The Potential for Healthy, Sustainable, and Equitable Transport Systems in Africa and the Caribbean: A Mixed-Methods Systematic Review and Meta-Study

Abstract

The Human Mobility Transition model describes shifts in mobility dynamics and transport systems. The aspirational stage, ‘human urbanism’, is characterised by high active travel, universal public transport, low private vehicle use and equitable access to transport. We explored factors associated with travel behaviour in Africa and the Caribbean, investigating the potential to realise ‘human urbanism’ in this context. We conducted a mixed-methods systematic review of ten databases and grey literature for articles published between January 2008 and February 2019. We appraised study quality using Critical Appraisal Skills Programme checklists. We narratively synthesized qualitative and quantitative data, using meta-study principles to integrate the findings. We identified 39,404 studies through database searching, mining reviews, reference screening, and topic experts’ consultation. We included 129 studies (78 quantitative, 28 mixed-methods, 23 qualitative) and 33 grey literature documents. In marginalised groups, including the poor, people living rurally or peripheral to cities, women and girls, and the elderly, transport was poorly accessible, travel was characterised by high levels of walking and paratransit (informal public transport) use, and low private vehicle use. Poorly controlled urban growth (density) and sprawl (expansion), with associated informality, was a salient aspect of this context, resulting in long travel distances and the necessity of motorised transportation. There were existing population-level assets in relation to ‘human urbanism’ (high levels of active travel, good paratransit coverage, low private vehicle use) as well as core challenges (urban sprawl and informality, socioeconomic and gendered barriers to travel, poor transport accessibility). Ineffective mobility systems were a product of uncoordinated urban planning, unregulated land use and subsequent land use conflict. To realise ‘human urbanism’, integrated planning policies recognising the linkages between health, transport and equity are needed. A shift in priority from economic growth to a focus on broader population needs and the rights and wellbeing of ordinary people is required. Policymakers should focus attention on transport accessibility for the most vulnerable.

1. Introduction

Travel behaviour, and the mobility systems that underpin it, have implications for population and planetary health [1,2,3]. These include, for example, the risk of road traffic injury and death, the risk of respiratory diseases or other non-communicable diseases related to air pollution and sedentary modes of transport, or the implications of car use on climate change [4,5]. Mobility systems (policies, practices and infrastructure which permit or constrain movement) are key to facilitating equitable access to resources and opportunities, and their importance is well reflected in global agendas. For instance, the 2015 United Nations Sustainable Development Goal (SDG) target 11.2 aims to provide access to “safe, affordable, accessible and sustainable transport systems for all” [6]. Aligned with the SDGs, the United Nations has endorsed the New Urban Agenda [7], a concerted global effort to “promote age- and gender-responsive planning and investment for sustainable, safe and accessible urban mobility for all and resource-efficient transport systems for passengers and freight, effectively linking people, places, goods, services and economic opportunities”. The need for accessible, inclusive and sustainable transport systems with minimal harmful environmental impact is also highlighted in the United Nations Decade of Healthy Ageing (2021–2030) [8], and the Paris Agreement [9].

The determinants of travel behaviour are complex, involving multiple interacting levels of influence ranging from individual characteristics [10,11,12] through to built, natural, and social environments [13,14,15] acting alongside political and geopolitical forces [16,17]. Hérick de Sá and colleagues [18] recently proposed a framework to describe historical macro-level shifts in human mobility dynamics that shape urban form and overlying social networks, calling this the Human Mobility Transition. Beginning with walking in early humans, the transition moves through the expansion of auxiliary forms of transport (for example, horse riding), the emergence of motorisation—and what has been termed the automobility regime [19]—towards an aspirational stage termed ‘human urbanism’. In this last stage, a focus on active travel, universal public transport, and inclusive cities produces the conditions required for a healthy, sustainable, and equitable mobility system.

The continent of Africa (population 1.3 billion in 2019) is undergoing rapid and mostly unplanned urbanisation. This complex phenomenon is mainly driven by rapid population growth which can undermine principles of sustainable socioeconomic progress and environmental protection. Negative consequences include urban poverty, unsustainable land use, and health inequalities [20]. Caribbean countries (population 44 million in 2019) are not experiencing population growth to the same extent but are disproportionately vulnerable to climate change considering their <1% contribution to global greenhouse gas emissions. They have unique historical, geophysical, environmental, and economic characteristics typical of small islands, are highly dependent on tourism and experience climate change effects like economic losses and weakened infrastructure, forced migration, gender inequalities, poverty, and loss of natural resources [21,22]. Despite the differences between Africa and the Caribbean, both regions have complex mobility systems typified by the deficient provision of formal public transport, a dominant paratransit sector (privately owned informal transport modes serving the general public), congestion, air pollution, and high levels of traffic fatalities [23,24]. Currently, there is little evidence on the main factors that determine choice and use of transport modes in these contexts [25], limiting our understanding of the extent to which the human mobility transitions occurring in Africa and the Caribbean are linked to their epidemiological transitions with rising levels of non-communicable diseases. Given the contribution of travel behaviour and mobility systems to health, poverty, and the environment, greater evidence on the factors shaping travel behaviour (and their inter-relationships) could help with achieving better outcomes. Our systematic review involved a global health partnership of researchers based in Cameroon, Kenya, South Africa, Barbados, Jamaica, and the United Kingdom [26]. We aimed, firstly, to identify and synthesize existing empirical quantitative and qualitative evidence from a wide range of disciplines on the factors associated with routine or regular travel behaviour in Africa and the Caribbean. Second, by framing our review against the Human Mobility Transition, we aimed to explore the potential for realising ‘human urbanism’ in these regions by examining how the most reported factors related to a set of predefined indicators of healthy, sustainable, and equitable mobility systems. Finally, we aimed to describe and critique the theories and methods reflected in the reviewed literature and propose ways forward for this field of research.

2. Materials and Methods

This mixed-method systematic review was registered with PROSPERO (CRD42019124802) and was conducted in accordance with PRISMA [27], SWiM [28] and ENTREQ guidelines [29] (File S1). The review methods have been described elsewhere [30,31]; here, we summarise these and provide detail on methodological aspects unique to this analysis. A diagram depicting the methodological flow unique to this study is presented in Figure 1.

2.1. Theoretical Framework

We grounded our analysis in a socioecological approach, aligning with the perspective of Krieger [32] on understanding the social determinants of health. This moves from a linear view of putative determinants as proximal or distal, towards an appreciation that conceptual ‘levels’ of influence coexist, intermingle, and exert causal effects simultaneously. We drew inspiration from elements of complex systems thinking [33] which involves considering the “bigger changing picture” [34]. We conceptualised mobility systems as existing within a larger system influenced by a dynamic range of actors within and beyond the transport sector. We applied the concept of ‘system framing’ (the application of complex systems thinking to a range of study designs) [35] in terms of explicitly considering how putative determinants of travel behaviour interacted with one another and with elements of mobility systems. In addition, we used the Human Mobility Transition [18] (File S2) as a guiding framework, focussing on the conditions required to realise ‘human urbanism’.

2.2. Search Strategy and Selection Criteria

We included literature on routine or regular travel behaviour, as this is likely to impact the health of people, populations, and the planet. We chose to focus on post-2008 literature to maximise relevance to the current policy context, given that between 2002 and 2008 a succession of World Urban Forums [36] established a basis for subsequent progressions within transport and mobility. Eligibility criteria are detailed in

Table 1. Systematic review inclusion and exclusion criteria.

Domain	Inclusion Criteria	Exclusion Criteria
Study design	Studies must contain empirical data (primary or secondary) and present an analysis of these data. All study designs (quantitative and qualitative) are eligible	Literature reviews, narrative overviews, commentaries, opinion pieces, or any format not providing sufficient information to allow for data extraction
Participants	The general population living in African and Caribbean countries. No age or sex/gender restrictions	Studies focussed on specific or unique population segments in which travel is likely to be atypical: people with specific health conditions; professional travellers (e.g., bus drivers, professional cyclists); tourists, refugees, asylum seekers or migrants, or victims of trafficking—although relevant data pertaining to these population groups was included when part of an included study focussing on the general population
		Studies investigating non-human travel such as food or freight
Exposures	Both correlates (where causality is uncertain), as well as purported causal influences on travel behaviour
Comparators	All eligible, if used
Outcomes	Routine or regular travel behaviour, including: time spent in all travel or particular travel modes; number of trips; choice or use of particular travel modes or combinations of modes; mode share	Studies focussed on single travel purposes: school-related travel; travel to administer or receive healthcare
		Studies focussed on hypothetical (rather than actual) use of transport modes
		Studies without a primary focus on travel per se: road traffic accidents, injuries, or road safety as the main outcome
Timing	1 January 2008–31 January 2019
Setting	Africa and the Caribbean (File S3)	Studies set in contexts in which travel is likely to be atypical: war, political crises or natural disasters
Language	All languages considered

.

We searched MEDLINE, Transport Research International Documentation (TRID), SCOPUS, Web of Science, LILACS, SciELO, Global Health, Africa Index Medicus, CINAHL and MediCarib. A full search strategy is contained in File S4. Searches were conducted in February 2019, covering the period 1 January 2008 to 31 January 2019.

We conducted 100% double screening at both the title and abstract stage, and the subsequent full text stage, with discrepancies resolved by a third team member. To identify additional citations, we mined literature reviews, conducted forward (citation screening using Scopus and Web of Science) and backward (reference list) screening of eligible studies, and contacted topic experts from the areas of transport, urban planning and health, and anthropology. All additional citations were single screened by one team member, resulting in the final set of eligible studies. Using a bespoke data extraction template (File S5), team members undertook data extraction for the full set of eligible studies. Following this, we double extracted a set of fields for a randomly selected 20% of studies to check for consistency. When unacceptable (>50%) disagreement was identified, the data extraction was repeated in full by a small group of reviewers and the first set of extracted data was discarded.

Critical Appraisal Skills Programme (CASP) checklists [37,38] were used to appraise the trustworthiness, relevance, and results of eligible studies. For quantitative studies (and the quantitative element of mixed-methods studies), an adapted CASP cohort study checklist was used (File S6), which included items on biases, confounding and generalisability. For qualitative studies (and the qualitative element of mixed-methods studies), the CASP qualitative checklist was used (File S7), which included items on design and methods, ethics, and the participant-researcher relationship. Each study was appraised by one member for the purpose of gaining insight into the quality of the extant literature overall. We did not use quality as a basis on which to exclude studies.

We consulted within our multi-country, multi-disciplinary review team, as well as topic experts, to identify relevant sources of grey literature. After a pilot study, we searched World Health Organization, UN-Sustainable Development Platform, UN-Habitat, Institute for Transportation and Development Policy, The Carbon Trust, The World Cycling Alliance, Sub-Sahara Africa Transport Policy Programme, Prospero, and OpenGrey websites. In addition, we conducted Google advanced searches using keywords combined with specific country domains (File S8). Searches were conducted in August 2019 and yielded 164 relevant publications which we screened in duplicate. A final set of 33 grey literature publications was included for data extraction. We did not assess quality of the grey literature as this was included to complement and enrich the results from the peer reviewed dataset, rather than forming part of the main body of data analysed. We did note that most of the grey literature was sourced via platforms with internal quality controls for publication, such as the UN-Habitat repository.

2.3. Operationalisation of Exposures and Outcomes

We operationalised the ‘human urbanism’ stage of the Human Mobility Transition into a set of indicators signifying the opportunity for, or challenge to, healthy, sustainable, and equitable mobility systems (

Table 2. Indicators for the ‘human urbanism’ stage of the Human Mobility Transition.

Indicators Signifying Opportunity for a Healthy, Sustainable, and Equitable Mobility System	Indicators Signifying a Challenge to a Healthy, Sustainable, and Equitable Mobility System
High use of active forms of travel	Poorly controlled urban growth and sprawl
Universal public transport	Socioeconomic barriers to mode choice
Transport accessibility for all, including marginalised groups	Mode-shifting to private vehicles

). For each indicator, we then explored the relationships of exposures (i.e., putative determinants) with potential outcomes. For example, we explored the relationship of cost of travel mode (exposure) with public transport use (outcome) in relation to universal public transport (indicator).

To aid this process, we grouped the exposures into six thematic sets: individual characteristics; travel mode characteristics; built environment; natural environment; socio-cultural environment; and policy or political environment (

Table 3. Examples of exposures grouped into thematic sets.

Thematic Set	Examples
Individual characteristics	Age, sex, ethnicity, marital status, socioeconomic status, health status
Travel mode characteristics	Accessibility, safety, capacity, service quality
Built environment	Transport infrastructure, urban layout, road blockage
Socio-cultural environment	Social norms, stigmatisation or aspirations related to particular travel modes, general crime levels
Natural environment	Weather, wildlife, air pollution
Policy or political environment	Policy environment, economic climate, historical legacies

). A list of potential outcomes is presented in

Table 4. Potential outcomes explored under each indicator.

Indicator	Potential Outcomes
High use of active forms of travel	Walking use and mode share Cycling use and mode share Other active modes use and mode share
Universal public transport	Public transport use and mode share 1
Transport accessibility for all, including marginalised groups	Distance to access travel modes Number of trips 2 Coping strategies to deal with transport inaccessibility
Unplanned urban growth and sprawl	Area of residence Distance travelled Aggregate travel time Time spent in specific travel modes
Socioeconomic barriers to mode choice	Use of different travel modes Choice of travel mode Affordability of travel mode Price stability Mobility budget Coping strategies to deal with socio economic barriers
Mode-shifting to private vehicles	Private car use and mode share Private motorbike use and mode share Stigmatisation or aspiration of particular modes Attitudes towards travel modes Car as a symbol of status Motorbike as a solution to congested traffic

¹ This includes both formal and informal public transport. ² This includes those with zero trips who did not travel.

. When exploring these relationships, we acknowledged their bi-directional nature. For example, cost could be reported as high being a barrier to modal choice, as well as low facilitating transport accessibility.

2.4. Synthesis

We use the term ‘synthesis’ to refer to the process of compiling study findings within either quantitative or qualitative strands. For both qualitative and quantitative data, this took the form of a narrative synthesis. For both data types, we extracted individual study findings as the basic unit of synthesis (i.e., exposure-outcome relationships) in relation to the indicators described in Table 2. We described the overall direction of effect and number of studies that reported this (i.e., vote counting) [39]. Though this did not fully capture the nuance of the qualitative data, we considered it appropriate for this analysis due to our familiarity with the studies developed through two separate meta-ethnographic reviews conducted previously using this dataset [30,31]. Our appraisal of the certainty of evidence was bespoke, and is described in detail elsewhere [30,31]. Briefly, this entailed extending our vote counting exercise to incorporate elements of Bradford Hill’s principles of causation [40]. For the grey literature dataset, we conducted a thematic analysis using themes and codes that mimicked the exposures identified in the peer-reviewed dataset, and we extracted segments of text according to these thematic sets.

2.5. Analysis

We use the term ‘analysis’ to refer to the process of drawing together, integrating and understanding findings from the quantitative or qualitative strands. We conducted our analysis in two stages, based on a convergent design [41] where qualitative and quantitative strands were analysed and integrated in parallel. The grey literature was included in the second analysis stage only to concurrently enrich the insights gained from each of the strands.

2.5.1. Analysis 1: Charting and Prioritising

We charted our study findings according to study design (quantitative, mixed-method or qualitative) and exposures reported, grouping these exposures into the thematic sets described in Table 3. Where necessary, some exposures were grouped together (for example, street lighting, quality of sidewalks or cycle paths or street-crossings, and speed management methods such as speed bumps were grouped together as transport infrastructure under the built environment thematic set). We quantified the number of times each exposure was reported, by study design and as a proportion of the total set of eligible studies. From this, we prioritised the most reported exposures (based on percentage of frequency) in each thematic set for more detailed examination in the next stage of analysis. Studies that did not examine the prioritised exposures were excluded from the second stage.

2.5.2. Analysis 2: Meta-Study Exploring Factors Signifying Opportunity for or Challenge to Healthy, Sustainable, and Equitable Mobility Systems

We used the principles of meta-study [42] to describe and compare findings across studies in relation to the ‘human urbanism’ stage of the Human Mobility Transition (Table 2), using the prioritised exposures from the previous analysis stage. Meta-study involves four steps: meta-data analysis; meta-method; meta-theory; and meta-synthesis. Meta-data analysis examines the empirical findings; meta-method investigates the methods used; meta-theory analyses the underlying theories, assumptions, and principles of the studies; and meta-synthesis brings the previous three steps together to iteratively integrate and understand the studies’ findings as a whole. The meta-study approach is appropriate for synthesizing diverse evidence from distinct perspectives and disciplines to generate new understandings of a phenomenon [43] In using this approach we sought to retain our ‘system framing’ in terms of considering the myriad ways in which the putative determinants of travel behaviour and mobility systems could interact with each other. The meta-synthesis was initially developed by a small group of reviewers and further refined with the wider review team. In this analysis stage, we added grey literature documents, but only to the meta-data analysis and meta-synthesis steps, as research methods and theories were rarely reported in this literature. This evidence was intended to bolster understanding of the relationships identified from the peer-reviewed dataset.

3. Results

We identified 34,633 studies through database searching, 395 studies through mining reviews and 4229 studies through backward and forward reference searching. We contacted 27 topic experts, of whom 8 responded and provided an additional 147 studies. In all, 129 studies met eligibility criteria and were included in Analysis 1: 78 quantitative (60%), 28 mixed-methods (22%), and 23 qualitative (18%). Of these, 118 studies examined prioritized exposures and were included in the meta-study (Analysis 2). An additional 33 documents were identified through grey literature searching and included in the meta-study (Figure 2; File S9).

The 129 studies (i.e., papers) included in Analysis 1 yielded 168 datasets (i.e., units of data collection), as some studies were multi-country and contained more than one dataset. Most datasets (98%) were from Africa; datasets from the Caribbean were drawn from Jamaica, Suriname, and Panama. Datasets from Africa were drawn from Eastern Africa (38%), Western Africa (33%), Southern Africa (17%), Central Africa (5%) and Northern Africa (5%). The most represented countries by region were South Africa (93%) in Southern Africa, Ghana (38%) and Nigeria (36%) in Western Africa, and Uganda (21%), Kenya (21%) and Tanzania (18%) in Eastern Africa. Algeria and Cameroon were the most represented countries in Northern and Central Africa with 4 and 5 datasets, respectively (Figure 3). Most of the studies (87%) explored urban settings only, 7% explored rural settings only, and the remaining 6% explored both. The 33 grey literature documents included 29 reports (7 global-level, 14 regional-level, and 8 country-level), and four case studies from Lagos (Nigeria), Soweto (South Africa), Kampala (Uganda), and Dar es Salaam (Tanzania).

We found that the quantitative data provided insufficient detail on adjustment for confounding factors and relied predominantly on descriptive analyses and self-reported outcomes (though using validated and widely used travel behaviour research tools). Formal group comparisons were typically non-existent. Nevertheless, the quantitative data were useful in relation to exploring the travel behaviour of clearly defined groups within the local population. The qualitative data provided meaningful insight into participant experiences and helped to contextualise observed outcomes. However, we found them to contain insufficient details on recruitment strategy, data collection methods, and analysis rigour. Reporting and reflection on ethical issues and the relationships between researchers and participants were also poor in most qualitative studies.

3.1. Analysis 1: Charting and Prioritising

From 129 studies that met eligibility criteria, we identified 74 unique exposures (i.e., putative determinants) associated with travel behaviour (File S10). When grouped by thematic set, most exposures related to individual characteristics (examined in 45% of the cases), followed by travel mode characteristics (29%), built environment (13%), socio-cultural environment (6%), policy or political environment (5%), and natural environment (3%) (Figure 4). When examining exposures by study design, we observed that individual characteristics were examined in a higher proportion of the quantitative studies (56%) compared with mixed-methods studies (38%) and qualitative studies (29%). Conversely, qualitative studies were more likely to examine the policy or political environment than the other study design types (13% of qualitative, 4% of mixed-methods, 2% of quantitative studies) (File S11).

We used this first stage of analysis to prioritise frequently reported exposures in each thematic set for a deeper examination in the next stage. For the most part, exposures were included if their percentage of frequency was above the mean for the relevant thematic set. However, six additional exposures, with percentages below the mean, were also included due their direct relevance to ‘human urbanism’, as for example the individual ability to cycle. The selected exposures are listed in

Table 5. Selected exposures for Analysis 2.

Thematic Set	Selected Exposures
Individual characteristics	Age, sex, area of residence, socioeconomic status, attitudes towards particular travel modes *, ability to cycle *
Travel mode characteristics	Accessibility and availability of a service, safety and crime, cost, comfort, and convenience
Built environment	Transport infrastructure, spatial layout, traffic levels, aesthetics *, cycling-friendliness *
Socio-cultural environment	Social expectations, stigmatization, or aspiration of particular travel modes *
Natural environment	Weather, topography *
Policy or political environment	Transport sector, historical legacies, urban planning sector

* Additional included exposures for Analysis 2.

. Definitions and further details for each exposure are described in File S10.

3.2. Analysis 2: Meta-Study Exploring Factors Signifying Opportunity for or Challenge to Healthy, Sustainable, and Equitable Mobility Systems

Our meta-study included 118 studies examining the prioritised exposures from analysis 1 in relation to the Human Mobility Transition: 71 quantitative (60%), 27 mixed-method (23%), and 20 qualitative (17%). We also included 33 documents identified through grey literature searching.

3.2.1. Meta-Data

For our meta-data analysis, we focussed on drawing out the complexity of how exposures related to each other and to outcomes, with reference to the indicators signifying opportunity for, or challenge to, healthy, sustainable, and equitable mobility systems. As aligned with our conceptualisation of putative determinants [32], exposure-outcome relationships varied across different population groups depending on how exposures interacted with each other and with elements of mobility systems. In addition, the relationships we identified were connected to multiple indicators, acting concurrently ‘for’ one indicator but ‘against’ another.

• Indicator Signifying Opportunity for a Healthy, Sustainable, and Equitable Mobility System: High Use of Active Forms of Travel

Walking was the principal mode of transport in the studied cities, particularly in Africa [44]. Levels of cycling were low overall, but more common in smaller cities and peri-urban areas [44,45]. We found that lower individual or household socioeconomic status was related to high use of walking [46,47], and, to lesser degree, cycling [48,49,50,51,52,53,54]. Socioeconomic status was linked with location and nature of residence, in particular living in unplanned, poor settlements on the periphery of cities. Residents of these settlements typically relied exclusively on walking for travel within the settlement [55,56,57,58,59], and on combinations of walking and public transport (informal and formal) to travel beyond the settlement [60,61,62,63]. The inadequacy of public transport provision in these areas, and unaffordability of available public transport, made walking considerable distances unavoidable [63,64,65,66,67,68,69,70,71]. Overlaid across these indicators of disadvantage and marginalisation were stigmatising social norms that had developed around active travel; for example in Ghana, where walking was viewed as degrading [72]. Cycling was perceived as an indicator of poverty in Africa overall [51,54,73,74,75,76,77]. Thus, while we found evidence in line with high levels of active travel in these contexts, other evidence showed reasons for socioeconomic barriers to mode choice.

Girls and women were less likely to undertake active travel than boys or men [45,49,50,51,52,53,67,78,79,80,81,82,83,84,85,86]. This was due to cultural expectations (that increased with age) around household responsibilities, and because of safety concerns which stemmed in part from patriarchal views [47,54,87,88,89,90], with cycling often seen as particularly inappropriate for women [45,51,54,74,91,92,93]. We found that the relationship between age and active travel differed across the life course. For example, three studies from Ghana, Egypt, and Nigeria revealed that older boys walked more than younger boys during childhood, related to having fewer mobility constraints imposed by parents [45,51,94]. Practically speaking, active travel was often the only mode available to older boys who had not yet reached driving-license age and did not have the financial resources for a car, or even to pay for public transport [47,48,56,59,81,94,95]. In working age adults, active travel decreased with age, as seen in studies from Ethiopia, Egypt, and Uganda [48,51,96], and mode shifting from active travel to either public transport or private cars depended on socioeconomic status (i.e., being able to afford to buy and maintain a car) but was also influenced by the stigmatisation of active travel. Amongst the elderly, rates of walking were relatively high due to inability to afford motorised transport, shorter trip distances, and difficulties getting on and off public transport, as seen in narratives drawn from across the African continent [67,81].

Walking and cycling were perceived as unsafe; therefore when alternative modes were available, these safety concerns were associated with lower active travel use [44,45,50,51,54,75,87,97,98,99,100]. One study representative of eight different communities in Ghana showed that cycling was seen as unsafe because of vehicle driver negligence and because bicycles were viewed as illegitimate users of road space [45]. Evidence from two cities in Ethiopia revealed that cycling accidents and injuries were also due to casual attitudes towards the use of safety equipment by cyclists themselves [96]. Safety concerns around walking were centred on the perceived threat of personal violence from gangsters, thugs, thieves, and muggers [66,68,85,88,96,99,101,102]. Hazards and discomfort related to weather and topography were also salient: nine studies from Western, Eastern, and Southern Africa, and Panama, revealed that hot temperatures, humidity, the need to cross streams, flooding, and unfriendly terrain made walking challenging or undesirable [66,72,103,104,105]; whereas wind, hot temperatures, humidity, and hills were barriers for cycling [48,75,87,106]. Where alternatives were possible, severe weather events could promote mode shifting from active travel to public (or less commonly, private) motorised modes, as seen in Lagos (Nigeria) and Panama City (Panama) [101,106].

It appeared that transport infrastructure could influence active travel. For example, following the construction of 22 km of bicycle paths in Cape Town (South Africa), there was a 30% increase in the number of students who commuted by bicycle [107], with similar trends seen in Cameroon, Ghana, Mozambique, Nigeria, and Uganda [98,108,109]. By contrast, non-existent or poorly maintained pedestrian footpaths, bicycle lanes, pedestrian crossings, roadside barriers, street lighting, and signage were reported constraints for walking and cycling in both African and Caribbean regions [44,45,48,59,65,66,93,103,105,110,111,112,113,114], and sometimes contributed to a shift to motorised modes [54,101]. Despite awareness of its importance, active travel was not well represented in policy and planning, with evidence of bias against non-motorised modes by planners, lack of infrastructure provision, lack of commitment to promote pedestrian safety, and inexistent regulations to prevent blockage of and parking on footpaths [50,75,96,99,103,107,114,115,116].

• Indicator Signifying Opportunity for a Healthy, Sustainable, and Equitable Mobility System: Universal Public Transport

Rural, peri-urban, and unplanned settlements were typically not served well, or at all, by formal public transport, though paratransit (informal public transport) such as matatus and boda-bodas were widely available [56,62,69,71,74,117,118,119,120,121,122,123]. Introductory sections from multiple studies made it clear that in dense urban centres, formal public transport systems typically had colonial roots and were not designed to serve the local population; these deteriorated or collapsed in a post-colonial era, or due to civil war, and were replaced by paratransit in many places. Inadequate coordination among different actors in public transport delivery, corruption, poor management, slow bureaucratic procedures, and political meddling, coupled with low investment in road infrastructure (number of roads and maintenance), further contributed to the demise of formal public transport [120,124,125,126,127,128,129,130]. More recently, the worldwide economic recession and market liberalisation policies from the 1990s weakened an already struggling public transport sector [44].

Even where public transport existed, it was not affordable for the poorest urban and rural households, leading to a reliance on walking [91,107,131]. While public transport users were typically working class, different public transport modes followed different socioeconomic trajectories; for example, the likelihood of using buses decreased with vehicle ownership, income, and employment, whereas most rail (tram, train, and metro) users were employed or engaged in higher education [56,58,69,70,71,118,123,127,132,133,134,135]. Paratransit drew users from across the socioeconomic spectrum, apart from the richest [48,58,124,128,136]. One study from Jamaica found that women tended to be more reliant on public transport than men, who were more likely to have access to private vehicles [86]. Paratransit in particular was more likely to be used by women and younger people [58,122,137], though these groups reported problems with crime and harassment on paratransit and when using public transport in general [48,57,63,68,88,110,119,120,137,138,139,140,141]. Two studies from Nigeria showed that for women and the children accompanying them, additional risk was introduced through practical difficulties getting on and off transport due to dress and social norms related to sitting position [44,124]. Use of public transport increased with trip distance, time, and frequency [58,69,89,94,142]. High value was placed on public transport being readily available [48,63,97,101,120,122,128,137,139,143,144,145,146,147], and it could be faster and less stressful than using the car [110,132,135,136,148]. Reasons for not using public transport included unaffordability, poor coverage, delays, lack of fixed schedules or information, lack of transit stops, short operating hours, poor safety or driver behaviour (e.g., speeding) and lack of comfort [55,63,70,71,93,94,110,114,120,125,132,139,143,149,150,151].

In studies mainly drawn from Sub-Saharan Africa, paratransit was seen as providing a dense, frequent, flexible, convenient, and somewhat more affordable service [63,122,126,128,136,137,139,140], albeit being disorderly, unreliable, and unsafe [114]. Paratransit was valued for its ability to navigate traffic congestion, which was widespread and problematic during peak times [45,120,128,136,151,152], and because it could still access damaged or flooded roads [64,74,114,119,136,151]. Paratransit’s existence predominantly outside of regulatory structures added unique challenges in these contexts. Passengers did not benefit from the subsidies applied to government-controlled services like formal buses and trains [117,139], and operators from across Africa and the Caribbean were typically not able to obtain credit through formal channels [44]. This resulted in debt and cycles of poverty, and those policies which did exist were often hostile; for example, policies excluding paratransit from particular areas [77,114,127,153].

• Indicator Signifying Opportunity for a Healthy, Sustainable, and Equitable Mobility System: Transport Accessibility for All

Narratives drawn from across the African continent revealed that transport was unequally and inequitably accessible for different population groups. Low socioeconomic status individuals, people living rurally or peripheral to the city centre, women and girls, older adults, and those with physical disabilities had higher levels of immobility (inability to travel) than their counterparts [56,69,70,134,143,154,155,156]. Even when able to travel, these groups were disadvantaged in terms of transport accessibility, and this manifested in different ways. For example, low socioeconomic individuals and those living in peripheral areas tended to rely on walking and public transport to traverse long distances to access resources, opportunities, and livelihoods, and were often unable to afford motorised transport [61,69,70,71,118,154,157,158]. By contrast, in South Africa residents in more central locations were afforded more transport options facilitating trip making even by poor households [71]. For women and girls, harassment and crime during travel was a particular problem and travel was often restricted for this reason. In addition, women and girls typically had fewer funds available to pay for transport compared to male members of the household [69,72,88,89,90,124,141,152,156,159,160,161,162,163,164]. Older adults and those with physical disabilities experienced accessibility problems when getting on and off public transport due to overcrowding, high boarding platforms, low ceilings, and sliding doors [56,70,113,140,144]. The combination of immobility with cost and safety issues encountered when travelling resulted in unmet travel needs for these population groups.

• Indicator Signifying Challenge to a Healthy, Sustainable, and Equitable Mobility System: Poorly Controlled Urban Growth and Sprawl

The studies showed that informal, unplanned and poorly controlled urbanisation is a core feature of African cities [44,55,57,61,71,77,87,88,93,94,110,111,124,143,145,165,166,167,168]. Even though less is known about this phenomenon in the Caribbean cities, four studies from cities in Panama, Suriname, and Jamaica revealed similar emerging challenges in this region [79,106,169,170]. This informality was further reflected in livelihoods (the informal sector accounts for nearly 90% of total employment in sub-Saharan Africa), with associated diverse, irregular, and shifting mobility patterns [61]. Accessing livelihoods, social services, activities, or education therefore typically involved long distances and travel times [71,77,111,143,166,167]. As such, transport was perceived as a major and inescapable problem and source of stress by most peripheral residents [57,61,71,91,111,124,143,165,171]. For the wealthy, long distances could be overcome through private car ownership, but traffic congestion (generally and related to highway tolls) still resulted in long travel times [120,145]. The majority of residents from peripheral settlements needed to use multiple motorised modes and combinations of informal, formal, and private transport to reach destinations [61]. When areas were poorly serviced, relatively expensive motorcycle-taxis were often all that was available for the initial part of the journey [172]. Where buses were available, they tended to operate in mixed traffic without priority [106]. Due to a prevailing lack of integrated land-use planning (that has traditionally narrowly focused on economic development), construction of new transport infrastructure often disrupted existing neighbourhoods and resulted in the relocation of urban residents to the periphery. Residents of low-income neighbourhoods were especially prone to displacement and subsequent changes to mobility patterns [44,57,59,77,93,105,116,146,154,168,173]; this was the case, for example, with the construction of the Nairobi-Thika highway in Kenya, jointly financed by the African Development Bank and the Chinese government.

• Indicator Signifying Challenge to a Healthy, Sustainable, and Equitable Mobility System: Socioeconomic Barriers to Mode Choice

The mobility and mode choices of low socioeconomic individuals and households strongly depended on cost (either ticket cost or the cost to purchase and maintain a private car, motorbike, or bicycle) [55,70,87,97,111,138,165,174]. Being unable to afford to travel was common, and poverty was negatively and strongly associated with lack of access to motorised transport [48,56,60,69,111,159]. An inability to afford travel could reduce access to basic needs, including food, education, and healthcare [44,62,72,146,153,175]. This was particularly salient for women, children, and those who were older, sick, or disabled [90]. Women in particular experienced multiple financial barriers to travel [56,90,139,161,162,163]. The nature of household-related travel tended to require multiple stops, making travel more expensive due to having to pay multiple single fares [141]. Poor, unplanned, and peripheral neighbourhoods tended to have a lower number of motorised travel options available, and residents were highly price-sensitive in relation to transport [62,69,70,71,111,123,143]. Absolute and relative transport costs were high. Residents (particularly the poor and unemployed) spent a substantial proportion of their income on transport, although this varied by modes used [44,70,71,111,123,139,153,165,166,174,176]. Both in Africa and in the Caribbean, paratransit fares were unstable and could fluctuate depending on the weather, the direction of travel, and whether the route included a toll road [44,106,120], itinerant workers appeared to feel the impact of fluctuations in transport related costs more than other types of workers. In the absence of public transport subsidies, the fares charged by private operators were high in comparison with users’ incomes, indicating that public transport expenditure was socially regressive [111,154,177]. So-called ‘captive’ walking and, to a lesser degree cycling, was often the only mode available when motorised transport was unaffordable. Thus, these forms of active transport were seen as necessities that would be discarded when wealth increased.

• Indicator Signifying Challenge to a Healthy, Sustainable, and Equitable Mobility System: Mode-Shifting to Private Vehicles

Private vehicle ownership and use were low, but increasing rapidly. Three studies from Panama and Ghana showed that a switch to a private car was seen as a symbol of progress [60,106,122]. One study from Lagos revealed that subsidised petroleum prices and unrestricted imports of second-hand vehicles facilitated this burgeoning trend by putting car ownership within reach of the middle class [120]. As age and income increased, middle-class users switched to private cars or formal public transport, and away from walking and paratransit [60,62,94,95,126,139,168]. Correspondingly, household car ownership was associated with lower levels of active transport as seen in two studies from urban settings in Western Africa [49,84]. Car users were typically resistant to changing modes, perceiving public transport to be poor and cars as more reliable and convenient than other modes [60,71,178]. In addition, private motorbikes were seen as a way to circumnavigate traffic jams in two studies from the large cities of Dar es Salaam (Tanzania) and Accra (Ghana) [61,152]. Regardless, private vehicles provided mobility for a minority of the financially better off, with the majority relying on other forms of transport [60,71,77,94,117,119,138,165].

3.2.2. Meta-Method

In terms of the variety and number of methods (i.e., data collection strategies) employed, the quantitative studies were notably more homogenous than the other study design types (File S12). Overall, they employed five different methods, compared with 17 methods for mixed-methods studies and 23 methods for qualitative studies. Over 90% of the quantitative studies used surveys. By comparison, the most common methods reported in mixed-methods studies were surveys (utilised in 27% of those studies), followed by participant interviews (16%), key informant interviews (14%), and focus groups (10%). The qualitative studies were the most varied, with focus groups (utilised in 20%) being the most reported method. The vast majority (94%) of the quantitative studies used a single method, whereas 95% of the qualitative studies (along with all of the mixed-methods studies) used two or more methods.

3.2.3. Meta-Theory

A quarter (24%) of all the studies reported that a specific theory or perspective had been used to guide the research (File S13). Specification of theory was rare in the quantitative studies (14%), while 50% of the qualitative and 30% of the mixed-methods studies specified a theory. The mobilities paradigm [179] was the most reported theory in the qualitative studies, reported by 60%. Mixed-methods studies were more theoretically diverse, and included theories drawn from the behavioural sciences (13% of the studies), human geography (50%), sociology (25%), and political science (13%).

3.2.4. Meta-Synthesis

Throughout this last step, we built up our interpretation and understanding across varying contexts. Our main findings are summarised in

Table 6. Meta-synthesis findings.

Indicator	Findings
High use of active forms of travel	There were high levels of active travel, predominantly walking
Universal public transport	Public transport was widely available, but mainly in the form of paratransit rather than formal public transport Variability in levels of active travel and use of paratransit was underpinned by significant inequities, with a lack of choice and over-reliance on these modes in some groups, despite their associated hazards and opportunity cost
Transport accessibility for all, including marginalised groups	Transport accessibility was poor for marginalised groups, including low socioeconomic status individuals, people living rurally or peripheral to cities, women and girls, and the elderly, sick and disabled
Poorly controlled urban growth and sprawl	Poorly controlled urban growth and sprawl, with associated informality, was a salient aspect of the study settings, resulting in long travel distances and the necessity of motorised modes Sprawl was associated with class segregation with residents of low-income neighbourhoods especially prone to displacement and subsequent changes to mobility patterns
Socioeconomic barriers to mode choice	There were marked socioeconomic barriers to mode choice along with some barriers which were significant enough to prevent any travelling at all
Mode-shifting to private vehicles	While mode-shifting to private vehicles was increasing, it was not yet widespread. Cars were seen as preferable to public transport and as a badge of success, and ownership was becoming more financially attainable for more people

.

Across the indicators, we found evidence of existing assets to realise ‘human urbanism’, including high levels of active travel, good coverage of paratransit, and relatively low private car use. We also identified core challenges, namely urban sprawl and informality, socioeconomic barriers to travel and poor transport accessibility for some population groups.

4. Discussion

In this systematic review and meta-study, we drew together quantitative and qualitative evidence on travel behaviour in Africa and the Caribbean. Our meta-synthesis of the potential to realise the ‘human urbanism’ stage of the Human Mobility Transition revealed the complexity of exposures associated with travel behaviour in these contexts. The ways in which these factors intermingled had the potential to both help and harm the realisation of a healthy, sustainable, and equitable mobility system. For example, high levels of active travel may be seen as an existing asset through benefits to human health and planetary sustainability, but a differential reliance on walking manifested in particular population groups (i.e., women and girls, and low socioeconomic individuals) in ways that were not equitable. We operationalised a set of ‘human urbanism’ indicators in order to structure our analysis, also demonstrating that these indicators cannot exist in isolation and that all must be developed in order to realise the potential of the mobility system. Some well-intentioned actions ‘for’ some indicator could produce unintentional impacts ‘against’ others. For example, the regulation of paratransit is often seen as a stepping stone towards universal (formal) public transport, but could also reduce transport accessibility in low income groups, promote mode-shifting to private vehicles, and promote unjust governance of informal transport providers.

The meta-method and meta-theory components of our review revealed that the quantitative studies exploring travel behaviour typically focused on individual-level exposures, while the studies with a qualitative component reported a wider spectrum of exposures including features of the built environment, gender norms, and socio-political influences. The quantitative studies demonstrated considerable methodological similarity; surveys were the preferred (and in most cases, the only) method. By contrast, the qualitative and mixed-methods studies ranged in the type and number of methods they utilised in order to provide insight into participant experience. Most studies lacked an underpinning theory or perspective. The mobilities paradigm was the most used theory, which was appropriate for a focus on power relations, equity, and social justice. Overall, the lack of theory-driven research is concerning, as this research is not atheoretical; rather, the underlying values and assumptions have not been made explicit. We do not align ourselves with the notion of ‘objective science’ being somehow hermetic to social, cultural, and political forces. It is critical that these forces influencing travel behaviour are given more adequate theoretical consideration in future research.

4.1. Our Research in Context

Our finding that daily travel is dominated by walking and paratransit aligns with previous research [180], though active travel and paratransit are not well recognised in transport planning in these regions [181,182,183,184]. The challenges we identified around urban growth and sprawl and the significant difficulties travelling for some population groups, particularly the poor [30,185,186], are indicators of transport and mobility injustice, as examined by Jacobs [187] and Harvey [188]. Additionally, these findings highlight that ineffective mobility systems are ultimately a product of uncoordinated and reactionary urban planning, unregulated land use, and subsequent land use conflict, as seen in previous research [18,189]. Our findings align with Boeing and colleagues [190] and support their position that creating living spaces that meet size and population density thresholds, offer local job opportunities and local destinations to livelihoods, and guarantee inclusive access to public transport could provide an opportunity for healthier, more sustainable, and socially equitable environments. Previous research has shown that large African cities have a relatively less compact urban form than similar cities in other regions of the world [191]. In line with this, our findings revealed that access to livelihoods, social services, activities, or education regularly involves long distances and travel times. In order to achieve functioning mobility systems, it is recommended to implement policies that ensure sufficiently dense living areas with an optimal distribution of jobs and amenities.

While our findings were drawn from evidence in both Africa and the Caribbean, we found an important relative gap in the Caribbean region. This research gap could be due to differences in population size between Caribbean (44 million inhabitants in 2019) and Africa (1.3 billion inhabitants in 2019), which could also lead to the Caribbean often being studied as part of the Latin America and the Caribbean (LAC) region as a whole. The emerging spatial complexity in the Caribbean (as found by Sheller [183]) could well drive future travel behaviour research in this region, just as marked spatial complexity in Africa has fostered research on this continent.

4.2. Strengths and Limitations

The review’s strengths include an exhaustive search strategy developed with a medical librarian, allowing us to search ten databases from different disciplines with no language restrictions. We included both qualitative and quantitative peer-reviewed evidence alongside grey literature, using accepted guidelines for quality appraisal, conduct, and reporting. Our large, multidisciplinary, and geographically diverse team drawn from Africa and the Caribbean, inter alia, brought a variety of perspectives to the interpretation of review findings. The Human Mobility Transition allowed us to organise and frame the evidence. However, important limitations remain. First, the narrative summary of the qualitative data may be considered over-simplistic, although we note that it complements two separate meta-ethnographies [30,31] conducted using subsets of the same large dataset. Second, most of the retrieved evidence pertains to Africa, and mainly from countries in Western, Eastern and Southern Africa. Very few studies were found in relation to Central Africa, Northern Africa, and the Caribbean, limiting our capacity to draw conclusions for these regions. These identified gaps suggest directions for future research. Third, we acknowledge that our use of the Human Mobility Transition framework may imply a linear and universal process of change in which there is a singular path. We make no such claim; indeed, our findings reveal that multiple temporalities coexist. There are different travel behaviour narratives shaping different mobility systems that are true for a group of people at that moment in time and place. Regardless, there are remarkable consistencies in the data of mobility transition, and there are pressing health and environmental concerns that support a move towards ‘human urbanism’.

4.3. Further Research

This review highlights that a preponderance of studies exploring individual characteristics is a missed opportunity to investigate the myriad of other potential influences on travel behaviour. Despite the current climate emergency and the role mobility systems play in climate change [1,2,3], we found few studies reporting associations between travel behaviour and the natural environment in these settings. Similarly, in the context of rapid urbanisation and land-use challenges, we found a research gap in relation to the associations between travel behaviour and the policy or political environment. We found that the reviewed literature classified private vehicle ownership as an individual characteristic and as a proxy for higher socioeconomic status. This is a missed opportunity in the transport literature to explore motorisation as a commodity and as a social process with serious public health implications, as described by Woodcock and Aldred [192]. More focus on the political economy of transport and the commercial determinants of health, more qualitative and quantitative research on a wider range of exposures, more critical and creative use of study methods, and more thoughtful application of theory is needed to improve understanding of travel patterns and mobility systems in future research.

4.4. Contribution to the Field

We found evidence that current mobility systems are perpetuating cycles of poverty and disadvantage. As such, policymaker attention to active and public transport accessibility for all, and especially for the most vulnerable, is essential, and aligns with the SDGs. In addition, land-use planning remains a key issue for making progress towards ‘human urbanism’. This will require a shift in priority from economic growth and favouring of elites to a focus on broader population needs and the rights and wellbeing of ordinary people; a shift from individualised mobility with a car as an aspirational ‘good’ towards car-free societies [192,193]. Our findings suggest a need for policy and practice based on locally conceptualised ideas, informed by public health and climate evidence, that reflect the demands of the local community, and implemented by government-citizen partnerships (citizens’ active participation in decision making). Addressing these not only has the potential to target the real challenges rooted in the dynamics that occur in that place, but also to do it using already existing local assets. Overall, research and practice framing dysfunctional mobility as a structural or systemic challenge, and using a more holistic, systems thinking approach is needed.

In conclusion, we found evidence of existing assets (high levels of active travel, good coverage of paratransit, relatively low private car use) as well as core challenges (urban sprawl and informality, socioeconomic barriers to travel, and poor transport accessibility for some population groups) to the realisation of healthy, sustainable, and equitable mobility systems in Africa and the Caribbean. To realise ‘human urbanism’, integrated planning policies recognising the linkages between health, transport, and equity are needed.