Under-reporting of deaths and injuries resulting from road traffic crashes (RTCs) is a major global problem affecting many developing and developed countries around the world.1^–8 As road traffic injuries (RTIs) are projected to be the third leading cause of disability-adjusted life years lost globally by the year 2020,9 a critical first step towards reducing the RTI burden is the availability of reliable, accurate, and adequate data on RTCs and the resulting fatalities and injuries.

Road safety in India is the responsibility of the Transport Ministry and the Police Department, with the role of the Health Ministry mostly being limited to provision of trauma care after RTCs.1011 The National Crimes Record Bureau (NCRB) is the nodal agency responsible for the collection of data from the police in each state/city and compilation, analysis, and dissemination of injury-related information. According to NCRB, 105 725 people died and 452 922 were injured in RTCs in India in the year 2006.12 There is a suggestion that these numbers are under-reported, and some concerns have also been raised about the quality of these data.13^–16 As policy makers use these data for identifying and prioritizing issues related to road safety intervention,15 it is imperative to determine the accuracy of these data.

We describe the magnitude and pattern of under-reporting of RTIs and resulting fatalities to the police from two studies in the Indian city of Hyderabad. According to NCRB, 2941 RTCs resulted in 532 people dead and 2798 injured in Hyderabad in the year 2006.12 Hyderabad has a population of 3.8 million, excluding the surrounding areas that make up Hyderabad agglomeration,17 and had 1.2 million registered motor vehicles in 2001–2002 with the majority being motorized two-wheeled vehicles (77%).18

METHODS

These studies were approved by the ethics committee of the Administrative Staff College of India, Hyderabad, India.

Population-based study

The detailed methodology for the population-based study is reported in the companion paper.19 In brief, on the basis of census data for Hyderabad,20 we selected the study population from 50 clusters using a three-stage random cluster sampling procedure and one cluster of 49 homeless persons to represent the 5–49 years age group in the population.21 Trained interviewers obtained written informed consent from eligible people followed by a confidential interview from October 2005 to December 2006. Of relevance to this paper, reporting of RTCs to the police and type of treatment sought for the most recent RTIs in the preceding 12 months were documented in addition to the vehicles involved, time of RTC, and the number of people injured. Data on RTI-related death of a household member in the preceding 3 years were documented by interviewing the head/woman of the household. RTI was defined as any injury resulting from an RTC irrespective of the severity and outcome.19

Hospital-based study

A total of 781 consecutive RTI cases reporting to the emergency department of two large public hospitals and three branches of a large private hospital in Hyderabad were recruited from November 2005 to June 2006. People of all ages with RTIs (defined as in the population-based study) who either reported alive to the emergency department or were dead on arrival at hospital were included. Trained interviewers were posted round-the-clock in the emergency department and mortuary to capture all RTI cases. They documented contact addresses in detail to ensure that a follow-up was possible.

Interviews were conducted using a questionnaire designed for this study after written informed consent had been obtained from the injured person or the care giver, or a responsible adult family member in the case of death. Data were collected from the injured person where possible or from the care giver or a responsible adult family member. Of relevance to this paper, detailed data on RTCs that resulted in RTIs were documented, including reporting of the RTC to the police. All recruited RTI cases (families in the case of death) were followed-up for a 6 months from the date of discharge/death. Reporting of the RTC to the police was documented at follow-up for those who had not reported it to the police during the first interview. The reasons for not reporting the RTC to the police were also documented.

Data management and statistical analysis

Data were entered into an MS Access database, and SPSS was used for statistical analysis. The main outcome variable assessed was reporting of RTCs resulting in RTIs to the police, which is carried out in India by registering a First Information Report (FIR).22 Data on the most recent non-fatal RTI in the preceding 12 months for participants aged 5–49 years and fatal RTIs for all ages in the preceding 3 years from the population-based study, and non-fatal and fatal RTI cases from the hospital-based study for all ages were analysed. An RTI resulting in death irrespective of the time between the RTC and death was considered a fatal RTI. Follow-up could not be carried out for 24 (3.1%) non-fatal RTI cases in the hospital-based study. For these cases, the information provided in the first interview during the hospital admission was considered for analysis. Categories of type of road user included pedestrian, cyclist, and user of motorized two-wheeled vehicle (MTV), motorized three-wheeled vehicle (commercial passenger vehicles: auto-rickshaw and seven-seater), car/jeep, or other vehicle (bus/tempo/truck/lorry).

Univariate and multivariate analyses were performed separately for the two studies to understand the association of reporting the RTC resulting in RTI to the police with various characteristics to identify those that may play a significant role in determining this reporting. In the multiple logistic regression models, the effect of each category of a multi-categorical variable was assessed by keeping the first or the last category as reference, and all the variables were introduced simultaneously into the models. The reasons for not reporting RTCs to the police are reported for both studies. The 95% CIs of estimates for the population-based study were calculated taking into account the design effect of the cluster sampling strategy,23 and χ² test for significance is reported where appropriate.

RESULTS

Participation

A total of 10 459 (94.3%) of the 11 097 eligible participants aged 5–49 years in the population-based sample were interviewed; 2809 (26.9%) were 5–14 years of age, 1372 (13.1%) were 15–19 years of age, 6278 (60%) were 20–49 years of age, and 5376 (51.4%) were male.

Of the 781 prospective RTI cases in the hospital-based study, 610 (78.1%) were recruited from the two public hospitals, and 171 (21.9%) from the three branches of the private hospital. Of these, 640 (81.9%) were male, and 106 (13.6%) were fatal RTI cases; in 40 (37.7%) of these cases, the patient was dead on arrival at the hospital.

Non-fatal RTIs

In the population-based study, of the 1032 (9.9%) most recent non-fatal RTIs in the preceding 12 months, 336 (32.6%) were sustained by a pedestrian and 367 (35.6%), 226 (21.9%), 35 (3.4%), 15 (1.5%), and 52 (5%) by a user of an MTV, bicycle, motorized three-wheeled vehicle, car/jeep, and other vehicle, respectively. Only 20 of the 1032 most recent non-fatal RTIs (1.9%, 95% CI 1.1% to 2.7%) were reported to the police. Of 571 (55.4%) and 29 (2.8%) participants who had sought treatment as an outpatient or inpatient, 13 (2.3%, 95% CI 1.1% to 3.5%) and five (17.2%, 95% CI 3.5% to 30.9%), respectively, had reported the RTC resulting in the RTI to the police.

Of the 675 non-fatal RTIs in the hospital-based study, 166 (24.6%, 95% CI 21.3% to 27.8%) patients had reported the RTC resulting in the RTI to the police. Significantly less reporting of RTCs to the police by pedestrians (p<0.001), cyclists (p<0.001), MTV users (p<0.001) or users of motorized three-wheeled vehicle (p<0.003) was found in the population-based study than in the hospital-based study (fig 1). RTCs involving cars, jeeps, and heavy vehicles as the other vehicle and those with more than one person injured had significantly higher odds of being reported to the police in both studies (table 1). RTCs involving occupants of cars/jeeps/tempos/vans/heavy vehicles and those occurring between 22:00 and 06:00 in the population-based study were significantly more likely to be reported to the police, but this was not significant in the hospital-based study. The RTC was more than twice as likely to be reported to the police in the cases recruited from public hospitals than in those recruited from private hospitals (table 1). The results of multiple logistic regression analysis considering only people who sought outpatient or inpatient services for their RTIs in the population-based study were similar.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Percentage of road traffic crashes (RTCs) resulting in non-fatal injuries reported to the police for the type of road user in the population-based and hospital-based studies in Hyderabad. Motorized two-wheeled vehicle includes moped, luna, scooter, scooterette, and motorcycle. Motorized three-wheeled vehicles are commercial passenger vehicles and include auto-rickshaws and seven-seaters.

RTIs resulting in expenditure of more than 5000 Indian Rupees (US$124.40; £71.09; €91.56) were significantly more likely (p<0.001) to be reported to the police in both the population-based (18.6%) and hospital-based (46%) studies compared with those that resulted in a lower expenditure (0.9% and 25.1% for population-based and hospital-based studies, respectively). This variable was not included in the multiple logistic regression model, as the cost is a consequence of the RTI and we limited the model to the variables related directly to the RTC.

Fatal RTIs

Eighteen deaths of people of all ages from RTIs in the preceding 3 years were reported in the population-based study. Of these, eight (44%) died at the scene, two (11%) on the way to hospital, six (33%) in a health facility, and two (11%) at home. Fourteen (77.8%, 95% CI 65.1% to 90.5%) fatal RTIs were reported to the police. All RTCs that resulted in death at the scene were reported to the police (fig 2). Of the four fatal RTI cases not reported to the police, the deceased was male in three (75%) cases, a pedestrian in two (50%) cases, a user of a motorized three-wheeled vehicle in two (50%) cases, and 50 years of age or more in three (75%) cases. In three of these cases, a mistake made by the deceased was thought to have resulted in the RTC; two died within 6–12 h of the RTC and one on the way to hospital. The remaining person died >7 days after the RTC.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Percentage of road traffic crashes resulting in fatal road traffic injuries reported to the police in the population-based and hospital-based studies in Hyderabad.

Of the 781 RTI cases in the hospital-based study, 106 (13.6%) were fatal. Of these, 17 (16%) died at the scene, 22 (20.8%) on the way to hospital, 62 (58.5%) in hospital, and five (4.7%) at home/elsewhere. All but two of these cases (98.1%; 95% CI 95.5% to 100%) were reported to the police (fig 2). In the two cases not reported, death occurred >30 days after the RTC, one at a health facility and one at home/elsewhere (fig 2).

Reasons for not reporting the RTC to the police

In the population-based study, among all the non-fatal RTI cases (1032), those who had sought outpatient treatment (571), and those who had inpatient treatment (29), 843 (81.4%), 440 (77.1%), and seven (24.1%), respectively, mentioned that it was “not necessary” to report the RTC to the police.

Of the 511 RTI cases in the hospital-based study who did not report the RTC to the police, 500 (97.8%) provided a reason for not reporting (fig 3). Similar proportions reported “hit and run case” (the other party ran away after the RTC; 32.4%) and “not necessary” (31.8%), followed by “settlement between the parties” (20.4%) as the reasons for not reporting the RTC to the police. “Hit and run” was cited as the reason for not reporting the two fatal RTI cases to the police.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Reasons for not reporting road traffic crash resulting in road traffic injury to the police in the hospital-based study in Hyderabad.

DISCUSSION

Under-reporting of deaths and injuries resulting from RTCs is a major issue globally. We found gross under-reporting and selective reporting of non-fatal RTIs to the police in this urban Indian population. Only 17.2% and 2.3% of non-fatal RTIs requiring treatment as inpatient and outpatient, respectively, in the population-based study were reported to the police, and 24.6% of the non-fatal RTI cases admitted in the hospital-based study were reported to the police. Even fatal RTIs were under-reported, with 22% fatalities in the population-based study not reported to the police.

Police records are the main source of information on RTI-related deaths and injuries in India.12 The police are required to file a FIR when an RTC is reported to them, so that legal proceedings, if necessary, can be initiated. A FIR can be filed by any person irrespective of whether he/she has first-hand knowledge of the RTC.22 Our data suggest that a large proportion of the people involved in an RTC fail to report it to the police. Irrespective of the type of study, many people thought that it was not necessary to report an RTC to the police and that they needed to have information on the other party to report it. Settlement between the two parties involved in the RTC was also a common reason for not reporting it to the police. These data highlight the practical issues that need to be addressed to improve reporting to the police. Although mechanisms are in place to register FIRs in India, in general, people do not consider it easy to interact with the police. It is widely known that the drawn-out administrative process, including attendance at a court of law, is a major deterrent to people reporting RTCs to the police. Therefore, not reporting it or settlement between those involved are the preferred options for the majority. We found selective reporting of RTCs, as RTCs involving cars/jeeps and heavy vehicles, night time RTCs, and those with more than one person injured were reported in higher numbers in the population-based study, which was more or less similar to the hospital-based study. In addition, RTCs resulting in death and severe injuries were more likely to be reported to the police in the hospital-based study. The reporting to police also increased with increase in RTI-related expenditure. These data highlight that people do not feel the need to report RTCs to the police unless the RTI is serious or compensation is needed, in which case also only less than a quarter are reported to the police. RTI reporting to the police has been previously found to be lower in less developed countries, with 70% of inpatient and 25% of outpatient RTIs reported in the Netherlands8 and 82% of RTIs in Australia6 compared with 48% of hospital patients in India24 and 4% of serious injuries in Pakistan.7

Police also obtain information on RTC injuries and fatalities through hospitals. Injuries or deaths in an RTC reported or presented to a doctor are required by law to be documented in the medicolegal case register maintained by the doctor/hospital. The information from this is then expected to be given to the police on a regular basis for further action. We reviewed this register and the process at one of the study hospitals. The register includes the following information: name, age, sex, caste, occupation, address, identification marks, who brought the injured/dead to hospital, whether the police had been notified, whether a declaration to the police was required, and the nature of injury and treatment provided. No information is documented on the type of road user or the RTC per se. The medical officer completing the register decides if the case should be reported to the police on the basis of what he hears from the injured person or anyone else providing information. We also attempted to understand what happens in cases where information from the medicolegal case register is provided to the police. Communications with a number of police officials at different levels revealed that there is ambiguity about how this information is used or further documented by the police. Some suggested that the police attempt to contact the injured/family to seek details about the RTC. On the basis of the information gathered from them and enquiries at the RTC location, a decision is made about whether to register a FIR. On exploring this conditional FIR registration further, lack of human resources to deal with the RTC investigation and administrative work were elucidated as the main reasons for not doing so. This lack of human resources has been documented previously.1325 This suggests that the other reasons for under-reporting are hospitals not presenting all cases to the police, and police not recording all cases reported to them. These issues with the reporting systems are well acknowledged in India, although reporting system practice varies widely throughout the country.24

In our studies, we documented RTI reporting to the police by the respondent/care giver. We did not attempt to link the information provided by the respondent/care giver to the police records because of logistical problems in tracing records over the preceding 12 months against the background of poor and incomplete documentation by the police.13 It is unlikely that we have overestimated the under-reporting to the police because, in the event of someone else/hospital reporting an RTC to the police, the respondent/care giver would have been contacted by the police to seek further information to register a FIR.

Under-reporting of non-fatal RTIs by MTV users, pedestrians and cyclists highlighted in the population-based study is a concern, as these road users account for the majority of RTIs in India.1619 As policy formulation and road safety interventions are based on information available from police records, it is clear that application of these data available to the police is seriously limited for projecting the true nature of RTIs in India, as they are unlikely to be focused on the road users that need attention. Pedestrian injuries have been under-reported in linkage studies between hospital records/trauma registries with police crash data in Australia and UK,62627 and similarly for cyclist injuries in the Netherlands.8 Under-reporting of RTIs in children has been documented from Japan, UK, and the Netherlands.458 We did not find any significant association of age and sex with under-reporting of RTIs in this urban population.

If a fatal RTI is considered to be death within 30 days of the RTC, all fatal RTIs in the hospital-based study were reported to the police; however, if the definition is not limited, 2% of fatal RTIs were not reported. Under-reporting of 5% deaths by police compared with hospital records has been previously documented in Bangalore in India.28 The population-based study suggested 22% under-reporting of fatal RTI. All but one case was not reported because the RTC was considered to be due to a mistake made by the deceased, highlighting the need for better information sharing with the public about the importance of reporting fatal RTIs to the police. This under-reporting rate is lower than in Pakistan: the death rate in Karachi was underestimated in the police records by 179%.7

These data from population-based and hospital-based samples of the same population highlights the similarities and dissimilarities, and together provide a more comprehensive understanding of the complex process of RTI reporting to the police. It also highlights the difference in the magnitude of underestimation when only hospital-based data are considered for comparison with the police data. The urgent need to improve road safety information in India is recognized in the recent draft National Road Safety Policy, which includes recommendations to improve reporting of crashes to the police and linkages between the police and vehicle and driver registration databases.29 Our studies suggest that, in addition, improvements in the reporting system from the emergency departments and inpatient admission in hospitals for all fatal and non-fatal RTIs to the police and increasing awareness of the public about the need to report all fatal RTIs to the police irrespective of the circumstances of the RTC are needed to help to improve RTI estimates in India.