|Year : 2017 | Volume
| Issue : 3 | Page : 166-171
Using bayesian model to estimate the cost of traffic injuries in Iran in 2013
Elaheh Ainy1, Hamid Soori2, Mojtaba Ganjali3, Ayad Bahadorimonfared4
1 Department of Research Affairs, Safety Promotion and Injury Prevention Research Center of Shahid Beheshti University of Medical Sciences, Tehran, Iran
2 Department of Epidemiology, Safety Promotion and Injury Prevention Research Center of Shahid Beheshti University of Medical Sciences, Tehran, Iran
3 Department of Statistics, Statistics School of Shahid Beheshti University, Tehran, Iran
4 Department of Health and Community Medicine, Faculty of Medicine, Shahid Beheshti University of Medical Sciences, Tehran, Iran
|Date of Web Publication||12-Sep-2017|
Department of Research Affairs, Safety Promotion and Injury Prevention Research Center of Shahid Beheshti University of Medical Sciences, Next to Taleghani Hospital, Evin, Tehran
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Background and Aim: A significant social and economic burden inflicts by road traffic injuries (RTIs). We aimed to use Bayesian model, to present the precise method, and to estimate the cost of RTIs in Iran in 2013.
Materials and Methods: In a cross-sectional study on costs resulting from traffic injuries, 846 people per road user were randomly selected and investigated during 3 months (1st September–1st December) in 2013. The research questionnaire was prepared based on the standard for willingness to pay (WTP) method considering perceived risks, especially in Iran. Data were collected along with four scenarios for occupants, pedestrians, vehicle drivers, and motorcyclists. Inclusion criterion was having at least high school education and being in the age range of 18–65 years old; risk perception was an important factor to the study and measured by visual tool. Samples who did not have risk perception were excluded from the study. Main outcome measure was cost estimation of traffic injuries using WTP method.
Results: Mean WTP was 2,612,050 internal rate of return (IRR) among these road users. Statistical value of life was estimated according to 20,408 death cases 402,314,106,073,648 IRR, equivalent to 13,410,470,202$ based on the dollar free market rate of 30,000 IRR (purchase power parity). In sum, injury and death cases came to 1,171,450,232,238,648 IRR equivalents to 39,048,341,074$. Moreover, in 2013, costs of traffic accident constituted 6.46% of gross national income, which was 604,300,000,000$. WTP had a significant relationship with age, middle and high income, daily payment to injury reduction, more payment to time reduction, trip mileage, private cars drivers, bus, minibus vehicles, and occupants (P < 0.01).
Conclusion: Costs of traffic injuries included noticeable portion of gross national income. If policy-making and resource allocation are made based on the scientific pieces of evidence, an enormous amount of capital can be saved through reducing death and injury rates.
Keywords: Bayesian model, contingent value, cost, death, injured, revealed preference, road traffic injury, stated preference, willingness to pay
|How to cite this article:|
Ainy E, Soori H, Ganjali M, Bahadorimonfared A. Using bayesian model to estimate the cost of traffic injuries in Iran in 2013. Int J Crit Illn Inj Sci 2017;7:166-71
|How to cite this URL:|
Ainy E, Soori H, Ganjali M, Bahadorimonfared A. Using bayesian model to estimate the cost of traffic injuries in Iran in 2013. Int J Crit Illn Inj Sci [serial online] 2017 [cited 2020 May 30];7:166-71. Available from: http://www.ijciis.org/text.asp?2017/7/3/166/214402
| Introduction|| |
Traffic injuries with the annual occurrence rate of 26.5 cases/100,000 people are the second cause of fatality and first cause of year life lost in Iran., In general, in Iran, the amount of year life lost as a result of traffic injuries is higher than that in the other part of the world and Eastern Mediterranean region, and this problem is one of the serious issues of the country. Death victims of road injuries are 3 people per 10,000 vehicles in the world, while in Iran, it is 7.3 people per 10,000 vehicles, which is unfortunately increasing every year. Moreover, traffic injuries can put a pressure on the health-care system that is mostly due to the insufficiency of resources. Low- and middle-income countries (LIC-MIC) have a high amount of death burden related to traffic injuries. Further and depth research of the economic burden due to road traffic injuries (RTIs) and more probing about the relationship of the health system factors and RTIs burdens to prevention and intervention programs should be produced in LIC, MIC, and high-income countries (HIC). According to Transportation Research Center, Iran University of Science and Technology, cost of traffic injuries using the human capital method was estimated 180,000 billion internal rate of return (IRR) in 2012. Generally speaking, the human capital approach proposes a number which is less than the actual cost of injuries, because in this method, cost components including special damage, execution, lost work, medical services, human cost, underreporting by the police and forensic medicine organizations, intensity and degree of injury, underreporting by insurance company, social costs including lost time, consumable fuel, additional pollution, cost of pain accompanying the injury, or worsening life quality, elderly, children, or patients are not included or estimated in the analysis. In contrast, willingness to pay (WTP) method presents a more accurate figure for cost rate and is an appropriate way for increasing social welfare by reducing injury and death.,,,, Death of thousands Iranians during the past 10 years due to traffic injuries and also injury and disability of millions of people during this period of time can by themselves place the necessity for such estimation on the list of the most important health-related priorities in the country. In this study, to estimate the cost resulting from traffic injuries, WTP method (which is used in HIC) was used. Thus far, this method has been less applied in LIC and MIC due to the accuracy in collecting the required information. We aimed to use Bayesian model to estimate the cost of traffic injuries in Iran in 2013.
| Materials and Methods|| |
In a cross-sectional study on costs resulting from traffic injuries, 846 people per road user were randomly selected and investigated during 3 months (1st September–1st December) in 2013. Global report of road safety was reported for Iran 2013 in which the portions of pedestrians, two-wheeled motorcyclists, occupants of four-wheeled vehicles, and drivers of four-wheeled vehicles were 28%, 23%, 26%, and 23%, respectively. Totally, based on the mentioned percentages, 846 people per road user were randomly selected and investigated. The research questionnaire was prepared based on the standard for WTP method considering perceived risks, especially in Iran, and its validity was determined using content validity which gained 75% of the overall degree of agreement out of 10 domain experts (by determining validity index and content validity ratio of 0.79). Its reliability was specified using test–retest method as r = 0.88. The study questionnaires included three parts: The contingent value (CV) approach involved direct questions; participants were straight asked that how much they were willing to pay for fatality risk reduction. The stated preference (SP) involved hypothetical scenarios for all road users (public transport drivers, motorcyclist, pedestrians, and occupants). The revealed preference (RP) method elicits value from real evidence such as importance to vehicle safety based on his/her WTP to more safety to his/her family or added safety device to own vehicle. In addition, WTP approach was used to estimate the value of statistical life and cost of injuries. Inclusion criterion was having at least high school education and being in the age range of 18–65 years old. Risk perception was an important factor to the study and measured by visual tool. Samples who did not have risk perception were excluded from the study. After a brief explanation about the study, consent letter was obtained from the participants. The collected data were analyzed after their strict control. Final analysis of WTP was carried out using a Bayesian method using WinBUGS software (version 1-4-3) 2012, University of Cambridge, UK.
Here, we will use a Bayesian approach for estimating the parameters. In this method, hierarchical structure should be used to define the model. The following model is used for this purpose:
In which as before the model's error has an extreme exponential distribution (and εi and i are the annual rate of WTP for person i, in this model Wi has Weibull distribution).
To carry out Bayesian inference, specifying prior distributions for the unknown parameters is necessary; as usual we have location parameter.
In this structure, N18 (0,1000I18) refers to a multivariate normal distribution with zero mean and large variance. Furthermore, I18 refers to the 18 × 18 unit matrix. An exponential distribution with mean 1000 is also considered, as a prior distribution, for the scale parameter of the Weibull distribution. Using this structure, a noninformative structure is considered for modeling.
Therefore, the above-mentioned model leads us to the following hierarchical model:
Methods to handle missing covariates
Various approaches have been developed to deal with missing covariate problems in regression analysis when the data are missing at random. In the Bayesian approach, missing values can be filled in using predictive distributions. In this approach, an appropriate distribution is used for modeling the covariates with missing values.
In our data set, daily payment for injury reduction, payment for time reduction, and distance were continuous variables, have missing values. The assumption of lognormal distribution with mean m and variances is used for modeling these missing values. Furthermore, a Bernoulli distribution with mean p was considered for “having accident” variable which has some missing values. Hence, the hierarchical structure was as follows:
Therefore, in the presence of these prior distributions, it is necessary to obtain posterior distributions for the unknown parameters; then, Bayesian inference is formed according to these posterior distributions. Again, two methods have been used for modeling. The first method uses the imputation of zero values for WTP with the amount of 500 and the second one uses the assumption of having a value <1000; results of both imputations are the same.
We have used regression imputation for handling missing covariates. Regression imputation (also known as conditional mean imputation) replaces missing values with predicted scores from a regression equation. The basic idea behind this approach is intuitively appealing: use information from the complete variables to fill in the incomplete variables. Variables tend to be correlated, so it makes good sense to generate imputations that borrow information from the observed data. In fact, borrowing information from the observed data is a strategy that regression imputation shares with maximum likelihood and multiple imputation although the latter approaches do so in a more sophisticated manner. The first step of the imputation process is to estimate a set of regression equations that predict the incomplete variables from the complete variables. A complete-case analysis usually generates these estimates. The second step is to generate predicted values for the incomplete variables. These predicted scores fill in the missing values and produce a complete data set.
As mentioned before “Accident,” “Dis,” “PTR,” and “DPFR” are the covariates with missing values. The following regression model is used for the continuous covariates (“Dis”, “PTR,” and “DPFR”):
Where, for the binary covariate (Accident), a logistic model is used. After imputation of the missing values, a maximum likelihood approach is used to obtain the parameter estimates.
| Results|| |
Out of all the collected samples, only 846 questionnaires were completely filled out. Given that, one of the key questions of this research was risk perception; it was observed that 64 out of 846 people wrongly responded to risk perception. Therefore, they were removed and data of 782 samples were analyzed. Mean age of the studied people was 33.4 ± 9.9. Men and women formed maximum and minimum gender percent by 89.3% and 10.7%, respectively. Employment status showed that the maximum number of people (45.8%) was self-employed, 64.2% owned houses, and 57.3% owned cars. Moreover, maximum percentages of leaving and returning home were in the morning by 88.4% and at night and midnight by 51%, respectively. The value of statistical life for one death was estimated to be 19,713,584,609 IRR based on mean WTP among road users 2,612,050 IRR and based on reduction rate of death risk from 26.5/100,000 people to 13.25/100,000 people. For total death cases in 2013 based on 20,408 death cases or 26.5 death cases per 100,000 people, 402,314,106,073,648 IRR which was equivalents to 13,410,470,202$ based on the dollar free market rate of 30,000 IRR with purchase power parity was obtained. Injury cost was 769,136,126,165,000 IRR equivalents to 25,637,870,872 $ (considering annual traffic volume 113,518: Daily traffic volume 311 in 2013 × the mean of daily payment 31,030 IRR × 250 working days × 318,802 injured people in 2013). In sum, injury and death cases came to 1,171,450,232,238,648 IRR equivalents to 390,048,341,074$.
Findings in [Table 1], results of Bayesian estimation using 500 Toman (one of the Iranian currencies) or 5,000 IRR for zero values and regression imputation for missing values showed that WTP had a significant relationship with age, middle and high income, daily payment to injury reduction, more payment to time reduction, trip mileage, private cars drivers, bus, minibus vehicles, and occupants [Table 1].
|Table 1: Results of Bayesian estimation using 500 Toman or (5000 internal rate of return) for zero values and regression imputation for missing values|
Click here to view
Findings in [Table 2], results of Bayesian estimation using 1,000 Toman or 10,000 IRR for zero values and regression imputation for missing values showed that WTP had a significant relationship with age, middle and high income, daily payment to injury reduction, more payment to time reduction, trip mileage, private cars drivers, bus, minibus vehicles, and occupants [Table 2].
|Table 2: Results of Bayesian estimation using 1000 Toman or (10,000 internal rate of return) for zero values and regression imputation for missing values|
Click here to view
| Discussion|| |
Results showed that value of statistical life for a death case was 19,713,584,906 IRR. WTP was higher among the people who had middle and high income compared with those with low income. It also had a significant relationship with age, trip mileage, and all road users except pedestrians. In Bhattacharya's study, value of statistical life in Delhi was estimated to be 150.000$. Rate of WTP grew with income and risk reduction. In Henry's study, real value of accident cost was estimated using a method that reflected willingness of the society to pay. Pitel also calculated the number and cost of injuries caused by traffic injuries using the WTP method in Canada. Cost of the injuries was also estimated to be 2,062,000,000$.
Calculating value of statistical life is one of the most important components in determining the value of risk reduction., León showed that calculating value of statistical life is a key component in general policy-makings, which is used in the frequent evaluation of efficiency in environmental projects and field making which are effective in death. His results showed that social profit of survival would be 9.54 lives/year with the value of 5.5 million dollars. In the current study, in sum, injury and death cases came to 1,171,450,232,238,648 IRR equivalents to 390,048,341,074$. Transportation Research Center of Iran University of Science and Technology estimated that cost of traffic injuries using human capital approach to be 180,000 billion IRR in 2012. Statistics used in this method which is taken from the police and forensics medicine organizations are usually underreported. In human capital method, value of hypothetical income for those with no production and the retired people is equal to zero. Financial estimation in human capital approach is less than its real value.
Considering the accuracy of WTP method, in this study, cost of traffic injuries was approximately 6.5 times of the fortune reported by Iran University of Science and Technology. During the two recent decades, many countries have used the WTP approach. WTP is the value considered by people for social death reduction. Other studies have emphasized the accuracy of this method. If WTP is considered to aim at reduced risk, a sign for value of statistical life would be obtained. Value of risk change or WTP is based on fundamental assumptions and shows that the adopted decisions on resource allocation in the governmental sector should reflect citizens' demands and preferences. The value allocated for improving road safety (risk reduction) is the total amount paid by people for preparation.,,,,
Hensher showed value of statistical life as an efficient method for the economic analysis of safety benefit and promotion of road environment. Other studies have demonstrated that preference expression studies may be valuable in the field of policy-making.,,, New approaches are being implemented for calculating value of statistical life. However, no improvement has been made in calculation methods for lost output, medical costs, and other costs (human capital model).,
The presented statistics showed that this rate was much higher than the global statistics. It seems that the undertaken measures have not been efficient in terms of reducing traffic injuries, or priorities of resource allocation have not been correctly detected, and thus, the made investments have been spent in less effective areas, the result of which has been high traffic volume in the country. The mentioned capital can be spent in the related departments such as traffic police, road police, and other concerned organizations to improve traffic status and promote road monitoring in the country.
Using the three methods of CV, SP, and RP can be considered the innovation of this study. Study on specific age group 18–65 years old and certain education at least high school education were the study's limitations. Bayesian approach in analysis allows us to combine prior knowledge with observations to make predictions about the phenomenon under study. In Bayesian inference, all unknowns in a system are modeled by probability distributions that are updated using Bayes' theorem as evidence accumulates. We show that the use of Bayesian models opens new possibilities for cross-sectional research.
| Conclusion|| |
Costs of traffic injuries included noticeable portion of gross national income. If policy-making and resource allocation are made based on the scientific pieces of evidence, an enormous amount of capital can be saved through reducing death and injury rates.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Naghavi M, Abolhassani F, Pourmalek F, Lakeh M, Jafari N, Vaseghi S, et al.
The burden of disease and injury in Iran 2003. Popul Health Metr 2009;7:9.
Bhalla K, Naghavi M, Shahraz S, Bartels D, Murray CJ. Building national estimates of the burden of road traffic injuries in developing countries from all available data sources: Iran. Inj Prev 2009;15:150-6.
Khosravi A, Aghamohamadi S, Kazemi E, Pour Malek F, Shariati M. Mortality profile in Iran (29 provinces) over the years 2006 to 2010. Tehran: Ministry of Health and Medical Education; 2013.
Gore FM, Bloem PJ, Patton GC, Ferguson J, Joseph V, Coffey C, et al.
Global burden of disease in young people aged 10-24 years: A systematic analysis. Lancet 2011;377:2093-102.
Mathers C, Fat DM, Boerma JT. The Global Burden of Disease: 2004 Update: World Health Organization, Switzerland; 2008.
Dalal K, Lin Z, Gifford M, Svanström L. Economics of global burden of road traffic injuries and their relationship with health system variables. Int J Prev Med 2013;4:1442-50.
Brijs T, Karlis D, Van den Bossche F, Wets G. A Bayesian model for ranking hazardous road sites. J R Stat Soc Ser A 2007;170:1001-17.
Wang Y, Guo Z. Road safety evaluation based on fuzzy logic. J Tongji Univ 2008;36:47.
de Blaeij A, Florax RJ, Rietveld P, Verhoef E. The value of statistical life in road safety: A meta-analysis. Accid Anal Prev 2003;35:973-86.
Dionne G, Lanoie P. Public choice about the value of a statistical life for cost-benefit analyses: The case of road safety. J Transp Econ Policy 2004;38:247-74.
Nakahara S, Jayatilleke AU, Ichikawa M, Marasinghe A, Kimura A, Yoshida K. Feasibility of standardized injury surveillance and reporting: A comparison of data from four Asian nations. Inj Prev 2008;14:106-12.
Peden M. World Report on Road Traffic Injury Prevention. Geneva: World Health Organization; 2004.
Rahimi-Movaghar V, Zarei MR, Saadat S, Rasouli MR, Nouri M. Road traffic crashes in Iran from 1997 to 2007. Int J Inj Contr Saf Promot 2009;16:179-81.
Bhattacharya S, Alberini A, Cropper ML. The value of mortality risk reductions in Delhi, India. J Risk Uncertain 2007;34:21-47.
Le H, van Geldermalsen T, Lim WL, Murphy P. Deriving Accident Costs using Willingness-to-Pay Approaches-A Case Study for Singapore. In Australasian Transport Research Forum (ATRF), 34th
Pitel S, Solomon R. Estimating the number and cost of impairment related traffic crashes in Canada: 1999 to 2010. Oakville: MADD Canada. Apr; 9:2013. Available from: http://www.madd.ca/media/docs/estimating
. [Last accessed on 2013 Oct 25].
León G, Miguel E. Transportation choices and the value of statistical life. Barcelona, Spain National Bureau of Economic Research; 2013.
Trawen A, Maraste P, Persson U. Methods for Estimating Road Accident Costs-a Comparison of Costs for a Fatal Casualty in Different Countries. International Conference: Traffic Safety on Three Continents (No. VTI Konferens 18A); 2001.
Hultkrantz L, Svensson M. The value of a statistical life in Sweden: A review of the empirical literature. Health Policy 2012;108:302-10.
Litman T. Valuing transit service quality improvements: Considering comfort and convenience in transport project evaluation. Victoria Transport Policy Institute (www. vtpi. Org) 2007. Available from: http://citeseerx.ist.psu.edu/viewdoc/summary
. [Last accessed on 2014 Oct 14].
Chhotu AK, Kumar CS. Willingness to pay for better safety on state highways. Int J Comput Eng Res 2014;5:407-10.
Hensher DA, Rose JM, de Dios Ortúzar J, Rizzi LI. Estimating the willingness to pay and value of risk reduction for car occupants in the road environment. Transp Res Part A Policy Pract 2009;43:692-707.
Jennings WP, Jennings PR. Risk, the Willingness-to-Pay, and the Value of a Human Life. Journal of Insurance Issues. 2000 Oct 1:180-4.
Belin MŠ, Tillgren P, Vedung E. Vision zero – A road safety policy innovation. Int J Inj Contr Saf Promot 2012;19:171-9.
Hiselius LW. The Value of Road and Railway Safety: An Overview. Lund University; 2003 Jan 1. LUND UNIVERSITY publication.
[Table 1], [Table 2]