach. We used Geographic Information Systems to geolocate schools (n = 576) and alcohol outlets (n = 21,732) in Madrid. We defined the density of alcohol outlets as the number of establishments within an area of 400 m around schools measured using two buffering methods: crow flies' and street network distances. We evaluated the agreement between both methods visually and through regression models, including street connectivity, population density, and density of recreational venues as predictors of disagreement. The density of alcohol outlets around schools was higher using crow flies' distances compared to street network distances. The differences between methodologies were wider in areas of higher density of outlets, especially in the downtown areas, where there are higher population density and street connectivity. Our results suggest that the spatial characteristics and morphology of the study area (e.g., street connectivity and population density) should be considered when deciding the methodology to be used to measure alcohol outlet density. Future studies should explore the implications of different exposure measures in their association with drinking prevalence and consumption patterns among different geographical contexts.

Spatial Inequities in Life Expectancy in Small Areas of Buenos Aires, Argentina 2015-2017.

Studies of life expectancy (LE) in small areas of cities are relatively common in high-income countries but rare in Latin American countries. Small-area estimation methods can help to describe and quantify inequities in LE between neighborhoods and their predictors. Our objective was to analyze the distribution and spatial patterning of LE across small areas of Ciudad Autónoma de Buenos Aires (CABA), Argentina, and its association with socioeconomic characteristics. As part of the SALURBAL project, we used georeferenced death certificates in 2015-2017 for CABA, Argentina. We used a spatial Bayesian Poisson model using the TOPALS method to estimate age- and sex-specific mortality rates. We used life tables to estimate LE at birth. We obtained data on neighborhood socioeconomic characteristics from the 2010 census and analyzed their associations. LE at birth was higher for women (median of across neighborhoods = 81.1 years) compared to men (76.7 years). We found a gap in LE of 9.3 (women) and 14.9 years (men) between areas with the highest and the lowest LE. Better socioeconomic characteristics were associated with higher LE. For example, mean differences in LE at birth in areas with highest versus lowest values of composite SES index were 2.79 years (95% CI: 2.30 to 3.28) in women and 5.61 years (95% CI: 4.98 to 6.24) in men. We found large spatial inequities in LE across neighborhoods of a large city in Latin America, highlighting the importance of place-based policies to address this gap.

Ambient nitrogen dioxide in 47,187 neighborhoods across 326 cities in eight Latin American countries: population exposures and associations with urban features.

Background: Health research on ambient nitrogen dioxide (NO2) is sparse in Latin America, despite the high prevalence of NO2-associated respiratory diseases in the region. This study describes within-city distributions of ambient NO2 concentrations at high spatial resolution and urban characteristics associated with neighborhood ambient NO2 in 326 Latin American cities. Methods: We aggregated estimates of annual surface NO2 at 1 km2 spatial resolution for 2019, population counts, and urban characteristics compiled by the SALURBAL project to the neighborhood level (i.e., census tracts). We described the percent of the urban population living with ambient NO2 levels exceeding WHO Air Quality Guidelines. We used multilevel models to describe associations of neighborhood ambient NO2 concentrations with population and urban characteristics at the neighborhood and city levels. Findings: We examined 47,187 neighborhoods in 326 cities from eight Latin American countries. Of the ≈236 million urban residents observed, 85% lived in neighborhoods with ambient annual NO2 above WHO guidelines. In adjusted models, higher neighborhood-level educational attainment, closer proximity to the city center, and lower neighborhood-level greenness were associated with higher ambient NO2. At the city level, higher vehicle congestion, population size, and population density were associated with higher ambient NO2. Interpretation: Almost nine out of every 10 residents of Latin American cities live with ambient NO2 concentrations above WHO guidelines. Increasing neighborhood greenness and reducing reliance on fossil fuel-powered vehicles warrant further attention as potential actionable urban environmental interventions to reduce population exposure to ambient NO2. Funding: Wellcome Trust, National Institutes of Health, Cotswold Foundation.

Trends and determinants of social inequities in cardiovascular risk factors in Spain: a mixed-methods study.

This sequential mixed-methods study aims to: 1) assess spatial and temporal trends in cardiovascular risk factors by socioeconomic position from 2001 to 2020 in Spain; 2) explore public health professionals' perspectives regarding interventions that might have impacted these inequities; and 3) analyze determinants on social inequities in cardiovascular risk factors. First, we will measure the change in absolute and relative social inequities in eight cardiovascular risk factors through time trend analysis using repeated cross-sectional data from both National and European Health Surveys for Spain from 2001 to 2020. Second, we will interview key informants -both at the regional and national level-, to contextualize data obtained in phase 1 and capture the content and variation of policies across regions. Third, we will use econometric methods to analyze how these identified interventions have impacted these social inequities within and across regions.

Exercise facility availability and incidence of type 2 diabetes and complications in Spain: A population-based retrospective cohort 2015-2018.

BACKGROUND: To study the association between exercise facility availability and type 2 diabetes incidence and its complications, and to explore effect modification by socioeconomic status (SES) and sex in the Madrid adult population. METHODS: A multilevel longitudinal design, based on a population-based retrospective cohort including 1,214,281 residents of Madrid (Spain) aged 40-75 years from 2015 to 2018. Outcomes were type 2 diabetes incidence and macrovascular (cardiac ischemia and/or stroke) and microvascular (chronic kidney disease, retinopathy, and/or peripheral vascular disease) complications in those with diabetes at baseline. Exercise facility availability was defined as the count of exercise facilities in a 1000 m street network buffer around each participant's residence. Poisson regression models with robust standard errors were used to estimate the risk ratios (RR). Interactions were explored with SES tertiles and by sex. RESULTS: Residents living in areas with lower exercise facility availability showed higher risk of type 2 diabetes (RRtertile3vs1 = 1.25, CI95% 1.21-1.30) as well as macrovascular (RRTertile3vs1 = 1.09 CI95% 1.00-1.19), and microvascular (RRTertile3vs1 = 1.10 CI95% 1.01-1.19) complications. Associations were strongest in low SES areas for type 2 diabetes (RRtertile3vs1-LOW-SES = 1.22, CI95% 1.12-1.32; RRtertile3vs1-HIGH-SES = 0.91, CI95% 0.85-0.98) and microvascular complications (RRtertile3vs1-LOW-SES = 1.12, CI95% 0,94-1,33; RRtertile3vs1-HIGH-SES = 0.88, CI95% 0.73-1.05). CONCLUSIONS: Living in areas with lower availability of exercise facilities was associated with a greater risk of type 2 diabetes and its complications. Increasing exercise opportunities, particularly in low SES areas, could help reduce the social gradient of diabetes and its complications.

The impact of social violence on HIV risk for women in Colombia: A concurrent mixed methods study.

Gender, violence, and migration structurally impact health. The Venezuelan humanitarian crisis comprises the largest transnational migration in the history of the Americas. Colombia, a post-conflict country, is the primary recipient of Venezuelans. The Colombian context imposes high levels of violence on women across migration phases. There is little information on the relationship between violence and HIV risk in the region and how it impacts these groups. Evidence on how to approach the HIV response related to Venezuela's humanitarian crisis is lacking. Our study seeks to 1) understand how violence is associated with newly reported HIV/AIDS case rates for women in Colombian municipalities; and 2) describe how social violence impacts HIV risk, treatment, and prevention for Venezuelan migrant and refugee women undergoing transnational migration and resettlement in Colombia. We conducted a concurrent mixed-methods design. We used negative binomial models to explore associations between social violence proxied by Homicide Rates (HR) at the municipality level (n = 84). The also conducted 54 semi-structured interviews with Venezuelan migrant and refugee women and key informants in two Colombian cities to expand and describe contextual vulnerabilities to HIV risk, prevention and care related to violence. We found that newly reported HIV cases in women were 25% higher for every increase of 18 homicides per 100,000, after adjusting for covariates. Upon resettlement, participants cited armed actors' control, lack of government accountability, gender-based violence and stigmatization of HIV as sources of increased HIV risk for VMRW. These factors impose barriers to testing, treatment and care. Social violence in Colombian municipalities is associated with an increase in newly reported HIV/AIDS case rates in women. Violence hinders Venezuelan migrant and refugee women's access and engagement in available HIV prevention and treatment interventions.

Variation in youth and young adult homicide rates and their association with city characteristics in Latin America: the SALURBAL study.

Background: Latin America and the Caribbean (LAC) is one of the most urbanized and violent regions worldwide. Homicides in youth (15-24 years old, yo) and young adults (25-39yo) are an especially pressing public health problem. Yet there is little research on how city characteristics relate to homicide rates in youth and young adults. We aimed to describe homicide rates among youth and young adults, as well as their association with socioeconomic and built environment factors across 315 cities in eight LAC countries. Methods: This is an ecological study. We estimated homicide rates in youth and young adults for the period 2010-2016. We investigated associations of homicide rates with sub-city education and GDP, Gini, density, landscape isolation, population and population growth using sex-stratified negative binomial models with city and sub-city level random intercepts, and country-level fixed effects. Findings: The mean sub-city homicide rate per 100,000 in persons aged 15-24 was 76.9 (SD = 95.9) in male and 6.7 (SD = 8.5) in female, and in persons aged 25-39 was 69.4 (SD = 68.9) in male and 6.0 (SD = 6.7) in female. Rates were higher in Brazil, Colombia, Mexico and El Salvador than in Argentina, Chile, Panama and Peru. There was significant variation in rates across cities and sub-cities, even after accounting for the country. In fully adjusted models, higher sub-city education scores and higher city GDP were associated with a lower homicide rate among male and female (rate ratios (RR) per SD higher value in male and female, respectively, 0.87 (CI 0.84-0.90) and 0.90 (CI 0.86-0.93) for education and 0.87 (CI 0.81-0.92) and 0.92 (CI 0.87-0.97) for GDP). A higher city Gini index was associated with higher homicide rates (RR 1.28 (CI 1.10-1.48) and 1.21 (CI 1.07-1.36) in male and female, respectively). Greater isolation da was also associated with higher homicide rates (RR 1.13 (CI 1.07-1.21) and 1.07 (CI 1.02-1.12) in male and female, respectively). Interpretation: City and sub-city factors are associated with homicide rates. Improvements to education, social conditions and inequality and physical integration of cities may contribute to the reduction of homicides in the region. Funding: The Wellcome Trust [205177/Z/16/Z].