Skip to main content
Download PDF

Identification of determinant factors for crash severity levels occurred in Addis Ababa City, Ethiopia, from 2017 to 2020: using ordinal logistic regression model approach

BMC Public Health volume 23, Article number: 1884 (2023) Cite this article

Abstract

Background

Road traffic Injuries (RTI) are multifaceted occurrences determined by the combination of multiple factors. Also, severity levels of injuries from road traffic accidents are determined by the interaction of the composite factors. Even though most accidents are severe to fatal in developing countries, there is still a lack of extensive researches on crash severity levels and factors associated with these accidents. Hence, this study was intended to identify severity levels of road traffic injuries and determinant factors in Addis Ababa City, Ethiopia.

Methods

The study was conducted in Addis Ababa, the capital city of Ethiopia, using secondary data obtained from the Addis Ababa Police Commission office. The ordinal logistic regression model was used to investigate road traffic injury severity levels and factors worsening injury severity levels using the recorded dataset from October 2017 to July 2020.

Results

A total of 8458 car accidents were considered in the study, of which 15.1% were fatal, 46.7% severe, and 38.3% were slight injuries. The results of the ordinal logistic regression analysis estimation showed that being a commercial truck, college and above level educated driver, rollover crash, motorbike passengers, the crash day on Friday, and darkness were significantly associated factors with crash injury severity levels in the study area. On the contrary, driving experience (> 10 years), passenger of the vehicle, two-lane roads, and afternoon crashes were found to decrease the severity of road traffic injuries.

Conclusions

Road traffic injury reduction measures should include strict law enforcement in order to maintain road traffic rules especially among commercial truckers, motorcyclists, and government vehicle drivers. Also, it is better to train drivers to be more alert and conscious in their travels, especially on turning and handling their vehicles while driving.

Peer Review reports

Introduction

Road traffic injuries are a global disaster and among the top public health problems in recent years. Mortality from the road traffic accidents on the world remains high, with an estimated 1.35 million people dying each year worldwide which ranking it as the eighth leading cause of death for all age groups and the leading cause of death for children and young adults aged 5–29 years [1]. Among world regions, low-income countries share a larger amount of road traffic fatality rate (29.2/100,000 population), out of which Africa accounts for 20% of global road traffic deaths with nearly 272 000 deaths. Also, African RTIs have a significant number of the accidents under reported, mainly nonfatal injuries [1, 2]. From a total of the world’s RTI fatalities, 90% occurs in low- and middle-income countries. Even though these low- and middle-income countries have a low number of registered motor vehicles, the number of road traffic deaths is alarmingly higher than in other countries. For instance, in low-income African countries, the number of registered motor vehicles in 2016 was only 27%, however, the number of road traffic deaths is substantially higher, accounting for 59% [3, 4].

Statistics also show the disparities in road traffic deaths by the types of road users. Among the worlds’ road users, pedestrians and cyclists account for 26% of all road traffic deaths, unfortunately, Africa and Eastern Mediterranean pedestrians and cyclists’ mortality rates are as high as 44% and 36% respectively [5]. RTIs causes not only victims’ injuries and deaths, but also very significant amount of economic loss which is estimated to cost around United States Dollar (USD) 1,800 billion, or 3% of Gross Domestic Products (GDP) globally, while in low- and middle-income countries, economic losses are about 5% of their GDP per year. Moreover, road traffic crashes are predicted to become the seventh leading cause of death by 2030 unless continuous and sustained efforts and actions are taken to reduce road traffic accidents [4].

Despite an increase in the overall number of deaths and burden of the RTIs, the road safety issue was undermined and neglected, which does not receive the due attention it deserves. However, after extensive interventional measures taken in middle-and high-income countries, these measures have effectively reduced road traffic deaths in these countries, even though not a single low-income country has reduced in overall deaths and burden of RTIs [5, 6].

Having a better understanding of the contributing factors for road traffic injuries is very fundamental for road safety policies and crash prediction. Nevertheless, road traffic injuries are complex community health problems that are caused by the combination and interaction of multiple factors that make full understanding of the event and better insight toward the problem is challenging. So far, illegal drivers’ behavior, vehicle features, roadway characteristics, and other environmental conditions have been reported as major determinant factors [7].

In the literature, road traffic accident injury severity levels are determined by a composite factors and conditions. As mentioned, severe injuries or deaths were higher in traffic crashes with pedestrians at fault, drivers’ illegal behaviors, vehicles with unsafe status and crashes occurring at night with street lighting were some of the reported factors [8]. From the vast majority of pedestrian faults, crossing the road by violating the road traffic laws such as crossing outside of designated crossing areas was commonly reported [9].

In Ethiopia, there is a rapid growth in motorization and road infrastructure that has brought alarming road safety concerns in recent years. As reported by scholars, on average, around 13.34% of motorized vehicles and 10.4% of asphalt road development were registered as yearly growth rates. In the same study, on average, yearly around 9.28%, 11.39%, 10.76% and 5% were registered as fatal, serious and slight injuries, and property damage, respectively. Also, the economic losses due to RTIs in Ethiopia are estimated to be around 3.3 billion Birr per year due to the occurrences of road traffic accidents [10].

In the literature, scholars have reported focusing on identifying contributing factors of road traffic accidents and ways forwards [11].

Studies conducted in Ethiopia indicate that factors such as a younger driver, a less educated driver, being male, younger road users, driver’s experience, drunk and speedy driving, a motorcyclist or passenger without a helmet, vehicle size, dark lighting conditions, and a vehicle occupant traveling unrestrained in the back of a truck are major factors contributing to the RTI occurrences and worsening the injury severity levels [11,12,13,14].

Few road safety measures such as; increasing the number of speed breakers, strict enforcement of traffic rules, the imposition of heavy fines and punishments on drivers who violated the traffic rules, and striving to increase the growth rate of road infrastructures, etc. have been initiated in Ethiopia recently. However, the road traffic accidents induced mortality and occurrences of fatal injuries were increasing despite all these activities done so far [14, 15].

However, it has been reported that there is an urgent need to design a predictive model for road traffic accidents or crash severity. This will help manage these tragic accidents and improve road safety because predictive models can determine the frequency of accidents and the severity level of crashes [16]. Also, for an outcome that may be thought of as ordinal, it is often preferable to use all of the ordinal values, rather than condensing them into fewer categories or dichotomizing variables. It has been shown that collapsing categories reduces statistical power [17].

As a result, the ordinal regression model is an appropriate model for outcomes that can be considered ordinal because it is more sensitive to the ordinality of the data and has stable parameter estimates [18]. Thus, the outcome variable for the current study is road traffic injury severity levels, and categorized as slight, severe, and fatal injuries is ordinal data in nature.

Consequently, the current study was intended to assess injury severity levels induced by road traffic accidents and associated factors in Addis Ababa, Ethiopia.

Methods

Data description

The crash data used in this study were obtained from the Addis Ababa city police commission road safety department. The reported data from October 2017 to July 2020 were used. The collected and recorded road traffic accident data comprised accident circumstances, the severity of injuries, immediate outcomes of an accident, and property damage. Data were collected on an excel sheet using variables such as; date and time of an accident, biographic data of the victims such as: age, gender, drivers’ education, relationship with the car, drivers’ experience, location of injury, type of roads, injury severity outcome, pedestrian victim, type of road, type of vehicle, and environmental conditions (weather and air condition, light condition). The Collected data were exported into R software for the statistical analysis.

The original data set categorized injury severity levels into slight, severe, and fatal [19, 20]. Traffic police crash accident reports at accident scenes do not narrate or describe injuries in detail. This is because of the lack of traffic police qualifications and training on trauma assessment and management. In addition,, trauma experts needed to designate injury severity levels are not available in the crash scenes. Likewise, medical reports are hard to obtain because it is difficult to find out linked traffic police and medical patient charts at the same place. The result was that this study relied on the original classifications of slight, severe, and fatal injuries recorded on the traffic police reports [21]. This is similar to some previous studies that classified injuries in the same way [16, 22].

For the current study, the outcome variable was crash injury severity levels (Fatal, Severe, and Slight). For such ordered outcome variables, it is appropriate to analyze with an ordinal logistic regression model, which is a special type of the generalized linear model, and related to a response variable that follows the exponential family via an appropriate link function [23]. Whereas, the independent variables for road crash injury severity levels include: type of vehicle (two or three wheelers, public transport, commercial truck, automobile, and others), victims of crash (Pedestrian, Passenger, Driver of vehicle, Motorcyclist, Motor passenger), type of road (Straight, Zigzag, Curve, Tilted, Hill), and owner of the vehicle (Private, Government), etc.

Statistical analysis

The chi-square test (Χ2) of independence was used to determine the inclusion of variables in the model for further analysis (Table 3). The ordinal logistic regression model was used to assess an association between outcome and explanatory variables. The variables that have a statistically significant association with road crash severity level at 5% significance level were used for further analysis using the ordinal logistic regression model (Table 4).

Model description

The ordinal logistic regression model is a special type of the generalized linear model, which is appropriate for analysis of the ordered responses without losing information from collapsing (or ignoring) some categories of the ordered responses by typically maintaining statistical power. While analyzing data with this model, it would take into account the ordinal nature of the outcome, and an estimated odds ratio may address the questions asked of the analysis [24].

In other words, it takes into account the ordering of the categories of the response variable(s). In this model, we consider the probability of an individual event and all others above it in the ordinal ranking. Cumulative probabilities rather than probabilities for discrete categories are concerned. The goal of such a cumulative odd model is to simultaneously consider the effects of a set of explanatory variables across these possible consecutive cumulative splits in the outcome. One probability is monotonically increasing or decreasing as a function of x. Curves of probabilities for intermediate categories are unimodal with the modes (maximum) corresponding to the order of the categories. The response category does not affect conclusions regarding the relationship between y and x. Therefore, the specific combination of categories examined does not lead to substantially different conclusions regarding the relationship between responses and x [25].

This study applied the ordinal logistic regression model, relating injury severity to potential risk factors. Let Y represents an ordinal response variable with k ordered outcomes, and X represents a vector of explanatory variables, then the ordered logit model describing the relationship between these variables can be described [20, 22, 26]:

$$\mathrm{logit}\lbrack \,P\left(\mathrm y\leq\mathrm j\right)=\log\left\lfloor\frac{P\left(y\leq j\right)}{1-P\left(y\leq j\right)}\right\rfloor=\mathrm\alpha+\mathrm\beta\mathrm X,\,\mathrm{for\,j}=1,2,\,\dots,\mathrm k-1$$
(1)

Where:

P(y ≤ j) = P1 + P2 + … + Pj = are the cumulative probabilities,

αj = intercept parameter (threshold parameter) of the cumulative probability j,

β = is a column vector of parameters that describe the effects of the explanatory variable(s) on the dependent variable.

From model (1), since the relationship between all pairs of categories is the same, we obtain only one coefficient (beta) for the all the categories in the estimated model and different intercepts (alpha) for each category. Model (1) can then be expressed in terms of cumulative probabilities as [16, 26, 27]:

$$\mathrm{P}\left(\mathrm{y}\le \mathrm{j}\right)=\frac{\mathrm{exp}\left(\alpha j+\beta X\right)}{1+\mathrm{exp}\left(\alpha j+\beta X\right)},\mathrm{\, for\, j}=1, 2,\dots ,\mathrm{k}-1$$
(2)

The parameters in the model (2) are estimated using the maximum likelihood estimation method [26, 28].

Since we cannot observe y*, instead we can only observe the N levels of the response through the following structural model:

$$\mathrm y=\left\{\begin{array}{l}1\;if\;y\ast<\tau1\\2\;if\;\tau1<y\ast\leq\tau2\\3\;if\;\tau2<y\ast\leq\tau3\\.\\.\\N\;if\;\tau N<y\ast\end{array}\right.$$
(3)

Where the parameters; τi are the externally imposed endpoints of the observable categories. Then the ordered logit technique will use the observations on y, which are a form of censored data on y*, to fit the parameter vector β [22, 26, 29]. Also, each level of y has its own intercept, which is the negative of the corresponding threshold, but there is a common coefficient vector \(\beta\) [16, 22, 29].

The proportional odds assumption

This assumption states that, parameters should not change for different categories, or the correlation between the independent variable and dependent variable does not change for dependent variable categories. In other words, the effects of any explanatory variables are consistent across the different categories of the outcome variable. In ordinal logistic regression, there will be separate intercept terms at each threshold, but a single odds ratio (OR) for the effect of each explanatory variable [22, 24, 28, 29]. The proportional odds assumption for this study was tested and presented in (Tables 1 and 2).

Table 1 Test of parallel lines
Full size table
Table 2 Goodness of fit
Full size table

We have used R software version 4.2.3 for the analysis using the polr () function in the MASS package. The vglm () function in the VGAM package was used to fit various of threshold-based models to the ordinal data. The po test function in R implements tests in the MASS package and was used to fit the model. Also, ANOVA function to perform an ANOVA test was used to determine if the slope coefficients are equal for all levels of the response variable [16, 30,31,32].

Results

Sociodemographic characteristics of the study subjects

Out of the total number of 8458 accidents sustained, most drivers (86.4%) were male and in the age group of the 18–30 years which accounts for 46%. From the victims of the crash, pedestrians were dominant, accounting for 71.4%, followed by passengers of the vehicle 18.2% (Table 3).

Table 3 Distribution of road traffic accident injury severity levels in Addis Ababa, Ethiopia from October 2017–July 2020
Full size table

Injury severity levels

Most victims 3947 (46.7%) and 3237 (38.3%) had severe and fatal injuries, respectively (Table 3).

Descriptive statistics

An association between the injury severity levels and associated factors is shown in Table 3 using the chi-square test of association. Driver’s sex and age, driving experience, day of crash, type of crash, type of vehicle (two or three wheelers, public transport, Commercial truck, automobile, Others), vehicle ownership (private or government), division of road (round about, one lane, two lanes and others), and air condition (good air, foggy, cloudy and rainy) were found to be significantly associated with the severity level of crash injuries (Table 3).

In the current study, a test of proportional odds assumption or a non-significance test was performed to check that this assumption is not violated. This is confirmed in Table 1, where we accepted the null hypothesis, which assumes that the parameters are the same for all categories. From Table 1, the chi-squared statistic is 15,024.84 with 8400 degrees of freedom. The p-value for this test was 0.217, which is not significant. This implies that there is no evidence to suggest that the parameters differ for each category. In other words, the assumption of proportional odds is satisfied (Table 1).

On the other hand, other diagnostics that are used to detect a lack of fit can be seen in Table 2. The first row displays the values of Pearson chi-square statistics computed by a covariate pattern. A small p-value is an indication that something is wrong with the model. Yet, the reported p-value 0.897 compared with a value of 0.05 showed that the overall model is fit. A 5% significance level was used since our sample size is larger (n > 400) [27, 33].

Same as deviance X2 in the second row of the same Table 2. This means that there is no evidence to suggest that the model does not fit the data at the 0.05 level of significance (Table 2).

From the results, most accidents 2246 (26.6%), 2015 (23.8%), 1942 (23%), and 2255 (26.7%) were occurring in autumn, winter, spring, and summer seasons, respectively (Fig. 1).

Fig. 1
figure 1

Road traffic injury severity levels in different seasons in Addis Ababa, Ethiopia, from October 2017-July 2020

Full size image

Ordered logistic regression analysis

In this section, ordinal logistic regression analysis was done to test the association between explanatory variables, and outcome variables (injury severity levels of the road traffic accidents).

Accordingly, the day of crash, drivers’ education and driving experience, type of crash, type of vehicles, vehicle movement, owner of the vehicle, victim of crash, division of road and types of roads, time of crash, sub-city of the accident, and light condition were included in the model.

From the result of an estimated coefficient of ordered logistic regression analysis, log of odds was converted into an odds ratio for ease understanding (Tables 4 and 5).

Table 4 Factors affecting road traffic accident injury severity levels in Addis Ababa, Ethiopia, from October 2017- July 2020
Full size table
Table 5 Odd ratio coefficients and confidence intervals
Full size table

The odds of a crash injury being fatal/severe /slight is 1.57 times higher for Friday holding Monday as reference. The confidence interval for odds could be as minimum as 1.325 and as maximum as 1.874 with 95% confidence and shows that it is statistically significant since it excludes one.

The results shows that the odd of the crash being fatal/severe/slight injury is 1.446 times higher among college or higher-level educated drivers than among less educated drivers. The 95% confidence interval also suggests that the odd could be as minimum as 1.316 and as maximum as 1. 590. This could seem as a contradictory finding, which could happen for some reasons.

These possible reasons include compromising some of the driving ethics and traffic rules such as over speeding to arrive on time for meetings and other duties, working by phone while driving, and being engaged in busy work by well-educated drivers, which could make them spend a long time on the phone while driving.

The odds of a crash injury being fatal/severe is 2.32 times higher in Bole and 1.76 times higher in Yeka sub-cities than accidents happened in Akaki kality sub-city for some reasons. Both of these sub-cities have large residential and commercial areas with the number of cross-roads and large square areas congested with passengers and vehicles that could make these areas spot for road traffic accidents.

According to the result of the analysis, individuals who sustained rollover type of crash injuries were about 1.470 times more likely to have a fatal injury than their counterparts, OR 1.470 (95% CI; 1.111- 1.946). And also, the odds of the crash injury being fatal or severe among crash victims by dump truck, ambulance, vans, crane truck, and others were nearly four times more likely than two or three-wheeler vehicles, OR 3.877 (95% CI: 2.906–5.185) (Tables 4 and 5).

The odds of fatal or severe injury for individuals who had sustained a crash injury on the zigzag type of the roads were found to be nearly two times more likely as compared to the injury sustained on the straight road, OR 1.834 (95% CI: 1.150–2.931).

From the study, the division of roads was found to have a statistically significant association with crash injury severity levels. Division of roads into one-lane, two-lane, and others is about 0.581, 0.649, 0.636 times less likely to result in fatal/severe injuries as compared to a roundabout road. Being the passenger of the vehicle has a lower odd of 0.697 times resulting in fatal/severe injury as compared to the driver of the vehicle (Tables 4 and 5).

Table 5 shows the odds ratio coefficients and confidence intervals of the variables associated with road traffic accidents injury severity levels in Addis Ababa, Ethiopia (Table 5).

Discussion

This study was conducted to identify the major risk factors associated with the injury severity levels induced by road traffic accidents in Addis Ababa. All recorded road traffic accidents from October 2017 to July 2020 were included in the study. The ordinal logistic regression model was found to be the most suitable model and was used in this current study for conducting ordinal data analysis [24, 29].

It was indicated in the current study that the age of the driver was found to be statistically significant in determining the road traffic accident’s injury severity levels. Young drivers increase injury severity in road traffic accidents. This could be due to the desire to drive at high speed by young drivers, overconfidence, liability to distracted driving, and from less experience. This impact of driver age on the injury severity level has also been confirmed in previous studies [34, 35].

The findings of this study showed that 6039 (71.4%) of the total road traffic accidents happened to pedestrians. Out of these accidents, 1017 (12.0%) were fatal and 2862 (33.8%) were severe injuries. This could be attributed to the fact that pedestrians are less protected than passengers of the vehicle, poor adaptation to the road traffic rules, and lack of adequate crossings for pedestrians. The result of the study done in Tanzania showed that passengers are the dominant victims of road traffic accidents, contrary to the finding of the current study [36]. This discrepancy might be due to the violent behavior of the drivers, poor enforcement of the traffic rules on the road safety practice and less awareness among pedestrians on how they should behave on the roads.

Regarding the age of the road traffic accident victims, majority 2680 (31.7%) of them were within the age group of 18–30 years. This could be due to the fact that younger road users were the predominant group of the community in the study area, lack of pedestrian routes, and convenient public transportation. This finding was in line with previous studies done in Ethiopia so far [6, 37].

This study identified that the day of the week on which severe road traffic accident injuries occurred the most was Friday, followed by Sunday. This is in contrary to the study finding in Ghana, where Saturday was the day on which severe road traffic accident injuries most commonly occurred [16, 38]. This difference might be attributed to cultural and societal values, and activities being held on these days may vary from country to country. For instance; in Ethiopia, since Friday is the end of the week, private and government employees might travel from different work places to Addis Ababa to spend their weekend with their family, which can result in traffic congestion and can increase injury severity level. Furthermore, Sunday is the most popular day for various gatherings, such as weddings, religious programs, football games, and celebrations, which could increase the risk of drunk driving, and sustaining severe crash injuries.

Regarding the educational status of the driver, the probability of sustaining fatal or severe injury was 1.446 times higher among college-or higher level-educated drivers than among less educated drivers in the current study, OR 1.446 (95% CI; 1.316–1.590). However, this is different from the study conducted by Lidetu A, et al.in the Gamo Gofa zone, Southern Ethiopia, which stated that the majority of the fatal and severe crash injuries were from collisions by less educated drivers [12]. Possible reasons for this difference might be attributed to study setting and study period differences. On another point of view, compromising some of the driving ethics and traffic rules such as over speeding by well-educated drivers to arrive on time for meetings and other duties, might be common in the Addis Ababa city, which is the metropolitan city of Ethiopia and the seat of the African Union.

Moreover, this study revealed that commercial truck was found to have a significant impact, OR 1.590 (95% CI;1.368–1.849) on determining the RTI severity level, which was in agreement with previous studies [8, 39]. The possible explanation for this could be that the weight of the trucks is much greater, and even minor accidents can cause severe to fatal injuries. Trucks are larger and heavier than other vehicles, and drivers may be unable to stop their trucks quickly if they are driving very fast, which can cause severe to fatal injuries.

On the other hand, rollover (vehicle upside down) type of crash was likely to increase the probability of severe or fatal injuries among the victims of the road traffic accidents than vehicle -pedestrian collisions. This is similar to the findings of the previous studies reported in the literatures [40]. The possible reasons for this could be, those occupants who were not using restraints were more likely to be ejected from a rolling vehicle, and those who were ejected were more probably sustain severe or fatal injuries. In addition, drivers who did not attempt an avoidance maneuvre (braking or steering), and driving over allowed speed have the propensity to rollover crash that could cause severe or fatal crash injuries. This is because when a vehicle rolls over, it can crush the occupants inside by vehicle’s bodies and structures, since vehicles usually landed up on their side or roof [41].

The present study also considered that a government-owned vehicle has a significant impact (OR = 1.211) on the distribution of the probability of the severe or fatal crash injuries to the collision victims, which agreed with a study done in Ethiopia [22]. This could possible because of poor handling and utilization of the government-owned vehicles by the drivers in the study area.

The proportion of collisions occurring during curved maneuvers of the vehicle either to the left or right is likely to increase the probability of severe or fatal injuries as compared to the straight movement of the vehicle OR = 1.240 (95% CI:1.086,1.417). This was contrary to the study done in Singapore [20], which showed that an accident occurred while vehicle movement was either to the left or right had a lower probability of severe or fatal injuries. This difference might be attributed to the fact that the study participants being studied in Singapore were only involved in an out-of-control single vehicle crash.

This study’s findings showed that the probability of severe or fatal injuries from road crash accidents occurring during the evening and nighttime was significantly lower than the probability of severe or fatal injuries occurring in the morning, OR = 0.818 (95% CI: 0.704, 0.950) and OR = 0.788 (95% CI: 0.675, 0.920), respectively. This agreed with the findings of the study conducted in the Oromia region of Ethiopia [40]. In contrast, the current study identified that the probability of severe or fatal injuries was lower in crash accidents occurring during the afternoon, which opposed the findings of the study done in the Oromia region [40], but was in agreement with the study done in China [8].

Furthermore, some of the previous studies reported that the time of the crash was not statistically significant in the determination of the crash injury severity levels [21].However, there was another study that reported that the risk of death was higher among victims who sustained injuries between midnight and dawn (6 a.m.) than at other times of the day [42]. These findings are in disagreement with current study results. These discrepancies could be attributed to the differences of the study settings, like having a varying number of working shifts among communities of study areas, the culture of spending longer night hours at parties and drunk driving home after party attendance, and the availability of roadside lighting.

The present study also assessed the sub-cities that are prone to road traffic accidents and those that are more susceptible to severe to fatal crash accidents within the study area. Out of the ten sub-cities, Bole, Yeka and Arada sub-cities have a higher probability of severe or fatal injuries than Akaki Kality and other sub-cities, with OR = 2.32 (95% CI: 1.88, 2.87), OR = 1.76 (95% CI: 1.40, 2.22), and OR = 1.38 (95% CI: 1.092, 1.75), respectively. In these areas, heavy traffic volumes, crowds of passengers, intersections, numerous traffic signals, and mixed land use patterns for commercial and residential purposes probably increased crash accident occurrences and injury severity levels. This is in agreement with the study done in Hong Kong [43]. However, in contrast to the findings of another study conducted in Hong Kong, which indicated that commercial areas are safer than residential areas. This was because there were more pedestrian overpasses that separated motor vehicles from pedestrians who were crossing the street, and there were a lot of ground-floor shops in Hong Kong’s residential areas. Unfortunately, such road structure and building alignments with respect to the pedestrian’s road are scarce in Addis Ababa’s commercial areas, and the number of crash accidents in these areas is higher than in other areas.

The crash accidents occurring under dark light conditions in the environment had a significant influence on the severity of road traffic injuries compared with crash accidents occurring under good light conditions in Addis Ababa, with OR = 1.703 (95% CI: 1.434, 2.022). This was in agreement with previous studies [20, 35]. A possible explanation for this is that reduced traffic volumes after dark could lead to increased vehicle speeds, which are likely to increase the risk of a collision and the severity of injury to the victim of the crash. Additionally, drivers are more likely to be inebriated after dark and may also feel sleepier and drowsier, which increases the likelihood that they will be involved in a road traffic accident and sustain severe crash injuries [44,45,46].

Limitations of the study

This study depends on secondary information. Thus, injury severity-level categorizations were considered as they were registered on the road traffic accident records of the Addis Ababa police commission. Also, some very important data, such as seat belt and helmet utilization, drunk driving, the number of people who died from crashes per accident, and geometric and traffic information, were missing. Likewise, the vehicle’s GPS data, which could help to know the correct spot of an accident, the speed of the vehicle, and the number of passengers in the vehicle per accident, were not available. As a result, we were unable to estimate traveled vehicle hours and specific time-related estimations.

Conclusion

From the total crash victims, pedestrians share the highest number of severe to fatal crash injuries, mainly in the summer and autumn seasons. Moreover, younger drivers who were found in the age group of 18–30 years old were found to be more commonly affected than other victims of crash injuries. This study found that severe to fatal crash injuries were attributed to commercial trucks, government-owned vehicles, college and above level educated drivers, rollover crashes, curved maneuver of vehicles, motorbike passenger, crash day on Friday, and darkness. It is recommended that the city traffic police office should give special consideration to commercial truck and public transport drivers, motorcyclists, and all other drivers who drive without adequate lighting. However, to make these police control measures much more effective, it is very essential to complement them with specific communication campaigns and in-service training programs for traffic police, increasing community awareness of road safety and appropriate behaviors on the road through various media campaigns [47, 48].

Furthermore, attention has to be given to the handling of the government vehicle by the respective government offices to which these vehicles are affiliated, and licensing authorities must implement robust assessment procedures before qualifying drivers. In addition, this study found that the occurrence of traffic accidents and the probability of severe to fatal crash accidents are commonly occurring in the morning. As a result, much more attention and extensive road safety control measures have to be implemented to reduce the impacts of road crash accidents. Also, mixed-use land (i.e., commercial and residential) areas in the city have been sustaining a larger number of crash accidents and severe to fatal injuries. Hence, the road traffic department of the city has to consider the establishment of additional pedestrian overpasses in highly overcrowded areas, which would separate street-crossing pedestrians from vehicles on the road to reduce the occurrence of road traffic accidents. Lastly, the current study established that crash accidents occurring in dark light conditions highly contributed to crash injury severity levels compared with crash accidents occurring in good light conditions. It would be beneficial if roadway lighting or illumination were established in crash-prone areas of the city.

Availability of data and materials

The datasets used during this study are available from the corresponding author on reasonable request.

Abbreviations

ABO:

Ararso Baru Olani

AW:

Asfawosen Woldemeskel

CI:

Confidence Intervals

GDP:

Gross Domestic Product

GG:

Geleta Guta

M:

Mean

MA:

Micheal Alemayehu

OR:

Odds Ratio

Ref:

Reference

RTI:

Road Traffic Injuries

SD:

Standard Deviation

TB:

Tariku Bekelcho

USD:

United States Dollar

References

  1. World Health Organization. Global status report on road safety 2018: summary. France: World Health Organization; 2018. Available from: https://www.who.int/publications/i/item/WHO-NMH-NVI-18.20. Cited 2022 Aug 3.

    Google Scholar 

  2. World Health Organization. Road safety in the African region 2015. Geneva: World Health Organization; 2016. Available from: https://www.afro.who.int/sites/default/files/201706/Road_Safety_AFRO_for_web_0.pdf. Cited 2022 Aug 4.

    Google Scholar 

  3. World Health Organization- Regional Office for Africa. Status of road safety in the African region. Brazzaville; 2017. Available from: https://www.ssatp.org/sites/ssatp/files/publication/Status%20of%20Road%20Safety%20in%20the%20African%20Region.pdf. Cited 2022 Aug 4.

  4. United Nations. Road safety - considerations in support of the 2030 agenda for sustainable development. Geneva; 2018. Available from: https://unctad.org/system/files/official-document/dtltlb2017d4_en.pdf. Cited 2022 Aug 5.

  5. World Health Organization. New WHO report highlights insufficient progress to tackle lack of safety on the world’s roads. Geneva; 2018. Available from: https://www.who.int/news/item/07-12-2018-new-who-report-highlights-insufficient-progress-to-tackle-lack-of-safety-on-the-world’s-roads#:~:text=A%20new%20report%20by%20the,people%20aged%205%2D29%20years. Cited 2022 Aug 5.

  6. Abegaz T, Gebremedhin S. Magnitude of road traffic accident-related injuries and fatalities in Ethiopia. PLoS One. 2019;14(1):e0202240. https://doi.org/10.1101/382333.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. Mitra S, Sarkar AP, Saren AB, Haldar D, Saha I, Sarkar GN. Road traffic injuries: a study on severity and outcome among inpatients of a tertiary care level hospital of West Bengal, India. J Emerg Trauma Shock. 2018;11(4):247. https://doi.org/10.4103/JETS.JETS_138_17.

    Article  PubMed  PubMed Central  Google Scholar 

  8. Zhang G, Yau KK, Zhang X. Analyzing fault and severity in pedestrian–motor vehicle accidents in China. Accid Anal Prev. 2014;73:141–50. https://doi.org/10.1016/j.aap.2014.08.018.

    Article  PubMed  Google Scholar 

  9. Cinnamon J, Schuurman N, Hameed SM. Pedestrian injury and human behaviour: observing road-rule violations at high-incident intersections. PLoS One. 2011;6(6):e21063. https://doi.org/10.1371/journal.pone.0021063.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Deme D. Road traffic accident in Ethiopia from 2007/08-2017/18. Am Int J Sci Eng Res. 2019;2(2):49–59.

    Google Scholar 

  11. Zewde T. Determinants that lead drivers into traffic accidents: a case of Arba Minch city, south Ethiopia. Sci J Appl Math Stat. 2017;5(6):210–5. https://doi.org/10.11648/j.sjams.20170506.13.

    Article  Google Scholar 

  12. Anjulo LA, Kuma EP, G/Meskel BZ. Analysis of causes and consequences of road traffic accidents: the case of Gamo Gofa Zone, Southern Ethiopia. Eur J Bus Manag. 2019;11(28):14–5. https://doi.org/10.7176/EJBM/11-28-03.

    Article  Google Scholar 

  13. Misker D, Tunje A, Mengistu A, Abera F, Yalelet M, Gebrie M. Magnitude and factors associated with road traffic accident among traumatized patients in Arba Minch General hospital, 2017. Int J Pub Health Safe. 2017;2(127):2.

    Google Scholar 

  14. Baru A, Azazh A, Beza L. Injury severity levels and associated factors among road traffic collision victims referred to emergency departments of selected public hospitals in Addis Ababa, Ethiopia: the study based on the Haddon matrix. BMC Emerg Med. 2019:1–0. https://doi.org/10.1186/s12873-018-0206-1.

  15. Gebresenbet RF, Aliyu AD. Injury severity level and associated factors among road traffic accident victims attending emergency department of Tirunesh Beijing Hospital, Addis Ababa, Ethiopia: a cross sectional hospital-based study. PLoS One. 2019;14(9):e0222793. https://doi.org/10.1371/journal.pone.0222793.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  16. Asare IO, Mensah AC. Crash severity modelling using ordinal logistic regression approach. Int J Inj Contr Saf Promot. 2020;27(4):412–9. https://doi.org/10.1080/17457300.2020.1790615.

    Article  PubMed  Google Scholar 

  17. Ananth CV, Kleinbaum DG. Regression models for ordinal responses: a review of methods and applications. Int J Epidemiol. 1997;26(6):1323–33. https://doi.org/10.1093/ije/26.6.1323.

    Article  CAS  PubMed  Google Scholar 

  18. Norris CM, et al. Ordinal regression model and the linear regression model were superior to the logistic regression models. J Clin Epidemiol. 2006;59(5):448–56.

    Article  PubMed  Google Scholar 

  19. Yu M, Zheng C, Ma C, Shen J. The temporal stability of factors affecting driver injury severity in run-off-road crashes: a random parameters ordered probit model with heterogeneity in the means approach. Accid Anal Prev. 2020;144:105677. https://doi.org/10.1016/j.aap.2020.105677.

    Article  PubMed  Google Scholar 

  20. Zhou M, Chin HC. Factors affecting the injury severity of out-of-control single-vehicle crashes in Singapore. Accid Anal Prev. 2019;124:104–12. https://doi.org/10.1016/j.aap.2019.01.009.

    Article  PubMed  Google Scholar 

  21. Al-Ghamdi AS. using logistic regression to estimate the influence of accident factors on accident severity. Accid Anal Prev. 2002;34(6):729–41.

    Article  PubMed  Google Scholar 

  22. Meles HG, Gebrehiwot DB, Gebrearegay F, Wubet GG, Gebregergis T. Identification of determinant factors for car accident levels occurred in Mekelle City, Tigray, Ethiopia: ordered logistic regression model approach. Momona Ethiop J Sci. 2021;13(2):225–39. https://doi.org/10.4314/mejs.v13i2.3.

    Article  Google Scholar 

  23. Jeong H, Jang Y, Bowman PJ, Masoud N. Classification of motor vehicle crash injury severity: A hybrid approach for imbalanced data. Accid Anal Prev. 2018;120:250–61. https://doi.org/10.1016/j.aap.2018.08.025.

    Article  PubMed  Google Scholar 

  24. O'Connell AA. Logistic regression models for ordinal response variables. California (US): SAGE Publications; 2006. Available from: https://books.google.com.et/books?hl=en&lr=&id=0TDPPB2uiBAC&oi=fnd&pg=PP9&dq=19.%09O%27Connell+AA.+Logistic+regression+models+for+ordinal+response+variables.+sage%3B+2006&ots=uWSCsYt6dj&sig=j-5EQRzuBC1wbFEbJYy-n7wdClI&redir_esc=y#v=onepage&q&f=false.

  25. Anderson CJ. Logistic regression for ordinal responses. Edps/Psych/Soc. 2019;589:31–3.

    Google Scholar 

  26. Aidoo EN, Appiah SK, Boateng A. Brief research report: a Monte Carlo simulation study on small sample bias in ordered logit model under multicollinearity. J Exp Educ. 2020;89(4):742–50. https://doi.org/10.1080/0022093.2019.1708233.

    Article  Google Scholar 

  27. Adeleke KA, Adepoju AA. Ordinal logistic regression model: an application to pregnancy outcomes. J Math Stat. 2010;6(3):279–85.

    Article  Google Scholar 

  28. McCullagh P. Regression models for ordinal data. J Roy Stat Soc: Ser B (Methodol). 1980;42(2):109–27.

    Google Scholar 

  29. Fox J, Hong J. Effect displays in R for multinomial and proportional-odds logit models: extensions to the effects package. J Stat Softw. 2009;32:1–24. https://doi.org/10.18637/jss.v032.i01.

    Article  Google Scholar 

  30. Yee TW. The VGAM package for categorical data analysis. J Stat Softw. 2010;32:1–34.

    Article  Google Scholar 

  31. Venables WN, Ripley BD. Statistics and Computing. Modern applied Statistics with S-PLUS. 4th ed. New York: Springer New York; 2002. https://doi.org/10.1007/978-0-387-21706-2.

  32. Zeileis A, Kleiber C. Threshold regression models in R: the threshold package. J Stat Softw. 2013;53(8):1–29.

    Google Scholar 

  33. Fagerland MW, Hosmer DW. How to test for goodness of fit in ordinal logistic regression models. Stand Genomic Sci. 2017;17(3):668–86.

    Google Scholar 

  34. Leijdesdorff HA, Gillissen S, Schipper IB, Krijnen P. Injury pattern and injury severity of in-hospital deceased road traffic accident victims in The Netherlands: Dutch road traffic accidents fatalities, 2020. World J Surg. https://doi.org/10.1007/s00268-019-05348-6.

  35. Samerei SA, Aghabayk K, Shiwakoti N, Karimi S. Modelling bus-pedestrian crash severity in the state of Victoria, Australia. Int J Inj Contr Saf Promot. 2021;28(2):233–42. https://doi.org/10.1080/17457300.2021.1907597.

    Article  PubMed  Google Scholar 

  36. Walugembe F, Levira F, Ganesh B, Lwetoijera DW. A retrospective study on the epidemiology and trends of road traffic accidents, fatalities and injuries in three Municipalities of Dar es Salaam Region, Tanzania between 2014–2018. Pan Afr Med J. 2020;36(1):24. https://doi.org/10.11604/pamj.2020.36.24.21754.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Duko B, Tadesse F, Oltaye Z. Patterns of road traffic injury and potential consequences among patients visiting Hawassa University Comprehensive Specialized Hospital, Hawassa, Ethiopia. BMC Res Notes. 2019;12(1):1–4. https://doi.org/10.1186/s13104-019-4192-5.

    Article  Google Scholar 

  38. Ackaah W, Adonteng DO. Analysis of fatal road traffic crashes in Ghana. Int J Inj Contr Saf Promot. 2011;18(1):21–7. https://doi.org/10.1080/17457300.2010.487157.

    Article  PubMed  Google Scholar 

  39. Barua U, Tay R. Severity of urban transit bus crashes in Bangladesh (2010). J Adv Transp. 2010;44(1):34–41.

    Article  Google Scholar 

  40. Aga MA, Woldeamanuel BT, Tadesse M. Statistical modeling of numbers of human deaths per road traffic accident in the Oromia region, Ethiopia. PLoS One. 2021;16(5):e0251492. https://doi.org/10.1371/journal.pone.0251492.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. U.S. Department of Transportation. Characteristics of fatal rollover crashes. Springfield; 2002. Available from: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/809438. Cited 2023 May 7.

  42. Vorko-Jović A, Kern J, Biloglav Z. Risk factors in urban road traffic accidents. J Safety Res. 2006;37(1):93–8.

    Article  PubMed  Google Scholar 

  43. Zeng Q, Wen H, Wong SC, Huang H, Guo Q, Pei X. Spatial joint analysis for zonal daytime and nighttime crash frequencies using a Bayesian bivariate conditional autoregressive model. J Transp Saf Secur. 2020;12(4):566–85. https://doi.org/10.1080/19439962.2018.1516259.

    Article  Google Scholar 

  44. Tefft BC. Impact speed and a pedestrian’s risk of severe injury or death. Accid Anal Prev. 2013;50:871–8.

    Article  PubMed  Google Scholar 

  45. Summala H, Mikkola T. Fatal accidents among car and truck drivers: effects of fatigue, age, and alcohol consumption. Hum Factors. 1994;36(2):315–26.

    Article  CAS  PubMed  Google Scholar 

  46. Uttley J, Fotios S. The effect of ambient light condition on road traffic collisions involving pedestrians on pedestrian crossings. Accid Anal Prev. 2017;108:189–200.

    Article  PubMed  Google Scholar 

  47. Alonso F, Faus M, Fernández C, Useche SA. “Where have I heard it?” Assessing the recall of traffic safety campaigns in the Dominican Republic. Energies. 2021;14(18):2–11. https://doi.org/10.3390/en14185792.

    Article  Google Scholar 

  48. Zatoński M, Herbeć A. Are mass media campaigns effective in reducing drinking and driving? Systematic review – an update. J Health Inequal. 2016;2(1):52–60.

    Article  Google Scholar 

Download references

Acknowledgements

We would like to extend our deepest gratitude to the Addis Ababa Police Commission for their cooperation and facilitation while conducting this study.

Funding

This study received no funding.

Author information

Authors and Affiliations

  1. Department of Emergency Medicine and Critical Care, Arba Minch University, Arba Minch, Ethiopia

    Tariku Bekelcho & Ararso Baru Olani

  2. Department of Medicine, Ethiopian Police University, Sendafa, Ethiopia

    Asfawosen Woldemeskel

  3. Department of Emergency and Critical Care, Tirunesh Beijing General Hospital, Addis Ababa, Ethiopia

    Micheal Alemayehu

  4. Department of Water Resources and Irrigation Engineering, Arba Minch University, Arba Minch, Ethiopia

    Geleta Guta

Authors
  1. Tariku Bekelcho
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Ararso Baru Olani
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Asfawosen Woldemeskel
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Micheal Alemayehu
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Geleta Guta
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

TB: Conceptualization, curation of the data, designing the methods, analyzing and interpreting the results, writing the original draft of the manuscript, reviewing, and editing. AB, AW, MA, and GG: Participated in designing methodology, data analysis, visualization, reviewing, and editing of the manuscript. All authors contributed to manuscript preparation, critically read, and forwarded constructive comments on the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Tariku Bekelcho.

Ethics declarations

Ethics approval and consent to participate

The ethical clearance for this study was provided and approved by the ethical review board of Ethiopian Police University, with an IRB number of EPU/HR-4/12. Then, the authors were granted permission from the Addis Ababa Police Commission through its data administrator to use the Addis Ababa Police commission data via a permission letter. The data were totally anonymous; any personal name was not encoded, and the identifiers of the injured individuals were simply assigned serial numbers. Patient informed consent was not obtained because the waiver of patient informed consent was obtained from an institutional review board of the Ethiopian Police University due to the retrospective nature of the data, in which all the information was collected from the records. All the data collection procedures were conducted according to the principles of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Bekelcho, T., Olani, A.B., Woldemeskel, A. et al. Identification of determinant factors for crash severity levels occurred in Addis Ababa City, Ethiopia, from 2017 to 2020: using ordinal logistic regression model approach. BMC Public Health 23, 1884 (2023). https://doi.org/10.1186/s12889-023-16785-3

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12889-023-16785-3

Keywords

BMC Public Health

ISSN: 1471-2458

Contact us