A multi-scalar perspective on health and urban housing: an umbrella review

With more than half the world’s population living in cities, understanding how the built environment impacts human health at different urban scales is crucial. To be able to shape cities for health, an understanding is needed of planetary health impacts, which encompass the human health impacts of human-caused disruptions on the Earth’s natural ecosystems. This umbrella review maps health evidence across the spatial scales of the built environment (building; neighbourhood; and wider system, including city, regional and planetary levels), with a specific focus on urban housing. Systematic reviews published in English between January 2011 and December 2020 were searched across 20 databases, with 1176 articles identified and 124 articles screened for inclusion. Findings suggests that most evidence reports on health determinants at the neighbourhood level, such as greenspace, physical and socio-economic conditions, transport infrastructure and access to local services. Physical health outcomes are also primarily reported, with an emerging interest in mental health outcomes. There is little evidence on planetary health outcomes and significant gaps in the research literature are identified. Based on these findings, three potential directions are identified for future research.


INTRODUCTION
The influence of the urban built environment on human health is complex and multifaceted. Health is determined by local environmental exposure and behaviours of individuals, which are socially and spatially patterned (Marmot 2005). Material aspects of the built environment give rise to various social processes that may directly and indirectly shape the health of its residents (Rydin et al. 2012;Bai et al. 2016). This poses difficulties to both conceptualising the relation between health and the urban built environment, and when seeking to design and implement actions to protect and improve health.
The urban built environment may be broadly defined at three scales: building, neighbourhood, and the wider urban system encompassing processes that operate across the city and regional levels and extend as far as the planetary level. Addressing health at each spatial scale involves different types of planning and stakeholders, as well as governance frameworks. Hence, understanding the confluence of different scales is necessary to improve knowledge and actions in local urban environments (Ramaswami et al. 2016). A multi-scalar perspective of the local built environment provides a comprehensive view of how health interventions may be implemented more efficiently at a particular scale. Multiple health objectives and outcomes, as well as health risks, may be considered simultaneously across the spectrum of policymaking and systematically reflected in decision-making. This requires a thorough understanding of the interrelations between the different scales of the built environment in which a policy intervenes, and this reaches beyond the local level and across social and political boundaries. Ultimately, local-level health considerations should be linked to the health of the wider urban system, including planetary health, whereby environmental processes can both be disturbed by and threaten to disturb local living conditions and health (Rydin et al. 2012;Ramaswami et al. 2016;Whitmee et al. 2015).
Urban housing is a primary area for health intervention in the built environment. Current urbanisation estimates suggest that 68% of world's population will live in cities by 2050. At the urban level, housing covers on average 70% of land-use, thus it is a substantial sector for policymaking (UN 2018). Traditionally, approaches to research supporting health-focused policy action in the housing sector have considered dwellings as atomised units of exposure (Thomson & Thomas 2015;Ige et al. 2019), although some recent work has begun to consider wider connections (Bird et al. 2018;Carlin et al. 2017;Núñez-González et al. 2020;Pérez et al. 2020). This paper aims to synthesise the current evidence on the built environment and health nexus, focusing on the urban housing sector as the entry point. The paper adopts the World Health Organization's (WHO) (2018) relational definition of housing and includes the physical structure of the dwelling, its immediate environment and the community (WHO 2018). Many systematic reviews already exist of the health impact of interventions and/or risks within the house or built environment. However, addressing the broader issues for policy implementation and decision-making across scales requires a broad synthesis of the evidence. Umbrella reviews are useful for efficiently summarising and comparing the evidence, assessing such considerations of aspects of health across built scales and identifying research gaps (Aromataris & Munn 2020). settings, vulnerable groups and communicable disease. There is limited discussion of implications for policymaking and economic evaluations of health interventions, i.e. the cost of intervention versus health outcomes. Urban health interventions have focused to date on treating the effects of health conditions; however, there is potential for constructive interventions in the built environment (at various scales) which improve health and/or reduce environmental health risks if the policy focus is on dealing with the causes of health conditions.

ELIGIBILITY CRITERIA
Housing as defined here (WHO 2018) included the dwelling and its immediate and wider surroundings, such as the neighbourhood services and facilities necessary for residential living, including green space, transportation, access to public services and facilities, and their connections to the city. Specific categories of housing such as student or elderly accommodation, residential care and shelters for the homeless were excluded. Systematic review articles published between January 2011 and December 2020 were eligible. Articles included meet the Cochrane Reviews criteria for systematic reviews: (1) clearly stated set of objectives with predefined eligibility criteria for included studies; (2) an explicit methodology; (3) a systematic search used to identify included studies; (4) an assessment of the validity of findings; and (5) a systematic presentation of results (Higgins et al. 2019).

DATA EXTRACTION
From the initial number of 2413 articles found, 1029 duplicates and 208 studies published prior to January 2011 were removed, leaving 1176 for manual screening of titles and abstracts by the authors (Figure 1). Articles were rated by two reviewers independently. A total of 264 articles were selected for full text review, of which 124 were agreed by the reviewers to meet the inclusion criteria and thus included in this umbrella review. The reasons for exclusion are summarised in the PRISMA diagram shown in Figure 1.

QUALITY ASSESSMENT
The quality of included studies was assessed using the JBI manual for evidence synthesis of umbrella reviews (Aromataris & Munn 2020). The following aspects were recorded: the topic or intervention of each paper, the context or geographical location, reported outcomes, study type (i.e. experimental or observational studies) and a critical assessment of how the review evidence was synthesised. Information was also collated on (1) the built scale at which the study was focused (building, neighbourhood, urban system); (2) any mentioned of planetary outcomes or links to planetary outcomes; and (3) stakeholders (i.e. public/private stakeholders at a local, city, regional or national level) and governance mechanisms (i.e. policy, regulation, legislation, standard or guidance).

DESCRIPTION OF STUDY QUALITY AND CHARACTERISTICS
The 124 articles primarily included research from high-income countries (predominantly the UK, Europe and North America); few included studies from low-income countries (Alaazi & Aganah 2020;Henson et al. 2020;Jung et al. 2017;Katoto et al. 2019;Quansah et al. 2017;Shuvo et al. 2020;Turley et al. 2013), and these were mainly in relation to urban slum interventions or impacts. The types of studies included in each review were diverse, but mainly collated evidence from cross-sectional population surveys. The majority of reviews assessed the quality of evidence in terms of potential publication bias and used criteria for appraising studies for inclusion. While the majority commented on the poor quality and potential bias of included papers, quality was not always stated. Few reported research from randomised controlled trials or quasi-experimental studies, and yet the quality of studies was generally reported by authors as medium quality. Study designs were generally broad, inclusive of qualitative and non-experimental evidence. The built environment was often broadly defined, although a few studies incorporated geographical information system (GIS) indicators when assessing outcomes at the neighbourhood or city scale (Gong et al. 2016;Ma et al. 2020;Malambo et al. 2016;McCrorie et al. 2014;Nordbø et al. 2018).
The majority of studies focused on the neighbourhood scale (n = 80), followed by the dwelling (n = 31) and urban system (n = 32) level. Some studies addressed multiple scales, but few identified these from a scale perspective (Levy-Storms et al. 2018). Most focused on non-communicable disease and physical health, especially at the neighbourhood level; mental health was mostly examined at the neighbourhood level. At the dwelling level, physical and mental health outcomes featured with a greater focus on cardiorespiratory outcomes related to indoor air pollution and other hazards. Many studies focused on the general population, but some were of more specific populations, including children (n = 20), older adults (n = 9), mothers and pregnant women (n = 3), the socially disadvantaged (n = 5) or people with a specific chronic disease (i.e. HIV, diabetes, asthma) (n = 4). A narrative synthesis of the findings follows below, structured under the three broad levels of housing at scale: building, neighbourhood and urban system. Figures 2-4 show the number of studies identified at each spatial level.      Turcu et al. Buildings and Cities DOI: 10.5334/bc.119

BUILDING DETERMINANTS OF HEALTH (31 STUDIES)
A total of 31 (25%) of the systematic reviews reported factors that affect health at the building or dwelling level under three broad areas: (1) indoor environmental quality (12 studies) such as air quality (e.g. indoor pollution from location and construction materials) and occupant hazards (e.g. burning fuel, cooking, heating); (2) dwelling condition (eight studies) such as home improvements (e.g. energy retrofits, bathroom/kitchens) and soundness (e.g. collapse and fire hazard); and (2) dwelling design (11 studies), including green buildings (e.g. green walls, green design, green standards), building type (e.g. high-rise, apartment, cohousing, shared facilities) and outdoor space (e.g. gardens) ( Table 1; and see Table S1 in the supplemental data online).

Indoor environmental quality
A total of 12 systematic reviews were included in this category looking at air quality (seven studies) and occupant hazards (five studies). The air quality subcategory included studies of outdoor air pollution in the vicinity of the dwelling (e.g. originating from such sources as a nearby traffic) (Vardoulakis et al. 2020), its infiltration into the dwelling, and the associated risks for the cardiovascular and respiratory health of the residents, including asthmatic and allergic exacerbations (Tagiyeva & Sheikh 2014;Van Boven et al. 2020). The extent to which outdoor air pollution enters the dwelling is related to the building structure, its fabric, operation and ventilation/air purification technologies (Van Boven et al. 2020). This subcategory also included studies of radon, lead, volatile and semi-volatile organic compounds (VOCs, SVOCs) (Ajrouche et al. 2017;Naldzhiev et al. 2020;Nussbaumer-Streit et al. 2020) as well as of biological agents (Sharpe et al. 2015). The occupant hazards subcategory looked at wood and biomass burning for heating and cooking as well as cookstove efficiency, which are important determinants of poor household air quality and impact on a wide array of health endpoints

Dwelling condition
Eight studies were grouped in this category under home improvements (six studies) and soundness (two studies). Of the six studies on home improvements, three focused on energy efficiency retrofitting (Ige et al. 2019;Pega & Wilson 2016;Naldzhiev et al. 2020;Bailey et al. 2019;Fenwick et al. 2013;Jevons et al. 2016), with one including interventions to reduce falls (Carnemolla & Bridge 2020). The two soundness studies reported on injuries from fire hazards (Senthilkumaran et al. 2019;Bailey et al. 2019;Fenwick et al. 2013;Jevons et al. 2016). Of all studies in this category, only those relating to home improvements reported on multiple health impacts; in addition to physical health, three studies looked at mental health (Garin et al. 2014;Ige et al. 2019;Fenwick et al. 2013) and one focused on the elderly (Garin et al. 2014).

Dwelling design
A total of 11 studies were identified under this category, further subcategorised into three groups: green buildings (three studies) looking at green walls, green design and green standards; building type (six studies) to do with high-rise, type of dwelling, shared facilities, etc.; and outdoor space (two studies). The green buildings studies assessed a range of health outcomes, including communicable disease (associated with water systems, vector-borne disease), allergens and air quality (Allen et al. 2015; Houghton & Castillo-Salgado 2017, 2019). The building type studies

Summary of building results
All systematic reviews in this category reported on physical health non-communicable outcomes associated with respiratory health (derived from exposure to allergens and air pollution) such as asthma, allergic, cardiovascular, blood pressure or thermoregulation conditions; and safety/ physical injury from specific housing types (e.g. duplex living). Only four reviews reported on communicable disease risks associated with water systems and shared sanitation as determinants for vector-borne disease. Compared with physical health, mental health outcomes were less reported (six studies) and in relation to a dwelling condition (e.g. thermal comfort) and design (e.g. type of building). Some reviews focused on vulnerable groups such as children and the elderly (indoor environmental quality and dwelling condition studies), the elderly (dwelling condition) and the disabled (dwelling design).
There was limited consideration of planetary health impacts (six studies). Studies reported on associated carbon emissions from traffic (Vardoulakis et al. 2020)

NEIGHBOURHOOD DETERMINANTS OF HEALTH (80 STUDIES)
A total of 80 studies (64.5%) reported on various aspects of the neighbourhood, which are broadly defined here as an urban area made of residential and other buildings, as well as the supporting infrastructure for everyday living and its resident community. These were organised under five broad categories: (1) green and blue infrastructure (34 studies), looking at greenery, water collection and waste; (2) physical conditions (16 studies), reporting on the effects of soundscape (i.e. noise from traffic), urban design (i.e. street layout, lighting, walkability), and neighbourhood renewal; (3) transport (14 studies), including traffic exposure, travel mode and mobility aspects; (4) access to local services (seven studies), such as shops, healthcare and education; and (5) socio-economic conditions (nine studies), reporting on effects of disadvantage, social capital and crime on health outcomes ( Table 2; and see Table S2 in the supplemental data online).

Green and blue infrastructure
A total of 34 studies were concerned with the relationship between green and blue infrastructure and health, grouped under greenery (32 studies) regarding various green aspects of neighbourhoods, including parks, greenspace, vegetation, trees, etc.; and water collection and waste (two studies) reporting on sewerage systems and water-borne pathogens (Jung et al.  Changes in cycling behaviour (Stewart et al. 2015); health associated with physical activity, air pollution and injuries (Stankov et al. 2020;Patterson et al. 2019;Möller et al. 2020;Macmillan et al. 2018 (Saitta et al. 2019), and older adults (Chastin et al. 2015). Five studies assessed various mental health outcomes, including stress and anxiety more generally (Calogiuri & Chroni 2014;Felappi et al. 2020;Gascon et al. 2017;Kondo et al. 2018a;Chastin et al. 2015), and in children and adolescents (Vanaken & Danckaerts 2018). One study assessed the mediating effects of vegetation on the relationship between stress and noise (Dzhambov & Dimitrova 2018). Six studies reported other types of outcomes in addition to health including benefits to the economy and society (Venkataramanan et al. 2019

Physical conditions
A total of 16 studies were grouped under this category considering soundscapes from road traffic mainly (six studies), aspects of urban design (eight studies) and neighbourhood renewal (two studies). Of these, six reviews examined the relationship between soundscape and health impacts, Two studies reported on health outcomes associated with neighbourhood renewal: they focused on health impacts of housing regeneration schemes and socio-economic determinants of health (Thomson & Thomas 2015), and on impacts on mental and wellbeing from improvements in neighbourhood infrastructure such as access to transport and street greening (Moore et al. 2018).

Transport
This category encompasses 14 studies grouped in three subcategories: traffic exposure (six studies), travel mode (six studies) and mobility (two studies). Studies looking at traffic exposure reported on cardiovascular outcomes (Malambo et al. 2016) in addition to mortality and other physical health impacts (Rugel & Brauer 2020) . Six studies reported on impacts of travel mode on health. Three studies were specifically concerned with cycling interventions, two looked at the potential for general public health impacts (Stewart et al. 2015;Stankov et al. 2020), and one at physical activity and diet-related health outcomes (Macmillan et al. 2018). One study considered public transport and weight-related health outcomes (Patterson et al. 2019). The remaining two studies looked at transport costs and health impacts (Möller et al. 2020), and mode of travel and health impacts in children from lower socio-economic groups (Ma et al. 2020). Two reviews looked at transport mobility issues such as barriers to transport arising from congenital cardiovascular conditions in children and adults (Davey et al. 2020), and transport mobility restrictions and health impacts including premature mortality in older adults (Rosso et al. 2011).

Access to local services
Seven studies were grouped in this category. Five studies focused specifically on access to and availability of local shops, with four looking at grocery shops and local food outlets such as supermarkets, farmers' markets, restaurants and community kitchens ( congenital heart disease in children and adults (Davey et al. 2020), and another reported on the health of children from lower socio-economic groups and exposure to traffic from their route to school (Ma et al. 2020). Two studies considered body mass index (BMI)-related outcomes in relation to grocery stores and supermarkets (Abeykoon et al. 2017;Malambo et al. 2016), one of which also considered blood pressure, diabetes mellitus and metabolic syndrome associated with fast-food restaurants (Malambo et al. 2016). Two studies reported on mental health outcomes: one looked at depression in the elderly and retail availability in the neighbourhood (Garin et al. 2014); and another looked at self-reported health and psychological health and grocery shops (Abeykoon et al. 2017). One study linked community kitchens with wellbeing benefits such as social engagement and community cohesion (Iacovou et al. 2013).

Socio-economic conditions
The socio-economic characteristics of neighbourhoods' impact health as they are underlying factors of: disadvantage and competition for scarce resources among neighbours; trust, social capital and collective action which can overcome challenges; and 'contagious' or 'epidemic' behaviours that makes neighbours engage in similar behaviours (Smelser & Baltes 2001). Nine studies were identified in this category reporting on disadvantage (three studies), social capital (four studies) and risky behaviours (two studies).
While only two studies reported on general physical health outcomes related to gentrification (Bhavsar et al. 2020) and adolescent pregnancy (Smelser & Baltes 2001), the overall focus in this category and touched upon by all studies was on mental health conditions (e.g. depression, anxiety, self-reported health, etc.) (five studies), health-risk behaviour (e.g. smoking, physical inactivity and early sex initiation) (two studies), and wellbeing (e.g. social health, loneliness) (four studies) outcomes. Moreover, many studies reported health impacts on vulnerable groups such children ( One paper reported on the health impacts on all these three groups and also referred to planetary health outcomes in relation to environmental equity aspects of 'green gentrification' which can result in displacing vulnerable residents and augment their need for more emergency room or mental health visits as well as their food insecurity (Bhavsar et al. 2020).
The impact of disadvantage on health was reported in relation to neighbourhood deprivation and health-risk behaviour such as smoking and physical inactivity (Algren et al. 2015); ethic segregation and mental health outcomes such as depression, anxiety, suicidality and suicide, psychotic experiences, and schizophreniform/psychotic disorders (Smelser & Baltes 2001); and gentrification and self-reported health, physical and mental health outcomes and health-related behaviour, with a specific focus on negative health outcomes for ethnic groups, children and displaced residents (Bhavsar et al. 2020). Four studies reported on social capital-related health outcomes looking at the amount of social capital in the neighbourhood and health outcomes in children and adolescents (Vyncke et al. 2013

Summary of neighbourhood results
Across all reviews in this group, three overall findings were apparent: (1) the green and blue infrastructure category received the most attention to date (34 studies); (2) physical health outcomes (respiratory and cardiovascular) were predominantly reported; however, mental health outcomes were also reported by 27 studies, especially in relation to greenspace (15 studies) and neighbourhood socio-economic conditions (nine studies)-interestingly, no transport study reported mental health outcomes; and (3) in contrast to the building level/group, nine studies reported on the socio-economic determinants of health and discussed these in relation to various vulnerable groups including children (15 studies), elderly (10 studies), disabled people (three studies), women (three studies), low income (two studies) and ethnic minority groups (two 745 Turcu et al. Buildings and Cities DOI: 10.5334/bc.119 studies); the intersectionality of health outcomes was also considered in three studies reporting on adverse health outcomes on low income and children or women.

URBAN SYSTEM DETERMINANTS OF HEALTH (32 STUDIES)
Local built environments are part of the wider urban system which extends beyond buildings and neighbourhoods to the whole city, immediate but also distant built or unbuilt hinterlands of regions, nations and, ultimately, to the planet. These are parts of the urban system connected by complex relations and feedback loops, which in turn can influence outcomes at the local level. In the housing sector, this occurs through, but it is not limited to, (1) planned action of housing at scale via spatial planning; (2) the unplanned, albeit regulated, interactions of agents and institutions seeking and providing housing represented by the housing system; and (3) via the consequential desired and undesired impacts on natural ecosystems. A total of 32 systematic reviews (25.8%) reported on urban system determinants of health categorised as spatial planning (17 studies); housing system (11 studies): and ecosystems (four studies) ( Table 3; and see Table S3 in the supplemental data online).

Spatial planning
A total of 17 studies were identified in this category reporting on health outcomes and informality (three studies), urban infrastructure (three studies), type of development (three studies) and master planning (eight studies). The three studies reporting on informality assessed the health impacts of strategies to improve the infrastructure, conditions and land tenure of slums, including communicable and non-communicable disease prevention, the risk of injury from chemical and biological hazards, as well as social impacts such as quality of life, education and employment (Alaazi & Aganah 2020;Turley et al. 2013;Henson et al. 2020). Three studies looked at urban infrastructure: one study assessed urban drinking water and gastroenteritis risk (Beaudeau 2018); one study reported on the structural soundness of the city in the face of earthquakes and subsequent building collapses (Doocy et al. 2013); and one study looked at urban exposure to overhead powerlines (Habash et al. 2019). Type of development effects were reported in three studies in relation to walkability, physical activity behaviours and obesogenic health impacts (Berghauser Pont et al. 2020;Chandrabose et al. 2019;Cyril et al. 2013). Eight studies focused on aspects of master planning at the city level and health outcomes such as morbidity and mortality related to non-communicable disease risks (i.e. physical inactivity), injury and mental health. More specifically, these studies focused on nature-based approaches (Kabisch et al. 2017), age-friendly infrastructure for the elderly or children (Sánchez-González et al. 2020;Nordbø et al. 2018), and smart city technologies (Rocha et al. 2019). Strategic planning and smart growth approaches were also included (McCrorie et al. 2014;Durand et al. 2011;Gong et al. 2016;Salgado et al. 2020).

Housing system
Housing system studies (n = 11) were categorised into two categories: vulnerability (eight studies) and policy (three studies). In the vulnerability subcategory, three studies looked at how foreclosure, either directly experienced or general risk in the neighbourhood, negatively affected physical and mental health, as well as health-relevant behaviours (including substance misuse and violence) (Downing 2016; Tsai 2015; Vásquez-Vera et al. 2017); one study looked at how combinations of tenure precarity and poor physical characteristics of dwellings may combine to influence mental health (Singh et al. 2019); one study assessed how permanent supportive social housing may benefit both health (e.g. mental health, hospital admissions) and health-supporting social conditions (e.g. employment and income) (Aubry et al. 2020); and three studies looked at vulnerability by specifically focusing on a population subgroup (people living with HIV) (Aidala et al. 2016), a risk type (cold weather) (Tanner et al. 2013), and an outcome (congenital heart disease) (Davey et al. 2020). Of the three studies classified in the policy subcategory, two looked at how a range of material housing changes and support contributed to changes in the risk of obesity (Tseng et al. 2018) and diabetes (Barnard et al. 2015). A third study considered strategies that combined material (e.g. built infrastructure) and social (e.g. community networking and empowerment) aspects to create healthy environments (e.g. availability of healthy food; encouragement of physical activity) that would in turn improve community health (Chaparro et al. 2020).

Natural ecosystems
Four studies reported on natural ecosystems such as air, water and climate. Two studies considered air pollution and its effects on cardiorespiratory outcomes in people of different age groups in Sub-Saharan Africa (Katoto et al. 2019), and the health and health equity benefits of interventions aiming to reduce air pollution levels (Benmarhnia et al. 2014). One study assessed the mental health benefits of wider natural ecosystems for residential areas (Gascon et al. 2017), and one paper looked at microclimate influences on urban heat islands and the resulting impacts of mortality and cardiorespiratory morbidity (Schinasi et al. 2018).

Summary of urban system results
More than half (17 studies) of all studies in this group focused on the impact of spatial planning on both health exposure and health interventions. Physical health impacts were discussed in relation to both communicable and non-communicable diseases, more specifically general health outcomes, mortality and morbidity, walkability and physical activity outcomes, drinking water 747 Turcu et al. Buildings and Cities DOI: 10.5334/bc.119 quality and communicable disease, resilience to natural disasters, cancer, reproductive health, obesity and cardiorespiratory. Mental health impacts were discussed in relation to informality and master planning aspects; housing system's vulnerability (mainly measures of housing insecurity) and policies (material change and material and social support) aspects, and the natural ecosystem. There was some evidence reporting on the health of particular demographic groups such as the elderly and children (under spatial planning), and vulnerable groups including those with underlying health conditions (HIV, heart disease) and especially those with insecure housing tenure (under housing system).
Surprisingly, less than one-sixth of studies (n = 5) discussed planetary health outcomes. These reported on the impacts of rapid urbanisation on the environment and implications for the SDGs (Henson et al. 2020;Chandrabose et al. 2019); and climate change impacts of air pollution (Benmarhnia et al. 2014;Katoto et al. 2019) and microclimates (Schinasi et al. 2018).

DISCUSSION
This paper provides an overview of the last decade's evidence on health and urban housing from a spatial scale perspective. Most of the systematic review evidence (64.5%) reviewed by this paper focused on the neighbourhood level of the local built environment, while the building and urban system levels accounted for the rest in equal shares. Across the three scales, however, five out of 11 categories of built environment determinants of health have received most attention to date: green and blue structure (32/124); spatial planning (17/124), neighbourhood's physical conditions (16/124) and transport (14/124), and a dwelling's indoor environmental quality (12/124). The number of reviews focusing on health impacts at the building level was less than anticipated; this can be explained by the fact that existing evidence is not yet published in English and/or summarised by systematic reviews during the period 2011-20.
Three overall observations can be made across all scales and all studies. Physical health outcomes remained predominantly reported by systematic reviews, primarily in relation to noncommunicable diseases in high-income settings. This may be because the majority of reviews focused on developed contexts where non-communicable diseases form the predominant burden of health. Mental health outcomes were included in 40 studies (one-third of all studies): 27 were at the neighbourhood level, and six and seven at the building and urban system scales, respectively. Only one-fifth of studies (27/124) reported some planetary health outcomes, and again mostly at the neighbourhood level (16/27), and 6/27 at the building and 5/27 at the urban system levels.

WHERE NEXT FOR RESEARCH?
This umbrella review indicates three potential directions for further research, more generally, and systematic review research, more specifically. First, research on urban health usually involves two distinct communities of scholars, health scientists and urban scientists, who can come from completely different research paradigms. This requires time to learn or synthetise across disciplines, transdisciplinary methods to account for the variety of entry points, but also relational thinking to acknowledge the multiple connections between the different elements of the urban system and the continuum of health outcomes, e.g. physical-mental-wellbeing. The studies identified by this paper come from teams of primarily health scientists, hence grounding findings with urban scientists can be challenging. There is also a predominant focus on the 'negative psychology' approach to health whereby treating the effects of a particular condition (i.e. cardiovascular, respiratory, BMI, blood pressure) is in focus, as opposed to 'positive psychology' approach when the cause of the condition (poverty/deprivation, vulnerability) is analysed (Seligman 2004). While some mental health and wellbeing outcomes are mentioned at the neighbourhood and wider urban system levels, they need better understanding. Here, a 'flourish' approach to health can be explored, whereby the focus is on people, rather than their health, under the assumption that improving the wider quality of life and social health would make for healthier people (Seligman 2011).
Second, there are three obvious gaps in the literature: the intersectionality of health outcomes, lack of evidence from low-income settings and little current discussion of communicable disease 748 Turcu et al. Buildings and Cities DOI: 10.5334/bc.119 burden. While there are often clear pathways for increased risk for vulnerable population groups, review evidence for these groups appears to be limited and tends to be focused on children and the elderly; there is little consideration of how, e.g., age, gender and socio-economic status may intersect in the built environment and affect health. There was also very limited evidence from low-income countries and/or communicable disease outcomes; where reviews were identified they reported health outcomes associated with indoor air pollution from cooking stoves and informal living, lack of water infrastructure and sanitation, and vector-borne disease. The current COVID-19 pandemic will certainly move the focus back onto communicable disease outcomes.
Third, planetary health impacts are mainly reported in relation to carbon emissions. Expanding understanding beyond this point is another research direction worth exploring. Research reporting on planetary impacts in conjunction with human health ones, across all scales of the built environment, can reinforce advocacy for urban sustainability transitions. If the challenges of the health-climate crisis is to be met, the nexus of human and planetary health needs better understanding of unintended consequences, better policymaking and urban governance at all levels (Crane et al. 2021).

A ROLE FOR POLICY AND URBAN GOVERNANCE
Urban health research is closely associated with relevance to policymaking (Hawkes et al. 2016;Sallis et al. 2016;Schneidera & Blythb 2017). This paper found that many studies note implications for policymaking and urban stakeholders (e.g. urban planners, landscape architects, communities, residents, etc.), but discussion is rather general. This may be explained by the fact that most reviews take a health perspective whereby roles outside health in implementing or changing exposure risks in the built environment are not considered. In what follows, this paper contributes to expanding policy understanding in this area by exploring different types of policy interventions and actors involved across the scales of the built environment.
At the building level, exposure to many identified environmental risk factors is long-term and difficult to modify without substantial investment of time and resources. For example, remedial factors to address risks from exposure to low indoor temperatures via energy retrofitting to protect against winter cold may take substantial investment. Likewise, factors related to a building's condition have clear benefits in terms of health when considering safety measure (e.g. injury from fire, falls, etc.) and building regulations usually address these issues, but regulation and compliance may be challenging in some settings or differ amongst sectors and professions. The building level is usually addressed by architects, designers, developers, building contractors, owners of individual structures and health practitioners.
At the neighbourhood level, the weight of evidence suggests positive health outcomes are associated with green space, although these may be confounded by socio-economic status, i.e. wealthier neighbourhoods having higher density of green space and living in deprived/poor neighbourhoods is linked to adverse health impacts for vulnerable groups such as children, the elderly and disable people, women and ethnic minority groups. While it may be challenging to add new green space to established cities, modifications to existing green space can be made to encourage physical activity, along with fair and equal access for all socio-economic groups and education to effect behaviour change. Actors involved in the governance of the neighbourhood level primarily include local government, local planning and local health trusts, communities, civil society and business organisations. At the urban system level, the evidence presented points to at least two important 'alignments', e.g. between health, spatial planning and housing policy; and between climate change and health outcomes. City-level spatial planning policies can impact on health by providing adequate levels of affordable or social housing and so de-risking housing security, an important socio-economic determinant of health, while housing policy at the regional or national level can help to absorb shock-related health impacts from events such as the financial crisis in 2007-08 or the current COVID-19 pandemic. Furthermore, increasing greenspace and energy retrofitting are associated with clear health and climate change-positive outcomes; high density or shared living can reduce pressure on resources and associated carbon emissions, but some evidence suggests mental 749 Turcu et al. Buildings and Cities DOI: 10.5334/bc.119 health-negative outcomes from overcrowding and impacts on physical health from limited space. The latter is easier to address in policy terms, e.g. the provision of easily accessible parks and recreational facilities to allow physical activity, while overcrowding is harder to address and requires long-term action and investment, but policy measures such as standards and regulations can help. The governance of the urban system involves all the actors at the building and neighbourhood level and much more, i.e. regional and national governments involved in strategic policymaking, but also global organisations including international institutions such as the WHO and European Union.
The above examples frame urban health within the wider process of policymaking in the build environment. Policymaking is a complex and, most importantly, a political process; it is not something happening at a particular time, in a particular spot, but part of wider multilevel governance frameworks, which frame the complexity of the urban system and local built environment. As seen above, the governance of urban health involves policymakers responsible for health-or housing-related policy and regulation, and other stakeholders involved in health interventions such as government agencies, architects, builders, housing providers, developers, engineers, urban planners, industry regulators, financial institutions as well as social services, community groups and public health professionals. These stakeholders are ultimately required to ensure that housing is built, maintained, renovated, used and demolished in ways that support health.
By taking a scale perspective on health, the connections between different policies and levels of policymaking become apparent. For example, national government needs to align health with SDGs, ensure geographical equity and combat siloed approaches, while local government must ensure that public health and spatial planning work together to strengthen the link between people and places and break administrative boundaries to reap the benefits of planetary health. Also, the wider the scale, the more complex the array of actors involved and the dynamic lines of power and networks inside and outside policymaking (Bulkeley & Kern 2006). Communities may hold important knowledge about public health in their locality, but may not have a voice, especially if marginalised; civil society organisations may lobby or support government and communities, while professional communities such as urban planners share many communalities with public health professionals including an interest in the public good, and use of evidencebased and long-term assessment approaches (WHO 2020). If evidence fails to engage with the multi-scalar, multilevel nature of urban governance, which makes links across scales from local to supranational, and where power is distributed across horizontal and vertical networks which do not operate in a hierarchical manner (Bulkeley & Betsill 2005). This paper suggests that a better grounding of urban health research in existing urban governance landscapes would not only support a faster and more efficient implementation of health interventions at the local level, but also gauge potential synergies and tensions at the urban system level with other pressing urban challenges such as climate change.

STRENGTHS AND LIMITATIONS
This paper is novel for synthesising evidence on housing from the perspective of scales within the urban built environment. Most research has either considered housing from a narrow understanding of the individual's dwelling or disconnected the built environment from its purpose of providing liveability (Giles-Corti et al. 2019). To the present authors' knowledge, this is the first umbrella review to take a comprehensive view of the local built environment and housing at scale while looking at both physical and socio-economic characteristics that define the complex urban system. This review is limited to systematic review evidence and generic geographical regions. Evidence derived purposely for specific countries was excluded because of the specific content and context. As such, the authors acknowledge the broad conclusions made in this review. The intent was not to assess the size of health impacts or the effectiveness of interventions and make no assessment of the importance of one health determinant or built factor over another. Furthermore, the study excluded research on shelters for the homeless, residential care or student accommodation, and rural housing, which also pose built environment, societal and health concerns. 750 Turcu et al. Buildings and Cities DOI: 10.5334/bc.119

CONCLUSIONS
This paper provides an umbrella review of reported health impacts across three broad spatial scales and notes that most research has focused on health at the neighbourhood level. Discussions of planetary health and policy implications have been limited, and only a few studies have evaluated the economic implications of health interventions. This study is the first of its type: it applies a multiscalar perspective to health, it suggests directions for potential future research, and it expands the discussion of urban governance for health.