Bangladesh road crash costing discussion document. April 2003 TRL

Bangladesh road crash costing discussion document April 2003 TRL

Preface This discussion document is based on the data collected and draft reports produced by the Institute of Child and Mother Health (ICMH) and the Bangladesh Transport Foundation. Data analysis was conducted by the ICMH, with the assistance of Barry Sexton, TRL statistician. This document has been produced by the TRL project coordinator as guidance for ICMH, the lead local counterpart who have been commissioned to produce a case study report. This report is to be produced after the draft findings presented here have been discussed at a local workshop. This study has highlighted the many assumptions required in a road crash costing and the practicality of these assumptions should be confirmed locally. ii

Contents 1 Background... 1 1.1 Research study (Phase 2)... 1 1.2 Objectives... 1 2 Survey methodology... 1 2.1 Household surveys... 1 2.2 Other data sources... 3 3 Road traffic casualty statistics... 4 3.1 Background... 4 3.2 Study Findings... 4 4 Property damage costs... 7 4.1 Background... 7 4.2 Vehicle Damage costs... 7 5 Administration costs... 9 5.1 Background... 9 5.2 Study estimate... 10 6 Medical costs... 10 6.1 Background... 10 6.2 Study findings... 11 6.3 Medical cost calculations... 11 7 Lost output... 12 7.1 Background... 12 7.2 Lost time... 13 7.3 Average income... 13 7.4 Lost output calculations... 13 8 Human costs... 13 8.1 Socio-economic impacts... 14 9 Total costs... 15 9.1 Average costs... 15 9.2 National cost... 15 9.3 Cost conclusions... 16 10 References... 16 1 Background... 1 1.1 Research study (Phase 2)... 1 1.2 Objectives... 1 2 Survey methodology... 1 2.1 Household surveys... 1 2.2 Other data sources... 3 3 Road traffic casualty statistics... 4 3.1 Background... 4 3.2 Study Findings... 4 4 Property damage costs... 7 4.1 Background... 7 4.2 Vehicle Damage costs... 7 iii

5 Administration costs... 9 5.1 Background... 9 5.2 Study estimate... 10 6 Medical costs... 10 6.1 Background... 10 6.2 Study findings... 11 6.3 Medical cost calculations... 11 7 Lost output... 12 7.1 Background... 12 7.2 Lost time... 13 7.3 Average income... 13 7.4 Lost output calculations... 13 8 Human costs... 13 8.1 Socio-economic impacts... 14 9 Total costs... 15 9.1 Average costs... 15 9.2 National cost... 15 9.3 Cost conclusions... 16 10 References... 16 iv

SECTION 1: INTRODUCTION 1 Background Bangladesh road crash costing Discussion Document Road crashes are already a leading cause of death in the world and the casualty numbers are expected to increase before they improve. Even in low income countries where motorisation is low, road crashes are a leading cause of death amongst males in the prime of life. Road crashes also account for a huge economic loss, as well as human suffering. While road crashes have traditionally been estimated to cost 1 per cent of gross national product (a smaller percentage than in highly motorised countries), this amount has begun to be questioned. Mohan has argued that low income countries (LIC) have higher road crash costs than do high income countries (HIC) as:- Medical care (of similar levels) costs more in LICs than in HICs The scarcity of rehabilitation resources will result in more and longer term disability in LICs The lack of care facilities will mean more families in LIC will have to take care of injured. (Mohan, 2001) 1.1 Research study (Phase 2) Funded by the UK Department of International Development (DFID), this study was intended to update the road crash costing methodology used in developing countries. The case study countries included South Africa, India, Ghana and Bangladesh. Babtie Ross Silcock led the study with TRL responsible for coordinating the Bangladesh and India research. 1.2 Objectives In addition to updating the guidelines on how to conduct a road crash costing, this research also sought to: develop an accurate estimate of the number of road traffic casualties nationwide, document the social impacts of RTC, especially on the poor, estimate the cost of road traffic crashes to the country, and investigate the sensitivity of cost estimates. 2 Survey methodology 2.1 Household surveys The main source of data for both the cost estimates and the incidence rate was from household surveys. This approach allowed such key issues as under-reporting, relative involvement and impact on poor, and social impacts to be addressed. 2.1.1 Sampling Technique The population sample was all Bangladeshi citizens and a stratified sampling technique was followed during the selection of the study area. Dhaka and Rajshahi were selected from six metropolitan cities of Bangladesh. For the urban sample, 12 district towns were selected from the total 6 divisions, i.e. 2 district towns were chosen from each division. From the 12 district towns, a total of 24 upazilas (an administrative division of population about 300,000) were selected randomly, and from these upazilas, 5 unions were selected and 2 villages were chosen from each union for surveying. The names of the survey areas are shown in Appendix A which also shows the number surveyed and relative share of local population. 1

2.1.2 Questionnaire Interviews were conducted in person with the senior female member of the households or head of the households as they were believed to be more knowledgeable about the injury incidence and consequences of the household members. Four questionnaires were developed. One screening questionnaire was used with all households surveyed and established the baseline socio-economic data for the study. The other three questionnaires were for victims and were organised into bereaved family, seriously injured and slight injury questionnaire which was much shorter. Road traffic casualty severity definitions were: Slight Injury: Severe Injury: Death Injury: Injury receiving medical treatment (but not hospitalisation) or resulted in the loss of at least one day of usual work. Injury caused hospital admission for more than one day Injury caused death within 30days of the incident. 2.1.3 Pilot testing A small pilot testing of the whole study was done in a metropolitan area (part of Dhaka City), an urban area (district town of Munshiganj) and rural area (rural area of Sherpur Upazila) to finalise the methodology, field test of the questionnaire, and provide practical training to the data collectors. Investigators, consultants and supervisors were also involved in the pilot testing. 2.1.4 Project workshops An introductory workshop was conducted at which draft questionnaires and the sampling methodology was discussed. A second workshop was conducted after the pilot stage and included representatives from the following organisations: National Institute of Preventive and Social Medicine (NIPSOM) Bangladesh University of Engineering &Technology (BUET) Bangladesh Safe Community Foundation (BSCF) Rehabilitation Institute and Hospital for the Disabled (RIHD) Roads and Highways Department (RHD) : Road Safety Division. Institute of Child and Mother Health (ICMH) Dhaka Medical College Hospital (DMCH) Directorate General of Health Services (DGHS) 2.1.5 Full survey A total of 22 data collectors and 6 supervisors were recruited for three months. Preference was given to those with graduate degrees and experience in data collection for health surveys. A 6-day extensive training course on both conceptual and practical matters was conducted at ICMH. Investigators worked as trainers. The data collectors were divided into 5 teams each under one supervisor. Police department, Hospital directors and Health care managers of various level and other stakeholders were contacted in advance. One letter was issued from the directors of ICMH asking full cooperation from their (stakeholder) part. 2.1.6 Quality assurance Household and victim questionnaires were reviewed the same day they were completed and checked for omissions, specific errors, and inconsistencies. Early analysis was to be undertaken for checking expected patterns, unexpected results and revealing problems with study design/questions. Where necessary, the household would be re-interviewed on the questionnaire subset in question. One supervisor was assigned to each team of data collectors, except where there was one group which had six data collectors and thus required two supervisors. There were also frequent field visits by investigators and research associates which allowed for meetings with data collectors at regular intervals for problem solving. 2

Data were analysed at the department of Epidemiology and Biostatistics of Institute of Child and Mother Health. Microsoft Access was used for data entry and SPSS software used for analysis. 2.1.7 Population and Survey Distribution The number of households and population included in the survey are shown in Table 2.1. Additional data on survey locations is shown in Appendix A. Table 2.1 Bangladesh household survey Survey sample No. Households Urban Metropolitan Other Urban Average family size Population Share 15,083 9,161 4.77 4.99 71,980 45,699 17.3% 11.0% Rural 58,956 5.04 297,141 71.6% Total 83,200 4.99 414,820 100% 2.2 Other data sources While the household surveys were the main data source for the study, additional data were collected from other surveys and published data. These are discussed in the individual cost component chapters. 3

SECTION 2: ROAD TRAFFIC CASUALTY INCIDENCE 3 Road traffic casualty statistics 3.1 Background Police are officially responsible for reporting road crashes and casualties. However, the problem of under-reporting is well-known and has been previously addressed in DFID funded road safety research. A survey of road traffic casualties admitted to the two main trauma hospitals in Dhaka concluded that as few as between 3 and 13 per cent of road traffic injuries were being reported by police (Aeron-Thomas, 2000). Hospitals, however, are not a good source of data to check the under-reporting of road deaths as families may be reluctant to take bodies to hospital in order to avoid any post mortem requirement which could postpone burial. The Bangladesh Bureau of Statistics (BBS) undertook a household survey and estimated the national transport injury related death incidence rate to be over 5 per 100,000, almost twice that reported by the police. Males had a death rate some eight times more than females with urban males having the highest transport related death rate, with almost 13 per 100,000 (BBS, 1996). In 2001, the International Centre for Diarrhoeal Disease Research (ICDDRB), estimated there to be 95,000 unintentional injury deaths in Bangladesh, and attributed them primarily to road traffic crashes (ICDDRB, July 2001). However this estimate was based on the use of the 1996 Global Burden of Disease Indian subcontinent mortality patterns, adjusted for Bangladesh. Official data from the two most recent years are presented in Table 3.1 and shows: 1. a decrease of 24 percent in all reported casualties in the last year, including a 20 per cent decrease in road deaths. 2. fewer serious injuries and even fewer slight injuries than deaths reported to the police (the opposite of what would be expected by the injury pyramid ). Table 3.1: Bangladesh official recorded road traffic casualties 2000 2001 Fatal 3058 2388 Serious 2270 1661 Slight 1215 904 Total 6543 4953 Source: Bangladesh Police: Road Traffic Accidents National Report 2000 and 2001 3.2 Study Findings 3.2.1 Incidence Table 3.2 shows the number of casualties by severity type and incidence rate estimated by the household surveys. Appendix A includes more background information on the death rate calculations. Rural households were involved in almost three-quarters of fatal road crashes (73%), but with 30 per cent occurring in urban areas. Table 3.2 Household survey findings No. casualties surveyed urban rural national Fatal 56 148 204 Serious injury 156 222 379 Slight injury 655 689 1346 Killed/serious injury 212 370 583 All casualties 868 1059 1819 3.2.2 Comparison with official statistics Under-reporting Table 3.3 shows the level of under-reporting by the individual casualty severities. As there was a wide range of incidence found by the household surveys, both the best estimate available (weighted mean) 4

and a conservative estimate (lower estimate of the 95 per cent confidence interval) is shown for comparison. The study has found that the actual number of road deaths occurring in Bangladesh is at over 8,000 and currently estimated to be 12,786, which is at least 2.6-4.2 times greater than that included in official statistics. Table 3.3 Comparison of official statistics (2000) with study findings Police Study estimate Recorded by police Adjustment factor Statistics low* mean low mean Low mean Fatal 3058 8,084 12,786 37.8% 23.9% 2.6 4.2 Grievous injury 2270 121,397 165,469 1.9% 1.4% 53.5 72.9 Minor injury 1215 235,797 351,622 0.5% 0.3% 194.1 289.4 Fatal/Grievous injury 5328 129,481 178,256 4.1% 3.0% 24.3 33.5 All casualties 6543 365,277 529,878 1.8% 1.2% 55.8 81.0 *lower estimate of 95% confidence interval. With such a low number of road injuries being recorded by the police (as shown by the comparison with police reported deaths), the adjustment factors are much higher. The number of casualties being hospitalised is at least 53 times that reported and best estimated at 73 times larger. Slight injuries have an even larger adjustment factor as they will be the most common injury type and the least likely to be reported to the police. It should be noted that victim households were asked whether the crash had been reported to the police. While the majority of bereaved families (70%) claimed the police were informed of the fatal crash, only 23 per cent of seriously injured and 6 per cent of slightly injured said they had notified the police. If families were answering honestly, then many more casualties are being reported to the police but are not being included in the official statistics. The current Roads and Highways Department (RHD) road crash costings do adjust for underreporting. It assumes that 49 per cent fatal, 19 per cent grievous, and 7 per cent of simple crashes (not casualties) are reported. Table 3.3. shows that much higher adjustment factors are needed to account for the lower share of casualties being recorded by the police. 3.2.3 Characteristics of road traffic casualties The key characteristics of those killed and seriously injured in road crashes are summarised below: Sex and age distribution Females accounted for about one-sixth of road deaths (17%) and injuries (19%). This was consistent with police databases and hospital studies. Females appear to be at the greatest risk to a road death when young with girls accounting for over half of all female road deaths, over three times the share of male road deaths occurring to boys (See Table 3.4). Table 3.4 Road casualties by sex and age Killed Seriously Injured Male Female Total Male Female Total no. share no. share No. share No. Share No. share no. share 15 & under 29 17% 19 53% 49 24% 30 9% 10 19% 40 11% 16-45 104 63% 12 33% 117 57% 248 77% 34 65% 282 75% over 45 33 20% 5 14% 38 19% 46 14% 8 15% 54 14% Total 166 36 204 324 52 376 Road user type Among road deaths, pedestrians accounted for over four of every ten, slightly less than that reported by the police. Bus passengers were the second largest category with car and taxi occupants accounting for less than 5 per cent of road deaths. 5

Among the seriously injured, while pedestrians remained the prime casualty (24%), almost twice as many rickshaw and bicycle users were reported seriously injured than were mini/bus occupants. This trend continued with slight injuries with 43 per cent involving rickshaw and cyclists compared to 9% bus occupants. While the police reported similar pedestrian percentages for fatal and seriously injured (44%, 21%), they reported pedestrians accounting for only 8 per cent of minor/slight injuries, one third the share found by the household Table 3.5: Road user type casualty distribution (%) Death Serious Slight pedestrian 41% 24% 24% Bicycle 4% 6% 13% Rickshaw/pushcart 7% 23% 31% motorcycle 3% 10% 9% babytaxi/scooter 8% 13% 10% Car 3% 2% 2% Taxi 2% 1% 1% minibus/bus 23% 15% 9% truck/lorry 6% 3% 0% Others 2% 3% 0% surveys. As feared, police data was also found to overestimate 3 or 4 wheeled motor vehicle involvement and underestimate the involvement of NMV and motorcycles in injury collisions. Table 3.6: Vehicle involvement in casualty collisions Fatal Serious Minor Police Survey Police Survey Police Survey NMV 14% 13% 15% 25% 9% 45% 2 wheel motor vehicle 4% 7% 5% 9% 5% 11% 3-4 wheel motor vehicle 81% 80% 80% 68% 86% 47% 6

SECTION THREE: CRASH COST COMPONENTS This section includes a chapter on each of the main cost components Property damage Administration costs Medical costs Lost output Human costs The first two cost components are associated with the crash incident while the remaining three (medical costs, lost output and human costs) are classified as casualty related costs. Human costs are an indirect cost while the other four are resource costs. 4 Property damage costs 4.1 Background Property damage is a major component of road crash costs, and by far the largest tangible resource cost. In addition to vehicle damage costs, property damage costs include other vehicle related costs such as lost business, goods damage and damage to other property, i.e. traffic signs, bridges, fences. Vehicle damage accounts for the vast majority of all damage costs and should be the main concern of any cost calculation of property damage. The most recent RHD property damage costs are based on information collected as part of the Vehicle Operating Cost surveys conducted. Based on a sample of 95 vehicle claims from all motorised vehicle types, i.e. auto rickshaw, large bus, an average damage cost was derived which included both vehicle damage costs and an estimate of lost earnings. It should be noted that for all vehicle types save the large bus and car, the lost earnings were similar in size to the vehicle damage cost incurred. Earlier costings in Bangladesh had attempted to collect damage repair costs from garages and also from BRTA who are responsible for conducting post crash inspections. Neither of these approaches proved practical or reliable. 4.2 Vehicle Damage costs 4.2.1 Data sources The study sought to collect information on crash related property damage costs from three sources: Household survey Transport associations Insurance companies Household survey Victim households were questioned on any assets damaged in the crash. However, the vast majority of the victims were pedestrians, rickshaw occupants, and bus passengers. Less than 10 per cent of the bereaved or seriously injured families reported damaged assets of Tk 10,000 or more. Transport Associations Information on vehicle damage cost data was requested from the main offices of Bangladesh Sarak Paribahan Samity (Bangladesh Road Transport Association Public Transport Owners Association) and Bangladesh Sarak Paribabhan Sramik Federation (Bangladesh Road Transport Worker s Association). The latter represents drivers and helpers. Very few of the public transport vehicles were reported to have comprehensive insurance (5% estimated), with third party insurance coverage being much more popular. Members are not required to inform their associations of any road traffic crashes and the vast majority were believed to settle on the spot. Only about 10 per cent were thought to involve the association in the settlement negotiations. No records are kept and the above information is based on opinion, rather than recorded data. 7

Insurance companies The main focus for property damage cost data collection was with insurance companies as insurance claims are the most common source of vehicle damage estimates in both HICs and LICs. A total of nine companies provided claim data with one company providing data from 10 regional branches. They were requested to provide information on claims settled in the previous two years. The survey pro forma asked for the vehicle type involved, crash severity and the associated costs of vehicle damage, goods, medical costs, and other. Information on 1801 insurance claims was collected for this study. This included 117 fatal, 256 grievous and 1428 damage-only crashes. However some of these had insufficient information, did not involve any vehicle damage, or were from claims dating back many years. Accordingly, only 988 vehicle damage claim costs could be used. The survey findings were hampered by the lack of vehicle type data being provided for most of the claims. Thus while it was possible to analyse claims by crash severity, it was not possible to produce average vehicle damage costs by vehicle type. Due either to lack of availability or small sample size, insurance claims are the only practical source of vehicle damage costs currently, or at least readily, available in Bangladesh. However, insurance claim data cannot be assumed to be representative of all crashes. Relatively few vehicle owners claim through insurance companies and these will mainly be private car owners. Average vehicle damage costs should reflect all damaged vehicles, including commercial vehicles, 2-3 wheel motor vehicles and non-motorised vehicles. While VOC surveys did collect cost data on commercial vehicles and 2-3 wheel motor vehicles, they were small samples and no previous damage cost information had been collected on non-motorised vehicles. Household surveys showed a larger involvement of motorcycles and NMVs which could warrant additional surveys. 4.2.2 Property damage cost calculations The calculations explained below are presented in the Veh worksheet in the draft April 2003 Bangladesh crash costing calculations workbook which should be viewed with this report. As the calculations involved are more complicated than the other components, the process has been explained by steps. Step 1 Calculate vehicle involvement by type Given the relatively few crashes recorded in police statistics, the household survey data was used to identify the number and type of vehicles involved in road crashes. The starting point is to identify the vehicles which were listed as the victim s vehicle and the other vehicle involved. This step also allows the share of crashes involving pedestrians to be identified, as these are assumed to involve less damage. Step 2 Calculate basic vehicle type shares To simplify the calculations, basic vehicle type categories should be used as shown in Table 4.1.Table: 4.1 Vehicle involvement in road traffic crashes Average Vehicle type crash involvement number of vehicles Non motorised vehicle Two/three wheel motor vehicle Car/taxi Fatal 1.60.13.16.26.45 Serious 1.65.27.25.19.30 Slight 1.67.45.23.06.27 Minibus/ bus/lorry Step 3 Estimate average claim cost The average claim costs (per vehicle) by crash severity were derived from the sample of insurance data (See Table 4.2). Table 4.2 Vehicle damage insurance claims no. total cost average cost Fatal 76 4,153,693 54,654 Grievous 157 15,683,397 99,894 Minor Injury/Damage 755 68,486,003 90,710 Total 988 88,323,093 89,396 Source: Insurance companies 8

Step 4 Estimate average vehicle damage cost per damaged vehicle Not all vehicles involved in crashes will be damaged, especially if the other party is a pedestrian, which is often the case in Bangladesh. In order to avoid over-estimating the vehicle damage cost incurred, assumptions were made about the damage cost to vehicles. (Bangladesh is not unique in having a large share of crashes involving pedestrians. For the past several years, South Africa has adjusted its insurance claim estimate to reflect the share of collisions involving pedestrians). In an attempt to reduce the vehicle damage costs, the following arbitrary assumptions were applied. In crashes involving NMVs and 2 or3 wheeled motor vehicles, the average damage cost was 25 per cent the average insurance claim cost and for all other motor vehicles (cars, buses, and lorries), the average damage cost was estimated to be 50 per cent that found from insurance claims. Step 5 Calculate average vehicle damage cost The average vehicle damage cost will be calculated by multiplying the basic vehicle type shares by their adjusted/estimated vehicle damage costs. This will then be multiplied by the average number of vehicles involved to produce a vehicle damage cost per crash. This will be determined for each crash severity as the distribution of vehicle types will vary. Step 6 Estimate cost of damaged goods and calculate average property damage cost per crash by severity Previous Bangladesh costings have included an estimate of lost goods in crashes as part of the property damage costs. Insurance claims provided information on the value of goods damaged in the crash (See Table 4.3). However, these costs cannot be assumed to represent road traffic casualties, the majority of whom will be non-vehicle owners. An average damaged goods cost of Taka 38,000 was assumed to apply for only those crashes involving commercial vehicles. Table 4.3 Damaged goods claims (Taka) no. total cost Average cost Fatal 6 209,597 34,933 Injury 20 645,708 32,285 Damage 125 4,898,293 39,186 Total 151 5,753,598 38,103 Source: Insurance companies The average property damage cost per crash will include the average vehicle damage cost and an estimate of the cost of damaged goods. Average property damage cost per fatal crash is estimated to be Tk 55,433 Average property damage cost per grievous injury crash is estimated to be Tk 73,210. Average property damage cost per simple injury crash is estimated to be Tk 60,629. 5 Administration costs 5.1 Background The costs incurred by the police (scene management and investigation), insurance companies (claim handling) and judiciary (prosecution) in responding to road crashes are the associated administration costs of a road crash. Traditionally, they have accounted for a very small share of total crash costs. TRL Overseas Road Note 10 Costing Road Accidents in Developing Countries offered the then current UK values as shown in Box 1. Box 5.1: Total crash resource cost Fatal crash: 0.2% Serious injury crash: 4% Slight injury crash: 14% Damage only crash: 10% (TRL, 1995) A general insurance guideline held administration costs to be 15 per cent of claim costs, with half this amount from claim handling. A recent estimate of the costs of rural crashes in Iran accepted 10.8% of total rural road crash costs to be administration related (Ayati et al, 2002). 9

Previous Bangladesh road crash costings have not bothered to calculate the administrative costs on the basis that few crashes were attended and investigated or involved insurance claims or prosecution. The household survey found 20 per cent of bereaved families and less than 10 per cent of those seriously injured to have received private settlements. Only 2 bereaved families (1%) reported receiving any settlement from an insurance company. As reporting, investigation, settlement and prosecution will occur, if only in a minority of cases, a token amount of Tk 1000 was accepted in recent road crash costings as a best guess estimate of average administration costs. It should be noted that the insurance data collected included several fatal claims dating back to the early 1990s. Thus some bereaved families were having to wait almost 10 years to receive the statutory Tk 20,000 death benefit and this can be assumed to represent much follow up time on the part of the bereaved families. 5.2 Study estimate The current study invested a small amount into the investigation of police resource costs expended on road traffic crashes. Two police stations were visited in an attempt to collect police costs from attending and investigating road crashes. Officers from Savar Police Station identified the costs involved to include the following: Investigation time, Transportation of dead and injured, Processing post mortems, and Transferring bodies to families when latter could not afford to pay The police estimated an average of Tk 1200-1500 was spent on fatal and grievous injury crashes. When this was combined with salary costs and staff allowances, an average total of Tk 4000-5000 was spent on fatal and grievous injury crashes. The police stated that very few slight injury crashes were reported to them. However, according to the household survey, less than one-fourth of those seriously injured in road crashes were reported to the police and official police statistics show a very small percentage of the estimated seriously injured being included in the computerised database. The study findings have highlighted the difficulty in estimating the administration cost rather than the intended strengthening of the estimate. On this basis, the previous estimate of Tk 1000 will be used to represent all the administration costs involved, although it is acknowledged to be a token value and based on very little information. Average administration cost for road traffic collisions is estimated to be Tk 1000 6 Medical costs 6.1 Background Medical costs include the any at-scene treatment through to rehabilitation, i.e. first aid and rescue services, hospital costs (food and bed, x-rays and tests, operations, medicines, doctors services) and also any rehabilitation related costs, including prosthetics. The discounted value of funeral costs are also included as a medical cost. While medical costs are often only a small amount of total costs, this does not reflect their importance to either the hospitals or the victims and their families. With a minimal average per capita health expenditure, hospital resources cannot afford to be misused. For all those apart from vehicle owners, medical costs will be the largest direct cost and may have the most immediate impact on the family. 6.1.1 Data sources Information on medical costs of road traffic casualties was expected to be collected from three sources: Household survey Insurance claims Hospital studies 10

In the household survey, bereaved and seriously injured families were asked several questions about the treatment received and the medical costs involved. The review of road crash claims from insurance companies asked for medical costs separately to avoid grouping them with vehicle damage. ICMH was also working on a Burden of Injury study for the Ministry of Health which involved surveying 60 hospitals (45 upazilla headquarters and 15 district hospitals). Hospital checklists collected information on: 1. Total number of patients and injuries treated in emergency room and admitted as in-patients in 2000. 2. Age and sex of all in patients with their diagnosis and length of stay over a one-month period. 3. Injury profile of injury in-patients for one year period, including length of stay and outcome. 4. Age, sex, diagnosis, and outcome of injury patients in Emergency Unit. 5. Ambulance usage 6. Injury In-patient detailed profile, including cost of treatment and socio-economic characteristics of patient. However it should be noted that all sources will have their weaknesses as seen in Table 6.1. Table 6.1: Data weaknesses Source Weaknesses Hospital Usually only limited to running costs and not capital costs Does not include hospital subsidies Requires special survey Does not include out-of-pocket additional costs incurred by family Victim Possible overestimation of costs involved Requires special survey Does not include subsidised hospital costs Insurance Small sample, limited by insurance coverage and unlikely to be representative Previous costing Previous costings have relied on the average monthly in-patient cost provided by the Centre for the Rehabilitation of the Paralysed (CRP). Assumptions were made about the length of in-patient stay, out-patient costs and treatments, as well as transport costs from road crash scene. 6.2 Study findings 6.2.1 Type of treatment Road deaths Although two-thirds of road deaths were reported to have occurred at the scene of the crash, onequarter take place in the hospital. Approximately 10 per cent were admitted to hospital and three died after 30 days (1.5%). Seriously injured Most of the seriously injured were treated at emergency units in hospitals (74%) with another 16 per cent visiting doctors. Only 7 per cent reported seeing an allopathic quack. Over half did seek secondary treatment but the vast majority of these visited hospitals or physicians. The median stay in hospital was 7 days with three additional visits for treatment. Slightly injured For slight injuries, almost half sought treatment from a general practitioner (48%) while another 20 per cent visited an allopathic quack and 18 per cent went to an emergency unit. For the vast majority of those slightly injured (88%), that was the only type of treatment received with a few seeking additional assistance from doctors (5%) and allopathic quacks (4%). 6.3 Medical cost calculations The costs used here are mean values which are believed to be the best estimate currently available. Road casualties involve a wide range of injuries with varying levels of medical needs and costs. Lower and upper bounds based on the 25 and 75 per centile are shown in the worksheets. 11

6.3.1 Road death Funeral costs The average cost of a funeral cost was Tk 10,622, although 10 per cent reporting no related costs while another 10 per cent spent Tk 20,000 or more. As death is inevitable, so are funerals and their costs and so the discounted value of funeral costs (Tk 7,515) has been included in road crash costings. The impact of having to pay for a funeral decades in advance is considered in the social impacts, i.e. human costs. Over 86 per cent reported no travel costs involved with treatment visits and it is believed the transport of the body to the cemetery is included in the funeral related costs. Average medical cost of a road death: Tk 4259 (medical treatment) + Tk 7515 (discounted funeral cost) = Tk 11,774 6.3.2 Seriously injured Ten per cent of SI paid Tk 1200 or less on medical costs while the top 10 per cent reported spending at least Tk 35,000.. It has been assumed that Other includes the cost of post crash transport. Average medical cost of a serious injury: Tk 18,803 6.3.3 Slightly injured Slight injuries were only asked about their total medical costs. Ten per cent paid Tk 80 or less and the top ten percent paid Tk 1800 or more. Average medical cost of a slight injury: Tk 1,431 6.3.4 Comparison with other sources The road crash related medical costs from different sources are shown in Table 6.2 The review of 1051 motor vehicle insurance claims identified 51 medical cost claims. These do not include compensation awards for deaths (Tk 20,000), but only the amount paid for specific medical expenses. The small sample size helps explain the inconsistency in costs and only the most costly minor injuries would be expected to be pursued with insurance companies. Table 6.2 Comparison of medical cost estimates Household survey Insurance claims RHD No. Avg cost No. Average cost Road Accident Costs Fatal 204 11,774 10 13,474 300 Grievous Injury 379 18,803 33 5,558 4750 Minor injury 1346 1,431 8 8,788 113 The average in-patient cost previously used in the RHD costing (Tk 4750) was based on an assumed average in-patient stay of 10 days. The cost of injuries was estimated at approximately four times higher than the RHD values. Data concerns Household survey medical costs are based on memory recall as no households referred to written records or receipts. Out of pocket expenses may have been exaggerated by households in an attempt to obtain compensation. 7 Lost output 7.1 Background Lost output refers to the missed income from work foregone as a result of a crash. For road deaths, it is the lost working years while for the injured, it is the time spent recovering and also any time 12

required spent finding new work. Lost output costs are also incurred by the non-injured through lost travel time and from any work foregone from need to provide caring for the injured. 7.2 Lost time 7.2.1 Road deaths According to the survey, the average age of a road death was 26.2 years while a retirement age of 57 was used (this was the retirement age currently used by the RHD costing). This meant that the average road death lost a total of 30.8 years of working life. 7.2.2 Road injuries The household victim survey included questions about the number of days spent recovering and looking for new work. The amount of time spent looking for a new job was quite high and increased the amount of lost time by 50 per cent for grievously injuries and doubled that for slight injuries. Table 7.1 Amount of lost days of work to road traffic injuries No. recovery days No. days looking for Total missed days new employment Seriously injured 71 36 107 Slightly injured 6 7 13 7.3 Average income The average before crash incomes are shown in Table 7.2. It should also be noted that the injured reported a decrease in average income after the crash and this is discussed in the section on social-economic impacts of road traffic crashes. Table 7.2 Average monthly income Taka Fatality 3,729 Serious injury 3,726 Slight injury 2,958 7.4 Lost output calculations 7.4.1 Road death Based on the social discount rate of 12 per cent and average growth rate of 3 per cent, both of which were used in RHD s previous year s costing, Road death lost output= Tk 448,903 7.4.2 Serious injury Those seriously injured in road crashes were out of work for an estimated 107 days (3.6 months). Serious injury lost output = Tk 13,289 7.4.3 Slight injury The slightly injured were estimated to be out of work for 13 days (0.4 months). Slight injury lost output = Tk 1282 8 Human costs The cost components previously discussed, i.e. medical costs, property damage, administration costs and lost output, are all economic costs. They do not represent the total cost of road crashes as they 13

do not reflect the associated costs of pain, grief and suffering will cause nor the value of the quality of life lost from non-work days and years (retirement period). While human costs are regularly valued at larger sums than the direct costs in HICs, this is not the case in LICs. Previous RHD costings adopted the Indian approach and accepted human costs to be 20 per cent of economic costs calculated. This approach was maintained by this study as no other estimate of human cost was found to have been calculated in other cost of illness studies. It should be considered a conservative estimate, especially in light of the socio-economic impacts summarised below. 8.1 Socio-economic impacts The poor were defined as the bottom 45 per cent in terms of per capita income (household income divided by the number of household members). This definition was based on the finding of a recent study by the Bangladesh Bureau of Statistics. Both rural and urban poor were over-represented in fatal crashes, as poor households accounted for 57 percent of households involved in fatal crashes. 8.1.1 Household role of victim When asked about the social role of the victim in the household, most of those killed were described as children of the head of the household (44%). With the age distribution given, many of those killed appeared to be young adults living with their parents. Many do not appear to have started earning yet as 36 per cent were described as not contributing to the household income and students accounted for the largest number of road deaths (28%). Parents may well be losing their social security source as their children are dying before them. Poor families are more likely to lose their head of household or their spouse and thus suffer immediate economic impacts. Among the poor, 32 per cent of road deaths occurred to head of households or spouses, compared to 21 per cent for non-poor. What were impacts on victim and household? Over 70 per cent of poor households reported their household income, food consumption and food production to have decreased after a road death. Among the non-poor, the impacts were less with 57 per cent reporting household income to have decreased and slightly fewer (54%) stating their food consumption and production had worsened. Three quarters of all poor households affected by a road death reported a decrease in their living standard, compared to 59 per cent of non-poor households. What was the coping strategy adopted? Some 61 per cent of poor households were forced to arrange a loan after a road death, while only 34 per cent of non-poor had to borrow. A smaller number (35%) of poor bereaved households sold an asset, compared to 21 per cent of non-poor. Few households reported having to give up work or take on extra work after a road death, but again poor families were more likely to report these impacts. Only a minority of bereaved households received any compensation. Approximately one-quarter of poor households received private compensation, compared to 14 per cent of non-poor. Compensation from insurance companies was virtually non-existent with only one poor and one nonpoor bereaved households reporting receiving any insurance compensation. 14

SECTION 4: CONCLUSIONS This section includes summaries of data from the chapter on average and national cost calculations as well as discussing the implications of the study and the suggested follow-up. 9 Total costs 9.1 Average costs While the calculations of the average cost components have already been discussed, Table 9.1 Casualty rate per crash by severity medical costs, lost output and human costs fatal serious slight have been estimated on a per casualty basis Fatality 1.20 0 0 and need to be adjusted for a crash basis. Serious 0.30 1.53 0 The only data available was the official police Slight 0.16 0.39 2.09 data and so casualty rates were based on the most recent two-year period for which data All casualty 1.66 2.09 2.09 was available. Casualty costs were multiplied by the casualty rates shown in Table 9.1 in order to produce average crash costs shown in Table 9.2 Table 9.2 Average costs of crashes (Tk) Property damage Admin Lost output Medical Sub-total Human Costs total Fatal 55,433 1,000 541,297 19,998 617,728 123,546 741,274 Grievous 73,210 1,000 20,888 29,325 124,423 24,885 149,307 Minor 60,629 1,000 2,592 2,986 67,207 13,441 80,648 Damage only 6,063 1,000 7,063 7,063 The estimate of the average damage only collision was based on an assumption of a damage cost 10 per cent that of an average insurance claim plus the administration cost. While the insurance claim data showed much higher vehicle damage costs in damage only collisions, this was not considered to be representative of all damage only collisions, and a conservative assumption was made. 9.2 National cost Applying the casualty rates in Table 9.1 to the total number of casualties identified from the household survey incidence findings produces the following estimates of the number of road traffic crashes shown in Table 9.4. Table 9.4 Road crashes by severity type Number Fatalities Grievous Minor Total casualties Fatal 10,692 12,792 3,230 1,736 17,758 Grievous 106,062 162,234 41,765 204,001 Minor 147,660 308,124 308,124 All casualty crashes 264,414 12,792 165,464 351,624 529,880 The number of property damage crashes was based on multiplying the number of minor crashes by a factor of 3. This was the assumption used in the previous RHD costing and no additional information was available on the incidence of property damage crash occurrence. The total cost to the country is shown in Table 9.5 Table 9.5 Estimated National cost of road traffic crashes (Tk 000) Average cost Number Cost Fatal 741.3 10,692 7,925,773 Grievous 149.3 106,062 15,835,765 Minor 80.6 147,660 11,908,523 All casualty crashes 264,414 35,670,061 Property damage 7.1 442,981 3,128,729 Total crashes 855,056 38,798,790 15

9.3 Cost conclusions The national cost of road traffic crashes in Bangladesh has been estimated at Taka 38 billion (UK 455 million). This is approximately 2.9 times larger than the previous estimate. It should also be noted to be a conservative estimate as it has not factored in the following: Number and cost of the permanently disabled Travel time lost to road crashes Value of prevention, i.e. how much public would be willing to spend to reduce risk of road crashes It was also conservative in estimating property damage costs and used arbitrary assumptions to avoid overvaluing these costs. Furthermore, the poor have been shown to be over-represented in road traffic crashes. Their vulnerability to road traffic crashes and the associated socio- economic consequences is believed to be a strong argument for increasing the estimate of human costs, especially by donor agencies on foreign funded road rehabilitation programmes. 9.3.1 Summary discussion In summary, the key findings should be acknowledged: 1. The number of road deaths and injuries is severely under-estimated by the official estimates, and very large adjustment factors are needed. 2. The total cost of road crashes to the country is also much larger than previously estimated, despite the use of conservative assumptions. 3. There are many assumptions involved in valuing crash cost components and the estimates are very sensitive to the data sources. 4. The poor are at increased risk to both the incidence and the impacts of road crashes. Final workshop The preliminary findings and the assumptions involved need to be discussed at the final workshop. A draft workshop agenda and list of workshop participants has already been provided to IMCH. It has been proposed that after the study findings are presented, the following two workshop sessions be undertaken. Suggested research workshop discussion questions: 1. What additional research should be undertaken? 2. How should costs be updated? 3. How can the incidence and impacts on the low income be better understood? 4. Who should undertake the research and how should it be coordinated and disseminated? Suggested implementation workshop discussion questions: 1. How can road crash costs be used now? 2. What additional data or activities are needed to make better use of costs? 3. How can these be achieved? 4. How can under-reporting be accommodated in costing road safety interventions? How can road safety benefits for the low income be promoted? 10 References Aeron-Thomas A (2000), Under-reporting of road traffic casualties, unpublished TRL report, Crowthorne, UK. Ayati E, and W Young (2002), Estimation of the cost of rural road crashes in a developing country, Traffic Engineering and Control, February 2002, pg 56-60. Bangladesh Bureau of Statistics (1996), Accident and Injury, BBS, Dhaka. Mohan D (2001), Social cost of Road Traffic Crashes in India, 1 st Safe Community-Conference on Cost Calculation and Cost-effectiveness in Injury Prevention and Safety Promotion, 30 September-3 October 2001, Viborg County, Denmark. TRL (1995), Costing Road Accidents in Developing Countries. Overseas Road Note 10, TRL, Crowthorne, UK 16

Appendix A: Bangladesh household survey incidence findings Road deaths Annual within District fatalities households population RTA Death rate Location type weight diff from wtd diff over 5yrs per 100,000 wtd ave squared 1 32 10084 48,220 13.3 Dhaka metropolitan 0.67 4.4 12.86 6 0 4999 23,760 0.0 Rajshahi metropolitan 0.33-8.9 26.10 sum 15083 71,980 avg househld size 4.77 average 6.6 weighted ave 8.9 0.17 var of ave. 44.0 weighted var 39.0 2 6 876 4,425 27.1 Mymensingh urban 0.10 16.6 26.73 3 3 748 3,436 17.5 Manikanj urban 0.08 7.0 3.64 4 2 766 4,100 9.8 Chittagong urban 0.09-0.7 0.05 5 3 730 3,628 16.5 Cox's Bz urban 0.08 6.0 2.89 7 4 747 3,777 21.2 Sirajganj urban 0.08 10.7 9.42 8 4 756 3,769 21.2 Dinajpur urban 0.08 10.7 9.48 9 0 750 3,659 0.0 Jessore urban 0.08-10.5 8.83 10 0 755 3,518 0.0 Bagerhat urban 0.08-10.5 8.49 11 0 768 3,984 0.0 Sunamganj urban 0.09-10.5 9.62 12 2 751 3,790 10.6 Mauli bazar urban 0.08 0.1 0.00 13 0 764 3,756 0.0 Bhola urban 0.08-10.5 9.07 14 0 750 3,857 0.0 Jhalokati urban 0.08-10.5 9.31 sum 9161 45,699 avg househld size 4.99 average 10.3 weighted ave 10.5 0.11 var of ave. 8.6 weighted var 8.9 2 21 4997 22,440 18.7 Mymensingh rural 0.08 8.8 5.79 3 20 4981 25,044 16.0 Manikanj rural 0.08 6.0 3.04 4 30 4958 26,891 22.3 Chittagong rural 0.09 12.4 13.80 5 5 4989 25,531 3.9 Cox's Bz rural 0.09-6.0 3.14 7 4 4990 24,447 3.3 Sirajganj rural 0.08-6.7 3.68 8 9 5030 22,350 8.1 Dinajpur rural 0.08-1.9 0.27 9 16 4994 24,233 13.2 Jessore rural 0.08 3.2 0.86 10 3 5014 24,356 2.5 Bagerhat rural 0.08-7.5 4.61 11 10 4976 29,225 6.8 Sunamganj rural 0.10-3.1 0.96 12 18 4996 26,789 13.4 Mauli bazar rural 0.09 3.5 1.09 13 0 4887 24,546 0.0 Bhola rural 0.08-10.0 8.20 14 12 4144 21,289 11.3 Jhalokati rural 0.07 1.3 0.12 sum 58956 297,141 avg househld size 5.04 average 10.0 weighted ave 10.0 0.72 var of ave. 4.1 weighted var 4.1 Total sum 414,820 NATIONAL Overall average 9.9 var of ave. 2.8 using weighted estimates within strata stratified ave 9.42 stratified ave 9.84 strat. var of strat. var of ave 3.53 strat. SD of ave 1.88 ave 3.41 strat. SD of ave 1.85 95% CI from 5.74 95% CI from 6.22 to 13.10 to 13.45 17

SERIOUS INJURIES actual RTA Serious within District severe households population Injury rate Location type weight diff from wtd diff Code over 1 yr per 100,000 wtd ave squared 1 115 10084 48,220 238.5 Dhaka metropolitan 0.67 57.9 2244.61 6 15 4999 23,760 63.1 Rajshahi metropolitan 0.33-117.5 4555.34 sum 15083 71,980 avg househld size 4.77 average 150.8 var of ave. 7687.7 weighted ave 180.6 0.17 weighted var 6800.0 2 6 876 4,425 135.6 Mymensingh urban 0.10-52.6 267.85 3 5 748 3,436 145.5 Manikanj urban 0.08-42.7 136.90 4 2 766 4,100 48.8 Chittagong urban 0.09-139.4 1743.61 5 7 730 3,628 192.9 Cox's Bz urban 0.08 4.8 1.80 7 13 747 3,777 344.2 Sirajganj urban 0.08 156.0 2011.37 8 8 756 3,769 212.3 Dinajpur urban 0.08 24.1 47.78 9 12 750 3,659 328.0 Jessore urban 0.08 139.8 1564.18 10 5 755 3,518 142.1 Bagerhat urban 0.08-46.1 163.33 11 8 768 3,984 200.8 Sunamganj urban 0.09 12.6 13.87 12 8 751 3,790 211.1 Mauli bazar urban 0.08 22.9 43.47 13 3 764 3,756 79.9 Bhola urban 0.08-108.3 964.28 14 9 750 3,857 233.3 Jhalokati urban 0.08 45.2 172.08 sum 9161 45,699 avg househld size 4.99 average 189.5 var of ave. 644.3 weighted ave 188.2 0.11 weighted var 648.2 2 36 4997 22,440 160.4 Mymensingh rural 0.08 55.4 232.01 3 20 4981 25,044 79.9 Manikanj rural 0.08-25.1 53.27 4 19 4958 26,891 70.7 Chittagong rural 0.09-34.3 106.75 5 35 4989 25,531 137.1 Cox's Bz rural 0.09 32.1 88.47 7 19 4990 24,447 77.7 Sirajganj rural 0.08-27.3 61.24 8 15 5030 22,350 67.1 Dinajpur rural 0.08-37.9 107.97 9 26 4994 24,233 107.3 Jessore rural 0.08 2.3 0.43 10 15 5014 24,356 61.6 Bagerhat rural 0.08-43.4 154.49 11 28 4976 29,225 95.8 Sunamganj rural 0.10-9.2 8.31 12 56 4996 26,789 209.0 Mauli bazar rural 0.09 104.0 975.88 13 19 4887 24,546 77.4 Bhola rural 0.08-27.6 62.90 14 24 4144 21,289 112.7 Jhalokati rural 0.07 7.7 4.29 sum 58956 297,141 avg househld size 5.04 average 104.7 var of ave. 164.5 weighted ave 105.0 0.72 weighted var 168.7 total sum 414,820 overall average 147.4 var of ave. 254.3 using weighted estimates within strata stratified stratified ave 122.07 ave 127.28 strat. var of ave 323.69 strat. var of ave 299.19 strat. SD of ave 17.99 strat. SD of ave 17.30 95% CI from 86.80 95% CI from 93.38 to 157.33 to 161.19 18