Physical and social neighborhood disorder in Latin American cities: a scoping review

Abstract: Neighborhood disorder is an important aspect that may influence the health of residents in urban areas. The aims of this study were to map and systematize methods for measuring physical and social neighborhood disorder in studies conducted in Latin American cities. By means of a scoping review, articles published from 2000 in English, Spanish, and Portuguese with the following descriptors were mapped: neighborhood, physical disorder, and social disorder. Searches were conducted in MEDLINE (PubMed), LILACS (Virtual Health Library), Scopus, Web of Science, and Cochrane Library. Information on authorship, year, study type, locality, data source, target population, outcome, dominion, indicator, method, geographic unit, and unit of analysis was extracted. Variables from the disorder-related studies were extracted and grouped by similarity of content and themes. A total of 22 articles were identified, all published between 2012 and 2022, the majority in Brazil (n = 16). The perception of the individual was the most used method. The most frequent theme addressed in the physical disorder dominion was public streets (n = 20) and security (n = 15), in the social disorder dominion. A lack of consensus in the literature regarding variables used to measure physical and social neighborhood disorder in Latin American cities was detected. In addition to the need for standardization of the theme, studies to verify the sustainability of proposed measurement methods relevant to dynamically classify and compare urban neighborhoods and health impacts based on levels of exposure to physical and social disorder, are recommended.


Introduction
Urbanization is a global trend that is characterized as a dynamic process with differentiated patterns in each region of the world.Currently, 55% of the world population resides in urban areas.It is estimated that this figure will increase to 68% by 2050, with most of the growth occurring in low-and middle-income countries 1 .
Considered the most urbanized region in the world, Latin America has about 80% of residents in urban areas, which is a higher proportion than high-income countries 2 .This accelerated urbanization process has resulted in insufficient infrastructure, environmental deterioration, the formation of subnormal settlements and, most importantly, the region present the greatest socioeconomic inequality in the world 3 .

Information sources
Searches were conducted in December 2022 in MEDLINE (via PubMed), LILACS (via Virtual Health Library), Scopus (via CAPES Portal), Web of Science (via CAPES Portal), and Cochrane Library (via CAPES Portal) databases.
The references of the selected articles were checked to find additional studies not identified in the previous searches, considering the previously established eligibility criteria.

Search strategy
The search strategy was developed considering the inclusion criteria for the MEDLINE database (via PubMed), using the Health Science Descriptors (DeCS) and keywords: ("neighborhood") AND ("physical disorder" OR "social disorder") (Supplementary Material: https://cadernos.ensp.fiocruz.br/static//arquivo/suppl-en-00038423_3467.pdf).This search strategy was adapted according to the specificities of each database used.In all of them, the search was performed considering December 15, 2022 as the publication deadline.
The final search results were exported to Mendeley (https://www.mendeley.com),so that duplicate articles were removed.

Selecting sources of evidence
Article titles and abstracts were initially examined by one reviewer.The articles then selected were read in full by two independent reviewers, who identified whether the articles met the inclusion criteria.Disagreements were discussed with a third reviewer to reach consensus.

Data collection process
A structured instrument was designed to extract and synthesize the main elements found in each selected article, and Microsoft Excel (https://products.office.com/) was used for data tabulation.

Extracted information
Data extracted included authorship, year of publication, type of study (psychometric analysis; application of the method; association), locality, data source (on-site audit; secondary data; interview), target population (adolescents; youths; adults; older adults).Moreover, information such as outcome (when applicable), dominion of disorder (physical; social), type of indicator (simple; composite), method (demographic census; systematic social observation (SSO); perception of the individual; other), geographic unit of data collection of the disorder variables (street segment; census tract; neighborhood) and unit of analysis of disorder (individual; context), were systematized after reading the full articles (Box 1).The original variables of the physical and social disorder indicator were distributed into categories (Box 2).

Box 1
Key characteristics of the studies included in the scoping review (n = 22).

Summary of results
First, for each article included in this review, the dominion of disorder in the neighborhood, physical and/or social, defined according to each author, was identified.Only the studies by Höfelmann et al. 18,30 named the physical disorder dominion as physical neighborhood problems.
Next, the original variables used in the articles to measure disorder were extracted and grouped into categories by content similarity.For example, the variable "poorly lit streets" 13,31 was included in the category street lighting and "unsafe walking after dark in the neighborhood" 18,30 in the category walking after dark.
A total of 95 original variables were identified, 51 being physical disorder and 44 social disorder, which in turn were grouped into 41 categories.Finally, the variables were regrouped into themes: environmental factors, aesthetics, real estate and public facilities, neighborhood problems, security, and public streets.The environmental factors theme includes variables related to noise, odor, and pollution.Aesthetics includes variables that indicate whether a place is pleasant and the presence of trees.The real estate and public facilities theme includes items that characterize types of buildings, graffiti on buildings and public facilities.The neighborhood problems theme includes variables that are considered nuisances experienced by the residents, such as the presence of points of drug sales and consumers of alcohol and drugs.The security theme includes violent situations and presence of criminals.And in the public streets theme are the items that characterize problems in streets and sidewalks, such as holes, lack of public lighting, garbage, and others.The themes that presented the highest number of categories were public streets (n = 15 categories) for physical disorder, and security (n = 8 categories) and neighborhood problems (n = 7 categories) for social disorder (Boxes 2 and 3).

Box 3
Disorder-related characteristics of the studies included in the scoping review (n = 22).

Results
The search strategy identified 971 articles, of which 518 were excluded as they were duplicates.The titles and abstracts of the remaining 453 articles were read, and 428 were excluded for not meeting the inclusion criteria.Then, the 25 articles were read in full, and out of these, seven were excluded for not describing the disorder indicator (n = 6) and for not having as an objective to evaluate the disorder (n = 1).After checking the references of the selected articles, additional 78 articles were evaluated using the title and abstract.Out of these 74 articles were excluded for not meeting the inclusion criteria.Then, the remaining four articles were read in full, and all were included.In the end, 22 articles comprised the present review (Figure 1).Of the 22 articles included, 18 were association studies, three were psychometric analyses, and only one was an application of the method.In Brazil, the articles were developed in Belo Horizonte (Minas Gerais State) (n = 9), Florianópolis (Santa Catarina State) (n = 3), Rio de Janeiro (n = 1), Vespasiano (Minas Gerais State) (n = 1) and set of cities representative of the country (n = 2).The remainder was conducted in Bogotá, Colombia (n = 1), Mexico City, Mexico (n = 1), Santiago, Chile (n = 1) and other Latin American cities (n = 3).The publication period was from 2012 to 2022 (Box 1).
The data collection sources were: interview (n = 14), on-site audit (n = 4) and secondary data (n = 4).As for target population, the studies were conducted with adults (n = 12), older adults (n = 4), youths (n = 2) and adolescents (n = 2).The health outcomes of the included studies were quite varied, with self-rated health being the most frequent (n = 4) (Box 1).
Among the included articles, six assessed only the physical disorder dominion, four only the social disorder, and 12 assessed both physical and social disorder dominions.Regarding the disorder indicator, most articles presented composite indicators (n = 17) (Box 3).
Flowchart for article selection.
The main methods used to measure disorder were individual perception (n = 14), SSO (n = 4), secondary data from each country's demographic census information (n = 3) and one of the articles also used satellite images and administrative data.In addition, one of the studies employed a crowding source of information (Box 3).
The geographic units of data collection for the disorder variables were neighborhood (n = 14), census tract (n = 4) and street segments (n = 4); and the units of analysis for physical and social disorder were context (n = 14) and individual (n = 8) (Box 3).
Cad. Saúde Pública 2023; 39(9):e00038423 Of the six themes defined, four were present in both dominions of disorder: environmental factors, real estate and public facilities, neighborhood problems, and public streets.The aesthetic theme (n = 7) was observed only for physical disorder and the security theme (n = 14) for social disorder.Public streets (n = 20) and real estate and public facilities (n = 12) were the most frequent in the physical disorder dominion, and security (n = 15) and neighborhood problems (n = 13) in the social disorder dominion (Box 4).Points of sale and use of drugs

Discussion
This scoping review identified 22 articles published between 2012 and 2022 that assessed physical and social disorder in Latin American cities.Most studies were conducted in Brazil and used the perception of the individual to measure disorder.Public streets and real estate and public facilities were the most frequent themes to measure physical disorder, while for social disorder the themes were security and neighborhood problems.
As the number of studies evaluating disorders increased, there was also an increase in the choice of variables used to describe them 11 .Some studies suggested that physical disorder and social disorder might overlap 32,33 .However, most authors advocated a distinction between these components 34,35,36,37 .For example, drug outlets and drug use was reported in one of the included studies as physical disorder 13 .However, due to its behavioral nature, it was more described as social disorder 18,19,22,30,31,38,39,40,41,42 .Similarly, the presence of garbage, which was reported as social disorder in some studies 16,20,22,43 , was more often considered as physical disorder because it is a characteristic of the physical environment 12,13,15,18,19,21,30,31,38,39,40,41,42,44,45,46 .We also observed that the presence of abandoned or deteriorated property and equipment was used to measure social disorder 20,43 , whereas in most of the studies evaluated, it was considered as physical disorder 13,15,16,19,38,39,40,42,45 .Therefore, it was found that there is no consensus in the literature about the distribution of the variables for the evaluation of physical and social disorder in Latin America, which becomes a challenge for the systematization of research and comparison among studies.
Importantly, most of the studies evaluated used the individual's perception to obtain information of the disorder 13,15,17,18,19,20,30,31,38,39,40,41,42,43 .This measurement method has been frequently employed, usually in population studies, using simple and direct questions that make it possible to aggregate the responses and construct variables that characterize the perceived disorder, allowing the assessment of constructs that cannot be measured by other methods 39 , such as social disorder variables.However, individuals may respond differently based on their own behavior, thus resulting in common source bias.In other words, participants may be biased due to the stigma associated with low-income neighborhoods, being more likely to evaluate them with higher levels of disorder 47,48,49 .
Cad. Saúde Pública 2023; 39(9):e00038423 It should also be considered that perception may be associated with individual characteristics, such as sex, age, and length of residence 39 .The studies included in this review agree that the lack of objective measures of the environment is a limitation, as they may not always correlate with perceived measures 17,18,19,20,39,41 .
The SSO, among the selected studies, was performed only in Brazil 21,44,45,46 .This method allows recording, in a valid and reliable way, the physical attributes of the neighborhood, measuring characteristics that are not captured by census information, by other macro indicators, nor by the individuals' perception 50 .Moreover, it is a reproducible method that can work with other research techniques and survey quantitative and qualitative data in the same investigation.It also has the advantage of being the best option for areas that are difficult to access, where remote sensing or pre-collected images are not available 45,46 .The way in which SSO can be applied can also vary, as was observed in the study by Remigio et al. 46 , who developed a mobile app for systematic data collection in a large subnormal settlement in Rio de Janeiro.On the other hand, studies report that the use of SSO may have been limiting, as certain disorder items are subject to temporal variation.Thus, a more reliable measurement would require more than one observation, on different days and times, for the same street segment, as well as having more complex field logistics, which would result in high costs and extended data collection periods 21,44,45 .
Recently, there has been a growth in the development and use of new methods to assess neighborhood attributes through emerging technologies 49 .Among them is the virtual audit through Google Street View (https://www.google.com/maps), a digital alternative of SSO, which usually has a lower cost and less complex logistics.Some studies report that many variables of on-site audit can be assessed from remote imagery with reliability comparable to in-person assessment 51,52,53,54 .Fry et al. 55 evaluated the availability of Google Street View images in 371 Latin American cities and observed that localities with better socioeconomic conditions tended to have more consistent images.It is worth mentioning that none of the articles selected in this review performed the virtual audit, which, in turn, has been used in previous studies related to the food environment 56,57 and physical activity 58 .
The use of secondary data was also observed among the selected studies 12,16,22,59 .Population census measures, for example, besides having many variables, cover several municipalities in countries that perform them, as was observed in Brazil and Colombia 12,59 .However, they are collected only in certain periods and are not necessarily current.The geographic area is based on administrative units that may not represent social or geographic boundaries.Moreover, it usually contains limited variables about economic and structural factors, ignoring the social processes in the neighborhood 47,48 .In the study by Auler et al. 12 , conducted in three Brazilian capitals, the collection of neighborhood characteristics was carried out in person by the supervisors of the 2010 Demographic Census, representing a highlight in this set of information.
Among the selected studies, the geographic unit of analysis and data collection regarding disorder was mostly concentrated at the context level 12,13,16,18,21,22,30,39,40,41,44,45,46,59 , mainly in neighborhoods 13,18,22,30,39,40 .This geographical unit is a territory that can be subjectively or objectively delimited, where people live and interact socially, and is a measure of the macro-scale of the environment resulting from the aggregation of individual data or a smaller scale, which may reflect the characteristics of the context.For the assessment of larger areas, geographic information system-based measures are employed and consist of the set of tools for obtaining, storing, analyzing, and representing spatial data 5,38 .On the other hand, the micro-scale is differentiated by describing the urban configuration, in terms of presence and quality of infrastructure, such as items measured at the street level (e.g., sidewalks and trees).Individuals' perception and SSO constitute adequate methods for assessing smaller areas, and information obtained at the micro-scale can contribute to interventions with greater potential for change and lower costs 45,58 .
From this scoping review, it was also possible to establish recommendations for future research on neighborhood disorder.For reviews using a systematic process, we recommend the use of automated tools, such as text mining, which enables automatic extraction of concepts and keywords, allowing reviews to be completed more quickly, as well as minimizing the impact of publication bias and reducing the chances of losing relevant research (recommendation 1) 60 .Standardization of variables that compose the construct is also needed, since physical disorder relates to characteristics of the context (e.g., environmental factors, aesthetics, real estate and public facilities, and public streets) and social Cad.Saúde Pública 2023; 39(9):e00038423 disorder relates to aspects of interaction between people and the context (e.g., neighborhood problems and security) (recommendations 2-4).It should be noted that methods that use objective measures are better suited to assess physical disorder, while those that use subjective measures are better suited to assess social disorder.As mentioned earlier, no studies were selected that used virtual auditing.Thus, we suggest the use of new methods to measure disorder, such as virtual audits via Google Street View, which is an efficient alternative to on-site audits and is safer for the auditors, performed in less time and with less financial resources.Moreover, it allows covering more study sites, such as large or distant areas, as well as acquiring historical images for longitudinal studies and application in computer vision models (recommendation 5) 55,61 (Box 5).
The results obtained through this review made it possible to observe the advancement in studies on environmental disorder.However, there is still no consensus on the items that measure physical and social neighborhood disorder in Latin American cities, which indicates the need for method standardization and future studies that evaluate the psychometric properties of the disorder constructs, as well as greater sophistication in the analytical approaches used.We consider as fundamental systematic review studies, meta-analysis and new evaluative studies that verify the continuity, systematization and implementation of new methods of measurement and analysis in urban health to assess neighborhood disorder in a continuous and longitudinal way in Latin American countries, since environmental disorder is an important construct for understanding the relationships between physical and socioeconomic neighborhood conditions and health outcomes.This review had the limitation of using only the scientific literature, not including the gray literature.Also, the search strategy did not address the different terms used to describe disorder, such as neighborhood disturbances and problems, nor the terms used to describe methods of measuring disorder, which merits consideration in future work.The strengths of the study include the use of PRISMA-ScR guidelines to ensure a robust and replicable process and originality, as to our knowledge it is the first review on this theme in the Latin American context.

Conclusion
This review revealed that the most commonly used method to measure neighborhood disorder in Latin America is the perception of the urban environment.Most studies examined adults and assessed both disorders, generally with composite indicators using scales.Moreover, the item most evaluated for physical disorder was related to the characteristics of public streets, while for social disorder, it was those related to security.The need to standardize the variables used to measure disorder, considering physical and social peculiarities separately can be seen from the findings.Furthermore, mixed methods of measurement are relevant to broaden the understanding of the phenomenon.Combining perception, systematic observation, and other methods will allow for capturing urban aspects that affect citizens' health more accurately in future studies.= 20), enquanto no domínio social foi segurança (n = 15).Detectou-se ausência de consenso na literatura quanto às variáveis utilizadas para aferir a desordem física e social da vizinhança em cidades da América Latina.Além da necessidade de padronização do tema, recomendam-se estudos que verifiquem a sustentabilidade de métodos de mensuração propostos, relevantes para classificar e comparar, de forma dinâmica, vizinhanças urbanas e os impactos na saúde com base nos níveis de exposição à desordem física e social.

*
The study adapted the on-site audit strategies using smartphones to apply systematic social observation; ** Construct validation studies and internal consistency analysis; *** The outcome of the study was the cumulative homicide rate calculated from the sum of the homicides in the years 2003, 2004, and 2005 divided by the average population size in the three years and then multiplied by 10,000.

*
The simple indicator shows that the disorder variables were assessed separately, while the composite indicator shows that the disorder variables were grouped and presented in scales; ** Collection of the surrounding characteristics for this study was conducted by the 2010 Demographic Census supervisors; *** Aggregated into a geographic unit (e.g., census tract, municipality); # The source of mass collaboration information used in this study was Mexico City's 911 calling system, which receives emergency and non-emergency requests by phone and message, allowing direct request for services or crime reports;## The studies used the term physical problems to refer to physical disorder; ### The study used two indicators as a proxy for disorder obtained from the 2005 Demographic Census.
Classification of the original neighborhood disorder variables according to domain, theme and category.