Introduction
A Flexible Multilevel Data Structure
Identifying and Operationalizing Cities
SALURBAL Approach to Identifying and Operationalizing Cities
Level | Definition |
---|---|
Level 1 “city” | |
L1Admin (administrative) | “City” defined as a single administrative unit (e.g., municipio) or combination of adjacent administrative units (e.g., several municipios) that are part of the urban extent as determined from satellite imagery. Each L1Admin is defined based on its component level 2 units. |
L1Metro (metropolitan areas) | “City” defined following the exact definition that each country provides for metropolitan areas (if available), as a combination of either level 2 units or other units. |
L1UrbExt (urban extent) | “City” defined based on systematically identified urban extent based on built area; boundaries may not overlap exactly with administrative units. |
L1Excess (urban extent spillover) | “City” defined as in L1UrbExt but also including the urban extent spilling into a neighboring non-SALURBAL country. |
Level 2 “sub-city” | Administrative units (e.g., municipios) nested within L1Admin. In some cases, this may be a single unit for each city, and in other cases, it will be multiple units. In some cases, level 2 units may also be nested within L1Metro. |
Level 3 “neighborhood” | Smaller units such as census tracts that can be used as proxies for “neighborhoods” within a city. Level 3 units will be nested within level 2 units. They will also be approximately linked to L1UrbExt so that census data can be linked to the L1UrbExt for analyses. In some cases, level 3 units may also be nested within L1Metro. |
Country | Cities | Level 2 unit | Level 3 unitb |
---|---|---|---|
Argentina | 33 | Departamento/Partido/Comunaa | Radio Censal |
Brazil | 152 | Municipios | Setor Censitário |
Chile | 21 | Comuna | Zona Censal |
Colombia | 35 | Municipio | Sector Urbano |
Costa Rica | 1 | Canton | Unidad Geoestadistica Basica |
El Salvador | 3 | Municipio | Sector Censal |
Guatemala | 3 | Municipio | Sector Censal |
Mexico | 92 | Area Geoestadistica Municipal | Area Geoestadistica Basica |
Nicaragua | 5 | Municipio | Sector Censal |
Panama | 3 | Corregimiento | Barrio |
Peru | 23 | Distrito | Zona Censal |
L1Admin | L2 | L3 | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
N
| Population (in 1000s) | Total Na | Units per L1Admin | Population (in 1000s) Median (5th–95th percentile) | Total Na | Units per L2 | Population Median (5th–95th percentile)g | ||||
Median (5th–95th percentile) | Max | Median (5th–95th percentile) | Max | Median (5th–95th percentile) | Max | ||||||
AR | 33 | 304.2 (123.0–1466.5) | 14,791.1 | 110 | 1 (1–6) | 51 | 188.8 (28.8–605.7) | 29,792 | 218 (34–606) | 1493 | 883 (250–1692) |
BR | 152 | 231.4 (114.5–3070.4) | 19,987.8 | 422 | 1 (1–9) | 31 | 124.2 (15.6–798.2) | 164,107 | 183 (23–1074) | 18,953 | 646 (79–1251) |
CL | 21 | 215.3 (126.8–994.6) | 6213.8 | 81 | 1 (1–10) | 36 | 137.9 (27.6–319.2) | 3,918b | 39 (14–114) | 172 | 2200 (0, 7787) |
CO | 35 | 360.3 (119.9–2822.2) | 8546.8 | 84 | 1 (1–6) | 15 | 115.0 (12.8–895.6) | 4,679c | 22.5 (2–170) | 643 |
h
|
CR | 1 | 2367.0 (2367.0-2367.0) | 2367.0 | 29 | 29 (29–29) | 29 | 57.4 (22.7–251.6) |
f
| |||
SV | 3 | 261.9 (241.2–1704.8) | 1870.8 | 22 | 1 (1–18) | 20 | 79.5 (9.3–267.3) | 944 | 28.5 (4–108) | 137 | 2361 (1425-3031) |
GT | 3 | 242.0 (150.3–2633.0) | 2898.7 | 20 | 5 (1–13) | 14 | 94.1 (22.8–516.4) | 4025 | 86 (21–688) | 1485 | 677 (312–1106) |
MX | 92 | 351.7 (134.9–1855.9) | 20,014.5 | 406 | 2 (−15) | 76 | 67.6 (7.1–774.5) | 32,921d | 32 (4–319) | 638 | 1749 (6–5636) |
NI | 5 | 174.1 (117.6–936.0) | 1120.4 | 11 | 1 (1–5) | 6 | 76.8 (20.3–555.8) |
f
| |||
PA | 3 | 212.0 (209.4–1591.8) | 1745.1 | 82 | 18 (12–50) | 53 | 20.1 (2.2–66.2) | 1,800e | 18 (1–53) | 147 | 116 (3–1150) |
PE | 23 | 281.5 (127.7–876.8) | 9177.7 | 169 | 5 (2–18) | 51 | 55.9 (4.9–340.0) |
f
|
Linking Health and Environmental Data at Various Geographic Levels
Obtaining and Harmonizing Health Data
Mortality Data
Population Data
Survey Data
Domain | Variables | Definitions | Sourcea |
---|---|---|---|
Demographics | Age | Age in years | N/A |
Sex | Male or female | ||
Education | Education level as less than primary, primary completed, secondary completed, or more than secondary completed | ||
Diabetes | Diabetes | Presence of diabetes diagnosis by a health care provider among all adults (excluding diagnoses during pregnancy) | CDC [50] |
Gestational diabetes | Presence of gestational diabetes diagnosis among all adult female respondents with a history of pregnancy | ||
Diabetes treatment | Any pharmacological treatment among those with diabetes | ||
Hypertension | Hypertension | Presence of hypertension diagnosis by a health care provider among all adults (excluding a diagnosis during pregnancy) | CDC [53] WHO and NCD RisC [54] WHL [55] |
Gestational hypertension | Presence of gestational hypertension diagnosis among all adult female respondents with a history of pregnancy | ||
Hypertension treatment | Any pharmacological treatment among those with hypertension | ||
Systolic blood pressure (SBP) | Average of 2–4 SBP measured by survey interviewer | ||
Diastolic blood pressure (DBP) | Average of 2–4 DBP measured by survey interviewer | ||
Health status | General health status | Respondent’s self-rated health categorized as very poor to very good or excellent | OECD [56] CDC-BRFSS [57] |
Tobacco use | Cigarette smoking status | Cigarette smoking status as current, former, or never smoker among adults | CDC [58] GTSS [59] |
Alcohol use | Binge drinking | Varied by country: defined as 3 or 4 or 5 alcoholic drinks for women and 4 or 5 alcoholic drinks for men in the past 30 days on one occasion | CDC [60] WHO [61] |
Current drinking (30 days) | Any consumption of alcoholic beverages in the past 30 days | ||
Current drinking (12 months) | Any consumption of alcoholic beverages in the past 12 months | ||
Anthropometrics | Height (measured) | Measured | WHO [62] |
Weight (measured) | Measured | ||
Height (self-reported) | Reported by respondent | ||
Weight (self-reported) | Reported by respondent | ||
Body mass index (BMI based self-reported or measured height and weight) | Reported by respondent or measured | ||
Physical activity | Global physical activity | Total minutes of self-reported physical activity in the past week | IPAQ [63] GPAQ [64] |
Transportation physical activity | Total minutes of self-reported transportation-related physical activity in the past week | ||
Leisure physical activity | Total minutes of self-reported leisure physical activity in the past week | ||
Total walking | Total minutes of self-reported walking in the past week | ||
Nutrition | Fruit consumption frequency | Number of days per week in the last week | WHO [65] IARC [66] CDC [67] |
Vegetable consumption frequency | Number of days per week in the last week | ||
Soda consumption | Number of days per week in the last week | ||
Dessert foods consumption | Number of days per week in the last week |
Characterizing Urban Social and Physical Environments
Domain | Indicator | Definition | Level | Data source(s) |
---|---|---|---|---|
Economic | ||||
Poverty, income, and inequality | Poverty | Proportion of population living below the nationally defined income-based poverty level | L1–L3 | Census or national household surveys |
Income-based Gini Index | A measure of inequality in the distribution of income | L1 | Census or national household surveys | |
Employment | Unemployment | Proportion of persons 15 years or older in the labor force who are not working but seeking employment | L1–L3 | Census or national household surveys |
Labor force participation | Proportion of persons 15 years or older who are working or seeking employment | L1–L3 | Census or national household surveys | |
Social | ||||
Education | 15–17 years old in school | Proportion of 15–17 year-olds enrolled in school | L1–L3 | Census |
Adults with completed secondary education or more | Proportion of people 25 years and older with completed secondary education or higher | L1–L3 | Census | |
Education-based Gini Index | A measure of inequality in the distribution of education | L1 | Census | |
Gender empowerment | Female labor force participation | Proportion women 15 years or older who are working or seeking employment | L1–L3 | Census or National household surveys |
Female government leadership | Proportion of city leadership (e.g., city council members) who are female | L1 | National government sources | |
Violence and disorder | Violent deaths | Age-standardized homicide rate per 100,000 population of homicides | L1–L2 | Mortality |
Crime/safety | Proportion of individuals reporting being a victim of a crime in the past 12 months Safety perception score | L1–L2 | Selected national surveys, CAF Survey [68] | |
Social disorder | Social disorder/incivilities scale | L1–L2 | CAF Survey | |
Social cohesion and social capital | Election participation | Proportion of eligible individuals voting in the last presidential election | L1–L2 | CAF Survey |
Community organization membership | Proportion of individuals who are part of a community or neighborhood organization. | L1–L2 | CAF Survey | |
Neighborhood connectedness | Neighborhood connectivity scale/social support scale | L1–L2 | CAF Survey | |
Discrimination | Proportion of individuals reporting discrimination | L1–L2 | CAF Survey | |
Housing | ||||
Water connection | Proportion of households without piped water | L1–L3 | Census | |
Sewage connection | Proportion of households lacking a connection to the municipal sewer system or a septic tank | L1–L3 | Census | |
Overcrowding | Proportion of households with 3 people per room or more | L1–L3 | Census | |
Housing materials | Proportion of households with non-durable wall materials | L1–L3 | Census | |
Governmental, institutional, and organizational | ||||
Governance | Presence of participatory budgeting | L1 | Selected national sources | |
Property taxes: total revenue and as % of GDP and total tax revenue | L1/L2 | Lincoln Land Institute | ||
Social services and health care | Percent of population with health insurance | L1 | Selected country surveys | |
Percent of children with age-appropriate vaccine coverage | L1 | Selected country surveys | ||
Percent of households in poverty receiving public assistance | L1 | Selected country surveys |
Domain | Definition | Indicators | Level | Data source |
---|---|---|---|---|
Urban form and population metrics | ||||
Population | Measure of the number of people living per unit of an area or within a geographic boundary | Total population, population density, Gini coefficient of the population distributions | L1–L2 | Census or population projectionsa |
Population distribution | Measure of concentration population within geographic boundary | Gini coefficient of population distribution | L2–L3 | WorldPopb [69] |
Neighborhood centrality | Measure of the distance to the city center | Neighborhood centrality | L2–L3 | Local sources |
Urban landscape metrics | ||||
Area | Measure of the urbanized area inside a geographic boundary | Total urban area, percentage of urban area, coefficient of variation of urban patchb area, area-weighted mean urban patch area, mean urban patch area, effective mesh size | L1–L3 | |
Shape | Measure of compactness and complexity | Area-weighted mean shape index | ||
Fragmentation | Measure of fragmentation of urban expansion. It is the relative share of open space in the urban landscape | Number of patches, patch density, mean patch size, effective mesh size | ||
Isolation | Measure of the tendency for patches to be relatively clustered or isolated in space. It is the mean distance to the nearest urban patch within the geographic boundary | Area-weighted mean euclidean nearest neighbor distance | ||
Edge | Measure of fragmentation and shape complexity. It is the boundary between urban and non-urban patches | Edge density, area-weighted edge density | ||
Aggregation | Measure of the tendency of clumping of urban patches | Aggregation index | ||
Street design and connectivity metrics | ||||
Street density | Measure of street network density | Street density, large road density | L1–L3 | OpenStreetMap and OSMNx [70] |
Intersection density | Measure of the amount of intersections within the street network | Intersection density, intersection density 3-way, intersection density 4-way, streets per node average, streets per node standard deviation | ||
Street network length and structure | Measure of street network structure | Street length average, circuity average | ||
Transportation metrics | ||||
Bus rapid transit | Bus-based transit system that includes dedicated lanes, traffic signal priority, off-board fare collection, elevated platforms, and enhanced stations | Presence of BRT, BRT length, BRT daily users, BRT price per ride, BRT supply length, BRT demand, BRT payment capacity | L1–L3 | BRTData, OpenStreetMap, minimum wage of Latin America and local sources |
Subway, light rail, and/or elevated train (SLRET) transport systems | Mass rapid transit, including heavy rail, metro or subway | Presence of SLRET, SLRET length, SLRET daily users, SLRET price per ride, SLRET supply length, SLRET demand, SLRET payment capacity | OpenStreetMap and local sources | |
Aerial Tram transport system | Transport lift systems integrated into the city’s public transport network that provide mobility options for those living in hillside neighborhoods | Presence of aerial tram, aerial tram length | OpenStreetMap and local sources | |
Bicycle facilities | Public infrastructure for exclusive or shared use of bicycles | Total length of bike lanes, bike lane km per population, presence of Open Streets program and length of Open Streets programs | OpenStreetMap, CAF data, and local sources | |
Urban travel delay index | Measure of congestion | Measures the increase in travel times due to congestion in the street network | L2 | OpenStreetMap and Google Maps Distance Matrix API |
Gasoline price | Adjusted gasoline price | Price per gallon adjusted by minimum wage | L1 | Local sources |
Air pollution and green space metrics | ||||
Parks and green space | Measures of parks or green space availability | Parks area, parks density | L1–L3 | Local sources |
PM10, NOx, SO4, O3 | Annual mean value by existing monitoring station | Annual average in μg/m3 | L1–L3 | Local sourcesd |
PM2.5 | Annual mean value from satellite measurements | Annual average in μg/m3 | L1–L3 | |
Food environment | ||||
Density of chain supermarkets | Large food stores with availability of processed foods, frozen foods and fresh produce | Number of supermarkets /area | L1–L3 | Online searches of chain company websites |
Density of chain convenience stores | Stores with long opening hours and high availability of ultra-processed foods | Number of convenience stores/area | L1–L3 | Online searches of chain company websites |
A Typology of Multilevel Urban Health Questions
Question | Analytical approach and unit of analysis | Example |
---|---|---|
Between-city differences | ||
How much do summary health indicators vary across cities (within and between countries) and what factors are associated with this variability? | Multilevel analysis of city-level outcomes nested within countries (including variables at L1 and at the country level) | Does life expectancy vary across cities? Are these differences associated with city size and recent growth? |
How much does individual-level health vary across cities and what factors are related to this variability? | Multilevel analysis of individual-level survey outcomes nested within cities and countries (including variables at the individual level, at L1, and at the country level) | Does the probability of having diabetes vary across cities? How do individual-level factors, city, and country characteristics contribute to these differences? |
Within-city differences | ||
Description of small area variations in summary health within large cities and factors associated with this variability | Small area estimation methods for mortality or survey estimates and their association with neighborhood (L3) characteristics | How much does life expectancy vary within a city? Is this related to area-level poverty? |
How much does individual-level health vary across neighborhoods within cities and what factors are related to this variability? | Multilevel analysis of individual-level survey outcomes nested within neighborhoods (L3) and cities (L2 or L1), including variables at the individual-level, and at L3, L2, and L1 as appropriate | How do neighborhood features of the built environment associate with differences in physical activity levels? Do city-level factors (such as street connectivity) modify these associations? |
Impact of city context on inequities | Multilevel analysis of city-level outcomes stratified by education nested within countries (including variables at L1 and at the country level) or multilevel models for aggregate data Multilevel analysis of survey respondents nested within cities, including variables at the individual level, city level, and country level | Do mortality differences by education vary across cities? What city-level factors are associated with greater or smaller inequities? Do educational differences in diabetes prevalence vary across cities? Are city-level factors associated with smaller or larger inequities? |
Changes over time | ||
What longitudinal trends in summary health indicators are observed and to what extent do city or country characteristics modify these trends? | Longitudinal analyses of summary city-level health outcomes and their association with time invariant and time-varying city and country characteristics | How has life expectancy changed over time in cities? Are city growth and air pollution levels related to these trends? |
Are changes over time in city or neighborhood characteristics related to changes in individual-level health outcomes? | Longitudinal analyses of individual-level survey responses nested within neighborhoods and cities and their relation to L1, L2, or L3 time-varying characteristics | Do changes in a city’s urban landscape and in neighborhood crime levels affect changes in BMI? |