The Pattern of Road Accidents in Fog and the Related Factors in North of Iran in 2014-2018

Objectives: Road fog can form suddenly which necessitates more caution by the researchers as well as drivers due to low vision and related morbidities and mortalities. This study aimed at determining the pattern of accidents in fog and the relationship between driver, road, and accident with injury level in Guilan, Iran from 2014 to 2018. Methods: This is a retrospective descriptive-analytical study that investigated all accidents that had occurred in Guilan from 2014 to 2018. All of the data were entered into STATA software. The time-series estimators were used as the multivariate regression model for the presence or absence of any relationship between variables. Results: In fog, the fatality rate of trac accidents was signicantly higher than other climatic conditions (IRR = 1.471). The likelihood of fatalities in fog decreased with age while it increased when drivers were female (p = 0.001). Most of the accidents in fog condition had occurred in December (p=0.000), February (p=0.001) and November (p=0.012). The number of related injuries at 2 AM (p = 0.032), 3 AM (p = 0.019), and 2 PM (p = 0.02) signicantly reduced. Moreover, the rate of mortality at 2 AM (p = 0.005), 9 AM (p = 0.012), 11 AM (p = 0.004), 13 PM (p = 0.008) and 19 PM (p = 0.003) were signicantly lower. The rate of injuries was signicantly higher in Anzali (P = 0.042) and Rasht (P = 0.018), respectively. Conclusion: In the fog, the fatality rate was signicantly higher than that of other conditions. The pattern of accidents in fog decreased with age and increased in accidents involving female drivers. Most of the fatal accidents had occurred in fog in November, January and February, respectively. Road accidents in Anzali and Rasht cities were more fatal.


Introduction
Weather condition may not be a major factor in causing road tra c accidents (RTAs) but it is undoubtedly one of the important environmental components (1,2). According to the NHTSA reports, the weather condition is responsible for 22% of damages, 19% of injuries and 16% of RTAs between 2005 and 2014 (3). Fog is one of the most important atmospheric hazards which directly affects transportation safety (4,5). In China, the analysis of the data from 2005 to 2014 showed that the average nancial cost of RTAs was highest on foggy days compared with that of other bad weather conditions (6). In Korea, the results of the statistical analysis of RTAs in 2014 suggested that 0.5% of all RTAs occurred in foggy weather, which seems to be low compared with the occurrence of accidents on sunny days (80.3%), rainy days (9.6%) and snowy days (1.1%) (7). In another study, the rate of fatality was 6.9% on foggy days which was the highest rate compared with 1.9% on sunny, 2.5% on rainy, and 2.5% on snowy days (7). Furthermore, according to reports, the mortalities due to multi-vehicle chain crash are more likely to occur in foggy and snowy conditions (7), with a risk ratio (RR) of 6 times for rainy, 158 for snowy and 171 for foggy weather (8).
Reduced visibility in fog condition can be a real threat since the ndings of the Transport Laboratory Research have shown that visibility less than 150 meters is very dangerous for drivers (9). Compared with other weather conditions, fog disturbs drivers due to sudden and irregular occurrence, increasing the likelihood of occurrence of accidents (1) particularly accidents with higher severity. According to the CRII, in foggy weather, the risk of accidents has a 40% increase (10). In sudden fog, drivers are usually unprepared for fog. Moreover, the foggy condition may affect the driver's perception and according to reports, there is up to 60% overestimation of distances in the fog (11). Under this weather condition, especially in thick fog, images and distances are blurred which may cause a slowdown in tra c volume and an increased risk of RTAs occurrence. Moreover, the reduced speed of vehicles and the resulting delay can increase stress and mental pressure during driving. These important factors may lead to increased road hazards in foggy weather (1,4,11). In addition to perception, the driver's behavior and performance are also affected by low vision (1). Some drivers slow down their speed while some others may increase it. This behavioral difference in response to low vision disrupts the tra c ow (10). In clear weather condition, people tend to drive faster due to good visibility and clear weather than in fog (1). Lane-keeping performance in the foggy condition is another example of the driver's behavior in fog. Drivers in low visibility have a higher standard deviation in lane-keeping (poor keeping of the lane in which they drive), i.e. drivers in fog compared to drivers who drive in good weather condition with high visibility always change lanes concerning low visibility of tra c signs. This is extremely dangerous and multiples the risk of collision with vehicles coming from the front in the opposite direction (1,12,13). Other behaviors include headway distance control in the foggy and misty condition which can be a vital factor. Most drivers tend to reduce their headway distance in this situation because it is easier to notice them (14). A simulated study found that drivers have a shorter headway space in fog, most of whom are unaware of this. Dense fog increases the perception of risk and uncertainty of the environment which makes the drivers reduce their headway distance for more safety. Therefore, shortening the headway distance in fog is not mainly a result of slowing down but probably due to the mental sense of safety in drivers (2). As a result, this behavior increases the likelihood of a rear-end collision (2,15,16). Drivers' skills in fog vary according to their individual characteristics. Elderly drivers may be at greater risk of injury due to reduced ability to predict possible accidents in foggy weather (14). More experienced drivers tend to signi cantly reduce their speed in the fog more than novice ones (17). There is also evidence that usually older drivers get close to the front car 21% less than novice drivers to have a better vision but they do not have enough time to react appropriately if accident occurs (14). Most of these accidents occurred in the morning in December and February. Front to head and rear-end collisions were the most common types of accidents in fog condition (18).
The limited number of epidemiological studies on the fatal accidents in foggy weather in Iran as well as the differences between our country and other countries in related dimensions such as different scienti c views (19)(20)(21) and different safety levels emphasize the need for undertaking such studies. These remarkable differences include vehicle type, availability and use of seatbelt and airbag, features of road design, driving styles and response time of emergency services (EMS) (22)(23)(24). Despite the frequent fog in some roads in northern Iran (Guilan province) and its high risk of accident, few studies have been conducted in this area. The hotspots of RTAs need to be identi ed in the fog so with greater knowledge, the number of these accidents can be minimized through e cient tra c management in hazardous areas. The purpose of this study was to determine the pattern of crashes in fog and to detect the relationship between factors related to driver, road and accident with injury levels in Guilan from 2014 to 2018.

Methods
This is a retrospective descriptive-analytical study. Inclusion criteria were the occurrence of RTAs during the period of 2014 to 2018 in Guilan province in fog condition. RTAs with incomplete recorded data in the database were not entered into the study. The data required for this study were collected from two sources: 1. Meteorological data: This data included daily information about maximum, minimum and average temperature in Celsius, mean daily rainfall, and mean daily air pollution which were obtained from Guilan Meteorological Organization and from Guilan meteorological stations in https://www.wunderground.com website. This center has appropriate and up-to-date technical equipment to measure and record the related data. The data were sorted based on days from March 21, 2014 to March 20, 2018. Foggy days were found according to meteorological station data. 2. Data related to tra c accident variables: Data on the number of accidents with vehicle damages, as well as the number of accidents with mortalities and morbidities/injuries were collected and entered into Excel software which were then determined and sorted on a daily basis, using the PivotTable command (mortality, morbidity, and vehicle damage).
For a descriptive assessment of the status of road accident on foggy days, the accidents which had occurred in fog were separated from normal weather and the differences in the severity of accident, the time of accident, the age and sex of the driver at-fault, the time and day and month of the accident, the city of the accident, the situation of the road, geometrical status of the road and the type of vehicle were identi ed and compared. The number of deaths, injuries and accidents were collected daily. After data collection, all of the data were entered into STATA software and analyzed by multivariate regression model for the presence or absence of any relationship among variables. P and Z were used to examine the signi cance between dependent and independent variables. R2 statistic was used for studying goodness of t.

Results
A total of 467 accidents occurred in the foggy weather in the cities of Guilan during 2014-2018. Most of the accidents occurred in Rasht, Astara, and Talesh, respectively, and no accidents occurred in Lahijan in foggy weather. However, in per thousand accidents, the highest number of accidents belonged to Astara, Talesh, and Rasht, respectively ( Fig. 1).
The mean age of the drivers at-fault in non-foggy (normal) compared with foggy weather was 35.95% versus 37.03%. However, there was no signi cant difference between the two groups (p = 0.11). The fatalities of male drivers in normal compared with fog was 87.53% versus 91.6%. The difference was statistically signi cant (p = 0.005). In regression model, considering death rate due to road accidents in foggy weather in Guilan between 2014 and 2018, in terms of age, sex, road type and road geometry, IRR = 0.935 and p = 0.039 indicated that with increasing age, the likelihood of death in tra c accidents could signi cantly reduce. Whereas there was a signi cant increase in death rate (P = 0.001) in accidents involving a female driver (Table 1). and p = 0.012, respectively. However, in June, July, August and September, the fatality rate in foggy weather was signi cantly lower. In other months, no signi cant relationship between the month and the occurrence of accidents was noticed ( Table 2). After controlling for the variables of age, sex, road type and geometry in the regression model, IRR = 1.03 and p = 0.616 indicated no signi cant difference in the rate of injuries in foggy and normal weather conditions (Table 3). Considering the foggy condition and mortality rate of IRR = 1.47 and P = 0.078, the regression model showed that the mortality rate was higher in foggy weather in 90% con dence interval (Table 4). There was no signi cant relationship between the variables of age, sex, road type and road geometry in the regression model of injuries in road accidents in the foggy weather. The regression model of mortalities due to tra c accidents in Guilan from 2014 to 2018 proposed no statistically signi cant relationship between the site of accident and mortality rate. However, the highest mortality rate belonged to Someh Sara with IRR = 1.5% (insu cient data on the exact number of fatalities in foggy weather in the cities of Astara, Anzali, Langrud, Talesh, Fouman, Roudbar, Masal, Amlash, Siahkal, and Shaft). Moreover, the assessment of injuries in the model showed that the rate of injuries due to accidents which had occurred in the aforementioned period was signi cantly higher in Anzali (IRR = 5.923 and p = 0.042) and Rasht (IRR = 2.432 and p = 0.018), respectively (su cient data on the extent of injuries due to road accidents from 2014 to 2018 in the cities of Langroud, Some Sara and Amlash were not available).

Discussion
A notable nding of this study was that although the rate of injuries was signi cantly higher in Anzali and Rasht, accidents occurring in foggy condition caused huge damages and injuries in most of the cities. A study of all areas of the United States except Northwest and New England reported similar results (25). In an Iranian three-year study on the Firouzkooh-Sari road, 930 road crashes were identi ed over a distance of 131 kilometers. The highest risk of fatal accidents belonged to the distance of 98, 100, 106, 109 and 110 kilometers, and due to the installation of fog lights at 9 to 18 kilometers (before and after the mountain pass or road curve), these types of accidents were prevented to a huge extent (26). Therefore, it is likely that the differences in the magnitude and severity of accidents be due to the installation of lighting and fog lights and other road facilities in different cities of Guilan with varying numbers and qualities.
According to the results of the present study, the mortality rate due to RTAs decreased signi cantly with age. In fact, this type of accidents in northern Iran follows a speci c age pattern in terms of rate of mortality of drivers. In a similar research which scrutinized the relationship between all certain weather conditions including rain, fog, dust, etc. and tra c accidents, signi cantly less number of RTAs were observed in older drivers (40-60) (27). There are various reports insisting that older drivers may be at greater risk of injury due to the reduced ability to detect impending events and fail to maintain a safe headway distance (14). If we assume age and experience as two similar concepts, a study reported that more experienced drivers compared with novice drivers were signi cantly less likely to lower their speed (17). Experienced drivers are also less likely to approach the front vehicle than novice drivers, which gives them enough time to respond appropriately (14). However, these differences may be related to the different age range included in each study.
The pattern of RTAs in the foggy condition was signi cantly associated with the male sex. That is, men were more likely to have accident than women in such condition, however, the likelihood of death dramatically increased in accidents where a female driver was at-fault. In a similar study, female drivers were more likely to deviate from the straight path in fog and involved in a single-vehicle accident in dense fog. It is noteworthy that male drivers usually drive at high speeds in foggy conditions since they usually have more con dence in their driving skills than female drivers (28). But in a study in Pakistan's urban areas, men were more likely to be involved in tra c accidents than women in all types of weather (27). Also, according to a logistic model, sex was the only characteristic associated with the deceleration of the vehicle in foggy weather. Evidence suggests that the average speed of women is more affected by fog than men at speeds of up to 10 mph. This means that women are probably more cautious when driving in foggy conditions and have a more responsible attitude than men because male drivers are more likely to risk speeding when the immediate dangers are unclear (29,30). Still, more detailed studies are required.
In the present study, there were no signi cant differences in the number of injuries in foggy weather compared with other conditions, but the mortality rate was signi cantly higher than other climatic conditions. In one Iranian study, however, the number of accidents in rainy weather (73.8%) which led to damages was higher than the number of accidents leading to death, which was due to the slower speed of vehicles in such conditions compared with sunny weather (31). In a time series analysis in northern Iran, the rst day of rainfall after several dry periods was signi cantly associated with morbidities and mortalities due to accidents (32). This may imply that fatal accidents in fog are so severe that they lead to immediate death because of the severity of the injuries.
Most of the fatal accidents related to foggy weather occurred in the autumn (November) and winter (January and February) whereas the number of these accidents was signi cantly lower in warm months including June, July, August and September. A study on accidents in the USA in cold weather in winter season published similar ndings according to which the highest rate of deaths due to RTAs dedicated to cold days of winter (33). Furthermore, Abdel-Atyet al. (2011), found similar results where the highest number of accidents had been recorded for smog weather occurring during winter (34). According to the data obtained from the FARS database, 40% of fatalities occurred in fog, smoke, dust storm and sand storm, with most of them occurring in December, March, and April (35). In an Iranian study, dense fog (less than 40 meter vision sight) was highest in February (31). Therefore, in the present study, the pattern of RTAs in the fog had a seasonal pattern and had a signi cant increase in winter compared with summer. However, previous studies have reported that drivers tend to go faster in clean weather due to good visibility so they usually speed up in clear weather in foggy weather (1,5). Although the dense fog is often observed during the winter when the average speed of vehicles is slower and the number of tourists is remarkably lower than that of other times of the year (36), the number of accidents under foggy weather is surprisingly high which needs more investigations.
Studying the mortality rate of RTAs in foggy weather at different hours of the day in Guilan province from 2014 to 2018 revealed that the mortality rate signi cantly decreased at 2 AM, 11 PM, 13 PM and 19 PM. The number of injuries decreased dramatically at 2 AM, 3 AM and 2 PM. In the study by Black and Mote (2015), the highest number of fatalities in fog due to RTAs belonged to evening time (33), and in another study, the most frequent fatalities were in smog condition in the morning in winter (34). However, due to the lack of available data related to certain times of the day/night, an accurate analysis was not possible.
Examining the mortality and injury rates of RTAs in the fog revealed that the incidence of these accidents did not change signi cantly in Guilan from 2014 to 2018. Thus, it can be concluded that with advances in knowledge, technology and resource development, etc., there has been no reduction in these types of accidents within this time and there is a need for more effort to prevent such accidents. However, it should be noted that atmospheric phenomena can only exacerbate the cause of accidents and they should not be accounted for as the mere causes of these events. A study on bad weather conditions (snow, ice, rain, and fog) in Iran speci ed that these conditions caused a total increase in the risk of accidents by 6.6% (26).
Although natural phenomena such as fog, snow, cold weather, etc. cannot be prevented, some variables can be interfered with. For example, the driver's information on safe driving in such weather conditions can be improved by installation of not only the common road warning signs but also some signs containing brief messages about driving skills in foggy weather in hot spots with a higher likelihood of fog (37).
One of the limitations of this study was that the information obtained from tra c police about several variables such as the hour of accidents leading to mortality or morbidity and the city where accidents took place were incomplete. Moreover, the accidents which involved animals or hitting a tra c sign and the accidents not being reported by the driver were excluded from the study. Therefore, the number of recorded accidents may not showcase the actual number of accidents (under-report). However, our ndings were similar to those of related conducted studies in other countries.

Conclusion
In the fog weather, the mortality rate was signi cantly higher than that of other climates, and with increasing age, the likelihood of fatalities due to accidents in fog condition decreased while the likelihood of death increased signi cantly among female drivers. Most of the fatal accidents occurred in November, January, and February, respectively. The incidence of injuries due to accidents in Anzali and Rasht cities was signi cantly higher. In addition, our investigation on the fatalities and injuries due to tra c accidents in foggy weather demonstrated that the number of accidents in Guilan has not seen a signi cant change from 2014 to 2018. This indicates a low attention to this important, life-threatening issue in northern Iran.
Therefore, according to our ndings, it seems that some predictive factors such as age and sex can be effective in reducing fatalities due to road accidents in Rasht and Anzali during November, January and February.
Abbreviations NHTSA: [1] National Highway Tra c Safety Administration, CRII: Crash Risk Increase Indicator Declarations Acknowledgment The authors would like to offer their special thanks to Ms. Fatemeh Javadi for translating the manuscript.
Authors' contributions N KH designed the study, E HR performed the analyses and drafted the manuscript, H SK collected the data, L KE contributed to the interpretation of the results and writing the manuscript. All authors read and approved the nal manuscript.

None
Availability of data and material The data used and analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
This study has been approved by The Ethics Committee of Guilan University of Medical Sciences, Rasht, Iran (IR.GUMS.REC.1398.202).

Consent for publication
Not applicable

Competing interests
The authors declare that they have no con ict of interest  Figure 1 Frequency of tra c accidents in each city per 1000 accidents in foggy weather in Guilan from 2014 to