Rune Elvik

Incomplete accident reporting: Meta-analysis of studies made in 13 countries

A meta-analysis of studies of road accident reporting in official accident statistics made in 13 countries is described here. A rigorous comparison of reporting levels between countries is difficult because of differences in the definitions of reportable accidents, reporting levels, and data sources used to assess reporting levels. 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 percent for fatal injuries according to the 30-day rule, 70 percent for serious injuries (admitted to hospital), 25 percent for slight injuries (treated as outpatients), and 10 percent for very slight injuries (treated outside hospitals). Reporting levels vary substantially among countries, ranging from 21 to 88 percent for hospital-treated injuries. Reporting is highest for car occupants and lowest for cyclists. In particular, single-vehicle bicycle accidents are very rarely reported in official road accident statistics.

Area-wide urban traffic calming schemes: a meta-analysis of safety effects

This paper presents a meta-analysis of 33 studies that have evaluated the effects on road safety of area-wide urban traffic calming schemes. Area-wide urban traffic calming schemes are typically implemented in residential areas in towns in order to reduce the environmental and safety problems caused by road traffic. A hierarchical road system is established and through traffic is removed from residential streets by means of, for example, street closures or one-way systems. Speed reducing devices are often installed in residential streets. Main roads are improved in order to carry a larger traffic volume without additional delays or more accidents. The meta-analysis shows that area-wide urban traffic calming schemes on the average reduce the number of injury accidents by about 15%. The largest reduction in the number of accidents is found for residential streets (about 25%), a somewhat smaller reduction is found for main roads (about 10%). Similar reductions are found in the number of property damage only accidents. The results of evaluation studies are robust with respect to study design. There is no evidence of publication bias in evaluation studies. Study findings are found to have high external validity.

Risk of road accident associated with the use of drugs: A systematic review and meta-analysis of evidence from epidemiological studies

This paper is a corrigendum to a previously published paper where errors were detected. The errors have been corrected in this paper. The paper is otherwise identical to the previously published paper. A systematic review and meta-analysis of studies that have assessed the risk of accident associated with the use of drugs when driving is presented. The meta-analysis included 66 studies containing a total of 264 estimates of the effects on accident risk of using illicit or prescribed drugs when driving. Summary estimates of the odds ratio of accident involvement are presented for amphetamines, analgesics, anti-asthmatics, anti-depressives, anti-histamines, benzodiazepines, cannabis, cocaine, opiates, penicillin and zopiclone (a sleeping pill). For most of the drugs, small or moderate increases in accident risk associated with the use of the drugs were found. Information about whether the drugs were actually used while driving and about the doses used was often imprecise. Most studies that have evaluated the presence of a dose-response relationship between the dose of drugs taken and the effects on accident risk confirm the existence of a dose-response relationship. Use of drugs while driving tends to have a larger effect on the risk of fatal and serious injury accidents than on the risk of less serious accidents (usually property-damage-only accidents). The quality of the studies that have assessed risk varied greatly. There was a tendency for the estimated effects of drug use on accident risk to be smaller in well-controlled studies than in poorly controlled studies. Evidence of publication bias was found for some drugs. The associations found cannot be interpreted as causal relationships, principally because most studies do not control very well for potentially confounding factors.

Effects on Accidents of Automatic Speed Enforcement in Norway

Automatic speed enforcement by means of photo radar was introduced in Norway in 1988. The results of a before-and-after study of the effects of automatic speed enforcement on accidents are reported in this paper. The study controlled for general trends in the number of accidents and regression to the mean. A statistically significant reduction of 20 percent in the number of injury accidents was found. The number of property-damage-only accidents was reduced by 12 percent. This change was not statistically significant at the 5 percent level. The effect of automatic speed enforcement on the number of injury accidents varied according to the level of conformance with official warrants for its use. The warrants refer to accident rate (accidents per vehicle kilometer) and accident density (accidents per kilometer of road). A decline of 26 percent in injury accidents was found on road sections conforming with both warrants. On road sections not conforming with any of the warrants, injury accidents declined by 5 percent. The results of this study confirm the results of previous studies of the effects of automatic speed enforcement on accidents.

Publication bias and time-trend bias in meta-analysis of bicycle helmet efficacy: A re-analysis of Attewell, Glase and McFadden, 2001

This paper shows that the meta-analysis of bicycle helmet efficacy reported by Attewell, Glase, and McFadden (Accident Analysis and Prevention 2001, 345-352) was influenced by publication bias and time-trend bias that was not controlled for. As a result, the analysis reported inflated estimates of the effects of bicycle helmets. This paper presents a re-analysis of the study. The re-analysis included: (1) detecting and adjusting for publication bias by means of the trim-and-fill method; (2) ensuring the inclusion of all published studies by means of continuity corrections of estimates of effect rely on zero counts; (3) detecting and trying to account for a time-trend bias in estimates of the effects of bicycle helmets; (4) updating the study by including recently published studies evaluating the effects of bicycle helmets. The re-analysis shows smaller safety benefits associated with the use of bicycle helmets than the original study.

Why some road safety problems are more difficult to solve than others

Some road safety problems have persisted for a long time in nearly all motorised countries, suggesting that they are not easily solved. This paper documents the persistence over time of five such problems: the high risk of accidents involving young drivers; the high risk of injury run by unprotected road users; risks attributable to incompatibility between different types of vehicles and groups of road users; differences in risk between different types of traffic environment and speeding. A taxonomy of road safety problems is developed in order to identify characteristics of problems that can make them difficult to solve. It is argued that if a problem is not perceived as a problem, is attributable to a misguided confidence in road user rationality, involves social dilemmas, or is closely related to the physics of impacts then it is likely to be difficult to solve. Problems to which biological factors contribute are also likely to be difficult to solve. The characteristics that can make a problem difficult to solve are to some extent present for all the five problems shown to be persistent in this paper.

The safety value of guardrails and crash cushions: A meta-analysis of evidence from evaluation studies

Evidence from 32 studies that have evaluated the safety effects of median barriers, guardrails along the edge of the road, and crash cushions (impact attenuators) is summarized by means of a meta-analysis. Two hundred and thirty-two (232) estimates of safety effects are included in the meta-analysis. The presence of publication bias is tested by means of the funnel graph method. For most subsets of the data, no evidence of publication bias is found. Weighted mean estimates of safety effects are computed by means of the logodds method. Median barriers are found to increase accident rate, but reduce accident severity. Guardrails and crash cushions are found to reduce both accident rate and accident severity. The effects of guardrails and crash cushions on accident rate have been less extensively studied than the effects on accident severity. Current estimates of the effects on accident rate are highly uncertain because of methodological shortcomings of available studies. The effects of guardrails on accident severity are found to be quite robust with respect to study design and the number of confounding variables controlled in each study. In general, random variation in the number of accidents is the most important source of variation in study results.

The predictive validity of empirical Bayes estimates of road safety

This paper examines the predictive validity of empirical Bayes (EB) estimates of road safety. The predictive performance of EB-estimates was tested by applying five versions of EB-estimates of road safety: (1) A simple estimate derived from the empirical distribution of accidents in a population of sites, by which the number of accidents predicted for period 2 for sites that recorded k accidents in period 1 equals the number of accidents for sites that recorded k+1 accidents in period 1. (2) Estimates derived from the parameters of a negative binomial distribution fitted to an empirical distribution of accidents in a population of sites by means of the method of moments. (3) Estimates derived from the parameters of a negative binomial distribution fitted to an empirical distribution of accidents in a population of sites by means of the maximum likelihood technique. (4) Estimates derived by combining the predictions of an accident prediction model and the recorded number of accidents for a site. (5) Estimates derived by combining the predictions of a different version of an accident prediction model and the recorded number of accidents for a site. All versions of EB-estimates are compared to the traditional, naïve assumption of treating the recorded number of accidents as an unbiased estimate of the expected number of accidents. To test the predictive performance of EB-estimates, data for two periods was used. EB-estimates based on data for the first period were treated as predictions of the number of accidents in the second period for road sections that had 0, 1, 2, etc., accidents in the first period, the idea being that the more accurate the prediction, the more accurate the result of a before-and-after study. All versions of EB-estimates were found to give considerably more correct predictions of the number of accidents in the second period than relying on the count of accidents in the first period as a prediction of the count in the second period. Smaller prediction errors were associated with predictions based on accident prediction models than predictions not based on such models.

The importance of confounding in observational before-and-after studies of road safety measures

This paper discusses the importance of confounding in observational before-and-after studies of road safety measures. The importance of the approach taken to controlling for confounding factors is shown by means of examples. It is shown that the size of the effect on accidents attributed to a road safety measure can be profoundly affected by which confounding factors are controlled for in an evaluation study, and the way this is done. Simple before-and-after studies, not controlling for any confounding factors should never be trusted and are likely to overstate the effects of road safety measures.

A re-parameterisation of the Power Model of the relationship between the speed of traffic and the number of accidents and accident victims

This paper presents a re-analysis of the Power Model of the relationship between the mean speed of traffic and road safety. Past evaluations of the model, most recently in 2009, have broadly speaking supported it. However, the most recent evaluation of the model indicated that the relationship between speed and road safety depends not only on the relative change in speed, as suggested by the Power Model, but also on initial speed. This implies that the exponent describing, for example, a 25% reduction in speed will not be the same when speed changes from 100km/h to 75km/h as it will when speed changes from 20km/h to 15km/h. This paper reports an analysis leading to a re-parameterisation of the Power Model in terms of continuously varying exponents which depend on initial speed. The re-parameterisation was accomplished by fitting exponential functions to data points in which changes in speed and accidents were sorted in groups of 10km/h according to initial speed, starting with data points referring to the highest initial speeds. The exponential functions fitted the data extremely well and imply that the effect on accidents of a given relative change in speed is largest when initial speed is highest.