Studies about underreporting

Below follows a list of studies on underreporting of accident, put in alphabetical order according to the name of the first author. Each reference is followed with a short description of main contents of the study.

Agran PF, Castillo DN, Winn DG (1990). Limitations of data compiled from police reports on pediatric pedestrian and bicycle motor vehicle events. Accident Analysis and Prevention, Vol. 22, No. 4, pp 361-370.

A U.S. study where police reports are compared to a hospital monitoring system for children aged below 15. Under-reporting by police is conservatively estimated at 20% for pedestrians and 10% for cyclists.

Aptel I, Salmi LR, Masson F, Bourde A, Henrion G, and P Erny (1999). Road accident statistics: discrepancies between police and hospital data in a French island. Accident Analysis & Prevention, Vol. 31, No. 1/2, pp. 101-108.

In La Reunion, a French overseas dependency, an epidemiological study of injuries leading to hospitalisation or deaths has been performed from June 1993 to June 1994. The comparison between hospital data and police records showed that only 37.3% of non-fatally traffic-injured in-patients were recorded by the police.

Cryer PC, Westrup S, Cook AC, Ashwell V, Bridger P and Clarke C (2001). Investigation of bias after data linkage of hospital admissions data to police road traffic crash reports. Injury Prevention, Vol. 7, pp. 234-241.

This British research suggest that investigators could be misinformed if they base their analysis solely on police road traffic accident (RTA) data, and that information derived from a linked database to hospital admissions is less biased than that from police RTA data alone. A national linked dataset of road traffic crash data should be produced from hospital admissions and police RTA data for use by policymakers, planners and practitioners.

Elvik, R, Mysen, AB (1999). Incomplete accident reporting: meta-analysis of studies made in 13 countries. Transportation Research Record. 1999, 1665:133-140

A meta-analysis of studies of road accident reporting in official accident statistics is described. Based on 49 studies in 13 countries, it is concluded that reporting of injuries in official accident statistics is incomplete at all levels of injury severity. In rounded values, the mean reporting level in the countries included was found to be 95% for fatal injuries according to the 30-day rule, 70% for serious injuries (admitted to hospital), 25% for slight injuries (treated as outpatients), and 10% for very slight injuries ( treated outside hospitals). Reporting levels vary substantially among countries, ranging from 21 to 88% for hospital-treated injuries. Reporting is highest for car occupants and lowest for cyclists.

Ibrahim K and Silcock DT (1992). The Accuracy Of Accident Data. Traffic Engineering and Control, Vol. 33, No. 9, pp. 492-7.

The purpose of this British study is to review the extent of effort and time devoted by highway authorities to the checking of accident data. The study also investigates the different stages of validation of accident data which are carried out and the percentage of records with at least one error at each stage of data validation. Moreover, the percentage of under-reporting for various different types of accidents is examined.

International Road Traffic and Accident Database (1994), Underreporting of road traffic accidents recorded by the police, at the international level. IRTAD-report published by the Public Roads Administration, Norway.

This is a Norwegian study which finds that the under-reporting of traffic accidents is considerable and depends very much on the type of accident. The study indicates that under-reporting can vary from one country to another while the most serious accidents are nevertheless reported almost fully to the police.

Jacobs G, Aeron-Thomas A and A Astrop (2000). Estimating global road fatalities. TRL report 445. DFID.

This British study estimates that in 1999 between 750,000 and 880,000 people died from road crashes and that the majority of these deaths occurred in developing and transitional nations (85%). Estimates suggest that 23-34 million people are injured worldwide in road crashes - a value almost twice that previously estimated. The extent and impact of under-reporting, especially in developing countries, is highlighted with documented examples.

James HF (1991). Under-reporting of accidents. Traffic engineering and control, Vol. 32, No. 12, pp. 573-80, 583.

This British paper summarises the results of a variety of under-reporting studies. The factors affecting whether an accident was reported are identified and their effects discussed. Two postal surveys of motorcyclists and car drivers are analysed to examine the factors affecting reporting in more detail. Estimates of the total number of accidents not included in official figures for Great Britain are calculated and the national data adjusted for under-reporting. Additional casualty costs are also estimated.

Jarvis SN, Lowe PJ, Levene S, Cormack RM (2000). Children are not goldfish - mark/recapture techniques and their application to injury data. Injury Prevention, Vol. 6, pp. 46-50.

A British study. This paper seeks to illustrate serious drawbacks in the use of the mark/recapture technique when applied to injuries. The results indicate there were significant variations in matching proportions by class of accident (pedestrian v passenger v cyclist) and that recapture was influenced by the age of child.

Keigan M, Broughton J and Tunbridge R J (1999). Linkage of STATS19 and Scottish hospital in-patient data — analyses for 1980–1995. TRL Report 420

This report analyses the enhanced casualty information for the years 1980-95. Its aim is to provide an overview of trends in the clinical data over these sixteen years, and to give some insight into the scope for more detailed analyses. Various clear trends are identified, and some clear differences in the distribution of the clinical details are reported. The linkage used the computer program described by Stone (1984).

Lopez DG, Rosman DL, Jelinek GA, Wilkes GJ, Sprivulis PC (1999). Complementing police road-crash records with trauma registry data – an initial evaluation. Accident Analysis and Prevention, Vol. 32, pp. 771-777.

This Australian paper examines the consistency of hospital and police reporting of outcomes of road traffic crashes in Western Australia. Linkage rates were associated with gender, injury severity and the number of vehicles involved. Within the road user category, pedestrians were least likely to link. Of the linked records, police classification of injury severity was correct in 78% of cases.

Maas, MW (1982). De politieregistratie van verkeersgewonden in ziekenhuizen (police recording of road accident in-patients). SWOV, No. R-82-34, 45p.

A Dutch study which found during the years 1977-1979 the extent of under-reporting was constant and for the year 1979 the police data had a coverage of 83% of all road accident in-patients. Police data are therefore reliable for time series and for period studies of most of transport/age group combination.

Mills PJ (1989). Pedal cycle accidents - A hospital based study. TRRL Research Report RR220.

This report describes a one year hospital based British study undertaken to identify the main factors and the level of under-reporting of pedal cycle accidents, and to investigate these accidents in depth. Under-reporting rates are calculated for slight, serious and fatal accidents which are used to estimate the total number and actual cost of pedal cycle accidents.

Morrison A and Stone DH (2000). Capture-recapture: a useful methodological tool for counting traffic related injuries? Injury Prevention, Vol. 6, No. 4, pp. 299-304.

A British study. This paper estimates the completeness of official data sources on traffic related injuries (TRIs) by using the capture-recapture technique and to calculate an ascertainment corrected number of fatal and serious TRIs among Scottish young people aged 15-24 years. The appropriateness of the approach in this context is also assessed.

Nakahara S, Wakai S (2001). Underreporting of traffic injuries involving children in Japan. Injury Prevention, Vol. 7, No. 3, pp. 242-244.

A Japanese study seeking to clarify the magnitude of under-reporting of police data. The results reveal significant under-reporting by police of child vehicle occupant injuries. The true incidence of these injuries in preschoolers was twice as high as that provided by official police reports.

Rosman DL (1995). The Feasibility of linking hospital and police road crash casualty records without names. Accident Analysis and Prevention, Vol. 28, No. 2, pp. 271-274.

This Australian research re-matches the hospital admission and police records from the Western Australian Road Injury Database. The process shows that 90% of original links can be identified when a phonetic code of the family name of the casualty is used alongside age, sex, road user type and crash date. When names, or phonetic name codes, are not applied only about 50% of the original links are found.

Rosman DL (2001). The Western Australian Road Injury Database (1987-1996): ten years of linked police, hospital and death records of road crashes and injuries. Accident Analysis & Prevention, Vol. 33, pp. 81-88.

The Western Australian Road Injury Database links crash details from reports to police with the details of injuries to casualties contained in hospital and death records. This Australian article demonstrates the power of a linked system to answer complex research questions related to outcome and under-reporting.

Silke, E (1989). Invisible victims - counting the casualties of official reckoning. Surveyor, Vol. 171, No. 5038, pp. 8-9.

This British article discusses the possibility that the most vulnerable road users may be the least visible, to both statisticians and motorists. He examines some gaps in the existing accident figures, and considers whether a favoured engineering approach may be threatening rather than protecting cyclists and pedestrians.

Simpson, H F (1996). Comparison of hospital and police casualty data: a national study. TRL Report 173

This report summarises the results of a study to Iink casualties from a national sample of hospitals with police casualty records to provide information on the relative differences between reported and unreported casualties and the extent to which casualties are misclassified within national accident data. The linkage used the computer program described by Stone (1984).

Stewart, JR (1986). How many drinking drivers are there? American Association for Automotive Medicine (AAAM): Proceedings of the 30th annual conference, Montreal, Quebec, Canada, October 6-8, pages 33-43.

A U.S. study; estimators of drinking driver populations are developed by applying capture-recapture information. Several estimators are considered with different assumptions concerning the manner in which capture probabilities vary with time and over the population; application of these models to North Carolina driver history data yields drink driver population estimates.

Stone R. D (1984). Computer linkage of transport and health data TRRL Laboratory Report 1130

The report describes the development of a computer record matching algorithm using a 100 per cent sample of hospital in-patient records for Scotland for the year 1980. A total of 6093 unique matches were obtained, representing 70 per cent of health records. For these matches the casualty records in the national road accident data files were enhanced by length of stay in hospital together with clinical detail of inju& in ICD (International Classification of Diseases) and AIS (Abbreviated Injury Scale) forms. Some examples of uses of the matched data set are given.

Tercero F and Andersson R (2004). Measuring transport injuries in a developing country: an application of the capture-recapture method. Accident Analysis & Prevention, Vol. 36, No. 1, pp. 13-20.

The purposes of this joint Nicaraguan-Swedish study are to provide an estimation of the incidence of transport injuries in a defined local community in Nicaragua by using the capture-recapture method, and to compare results using this method when data at different levels of severity are utilized.”