Introduction

Nowadays in the world particularly in Low-Middle income countries (LMIC), traffic accidents and its consequent injuries are the most important and the main factors of disability and death [1, 2]. That, death casualty because of traffic accidents is over 1.36 million annually worldwide and around 50 million are also injured each year [3]. Traffic incident is the main reason for casualty among 5-20-year-old people around the world [4], as well. Although, corrective activities were implemented in all countries especially LMIC in 2013 to reduce the traffic accidents and to safe the roads, there isn’t still any significant reduction in the number of traffic accidents and its complications[5]. Therefore, World Health Organization (WHO) predicts that traffic accidents will be one of the biggest future problems of the world [6, 7].

Findings of studies show that traffic accidents are preventable [8-14] and discriminating the nature of cause and effect has a significant effect on preventing RTIs [10, 15]. Researches performed on traffic accidents mentioned three factors of human, road and vehicle as elements cause the accidents. Among mentioned factors, human factor is identified as the most important factor causes these incidents [16]. Human errors are the reason of 95% of traffic accidents [17]. Based on international reports, passengers of vehicles have the highest mortality rate among road users in 2020. About one third (32%) of mortality rate caused by accident is related to the ages of 15 to 17 [18]; this rate decreases substantially after getting driving license and by maturing and gaining experience in driving [19]. Higher rate of accidents among fresh young drivers is related to their failure to recogne driving dangers and their reaction towards them [20, 21]. Several studies consider the traffic incidents occurrence rate directly relevant to the level of driver’s skills in recognizing and reacting of driving hazards [21-23].

Hazard is defined as “the probability of a vehicle collision with a moving or stationary object or its deviation of road because of changing its speed or direction” [24]. In other words, hazard is defined as a dangerous, unsafe and unusual driving event [25]. Factors such as pedestrian, motorist or cyclist, unusual object on the road, pedestrian sidewalk on highway, the car that is parking, road construction sites, traffic control equipment, road side entrance to main road, front car breakdown or warning and other thereof factors are included as driving hazards [26, 27].

HP have been investigated by researches for decades (the first paper was presented by Spicer in 1964 cited by Plez and Krupat, 1974) [28]. Lots of studies have examined HP, measurement methods and factors affecting it, in which different definitions are given. In these studies, expressions related to hazard are often used interchangeably; such as hazard recognition [29], HP [20, 29, 30], hazard prediction[31], and awareness of hazard [32, 33]. Not only are these expressions independent [34], but sometimes studies that use these terms also offer adverse meanings. As an example, HP is defined as ability to recognize situations of the road danger [35], to predict the hazard signs [36], to identify and respond the dangers [30], to have the early detection of hazards [37] or the process of risk awareness [38]. In addition, several studies have also failed to identify correct and measurable terms for HP [29, 39].

Regrading to the importance of HP and its key role in declining accident occurrence, having an accurate and clear definition of the term can be helpful for authorities in setting transparent policies to prevent accidents and also for researches of this field. Therefore, by the purpose of conducting a systematic review in this field, current study tries to gather driving HP and its available definitions in different studies and to explicate HP concept and determine its dimensions.

Materials and Methods

Design

Present study is a systematic review and meta-analysis which was conducted to explain the concept of HP and to assess its dimensions in 2021 according to the book of “systematic reviews to support evidence-based medicine” [40].

Data Sources and Searching Strategy

Searching the information is done in databases: Science Direct, PubMed, Scopus, PROQUES, Web of knowledge and Google scholar search engine from January 2000 to September 2021. The related key words derived from Mesh were used in studies search. Also, after searching the databases, in order to identify and cover more published papers, some specialized Journals related to the topic were searched and the reference list of finalized papers to enter the study were investigated, manually (Table 1). Also search strategy was developed for other data bases according to the data bases conditions.

Inclusion Criteria

- All observational studies (descriptive-analytical, cross-sectional) entered the study.

- Studies investigating the motorist’s perspectives of HP were discarded.

- Studies were conducted from 2000 to 2021.

Exclusion Criteria

- Studies that investigated the HP from cyclists’ point of view were not included.

- Studies that were written in language other than English were excluded from the study.

- Papers presented in conferences and congresses were excluded from the study.

Review Process

After removing duplicate cases, the titles of all papers were investigated and the papers that seemed not to be consistent with study objectives were excluded. In the next stages, the abstract and full context of the articles were studied respectively and the studies that didn’t have the entry criteria and had weak relationships with the objectives were identified and discarded. All systematic review processes were done by two authors and the cases of dispute between them were referred to a third party. Endnote X9 was used to manage the organization of the resources. 

Data Extraction

According to the objectives and based on preliminary evaluation of extracting data, the extraction form was designed by Microsoft Word 2016. At the first, data from 5 papers were experimentally extracted using this form and the shortcomings and problems of the initial form were eliminated.  Data were extracted according to the author, year, country, methodology, sample size, participants, age, hazard perception, and definition variables.

Quality Appraisal

Critical Appraisal was done for 50 included articles in systematic review and meta-analysis. The studies reporting quality was independently assessed by two investigators (Y.H & H.S) according to 22-items STROBE checklist for the cross-sectional studies [41, 42] and JBI experimental checklist for experimental studies [43]. This checklist contains 22 questions. The checklist items score was 0, 1 and 2, according to matching the checklist question criteria with the contents of the articles. The minimum score for checklists is 0 and maximum score is 44. Studies were classified as good [the score in the range 30-44], medium [15-29] and poor [0-14] quality studies. The JBI checklist for experimental studies items score was 0 and 1. The minimum score for checklists is 0 and maximum score is 9.

Articles were assessed by two of authors and they categorized studies into three groups of high, moderate and low quality based on their overall score. Any disagreements between investigators were resolved by discussion and in consultation with research team.

Data Analysis

Qualitative Data Analysis

The Thematic Content-Analysis was applied which is a common method for identifying, analyzing, and reporting the patterns within the texts [44-47]. Coding was done by two researchers independently. Steps of the analysis included: reading the text several times, getting familiarized with data, identifying and extracting the primary codes, identifying the themes by placing similar codes together, revising the themes, naming and defining the themes, and ensuring the reliability of identified codes and themes by obtaining agreement between the two coders and resolving the disputes by discussion.

Data Analysis for Meta-analysis

At first, weighted average was estimated for studies that reported HP. In order to estimate the variable of the overall HP drivers in both groups of experienced and novice drivers, meta-analytical statistical methods were used. First, the data of selected studies was entered in Excel sheet. Stata 16 was used to perform the meta-analysis. Forest plot diagrams were used to report the results, where the size of each square represents the sample size and the lines on each side of the square represent 95% confidence interval for each study. Q and I² were used to evaluate the heterogeneity of the studies’ result. To interpret the I², thresholds recommended by Higgins et al. [48], were used (25% for low heterogeneity, 50% for medium, 75% for high heterogeneity). Due to the high heterogeneity of the studies random model was used. Funnel plot diagram and Egger’s regression test were used to measure publication bias.

This study was part of a Ph.D. Thesis supported by Isfahan University of Medical Sciences (ethical code: IR.MUI.RESEARCH.REC.1399.719).

Results

In the first stage of searching in different databases, 1167 articles resulted from studies were found. Subsequently, 225 articles were removed because of duplication and 942 articles were entered into the investigation process of title and abstract. After reviewing their titles and abstracts, 267 related articles were selected out of total number of papers. Finally, after screening complete context of the articles with entry criteria, 50 articles were entered the study as related qualified articles and among them, only 10 articles were used for meta-analysis that based on quality evaluation checklist as well as entry and exit criteria, they had the retention condition in study (Figure 1) [21, 22, 26, 27, 35, 36, 49-92].

Fig. 1. Flow diagram of the searches and Inclusion process.

A summary of extracted data for meta-analysis is also indicated in Table 2. To conduct meta-analysis, some studies that did not report the needed data were excluded from the study. The appendix shows a summary of the extracted data for meta-synthesis (Supplementary 1- characteristics of included studies and summarized in the study).

Articles about drivers HP of traffic accidents used in the study were mostly conducted in high-income countries in continent of Europe. The results demonstrated that Canada and Australia had the most studies in this issue. Also, South America had the fewest studies; there wasn’t any noticeable attention about this case in Africa so that there wasn’t any study about it in this continent. By investigating the time of article publication, it was shown that most of the papers were published between the years of 2009 till 2021 (Figure 2).

Fig. 2. Studies conducted in countries for measurement hazard perception of traffic accidents.

According to obtained results, the average score of drivers’ HP in investigated studies were estimated 3.33 [4.04-2.62 with CI 95, I²=89.72]. The results also demonstrated that there is a high level of heterogeny among studies results (Figure 3).

Fig. 3. The overall score of hazard perception in the included study.

Before investigated the heterogeneity of the studies, the score was 3.91 [2.81-5-01 with CI 95, I²=99.92]. After the heterogenic of studies was investigated through analyzing the sensitivity and one study was deleted from analysis because it reported a high average score that had excessive effect on average score and caused its deviation (Figure 4).

Fig. 4. Publication bias of the studies on hazard perception.

Reviewing the obtained results show that the beginners with average score of 3.42 [2.41-4.43 with CI 95, I²⁼89.68] in compare with experienced drivers with average score of 3.26 [2.23-4.26 with CI 95, I²⁼89.47] received lower score (Figure 5).

Fig. 5. Average scores of Hazard Perception based on participant skill.

Quality Appraisal Results

The average overall quality of reporting the articles with cross sectional methodology was 37.75 (rang=0-44) and the average overall quality of reporting the articles with experimental methodology was 7. Generally, the reporting quality of articles was estimated as a good category (Table 2).

Defining Hazard Perception and Its Components

The results obtained through reviewing the studies showed that there are several definitions about HP which are categorized in Table 3 to make its concept more understandable. The most definition is related to the concept presented in Horswill and McKenna article in 2004 (32) which refers to “the ability to predict the potential hazards of the road and traffic situations” [49-51, 54, 57, 65, 67-69, 75, 78, 79, 81, 87, 93]. Also, another study by Horswill conducted in 2016 redefined the HP in drivers as “the driver’s ability to predict and identify the road potential hazards” [21] (Table 3).

Identifying and determining the effective component in HP is one of the most important points needed in measuring the drivers’ HP that through investigations of current study, 7 effective components were identified (Figure 6).

Fig. 6. Influential component in measuring drivers’ Hazard perception.

Discussion

The current research was conducted to explain the concept of HP and to determine its effective components. In this study, 6 effective components and 28 definitions of drivers’ HP in traffic accidents were identified in these studies. Also, reviewing of obtained results from meta-analysis demonstrated that average score of drivers’ HP is 3.33 [4.04-2.62 with CI 95, I²=89.72].

The results obtained from meta-analysis indicated that average score of drivers’ HP is low and regarding to the importance of the issue, it’s not acceptable and more attention of countries needs to improve drivers’ HP of traffic accidents. Several factors can be effective in reducing drivers’ recognition of danger related to the condition of each country. The level of experience in driving is one of the factors that have a direct effect on increasing the HP of traffic accidents and can have a high influence to reduce the incidence of traffic accidents [36, 94, 95]. One of the interesting points that can be observed in current study is the high average score of HP among beginner drivers with the score of 3.42 [2.41-4.43 with CI 95, I²⁼89.68] compared to the experienced drivers with average score of 3.26 [2.23-4.26 with CI 95, I²=89.47] [53, 89, 96]. A review of recent decade studies shows that experienced drivers get higher HP score than beginner drivers and even perform better to respond the dangers and also to identify concealed or hidden dangers that contradicts the findings of current study [33, 36, 91, 97-101]. According to obtained results, the present study shows that currently, the beginner drivers also offer close or even better performance than experienced drivers which can be related to many strictness during basic driving trainings that improves drivers’ abilities. In addition, taking strict and graded tests regarded to vehicle at the time of issuing driver license can also be effective in this matter. Nowadays, compulsory and professional driving education in countries, taking HP tests from driving license applicants and necessary training appropriate to beginner drivers’ needs reduces driving accidents as well as increases the HP of drivers [102-106]. Another probable reason which causes higher average score of HP in beginner drivers can be the fear of an accident; this fear makes the fresh drivers to investigate the road with great sensitivity and accuracy to reduce probable accident. A study in Turkey shows that fear of traffic accidents danger is one of the most important factors effective on increasing demands for professional driving trainings; these training can have a very positive effect on increasing the individuals’ HP [102].

The results of study investigation show that many components can be effective on drivers’ HP. Hazard recognition and response time to traffic accidents are two identified effective components that are closely related. HP and response time are considered more than other components in measuring HP due to their repetitive frequency and their importance in articles. Both components are placed together in such a way that HP is a major principle in responding to the danger and takes precedence over it and the response time for identified danger can only be measured during this process [80]. It is also thought that early HP is performed better in drivers having good skill to recognize the danger [53]. In a study by using signal recognition theory of Wallis and Horsville, different criteria of making decision between experienced and beginner drivers causes them to be different in HP [107]. In other studies, it is also demonstrated that drivers with freer decision-making criteria can respond faster to the dangers [20, 94].

Visual search of road by driver is another component that can reduce or increase drivers’ HP especially in beginners. when beginner drivers are in dangerous traffic situation, their road visual search is less, while for experienced drivers, it is more [36].  In this regard, other studies also indicate that experienced drivers have more horizontal eye movement than beginner drivers [108, 109]. Different studies showed that the more experienced a driver is, the better they use their visual search on the road to prevent different traffic situations and hazards [110]. A study designed by Fisher and Pollatsek in 2006 demonstrated that beginner drivers see and distinguish high risk situations on the road less than experienced drivers and in this regard, they believed that necessary trainings can be used in order to enable the fresh drivers to recognize high risk situations on the road [111].

One of the most important identified components mentioned in drivers’ HP is the prediction of danger. In this component, researchers investigate the ability of drivers to predict what the potential hazard situations are and how likely they are to occur. This component can be said to be a completed and enhanced component of HP. This component taken from Situation Awareness Global Assessment technique (SAGAT) is the danger prediction test in which they try to show drivers some videos related to probability of traffic accidents occurrence then suddenly and at the time when the danger arises the video is blocked to allow the drivers to predict that situation, answer the questions and make decision [38, 58, 68, 112]. Therefore, Jackson et al., study argued that accident prediction is perhaps the most important aspect of HP as it clarifies the place of future dangers and the ability of their processing, as well [112]. Given that limited studies have conducted using HP but one of the advantages mentioned for that is the accurate record of hazards predicted by drivers that can be a powerful test to distinguish between beginner and experienced drivers [38, 58, 68, 112].

Drivers’ response sensitivity to situations of danger is another component that was identified during measurement of drivers’ HP. In response sensitivity, level of drivers’ sensitivity in reacting to hazards is measured to show that whether they consider the situation as a danger or not. One of the importance of this component is clarified when the hazard potential situations which seem to be unimportant in drivers’ point of view are not excluded from response sensitivity test and are analyzed for response sensitivity [81].  In this regard, the results of studies conducted by Ventsislava et al., in 2016 demonstrated that the more sensitive a driver is to traffic hazards, the better he can recognize the hazards and ultimately has a better perception of danger [35].

Limitations and Strengths

One of the limitations of the current study was using a language (English) to search the studies and documents. While it is possible that studies and documents about drivers’ HP are conducted in different countries and published in the language of those countries but they are not found and examined in the present study. Many different and influential factors can be effective in drivers’ HP but in current studies, it is tried to find and select components related to human factors and use them to evaluate the drivers’ HP. One of the strengths of the study is that it has tried to focus on the human factors influencing the occurrence of traffic accidents because identifying and defining these factors and training in these fields can increase the accuracy of drivers and reduce road traffic accidents.

Conclusion

Despite the general perception and belief about direct relationship between driving history and experience with increasing HP, current study results showed that increasing driving years has reverse relationship with driving HP and experienced drivers have much lower HP than beginners. The results also showed that although lots of attentions are focused on traffic accidents and its prevention in countries, drivers’ HP is still less. In this regard, due to the importance of HP discussion and its important role in traffic accidents, more focus and more accurate studies about identified components are needed. Therefore, it is suggested that the researchers should evaluate different areas of HP such as psychology, traffic engineering and so on. Also, one of the things that can be effective in improving the drivers’ HP is to make them to take training courses and to pass standard tests while they are getting the driving license. 

Declarations

Ethics approval and consent to participate: The Institutional Review Board and the Ethics Committee of Isfahan University of Medical Sciences, Isfahan, Iran, approved this study (IR.MUI.RESEARCH.REC.1399.719).

Consent for publication: Not applicable.

Conflict of interest: Conflict of interest the authors state they have no conflicts of interest to declare.

Sources of funding: This study was financially supported by Isfahan University of Medical Sciences (Ethics and Grant number: IR.MUI.RESEARCH.REC.1399.719).

Authors’ contributions: YHO & MY collected, reviewed papers, analyzed and prepared the figures. YHO & H.SB contributed in designing, analyzing and drafting the paper. YHO & MY & HSB contributed in categorizing the indicators, developing the tool and reviewing and YHO & MY & HSB Contributed in analyzing and editing the paper.

Acknowledgments: We are very grateful to Experts for contribution and time and invaluable comments. Also, we would like to thank the Research Administration of Isfahan University of Medical Sciences, Isfahan, who supported us in doing this study.