Burden of road traffic injuries and related risk factors in low and middle-income Pacific Island countries and territories: a systematic review of the scientific literature (TRIP 5)

Background In Pacific Island countries and territories, the burden of road traffic injuries and their attendant risks are considered significant but are poorly quantified. As with other low and middle-income countries, understanding the epidemiology of road traffic injuries in Pacific countries is critical to informing sustainable research and policy initiatives aimed at reducing this burden. Methods We undertook a systematic review and critical appraisal of the relevant epidemiological literature between January 1980 and December 2010, using key search strings for incidence and aetiological studies focusing on RTIs in less resourced Pacific countries. Results Nineteen studies were identified. The majority were descriptive and were unable to provide population-based estimates of the burden of road crash injury, or reliable information on risk factors using well-designed aetiological research methods. All studies were published more than 10 years ago, and all but three reported on data from Papua New Guinea, thereby limiting the generalisability of findings to the current status in the region. Studies undertaken in Papua New Guinea suggested that RTIs were more frequent among young males, with head injuries the most common cause of death or hospital admission. Two thirds of fatalities occurred at the crash site or soon after admission. Most road crash victims were passengers or pedestrians. Factors postulated to influence the risk of RTIs were travel in open-back utility vehicles, utility vehicle overcrowding, and alcohol. Conclusions This review suggests that, despite increasing awareness of the importance of addressing road safety among stakeholders in less resourced Pacific Island countries, road traffic injuries have not been a research priority with little relevant current evidence from the region to inform policy. Robust epidemiological research that can assess the magnitude and key determinants of road traffic injuries in these settings is essential to determine context-specific road safety initiatives that are relevant and affordable. Greater attention to harnessing routinely collected data (e.g., hospital information systems and police crash statistics) to inform policy is also required.


Background
Road traffic injuries (RTIs) are a significant public health challenge and projected to be the fifth leading contributor to the global burden of disease by 2030 [1]. Over 90% of fatal crashes occur in low and middle-income countries with substantial consequences, particularly for vulnerable road users such as pedestrians, cyclists, and riders of motorised two-wheelers [2][3][4]. In Pacific Island countries and territories, the published literature suggests that RTIs and their attendant risks are a significant but poorly quantified cause of death and disability [5][6][7]. It has been reported that up to half of all fatal injuries in the Pacific are due to road traffic crashes [8]. Previous commentators and researchers have also drawn attention to the impact of RTIs among Pacific populations, highlighting factors that require particular attention such as alcohol, motorisation, poor driving, seatbelt use, and poorly maintained vehicles and roads [9][10][11][12][13][14][15]. The Global road safety survey identifies Pacific Island countryspecific characteristics from the perspectives of government statistics and stakeholders [16]. However in order to develop effective road safety initiatives that are relevant for the Pacific context, it is important to assess the quality of local research evidence examining the burden and risk factors for road crashes, and gain a broader appreciation of its utility to influence national policy and behavioural change [17][18][19]. To this end, we undertook a systematic review of the published literature of epidemiological studies investigating the burden of and modifiable risk factors for RTIs in less resourced Pacific countries.

Methods
We identified and critically appraised epidemiological studies published between January 1980 and December 2010 focusing on RTIs. The countries of interest comprised 18 of the 22 member states of the Secretariat of the Pacific Community, encompassing all of the countries in the Pacific region but excluding high-income countries as classified by The World Bank [20]. Studies describing the burden of RTIs were included if they presented epidemiological data on road users who had sustained an injury as a result of a road traffic crash. Studies investigating risk factors for RTI were reviewed if they included a comparison group or time period (as in case crossover studies) with respect to the exposure of interest, and the study outcome measure was a road traffic crash resulting in an injury (fatal or non-fatal). Therefore the primary study designs of relevance included: case control, cohort, case crossover, or cross sectional research methodologies. The following electronic databases were searched building on strategies recommended for reviewing the literature addressing incidence and aetiological studies [21]: Medline, EMBASE, CINAHL, PsycINFO and the Australian Transport and Road Index database, and Transport Research International Documentation. A search string applied for the review is provided in additional material. The review was restricted to studies published in the English language. All studies identified were screened initially by title and abstract, and then more thoroughly in full text. (Figure 1) Adhering to PRISMA guidelines,1 (Additional file 2) studies meeting the inclusion criteria  Figure 1 Flow diagram of studies selected for review.
were critically appraised for the quality of the evidence using GATE-LITE™ [22] and summarised in evidence tables with the following headings: study design, participants, variables examined, key findings relevant to the review and study appraisal and comments. (Table 1).

Appraisal of study methods and quality
Although the majority of studies were case series, there was wide variation in study design and quality. Most of the studies conducted outside PNG were small except for one study in Fiji reviewing national road crash data (n = 2277) [37]. In general, PNG studies involved larger study samples (up to 5772) including three ecological studies drawing on the national road crash database [38][39][40]. The methodological quality of the studies or the ability to appraise this was compromised by inadequate reporting of population characteristics, case definitions, recruitment methodology (source, eligible and study population), exclusion criteria, data collection and analysis, response rates, and identification of potential biases and confounders. Most research designs failed to consider possible biases in analysis and the interpretation of findings. Since many studies reviewed post mortem records, potential cases may have been excluded if they had not presented to the hospital setting. The majority of studies were retrospective, and the quality of data sources, (e.g. completeness, accuracy) was unspecified.
This makes it difficult to ascertain the extent to which misclassification and underreporting may have been a source of information and recall bias. No studies reported whether outcome measures or blood alcohol concentration (BAC) measurements were assessed blind to exposure information, nor whether measurement bias was considered when estimating alcohol concentration from breath or blood samples.
Regarding measures of occurrence and effect, most of the studies reported unadjusted event counts and proportions, with only three studies providing populationbased rates [23,38,40], and an ecological study reporting effect estimates [40].
Acknowledging these limitations, Table 1 summarises studies investigating the burden and aetiological factors for RTIs in Pacific countries, all of which were published more than a decade ago. The key aspects of these are described next.

RTI-related fatalities in PNG
In a study of 1,279 post mortems from all causes of deaths in all ages conducted at Port Morseby General Hospital (PMGH) for the years 1962 to 1989, RTIs (n = 573) accounted for 45% of all deaths, 75% of whom were male [29]. Another post mortem study focusing only on traumarelated fatalities in all ages (n = 608) at PMGH (1976 to 1985), showed that RTIs were responsible for 60% of these deaths, 83% of whom were male [28]. Notable findings in other studies included the high proportion of road deaths that occurred at the crash site or before arrival to PMGH (66%) [24] and Goroka Hospital (81%) [25], as well as the disproportionately high number of fatal crashes on weekends and during early morning hours [23,28].

RTI-related studies in pacific countries other than PNG
RTIs contributed to 14% of injury related admissions (n = 100) to the accident and emergency (A&E) department in Yap hospital [35], and 25% of trauma-related spinal cord injury admissions (n = 75) to a rehabilitation unit in Fiji [31]. A Fiji study examining national (n = 2277) and Western Division) Province (n = 872) data reported deaths in 1% and 6% of crashes respectively [37]. In this study, 40% of those hospitalised had sustained head and neck injuries, while injured children were more likely to be pedestrians [37].

Potential contributors to RTIs
Although we did not identify any aetiological studies with appropriate control (unexposed) groups, several studies suggested factors that were likely to be important contributory factors to road crashes. An ecological study drew attention to vehicle type, noting increased risks of crashes involving buses and open-back utility vehicles compared with cars [40]. Several other studies also noted the high proportion of utility vehicles involved in crashes [27,38,39]. When combined with road user type, passenger fatalities were noted to be highest for utility vehicles and lower for buses and cars [40]. Between 25 and 60% of passenger deaths were due to falls from utility vehicles [23,25], attributed largely to vehicle overcrowding [40].
In contrast to passenger injuries, driver injuries were reported to be highest for cars, with some studies attributing up to 70% of RTI fatalities to driver error [24,37]. One review of 121 post mortem records in PNG identified that motorcycle drivers were five times more likely to die in a road crash than car drivers [23].
Several studies commented on the role of alcohol in RTIs. In a roadside survey of motor vehicle drivers (n = 893), and review of RTI related drivers assessed at the PMGH emergency department (n = 10), BAC levels above 80 mg/dL were detected in 24% and 80% of drivers, respectively [41]. Post mortem data on traumarelated cases revealed BAC levels greater than 80 mg/dl in 48% of drivers and 67% of pedestrians [28]. In 188 RTI-related patients involved in 109 crashes, 49% of crashes were associated with alcohol [30]. Several studies suggested that alcohol-related crashes were more common at night, early morning, and during weekends [23,28,30,41]. While not designed as a formal aetiological investigation, one study in Fiji attributed 4% of road crashes to alcohol and 2.5% to speed [37].

Discussion
The literature regarding the burden and risk factors in the Pacific region remains largely limited to descriptive studies in PNG with important limitations in their ability to examine the population-based burden of road crashes and investigate related risk factors. No RTI focused epidemiological studies have been published for over 15 years and few were conducted outside PNG. Notwithstanding these limitations, the published literature indicates that road crashes are a major public health problem. Studies from PNG reported that RTIs contributed to at least 40% of all deaths, 60% of trauma-related deaths, one third of trauma-related hospitalisations, and 40% of all head injuries. Most (two thirds) RTI-related deaths occurred prior to hospitalisation with related fatalities particularly high among pedestrians and passengers. Although head injuries were the most commonly reported RTI-related injury, chest injuries were reported to be particularly common among motor vehicle drivers. The available literature suggests that vehicle type (openback utility), utility vehicle overloading, and alcohol are potentially important risk factors for RTI.
This review was limited to scientific publications and the English language. Therefore, it excluded the grey literature and routinely collected data such as police annual reports and national hospital and mortality statistics, most of which report events but provide little critique or analysis. The diverse methodological characteristics of the studies identified made it inappropriate to undertake a metaanalysis, or infer that the findings from these largely PNG focused studies can be generalised to Pacific countries more generally.
Nevertheless, this review adds to the evidence base of previous reviews on RTI in less resourced settings [7,43,44], focusing in this case, on Pacific countries which are often absent or overwhelmed by statistics from countries with very large populations in the Western Pacific region. A defining feature of this review is the paucity of robust and contemporary epidemiological RTI studies from Pacific countries, suggesting research attention to RTIs in the region have not been a priority [45,46] and long overdue. However, the evidence available indicates that, as in other less resourced settings, RTIs contribute significantly to the burden of RTIs in Pacific countries and poor data should not delay action [47], The review particularly reveals the need to undertake well-designed aetiological studies that can quantify the contribution of important context-specific risk factors associated with serious road crashes. Indicative findings from the review suggest a key area requiring focused attention in PNG is improved safety considerations relating to the use of utility type vehicles [48]. While the approaches to address this must take into account the socio-economic and transportation implications for the communities involved [49,50], it has been argued that restricting passenger numbers through vehicle occupancy capacity limits may have the potential to reduce crash fatalities by almost 30% [40]. Other areas amenable to well-implemented legislative interventions in the region include the use of seatbelts [51] and deterrents to drink driving [16]. Given the high proportion of road deaths at the crash scene, greater attention is required to health system improvements including pre-hospital care and the training of first responders [2].
The World Report on road traffic injury prevention identifies the need for a comprehensive multi-sectoral, integrated, systems approach that focuses on improved information systems, institutional capacity strengthening, research that quantifies common modifiable risks, and resources for targeting these [2]. Reliable and sustainable injury data surveillance [52][53][54][55] including secondary data routinely collected from hospital and police records, comprises an important foundation for monitoring and evaluating road safety strategies in the Pacific context [19,39]. This would be strengthened by the standardisation of RTI case definitions, data collection processes, analysis and reporting from surveillance systems [56], while ensuring dissemination approaches link findings to interventions, policies, context-specific research and funding mechanisms. As noted elsewhere [57], a critical requirement for identifying and implementing effective road safety strategies is a skilled workforce integrated across relevant stakeholder agencies including legislation, economic policy, capital infrastructure, road engineering, vehicle design, and health care.

Conclusions
This review found that the burden of road traffic injuries in PNG and other less resourced Pacific countries is significant, however, the contribution of modifiable risks for RTI remain poorly quantified. While some studies revealed the likely importance of factors such as alcohol misuse and vehicle type, effective public health and road safety policy requires attention to population-based studies that can identify and quantify the contributors to RTI in the Pacific context. Improving the quality of secondary data sources from routinely collected hospital and police surveillance systems is essential.