Evidence From the Decade of Action for Road Safety: A Systematic Review of the Effectiveness of Interventions in Low and Middle-Income Countries

Introduction

With over 1.5 million deaths annually, road traffic injuries now account for the sixth highest cause of disability-adjusted life years lost globally (1). Road safety has become a major public health concern in many countries, resulting in increased attention from the international community. The United Nations integrated road safety into two of the 17 sustainable development goals (SDGs), SDG 3 and SDG 11 (2), and proclaimed 2011 to 2020 the Decade of Action for Road Safety (3).

Traditional approaches failed to capture the complexity of the road traffic systems. To address this, key principles of systems thinking were adopted in the UN Global Plan for the Decade of Action (3, 4). The plan provides an overall framework for countries to improve road safety (3) and is based on five pillars—road safety management, safer roads and mobility, safer vehicles, making road users safer, and improved post-crash response and hospital care—encouraging countries to pay more attention to vulnerable road users and shifting the responsibility of incidents from road users to the designers of the transport system (3).

Despite the increased global attention and progress in policy-making at national level, the number of road casualties increased in 87 low- and middle-income countries (LMIC) since 2013 (5, 6). LMICs bear more than 90% of deaths due to road traffic injuries, despite containing only 60% of the world’s motor vehicles. Death rates due to road traffic injuries in LMICs are three times higher than in high-income countries (HIC) (27.5 vs. 8.3 per 100,000 population) (6).

Even though findings about effective interventions from high-income countries (HIC) are potentially transferable to low resource settings, different context-specific factors such as costs, feasibility and barriers to implementation should be taken into account (7). Also, the traffic mix in LMICs is different than in HICs; the majority of evidence coming from HICs focuses on preventing injuries to vehicle occupants, but pedestrians and motorcyclists are the most vulnerable road users in LMIC settings (8).

During the past decade, few studies reviewed the effectiveness of road safety interventions in LMICs, however the results were limited to single interventions mostly in legislation and law enforcement category (9–11). A systematic understanding of interactions among different components of the system contributing to road safety is still lacking and very little is known about other road safety components, such as speed management, improving vehicle design, road infrastructure and post-crash response and their interactions in the system (12).

Following the end of the Decade of Action for Road Safety in 2020, we systematically reviewed the recent evidence on road safety interventions in LMICs. Our aim was to evaluate the effectiveness of road safety interventions in LMIC settings through the lens of the principles of the UN’s Global Plan for the Decade of Action.

Methods

Search Strategy and Selection Criteria

We conducted a systematic review and a narrative synthesis based on PRISMA guidelines as described in Supplementary Appendix A1 (13, 14). We included original research studies evaluating quantitative effects of road traffic safety interventions implemented in LMICs (5), reporting mortality, injury or crash as the primary outcome, online publication during January 2011–August 2019 in English, French, Spanish, Portuguese and Persian.

We used the World Bank income group classifications for fiscal year 2019, in which 138 LMICs are stratified by GNI per capita: low-income (under $1026), lower-middle income ($1026–$3,995) and upper middle-income ($3,996–$12,375) (5).

We developed a search strategy including terms to identify population, intervention, outcome and keywords using five pillars of the Global Plan as a framework (3). Then we used synonyms and variants of search terms to find relevant articles (15) in nine electronic databases including Medline, Embase, PsycInfo, Scopus, Web of Science, Cochrane library, Global Health Library, ProQuest and Transport Research International Documentation (TRID). We tested our search strategy by checking the inclusion of ten relevant studies in the field in our Medline search; then we modified the search for other databases as shown in Supplementary Appendix A2 Database search and retrieval of results from each database were done in August 2019.

After importing references to EndNote X9, we followed seven steps of deduplication in EndNote recommended by Bramer et al. (16). Then study authors (MT and ZT) independently conducted screening of titles, abstracts and full texts. We recorded the reasons for exclusion of studies and resolved disagreements by discussion to reach a consensus, or deferred to a third author (DC) for the remaining cases. We registered the protocol for our systematic review in the PROSPERO database (CRD42020197267).

Data Synthesis

The primary outcome measures were crash, injury or mortality due to road traffic crashes. Two reviewers (MT and ZT) independently performed a risk of bias assessment of screened studies using the Effective Public Health Practice Project (EPHPP) tool (17). Any disagreement in this step was resolved by discussion to reach a consensus or deferred to a third author (DC) for the remaining cases (see Supplementary Appendix A3). In this step, we included controlled trial, case-control, cohort, quasi-experimental (interrupted times-series, before and after) study designs but excluded other study designs. In the next step, using a pre-defined data extraction form, two authors (MT and ZT) extracted variables of interest. We tested the data extraction tool with ten included model studies. Extracted variables are described in Supplementary Appendix A4. We classified interventions into seven categories using an adapted road safety intervention categories as defined in the WHO Save LIVES road safety technical package; see more details in Supplementary Appendix A5 (18).

The decision on the synthesis method was made after the risk of bias assessment to ensure that a sufficient number of studies were available reporting standardized metrics. Due to variability in statistical tests and different outcomes across studies, we decided to synthesize data using vote counting based on the direction of effect. We used a harvest plot to visualize the distribution of the evidence (19).

Results

The initial search retrieved 22,534 records. After deduplication, and screening for title-abstracts and full-text, using a predefined list of inclusion/exclusion criteria and assessment for risk of bias, 33 studies were included in the systematic review, of which 18 rated strong based on EPHPP quality appraisal tool as outlined in Figure 3 and Supplementary Appendix A3. Included studies were from 17 LMICs shown in Figures 1, 2. Twenty-seven studies were conducted in upper middle-income countries, four studies were conducted in lower middle-income countries, and two in low-income countries.

FIGURE 1. Study selection flow diagram. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses, for Road safety interventions in low and-middle income countries, 2011–2019.

FIGURE 2. Geographical distribution of studies by type of road safety intervention in low and middle-income countries (systematic review, 2011–2019).

We organized the results around six categories by type of intervention as seen in Tables 1, 2. The categories with the most supporting evidence were enforcement and traffic laws, speed management, leadership (see harvest plot in Figure 3). Here we describe the range of effect in each category.

TABLE 1. Characteristics of included studies classified by the type of road safety intervention (systematic review, low and middle-income countries, 2011–2019).

TABLE 2. Summary of outcomes of road safety intervention (systematic review, low and middle-income countries, 2011–2019).

FIGURE 3. Summary of direction of health—related outcomes from included studies-harvest plot (systematic review, low and middle-income countries, 2011–2019). Distribution of the evidence for Six categories of interventions (Enforcement and traffic laws is presented with three subcategories) is plotted with the number of studies along the Y-axis. Each study is represented by a mark in each row for which that study had reported relevant results. Studies that rated STRONG using EPHPP quality appraisal tool are indicated with green bars, and studies rated MODERATE with yellow bars. The height of the bar indicates number of the studies.

Enforcement and Traffic Laws

We identified 18 studies in this category including interventions related to lowering blood alcohol concentration (BAC) limits, multifaceted road safety laws and mandatory use of safety equipment law.

Lowering BAC Limits

Of seven studies evaluating the effects of lowering BAC limits, four studies showed positive effects (see Figure 3). Brazil reduced the BAC limit for drivers and strengthened punishment by introducing the “Dry Law” (Lei Seca) in 2008. A 16% decrease in the number of mortality in Sao Paulo was reported 2 years after enactment of the Dry Law (20). However, looking at the trends of mortality rate, Volpe et al. found no significant change 5 years after enactment of the law (21). Despite the absence of any abrupt change in health-related outcomes, the law’s enforcement strengthened the pre-existing downward trends of mortality in some road users such as bicyclists and motorcyclists and pedestrians (22). Enabling effective enforcement of the law in 2012 led to a minor, yet steady decline in fatal crashes (23). Likewise, in China, enforcement of the criminal amendment for drinking and driving to the road safety law strengthened the pre-existing reduction trend in mortality and injuries (24, 25). Although enforcement of the law was better in cities, the reduction in years of life lost (YLL) was more evident in suburban areas (25). After enforcing the Life-Saving Law (“Ley Salvavidas”) in Mexico, a reduction was observed in the number of crashes as well as a downward trend in alcohol-related injury and mortality, but the authors could not find a strong association between the enforcement of the law and health-related outcomes (26).

Multifaceted Road Safety Laws

We found six studies evaluating the effects of multifaceted road safety laws, of which three showed positive effects (see Figure 3). In China, the road traffic safety law was enacted in 2003. Sheng et al. and Aney et al. found that while there was a reduction in the number of crashes and casualties, the law was more effective in preventing less severe crashes (27, 28). Similar results were observed in Mexico where a two-phase road safety initiative called IMESEVI—Iniciativa Mexicana de Seguridad Vial—was launched. Although there was no significant difference in injuries and mortality compared to the period preceding the implementation of the law, a reduction in the rate of crashes after the second phase of the intervention was reported (29). Enactment of the Brazilian Traffic Code in 1998 contributed to a reduction of 9.69 death per year from 1980 to 2014 with the highest decline among young adults (30). Similarly a multi-faceted road safety law in Ethiopia was followed by a rapid decrease of 19.2 and 12.4% in property-damage-only crashes and casualties respectively (31). And finally, in Botswana, a 30% levy on alcohol products complemented by the road traffic act contributed to a decrease in road traffic crashes and deaths, although the results were not sustained over time (32).

Taken together the results suggest that introducing a multifaceted road safety law was more effective on less severe crashes. However it is important to bear in mind this category includes various types of road traffic laws which impact health-related outcomes through different mechanisms, therefore results should be interpreted with caution.

Mandatory Use of Safety Equipment Laws

Four out of six studies in this category showed positive effects following the intervention (see Figure 3). In Serbia and Brazil, mandatory use of age-appropriate child restraints resulted in a reduction of injuries up to 8% among children (33, 34), but in Brazil the law had no effect on non-white children (34).

Two studies investigated the effects of a motorcycle helmet law on health outcomes—both emphasizing the diversity of outcomes among grographical and population subgroups. In Vietnam, law imposing severe penalties for offenders, resulted in a downward trend in potential years of life lost (PYLL) in 42 of the 61 provinces. Likewise, at national level a gradual reduction in PYLL (−18.1 per 100,000 person-months) was observed 6 months after the enactment of the law (35). In India, Gupta et al. studied the influence of the cultural setting that exempted women and Sikh men from wearing a helmet on road safety. Assessment of 224 male driver-female pillion crashes, revealed that women had lower compliance with the law (4.17% F, 61.8% M), and men had a 56 and 58% lower risk of deaths and serious head injury, compared to female counterparts (36).

Finally, in the case of seatbelt legislation we found one study from Iran. The results indicated with the increase of seatbelt use, a decline was observed in the severity of injuries in the first and second year after introducing the law (37).

Speed Management

We identified four studies evaluating the effects of speed management interventions, generally in favour of the interventions (see Figure 3). Establishing a new speed limit in Brazil strengthened the downward trend in the annual change in mortality after the intervention (38). Traffic-calming measures, another speed management intervention, mainly targeted safety of pedestrians. Two studies showed that transverse rumble strips and parallelogram-shaped pavements before pedestrian crosswalks are likely to reduce vehicle-pedestrian crash frequency by 25 and 21% respectively (39, 40). Likewise, the presence of speed tables, speed humps and speed bumps, was associated with a lower risk of pedestrian fatality (41). Meanwhile, fixed cameras detecting traffic violations in Colombia resulted in mixed effects: from a 57% increase in the total number of crashes and crashes with injury in the intervention group to a 5–6% reduction in the number of crashes (42, 43).

Leadership on Road Safety

We identified five studies evaluating the effects of education and public awareness interventions on road safety, four of which reported positive effects (see Figure 3). An initiative focusing on raising awareness among passengers and drivers of long-distance minibuses—Matatu in Kenya resulted in a 7.5 and 6% reduction in property-damage-only crashes and crashes with casualties respectively (44). Education initiatives were found to have minor to moderate effects on health-related outcomes. In China, an education initiative about bicycle safety for seventh-grade students reduced the incidence of bicycle injuries in children (45). Similarly, in Thailand, a safety-riding program targeting motorcyclists, reduced motorcycle-related injuries by 30% (46). Two studies from Uganda and South Africa showed that safety training and the provision of safe equipment through ride-sharing programs results in a reduction in crash risk (47, 48).

Survival After a Crash

We found two studies showing different results in this category (see Figure 3). In Iraq, the presence of prehospital trauma care was associated with an 8% deaths rate compared to 44% in absence of pre-hospital trauma care (49). In contrast, no significant change in the number of road traffic deaths was observed after the introduction of the central emergency medical service in India (50).

Infrastructure Design and Improvement

The most striking result to emerge from this category is the increase in the risk of pedestrian-vehicle collisions in presence of motionless green/red light signalization for pedestrian and vehicles. Whereas a negative association was found between police presence and the number of crashes in signalized intersections compared to non-signalized intersections (51).

Vehicle Safety Standards

We found only one study evaluating the effect of the antilock braking system (ABS). The results showed that vehicles with antilock braking system are less likely than vehicles with a conventional braking system to be involved in road traffic crashes due to brake failure (52).

Discussion

Our study aimed to systematically review the recent evidence on the effectiveness of road safety interventions in LMICs in light of principles of the Global Plan for the First Decade of Action for Road Safety. Eighteen out of 33 studies evaluated interventions in legislation and enforcement category, while some components of the road traffic system such as vehicle safety standards and road infrastructure and design were understudied. In most studies, only single interventions were assessed, omitting discussions around the complexity of road traffic system, relevant contextual factors, and its influence on the performance of interventions.

Our findings were in line with those of previous reviews, showing that most studies focused on changing road users’ behavior via enforcement, traffic laws, education and public awareness (9, 11, 53). Therefore, despite the recommendations of the Decade of Action to shift the burden of responsibility from road users to designers of the system, the role of designers in the system such as policy-makers, road managers, police, politicians, health sector, education system and etc. remained ignored (3). Our review exposes the scarcity of robust scientific evidence for some aspects of road safety—such as road infrastructure and design and vehicle safety standard—gaps that were revealed in previous literature and have remained unchanged after the Decade of Action (10, 11).

Another major finding of our review was that the existing scientific evidence in LMICs was focused on single interventions while little is known about the interaction between those interventions and components of the road transport system. This implies that the evidence around road safety is driven by component-oriented approaches while contextual factors are overlooked and road users are assumed to be the sole cause and changeable component in accidents (54). However a system is not merely a collection of single components (55) and road transport systems have characteristics of complex adaptive systems (56). In these systems, the relationships among the different components or stakeholders, emergent behaviours in the system, and contextual and historical factors are essential to improving performance (55, 57). In our review, single interventions were found to have short-term impacts, and their positive effect decreased over time. Salmon and Lenne argued that only focusing on one component of the system in isolation and ignoring other relevant factors will result in a diminished impact of the intervention (54). Furthermore, single interventions are less effective unless complemented with other interventions (58); Bambach et al, found that cumulative benefit of a combination of road safety interventions such as roadside barriers, helmet use, speed management and ban of alcohol consumption, resulted in synergies and a stronger positive effects (59).

Regarding evidence production, only 17 out of 138 LMICs were represented in our review. In line with previous literature, we found that a country’s income level was directly associated with representation in the production of scientific evidence: the number of studies from low-income and lower-middle-income countries was limited to 2 and 4 out of 33 studies respectively. These results reflect those of Perel et al. who also found that only 6 out of 236 road safety studies included in Cochrane systematic reviews were conducted in LMICs (60). Furthermore Zou et al. showed that among the top ten countries accounting for the production of 80.56% of road safety literature, only one LMIC—China—exists (61). This underrepresentation may partly be explained by factors related to the process of conducting research in LMICs and scientific publication systems (62). To measure the effectiveness of an intervention, researchers need information about characteristics of crashes, but many LMICs do not have a reliable information system recording this information (6). Another important factor affecting the conduct of research is the limited global funding opportunities for road safety. The available funding are donor driven and are dominated by HICs (63). The under-representation of LMICs in research publications goes beyond road safety research. In terms of the publication process, researchers in LMICs also face challenges. A review of the editorial boards of 27 global health journals showed that only 24% of editors were from LMICs and there was no editor-in-chief based in a low-income country (64). This lack of representation could be the result of institutional racism in publication process (65). In addition, authors from LMICs often cannot afford publication fees and consequently publish in journals that are not indexed in European or American (“global”) databases. This disparity leads to so-called “Academic-colonialism,” where the scientific publication system tends to empower researchers from HICs (66).

Finally, as mentioned before, the Global Plan for the Decade of Action encourages countries to set priorities for vulnerable road users—pedestrians, cyclists and motorized two and three-wheelers—accounting for half of the road traffic deaths globally (67). These users mostly benefit from speed management, alcohol control and enhanced visibility measures (68). In our review, we found 16 studies evaluating interventions such as lowering BAC limit, traffic calming measures, traffic signals and mandatory helmet law directly target vulnerable road users. However, we did not find studies evaluating the impact of proven effective interventions such as alcohol ignition interlock (69), street lighting (70), conspicuity aids and separate bicycle lanes (71) for vulnerable road users.

Selecting LMICs as the population for this review was both a strength and limitation of our study. We acknowledge that LMICs are very diverse in terms of resources and other contextual factors that might affect the subject of this study, however we used this term for general characteristics of countries in these income levels and to highlight the gaps in the literature among these contexts.

Conclusion

Our study has shown the lack of a systems lens in evaluating road safety interventions in LMICs. We found that the same components of the system are often studied in isolation and little is known about the interactions between different components at a systemic level. The majority of the interventions are trying to change road users’ behaviour with education, legislation and enforcement and there is very little focus on eliminating hazards from the road system. A holistic understanding of the road transport system requires shifting from the prevailing paradigm of “fix the road user” to the systems thinking approach to “fix the system” while accounting for synergies and interactions among system components. A prerequisite in the application of systems approach in road transport systems research and practice is the availability of a good quality crash data system. This entails collaborating with relevant stakeholders and taking steps to improve road safety data systems. Until countries have a strong information system to collect various crash characteristics, the integration of the different information systems and using methodological approaches accounting for missing data will be essential. We are currently investigating ways of improving road traffic death registration systems in LMICs using systems thinking approaches such as process mapping and modelling, social network analysis (72).

Addressing the evidence gap in scientific publications from lower-income countries is a long-term but necessary process. It requires capacity strengthening by creating research networks that reach also into LMICs. The scientific community needs to give more visibility to evidence from LICs led by local scientists, by providing funding opportunities and facilitating publication processes to collect good quality evidence.

With the beginning of the second Decade of Action for Road Safety in 2021, we expect our findings to be valuable for informing global and national efforts towards designing inclusive, safe, resilient and sustainable societies, as well as the comprehensive research agenda needed to support such practice. In light of our findings, we recommend taking a step further from component-oriented approaches to applying systemic approached including systems thinking approaches in research and practice. This entails efforts to study complexity of the system as a whole and identifying international, regional and local actors and processes to improve accountability for improving road safety in LMICs.

Author Contributions

MT and DC designed the study protocol and developed the search strategy. MT and ZT-K conducted screening and data extraction. All authors contributed to the synthesis of results and writing and editing drafts. MT wrote the first draft and all authors reviewed and provided inputs for subsequent drafts. All authors prepared and approved the final article.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

We gratefully acknowledge the support from Swiss Government Excellence Scholarships for Foreign Scholars and the SSPH + Global PhD Fellowship Program in Public Health Sciences funded by Marie Skłodowska-Curie Actions (Horizon 2020—COFUND).

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.ssph-journal.org/articles/10.3389/phrs.2022.1604499/full#supplementary-material

References