A systematic review of risk factors associated with road traffic crashes and injuries among commercial motorcycle drivers

Abstract To effectively reduce road traffic crashes (RTCs) and injuries interventions should be based on firm evidence regarding risk factors of RTCs and injuries in that specific population. Therefore, we undertook a systematic review to determine risk factors of RTCs and injuries among commercial motorcycle drivers. Searches were performed from inception to May 2022 in Medline, Embase, Cochrane Library, Web of Science Core Collection, PsycINFO and Cinahl, along with registers and reference lists. Inclusion criteria were commercial motorcycle drivers, quantitative observational studies, and RTCs and injuries. The search resulted in 1546 articles, of which 20 met the relevance and quality criteria. Of the 20 articles, 17 were cross-sectional, 2 were case-control studies, and one was a cohort study. Close to half of all articles (9) came from sub-Saharan Africa. Risk factors with consistent association with RTCs and injuries were young age, low education level, alcohol consumption, speeding, mobile phone use, non-helmet use, risky driving behaviours and long working hours. There was inconclusive evidence for driver’s training, work schedules, motorcycle ownership, experience, dependents number, and marital status. More robust designs such as case-control or longitudinal studies are required to gain a comprehensive understanding of the antecedents of RTCs among commercial motorcycle drivers.


Introduction
Motorcycles for commercial purposes are becoming popular, especially in Sub-Saharan Africa (SSA) and Latin America (Ehebrecht et al., 2018;Hagen et al., 2016) and are increasingly responsible for a large proportion of road traffic crashes (RTCs) and injuries in SSA (Bishop & Courtright, 2022;Ehebrecht et al., 2018).Compared to car drivers, motorcycle drivers are at an increased risk of a crash and crash-related injuries due to the unstable nature of the motorcycle and lack of external protection (Horswill & Helman, 2003;Keall & Newstead, 2012).The risk may be higher for commercial compared to non-commercial motorcycle drivers because they spend significantly more time on the road and thus are more exposed to traffic hazards (Ehebrecht et al., 2018;Nguyen et al., 2018).A common definition of a commercial motorcycle driver is someone who operates a motorcycle for commercial purposes.In this review, a commercial motorcycle driver includes drivers who carry passengers, deliver goods and services on motorcycles or both.It comprises traditional motorcycle drivers, who usually have a designated parking station to wait for customers, and delivery drivers who perform deliveries.Many traditional drivers have recently enrolled in ride-hailing services and can be serviced using apps (Bishop & Courtright, 2022;Chalermpong et al., 2022;de Souza Silva et al., 2018).A considerable number of motorcycle drivers in SSA work in informal jobs without protection and guarantees, do not use protective gear, e.g.helmets, operate poorly maintained vehicles and lack proper training (Da Silva et al., 2012;Francis et al., 2023;Tumwesigye et al., 2016).Altogether, these may significantly increase their crash and injury risk.
There have been some reviews on the risk factors of RTCs and injuries among two-wheelers working commercially, including commercial motorcycle drivers (Konlan & Hayford, 2022;McKinlay et al., 2022).McKinlay et al. 2022 conducted a review focusing on two-wheeled delivery drivers and identified work-related risk factors among significant contributors to injuries and unique to delivery drivers compared to non-commercial drivers.However, this review focused solely on delivery drivers whose risk profile may differ from motorcycle taxi drivers.Another review by Konlan and Hayford in 2022 qualitatively examined factors associated with motorcycle-related RTCs in Africa from 2016 to 2022.Their findings indicated working for extended periods, having little rest between days and ownership as important risk factors for RTCs among motorcycle taxi drivers.Similarly, this review was confined to countries in SSA, which limits their extrapolation to other countries, especially in Asia (Irawan et al., 2020;Nguyen-Phuoc, Oviedo-Trespalacios, et al., 2020) and Latin America (Hagen et al., 2016;Márquez et al., 2018;Nguyen-Phuoc, Oviedo-Trespalacios, et al., 2020) where commercial motorcycles are very common.In addition, both reviews were scoping in nature, with broader inclusion criteria and included case series lacking a comparison group and unsuitable for determining risk factors.
Although there is a growing body of research on motorcycle-related RTCs and injuries, no systematic review has summarised the evidence on risk factors for motorcycle-related RTCs and injuries among commercial motorcycle drivers.Therefore, the article aims to identify, appraise and summarise available epidemiological research on risk factors of RTCs and injuries among commercial motorcycle drivers.

Methods
The reporting of this review follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.The protocol was registered in the International Prospective Register for Systematic Reviews (PROSPERO) number (CRD42022332255).

Inclusion criteria
We applied the following inclusion criteria: • Population: motorcycle drivers working for commercial purposes, including drivers carrying passengers, motorcycle-delivery drivers and the hybrid (doing both) • Exposure: any risk factors for road traffic crashes or injuries • Outcome: road traffic crashes (including road traffic injuries) • Study design: cross-sectional, cohort and case-control studies • Setting: no restriction • Publication language: no restriction • Time: no restriction We excluded studies if they • reported on road traffic crashes and injuries among non-commercial motorcycle drivers • were case studies, qualitative studies, correspondence or commentaries • did not provide enough information to allow the discernment of commercial vs non-commercial motorcycle drivers.

Search strategy
The literature search was conducted using the following databases: Medline, Embase, Cochrane Library, Web of Science Core Collection, PsycINFO and Cinahl from inception to May 2022.We also searched registers, including the clinicaltrials.govand WHO international clinical trials search portal.The search combined three groups of keywords 1) motorcycles and 2) commercial or occupational, and 3) drivers or riders (see supplement 1 for an example of a search strategy).Additional relevant articles were identified via a review of the reference list and citations of included studies using SpiderCite (https://sr-accelerator.com/#/ spidercite) and Web of Science.The search strategy was developed in consultation with the Karolinska Institutet library services.

Selection of studies
The relevance of articles was assessed using a checklist developed based on the reported inclusion criteria.First, two authors (G.K. and D.K) independently screened articles by titles and abstracts and removed irrelevant articles.After that, (G.K. and D.K) performed a full-text review of the potentially relevant articles.Disagreements were resolved via discussion with the other authors (M.H., J.M).The reasons for exclusion at this stage were documented (see Supplement 2).

Risk of bias assessment
We used a modified Newcastle-Ottawa Scale (N.O.S.) to assess the quality of included studies.The criteria consider aspects related to the selection of participants, comparability of cases and controls, ascertainment of exposure and assessment of outcome.For further details, see supplement 3. Two authors (G.K., D.K.) independently assessed the included articles' quality.The studies were then graded as of low quality, medium quality, or high quality.Disagreements were resolved by discussion between M.H. and J.M.

Data extraction and analysis
For this review, the authors adapted the Cochrane Good Practice Data Extraction form for extracting relevant data for each study.Data were extracted from the articles by G.K. and checked for accuracy by D.K see sampling method or study design, where possible, such information was determined by two authors together (G.K and D.K) based on details available in the article.We summarised findings from the articles in a narrative synthesis.Bivariate and multivariable associations between a risk factor and RTCs or injuries are listed in Table 3 and indicated to be: significant and positive (high levels of the factor associate with increased risk of RTCs and injuries); significant and negative (high levels of the factor associate with reduced risk of RTCs and injuries); or non-significant.Due to the heterogeneity in methodologies employed for outcome assessment and reporting across the included studies, we conducted a synthesis of the evidence utilising a vote-counting approach that solely considered the direction of effect, irrespective of statistical significance (Campbell et al., 2020).We aggregated all relevant articles reporting risk factors in a consistent direction and compared them to those with divergent directions.Subsequently, if the number of studies demonstrating an association in a particular direction exceeded those indicating an opposing association, a conclusion of evidence supporting an association between the respective risk factor and RTCs was inferred in that direction.In instances where the article presented multiple multivariable models exploring the association between subsets of risk factors and RTCs, the association reported in the full model was considered.
Conceptually similar risk factors were operationalised and labelled differently across articles in several cases.However, where in the consensual judgement of two authors (G.K and D.K) differently labelled factors effectively operationalised the same construct, variations in labelling were standardised, and findings for those risk factors were listed together under a single label.The articles' labels and operationalisation of risk factors are presented in supplement 4. For studies with multiple reports, we selected one that provided the most comprehensive information.

Results
Figure 1 shows the PRISMA flowchart of the article selection process.After screening titles and abstracts of 1546 articles and 66 full texts, 26 articles were identified.A reference review and forward citation of included articles yielded an additional seven studies, resulting in 33 for analysis.A table of excluded full texts is provided in the supplementary file (Supplement 2).
Quality assessment was made of 33 relevant articles, as shown in Table 2 (see supplement 3 for more details), with an overall grading of high quality, i.e. low risk of bias (20 articles), medium quality (7 articles), and low quality (6 articles).The articles found to be of either low or medium quality were excluded in the qualitative synthesis, and thus N = 20 articles were included.
Of the 20 studies involving 7852 commercial motorcycle drivers, 17 were cross-sectional, 2 were case-control studies, and 1 was a cohort study.The articles described studies from 11 countries, with the highest representation from Vietnam (5), Brasil (4) and Uganda (3).Close to half of all articles (9) came from Sub-Saharan Africa.The five articles from Vietnam were from two large studies exploring RTCs among commercial motorcycle drivers.The earliest study was published in 2005, while over two-thirds were published in the last five years, i.e. between 2018 and 2022.Study characteristics are presented in Table 1.

Methods of studies included in the qualitative synthesis
Most studies (n = 19) were conducted in an urban setting and used face-to-face interviews (n = 17) for data collection.There were only three studies that used either selfadministered or computer-assisted questionnaires.A response rate was reported in 55% of all articles (n = 11) and varied from 40% to 99.5%.All articles adjusted for confounding variables such as age, motorcycle riding experience, educational level, and marital status.Most studies also adjusted for working hours, alcohol use, risky driving behaviours and employment status (full vs part-time).Almost all studies used self-reports to measure RTCs and injuries.

Participants
The analytic sample size ranged from n = 101 (Kieling et al., 2010) to n = 824 (Zheng et al., 2019).Participants' age ranged from 10 to 70 years, and more than half of the articles had samples with an average age of younger than 35 years.
Among the articles reporting on gender n = 17, men formed most participants ranging from 93.5% to 100% of all study participants.Average working hours were reported in six studies and ranged from 8.7-(Ospina-Mateus et al., 2021) to 13.5-h (Nguyen et al., 2018) in a day.Only three articles reported on motorcycle delivery drivers (Da Silva et al., 2012;Kieling et al., 2010;Zheng et al., 2019).

Results of risk factors for RTCs and injuries
Results of risk factors for RTCs and injuries are presented in supplement D. We grouped factors into six main  3. To simplify the presentation, we grouped unique risk factors that occupy a similar conceptual area together.For example, unique risk factors such as working hours and limited time to rest are grouped under working conditions.

Demographic factors
Age was the most included variable, although sometimes without presentation of its association with RTCs and injuries.Younger age was associated with RTCs in five (Almeida et al., 2016;Amorim et al., 2012;Ayinde et al., 2018;Da Silva et al., 2012;Tumwesigye et al., 2016) out of six articles in bivariate analysis, out of which three (Ayinde et al., 2018;Da Silva et al., 2012;Tumwesigye et al., 2016) reached statistical significance.At the multivariable level, four articles (Da Silva et al., 2012;Nguyen et al., 2018;Ospina-Mateus et al., 2021;Tumwesigye et al., 2016) reported a positive association between RTCs and young age (below 30).The remaining two articles (Kitara & Ikoona, 2022;Truong, Tay, et al., 2020) reported a negative association with increasing age, although one of the two articles did not reach statistical significance.Gender was examined as a risk factor in one article (Nguyen-Phuoc et al., 2019), which reported an increase in risk with the male gender, although the association was not statistically significant.

Socioeconomic factors
Many socioeconomic factors were examined.Regarding education level, all four articles reported an increased association of RTCs among drivers with low education status at the multivariable level (Almeida et al., 2016;Ayinde et al., 2018;Kitara & Ikoona, 2022;Nguyen-Phuoc et al., 2019).Motorcycle experience (as a driver of a motorcycle taxi or driving experience) was examined at the bivariate level by six articles and at the multivariable level by 7 articles.Four of six articles (Almeida et al., 2016;Amorim et al., 2012;Da Silva et al., 2012;Tumwesigye et al., 2016) reported an increased association of RTCs and injuries with more driving experience, while the other two (Sopoh et al., 2021;Wankie et al., 2021) reported a decrease.At the multivariable analyses, only two articles (Ospina-Mateus et al., 2021;Tumwesigye et al., 2016) reported an increased association of RTCs among drivers with more experience, while the remaining five found either a reduction, n = 2 (Sopoh et al., 2021;Wankie et al., 2021) or non-significance (Kitara & Karlsson, 2020;Nguyen-Phuoc et al., 2019;Nguyen et al., 2018).
Concerning marital status, three (Amorim et al., 2012; Nguyen-Phuoc et al., 2019; Sopoh et al., 2021) out of four articles reported an increased association of RTCs with being married or having a partner at the multivariable level.However, the association did not reach statistical significance in any of the three articles.At the bi-variable level, out of two articles, one found an decreased association of RTCs with being married or having a partner (Tumwesigye et al., 2016), and another reported an association without indicating the direction (Ayinde et al., 2018).
Only three articles examined the relationship between the number of dependents and RTCs.In the three articles, there was a positive association between having many dependents and RTCs in both bivariate and multivariable analysis (Amorim et al., 2012;Ayinde et al., 2018;Kitara & Ikoona, 2022), although statistical significance was reached in only one article (Ayinde et al., 2018).Formal training was found to be associated with a decrease in RTCs in one article (Tumwesigye et al., 2016) and an association without indicating the direction of effect in another (Ayinde et al., 2018), all examined at the bivariate level.

Behavioural factors
Alcohol use and RTCs was measured in various ways, such as lifetime alcohol use, current alcohol consumption, Alcohol, Smoking and Substance Involvement Screening Test (ASSIST) and through other measures such as Alcohol Use Disorder Identification Test (AUDIT).Alcohol use was reported in seven articles, and a positive association was found in six out of seven articles at the multivariable level (Ayinde et al., 2018;Da Silva et al., 2012;Kiwango et al., 2021;Sopoh et al., 2021;Tumwesigye et al., 2016;Wankie et al., 2021), with five articles (Ayinde et al., 2018;Kiwango et al., 2021;Sopoh et al., 2021;Tumwesigye et al., 2016;Wankie et al., 2021)reaching statistical significance.Smoking was found to be associated with increased RTCs in all four articles (Ayinde et al., 2018;Nguyen-Phuoc et al., 2019;Sopoh et al., 2021;Wankie et al., 2021) reporting on it, although the association was statistically significant in two articles (Nguyen-Phuoc et al., 2019;Wankie et al., 2021).
Mobile phone use was examined in two ways; using a mobile phone while driving and using handheld vs hands-free devices.Increased association with RTCs was reported in all three articles reporting on the use of a mobile phone while driving (Da Silva et al., 2012;Nguyen-Phuoc et al., 2019;Truong & Nguyen, 2019).Notably, an increase in association with RTCs was reported in handheld compared to hands-free devices (Truong & Nguyen, 2019).In the five articles (Ayinde et al., 2018;Da Silva et al., 2012;Nguyen-Phuoc, De Gruyter, et al., 2020;T. D. Nguyen et al., 2018;Tumwesigye et al., 2016) reporting on helmet use, non-use was associated with increased RTCs in all articles at both bivariate and multivariable analyses.However, only two articles (Ayinde et al., 2018;Nguyen et al., 2018) reached statistical significance in the multivariable analyses.Risky driving was assessed in many ways, either as a composite variable or as a single item, e.g.passing a red light, carrying two passengers or more, and aggressive driving.All six articles reporting risky driving behaviour reported an increased association of RTCs in multivariable analyses (Muni et al., 2019;Nguyen-Phuoc, De Gruyter, et al., 2020;Nickenig Vissoci et al., 2020;Ospina-Mateus et al., 2021;Wankie et al., 2021;Zheng et al., 2019).Speeding was found to be positively associated with RTCs in three (Da Silva et al., 2012;Ospina-Mateus et al., 2021;Wankie et al., 2021) out of four articles (Da Silva et al., 2012;Nguyen-Phuoc, De Gruyter, et al., 2020;Ospina-Mateus et al., 2021;Wankie et al., 2021).

Psychological factors
A positive association was found between having a mental disorder and RTCs in the one article examining it (Kieling et al., 2010).All two articles (Da Silva et al., 2012;Zheng et al., 2019) that reported on fatigue found an increased association with RTCs, although the association was statistically significant in one article (Kieling et al., 2010).

Environmental factors
Several different environmental factors were examined.Working conditions, including long working hours, long distances travelled in a day, and few rest days or hours, were positively associated with RTCs in all five articles in the bivariate (Almeida et al., 2016;Amorim et al., 2012;Da Silva et al., 2012;Nguyen-Phuoc et al., 2019;Tumwesigye et al., 2016) and multivariable analyses (Almeida et al., 2016;Nguyen et al., 2018;Ospina-Mateus et al., 2021;Tumwesigye et al., 2016;Zheng et al., 2019).Alternation of shift and working at night was associated with RTCs in two of the three articles (Da Silva et al., 2012;Wankie et al., 2021), while the remaining one reported a protective effect (Truong & Nguyen, 2019).Regarding payment per number of deliveries, two articles found a positive association with RTCs in bivariate analyses (Da Silva et al., 2012;Zheng et al., 2019), although only one was statistically significant (Da Silva et al., 2012).One article reporting on the effect of time pressure found a positive association with RTCs at the multivariable level (Zheng et al., 2019).
Often described as self-owned vs other categories, e.g.rented or under contract, ownership of the motorcycle was found to be associated with a decrease in RTCs in the only article (Tumwesigye et al., 2016) at the bivariate level and one (Ospina-Mateus et al., 2021) out of three (Kitara & Ikoona, 2022;Nguyen-Phuoc et al., 2019;Ospina-Mateus et al., 2021) at the multivariable level.Three articles (Kitara & Ikoona, 2022;Nguyen-Phuoc et al., 2019;Truong & Nguyen, 2019) at a multivariable level examined working on either a full-time or part-time basis, with two articles (Nguyen-Phuoc et al., 2019;Truong & Nguyen, 2019) reporting an increased association with RTCs among drivers working full-time.The article (Wankie et al., 2021) on poor road environment's effect on RTCs risk reported a positive association.

Vehicular factors
One article compared the effect of low versus high engine capacity on RTCs and reported an increased association of RTCs among drivers using motorcycles with low engine capacity in multivariable analyses (Tumwesigye et al., 2016).

Comparison of risk between commercial motorcycle drivers and delivery drivers
The mode of operation (passenger versus delivery) was examined in only three articles (Truong et al., 2020;Truong et al., 2020;Truong & Nguyen, 2019) from the same large study.A negative association was observed between carrying passengers and RTCs compared to delivery drivers.
The examination of risk factors for RTCs among motorcycle delivery drivers was limited to a few studies, with only three studies exclusively focusing on this specific subgroup.Furthermore, the majority of studies that reported on motorcycle taxi drivers also highlighted that these drivers engage in delivery activities at other times.Given the limited number of articles exclusively investigating delivery drivers and the mixed nature of work among motorcycle taxi drivers, we refrained from directly comparing the results between delivery drivers and motorcycle taxi drivers.

Main findings on risk factors' associations with RTCs and injuries
Over 50 risk factors were studied in the 20 articles in this review.While many risk factors showed a statistically significant association with RTCs, a consistent association with RTCs and injuries was found for the following: young age, low education status, alcohol consumption, speeding, mobile phone use while driving, non-helmet use, risky driving behaviours, long working hours, and payment per number of deliveries made.
Over half of all articles examining age, reported an increased association of RTCs among young drivers aged between 18 and 29 years.These results are consistent with those obtained from two scoping reviews of risk factors of RTCs among commercial motorcycle drivers (Konlan & Hayford, 2022;McKinlay et al., 2022).Increased risk of RTCs among young motorcycle drivers has been extensively documented (Berg, 2006;Vlahogianni et al., 2012).In younger drivers, this phenomenon may be due to several aspects, such as having less experience (Vlahogianni et al., 2012), poor perception of drivers' skills (Gicquel et al., 2017) and underestimating risks (Rutter & Quine, 1996).
Notably, more than half of all commercial motorcycle drivers in this review were below 35 years, suggesting an overall younger age demographic profile similar to that in SSA.This is of concern since, as documented, the risk is often higher in younger age groups.Research indicates that age restrictions based on engine size have proven effective in reducing the number of motorcycle-related crashes and injuries in high-income nations (Araujo et al., 2017).As such, similar interventions may be worth considering in low and middle-income countries (LMICs).Additionally, successful workshops for young motorcycle drivers that use practical training to address driver behaviour in traffic have been implemented in Asia (Radin Umar, 2006) and could be adapted for use in other parts of the world.Lastly, long-term behaviour modification programs, including communication campaigns that use persuasive and emotionally impactful messaging in collaboration with communities, have the potential to alter young drivers' attitudes towards road safety (Berg, 2006;Girma Behabu Bezabeh et al., 2020) and in turn improve their driving behaviour.Conversely, although studies indicate a U-shaped risk curve for age (Jakobsen et al., 2022), i.e. increased risk for the young and elderly, no study examined the risk of RTCs among older motorcycle drivers.Interestingly, in four studies, the age range was up to 60 years, indicating older drivers may be part of the profession and a need to consider them in future studies.
The effect of education level on RTCs has been a subject of debate.An increased risk of RTCs among car drivers with low education has been reported (Sami 2013).However, the risk was reduced when adjusted for age, hinting perhaps a confounding effect.Other factors, such as driving experience, may also confound the effect of low education status on RTCs.However, in this review, the effect of low education on RTCs remained even after adjustment for age and experience in the four articles (Almeida et al., 2016;Ayinde et al., 2018;Kitara & Ikoona, 2022;Nguyen-Phuoc et al., 2019) reporting on both risk factors.Drivers with a high level of education may have spent more time in the standard education system and thus be more exposed to information about road safety (Topolšek et al., 2019), potentially improving their driving behaviour.However, the cross nature of studies on this topic precludes drawing causation, and studies with robust design are needed to determine if a causal association exist.
Alcohol use while driving has been repeatedly emphasised as a main risk factor for RTCs and injuries among motorcycle drivers (Da Silva et al., 2012;Kiwango et al., 2021;Truong & Nguyen, 2019).Acute alcohol consumption can affect driving directly by impairing motor and coordination functions and distorting colour discrimination in red light crossing (Jurgen, 2013;Ogden & Moskowitz, 2004;Woratanarat et al., 2009).Alcohol use has also been shown to indirectly increase risk-taking behaviours such as speeding, non-helmet use and tailgating (Sewell et al., 2009;Staton et al., 2018;Woratanarat et al., 2009), which may significantly increase the crash and injury risk.Even after use, residual effects of alcohol have been reported to persist for up to 24 h (Alford et al., 2020) and thus may continue to compromise driving.Commercial motorcycle drivers, particularly in SSA. and Latin America, usually work for many days with minimal rest (Da Silva et al., 2012;Ehebrecht et al., 2018).Given that the residual effects of alcohol may persist for some time, alcohol use may not only affect the drivers during drinking but also extend into the following day and compromise their driving.There have not been many interventions focused on alcohol use among motorcycle drivers; the available ones have mostly been in high-income economies.Measures such as using peers to prevent drink-driving among motorcycle drivers in the U.K. have been effective in reducing drunk driving (Girma Behabu Bezabeh et al., 2020).Other efforts, including random breath-testing activities and sobriety checkpoints common among car drivers (World Health Organization, 2018), may also be tried for commercial motorcycle drivers.This could be accompanied by strong public awareness campaigns, using mass media and other strategic communications, on the risks of drink-driving and the presence of enforcement activities.Nonetheless, each country should determine what works for them as variables specific to local conditions can modify control measures' effects or applicability.
Surprisingly, there is limited literature on the effect of mobile phone use on motorcycle-related RTCs and injuries, even when motorcycles are becoming increasingly used for delivery and ride-hailing services in Asia (Chalermpong et al., 2022), SSA (Bishop & Courtright, 2022) and Latin America (de Souza et al., 2020).In this review, only three articles (Da Silva et al., 2012;Nguyen-Phuoc et al., 2019;Truong & Nguyen, 2019) reported on the risk of cell phones while driving, and all found an increased risk of RTCs and injuries among drivers who use cell phones.The study by Truong and Nguyen, (2019) found an increased risk among drivers using handheld devices compared to hands-free devices.Distracting activities, particularly phone use, seem to be an emerging risk factor frequently linked to road crashes in other types of road users (Jakobsen et al., 2022).A systematic review from 2018 on the effect of talking on a mobile phone on overall driving performance concluded that conversation on a handheld and hands-free phone resulted in performance costs compared to driving without using a phone (Caird et al., 2018).The authors attributed the decline in performance to a lack of attention caused by the additional cognitive load associated with conversation while driving.Thus, such distracting behaviours seem to compromise road safety regardless of whether the phone is handheld or not.With the new ways of organising work using digital platforms like Uber and Bolt, which depends on cellular technology in sourcing and responding to clients, mobile phone-related motorcycle crashes and injuries will be an area of heightened concern.
Using helmets is one of the most effective measures for reducing head injuries and fatalities among motorcyclists (World Health Organization, 2018).They do this by lessening the force of impact between the brain and the skull, distributing the impact over a larger area, and preventing direct contact between the head and any object it may collide with, such as the ground.However, for helmets to be effective, they must be accessible, meet the appropriate standards, and be worn correctly (Araujo et al., 2017).Unfortunately, helmet availability and regulation of standards are inconsistent, particularly in Sub-Saharan Africa (Girma Behabu Bezabeh et al., 2020;Bishop and Courtright, 2022), resulting in low usage rates and the use of ill-fitting and subpar helmets (Nguyen et al., 2018).To address this issue, various initiatives have been implemented to enhance the quality and increase the utilisation of helmets in Africa, with support from the Fédération Internationale de l' Automobile (F.I.A.) Foundation's Motorcycle Initiative (Girma Behabu Bezabeh et al., 2020).Through the Safe and Affordable Helmet Program, F.I.A. brings together helmet experts from Africa and other regions, such as Vietnam and the United States, to develop an innovative helmet specifically designed for motorcycle riders in low-income countries.This helmet complies with international standards (Standard ECE 22.05 [Regulation 22]), is suitable for hot and humid climates, and has a target retail price of approximately $20.These initiatives could be extended to other areas with similar weather conditions and helmet usage concerns, such as Latin America.
In addition to confirming previous knowledge regarding risk factors for motorcycle-related RTCs and injuries, our review adds some important new knowledge.We report on work-related factors as significant contributors to RTCs and injuries among commercial motorcycle drivers.Risk factors such as long working hours, time pressures and payment by delivery were significantly associated with increased risk of RTCs.These results are congruent with those from a scoping review of psychosocial work factors for road traffic accidents in low-and middle-income countries (Amoadu et al., 2023).Commercial motorcycle drivers often operate in an environment with high levels of work-related stress, with effort-reward imbalance (Amoadu et al., 2023;Ehebrecht et al., 2018;Francis et al., 2023).The work is organised around performance (Da Silva et al., 2012;Zheng et al., 2019), and earnings often correlate with how many trips a driver makes in a day (Francis et al., 2023).As a result, many of these drivers may be forced to work for extended periods in an effort to raise their earnings.In addition, many workers report irregular job schedules like nightshift and overtime and a lack of job security and guarantees.These factors combined increase the risk of fatigue and risky driving behaviours and limit what can be done when they fall victim to a crash or get injured.Although several countries in SSA.(Bishop & Amos, 2015;Francis et al., 2023) have enacted motorcycle taxi business regulations, their implementation and reinforcement scope are limited.Thus, efforts to enact motorcycle taxis and delivery regulations were unavailable and strengthening existing ones in areas such as contract setting, having standardised trip fares, regulating driving hours and rest periods, promoting motorcycle associations, and using better methods for planning delivery routes should be considered.

Implications for future research
The research on risk factors for motorcycle-related RTCs and injuries has significantly increased in recent years, covering a wide range of factors.However, most studies on commercial motorcycle drivers' risk factors have been cross-sectional and suffer from reverse causality.Only a few articles in this review employed case-control or longitudinal designs, which are more robust for determining causal relationships.Conducting randomised controlled trials to investigate specific risk factors is neither feasible nor ethical.However, well-designed case-control studies, particularly among motorcycle ride-hailing groups or commercial motorcycle taxi associations, are feasible and should be pursued.
One of the challenges in road safety research is the lack of standardised and validated measurements.Several studies in this review used different measurements for the same variable leading to inconsistencies and hindering accurate comparisons.Lack of standardisation also affects the operational definition of risk factors, further complicating comparisons across studies.Developing agreed-upon operational definitions would contribute to standardising measurements.Another concern was the heavy reliance on questionnaires and interviews to assess risk factors, RTCs, and injuries.These methods are susceptible to social desirability and recall bias, potentially distorting participants' responses.While surveys through questionnaires and interviews remain important in areas with limited comprehensive data systems, future research should aim to incorporate more objective assessments, such as physiological markers and direct roadside observations.Low and middle-income countries (LMICs) should also work on strengthening available data systems, such as police and hospital records.Additionally, future studies should consider additional confounding factors like traffic density, speed, and mileage.
Surprisingly, there was a lack of research on the effect of fatigue on RTCs among commercial motorcycle drivers, despite the long working hours and high job strain they often experience.Fatigue has been associated with increased risk-taking behaviours and RTCs, highlighting the need for further investigation into its prevalence, impact, and determinants among these drivers.The effect of driving experience on RTCs yielded mixed results, with some studies suggesting a protective effect while others reported an increased risk.Ongoing training and communication to improve drivers' beliefs and behaviours may be beneficial in addressing this issue.
Comparing delivery drivers to motorcycle taxi drivers was challenging due to limited research on delivery drivers and the overlapping duties between motorcycle taxi drivers and delivery services in LMICs.The shortage of research on RTCs and injuries among motorcycle delivery drivers is a matter that deserves attention.This mode of transport is increasingly utilised for delivery services worldwide (Chalermpong et al., 2022), and these drivers face unique risk factors and stressors (McKinlay et al., 2022).Therefore, research is needed to investigate the incidence and epidemiology of RTCs and injuries within this subpopulation.Additionally, future research focusing on commercial motorcycle drivers should strive to document drivers' primary occupation, whether operating the vehicle as a motorcycle taxi, i.e. carrying passengers exclusively, as a delivery vehicle or a combination thereof.Such documentation will ensure accurate comparison across studies.

Strengths and limitations
Our search strategy and eligibility criteria were comprehensive, including publications in any language and without restrictions based on publication status.To ensure expertise across all domains, our team comprises subject matter and methods experts who have undergone extensive training and calibration exercises for all stages of the review process.However, our review has some limitations.
Our review's primary limitations are derived from the included studies' limitations.Firstly, most of the studies included were cross-sectional, which poses challenges in establishing temporal relationships and causality.Secondly, a meta-analysis could not be conducted due to differences in data measurements of similar variables across studies.This was primarily attributed to the absence of standardised and validated measurements, which led to increased heterogeneity and made it difficult to pool the data from different studies.Additionally, the lack of standardised measurements further complicated the process of data aggregation.Thirdly, in estimating associations, we relied on observational studies that are susceptible to residual confounding.However, it is worth noting that the majority of studies in this review adjusted for confounding factors in their analyses.Fourthly, the narrative and vote-counting approach used in this review should be interpreted with caution as it does not consider the magnitude of effects or the differing relative sizes of the studies.Lastly, despite our efforts to minimise the risk of excluding relevant articles, there is a possibility that some were not included due to limitations in the search strategy and database selection.Nevertheless, this systematic review provides valuable insights into the risk factors associated with commercial motorcycle drivers and RTCs.

Conclusion
This systematic review identified risk factors for RTCs and injuries among commercial motorcycle drivers.The most consistent significant associations were observed for young age, alcohol use, speeding, non-helmet use and risky driving behaviours.Additionally, distracting activities such as using mobile phones while driving, long working hours, and payment per delivery were identified as recent risk factors, some of which were specific to certain driver occupations.However, several factors, such as driver's training, full-time vs part-time work, motorcycle ownership, driving experience, number of dependents, and marital status, had mixed or no associations.All the included studies were of high quality and most were cross-sectional.More robust design such as case-control or longitudinal studies are required to gain a more comprehensive understanding of the aetiologies of road crashes among commercial motorcycle drivers and to support the development of effective strategies for improving driver safety.

Table 1
(including age, level of experience, gender, level of education, partner status, employment (part-time or full time), taxi vs delivery status, income, motorcycle ownership, working hours, fatigue and risky driving behaviours (speeding, alcohol use, use of helmets, use of mobile phones while riding).Outcomes were defined as self-reported road traffic crashes or injuries regardless of the time.When articles failed to provide clear methodological information, e.g.methods of recruitment,

Table 1 .
Characteristics of included studies n = 33.

Table 2 .
Summary of results of quality assessment using modified newcastle ottawa scale for cross sectional, case control and cohort studies.

Table 3 .
risk factors for road traffic Crashes and injuries of commercial motorcycle drivers based on the high-quality studies, n = 20.