ANALYSIS OF ROAD TRAFFIC MORTALITY IN IRAN BY SETAREH KORDI B.Sc., Shahid Beheshti University,2006 M.Sc., Shahid Beheshti University,2009 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS IN DEVELOPMENT ECONOMICS UNIVERSITY OF NORTHERN BRITISH COLUMBIA August 2013 © Setareh Kordi, 2013 UMI Number: 1525712 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. Di!ss0?t&iori Publishing UMI 1525712 Published by ProQuest LLC 2014. Copyright in the Dissertation held by the Author. Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code. ProQuest LLC 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106-1346 Abstract The objective o f this study is to analyze demographic, socioeconomic and circumstantial profiles of 16,556 victims o f road traffic crashes in Iran during 2009. The descriptive analysis indicates that the vast majority o f victims (79%) were males, over 44% o f them were o f younger age (15-34 years), and 41% o f them were drivers. Also about 61% o f all deaths were because o f head injury and over 2/3 of the crashes happened in out-of-town areas. The results from the multinomial logistic regression models for those variables with significant bivariate associations show that males and people under 60 are more likely to die as a driver. Drivers and victims involved in out-of-town crashes mostly tend to die at the scene o f the crash. Also the estimated economic costs o f road traffic crashes in the study period were indirectly estimated at US$17 billion or 7% of the country’s gross domestic product. Table of Contents Abstract...............................................................................................................................................ii Table o f Contents............................................................................................................................. iii List of Tables..................................................................................................................................... v List o f Figures..................................................................................................................................vii Acknowledgement......................................................................................................................... viii 1 Introduction.................................................................................................................................1 1.1 The Significance o f Road Traffic Injuries and Fatalities................................................. 1 1.2 Why Study Road Traffic Mortality in Ira n ...................................................................... 6 1.3 The Purpose of this Study.................................................................................................. 7 1.4 Research Questions.............................................................................................................7 1.5 The key findings................................................................................................................. 8 2 Why Study Road Traffic Mortality in Iran............................................................................ 12 3 Literature review ..................................................................................................................... 21 3.1 Rapid Motorization, Economic Growth and Road Traffic Injuries.............................21 3.2 The Profiles of the Victims of Road Traffic Crashes.................................................... 25 3.2.1 Road User Status....................................................................................................... 25 3.2.2 Demographic Status of Victims...............................................................................27 3.2.3 Socioeconomic Status............................................................................................... 29 3.3 4 5 The Economic Cost of Road Traffic Injuries.................................................................30 Data &Methodology................................................................................................................ 34 4.1 The D ataset........................................................................................................................34 4.2 Methodology..................................................................................................................... 35 4.2.1 Descriptive Analysis................................................................................................. 35 4.2.2 Bivariate Analysis......................................................................................................35 4.2.3 Multivariate Analysis................................................................................................ 37 4.2.4 Estimating the Economic Cost o f Road Fatalities................................................. 37 Analyses o f D a ta..................................................................................................................... 40 5.1 Descriptive Analysis.........................................................................................................40 5.1.1 The Demographic Profiles of the Victims............................................................. 40 5.1.2 Socioeconomic Profiles of the V ictim s................................................................. 44 5.1.3 The Circumstantial Profiles o f the Crashes...........................................................46 5.2 Bivariate Analysis............................................................................................................. 51 5.2.1 The Association of Victim's Status with Demographic and Socioeconomic Factors....................................................................................................................................... 51 5.2.2 The Association of the Place o f Death with Demographic and Other Circumstantial Factors.............................................................................................................54 5.3 Multivariate Analysis........................................................................................................59 5.3.1 Multinomial Logistic Regression Results for Victim's Status, Gender, Age and Education.................................................................................................................................. 59 5.3.2 Multinomial Logistic Regression Results for the Place o f Death Versus Age and other Circumstantial Factors............................................................................................ 65 5.4 6 Calculating the Economic Cost o f Road Traffic Fatalities...........................................75 Conclusion................................................................................................................................. 77 6.1 Empirical Findings............................................................................................................77 6.2 Policy implications............................................................................................................81 6.3 Limitations of the study....................................................................................................82 6.4 Recommendations for future research........................................................................... 83 References........................................................................................................................................ 84 iv List of Tables Table 1-1Top ten leading causes o f death, 2004 and 2030 compared..........................................2 Table 1-2 Road traffic fatality rates (per 100,000 population), by WHO region and income group................................................................................................................................................... 3 Table 1-3 The total number o f crashes, killed and injured people due to road traffic crashes in Iran.......................................................................................................................................................7 Table 2-1 Road Traffic Injury death rate in Iran compared with other countries in 2 0 0 7 ...... 13 Table 2-2 Road Traffic Injury death rate in Iran compared with other middle-income countries in the Eastern Mediterranean Region in 2007.............................................................. 14 Table 2-3 The total number o f crashes, killed and injured people due to road traffic crashes in Turkey and Iran.................................................................................................................................17 Table 2-4 Mortality rate per 100,000 population in Turkey and Iran.........................................18 Table 2-5 Leading causes o f death in Iran.....................................................................................19 Table 3-1 GDP, motorization, and road traffic crashes in China, 1980-2005.......................... 24 Table 3-2 GNP and road crash costs by region (USSbillion)..................................................... 31 Table 5-1 The distribution of road traffic deaths by gender....................................................... 40 Table 5-2 The distribution of road traffic deaths by age groups................................................42 Table 5-3 The distribution o f victims by educational levels.....................................................44 Table 5-4 The distribution o f victims by occupation................................................................ 46 Table 5-5 The distribution o f victims by status.......................................................................... 47 Table 5-6 The distribution o f victims by mode o f transport...................................................... 48 Table 5-7 The distribution o f the location of crash..................................................................... 49 Table 5-8 The distribution o f victims by the place o f death....................................................... 50 Table 5-9 The distribution o f the reasons o f death...................................................................... 51 Table 5-10 Cross tabulation between status and gender..............................................................52 Table 5-11 Cross tabulation between status and age................................................................... 53 Table 5-12 Cross tabulation between status and education.........................................................54 Table 5-13 Cross tabulation between the place o f death and age..............................................55 Table 5-14 Cross tabulation between the reason o f death and age.............................................56 Table 5-15 Cross tabulation between the place o f death and status......................................... 57 Table 5-16 Cross tabulation between the place o f death and the reason o f death...................58 Table 5-17 Cross tabulation between the place o f death and location o f the crash..................59 Table 5-18 The odds ratios of different status for males compared to females....................... 60 Table 5-19 The odds ratios of comparing the change in the status o f the victims for those aged 60 years old or above compared to those under 60 years o ld ............................................61 Table 5-20 The odds ratios of comparing the change in the status o f the victims for those with primary level o f education compared to those with no level o f education................................62 Table 5-21 The odds ratios o f comparing the change in the status o f the victims for those with high school level o f education compared to those with no level o f education......................... 63 Table 5-22 The odds ratios o f comparing the change in the status o f the victims for those with university level o f education compared to those with no level o f education.............................64 Table 5-23 The odds ratios of comparing the change in the place of death for those aged 60 years old or above compared to those under 6 0 ...........................................................................65 Table 5-24 The odds ratios o f comparing the change in the place of the death for driver compared to pedestrian................................................................................................................... 67 Table 5-25 The odds ratios o f comparing the change in the place of the death for passenger compared to pedestrian................................................................................................................... 69 Table 5-26 The odds ratios o f comparing the change in the place of death when the reason of death changes from bleeding to head injury................................................................................. 70 Table 5-27 The odds ratios o f comparing the change in the place o f death when the reason of death changes from bleeding to multiple fractures.......................................................................72 Table 5-28 The odds ratios of comparing the change in the place of death for crashes occurred in in-town areas compared to those occurred in out-of-town areas............................73 List of Figures Figure 1-1 Population, road traffic deaths and registered vehicles by income group................4 Figure 2-1 Road traffic crash trends in Iran...................................................................................15 Figure 2-2 Trend o f the total number o f fatalities due to road traffic crashes in Iran .............. 16 Figure 2-3 Trend o f the total number o f injured people due to road traffic crashes in Iran.... 16 Figure 3-1 Fatality Risk and Per Capita Income, 1975-2000..................................................... 23 Figure 5-1 The distribution o f road traffic deaths by sex and age group..................................43 Acknowledgement I would like to express my deep gratitude to Dr. Jalil Safaei for giving me the opportunity to be his student, providing me with his support, and having endless patience and understanding. His commitment to excellence, together with his excellent knowledge, has been essential for the completion of this work. I have truly enjoyed being his student. I would like to express sincere appreciation to the members o f the committee Dr. Baotai Wang and Dr. Mamdouh Shubair. Appreciation is also expressed to the Road Maintenance and Transportation Organization o f Iran for providing the dataset o f this study. Finally, I dedicate this thesis to my family for all their unconditional support which have greatly empowered me throughout the course o f my life. Also, special thanks are due to my best friend Saeed. Chapter 1: Introduction 1.1 The Significance of Road Traffic Injuries and Fatalities The problem o f road traffic injury existed before motor vehicles have been invented. However, with the emergence of motor vehicles (e.g. cars, buses, trucks and motorcycles), the number o f road traffic injuries has increased rapidly. The first road traffic injury was recorded in New York City on May 30, 1896 between a cyclist and a motor vehicle (World’s first road death, 2011). One month later in the same year, a London pedestrian was recorded the first death due to road traffic crashes1 (World’s first road death, 2011). Since then, road traffic crashes have been reported all around the world. Road traffic injuries became the second leading cause o f death among children aged 5-14 years, the first leading cause o f death among young people aged 15-29 years and the third leading cause of death for people aged between 30-44 years around the world in 2004 (WHO, 2009). Most recent data indicates that each year about 1.3 million people die from road traffic injuries and 20-50 million people are injured worldwide (WHO, 2011). Around 90% o f these deaths and injuries take place in low-income and middle-income countries, which comprise only 48% o f the world's registered vehicles (WHO, 2009). Pedestrians, cyclists and motorcyclists, which are categorized under “vulnerable road users”, amounts for nearly half (46%) o f those dying on the roads worldwide (WHO, 2011, P.3). In 1 Peden et al. (2004) use the term "crash" instead o f "accident". They believed that the term "accident" denote to an event which is spontaneous, unpredictable and unmanageable while the term "crash" can give the idea of an event that can be controlled and analyzed (Peden et al., 2004, P.7). 1 addition to grief and suffering, road traffic crashes cause considerable economic losses to victims, their families and the society as a whole. Such loss has been estimated to add up to 1 to 3 percent o f gross national product in most countries (WHO, 2011). Table 1-1 compares the ten leading causes o f death in 2004 with those estimated for 2030. While road traffic deaths in recent decades have become stable or decreased in many high-income countries, they are predicted to increase significantly in low-income and middle-income countries (WHO, 2009). WHO predicts that road traffic injuries will increase and become as the fifth leading cause o f death in 2030, and the number o f deaths are estimated to increase to 2.4 million fatalities annually (WHO, 2009). Table 1-1 Top ten leading causes of death, 2004 and 2030 compared Rank 2004 Disease or Injury 1 2 Ischaemic heart disease Cerebrovascular disease 3 Lower respiratory infections 4 5 6 7 8 9 10 Rank 1 2 3 Chronic obstructive pulmonary disease Diarrhoeal diseases HIV/AIDS Tuberculosis Trachea, bronchus, lung cancers 2030 Disease or Injury Ischaemic heart disease Cerebrovascular disease Chronic obstructive pulmonary disease 4 Lower respiratory infections 5 6 Road traffic injuries Trachea, bronchus, lung cancers 7 Diabetes mellitus Hypertensive heart disease 8 9 Stomach cancer Road traffic injuries HIV/AIDS Prematurity & low-birth weight 10 (Source: Global Status Report on Road Safety, 2009, P.ix) Road traffic mortality rates in different parts of the world are significantly different. They also vary with the level o f income in each country. Table 1-2 shows road traffic injury mortality rates per 100,000 population categorized by income groups for each o f the six WHO regions - namely, Africa, the Americas, South-East Asia, Europe, Eastern Mediterranean, and Western Pacific. 2 From Table 1-2, it could be concluded that the middle-income countries o f the Eastern Mediterranean Region have the highest mortality rate (35.8 per 100,000 population), followed closely by the low-income and middle-income countries o f the African Region at 32.3 and 32.2 per 100,000 population, respectively. The high-income countries o f the western Pacific Region have the lowest mortality rate (7.2 per 100,000 population), followed closely by the high income countries of the European Region (7.9 per 100,000 population). As Table 1-2 shows, the global mortality rates for low-income and middle-income countries are much higher than mortality rates in high-income countries. Table 1-2 Road traffic fatality rates (per 100,000 population)2, by WHO region and income group Middle-income Low-income Total WHO region High-income 32.2 32.3 32.3 African 13.4 17.3 15.8 The Americas 16.5 16.6 16.7 South-East Asia 13.4 7.9 19.3 12.2 European 35.8 27.5 32.2 Eastern 28.5 Mediterranean 15.7 7.2 16.9 15.6 Western Pacific 18.8 10.3 19.5 21.5 Global (Source: Decade of Action for Road Safety 2011-2020- Saving Millions of Lives, 2011, P.4) Note: A dash in a cell indicates no country existed in the respective income category and region. Interestingly, the majority of road traffic deaths happen in Middle-income countries. As Figure 1-1 shows, about 49.6% of all road traffic deaths occur in middle-income countries, while they just comprise around 38.7% o f all registered motorized vehicles around the world. Low-income countries account for 41.9% of the world's road traffic deaths, while they have only 9.2% o f the world's registered vehicles. Therefore, the road 2 The definition of road traffic fatality is any person killed immediately or dying within 30 days as a result o f a road traffic injury accident (Jones et al., 2008). 3 traffic deaths o f low-income countries along with those o f middle-income countries amount to about 91.5% o f the world's fatalities on road, while these countries comprise only 47.9% o f the registered vehicles around the world (WHO, 2009). Figure 1-1 Population, road traffic deaths and registered vehicles by income group Population Road traffic deaths' Registered vehicles Note: H1C stands for high-income countries, MIC for middle-income countries and LIC for low- income countries. (Source: Global Status Report on Road Safety- Time for Action, 2009, P. 22) A number o f factors have been identified by Peden et al. (2004) as risk factors for road traffic crashes. These factors include speed, pedestrians and cyclists, young drivers and riders, alcohol, medicinal and recreational drugs, driver fatigue, hand-held mobile telephones, inadequate visibility, and road-related factors (Peden et al., 2004). Finch (1994) concluded that increasing the mean o f traffic speed by 1 km/h leads to a 3% increase in the incidence o f the injury crashes or to a 4-5% increase in the fatal crashes (Finch, 1994). Taylor et al. (2000) also indicated that there is a positive correlation between the speed o f traffic and the frequency o f accident. The higher the speed, the greater the frequency o f accident (Taylor et al., 2000). Compton et al. (2002) concluded that the relative risk o f crash involvement begins to increase considerably at a blood alcohol concentration (BAC) level o f above 0.03 g/dl (Compton et al., 2002) 4 Besides risk factors influencing crash involvement, Peden et al. (2004) introduced the risk factors influencing injury severity. These factors are lack o f in-vehicle crash protection, non-use o f crash helmets by two-wheeled vehicle users, non-use of seat-belts and child restraints in motor vehicles, and roadside objects (Peden et al., 2004). Not using seat-belts is a main risk for car occupants. In frontal-impact crashes, the most frequent and serious head injuries happens to occupants that are not using seat-belts (Kajzer et al., 1992). Because of the heavy burden of road traffic injury on the victims, their families and the economies, and because it is a serious issue for public health, international organizations such as World Bank and World Health Organization have taken some actions to identify its major determinants and risk factors in order to come up with effective intervention strategies to prevent road traffic injuries. The publication of “the World Report on Road Traffic Injury Prevention”, on World Health Day in 2004, the recognition of the third Sunday in November o f every year as the “World Day o f Remembrance for Road Traffic Victims” (United Nations General Assembly Report, 2007), and the declaration o f the period 2011-2020 as “the Decade of Action for Road Safety” by the General Assembly On May 10, 2010, with a purpose of decreasing the level o f road traffic fatalities worldwide by implementing policies at the national, regional and global levels (The United Nations and Road Safety, 2011) are among major actions of the international organizations. 5 1.2 Why Study Road Traffic Mortality in Iran As I discussed earlier, road traffic injuries are a serious global public health issue. In general, the death rate in developing countries is higher than that o f developed countries. It is anticipated that this gap is going to increase further over the following few decades as the death rate in developing countries will continue to increase while it will continue to fall in the developed countries (Bhalla et al., 2008). Iran as a developing country is no exception. World Bank in its road safety project on Iran in 2006 addressed that "the road safety situation in Iran is one of the worst in the world and it has severe consequences on the population and the economy" (World Bank, 2006, P. 2). In 2007, Iran ranked as the eleventh worst country among 178 other countries in terms o f road traffic death rate per 100,000 population with the rate o f 35.8 per 100,000 population (WHO, 2009). Table 1-3 shows the total number of crashes, killed people, and injured people for the year 2001 and 2010 and it also shows how much these numbers has changed within those years. As shown in that table, the total number o f crashes in 2010 compared to 2001 has been increased by 125%. Subsequently, the number o f killed people and injured people has been increased by 18% and 166%, respectively. Although the number o f deaths has not increased as much as the number o f crashes or injuries, it is still growing. The mortality rate per 100,000 population for year 2001 was 30.4 and it has increased to 31.1 in 2010. Moreover, road traffic injuries ranked as the third leading cause o f death in Iran and accounted for 10.3% o f all deaths in 2002 after myocardial infarction and cerebral vascular diseases (Bhalla et al., 2008). It can be concluded that the road safety situation in Iran is very bad and each year instead of improving is just deteriorating. I will discuss the road traffic injury situation in Iran in depth in chapter 2. 6 Table 1-3 The total number o f crashes, killed and injured people due to road traffic crashes in Iran Year 2001 2010 Total Crashes % increased from 2001 to 2010 346,853 782,170 125.50% Iran % increased from 2001 to Total killed 2010 19,727 17.85% 23,249 Total Injured % increased from 2001 to 2010 117,566 312,745 166.02% (Source: Adopted from Iran Statistical Year Book and Statistical Year Book of Iran Road Maintenance & Transportation Organization, 2012) 1.3 The Purpose of this Study The main objective o f this study is to examine the demographic, socioeconomic and circumstantial profiles o f the victims o f road traffic fatalities and investigate various road crash situations that may be related to specific demographic, socioeconomic and circumstantial profiles. To my knowledge, no previous study on road traffic injuries in Iran has systematically analyzed the demographic, socioeconomic and circumstantial profiles o f the road traffic victims. This information would be valuable for the formation o f policies in order to decrease road traffic crashes and target those policies to the most vulnerable groups prone to road traffic injuries. Moreover, this study attempts to estimate the economic burden o f the road traffic fatalities in Iran. 1.4 Research Questions 1- What are the demographic, socioeconomic, and circumstantial profiles o f the road crash victims? 2- Which demographic or socioeconomic factors affect the victim’s status? 3- Which demographic or circumstantial factors affect the place o f death? 4- How much is the economic cost or burden o f road traffic fatalities? 7 1.5 The key findings The dataset o f this study consists o f a total o f 16,556 individuals who died from the road traffic crashes in Iran during March 21 to November 21, 2009. The male to female death ratio was 3.79. Male deaths accounted for 79% of the total deaths. Age group 15 to 24 years comprises the highest number o f deaths (22% of the total deaths) followed by the age group 25 to 34 years with about 20% o f the total deaths. It is observed that road mortalities increase by age and peaks in the age group o f 15 to 24 years for males and in the age group o f 25-34 for females, and decline for older age groups. The difference in the male to female mortality is more pronounced among young adults. These differences are relatively smaller among children (<14 years) and the frail elderly (>85 years) groups. The analysis o f the socioeconomic profile of the victims shows that about half o f the victims had no education or only primary education. It also shows that those with business occupations and “blue-collar” jobs were over-represented among the victims. More frequent driving (by businessmen) and commute (by blue-collar workers) is noted as causing such outcome. The analysis o f the circumstantial profiles o f the road traffic crashes reveals that the largest group o f victims are drivers, followed by passengers and pedestrians. Moreover, it is shown that more than two thirds of the crashes are those that happened out-of-town areas. Over half o f the victims died at the scene o f the crash and almost 40% o f deaths happened in hospitals, due to the severity o f the inflicted injuries. Head injury figured as the main reason o f death accounting for over 61% o f all deaths. Such outcomes are taken as evidence o f acute severity of crashes in Iran that may have been caused by lack of 8 appropriate protective gear by drivers and passengers on the one hand, and lack o f timely road emergency services, on the other. Bivariate analysis were done to test the potential relationship between some circumstantial factors with demographic or socioeconomic variables. Strong statistically significant associations are found between victim’s status (as driver, passenger or pedestrian) and gender, age and education. The relationships between circumstantial, demographic and socioeconomic profiles o f the victims, has been demonstrated using cross tabulation between status and gender, status and age, and status and education. It can be seen that among female deceased 67.2% o f them were passenger and among male deceased over half o f them were driver (51.9%). Deceased aged under 60 years old are mostly tend to be the driver victims o f road traffic crashes. Over 50% o f deceased aged 60 years old or above were pedestrian. 53% o f driver had high school level of education. 49% o f pedestrian had no education and around 39% o f passenger had high school level o f education. Moreover, strong statistically significant associations are found between the place o f death and one demographic variable (age) and 3 other circumstantial variables, namelystatus, the reason o f death, and the location of the crash. From the related cross tabulation table between the place o f death and age, the place o f death and status, the place o f death and the reason o f death, and the place of death and the location o f the crash, it can be concluded that over 50% o f deceased aged under 60 years old died at the scene o f the crash which could be explained by their reason o f death which is mostly head injury (over 65%). Over 50% o f driver died at the scene of the crash. Among pedestrian, 53% o f them died in the hospital and over 58% o f passenger died at the scene o f the crash. 9 It also shows that about 53% o f people with head injury passed away at the scene o f the crash. 58% o f deceased were involved in the crashes happened in out-of-town areas died at the scene o f the crash which could be because of the higher severity o f the crashes happen in out o f town areas as traffic speed is normally higher on out-of-town roads and the quality and safety of those roads are typically poorer than roads in-town. Moreover, the monitoring o f traffic is usually less frequent and effective than that o f the in-town roads. To further explore the relationships between circumstantial, demographic and socioeconomic profiles o f the victims, I did a multivariate analysis using multinomial logistic regression models to estimate the odds ratios for different categories o f the victim’s status and the place of death as affected by gender, age, education and other circumstantial factors as appropriate. The multivariate analysis o f victims’ statues indicates that males and those under 60 years o f age are most likely to die as drivers. Whereas, females and those aged 60 or above are most likely to die as pedestrians or passengers (see Table 5-18, and 5-19). Interestingly, the analysis shows that those with more education are also more likely to die as drivers (see Table 5-20, 5-21 and 5-22). The multivariate analysis o f the place of death provides plausible results as well. The chances o f dying at the scene o f the crash and on the way to hospital or in the hospital are greater for people younger than 60 years old, drivers, victims with head injuries and multiple fractures, and victims o f crashes happened on out-of-town roads. Finally, as an add-on analysis, this study presents a rough estimation o f the economic burden of road traffic fatalities in Iran for the study period based on existing methodology. The total economic cost as a result o f road traffic fatalities within the time 10 frame of this study has been estimated around US$17 billion which amounts to 7% o f GDP o f Iran during March to November 2009. This is a substantial cost imposed on a developing country, and yet it does not capture the huge loss of life and suffering inflicted on the survivors and their families. In this chapter, I discussed the magnitude of road traffic injury problems around the world and brought an introduction to the country which will be the focus o f this study. I also presented the purpose o f this study, research questions and key findings. Chapter 2 will be focused on the road traffic injury and mortality situation in Iran. Chapter 3 provides the literature on road traffic injuries and mortalities. Chapter 4 describe the sample data in terms o f demographic and socioeconomic profiles o f the victims and the circumstances o f the crashes related to their death. It also specifies the methodology for further analysis of the data in chapter 5. The later chapter provides the results for analysis o f the associations between the mortality conditions and the factors deemed related to fatality. It also provides the results from logistic regression model to determine the potential role of demographic, socioeconomic and circumstantial factors in the conditions of fatality. Moreover, it provides a rough estimation o f the economic cost o f road traffic fatalities in Iran. Chapter 6 concludes the thesis. 11 Chapter 2: Why Study Road Traffic Mortality in Iran Iran as a populated and vast country does not have enough railway lines, and as a result more than 90% o f all the freights and people traveled through the roads (Ayati, 2009). The transport sector is an important sector given the high urbanization rate (about 69% in 2011) in Iran (United Nations, 2012). While most o f the country is mountainous land area, the distances between major cities and between these cities and the sea in the north and south o f the country is large, and while the country is located at one of the most important corridors o f international trade routes, the transport networks between cities and within cities are not enough regarding quantity and quality and even not good from safety perspective. These inadequacies along with a general disregard for traffic regulations due to low education, old vehicles, traffic jam, and low emergency services have created a very bad situation in Iran. In other words, each person who goes out to the street or highway in Iran runs a considerable risk o f injury or death. In 2009, WHO published the Global Status Report on Road Safety which is the first assessment on the status of road safety around the world. The report covered the data from 178 countries and areas. Table 2-1 compares the fatality rate per 100,000 population in Iran with those o f a selected number of countries around the world. Table 2-1 lists the countries from the highest to the lowest road traffic fatality rate, excluding a large number o f countries in the middle. As can be seen from this table, Iran places as the 11,h worst country in terms o f road traffic death rate per 100,000 population. 12 Table 2-1 Road Traffic Injury death rate in Iran compared with other countries in 2007 Eritrea Road Traffic Death Rate per 100,000 population 48.4 Cook Islands 45 2^ Egypt 41.6 y d Libyan Arab Jamahiriya (the) 40.5 4'* Afghanistan 39 5 ,h Iraq 38.1 6'* Angola 37.7 -j lh Niger (the) 37.7 8 ih 37.1 g th Gambia 36.6 10'* Iran 35.8 11'" Switzerland 4.9 174'" Singapore 4.8 175'" Netherland 4.8 176'" Uruguay 4.3 175'* Malta 3.4 176'* San Marino 3.2 177'* Country United Arab Emirates Rank r' 1.7 Marshal Island 178'* (Source: Adopted from Global Status Report on Road Safety- Time for Actions, WHO 2009, P. 240) Based on WHO classification, Iran is located in the Eastern Mediterranean Region. As I mentioned earlier, the middle-income countries o f the Eastern Mediterranean Region have the highest mortality rate (35.8 per 100,000 population). The Eastern Mediterranean Region comprises 22 countries. Table 2-2 compares the death rate per 100,000 population o f Iran with those of 10 other middle-income countries in the region for which data is available. As can be seen from the table, in the Eastern Mediterranean Region, Iran stands after Egypt, Libya, and Iraq for the highest death rate due to road traffic injuries (WHO, 2009). 13 Table 2-2 Road Traffic Injury death rate in Iran compared with other middle-income countries in the Eastern Mediterranean Region in 2007 Road Traffic Death Rate per 100,000 Population 41.6 Egypt 40.5 Libya 38.1 Iraq 35.8 Iran 34.7 Sudan 34.5 Tunisia 34.2 Jordan 32.9 Syria 28.5 Lebanon 28.3 Morocco 21.3 Oman (Source: Adopted from Global Status Report on Road Safety- Times for Actions, 2009, P. 240) Country Figure 2-1 shows the trend o f the absolute number of crashes in Iran from 2001 to 2010. A detailed look at the recent Iranian road traffic incidents for some recent years shows that the number of crashes occurred in-town or out-of-town areas has increased and reached to the peak o f 646,851 and 165,130 respectively, with a total o f 811,981 crashes during 2006-2007 (Iranian Calendar). Since then, the total number of crashes started to decrease but it has remained very high. For example, during March 2009 to March 2010, the total number o f crashes was about 702,512. 14 Figure 2-1 Road traffic crash trends in Iran 900000 800000 700000 600000 • 500000 400000 - • - T o t a l Out-of-Town Crashes : 200000 ■Total Crashes | 100000 (Source: Adopted from Iran Statistical Year Book, 2012) The number o f crashes in out-of-town areas is lower than that o f in-town areas but because of the higher speed in out-of-town areas, the severity o f crashes is much higher. Therefore, the number of people killed in out-of-town crashes is much higher than those killed in in-town crashes. During March 2009 to March 2010, 77% o f the crashes occurred in-town areas but it just accounts for 30.6% of deceased people in that year. On the other hand, 23% of the crashes occurred in out-of-town areas, while it accounts of 69.4% percent of the deceased people in the study year (Iran Statistical Year Book, 2012). Figures 2-2 and 2-3 demonstrate the trend of the absolute number o f fatalities and injuries in Iran during the decade o f 2001-2010, respectively. The highest number of deceased people during this period occurred in the 2005-2006 (Iranian Calendar) which was about 27,746 people. After that the number of fatalities started to decrease but it still remains very high. For example, during March 2010 to March 2011, the total number of fatalities was as high as 23,249. As can be seen from Table 2-3, the number o f injured 15 people due to road traffic crashes has been increasing throughout the decade despite a dip during 2007-2008. Figure 2-2 Trend of the total number of fatalities due to road traffic crashes in Iran 30000 25000 20000 15000 10000 5000 t o t a l K illed O 1-4 rl v4 O O (Source: Adopted from Iran Statistical Year Book of Road Maintenance & Transportation Organization, 2012, P. 175) Figure 2-3 Trend of the total number o f injured people due to road traffic crashes in Iran 350000 *' 300000 ............... ................ 250000 200000 * .... 150000 100000 ' 50000 H 0 J ....... ........ Total Injured * * * * * v*/ 4? 4 # •**“ ■ (Source: Adopted from Iran Statistical Year Book o f Road Maintenance & Transportation Organization, 2012, P. 175) To get a better sense o f the severity o f road crashes in Iran, the road traffic accidents of Iran will be compared with that of Turkey, which is one o f Iran’s neighboring countries with comparable demographics. Table 2-3 shows the total number of crashes, fatalities and injured people due to road traffic crashes and the fatality rate per 100,000 population in some selected years in Turkey and Iran based on their official data. In 2001, as a result o f 442,960 crashes in Turkey, 4,386 people were killed and 116,203 people were injured. As Table 2-3 shows 9 years later in 2010, the total number o f crashes reached to 1,106,201 in Turkey which led to only 4,045 deaths and 211,496 injured ones. Therefore, during 2001 to 2010, while the number of crashes has increased by 150%, the number o f deceased people has decreased by 8% in Turkey. On the other hand, during the same period, the number o f crashes in Iran has been increased by 125%, which has led to an increase in the number of deceased people by 18%. It is evident that while the number o f crashes in Turkey is higher than that of in Iran, the number o f people killed due to road traffic crashes in Turkey is much lower than that o f in Iran. Table 2-3 The total number of crashes, killed and injured people due to road traffic crashes in Turkey and Iran Year 2001 2007 2008 2009 2010 Total Crashes 442,960 825,561 950,120 1,053,346 1,106,201 Turkey Total killed 4,386 5,007 4,236 4,324 4,045 Total Injured 116,203 189,057 184,468 201,380 211,496 Total Crashes 346,853 750,250 780,352 702,512 782,170 Iran Total killed 19,727 22,918 23,362 22,974 23,249 Total Injured 117,566 245,418 272,877 295,179 312,745 (Source: Adopted from Turkish Statistical Institute, Iran Statistical Year Book and Iran Statistical Year Book of Road Maintenance & Transportation Organization, 2012) Table 2-4 compares the fatality rate per 100,000 population for Turkey and Iran for the same years as in Table 2-3. It can be seen that the fatality rate in Turkey has been generally decreasing while that of Iran has been increasing till 2008 before showing a negligible decline in the following two years. Thus, in 2001, the fatality rate per 100,000 population in Turkey was 6.4 and it has been decreased to 5.5 in 2010 which means 17 fatality rate in Turkey decreased by 14.06% during 9 years. On the other hand, the fatality rate in Iran has increased from 30.4 in 2001 to 31.1 in 2010 which shows an increase o f 2.30% during those 9 years. More importantly, comparing the fatality rate of Turkey in 2010 with that o f Iran, it can be seen that the fatality rate in Iran during those years is more than 5 times the rate in Turkey. This clearly shows that the road safety situation in Iran is far below its neighbouring country, Turkey. Table 2-4 Mortality rate per 100,000 population in Turkey and Iran Year 2001 2007 2008 2009 2010 Mortality rate per 100,000 population in Turkey 6.4 7.1 5.9 6.0 5.5 Mortality rate per 100,000 population in Iran 30.4 32.0 32.2 31.2 31.1 (Source: Adopted from Turkish Statistical Institute, Iran Statistical Year Book and Iran Statistical Year Book of Road Maintenance & Transportation Organization, 2012) The above data demonstrate that Iran is one of the countries around the world with a very high record of road traffic crashes and fatalities. The burden o f such fatality can be better understood by looking at the leading causes of death in this country. According to Bhalla et al. (2008), road traffic injuries ranked as the third leading cause of death in Iran accounted for 10.3% o f all deaths in 2002 after myocardial infarction and cerebral vascular diseases (See Table 2-5 below) (Bhalla et al., 2008). However, based on WHO data for 2002, road traffic crashes resulted in 2.1% o f all global deaths (Peden et al., 2004). Therefore, it is obvious that the proportion o f deaths because o f road traffic injuries in Iran is way higher than that of the world average. 18 To shed more light to the high number o f deaths in Iran, Bhalla et al. (2008) compared the number o f deaths in the 2003 due to a devastating earthquake in Bam (South o f Iran) in which 28,745 people lost their lives with the number of people who died in road traffic crashes in 2005 (i.e. 30,721 people) (Bhalla et al., 2008). The tragedies o f Barn's earthquake were broadly covered by the international and internal media. However, the road traffic fatalities which are preventable receives little attention. Table 2-5 Leading causes o f death in Iran Rank Cause o f Death # o f Deaths % Total Deaths 1 Myocardial infarction 68892 23% 2 Cerebral vascular diseases 33922 11.3% 3 Road traffic injuries 30721 10.3% 4 Other cardiac diseases 11459 3.8% 5 Stomach cancer 7799 2.6% 6 Chronic lung and bronchus disease 5297 1.8% 7 Cancer o f trachea, bronchus and lung disorders related to short gestation 45% 1.5% 8 Low birth light 4443 1.5% 9 Pneumonia 4413 1.5% 10 Intentional self-harm 4344 1.5% (Source: Bhalla et al., 2008, P. 19) The causes o f road traffic crashes in Iran have also been analyzed by Tavakoli Kashani et al. (2012). The study indicates that the lack o f using seat belt is the most important factor for the severity o f injuries on the two-way traffic rural roads. The other factors noted by that study to have influenced the severity o f injuries on those roads include improper overtaking and speeding. Overtaking usually lead to more severe injuries as it takes place by driving in the opposite lane. The study identifies “inattention to traffic ahead”, 19 “vehicle defect”, and “movement of pedestrians, livestock and unauthorized vehicles on freeways” as the main reasons for the serious crashes on freeways where wearing seat belts are more strictly enforced (Tavakoli Kashani et al., 2012, P. 40). Overall, it could be concluded that road traffic crashes is one of the most important causes o f death in Iran as well as other countries around the world. However, Iran is one o f the countries which experiences one o f the highest rate o f road traffic mortality. Every year many people in Iran lose their lives because of road crashes and many children become orphans as they lose their parents in road traffic crashes. Therefore, studying the demographic, socioeconomic and circumstantial profiles o f the victims o f road traffic fatalities and investigating various road crash situations that may be related to specific demographic, socioeconomic and circumstantial profiles of the victims is important for the formation of policies in order to decrease road traffic crashes and target those policies to the most vulnerable groups. 20 Chapter 3: Literature review The previous chapters indicated the magnitude o f road traffic injury problem around the world and the devastating situation o f road traffic injuries in Iran. To put the problem in a greater context and have a sense of how it has evolved overtime, this chapter will review the literature on road traffic injury and fatality. It begins with the review o f the work on the effects of motorization and economic development on road traffic injuries, then it reviews the literature on the profiles o f those affected by road traffic injuries. The last part o f the chapter is dedicated to review o f the literature regarding the economic cost o f road traffic fatalities. 3.1 Rapid Motorization, Economic Growth and Road Traffic Injuries In this section, a number o f studies that have explored the relationship between the development, motorization and road traffic injuries will be reviewed. Soderland and Zwi (1995) examined cross-sectional data on road traffic deaths from 83 countries in 1990. Using multiple regression analysis, they studied the relationship between road traffic death and a number o f independent variables including per capita income and the number of registered vehicles for each individual country (Soderland and Zwi, 1995). The study finds that an increase in the gross national product per capita results in increase in the number of road traffic mortality rate. The study also finds a non-linear relationship between road traffic mortality and the number o f registered vehicles. With increase in income per capita up to US$5,000, the number o f vehicles increases more than proportionately, causing greater traffic fatality. However, increases in income 21 beyond US$5,000 lead to less than proportionate increases in the number of vehicles and traffic mortality (Soderland and Zwi, 1995). Ingram and Liu (1999) also confirm that “income is a strong determinant of vehicle ownership at both the country and city level, and that both national and urban motor vehicle ownership increase at about the same rate as income” (Ingram and Liu, 1999, P.2). Kopits and Cropper (2003) uses panel data from 1963-1999 for 88 countries to show that during the first stages o f development, as income increases (approximately up to $8,600 in 1985 international dollars), the vehicle fleets rise and the traffic mortality rates tend to increase as well. But for the GDP (per capita) levels higher than $8,600, motorization grows slowly and governments and individuals invest more in road safety. Therefore, fatalities per vehicle decreases which leads to decline in fatality rate (Kopits and Cropper, 2003). As well, Anbarci et al. (2006) illustrates the relationship between traffic fatality risk and per capita income. Figure 3-1 shows a positive correlation between fatalities and per capita income up to an income level range between $10,000-$ 11,000, and a negative correlation at levels higher than the pre-mentioned range. 22 Figure 3-1 Fatality Risk and Per Capita Income, 1975-2000 is so as s» * 1* is s 0 0 10000 19000 yoflpft wmwi yppflQ Par CapNa tneomi (Pn>k1S9Q (Source: Anbarci et al., 2006, P: 328.) Zhao (2006) explores the relationship between rapid motorization and road traffic deaths in China. He reports that during 1980-2005, China's GDP increased around 40 times averaging around 10% growth each year and China's GDP per capita increased around 30 times. The total population grew by 32% and the urban population increased three times. The total number o f motor vehicles increased about 18 times. Table 3-1 shows how the GDP per capita, the number of motor vehicle ownership, the number o f crashes and the number of deaths due to road traffic crashes are related for selected years during 19802005 in China (Zhao, 2006). 23 Table 3-1 GDP, motorization, and road traffic crashes in China, 1980-2005 year Per capita GDP (US dollar/person) Motor vehicle ownership (10,000 units) Number of crashes Number of deaths Number of deaths per 100,000 population 1980 59 178 116,692 21,818 2.21 1985 108 321 202,394 40,906 3.89 1990 207 551 250,297 49,271 4.31 1995 638 1,040 271,843 71,494 5.20 93,853 7.27 98,738 7.60 2000 995 1,609 616,971 2005 1,779 3,160 450,254 (Source: Adopted from Zhao, 2006, P. 1&4) Nakshabandi (2007) studies the changes in road traffic crashes, injuries and fatalities in Dohuk city, Kurdistan region in Iraq and investigated the reasons for the increase after 2003. After the war, the number of imported cars significantly increased which was the result o f the removing the regulations on importing used cars. The number of registered vehicles in Dohuk in 2004 compared to 2003 increased about 132.2%. The study shows that with the increase in the number o f registered motor vehicles, the number o f road traffic fatalities increased 66.3% in that year (Nakshabandi, 2007). Finally, Naghavi et al. (2009) explores the reasons of why road traffic injury is so high in Iran. Among other things, it finds the rapid increase of vehicle manufacturing as a significant contributing factor to road traffic crashes. The study notices that the most significant increase in the production of car and motorcycle occurred in the construction period after the war. Since 2002, every year more than 1 million cars and 1.5 million motorcycles produced in Iran which could be an important contributing factor in raising the scale o f road traffic injuries problem in Iran (Naghavi et al., 2009). 24 The review o f the above empirical studies indicates that rising income with economic development and the associated increases in the number o f vehicles have a positive effect on people’s mobility and exposure to risk which results in greater road traffic injuries. 3.2 The Profiles of the Victims of Road Traffic Crashes All different kinds of road users are at risk o f being injured or killed in road traffic crashes, but the fatality rates between different groups o f road users are quite different. The risk o f being injured or killed for the vulnerable road users such as pedestrians and two-wheeler users are greater than that o f vehicle occupants. In particular, this is the case in low-income and middle-income countries, because o f the mixture o f different kinds of road users and the lack o f division between them (Peden et al., 2004). The proportion of male and female deaths because o f road traffic crashes is quite different in each country. The victim's age and their socio-economic status are also important factors in determining the exposure to these kinds o f crashes. 3.2.1 Road User Status Odero et al. (1997) reviewed published and unpublished reports on road traffic crashes in developing countries for the period o f 1966 to May 1994. It reviewed 38 studies that had examined casualties by the road user groups. The study concludes that in 75% of the studies pedestrian fatalities were ranked first accounting for 41% to 75% o f casualties. In 62% of the studies, the passenger’s fatalities were placed second (35-51%) among the other road users. They also found that driver deaths were placed third in 55% of the studies (Odero et al., 1997). 25 Wong et al. (2002) studied all road traffic accident deaths that occurred in 1995 in Singapore. They found that the first common victims' group were motorcyclists with about 39.8% of all deaths. The second common victims' group were pedestrians who constituted about 27.9% o f all cases (Wong et al., 2002). Odero et al. (2003) studied road traffic injuries in Kenya. They reviewed published and unpublished reports between 1971 and 3990. Each year around 3,000 people were losing their lives on Kenyan roads. They found that pedestrians and passengers are the most vulnerable road users in Kenya. Around 42% o f crash fatalities were among pedestrians and the combination of pedestrian and passenger deaths represented 80% o f all fatalities in each year. Pedestrian deaths were more likely to occur in urban areas, while passengers were mostly involved in deadly crashes in rural areas (Odero et al., 2003). Montazeri (2004) studied Iranian road traffic mortality data between March 1999 and March 2000. He classifies the status o f road-users as pedestrian, car occupant, car driver, motorcyclist and unknown. The highest mortality were among pedestrians with about 33% of all deaths. The percentage o f deaths among the other road users were: car occupants (29%), car drivers (16%), motorcyclists (12%) and unknown (10%) (Montazeri, 2004). Moharamzad et al. (2008) studied post-mortem records of road traffic accident victims in order to investigate their mortality pattern in the City o f Yazd, Iran. They reviewed hospital records o f 251 victims who were admitted to a tertiary trauma hospital from 2006 to 2007. They found that pedestrians and motorcyclists comprised the most common type o f road user's victims (39.8% and 33.1%, respectively) followed by the car occupants (24.3%) (Moharamzad et al., 2008). 26 3.2.2 Demographic Status of Victims Odero et al. (1997) reviewed 46 studies that examined the effect o f sex on road traffic fatality in developing countries and came to the conclusion that males comprised between 67 to 99.5% (mean 80%) of all crashes. In all studies, the male to female ratio was bigger than 2 and it was bigger than 3 in 83% of the studies. Moreover, when they considered traffic fatalities by type of road users, the number of male deaths was still higher in each sub-group, particularly among drivers, about 87% of them were male (Odero et al., 1997). Wong et al. (2002) in their study on all road traffic accident deaths that occurred in 1995 in Singapore found that motorcyclists were the most common victims' group with the median age o f 24 years and they were mostly men, about 89 out of 90 cases. Pedestrian were the second victims' group with the median age o f 51 years. Also, male deaths comprised about 82.3% of all deaths in their dataset (Wong et al., 2002). Zadeh et al. (2002) studies the traffic accident deaths that happened in Tehran during March 2000 to March 2001. The total number of cases was 2,128 in Tehran. The male to female ratio was 4.1 to 1. Men comprised 80.5% of all cases and women accounted for 19.5% of all deaths. They also indicated that less than half of the victims fall in the age group between 21 to 50 years (48.2% o f all deaths) who are an economically active group in the country (Zadeh et al., 2002). Montazeri (2004) studies Iranian road traffic mortality data between March 1999 and March 2000. Among 15,482 deaths, 79% o f them were men. The highest number of deaths were among young people aged between 21 and 30 (21% o f all deaths) followed by the age group o f 10 to 20 (17% o f all deaths) (Montazeri, 2004). 27 Similarly, Moharamzad et al. (2008) finds that o f 251 cases of road fatality in Yazd (Iran) during 2006-2007, 202 of them were males and 49 were females. Victims aged between 21 to 30 years were the most common age range (30.7% o f all deaths) followed by the age group o f 11 to 20 years (21.9% o f all deaths) (Moharamzad et al., 2008). Finally, several studies have paid attention to the elderly victims. In 2002, 193,478 older people (aged 60 years and above) have been killed due to road traffic crashes worldwide which accounts to 16% o f the total global deaths (Peden et al., 2004). Some countries have a higher proportion o f older people deaths than the global average. For example, Mitchell (2002) studies the kinds o f crashes that older people have been involved in 1998 in the United Kingdom. He shows that people aged 60 and above account for 25.4% o f all traffic deaths and they only comprised 20.5% o f the total population. He also shows that pedestrians are the most vulnerable groups o f people over sixty. Almost 47% o f all pedestrian deaths and 53% o f bus passenger deaths were people aged 60 years and above (Mitchell, 2002). Bhalla et al. (2008) studied road traffic injuries in Iran by age and sex groups in 2005. It finds that the death rate rises with age for both males and females and is the highest for the most elderly age groups. The most vulnerable group among elderly people in their study was pedestrians. Kubitzki and Janitzek (2009) studied the safety and mobility o f older road users. Their finding also confirms that the mortality rates among people aged 65 and above are higher than people aged below 65 in Europe. In Europe 40% of all pedestrian deaths were aged 65 and over and in Germany 49% o f all pedestrian deaths were older people (Kubitzki and Janitzek, 2009). 28 3.2.3 Socioeconomic Status Another characteristic o f the victims that has an effect on their exposure to road traffic injury is their socioeconomic status. Socioeconomic status is evaluated by education, income or occupation (or by grouping occupations into social classes) (Laflamme et al., 2009). Hasselberg et al. (2001) studied the effect o f socioeconomic status o f children and youth on their involvement in road traffic injuries. Their study is based on the Swedish Population and Housing census of 1985. They examined all children aged 0 to 15 years old who were about 1.5 million people in 1985. They classified them into five classes of road users over eight years, and also categorized them into seven socioeconomic groups based on their parent’s socioeconomic status. Their results show that the injury risks o f pedestrians and bicyclists were 20% to 30% higher among the children of lower socioeconomic status than those of higher socioeconomic status (Hasselberg et al., 2001). Whitlock et al. (2003) investigated the association between motor vehicle driver injury and their socioeconomic status in New Zealand during 1988 to 1998. Their sample consisted o f 10,525 adults. They found that driver injury risk was negatively correlated with both occupational status and educational level. Drivers with university or polytechnic degree were likely to experience a driver injury two times less than drivers who had been to secondary school for less than two years. People with the lowest occupational level were four times more likely to be at risk of driver injury than those in the highest occupational level (Whitlock et al., 2003). Aeron-Thomas et al. (2004) studied the effect of wealth on involvement in road traffic accidents in Bangalore, India. They found that the poor are more involved in road 29 traffic deaths both in rural and urban areas. The death rate per 100,000 for the poor in urban areas was 13.1 compared to 7.8 per 100,000 for the non-poor in the urban area. A similar difference has been reported for the rural area as well. The death rate per 100,000 for the poor was 48.1 compared to 26.1 for non-poor in rural area (Aeron-Thomas et al., 2004). Montazeri (2004) in his cross-sectional analysis o f road mortality in Iran showed that most o f the deaths occurred among the people with no education (29% o f all deaths) followed by the people with primary education (23% o f all deaths). People with university education comprised about 4% of all deaths. Therefore, people with lower education are more at risk o f being killed from road traffic crashes (Montazeri, 2004). Laflamme et al. (2009) reviewed 44 articles with 33 o f them from European countries mostly from the northern part o f the Europe. They showed that the lower the socioeconomic status, the higher the risk of road traffic injuries even in high-income countries (Laflamme et al., 2009). 3.3 The Economic Cost of Road Traffic Injuries Studying the economic cost o f road traffic crashes is an important issue for each country in order to properly allocate its resources and make sure that any investment has been used properly. Jacob et al. (2000) has estimated the economic cost o f road traffic crashes in developing and emerging countries to be about US$65 billion per year which is a bit higher than the foreign aid received by these countries. They also show that road traffic crashes cost highly motorized countries around US$453 billion (Jacob et al., 2000). (See Table 3-2 below.) 30 Table 3-2 GNP and road crash costs by region (USSbillion) Region Africa Asia Latin American/Caribbean Mille East Central and Eastern Europe subtotal Highly motorized countries Total Regional GNP 1997 370 2,454 1,890 495 659 5,615 22,665 Estimated annual crash costs % o f GNP Cost 3.7 1% 24.5 1% 18.9 1% 7.4 1.5% 9.9 1.5% 64.5 2% 453.3 517.8 (Source: Jacob et al. 2000, P. 11) Al-Masaeid et al. (1999) examined the economic costs o f road traffic accidents in Jordan during 1996. The traffic accident cost was estimated to be about JD 103 million (US$146.3 million) which was about 2.84% o f its Gross National Product (GNP) o f JD 5,146 million (Al-Masaeid et al., 1999). Elvik (2000) estimated the road traffic accidents cost in 12 OECD countries. On average, he estimated the total costs o f road accidents, including an economic valuation o f lost quality o f life o f about 2.5% o f the gross national product (Elvik, 2000). Zhou et al. (2003) examined the productivity losses and injury costs in China. The study uses the "potentially productive years o f life lost" to calculate the injury costs in 1999. It finds that injuries account for aboutl2.6 million years o f life o f which 25% was due to motor vehicle fatalities (Zhou, 2003, P. 124). Connelly and Supangan (2006) studied the economic costs o f road traffic crashes in Australia, states and territories. They estimated the road traffic crashes' cost in 2003 to be about US$17 billion, which was about 2.3% o f Australia Gross Domestic Product (GDP) (Connelly and Supangan, 2006). There are limited numbers of studies in the literature for estimating cost o f road traffic accidents in Iran. Ayati (2009) uses a comprehensive model to estimate the traffic 31 accident cost o f Iran in 2004 to be about US$11.4 billion (Ayati, 2009). Hejazi et al. (2012) also use the method o f Ayati with minimum variables to estimate the economic cost of traffic accidents of Iran in 2001 to be over US$13.6 billion. They use only three variables for estimating road crash costs which can be taken from insurance companies and police records (Hejazi et al. 2012). As it is notable from above studies, different researchers have come up with different cost estimates for various years. The reason of this discrepancy within studies in the literature is mainly because of using different calculation methods for estimating the following three components: 1- The direct costs, 2- The value o f lost output or productive capacity, and 3- The lost quality o f life (Elvik, 1995). More details about the valuation of road traffic fatalities is provided in Methodology section. This chapter reviewed the literature on road traffic injury and fatality in Iran and around the world to put the problem in a greater context and have a sense o f how it has evolved overtime. To my knowledge, no other study in the literature has reported comprehensive analysis on the demographic, socioeconomic and circumstantial profiles o f the deceased to examine the potential associations between certain circumstantial variables and demographic variables as well as socioeconomic variables with using multinomial logistic regression models to estimate the odds ratios o f different categories o f victim’s status and place of death as affected by gender, age, education and other circumstantial factors as appropriate. Addressing these shortcomings o f the literature is the main purpose o f my study. More specifically, there is no study working on identifying the occupational profile o f the road traffic victims and detecting the most common victims' group among elderly people in the literature o f road traffic injuries in Iran. Also additionally there is no analysis for 32 the time frame o f this study, March 21 to November 21 of 2009 in the literature. The calculation o f economic burden of road traffic accidents for the year 2009 is another novelty of this research. To overcome the mentioned gaps in the literature, I will do some descriptive analysis o f the data by categorizing the individual victims into various demographic, socioeconomic and circumstantial profiles. This analysis helps to recognize the prevalence o f road fatality among different genders and age groups. It also allows to identify the educational and professional profiles o f the road crash victims. The descriptive analysis also sheds light on the circumstantial factors related to victims’ statuses and their modes o f transportation, and helps to identify where the crashes happened most often and what have been the common reasons o f their deaths. The analysis also identifies the most common victims’ group among elderly people. Additionally, multinomial logistic regression models will be used to estimate the odds ratios o f different categories of victim’s status and place o f death as affected by gender, age, education and other circumstantial factors as appropriate. Finally, to emphasize the huge costs o f deadly road crashes to the Iranian society and get attention o f the policy makers, I will calculate the economic cost of road traffic accidents in Iran using the road traffic mortality data o f this research and some other macroeconomics indicators o f the country for the eight months of 2009. 33 Chapter 4: Data &Methodology 4.1 The Dataset To conduct this study, I obtained the road mortality data during the first eight months of the Persian calendar (i.e. March 21 to November 21) of 2009 from the Road Maintenance and Transportation Organization o f Iran. This organization collects the road traffic mortality data from the Forensic Medicine Organization (FMO) o f Iran. The FMO is responsible for issuing death certificate, so each death has to be reported to this organization (Montazeri, 2004). It took me around 5 months to obtain the related data because these data are not available to the public and they are confidential. Therefore, I had to convince the responsible officials about the importance o f the study and why I chose this topic. I talked to many different organizations and different people and finally the Road Maintenance and Transportation Organization gave me the data o f this study. The dataset after cleaning for missing entries consists o f 16,556 individuals who lost their lives on roads during March 21 to December 21, 2009. It provides detailed information on the demographic, socioeconomic and circumstantial profiles o f each victim of road traffic crashes. The dataset includes gender, age, victim's educational level (illiterate, primary, high school or university level), victim's professional profile (children, student, house-keeper, white-collar worker, blue-collar worker, farmer, business man, retired or unemployed). As well, it identifies, victim's status (driver, pedestrian or passenger), victim's mode of transport (pedestrian, bicycle, motorcycle, car, bus or truck), the location o f the crash (in-town or out-of-town areas), the place o f death 34 (at the scene, on the way to the hospital, in the hospital or at home), and the reasons of death (head injury, bleeding or multiple fractures). The definition of road traffic fatality is any person killed immediately or dying within 30 days as a result of a road traffic injury accident (Jones et al., 2008). 4.2 Methodology 4.2.1 Descriptive Analysis As a first step, I do a descriptive analysis o f the data by categorizing the individual victims into various demographic, socioeconomic and circumstantial profiles. This analysis helps to recognize the prevalence o f road fatality among different genders and age groups. It also allows to identify the educational and professional profiles o f the road crash victims. The descriptive analysis also sheds light on the circumstantial factors related to victims’ status and their mode o f transportation, and helps to identify where the crashes happened most often and what have been the common reasons of their deaths. 4.2.2 Bivariate Analysis Following the descriptive analysis, I do a bivariate analysis to examine the potential associations between certain circumstantial variables and demographic variables as well as socioeconomic variables. To test the existence o f associations, Pearson's Chi-square ( X 1) test is used to test the statistical significance o f the associations. The statistical software SPSS 20 is used for this purpose. More specifically, the relationship between victim's status and two demographic variables, namely- gender and age- and one socioeconomic variable (education) has to be 35 examined. Victim’s status consists o f three categories - driver, pedestrian, and passenger. Victims age is dichotomized into two categories o f under 60 years old and 60 years old or above. I used 60 years old as a threshold because this is the retirement age in Iran and people usually change their mode o f transportation at this age. Victim's educational level consists o f four categories- illiterate, primary, high school or university level. The Pearson's Chi-square test is used separately to test the association between the victim's status with each of the independent variables, namely- victim's gender, victim's age and victim's educational level. If the calculated p-value for each association would be less than 0.05, the null hypothesis o f no relationship between status and each variable (namely, victim's gender, victim's age and victim's educational levels) is rejected in favor o f a statistically significant relationship between victim's status and each o f those independent variables. Also, the associations between the place o f death and the demographic variable, age, and 3 circumstantial variables, namely- status, the reason o f death, and the location of the crash- are examined. The place of death consists o f four categories- at the scene o f the crash, in the way to the hospital, in the hospital, and at home. Victim's age were dichotomized into two categories o f under 60 years old and 60 years old or above. Victim's status consists o f three categories- driver, pedestrian, and passenger. The reason o f death consists o f three categories- bleeding, head injury, and multiple fractures. At the end, the location o f the crash consists o f crashes happened in in-town areas and those happened in out-of-town areas. The descriptive and bivariate analyses appear at the beginning o f Chapter 5. 36 4.2.3 M ultivariate Analysis Following the bivariate analysis, I explore further the relationship between certain circumstantial variables and demographic as well as socioeconomic variables using multinomial logistic regression models to estimate the odds ratios for different categories. The statistical software Stata 12 is used for this purpose. The multivariate analysis will consider the variables that are found to be associated by the Pearson’s Chi-square tests in the bivariate analysis. The multinomial logistic regressions are used to estimate the odds ratios for different categories o f victims' status as affected by gender, age and education. Moreover, multinomial logistic regressions are used to estimate the odds ratios for different categories of the place o f death as affected by age, status, the location o f the crash and the reason of death. 4.2.4 Estimating the Economic Cost of Road Fatalities As mentioned in section 3.3, there are three components to the total economic valuation o f traffic accident fatalities: 1- The “direct costs” consisting mainly o f property damages, medical costs, and non-medical costs, 2- “The value o f lost output or productive capacity”, and 3- “The lost quality o f life” which is the economic valuation o f the grief, pain, and suffering related to death (Elvik, 1995, P.238). According to Goodchild et al. (2002), direct costs which consists of accident costs, medical costs and non-medical costs are expenses linked to the occurrence and prevention of the accident. Accident costs only refer to the physical property damages as the result o f the crash and the related data could be obtained from the insurance companies. Medical 37 costs include the related expenses on the medicine and services provided to the patient after accident and also expenses related to the capital investment in staff training and research. Non-medical costs consist of informal care, family help and family counselling (Goodchild et al., 2002). To calculate the value o f lost productive capacity, researchers use human capital approach. Lost productive capacity considers the value of the production that would have been produced if the person would be still alive (Kostyniuk, 2006). Therefore, “production losses are measured as the discounted stream o f future income foregone by the individual” (Goodchild et al., 2002, P. 13). To calculate the lost quality o f life, researchers use willingness-to-pay approach. There are two varieties to this approach: (1) the individual willingness-to-pay approach and (2) the social willingness-to-pay approach (Elvik, 1995). The individual willingnessto-pay approach is estimated either by studying the money that people spend (in dollars or time) to decrease the risk o f dying on traffic accidents (Kostyniuk, 2006) or by means of questionnaires (Elvik, 1995). The social willingness-to-pay approach is inferred "from the valuation implicit in public decision making, for example, concerning speed limits or the regulation o f hazardous products" (Elvik, 1995, P.238). As it is notable from the above explanation, It is challenging to derive an accurate estimation o f the economic cost o f road traffic accidents. However, Ayati (2009) has come up with a reasonable estimation for Iran. Ayati (2009) calculated the economic cost o f road traffic fatalities in a very comprehensive study on Iran. The study used a coefficient similar to the one used by Goodchild et al. (2002) for calculating the lost quality of life. The calculated coefficient by Goodchild et al. is 4.76. Ayati also assumed the lost productive capacity coefficient is equal to that o f loss of quality o f life. Therefore, 38 adding these two coefficients gives the value o f 9.52 which should be multiplied by GDP per capita to come up with the average annual economic cost that the Iranian society will suffer from losing each victim (Ayati, 2009). Since the calculation o f the economic burden o f road traffic fatalities is not the main subject o f this research, I will use the same coefficient as Ayati (2009) to calculate the average annual economic cost of each victim in this study. After calculating the average economic cost of losing one person in a year, I will calculate the total years o f life lost due to road traffic fatalities by subtracting the age o f each victim from the life expectancy in Iran in year 2009 which is the most updated data. Based on WHO data, life expectancy in Iran was 72 for males and 75 for females in 2009 (WHO, 2013). And at the end, I will multiply the total number of years o f life lost to the average economic cost o f losing one person in eight months (the time frame o f the study) to estimate the total economic cost o f road traffic fatalities in Iran in the study period. 39 Chapter 5: Analysis of Data 5.1 D escrip tiv e A nalysis In the first section of this chapter, the descriptive analysis of demographic, socioeconomic, and circumstantial profiles o f the victims o f road traffic crashes will be examined. The dataset consists o f a total of 16,556 individuals who died from the road traffic crashes in Iran during March 21 to November 21,2009. 5.1.1 The Demographic Profiles of the Victims Gender The proportion o f male victims is much higher than female victims, so that the male to female death ratio is 3.79. Table 5-1 shows the total numbers and the percentages o f male and female deaths in the study period. Table 5-1 The distribution of road traffic deaths by gender G ender Frequency Percentage Male 13,099 79.12 Female 3,457 20.88 Total 16,556 100.00 Other studies on road traffic crashes in Iran also indicate greater rates for male deaths compared to female deaths. For example, Zadeh et al (2002) shows that men comprised 40 80.5% o f all cases and women accounted for 19.5% o f all deaths with the male to female ratio o f 4.1 to 1 for the year 2000-2001. Similarly, Montazeri (2004) demonstrates that men accounted for 79% o f all road traffic deaths during the year 1999-2000. Similar gender differences are also observed around the world as well. Based on the WHO data in 2002, males comprised 73% o f all road traffic deaths globally with the mortality rate per 100,000 population o f 27.6 compared to 10.4 o f women (Peden et al, 2004). One o f the reasons for this gender difference could be lesser exposure to road traffic crashes for females. Women in Iran are allowed to drive but they are using cars much less than male drivers because many families cannot afford to have more than one car and normally males members of the family are those who usually use the car. This leads females to less exposure to road traffic crashes than males. Age The age distribution o f the victims is reported in Table 5-2. As the table shows, the percentage o f the deceased people within various age groups are quite different, with adolescents and young adults making up the majority o f victims. For example, age group 15 to 24 years comprises the highest number of deaths (22.33 % o f the total deaths) followed by the age group 25 to 34 years that accounts for about 20.28% of the total deaths. 41 Table 5-2 The distribution of road traffic deaths by age groups Age Frequency Percentage <1 78 0.47 1-4 532 3.21 5-14 1,063 6.42 15-24 3,696 22.33 25-34 3,356 20.27 34-44 2,254 13.61 45-54 1,930 11.66 55-64 1,425 8.61 65-74 1,113 6.72 75-84 880 5.32 85+ 160 0.97 Not available 68 0.41 16,556 100.00 Total Findings by other studies on road traffic crashes in Iran are comparable, although their age groupings do not correspond to the ones of this study as described in Table 5-2. For example, in the study by Zadeh et al. (2002), most o f the victims (48.2% o f all deaths) fall in the age group between 21 to 50 years. Moreover, Montazeri (2004) and Moharramzad et al. (2008) find that victims aged between 21 to 30 years old were the most common victims o f road traffic crashes in Iran followed by the age group o f 11 to 20 years old (Montazeri, 2004 and Moharramzad et al., 2008). Globally, based on the WHO data in 2002, adults aged between 15-44 years comprise around 50% o f all road traffic deaths globally (Peden et al., 2004). 42 Figure 5-1 shows the distribution of victims by both gender and age. The number of the deceased people within each age group shows that the number o f male deaths in each group is consistently higher than female deaths. It also shows that road mortalities increase by age and peaks in the age group o f 15 to 24 years for males and in the age group of 25-34 for females, and decline for older age groups. The difference in the male to female mortality is more pronounced among young adults. These differences are relatively smaller among children (<14 years) and the frail elderly (>85 years) groups. As mentioned before, gender differences could be explained by the differences in the extent o f exposure to road traffic crashes. This pattern is similar to the finding o f the study by Bhalla et al. (2008) for the year 2005. Figure 5-1 The distribution o f road traffic deaths by sex and age group S ex 3 .000 - * § o t.ooo- 43 5.1.2 Socioeconomic Profiles of the Victims Two measures o f socioeconomic status were available in the dataset: the educational level of the victim and his/her professional or occupational status. Table 5-3 demonstrates the distribution of victims’ educational levels. As shown in the table, about half of the victims had no education or only primary education. Almost another 40% had high school education. These proportions are similar to the findings by Montazeri (2004). The latter study shows that 29% of the victims had no education, and 23% had primary education. One o f the reasons for victims with high school education to have the highest percentage o f death is the fact that the proportion of people with high school education is the highest in the general population o f Iran. Table 5-3 The distribution of victims by educational levels Frequency Percentage Illiterate 4,268 25.78 Primary 3,950 23.86 High School 6,519 39.38 University 1,282 7.74 Not available 537 3.24 Total 16,556 100.00 Educational Level Table 5-4 demonstrates the occupational profile o f the victims for 9 categories of children, student, housekeeper, white-collar worker, blue-collar worker, farmer, businessman, retired, and unemployed. Over a quarter of the victims were businessmen followed by blue-collar workers (18.78%), housekeepers (13.56%), and student 44 (11.01%). To my knowledge, no other study on Iran has reported on the occupational profile o f the deceased. The higher percentage for businessmen may be explained by the greater frequency of commuting and the popularity of using motorcycle among some o f them that are working in the market. The reason o f commuting with motorcycle is the existence of the heavy traffic jam in the big cities and prevention of cars to get into some regions o f the big cities during the day in order to reduce traffic jam. The only way to get to those areas is public transportations or motorcycles. Comparing the victims with “blue-collar" occupations and those with “white-collar” occupations clearly shows that the frequency of death is much higher (3.34 times) than that for victims with white-collar jobs. Such finding is comparable to the result by Whitlock et al. (2003), which finds that people with the lowest occupational level were four times more likely to being at risk of driver injury than those in the highest occupational level in New Zealand. 45 Table 5-4 The distribution of victims by occupation Frequency Percentage Children 893 5.39 Student 1,823 11.01 Housekeeper 2,235 13.50 White-collar worker 881 5.32 Blue-collar worker 3,110 18.78 Farmer 1,170 7.07 Businessman 4,191 25.31 Retired 693 4.19 Unemployed 551 3.33 Not available 1,009 6.09 Total 16,556 100.00 Occupational Status 5.1.3 The Circumstantial Profiles of the Crashes This section describes the status o f the victims as drivers, passengers or pedestrians. It also describes the victim's mode of transport. The location of the crash, the place o f death and the reason o f death are also described in this section. I use the term circumstantial to refer to these various aspects o f the road traffic crashes. Victims Status The distribution o f the victims status is shown in Table 5-5. As shown in the table, over 40% o f the deceased were drivers, and more than a third were passengers. The driver category consists o f both car driver and motorcycle driver. Such results cannot be compared with other studies on Iran because they use different categories for status from what I have used here as available in my dataset. Notwithstanding different 46 categorizations of status, studies on other developing countries find pedestrians as the dominant group among the victims o f road traffic crashes. For example, Odero et al. (2003) finds that 42% o f all road traffic deaths in Kenya were pedestrians and the combination o f pedestrian and passenger deaths represented 80% o f all deaths in each year. Table 5-5 The distribution of victims by status Status Frequency Percentage Driver 6,817 41.18 Pedestrian 3,855 23.28 Passenger 5,717 34.53 Not available 167 1.01 Total 16,556 100.00 Mode o f Transportation Table 5-6 shows the distribution o f the victim's mode of transport. It includes the categories for pedestrian, bicycle, motorcycle, car, bus, and truck. The occupants o f the cars are the most common group with 34.39% o f all deaths followed by the riders o f motorcycle (27.4%) and pedestrians (23.28%). These findings are comparable to those by Bhalla et al (2008) for car occupants, but not for other groups. The latter study reports 36% for car occupants, but 15% and 29% for riders o f motorcycles and pedestrians, respectively (Bhalla et al., 2008). 47 Table 5-6 The distribution of victims by mode o f transport Victim's Mode of Transport Frequency Percentage Pedestrian 3,855 23.28 Bicycle 136 0.82 Motorcycle 4,505 27.21 Car 5,694 34.39 Bus 452 2.73 Truck 1,756 10.61 Not available 159 0.96 Total 16,556 100.00 Location o f Crash The fatal crashes recorded in the dataset consist of both crashes that happened in "intown" areas and those which happened in "out-of-town" areas. As Table 5-7 shows, more than two thirds of crashes are those that happened in out-of-town areas. Such result is to be expected as traffic speed is normally higher on out-of-town roads and the quality and safety o f those roads are typically poorer than roads in-town. Moreover, the monitoring o f traffic is usually less frequent and effective than that o f the roads in in-town areas. Relatively similar results are observed in other developing countries. For example, In Kenya, 60% o f all fatal crashes happened in out-of-town areas and 40% of them took place in in-town areas (Odero et al., 2003). 48 Table 5-7 The distribution of the location o f crash Location F req u e n c y Percentage In-town 5,034 30.41 Out-of-town 11,408 68.91 Not available 114 0.69 Total 16,556 100.00 When the distribution o f the location o f crash is examined by the victim's mode of transport, it is observed that about 46% o f pedestrian deaths takes place in in-town roads compared with roughly 13% happening in out-of-town roads. On the other hand, over 44% o f car occupants victims died in out-of-town roads compared with 12.5% o f victims who died in in-town roads. Odero et al. (2003) also finds that pedestrian deaths were more likely to occur in urban areas, while passengers were mostly involved in deadly crashes in rural areas. Place o f Death In the majority o f cases the victims die at the site o f the crash. However, those who get injured might die while being transported to the medical centers, or in the hospital during operations or shortly after. If the injured people die within 30 days after a crash, they are considered as a victim of road traffic crashes (Jones et al., 2008). Table 5-8 shows where the deaths occurred. The places o f death consist of four categories: whether the victims die at the scene o f the crash, in the way to the hospital, in the hospital or at home. As can be seen from Table 5-8, over half o f the victims died at the scene of the crash and almost 40% o f deaths happened in the hospital. Such statistics 49 indicate the acute severity o f the inflicted injuries. Montazeri (2004) also finds that 57% o f deaths occurred in pre-hospital (at the scene o f the crash or in the way to the hospital). Table 5-8 The distribution of victims by the place of death Frequency Percentage At Scene 8,355 50.47 On the way to the hospital 1,427 8.62 Hospital 6,577 39.73 Home 120 0.72 Not available 77 0.47 16,556 100.00 Place of Death Total Reason o f Death The forensic or medical reasons for the death o f the victims in the dataset include head injury, bleeding, and multiple fractures. As Table 5-9 shows, head injury was the main reason o f death accounting for over 61% o f all deaths. The second main reason of death was multiple fractures accounting for over a quarter o f deaths. Once again, such findings reflect the severity o f the crashes in Iran. Montazeri (2004) and Moharamzad et al. (2008) also find head injury as the most common cause o f death by reporting proportions such as 66% and 58.1%, respectively. 50 Table 5-9 The distribution of the reasons of death Frequency Percentage Head Injury 10,136 61.22 Bleeding 1,380 8.34 Multiple Fractures 4,167 25.17 Not available 873 5.27 Total 16,556 100.00 Reason of Death 5.2 Bivariate Analysis This section examines possible associations between certain circumstantial variables and demographic variables as well as socioeconomic variables. Such examination would be helpful in finding evidence of relationships among the variables and help us better understand the potential risk factors related to road mortalities. To test the associations, Pearson's Chi-square test has been employed. 5.2.1 The Association of Victim's Status with Demographic and Socioeconomic Factors The relationship between victim's status and two demographic variables, namely- gender and age- and one socioeconomic variable (education) were examined. Since victim's status consists of three categories - driver, pedestrian, and passenger, this creates two 3 by 2 cross-tabulations: one for status versus gender, another for status versus age. The Pearson's Chi-square statistic for testing the existence of a relationship between status and gender is found to be statistically significant ( j 2 = 2972.40, P < 0.001). Since the calculated p-value is much smaller than 0.05, the null hypothesis o f no relationship 51 between status and gender is strongly rejected in favor o f a statistically significant relationship between status and gender. Table 5-10 shows the cross tabulation between status and gender. As can be seen, 67.2% o f female deceased were passenger and among male deceased over half o f them were driver (51.9%). Table 5-10 Cross tabulation between status and gender Gender Total Female Male 1.3% ^ \ 9 8 . 7 % '" - ''\ J 0 0 . 0 % Driver Status Pedestrian 2.6% 51.9% 41.6% ' V' ' \ 2 6 . 8 % ^ \7 3 .2 % " ^ \K ) 0 .0 % 30.2% 21.8% 23.5% ^ \4 0 .3 % Passenger 67.2% 59.7% "S''v\ 2 0 . 9 % N J O O .0 % 34.9% 26.3% 79.1% ^ \ 1 0 0 . 0 % Total 1 0 0 .0 % ^ \ 100.0% 1 0 0 .0 % ^ \ Note: The percentages in the upper portion of the cells show the conditional distribution of victims across gender given each status. Whereas the percentages in the lower portion o f the cells show the conditional distribution of victims across status given each gender category. To examine the relationship between the status and age, I dichotomized the victim's age into two categories of <60 years old and >=60 years old. Age 60 has been chosen because it is the retirement age in Iran and usually the change in the choice o f transportation is observed for most o f the retired people. This resulted in a 3 by 2 cross-tabulation with 2 degrees o f freedom. Here too, the Pearson's Chi-square test value is very high i X 1 = 1717.95, P < 0.001). Therefore, victim's status is found to be strongly related to age as categorized by <60 and >=60 years of age. Table 5-11 shows the cross tabulation between status and age. As the table shows, among deceased aged under 60 years old, over 45% of them were driver and among deceased aged 60 years old or above over 50% 52 o f them were pedestrian. Overall, it could be concluded that people aged under 60 years are more likely to die as a driver compared to those aged 60 years or above that are mostly tend to be a pedestrian victims o f the crashes. Table 5-11 Cross tabulation between status and age Age <60 Total >=60 91.1% 8.9% 100 . 0 % Driver 45.8% Status Pedestrian 21.7% 61.1% 17.3% 38.9% 53.7% 88 . 0 % Passenger 41.7% 36.8% 100 .0 % 23.5% 100 .0 % 12 .0 % 24.6% 83.0% 34.7% 17.0% 100 .0 % Total 100 . 0 % 100 . 0 % 100 .0 % Note: The percentages in the upper portion of the cells show the conditional distribution o f victims across age given each status. Whereas the percentages in the lower portion of the cells show the conditional distribution of victims across status given each age category. Examining the relationship between status and education entails a 3*4 cross-tabulation as education consists o f 4 categories o f no education, primary, high school and university education. The result o f the Pearson's Chi-square test supports the existence o f relationship between status and education (%2= 1843.41, P < 0.001). Table 5-12 shows the cross tabulation between status and education. As shown in the table, over 50% o f drivers had high school level of education. 49% o f pedestrians had no education and around 39% o f passengers had high school level o f education. This result could be the case since most of the cab driver and people who work with their motorcycle have high school and primary level of education and usually people with university level o f education work in the office and they tend to travel less than blue-collar workers. As 53 demonstrated in the table, white-collar workers comprise 5% of the victims and bluecollar workers encompass 19% of the victims. Table 5-12 Cross tabulation between status and education Education Illiterate Primary High University Total school Driver ^ 1 2 .8 % 'S XV25.7% s^2-6% \ R . 8 % " \ 400 .0% 2 0 . 1 % \ 4 3 .4 % \ v 5 4 . 0 ° / o \ 45.5% s s 41.7% X^ Status Pedestrian Passenger Total S\ 4 8 . 8 % S x s 25.3% N\2 1 .7 % Xs4.1% \ ( 0 0 . 0 % 43.1%S \ 24.1%SXV 1 2 . 5 % \ 1 2 .1 % X . 23.5% x ^ S\ 2 8 . 1 % SN s 23.1% S \3 9 .0 % N\ 9 . 8 % 3 6 .8 % \ 3 2 .5 % \ 33.5% X100.0% 4 2 . 4 % \ 34.8%\ Xy26.6% S \2 4 .7 % SN s 40.6% \ 8 . 1 % X s100.0% 1 0 0 .0 % \ 1 0 0 .0 % \ 1 0 0 .0 % \ 1 0 0 .0 % \ lO O .O V 'x Note: The percentages in the upper portion of the cells show the conditional distribution o f victims across education given each status. Whereas the percentages in the lower portion of the cells show the conditional distribution of victims across status given each education category. 5.2.2 The Association of the Place of Death with Demographic and Other Circumstantial Factors The relationship between the place o f death and one demographic variable (age) and 3 other circumstantial variables, namely- status, the reason o f death, and the location o f the crash- were examined. The place of death consists of four categories- at the scene of the crash, in the way to the hospital, in the hospital, and at home. Victim's status consists of three categories- driver, pedestrian, and passenger. The reason of death consists o f three categories- bleeding, head injury, and multiple fractures. And the location o f the crash consists of crashes happened in in-town areas and those happened in out-of-town areas. 54 To examine the relationship between the place o f death and age, I dichotomized the victim's age into two categories o f under 60 years old and 60 years old or above. This resulted in a 4 by 2 cross-tabulation with 3 degrees of freedom. Here too, the Pearson's Chi-square test value is high (%2 = 384.89, P < 0.001). Therefore, the place of death is found to be strongly related to age categorized as <60 and >=60 years old. Table 5-13 and Table 5-14 shows the cross tabulation between the place o f death and age and the reason o f death and age, respectively. As the table shows, over 50% of deceased aged under 60 years old died at the scene of the crash which could be explained by their reason o f death which is mostly head injury (over 65%). Table 5-13 Cross tabulation between the place o f death and age Age60 Total >=60 <60 At Scene ' S\ 12.3% 'V' \ i o o . o % v 87.7% On the way to the Place of Death 50.6% 36.8% 53.5% V'v''v x 85.6% ' V sv v 14.4% Vx\ hospital 8.9% 7.4% 8.7% Hospital V\ 7 7 . 2 % V\ 2 2 . 8 % ' V\ U ) 0 . 0 % 3 7 . 2 % ^ \ ^ 53.7% Home '''\ 4 9 . 6 % 0.4% Total >K)0.0% ' 'v . 40.0% 50.4% ^ \ 1 0 0 . 0 % 2.2% N .8 3 .0 % ^ \ 1 7 . 0 % 0.7% ^ \1 0 0 .0 % 1 0 0 .0 % ^ \ 1 0 0 .0 % ^ \ 1 0 0 .0 % ^ \ Note: The percentages in the upper portion of the cells show the conditional distribution of victims across age given each category o f the place o f death. Whereas the percentages in the lower portion o f the cells show the conditional distribution of victims across the place of death given each age category. 55 Table 5-14 Cross tabulation between the reason of death and age Age60 Total <60 Head Injury Reason of Bleeding Death >=60 \8 5 ,0 % S N s 15.0% \ u ) 0 . 0 % 6 5 .9 % 's\ 5 8 . 7 % v\ N \8 2 .7 % N \1 7 .3 % ^ 8 .6 % ^ \ 9 . 1 ° / i \ Multiple Fractures 64.7% \ \ 400 .0% 8 .7 % ^ \ s 80.0% ' \ v 20.0% " \ i o o . o % 2 5 . 5 ° / o \ 3 2 . 2 % \ 2 6 .6 ° /is \ Total ' V\ 8 3 . 5 % 6.5% x \ j o o . o % 1 0 0 .0 % \ 1 0 0 .0 % \ 1 0 0 .0 % \ Note: The percentages in the upper portion of the cells show the conditional distribution o f victims across age given each category of the reason of death. Whereas the percentages in the lower portion of the cells show the conditional distribution of victims across the reason of death given each age category. The Pearson's Chi-square statistic for examining the existence o f a relationship between the place o f death and status is found to be %2 = 569.25 and P < 0.001, indicating a statistically significant relationship between the place o f death and status. Table 5-15 shows the cross tabulation between the place o f death and status. As shown in the table, over 50% of drivers died at the scene o f the crash. And about 53% o f pedestrians died in the hospital and over 58% o f passengers died at the scene o f the crash. 56 Table 5-15 Cross tabulation between the place of death and status Status Total Driver At Scene ^ \4 3 .5 % Pedestrian \^ 6 .6 % N ' \ 3 9 . 8 % 5 3 . 0 ° A \ 35.9% On the way to the Place of hospital Death Hospital Passenger \ ^ 8 .0 % ^ \ 5 0 .7 % ^ \ ,,40.9% \ ^ 4 . 2 % ^ s M .9 % N\lo o .o % 8.7% 8 .7 % ^ ^ 8.5% 8.9% S\3 9 .8 % S N ^ ].4 % ^ \ 2 8 . 8 % 3 8 .1 % ^ 5 3 .2 % \ 3 2 .9 % ^ ^ 3 9 .9 ° /o \ v^!8.5% S\ 6 0 . 5 % ^ \ 2 1 . 0 % Home Total \io o .o % 0.3% 1 . 9 % ^ \ 0.4% S>\ 4 1 . 7 % rvN23.5% '\m 0 % s \io o .o % 0.7% s s 34.9% \ i o o . o % 1 0 0 . 0 % \ 1 0 0 .0 % s >40 0 .0% \ 1 0 0 .0 % \ Note: The percentages in the upper portion of the cells show the conditional distribution o f victims across status given each category o f the place o f death. Whereas the percentages in the lower portion of the cells show die conditional distribution of victims across place of death given each status. The Pearson's Chi-square statistic for examining the existence o f a relationship between the place of death and the reason o f death is found to be / 2 = 151.308 and P < 0.001. Therefore, there is a statistically significant relationship between the place o f death and the reason o f death. Table 5-16 shows the cross tabulation between the place o f death and the reason of death. As the table shows, about 53% o f people with head injury passed away at the scene of the crash. 57 Table 5-16 Cross tabulation between the place of death and the reason of death Reason of Death Head Injury Bleeding Total Multiple Fractures At Scene On the way to the Place of hospital \ s 27.7% X ' n 100.0% ^ \6 6 .1 % X \ 6 . 2 % 5 2 . 8 % \ 3 6 . 3 % \ 5 3 . 7 ° / o \ 51.6% \ \6 1 .4 % \ s J 2 . 2 % N\ v 26.3% \ m o % 8 .3 ° /< /\ 12.2% 8 .7 % ^ \ 8 .8 % ^ \ S N v63.5% N S v 11.3% N S s 25.1% s \ j o o . o % Death Hospital 3 8 .5 % \ 5 0 .5 % \ 3 7 .1 % \ 3 9 .2 % ^ \5 1 .4 % N \ 1 9 . 4 % V\ 0.4% 1.0% Home ^ ^ 6 4 .6 % Total s 29.2% V\ 0.5% 8.8% S \ 2 6 . 6 % 1 0 0 . 0 % \ 100.0%N\ 1 0 0 .0 % \ 400 .0% 0.5% s\ \ h >o.o% 1 0 0 .0 % \ Note: The percentages in the upper portion of the cells show the conditional distribution o f victims across the reason of death given each category of the place of death. Whereas the percentages in the lower portion o f the cells show the conditional distribution o f victims across place o f death given each reason o f death. Finally, the Pearson's Chi-square statistic for examining the existence o f a relationship between the place o f death and the location o f the crash is found to be x 2 ~ 1002.90 and P < 0.001, which suggests a statistically significant relationship between the place of death and the location o f the crash. Table 5-17 shows the cross tabulation between the place of death and location of the crash. As shown in this table, about 58% o f deceased were involved in the crashes happened out-of-town died at the scene o f the crash. This could be explained by higher severity of the out-of-town crashes as speed limit is normally higher there and the quality and safety o f roads are typically poorer than intown roads. Moreover, the monitoring o f traffic is usually less frequent and effective than that o f the in-town roads. 58 Table 5-17 Cross tabulation between the place of death and location of the crash Location Total At Scene In-Town Out-of-Town SS' \ 2 0 . 5 % ^ \7 9 .5 % 3 4 . 1 % ^ \ 58.3% On the way to the V' \ s 26.1% Place of hospital 7.4% Death hospital \ Home Total 's\ i o o . o % 50.9% 73.9% ' >'\ l 0 0 . 0 % 8.7% 9.2% 55.9% 'v\ i o o . o % 44.1% 57.2% 32.0% 3 9 .7 % ^ \ "^ \5 5 .6 % "V'\ 4 4 . 4 % 'v\ i o o . o % 1.3% 0.5% 0.7% ^ \3 0 .6 % 69.4% ^ \ 1 0 0 . 0 % 1 0 0 .0 % ^ \ 1 0 0 .0 % ^ \, Note: The percentages in the upper portion of the cells show the conditional distribution o f victims across the location of the crash given each category of the place of death. Whereas the percentages in the lower portion of the cells show the conditional distribution of victims across place of death given the location of the crash. 5.3 Multivariate Analysis 5.3.1 Multinomial Logistic Regression Results for Victim's Status, Gender, Age and Education To further explore the relationship between victim's status with gender, age and education, I run a multinomial logistic regression model to estimate the odds ratios for different categories o f victims' status as affected by gender, age and education. Odds Ratios for Gender Table 5-18 shows the logistic regression results for the odds ratios o f various combinations of victim's status for males compared to females, adjusted for age and education. 59 Table 5-18 The odds ratios of different status for males compared to females Odds comparing males to females Coefficient Odds P-Value Driver-Passenger 3.92286 50.5447 <0.001 Driver-Pedestrian 3.24147 25.5712 <0.001 Pedes trian-Passenger 0.68139 1.9766 <0.001 The following conclusions can be made from the above table. The odds ratio (the relative risk ratio) o f being dead as a driver versus passenger is 51 times greater for males than females, after adjusting for age and education in the Multinomial Logistic regression analysis model; the odds ratio of being dead as a driver versus pedestrian is 25.6 times greater for males than females, and the odds ratio o f being dead as a pedestrian versus passenger is 1.98 times greater for males than females. As a whole, the odds ratio for males compared to females is the highest as a driver vs. passenger followed by driver vs. pedestrian and pedestrian vs. passenger. Such results imply that it is very likely for males to be the victims in crashes in which they are drivers. Odds Ratios for Age Table 5-19 reports the logistic regression results for the odds ratios o f various combinations o f victim's status as affected by age- categorized as <60 and >=60 years old, adjusted for gender and education. 60 Table 5-19 The odds ratios of comparing the change in the status of the victims for those aged 60 years old or above compared to those under 60 years old Odds comparing people aged Coefficient Odds P-Value Pedes trian-Driver 1.25266 3.4996 <0.001 Pedes trian-Passenger 1.22741 3.4124 <0.001 Passenger-Driver 0.02525 1.0256 0.713 60 or above to those under 60 From the above table, the following conclusions can be made. The odds ratio o f being dead as a pedestrian versus driver is 3.5 times greater for people aged 60 years old or above than those under 60 years old; the odds ratio o f being dead as a pedestrian versus passenger is 3.41 times greater for people aged 60 years old or above than those under 60 years old; and the odds ratio of being dead as a passenger versus driver is not statistically significant, after adjusting for gender and education in the Multinomial Logistic regression analysis model. Overall, the odds ratio for people aged 60 years or above compared to those under 60 years old is the highest as a pedestrian vs. driver followed by pedestrian vs. passenger. These findings indicate older people are more likely to be the pedestrian victims of crashes. Odds Ratios for Education Table 5-20, 5-21 and 5-22 report the logistic regression results for the odds ratios of various combinations of victim's status as affected by education- categorized as illiterate, primary, high school, and university, adjusted for gender and age. Since education consists o f 4 different categories, the odds ratios o f various combinations o f victim's 61 status are reported for primary, high school and university education, using “no education” category as the reference group. Therefore, there are 3 different tables for results. Table 5-20 The odds ratios o f comparing the change in the status o f the victims for those with primary level of education compared to those with no level of education Odds comparing primary level to no education Coefficient Odds P-Value Driver-Pedestrian 0.95730 2.6047 <0.001 Driver-Passenger 0.69915 2.0121 <0.001 Passenger-Pedestrian 0.25815 1.2945 <0.001 The following conclusions can be made from the above table. The odds ratio o f being dead as a driver versus pedestrian is 2.6 times greater for people with primary level of education than those with no education; the odds ratio o f being dead as a driver versus passenger is 2 times greater for people with primary level o f education than those with no education; and the odds ratio of being dead as a passenger versus pedestrian is 1.3 times greater for people with primary level o f education than those with no education, after adjusting for gender and age in the Multinomial Logistic regression analysis model. From the three results mentioned above, it can be seen that the odds of being dead as a driver vs. pedestrian or a driver vs. passenger or a passenger vs. pedestrian are consistently higher for those with primary level o f education compared to those with no education when gender and age are controlled. 62 Table 5-21 reports the logistic regression results for the odds ratios of various combinations of the victim's status as affected by education changing from illiterate to high school. Table 5-21 The odds ratios of comparing the change in the status of the victims for those with high school level of education compared to those with no level of education Odds comparing high school to no education Coefficient Odds P-Value Driver-Pedestrian 1.56469 4.7812 <0.001 Driver-Passenger 0.78018 2.1819 <0.001 Passenger-Pedestrian 0.78451 2.1913 <0.001 The following conclusions can be made from the above table. After adjusting for gender and age in the Multinomial Logistic regression analysis model, the odds ratio of being dead as a driver versus pedestrian is 4.8 times greater for people with high school level of education than those with no education; the odds ratio o f being dead as a driver versus passenger is 2.2 times greater for people with high school level o f education than those with no education; and the odds ratio of being dead as a passenger versus pedestrian is 2.2 times greater for people with high school level o f education than those with no education. From the three results mentioned above, it can be seen that the odds o f being dead as a driver vs. pedestrian or a driver vs. passenger or a passenger vs. pedestrian are consistently higher for those with high school level of education compared to those with no education when gender and age are controlled. 63 Table 5-22 reports the logistic regression results for the odds ratio o f various combinations of the victim's status as affected by their educational level changing from illiterate to university level. Table 5-22 The odds ratios of comparing the change in the status of the victims for those with university level of education compared to those with no level of education Odds comparing university level to no education Coefficient Odds P-Value Dri ver-Pedestrian 1.58144 4.8620 <0.001 Passenger-Pedestrian 0.99320 2.6999 <0.001 Driver-Passenger 0.58824 1.8008 <0.001 From the above table, the following conclusions can be made. After adjusting for gender and age in the Multinomial Logistic regression analysis model, the odds ratio o f being dead as a driver versus pedestrian is 4.9 times greater for people with university level of education than those with no level of education; the odds ratio of being dead as a passenger versus pedestrian is 2.7 times greater for people with university level of education than those with no level of education; and the odds ratio o f being dead as a driver versus passenger is 1.8 times greater for people with university level o f education than those with no level of education. As a whole, from the results shown in Tables 5-20, 5-21 and 5-22, it can be concluded that with increase in the level o f education the odds ratios of being death as a driver vs. pedestrian and passenger vs. pedestrian increase more than the odds ratio o f being death as a driver vs. passenger. These results could be interpreted by the fact that with the increase in the level o f education people are more likely to travel as a driver or passenger 64 than pedestrian. Interpretation of the results of Tables 5-20, 5-21 and 5-22 is not straightforward. One would expect lower odds ratios for victims with greater education. However, greater education is associated with greater mobility, which increases the risk of road traffic fatality. The dataset I had access to does not allow to separate such effects. 5.3.2 Multinomial Logistic Regression Results for the Place of Death Versus Age and other Circumstantial Factors To further explore the relationship between the place o f death with age and other circumstantial factors such as status, the reason o f death, and the location of the crash, I run a multinomial logistic regression model to estimate the odds ratios for different categories o f the place o f death as affected by age categorized as <60 and >=60 years old, status, the reason o f death, and the location of the crash. Table 5-23 shows the logistic regression results for the odds ratios o f various combinations of the place of death as affected by age. Table 5-23 The odds ratios of comparing the change in the place of death for those aged 60 years old or above compared to those under 60 Odds comparing people aged 60 or above to those under 60 Coefficient Odds P-Value Home - At scene 1.41827 4.1300 <0.001 Home- On the way to the hospital 1.40792 4.0875 <0.001 Home - Hospital 0.97387 2.6482 <0.001 Hospital-At scene 0.44440 1.5595 <0.001 Hospital- On the way to the hospital 0.43405 1.5435 <0.001 On the way to the hospital- At scene 0.01035 1.0104 0.907 65 From Table 5-23, the following conclusions can be made. After adjusting for status, the reason of death, and the location of the crash in the Multinomial Logistic regression analysis model, the odds ratio of dying at home versus dying at the scene of the crash is 4 times greater for those aged 60 years old or above than those under 60 years old; the odds ratio o f dying at home versus dying on the way to the hospital is 4.1 times greater for those aged 60 years old or above than those under 60 years old; the odds ratio of dying at home versus dying in the hospital is 2.6 times greater for those aged 60 years old or above than those under 60 years old; the odds ratio of dying in the hospital versus dying at the scene of the crash is 1.6 times greater for those aged 60 years old or above than those under 60 years old; the odds ratio o f dying in the hospital versus dying on the way to the hospital is 1.5 times greater for those aged 60 years old or above than those under 60 years old; and the odds ratio o f dying on the way to the hospital versus dying at the scene of the crash for those aged 60 years old or above than those under 60 years old is not statistically significant. From the above mentioned results, it can be seen that the odds ratios for people aged 60 years old or above compared to those under 60 years old is the highest for dying at home vs. dying at the scene o f the crash followed by dying at home vs. on the way to the hospital and dying at the home vs. dying in the hospital after adjusting other variables. These findings indicate that people aged 60 years old or above are more likely to die at home compared to those under 60 years old that are more likely to die at the scene of the crash followed by on the way to the hospital. As shown in Table 5-14, people under 60 years old relative to those 60 years old or above are mostly died as a consequence of head injury. Also dying from multiple fractures is relatively more common among people 60 years old or above comparing to 66 below 60. These two facts are in accordance with the result of Table 5-19 which shows that people under 60 are more likely to die as a driver. All in all, we can conclude that people under 60 years old are mostly involved in the crashes with higher severity in which they are driver with the casualty to their head which lead them to die at the scene o f the crash followed by on the way to the hospital. On the other hand, older people are mostly died as a pedestrian and involved in the crashes with lower severity which lead them to die after the crash and in their home followed by in the hospital. Odds Ratios for Status The logistic regression results for the odds ratios o f various combinations o f the place of death as affected by the victim's status are reported below. Table 5-24 reports the results for driver vs. pedestrian and Table 5-25 reports the results for passenger vs. Pedestrian. The pedestrian is considered as the reference category. Table 5-24 The odds ratios of comparing the change in the place of the death for driver compared to pedestrian Odds comparing the change in place of death of driver to pedestrian Coefficient Odds P-Value At scene- Home 1.22676 3.4102 <0.001 Hospital-Home 0.91808 2.5045 <0.001 On the way to the hospital-Home 0.85355 2.3480 <0.001 At scene- On the way to the hospital 0.37322 1.4524 <0.001 At scene - Hospital 0.30869 1.3616 <0.001 Hospital- On the way to the hospital 0.06453 1.0667 0.436 67 From Table 5-24, the following conclusions can be made. After adjusting for age, the reason o f death, and the location o f the crash in the Multinomial Logistic regression analysis model, the odds ratio of dying at the scene o f the crash versus at home is 3.4 times greater for driver than pedestrian; the odds ratio o f dying in the hospital versus at home is 2.5 times greater for driver than pedestrian; the odds ratio o f dying on the way to the hospital versus at home is 2.3 times greater for driver than pedestrian; the odds ratio o f dying at the scene o f the crash versus on the way to the hospital is 1.4 times greater for driver than pedestrian; the odds ratio of dying at the scene o f the crash versus in the hospital is 1.36 times greater for driver than pedestrian; and the odds ratio of dying in the hospital versus in the way to the hospital is not statistically significant. From the above results, it can be concluded that the odds ratio o f death as a driver compared to a pedestrian is the highest for dying at the scene o f the crash vs. at home followed by dying in the hospital vs. at home, on the way to the hospital vs. at home, at the scene o f the crash vs. on the way to the hospital and at the scene of the crash vs. in the hospital. The implication is that drivers are seriously impacted in a crash leading to immediate death at the scene, or shortly after on the way to hospital or in the hospital. 68 Table 5-25 The odds ratios o f comparing the change in the place o f the death for passenger compared to pedestrian Odds comparing the change in place of death o f passenger to pedestrian Coefficient Odds P-Value At scene- Home 1.10762 3.0271 <0.001 On the way to the hospital-Home 0.69692 2.0076 <0.001 At scene - Hospital 0.43980 1.5524 <0.001 At scene- On the way to the hospital 0.41070 1.5079 <0.001 Hospital-Home 0.66782 1.9500 0.051 On the way to the hospital-Hospital 0.02911 1.0295 0.739 From Table 5-25, the following conclusions can be made. After adjusting for age, the reason o f death, and the location o f the crash in the Multinomial Logistic regression analysis model, the odds ratio o f dying at the scene o f the crash versus at home is 3 times greater for passenger than pedestrian; the odds ratio o f dying on the way to the hospital versus at home is 2 times greater for passenger than pedestrian; the odds ratio o f dying at the scene of the crash versus in the hospital is 1.6 times greater for passenger than pedestrian; the odds ratio of dying at the scene of the crash versus on the way to the hospital is 1.5 times greater for passenger than pedestrian; and the odds ratio o f dying in the hospital versus at home and the odds ratio o f dying on the way to the hospital versus in the hospital for passenger than pedestrian are not statistically significant. From the above mentioned results, it can be seen that the odds ratio o f death as a passenger compared to as a pedestrian is the highest for dying at the scene o f the crash vs. at home followed by dying on the way to the hospital vs. at home, at the scene o f the crash vs. in the hospital and at the scene o f the crash vs. on the way to the hospital. The findings indicate that passengers as occupants of cars may be more seriously injured than 69 pedestrians, leading to greater odds o f dying at the scene of the crash or on the way to the hospital or while in hospital. Odds Ratios fo r the Reason o f Death Table 5-26 and table 5-27 shows the odds ratios o f various combination o f the place of death as affected by the reason o f death. The former shows the results for the changes in the reason o f death from head injury to bleeding, and the latter shows the results for the change in the reason o f death changes from bleeding to multiple fractures. Bleeding is used as the reference category. Table 5-26 The odds ratios of comparing the change in the place of death when the reason o f death changes from bleeding to head injury Odds comparing head injury to bleeding as a reason o f death Coefficient Odds P-Value At scene- Home 1.44563 4.2445 <0.001 At scene- On the way to the hospital 0.79084 2.2052 <0.001 Hospital-Home 0.76921 2.1581 <0.001 At scene- Hospital 0.67642 1.9668 <0.001 On the way to the hospital- Home 0.65479 1.9247 <0.001 Hospital- On the way to the hospital 0.11442 1.1212 0.231 From the table 5-26, the following conclusions can be made. After adjusting for age, status, and the location o f the crash in the Multinomial Logistic regression analysis model, the odds ratio of dying at the scene o f the crash versus dying at home is 4 times greater for the victims suffering from head injury than bleeding; the odds ratio of dying at the scene o f the crash versus dying before reaching to the hospital is 2 times greater for 70 the victims suffering from head injury than bleeding; the odds ratio o f dying in the hospital versus dying at home is 2 times greater for the victims suffering from head injury than bleeding; the odds ratio of dying at the scene o f the crash versus dying in the hospital is 1.97 times greater for the victims suffering from head injury than bleeding; the odds ratio o f dying on the way to the hospital versus dying at home is 1.92 times greater for the victims suffering from head injury than bleeding; and the odds ratio o f dying in the hospital versus dying on the way to the hospital when the reason of death changes from bleeding to head injury is not statistically significant. From the above results, it can be concluded that the odds ratio o f dying at the scene of the crash vs. at home is the highest followed by dying at the scene of the crash vs. on the way to the hospital, in the hospital vs. at home, at the scene o f the crash vs. in the hospital, and on the way to the hospital vs. at home when the reason of death changes from bleeding to head injury. Once again, such results echo the severity o f crashes which more often lead to head injuries (most likely due to lack of protective gear) causing death at the scene, or shortly after on the way to hospital or in the hospital. Table 5-27 shows the odds ratio o f comparing the change in the place o f death when the reason o f death changes from bleeding to multiple fractures. 71 Table 5-27 The odds ratios of comparing the change in the place of death when the reason of death changes from bleeding to multiple fractures Odds comparing multiple fracture to bleeding as a reason of death Coefficient Odds P-Value At Scene-Home 1.32997 3.7809 <0.001 At Scene-Hospital 0.77391 2.1682 <0.001 At Scene-On the way to the hospital 0.77060 2.1611 <0.001 On the way to the hospital-Home 0.55937 1.7496 0.124 Hospital-Home 0.55606 1.7438 0.116 On the way to the hospital- Hospital 0.00331 1.0033 0.975 From table 5-27, the following conclusions can be made. After adjusting for age, status, and the location of the crash in the Multinomial Logistic regression analysis model, the odds ratio o f dying at the scene o f the crash versus dying at home is 3.8 times greater for the victims suffering from multiple fractures than bleeding; the odds ratio o f dying at the scene o f the crash versus dying at hospital is 2.2 times greater for the victims suffering from multiple fractures than bleeding; the odds ratio o f dying at the scene of the crash versus dying on the way to the hospital is 2.2 times greater for the victims suffering from multiple fractures than bleeding; and the odds ratio o f dying on the way to the hospital versus dying at home, the odds ratio o f dying in the hospital versus dying at home and the odds ratio o f dying on the way to the hospital versus dying in the hospital when the reason o f death changes from bleeding to multiple fractures are not statistically significant. From the above results, it is evident that the odds ratio o f dying at the scene o f the crash vs. at home is the highest followed by dying at the scene o f the crash vs. in the 72 hospital, and at the scene o f the crash vs. on the way to the hospital when the reason o f death changes from bleeding to multiple fractures. Odds Ratios for the Location o f the crash Table 5-28 shows the logistic regression results for the odds ratios of various combinations o f the place o f death as affected by the location of the crash. The location o f the crash is categorized as crashes occurred in in-town areas or those occurred in outof-town areas. Table 5-28 The odds ratios of comparing the change in the place of death for crashes occurred in in-town areas compared to those occurred in out-of-town areas Odds comparing out-of-town to intown as a place of death Coefficient Odds P-Value At scene-Hospital 0.93847 2.5561 <0.001 On the way to the hospital-Hospital 0.75309 2.1235 <0.001 At scene-Home 0.74633 2.1092 <0.001 On the way to the hospital- Home 0.56095 1.7523 <0.001 At Scene-On the way to the hospital 0.18538 1.2037 <0.001 Home-Hospital 0.19214 1.2118 0.470 The following conclusions can be made from the Table 5-28. After adjusting for age, status, and the reason o f death in the Multinomial Logistic regression analysis model, the odds ratio o f dying at the scene o f the crash versus dying in the hospital is 2.6 times greater for the victims being involved in the out-of-town crashes than those involved in in-town crashes; the odds ratio of dying on the way to the hospital versus dying in the hospital is 2 times greater for the victims who were involved in the crashes that occurred 73 in the out-of town areas than those involved in the crashes occurred in in-town areas; the odds ratio o f dying at the scene o f the crash versus dying at home is 2 times greater for the victims being involved in the out-of-town crashes than those involved in in-town crashes; the odds ratio o f dying on the way to the hospital versus dying at home is 1.7 times greater for the victims being involved in the out-of-town crashes than those involved in in-town crashes; the odds ratio o f dying at the scene o f the crash versus dying on the way to the hospital is 1 times greater for the victims being involved in the out-oftown crashes than those involved in in-town crashes; and the odds ratio of dying at home versus dying in the hospital for the victims being involved in the out-of-town crashes than those involved in in-town crashes is not statistically significant. From the above results, it can be concluded that the odds ratio o f dying at the scene of the crash vs. in the hospital is the highest followed by dying on the way to the hospital vs. in the hospital, at the scene o f the crash vs. at home, on the way to the hospital vs. at home, and at the scene of the crash vs. on the way to the hospital for the crashes happened in out-of town areas than those happened in in-town areas. Some possible reasons for such results include the lack o f the emergency services in out-of-town areas, long distances to emergency hospitals and, of course, the greater severity of crashes as a result o f higher speeds in the out-of-town areas. 74 5.4 Calculating the Economic Cost of Road Traffic Fatalities As I mentioned in chapter 4, to estimate the economic cost of road traffic fatalities in the study period, I followed Ayati’s methodology. Ayati (2009) in a very comprehensive study calculates the economic cost o f road traffic fatalities in Iran. He assumed the lost productive capacity coefficient is equal to that of loss o f quality o f life. The coefficient that he used for the loss quality of life is the same as the one estimated in the study by Goodchild et al. (2002) which is 4.76. Therefore, adding these two coefficients resulted in the value 9.52 which should be multiplied by GDP per capita to come up with the average annual economic cost that the Iranian society will suffer from losing each victim (Ayati, 2009). Since the calculation of the economic burden of road traffic fatalities is not the main subject o f this research, I will use the same coefficient as Ayati (2009) to calculate the average annual economic cost of each victim in this study. To do that, multiplying GDP per capita o f Iran during March 21 to December 21, 2009 (8 months) by 9.52 will result in the average economic cost o f losing one person in the study period as US$29,756. The total economic cost of road traffic fatalities is the product o f average economic cost of losing one person by the total number o f years o f life lost. To calculate the total number o f years o f life lost due to road traffic fatalities in the study period, I subtracted the age of each deceased based on their gender in the dataset from the life expectancy in Iran in year 2009 - life expectancy was 72 for males and 75 for females in 2009 (WHO, 2013) - and then summing up all o f those differentials results in the total number o f years o f life lost. 75 Therefore, the total number o f years o f life lost in the study period calculated as 585,556 years. Multiplying this number by the average economic cost o f losing one person (US$29,756) will amount to a total economic cost of around US$17 billion. This is a very substantial amount which accounts for 7% o f the GDP o f Iran during March to November 2009. 76 Chapter 6: Conclusion 6.1 Empirical Findings This study examined an exclusive dataset o f a total o f 16,556 individuals who died from the road traffic crashes in Iran during March 21 to November 21, 2009. These data that are not publicly available were obtained directly from the Road Maintenance and Transportation Organization of Iran through persistent personal contact with that organization. Taking advantage o f the detailed information in this dataset, I have been able to analyze demographic, socioeconomic and circumstantial profiles o f the victims of road traffic crashes, which has expanded the scope o f analysis beyond those undertaken in previous studies on Iran. The analysis of demographic profile of the victims indicates that the vast majority (79%) o f the victims were male and o f younger age (15-34 years). Less driving and travel and, therefore, exposure to road traffic crash risks is noted as a main reason for lower fatality rates for women. The analysis of the socioeconomic profile of the victims shows that about half o f the victims had no education or only primary education. It also shows that those with business occupations and “blue-collar” jobs were over-represented among the victims. The higher percentage for businessmen and blue collar workers may be explained by the greater frequency o f travel and mostly commuting with motorcycle which increase exposure to road traffic risks for these groups. The reason o f commuting with motorcycle is the existence of the heavy traffic jam in the big cities and the ban on cars in entering in some regions o f big cities during the day to reduce traffic jam. The only remaining ways 77 o f transportation in those regions are public transportations, motorcycles and authorized vehicles to avoid receiving high penalties. The analysis o f the circumstantial profiles o f the road traffic crashes reveals that the largest group o f victims are drivers, followed by passengers and pedestrians. Moreover, it is shown that more than two thirds o f the crashes are those that happened out-of-town areas. Over half o f the victims died at the scene o f the crash and almost 40% o f deaths happened in hospitals. Head injury figured as the main reason o f death accounting for over 61% o f all deaths. Such outcomes are taken as evidence of acute severity o f crashes in Iran that may have been caused by lack o f appropriate protective gear by drivers and passengers on the one hand, and lack of timely road emergency services, on the other. Bivariate analysis were done to test the potential relationship between some circumstantial factors with demographic, socioeconomic, and other circumstantial variables. Strong statistically significant associations are found between victim’s statues (as driver, passenger or pedestrian) and gender, age and education. The relationships between circumstantial, demographic and socioeconomic profiles of the victims, has been demonstrated using cross tabulation between status and gender, status and age, and status and education. It can be seen that among female deceased 67.2% o f them were passenger and among male deceased over half o f them were driver (51.9%). Deceased aged under 60 years old mostly tend to be the driver victims o f road traffic crashes. Over 50% of deceased aged 60 years old or above were pedestrian. Over 50% of driver had high school level of education. 49% o f pedestrian had no education and around 39% o f passenger had high school level o f education. 78 Moreover, strong statistically significant associations are found between the place o f death and one demographic variable (age) and 3 other circumstantial variables, namelystatus, the reason o f death, and the location o f the crash. From the related cross tabulation table between the place o f death and age, the place of death and status, the place o f death and the reason o f death, and the place of death and the location of the crash, it can be concluded that over 50% o f deceased aged under 60 years old died at the scene o f the crash which could be explained by their reason of death which is mostly head injury (over 65%). Over 50% o f driver died at the scene of the crash. Among pedestrian, 53% o f them died in the hospital and over 58% o f passenger died at the scene o f the crash. It also shows that about 53% o f people with head injury passed away at the scene o f the crash. 58% of deceased were involved in the crashes happened in out-of-town areas died at the scene o f the crash which could be because o f the higher severity o f the crashes happened in out o f town areas as traffic speed is normally higher on out-of-town roads and the quality and safety of those roads are typically poorer than roads in-town. Moreover, the monitoring o f traffic is usually less frequent and effective than that o f the roads in in-town areas. To further explore the relationships between circumstantial, demographic and socioeconomic profiles o f the victims, I did a multivariate analysis using Multinomial Logistic regression models to estimate the odds ratios for different categories o f victim’s status and place of death as affected by gender, age, education and other circumstantial factors as appropriate. The multivariate analysis of victims’ statues indicates that males and those under 60 years o f age are most likely to die as drivers. Whereas, females and those aged 60 or 79 above are most likely to die as pedestrians or passengers. Interestingly, the analysis shows that those with more education are also more likely to die as drivers. The multivariate analysis of the place o f death provides plausible results as well. The chances o f dying at the scene of the crash and on the way to hospital or in the hospital are greater for people younger than 60 years old, drivers, victims with head injuries and bleeding, and victims o f crashes in out-of-town areas. Finally, as an add-on analysis, this study presents a rough estimation o f the economic burden o f road traffic fatalities in Iran for the study period based on existing methodology to make it more sensible to policy makers the huge amount o f resources that the Iranian society lose as a result of deadly road crashes. The total economic cost as a result o f road traffic fatalities within the time frame o f this study has been estimated around US$17 billion which amounts to 7% of GDP of Iran during March to November 2009. This is a substantial cost imposed on a developing country. The finding o f this study is comparable to the similar study done by Ayati (2009). Ayati (2009) has estimated the traffic economic cost o f Iran at US$11.4 billion in 2004, which amounted to 7% of GDP o f Iran for that year. My study provides a more recent estimate o f the economic cost of road traffic fatality in Iran than other studies since I used the more recent road traffic data o f the year 2009. The higher estimated cost in absolute terms in this study reflects inflation o f prices over the 2004-2009 period (since I used the nominal GDP) as well as the greater number o f fatalities 80 6.2 Policy implications Iran has one o f the highest numbers o f road traffic mortalities in the world. Dealing with road safety should become a high priority policy in Iran for the politicians. As we saw, the issue has a significant cost (US$17 billion) for the country as well. The occurrence o f the high head injury as a first reason o f death could be because o f not using seat-belts by car occupants, helmets by motorcyclists and bicyclists, and also lack o f air bags in the cars. Usage o f the seat-belts for the front seat is compulsory but not for the back seat, also wearing motorcycle helmets is not compulsory in Iran. Police should enforce the usage o f the seat-belts for the back seat and oblige wearing the helmets by motorcyclists and cyclists. Also, carmakers in Iran should provide air-bags in all newly built cars. In Iran most of the cars are either fully home-made or are imported that have been partially assembled. Firstly, the majority of the home-made cars do not have air-bags and these are the ones that are mostly involved in deadly crashes. Secondly, local carmakers that assemble foreign brand cars provide the option o f removing the air­ bags o f these cars in case the customer wants to reduce the price. These two facts have caused very low penetration of air-bags feature in cars that are available in Iran. Moreover, the use of child safety seats is neither compulsory nor common in Iran and this should be changed by legislation and providing education to families. The study also shows that most o f the elderly people are involved in deadly crashes as a pedestrian. This is because in Iran the distance between zebra crossings is high and people usually cross the roadways. Also pedestrian bridges and underpasses usually do not have escalators which lead older people to jaywalk across the street or highway. But since older people are slower, they are more in danger from these hazardous behavior. 81 Therefore, providing escalators for pedestrian bridges and underpasses and decreasing the distance between zebra crossings should become a priority for policy makers. From the results o f this study, we saw that most of the victims involved in out-oftown crashes died at the scene o f the crash or right after that before reaching to the hospital which shows the lack o f emergency services in out-of-town roads. Therefore, policy maker should increase the emergency services in out-of-town areas to decrease the number o f deaths in out-of-town roads. From the multivariate analysis o f victim status, we saw that the vast majority o f victims were young people aged 15-34. This could be because o f the lack of driving experience. One reason could be the fact that obtaining driving license in Iran is much easier than in developed countries and most young drivers start driving without having enough education and skills. Also, fines and punishments for speeding, which is common within this group, are not preventive enough. 6.3 Limitations of the study Although some o f the novelties of this research are coming from the fact that the database that I have obtained from the Road Maintenance and Transportation Organization o f Iran was confidential and obtained after several months o f negotiation with them about the importance o f this study, there are several shortcomings related to this database. First o f all, as I did not have access to the large time series, I could not analyse the road traffic mortalities' trend and its changes during the time. In addition, the database lacks income level of deceased people. Having income level leads to a more reliable analysis o f socioeconomic situation of road traffic victims. 82 Also, the database is not detailed enough. For example, there is no distinction between car, motorcycle or bicycle drivers, which makes it difficult to interpret some of the results in Tables 5-20, 5-21 and 5-22. For instance, with greater differentiation among the vehicles, the effect of changes in the education level on the different categories of drivers (i.e. car drivers, motorcyclists or bicycle drivers) could be better analyzed. Moreover, there is no detailed information about collision types (head-on collision, run-off-road collision, collisions involving pedestrians, cyclists, or animals, etc.). Collision's time and weather condition at the time o f the accident is not available in the database either. Also, the database does not provide the location o f the roads that accidents happened in. Such shortcomings of the dataset did not let me find out on which roads accidents happen mostly and why. Identifying the most dangerous roads is valuable in allocating resources. Since, it would make it more obvious to the policy makers which roads need to be redesigned and how they should improve the safety o f those dangerous roads. It also would help policy makers with urban planning, road design and traffic flow management to help make roads more safe. Having all o f these missing information would allow a more comprehensive analysis of the road traffic mortalities in Iran. 6.4 Recommendations for future research There are many interesting subjects that could be the area o f future research if one can overcome before mentioned limitations o f this study. For example, it would be useful if one could obtain the time series o f road traffic crashes during an extended time frame to analyze the changes in the demographic, socioeconomic and circumstantial profiles o f the victims within years. Being able to observe these changes over time allows researchers 83 and policy makers to find out the trends o f road mortalities in different groups and to analyze the effectiveness of the policies that have been implemented to decrease the number of fatalities. For example, it would be interesting to analyze the effectiveness of preventive advertisements or educational programmes on different socioeconomic groups over time. Comparing road traffic accident trends o f Iran with other similar developing countries to find out which countries had more effective policies to reduce the mortalities is also an interesting subject for future research. Also, it would be very fruitful for setting policies if one can analyze the timing and exact place of the accidents and also the weather condition at the time of the accidents. This information can help to obtain ranking o f the most hazardous roads, time o f travel and weather condition for different profile of transport system users in Iran. Identifying the most hazardous roads provides the opportunity to pay more attention on the condition of those dangerous roads and how they can be improved. Especially, when the weather condition and the time o f the accidents are considered, it can be observed in which conditions most of the accident occur. Identifying the problems is the first step in improving the road standards and after a while with analyzing the accident's trend in those roads, it can be seen that if the improvement were effective or it still needs more work and attention. Finally, collecting information on whether drivers or passengers had seat belts on at the time o f crash and whether motorcyclist had helmets or other protective gear allows a better analysis o f the reasons of injury and fatality among the road traffic victims. 84 References Aeron-Thomas, A., Jacobs, G.D., Sexton, B., Gururaj, G., & Rahman, F., (2004). “The Involvement and Impact of Road Crashes on the Poor: Bangladesh and India Case Studies”. Transport Research Laboratory, Published Project Report, PPR010; Crowthome, United Kingdom. Al-Masaeid, H.R., Al-Mashakbeh, A.A., & Qudah, A.M., (1999). “Economic costs of traffic accidents in Jordan”. Accident Analysis and Prevention, 31, 347-357. Anbarci, N., Escaleras, M., & Register, C., (2006). “Traffic Fatalities and Public Sector Corruption”. KYKLOS, 59 (3), 327-344. Ayati, A., (2009). “Theory & Practice o f Traffic Accident Cost”. Transportation Research Institute, Ministry of Road and Transportation, Iran. Bhalla, K., Shahraz, S., Naghavi, M., Bartels, D., & Murray, C., (2008). “Road Traffic Injuries in Iran”. Harvard University Initiative for Global Health Road Traffic Injury Metrics Group (http://isites.harvard.edu/fs/docs/icb.topic510711 .files/ReportRoadTrafficIniuriesinlra n.pdf. accessed August 10,2011) Compton, R.P., Blomberg, R.D., Bums, M., Peck, R.C., & Fiorentino, D., (2002). “Crash risk o f alcohol impaired driving”. In: Mayhew DR, Dussault C, eds. Proceedings o f the 16th International Conference on Alcohol, Drugs and Traffic Safety, Montreal, 4-9 August 2002. Montreal, Societe de l’assurance automobile du Quebec, 2002:39-44 (http://www.saaQ.gouv.Qc.ca/t2002/actes/pdf/(06a').pdf. accessed April 10, 2013) Connelly, L.B., & Supangan, R., (2006). “The economic costs o f road traffic crashes: Australia, states and territories”. Accident Analysis and Prevention, 38, 1087-1093. Elvik, R., (1995). “an analysis o f official economic valuations of traffic accident fatalities in 20 motorized countries”. Accident Analysis and Prevention, 27, 2, pp. 237-247. 85 Elvik, R., (2000). “How much do road accidents cost the national economy? Accident Analysis and Prevention, 32, 849-851. Finch, D.J., Kompfher, P., Lockwood, C.R., & Maycock, G., (1994) “Speed, speed limits and accidents”. Crowthome, Transport Research Laboratory, (Project Report 58). Goodchild, M., Sanderson, k., & Nana, G., (2002) “Measuring the total cost o f injuries in New Zealand: a review o f alternatives methodologies”. Business and Economic Research Ltd, New Zealand, BERL# 4172. Hasselberg, M., Laflamme, L., & Ringback Weitoft, G., (2001). “Socioeconomic Differences in Road Traffic Injureis During Childhood and Youth: a Closer Look at Different Kinds o f Road User”. Journal of Epidemiol Community Health, 55,858-862. Hejazi, R., Shamsudin, M.N., , Radam, A., Rahim, K.A., Ibrahim, Z.Z., & Yazdani, S., (2012), “Estimation of traffic accident costs: a prompted model”. International Journal of Injury Control and Safety Promotion, DOI: 10.1080/17457300.2012.720578. Ingram, G.K., & Liu, Z., (1999). “Determinants of Motorization and Road Provision”. (http://elibrarv.worldbank.org/docserver/download/2042.pdf?expires-1314639301&i d=id&accname-guest&checksum=5CBEAD939A58BA0ADB61E8D90C9A35D0. accessed June 05,2011). I.R. o f Iran Road Maintenance & Transportation Organization, (2012). “Statistical Year Book o f Road Maintenance & Transportation Organization”. Iran, ('http://www.rmto.ir/NewTTO/e-archive/e-archivef7default.asp?ParentID=750. accessed September 05,2012). Jacobs, G., Aeron-Thomas, A., & Astrop, A., (2000). “Estimating global road fatalities”. Crowthome, Transport Research Laboratory, (TRL Report, No. 445). Jones, A.P., Haynes, R., Kennedy, V., Harvey, I.M., Jewell, T. & Lea, D., (2008). “Geographical variations in mortality and morbidity from road traffic accidents in England and Wales”. Health & Place, 14, 519-535. 86 Kajzer, J., Yang, J. K., & Mohan, D., (1992). “Safer bus fronts for pedestrian impact protection in bus-pedestrian accidents”. International Research Council on the Biomechanics o f Impact, 20,13-23. Kopits, E., & Cropper, M., (2003). “Traffic Fatalities and Economic Growth”. Washington, DC, The World Bank (Policy Research Working Paper No. 3035). Kostyniuk, L. P., Miller, L.L., Molnar, L.J., & Nation, A.D, (2006). “Societal Costs of Traffic Crashesand Crime in Michigan: 2006 Update”. The University of Michigan Transportation Research Institute. Report No. UMTRI-2006-23. Kubitzki, J., & Janitzek, T., (2009). “Safety and Mobility o f Older Road Users”. Allianz Deutschland AG, Munich. Laflamme, L., Burrows, S., & Hasselberg, M., (2009). “Socioeconomic Differences in Injury Risks- A Review o f Findings and a Discussion o f Potential Countermeasures”. World Health Organization Europe. Mitchell, K., (2002). “Older People and Road Safety: Dispelling the Myths”. World Transport Policy and Practice, 8 (2), 17-26. Moharamzad, Y., Taghipour, H., Hodjati Firoozabadi, N., Hodjati Firoozabadi, A., Hashemzadeh, M., Mirjalili, M., & Namavari, A., (2008). “Mortality Pattern According to Autopsy Findings Among Traffic Accident Victims in Yazd, Iran”. Chinese Journal of Traumatology, 11 (6), 329-334. Montazeri. A. (2004). “Road-Traffic-Related Mortality in Iran: a Descriptive Study”. Public Health, 118, 110-113. Naghavi, M., Shahraz, S., Bhalla, K., Jafari, N., Pourmalek, F., Bartels, D., Puthenpurakal, J. A., & Motlagh, M. E. (2009). “Adverse Health Outcomes o f Road Traffic Injuries in Iran After Rapid Motorization”. Archives o f Iranian Medicine, 12 (3), 284-294. Nakshabandi, M.M., (2007). “Casualties and Deaths From Road Traffic Accidents in Dohuk, Iraq”. Dohuk Medical Journal, 1 (1), 15-22. 87 Odero, W., Gamer, P. & Zwi, A., (1997). “Road Traffic Injuries in Developing Countries: a Comprehensive Review o f Epidemiological Studies”. Tropical Medicine and International Health, 2 (5), 445-460. Odero, W., Khayesi, M., & Heda, P.M., (2003). “Road Traffic Injuries in Kenya: Magnitude, Cause and Status of Intervention”. Injury Control and Safety Promotion, 10, 53-61. Peden, M., Scurfield, R., Sleet, D., Mohan, D., Hyder, A.A., Jarawan, E., & Mathers, C., (2004). “World Report on Road Traffic Injury Prevention”. Geneva, World Health Organization. Soderland, N., & Zwi, A.B., (1995), “Traffic-related mortality in industrialized and less developed countries”. Bulletin of the World Health Organization, 73 (2), 175-182. Statistical Center o f Iran, (2012). “Statistical Year Book”, Iran, (http://salnameh sci org.ir/AHUscr/DirectorvTreeComplete.aspx. accessed October 05, 2012). Tavakoli Kashani, A., Shariat-Mohaymany, A., & Ranjbari, A., (2012). “Analysis of Factors Associated with Traffic Injury Severity on Rural Roads in Iran”. Journal o f Injury and Violence Research, 4(1), 36-41. The United Nations and Road Safety, (2011). “Background”, (http://www.un.org/en/roadsafetv/background.shtml. accessed August 01, 2011). Turkish Statistical Institute, (2012). “Road Traffic Accident Statistics” (http://www.turkstat.gov.tr/VeriBilgi.do?alt id=52. accessed October 05, 2012). Taylor, M.C., Lynam, D.A., & Baruya, A., (2000). “The effects of drivers’ speed on the frequency o f road accidents”. Crowthome, Transport Research Laboratory (TRL Report 421). United Nations, (2012). “Population o f Urban and Rural Areas and Percentage Urban, 2011”, (http://esa.un.org/unup/CD-ROM/Urban-Rural-Population.htm. accessed October 13. 2012). 88 United Nations General Assembly, Resolution A/62/257, (2007). “Improving Global Road Safety”, (http://www.who.int/roadsafetv/news/un se report 2007 en.pdf. accessed July 05,2011). Whitlock, G., Norton, R., Clark, T., Pledger, M., Jackson, R., & MacMahon, S., (2003). “Motor Vehicle Driver Injury and Socioeconomic Status: a Cohort Study With Prospective Retrospective Driver Injuries”. Journal o f Epidemiol Community Health, 57,512-516. Wong, E., Leong, M.K., Anantharaman V., Raman, L., Wee, K.P., & Chao, T.C., (2002). “Road Traffic Accident Mortality in Singapore”. The Journal o f Emergency Medicine, 22 (2), 139-146. World Bank, (2006). “Iran, Islamic Republic o f - Road Safety Project”. Washington D.C. The World bank. (http://www- wds.worldbank.org/extemal/default/WDSContentServer/WDSP/IB/2006/Q2/15/00010 4615 20060216090521/Rendered/PDF/ProiectOInform 1ntO 10Appraisal0Stage.pdf. accessed October 15, 2012). World Health Organization, (2009). “Global Status Report on Road Safety-Time for Action”. Geneva, World Health Organization. World Health Organization. (2011). “Decade o f Action for Road Safety 2011-2020- Saving Millions o f Lives”. (http://www.who.int/violence injury prevention/publications/road traffic/saving mil lions lives en.pdf. accessed August 17, 2012). World Health Organization. (2013). “Countries: Iran”, (http://www.who.int/countries/im/en/. accessed April 02, 2013). World’s First Road Death. (2011). London, Road Peace, (http://admin.roadpeace.org/articles/worldfir.html. accessed July 17,2011) 89 Zadeh, H. S., Vahabi, R., Nazparvar, B., & Amoei, M. (2002). “An Epidemiological Study and Determination o f Causes of Traffic Accident-Related Deaths in Tehran, Iran (during 2000-2001)”. Journal o f Clinical Forensic Medicine, 9, 74-77. Zhao, S., (2006). “Rapid Motorization and Road Traffic Accidents in China”. University of Technology, Dalian, China Dalian 116024, (http://18.181.0.3 l/afs/sipb/proiect/wine/dosdevices/z:/afs/athena/course/l 1/11.95 l/ol dstuff/albacete/Other Documents/Europe%20Transport%20Conference/traffic and t ransnort /rapid m otorizationl360.pdf. accessed October 10, 2012). Zhou, Y., Baker, T. D., Rao, K., & Li, G., (2003). “Productivity Losses From Injury in China”. Injury Prevention, 9,124-127. 90