The debate around the merging of road casualty and health data is a growing one. It is argued that there is a need for joined-up working between the Department for Transport and Department of Health as the DfT's statistics are not a true reflection of the percentage of road casualties.
There has been a question mark over the government's casualty statistics, as measured by the police data, collected by DfT as other data sources such as insurance claims and hospital admissions were either rising or not falling so sharply. Civil servants are currently trying to match the two data sets to get closer to the truth – and that must be the way forward.
Best practice
The question of best practice guidelines is also another consideration. It has been recommended that the DfT considers the production of new best practice guidance, with a focus on evaluation methods, for road-safety education and public information campaigns, which could be provided by an independent body along similar lines to the National Institute for Clinical Excellence, which provides guidance on health issues for the NHS.
While there is little debate over the number of road deaths, the drop in serious injuries over the past decade or so has not been mirrored in the, admittedly not directly comparable, health and insurance data. Other figures appear to contradict the trend of falling injuries. For example, annual hospital admissions for road injuries rose from about 34,000 to about 40,000 from 1995 to 2007, while the police data showed the number of people seriously injured falling from nearly 40,000 to under 25,000 over the period. Compensation claims by the government for NHS treatment costs after road accidents rose 37% over five years to 537,000 in 2007-08.
Now the practice of relying on one data set in the face of conflicting others has been highlighted – something that the Statistics Authority and MPs are picking up on – pressure is increasing on the DfT to improve data. This means that the government could be forced to review its approach, changing its narrow definitions and simplistic claims of success.
"We have only the haziest idea of how many people are actually seriously injured on the roads. It could be 26,000, 80,000, or 220,000," says Nigel Hawkes, director of Straight Statistics, the charity formed to improve the use of statistics by politicians and businesses. Many newcomers to the subject find this uncertainty hard to believe. Surely, they say, ministers, safety campaigners, journalists and others must have something solid to go on? This may not be the case. For any well-meaning enthusiasts who are keen to work on improving road safety, there is often little or no evidence to justify their endeavours or data to monitor their success.
The last government was riding a wave of apparent success on the most important indicator – the number killed or seriously injured on the roads (KSI). A 40% fall in KSI from mid-90s levels was sought by 2010 and, just in advance of the last election, such a fall was announced.
Charities and industry bodies have done little to press home the need for good, reliable data, often seeming to prefer the world of partial information. Road Safety GB, for example, which helps local government road safety teams, has inconsistent messages at its core. The launch of one initiative to reduce urban speeds, said: "Two-thirds of all crashes in which people are killed or seriously injured happen in built-up areas where the speed limit is 30mph or less", while another of its briefing documents stated that "over half of all deaths on the roads occur on rural (non built-up) roads." A more definitive line would be of benefit to everyone, and yet where should action begin? The Parliamentary Advisory Council for Transport Safety, set up to advise on transport safety issues, barely hinted at the concerns surrounding the police-recorded data in its latest briefing on road safety issues, and The Highways Agency have also been accused of being largely unconcerned.
Lack of information
The problems faced in the UK are modest compared with many other countries – as we have at least some data and a relatively low incidence of serious injury on our roads – but our data is not adequate to understand the dynamics of road activity. It would be useful to understand, for instance: the shift in injuries from those in vehicles to those outside; the number of accidents happening to people going about their daily business, in contrast to boy racers having a thrill; and much more on drink- and drug-driving.
More data is becoming available on topics such as congestion and speeds and is being viewed holistically on web-based services. This will increase our chances of choosing the most effective road-safety policies, the right legislation, best publicity campaigns and children's education, especially if those working in the field push hard for improvements.
Accident hotspots Child casualties
Road-safety data is often presented within strictly confined geographical limits, typically those of a police-force area, almost as if the force boundary is the edge of a flat earth. This might make sense when considering engineering interventions with budgetary constraints or when looking at enforcement, but not when it comes to education and public interventions. People live in real communities that do not respect the neat, dotted lines on maps and they want to understand the safety situation in their area.
Road Safety Market Analysis and Segmentation Tools (Mast) was set up by the Thames Valley Safer Roads Partnership with support from the Department for Transport. It combines crash figures with socio-demographic analysis to help those working on road safety deliver intervention strategies to reduce casualties as efficiently as possible. The Mast data combines the postcode data in police accident reports, known as Stats19s, with socio-demographic information from Experian's Mosaic system.
The type of data offered by Mast is needed to stimulate joined-up thinking when planning interventions, as no campaign should be complete without a knowledge of what type of people are driving where. Do the crashes in an area involve locals? Do people who live in one area tend to have more crashes than in other parts of the country? The Mast data can be analysed by casualties, crashes or vehicles. The "risk" of an accident can be based on the population in an area or the miles driven.
Socio-demographic profiling has become increasingly popular in policy arenas in recent years. The merger of such profiles, which describe the types of people living in each postcode area alongside other data, has shed light on a wide range of issues and there is no reason why it should not also add to our understanding of road safety. A report last August from the Mast team revealed the profile of accidents involving children (above, showing chances of a child being injured).
The new model does require subscription fees but, as Mast's Richard Owen says, "the project aims to make some of the data freely available to anyone when funding allows". The system will always be subject to concerns about data quality – but at least the quality of debate will be raised.
Police data Reliability problems
The target mentality of the last Labour government was welcomed as it focused minds on delivery and improvements we hoped to see as a result of policy interventions, but it also led to what became known as gaming: those compiling the data were influenced, to a greater or lesser extent and sometimes subconsciously, to bend the data so they met their targets.
There is no proof that this happened with the police-collected Stats19 data (the name comes from the form the police fill in when reporting an accident), but the 40% drop in serious injuries over the past decade or so has not been mirrored in the – admittedly not directly comparable – health and insurance data. The number of accidents resulting in serious injury is now conceded by the government to be much higher than is shown in the Department for Transport's figures. The police figures demonstrate, in broad terms, about 2,000 road deaths a year, about 20,000 serious injuries and approximately 200,000 minor injuries. Official estimates now suggest that there are about 800,000 injuries a year.
There is little dispute over the number of road deaths. In any case, compared to the numbers injured, the figures are small. But cynics suggest that it would be easy for a police officer, aware of the desire to cut the number of "seriously injured", to mark down the casualty as a lesser grade of injury. At the margin, with unclear rules, such an action makes little difference to anyone but it does help to hit the target. The fall in the number of seriously injured could, of course, have been innocent, reflecting changing policing methods or reporting behaviours of the public.
The main statistics were reviewed by the UK Statistics Authority (UKSA), which issued a pretty damning set of conclusions: undercounting of the police data was acknowledged and definitions of injury types needed to be tightened and a business case was to be made for investment in better data. Some changes were made, others are promised, and the department now has to bring together all the relevant data on accidents in one report. The National Travel Survey has included questions on accidents since 2007.
An electronic system for recording accidents and incidents which is in the pipeline, called Crash, might help, according to Richard Alldritt of the UK Statistics Authority. Instead of licking their pencils and scribbling in their notebooks, police will enter details of accidents on a hand-held computer. Crash should be available to all forces next year, says the National Policing Improvement Agency.
