CHAPTER THREE RURAL ROAD ACCIDENT FACTORS
3.1 This chapter examines literature dealing with the main features of accidents on rural roads and focuses on some other factors which are particularly relevant to the rural situation.
3.2 The OECD (1999) suggests that much of the rural road safety problem can be categorised into three accident types:
- Single vehicle accidents (especially running off the road)
- Head-on collisions
- Junction accidents
3.3 The OECD (1999) also points to the three hazard factors which contribute to accidents:
- Driver behaviour
- Vehicle factors
- Road environment
3.4 In terms of the relative importance of these factors, Sabey and Staughton (1975), reporting on a detailed study of over 2,000 accidents between 1970 and 1974, showed that 95% of accidents involved a human factor with 28% involving a road factor and just 8.5% involved a vehicle factor. Whilst this gives an indication of the relative importance of these factors, Sabey (1980) cautions that
"It is too easy to conclude that all effort should be applied to influencing human behaviour directly, without taking into account the detailed circumstances, the multiplicity of factors which lead to accident occurrence, and the chances of success of measures applied."
3.5 For the purposes of this study, driver behaviour and road environment factors have been reviewed, whilst vehicle factors have not been examined. Additional factors, such as tourists/visitors and emergency service response times are also considered.
Driver behaviour factors
3.6 It is widely recognised that road user behaviour, and particularly driver and motorcycle rider behaviour, is a major factor in road accidents. The IHT point out that car drivers and motorcycle riders are most likely to be involved in accidents on rural roads and that their "skill, judgement, anticipation, state of mind and physical well-being" all contribute to the driver/rider's ability to avoid an accident ( IHT, 1999). The IHT also highlighted that speed, perceptual difficulties and drink driving are the most prevalent factors contributing to accidents.
Relationship between speed and accidents
3.7 In general, a reduction in mean speed is associated with a decrease in accident rate. An often quoted rule of thumb found by Finch et al (1994) states that a reduction of 1mph in mean speed leads to a 5% reduction in accident rate.
3.8 It is also the case that whilst the impact of speed on rural road safety is often the "subject of heated debate" ( OECD, 1999), high speed contributes to more severe accidents. On this basis, all countries have imposed speed limits based, to some extent, on the road characteristics.
3.9 It should be noted that a change in the speed limit without suitable enforcement will be unlikely to reduce speed by the full difference between limits (Taylor et al, 2000). A change from the national speed limit from 60mph to 40mph on a single carriageway road with no accompanying changes to the road character is unlikely to make the road appear urban.
3.10 Modern vehicles are capable of travelling at speeds far greater than the posted legal limits, and many drivers state that it is easy to exceed legal speed limits without realising it ( AA Foundation, 2001). As a counter-measure to this problem, in-vehicle technologies have been developed to warn drivers when they are exceeding the speed limit (e.g. see Carsten, 2001). Road design measures such as vehicle activated warning signs ( see para. 4.36) might also be an appropriate way to do this and these might help people who are motivated to keep within the speed limit to do so. However, their effect will probably be limited to a small distance downstream.
3.11 In a comprehensive study following on from the MASTER (Managing speeds of traffic on European Roads) Project (European Commission, 1998), Taylor et al (2002a) established relationships between speed and accidents on four relatively homogeneous groupings of rural single carriageway roads in England. The road groups are summarised in Table 3.1.
Table 3.1 - Road Groups in Taylor et al (2002a)
Roads which are very hilly, with a high bend density and low traffic speed. These are low quality roads.
Roads with high access density, above average bend density and below average traffic speed. These are lower than average quality roads.
Roads with a high junction density, but below average bend density and hilliness and above average traffic speed. These are higher than average quality roads.
Roads with a low density of bends, junctions and accesses and a high traffic speed. These are high quality roads.
3.12 Taylor et al
(2002a) found that accident frequencies in all road groups increased with mean speed to the power of approximately 2.5 - indicating that a 10% increase in mean speed results in a 26% increase in injury accidents. Two other factors were found to increase the frequency of injury accidents: density of sharp bends and the density of minor crossroad junctions. The effect of mean speed was found to be particularly large for junction accidents which suggested that there is substantial potential for accident reduction by reducing speeds at junctions. However, this has to be considered in the context of Scotland's lower than average number of accidents at rural junctions ( see para. 3.39
Speeding, excessive and inappropriate speed
3.13 In a recent study, Mosedale and Purdy (2004) examined contributory factor data, based on the system devised by Broughton et al (1998), for around 25% of all recorded injury accidents in Great Britain between 1999 and 2002, i.e. only data from those police forces that chose to participate in the trial of contributory factors.
3.14 Mosedale and Purdy (2004) looked at the occurrence of excessive speed as a contributory factor for accidents involving different vehicle types. They found that for all severities, all motorised vehicle types showed greater incidence of excessive speed on rural roads than on urban roads. Table 3.2 summarises the results. Overall, they found that excessive speed is a contributory factor in twice as many rural road accidents (18%) as urban road accidents (9%). It is not clear from the report what definitions of 'rural' and 'urban' were used.
Table 3.2 - Percentage of accidents on urban and rural roads with excessive speed as a contributory factor 1999-2002 (Mosedale and Purdy, 2004)
3.15 Data for Great Britain show that whilst nearly half of all vehicles on non built-up dual carriageways are exceeding the speed limit, the proportion of vehicles speeding on non built-up single carriageways is much less at 8% (DfT, 2003). The same data show that the percentage of vehicles exceeding the speed limit by 10mph or more has, in recent years, reduced on single carriageways but increased on dual carriageways. Articulated vehicles (which have lower speed limits on both dual and single carriageways) are particularly likely to exceed the speed limit on both dual and single carriageways. The data also show that motorcyclists are particularly likely to exceed the speed limit.
3.16 Webster and Wells (2000) provide a synthesis of research into the characteristics of speeders. The main findings of relevance were:
- Whilst the majority of drivers admit to speeding at some times, more speeders are young males in non-manual occupations.
- Company car drivers, drivers in large cars and drivers with higher annual mileages are more likely to drive faster.
- Speed choice (relative to the mean) seems to be consistent in all situations (ie drivers who drive fast in urban settings tend to drive fast in rural settings).
- Drivers may justify speeding by assuming that speed limits are unrealistic and many drivers fail to spot speed limit signs and are unable to 'read' the speed limit from the road design (ie roads are often not 'self-explaining').
- A large body of research exists which associates speeding behaviour with accidents. In particular, both speed over the speed limit and speed over the norm for any given road are critical factors in determining accident risk.
3.17 A recent Scottish Executive research project on speeding drivers (Stradling et al, 2003) contains several interesting findings. It showed, amongst other things, that levels of excess and excessive speed on faster rural roads have increased over the last decade. It also showed that, on faster roads, male drivers were more likely to choose a speed in excess of the limit than female drivers. This is in line with many previous studies which have concluded that male drivers are more likely to speed than females.
Alcohol and drugs
3.18 The impairment effects of alcohol on driving are well known and even at low blood concentrations, alcohol can have an impact on driver performance (Moskowitz and Fiorentino, 2000). However, the OECD suggested that there were no definitive studies to show that drink driving was more predominant in or specific to rural areas ( OECD, 1999).
3.19 Research carried out for the Scottish Executive found no concrete evidence of a real difference in drink driving prevalence between urban and rural areas (Anderson and Ingram, 2001). However, the same study did find evidence that there was a public perception that drink driving was more of a problem in rural areas.
3.20 Tunbridge et al (2001) reported on a study looking at the incidence of alcohol and other drugs in road accident fatalities. They found that alcohol was present in 31% of casualties and at least one medicinal or illicit drug was present in 24% of the sample. The study looked at regional variations but not explicitly at urban/rural splits.
3.21 Ingram et al (2000) carried out a survey to estimate the prevalence of drug-driving in Scotland. The results showed that 9% of respondents reported having driven under the influence of drugs and 5% had done so in the last 12 months. No breakdown between urban and rural areas is given due to the relatively small number of drug drivers detected in the survey.
3.22 National statistics ( ISD Scotland, 2002) show that whilst drug misuse is more prevalent in large urban areas, it is not confined to such areas. It is therefore reasonable to expect that a degree of drug-driving takes place outside urban areas.
Fatigue and Sleep-related Crashes
3.23 Driver fatigue and sleepiness is now recognised as being a significant accident causation factor. Crashes involving the driver falling asleep are rare on urban roads because driving conditions are relatively stimulating. These crashes typically occur on rural roads, especially motorways, where the driving task can be monotonous and undemanding (Horne and Reyner, 2001).
3.24 Identifying accidents caused by fatigue or sleepiness is a difficult task, principally because the driver involved either dies or will not admit to having fallen asleep. The best method devised to date uses circumstantial evidence from detailed police accident files to classify accidents as possibly or probably 'sleep-related' based on a series of criteria (Horne and Reyner 1995). The aim in this method is to exclude all other likely causes to arrive at those crashes where sleep is the most likely factor. Horne and Reyner (1995) assert that this method is likely to under-report sleep-related crashes because it excludes crashes which may be sleep-related if, for example, bad weather is also present. However, as the method relies on very detailed police records being available, it is likely to be less robust when applied to slight injury or damage only crashes.
3.25 Researchers have used this method to examine accidents on several UK motorway sections and rural dual and single carriageway roads (Reyner et al, 2001 and Flatley et al, 2004). These studies have estimated the proportions of crashes which are sleep-related to range from 3% on a rural 'B' road to 30% on a section of English motorway. The recent study (Flatley et al, 2004) highlights that the problem is not confined to motorways and estimates that 16% of all accidents on rural non-motorways could be sleep-related.
3.26 Sleep-related crashes tend to occur at night and to be more severe due to higher impact speeds. They tend to involve drivers who are at work or driving company cars (Maycock, 1996, 1997) and younger male drivers (under-30) are particularly at risk (Reyner et al, 2001).
3.27 Flatley et al (2003) found that the occurrence of sleep-related crashes was related to traffic density although they found different relationships on motorways (where more sleep-related crashes occurred when traffic density was lower) and on 'A' and 'B' roads (where more sleep-related crashes occurred when traffic density was higher). This might suggest that on lower trafficked Scottish 'A' and 'B' roads, the problems associated with sleep-related crashes may be less than elsewhere.
3.28 Research has shown that the onset of fatigue is exacerbated by the time on task (ie the length of time actually driving), sleep deprivation and circadian rhythm (time of day effect). Drivers are therefore advised to avoid long journeys if tired and in the early hours, to take a break of at least 15 minutes every 2 hours and stop in a safe place if sleepy. Based upon research by Horne and Reyner (1997), drivers are also advised that a caffeine drink and a short nap are the most effective countermeasures to fatigue although the benefits may be relatively short-lived (DfT, 2004b).
3.29 Obviously, for drivers to follow the above advice, there needs to be adequate provision of convenient stopping places such as lay-bys, rest areas and service facilities which are available 24 hours. This may be particularly difficult to achieve in the more rural parts of Scotland as traffic volumes would not sustain commercially viable service areas. Also, on one strategic trunk road in Scotland, existing planning policies have restricted the provision of roadside services although it is understood that this policy is currently under review ( SODD, 1996 and Scottish Executive, 2004b).
3.30 Jackson (2004) highlights the particular problems faced by professional drivers and gives advice on how awareness of fatigue and sleepiness can be raised within industry. In particular, Jackson (2004) discusses the 'Awake' programme which aims to minimise fatigue, raise awareness, reduce accident risk, improve performance and improve quality of life. It seeks to achieve this by addressing the main reasons why drivers continue to drive when they are tired. Jackson (2004) describes these reasons as being: driver tiredness is not taken seriously enough; drivers overestimate their capabilities; drivers have a poor knowledge on when to act on their sleepiness; and drivers have mixed knowledge of effective countermeasures.
Distraction and inattention
3.31 Several research studies have examined the detrimental effects of driver distraction on performance. These distractions can be either external to the vehicle (for example an advertising sign) or internal to the vehicle (for example tuning the radio or using a mobile phone).
3.32 Brown et al (1969) first identified the potential driver impairment effects of distraction through phone conversations. With the recent widespread availability and use of mobile phones, this issue has become more prominent.
3.33 Burns et al (2002) reported on a simulator study into the effects of mobile phone use on driver performance. The study compared the effects of having phone conversations to the effects of alcohol consumption and it was found that certain aspects of driving are impaired more by using a phone (whether hands-free or not) than by having a blood alcohol level at the legal limit (80mg/100ml). It therefore concluded that driving behaviour while talking on a phone is not only worse than normal driving, but can be described as dangerous.
3.34 The use of a hand-held mobile phone whilst driving was made an offence in the UK in December 2003.
3.35 Wallace (2003) carried out an extensive literature review on driver distraction with the emphasis on external-to-vehicle distractions and particularly advertising billboards. His study also included an investigation of potential accidents involving external-to-vehicle distraction in Central Scotland. The study found less than 1% of accidents in the database were attributable to external-to-vehicle distraction although Wallace suggests that distraction is likely to be under-represented as a cause in accident databases. In addition, he found very little data on billboard distraction pertaining to the British or even European environment. In contrast to this, it is understood that some countries e.g. France, deliberately erect features of this type on motorways with the aim of keeping drivers more alert.
Seat Belt use
3.36 Several studies have shown the benefits of seat belt use in terms of casualty and severity reduction (eg Tunbridge, 1989). In addition, Williams et al (1991) found a 19.5% reduction in the number of road deaths in Scotland after the introduction of compulsory front seat belt usage in 1983.
3.37 Studies have shown that seat belt wearing rates vary by time, location and car occupant demographics. For example, Broughton (2003) reported on seatbelt wearing rates in England and found that wearing rates were higher for women than for men, higher on non-built-up roads, and rise with increasing age with the highest rates amongst older drivers. Broughton (2003) also reports that seat belt wearing rates are lower for rear seat passengers, especially amongst young adults but again were higher on high speed rural roads than urban roads.
3.38 Recent research into seat belt wearing rates in Scotland (Burns et al, 2003) suggested that the following groups should be targeted in campaigns: all rear seat passengers, front seat male passengers and young male drivers. As in Broughton (2003), Burns et al (2003) found considerable variation in wearing rates between geographical areas with the more densely populated areas having lower rates. They also found evidence to suggest that the significance of age and gender varied by geographical location with some areas showing marked variation between male and female compliance. However, they found no significant difference between wearing rates on roads with 30mph limits and those with higher limits.
Road environment factors
3.39 Research has shown Scotland's higher than average accident severity rates for rural areas can be partly accounted for by differences in the road network between Scotland and the rest of Great Britain ( MVA, 1997). The three most important factors identified were: the estimated traffic flow at the time of the accident; junction type and; road type. In particular, Scotland had a greater proportion of accidents on single carriageways, on links (ie not at junctions) and at lower flows than the rest of Great Britain.
3.40 Scotland's road network consists mainly of single carriageways - over 97% of the public road length is single carriageway and a large proportion of this will be in non built-up areas. For example, 80% of local authority A roads, 85% of B roads and 88% of C roads have speed limits greater than 40mph. Roads with speed limits of 40mph or less (ie built-up roads) make up just one-third (33%) of the total road length and over three-quarters (77%) of these are unclassified (Scottish Executive, 2004c). Table 3.3 shows that Scotland's road network has more than double the proportion of rural A class single carriageway roads as England and double the proportion of B roads (which are almost all single carriageway) as England.
Table 3.3 - Proportions of road network by road class/type in Great Britain (from Transport Statistics Great Britain, 2004)
3.41 The high proportion of rural single carriageways in Scotland is important because all three of the main accident categories identified by the OECD (1999) are more common on single carriageway roads.
3.42 The EuroRAP project, which is discussed in detail later, points out that single carriageways, at-grade junctions and low traffic flow roads are all features associated with high accident rates ( EuroRAP, 2002).
Dual and single carriageway A roads
3.43 Two main studies in the UK have examined accidents on rural 'A' class single and dual carriageway roads.
3.44 Walmsley et al (1998a, 1998b) and Walmsley and Summersgill (1998) developed detailed models for schemes on modern rural single and dual carriageway trunk roads in England that were subject to the national speed limit. The models for single carriageways distinguish between wide (10m) and standard (7.3m) roads. They take account of traffic flow, length, the numbers of major junctions and the numbers and types of minor junctions, as well as bendiness, hilliness, the numbers of accesses, presence or absence of a hard strip and (for dual carriageways) a median safety barrier.
3.45 They concluded that modern roads are safer than older roads due to improvements in road design. In particular this is due to the modern practice of providing hard-strips, safety barriers and the construction of fewer, better designed junctions.
3.46 In Cambridgeshire, Hughes et al (1996, 1997) reported on similar studies for rural single and dual carriageways. The main findings of these studies along with those mentioned in paragraph 3.44 are discussed later.
Rural Single Carriageway Roads
3.47 Barker et al (1998) examined all the reported injury accidents occurring in 1994-95 on rural single carriageway roads ( RSCRs) in Great Britain. This updated a previous study of 1988-89 data with which comparisons were made. A few of the more notable findings were:
- Most accidents occurred on A roads (53%); on 2-lane roads (90%); in 60 mph speed limits (96%); and away from junctions (63%). They mostly involved 2 vehicles and no pedestrians (55%); occurred during daylight (71%); and occurred in fine weather (53%).
- Compared with accidents in built-up areas, those on RSCRs were more severe; only about a half as likely to be at a junction; almost one sixth as likely to involve a pedestrian and a quarter as likely to involve a pedal cycle; but they were three times as likely to involve a single vehicle (with no pedestrian).
- The accident severity ratio (ratio of KSI accidents to all injury accidents) for RSCRs was higher on the more major, wider roads, away from junctions, in the dark and in fine weather. It was also greater when male drivers, the oldest drivers, two-wheeled vehicles, public service vehicles, or heavy goods vehicles were involved.
- The most frequently involved vehicle manoeuvre was 'going ahead - other' (44% of accident-involved vehicles at junctions and 44% away from junctions). Away from junctions 'going ahead on a left/right hand bend' featured next most frequently (35%) while at junctions, 'turning right' did so (21%).
- Single-vehicle accidents accounted for almost one third of all accidents. They were more likely than other accidents to be associated with B/C roads, night-time, the youngest drivers and with 'going ahead on a bend'. Young drivers were also disproportionately associated with positive breath tests and with accidents in the dark.
- Accidents involving vehicles doing the faster manoeuvres ('going ahead', 'overtaking') were more likely than other accidents to involve young drivers, male drivers, TWMVs, a pedestrian, skidding, leaving the carriageway, hitting objects on or off the carriageway, and to be more severe. Accidents involving slower manoeuvres ('right-turns', 'stopping', 'waiting') were disproportionately associated with female and older drivers. 'Parked' and 'stopping' manoeuvres were disproportionately associated with PSV/ HGVs.
- 4% of all accidents involved a pedestrian. More than half of these involved a single vehicle, 'going ahead', not at a junction.
- 8% of all accidents involved two (non-overtaking) vehicles travelling in opposite directions on bends, not at a junction.
- High-performance cars were disproportionately involved in non-junction accidents, in single-vehicle accidents and in overtaking accidents. Their accident involvement was particularly associated with male drivers and with drivers in the 25-39 years age group.
- Comparisons with the earlier study years of 1988/89 (Taylor and Barker, 1992) showed that accident characteristics were remarkably similar between the two periods.
3.48 EuroRAP, an international not-for-profit association formed by motoring organisations and road authorities throughout Europe aims to provide comparative safety ratings for European roads. To date, roads in Great Britain, Sweden, the Netherlands and Spain have been assessed and rated. The research has highlighted many aspects of road design and use which could be improved to reduce the number of deaths and serious injuries on European roads.
3.49 EuroRAP uses two protocols (Lynam et al, 2003):
- Measurement and mapping of the rate at which people are killed or seriously injured
- A standard road inspection for safety features known as the Road Protection Score ( RPS)
3.50 The risk mapping allows decisions to be taken on road improvement policies and also presents information for individual road users. It is hoped that by informing road users of the level of risk presented, they can make decisions not only on route choice but also to modify their behaviour to minimise risk. Lynam et al (2003) also point out that there must be a recognition that infrastructure changes cannot eliminate all risk and that road users must take a share in responsibility for a safe road system.
3.51 EuroRAP provides a clear indication of the effect of traffic flow on accident rates, especially on single carriageway roads. Table 3.4 shows the fatal and serious accident densities (ie accidents per kilometre) for single carriageway roads in the EuroRAP countries by traffic flow. It can be seen that as flow increases, the accident density generally increases.
Table 3.4 - Fatal and serious accident densities (accident per km) for single carriageway roads by flow group (from Lynam et al, 2003)
Flow ( AADT)
3.52 The latest EuroRAP data for Great Britain ( AA Foundation, 2004) highlights several roads in Scotland which are either 'medium to high risk' or 'high risk'. This latest data also highlighted the contribution that motorcyclists make to fatal and serious injuries on some roads and contained a separate analysis for roads with motorcyclist accidents removed.
3.53 The EuroRAP Road Protection Score ( RPS) indicates the extent to which road design protects the user in the event of an accident occurring (Lynam et al, 2003). The main RPS results show that:
- On many roads, but especially single carriageways, there is substantial scope to improve the potential for injury prevention.
- Many roads score poorly for run-off protection - the suggestion is that fatalities are likely to occur unless barriers or wide safety zones can be provided
- The lowest scoring roads score poorly for all three accident types (head-ons, single vehicle run-offs and junction accidents)
- Protection on single carriageways is limited by narrow safety zones, poor access provision (ie junctions) and by the lack of medians to "limit the interaction of opposing traffic streams". The authors point to median treatment in Sweden and the Netherlands as good examples.
3.54 At present, the RPS does not include the contribution that road design, signs and markings can make to influencing driver behaviour and largely measures the 'passive safety of the road'. However, the RPS demonstrates that an appropriate balance between speed and road design can produce high levels of protection on most road types (Lynam et al, 2003).
3.55 The effect of width on accident rate was reviewed in an unpublished report by TRL in 1992, based on research on single-carriageway roads in the US. In general, wider roads were found to have a lower accident risk (e.g. Zegeer et al, 1981). The number of lanes had a similar effect to road width, but the two variables are correlated. More recently, Hughes and Amis (1996) found that for rural single carriageways in Cambridgeshire, a 1m increase in carriageway width was associated with a 19% decrease in accidents. Walmsley and Summersgill (1998) found that there was a 22% lower accident risk on 10m wide single carriageway trunk roads compared with 7.3m wide roads, for the same level of flow.
3.56 In research into road appearance which involved showing pictures in a questionnaire survey (Highways Agency, 2002), carriageway narrowing was found to reduce mean estimated driving speeds by as much as 7mph in one location. Other research projects in which reported speed has been measured (e.g. Fildes et al., 1987), and in which actual vehicle speeds have been measured (e.g. Kolsrud, 1985; Vey and Ferreri, 1968; Yagar and Van Aerde, 1983) support this finding.
Horizontal and vertical alignment
3.57 Walmsley and Summersgill (1998) found that bendiness and hilliness had only a small effect on accident risk on rural trunk roads, with a higher accident risk on roads that had poor alignment, probably because the trunk roads on which the models were based had relatively low values of hilliness and bendiness. They concluded that the effect of adding an extra bend or hill would be to increase accidents by less than 1 per cent for a single-carriageway scheme and about 4% on a dual-carriageway.
3.58 By contrast, Barker et al (1998) found that 1/3 of non-junction accidents occurred on bends.
3.59 Shrewsbury and Sumner (1980) showed that accident rates on rural roads decreased with increasing horizontal radius. Hughes and Amis (1996) found that for rural single carriageways in Cambridgeshire, an increase in bendiness of 1 degree per km was associated with a 1% increase in accidents. Hughes et al (1997) found a similar effect for rural dual carriageways in Cambridgeshire. More recent work by Taylor et al (2002a) on a variety of rural road types suggests that more sharp bends per kilometre are associated with a higher accident risk ( see para 3.12). Sight distance is correlated with bendiness and has therefore not been considered separately by most authors.
3.60 Volume 6 of the Design Manual for Roads and Bridges gives advice and sets the geometric standards by which new trunk roads are designed. Advice note TA85/01 ( DMRB 6.1.3) points out that high speed differentials occur at crests and sags in the alignment and this can result in increased numbers of accidents particularly where visibility is restricted.
3.61 Gradient was found to have a mixed effect (Shrewsbury and Sumner, 1980). Steeper downhill gradients were found to be associated with higher accident rates, but steeper uphill gradients had much less effect.
3.62 Walmsley and Summersgill (1998) found that the presence of a hard strip on rural trunk roads was associated with a reduction in accident risk of 16% on dual-carriageways and 18% on single carriageways.
3.63 Hughes et al (1997) found that nearside kerbing was associated with a reduction in accident risk on rural dual carriageways, whereas Walmsley and Summersgill (1998) found the opposite.
3.64 On rural roads, verge width was found to be important in some early studies (Zegeer et al, 1981, McLean 1985), but not others.
3.65 Safety barriers are required in appropriate circumstances on trunk roads with a speed limit above 50mph ( IRRRS, 2002). However, this requirement takes no account of the level of traffic flow. Roads with low flow may also have lower risk, but a relatively high proportion of Scottish roads will have steep embankments close to the road and these will require protection.
Junctions and accesses
3.66 The trunk road accident models by Walmsley and Summersgill (1998) take into account the effect of major junctions (roundabouts or traffic signals) and three- or four-arm major/ minor junctions. Accidents were found to increase with the number of junctions per kilometre. Walmsley and Summersgill found that adding an extra access to a typical single-carriageway trunk road scheme would have only a small effect (less than 1 per cent). On dual carriageways, it was the number of offside accesses (that is, across the reservation) which were significant. Adding an extra access to a typical dual-carriageway scheme would increase accidents by 2 to 3 per cent.
3.67 Hughes et al (1997) examined several different factors associated with increased accident frequency at rural dual carriageway junctions. These included the number of vehicles entering and leaving the main road at grade-separated junctions; minor road traffic flow at T-junctions, vertical alignment issues, and issues associated with gaps in the central reserve. They found that increasing the distance between junctions, providing a wide verge on the off-side of slip roads, and/or increasing on-slip merge lengths decreased the accident frequency.
3.68 Hughes et al (1997) also found that older drivers had greater involvement in accidents involving a right turn or crossing the main carriageway at junctions.
3.69 In their complementary study of single carriageways, Hughes and Amis (1996) found a significant proportion of accidents at private accesses on some routes. These accidents tended to be more prevalent at 'business' accesses, tourism spots and Sunday market locations where traffic movements were higher.
3.70 Hughes and Amis (1996) also found that accident frequencies at rural T-junctions are influenced by major road traffic flow, minor road traffic flow and carriageway width.
Weather/climate and seasonality
3.71 Seasonality of accidents will largely be influenced by traffic flow and weather conditions. Summer flows will be swelled by tourists, leading to more accidents in summer ( see paragraph 3.98). However, winter conditions may be more severe, thereby increasing the accident rate for those who still use the roads although this will be off-set to some degree by drivers choosing not to travel in the worst conditions. Poor weather can also affect emergency response times and the cold may decrease the chances of survival.
3.72 Green (1980), in a study to examine the effects of darkness on accident rates, studied the number of accidents in the five working days before and after the Sundays in 1975, 1976 and 1977 when the clocks changed. The study examined six regions of Great Britain, including Scotland, separately and the data was confined to non built-up roads. Green (1980) found that in the evening period studied, the frequency of all injury accidents is about 50 per cent higher and of fatal and serious accidents about 100 per cent higher. Green (1980) also noted that "the changes appear to be consistent over the country".
3.73 The evidence that casualty rates, particularly fatal and serious injuries, are higher in darkness has led to several investigations of the potential road safety effects of adopting so called Single/Double Summer Time ( SDST). SDST would involve setting clocks to one hour ahead of Greenwich Mean Time ( GMT +1) from October to March and two hours ahead ( GMT +2) from March to October.
3.74 A recent study into the potential effects of adopting SDST (Broughton and Stone, 1998) found that the effects of darkness are greater for pedestrians than for vehicle occupants and greater for fatalities than non-fatalities. Overall, Broughton and Stone (1998) predicted that KSI casualty rates for the whole of Great Britain for the period 1991-94 would have been 0.8% lower had SDST been in place. The predicted reduction for Scotland was slightly lower at 0.7%. However, it should be noted that the separate analysis for Scotland was limited by sparse data - particularly in the morning. The data could not be disaggregated by severity, time of day or into pedestrian and vehicle occupants. Therefore, the effect of SDST on rural casualty rates in Scotland is not clear.
3.75 On the basis of the evidence, several road safety organisations support the adoption of SDST. For example, RoSPA has suggested that SDST be introduced on a trial basis for two to three years so that the effects can be directly measured ( RoSPA, 2003).
3.76 In 2003, there were 222 reported injury accidents in Scotland involving an animal other than a dog in the carriageway (Scottish Executive, 2004a); this represents approximately 1.5% of all reported injury accidents. However, there is no breakdown by animal type or by built-up or non built-up roads given in the national statistics.
3.77 Staines et al (2001) report on a study into road accidents involving deer in Scotland. They found, albeit from limited data, that deer-related incidents are less common in the south and south-east and more common in the Highlands and south-west Scotland. The highest proportion of deer accidents is recorded on trunk or other A class roads and their incidence peaks in May-June and September-November. The highest proportion of accidents occurs between 2000-2400hrs with additional peaks at dawn and dusk. Organisations consulted considered that deer accident rates had increased in recent years but many felt that there were other issues associated with road traffic accidents which were of higher priority.
3.78 In order to more accurately quantify the problem of deer-related crashes, the Deer Initiative has created a web based database ( www.deercollisions.co.uk ) to allow incidents to be directly reported. This site also contains advice to drivers to help avoid accidents with deer.
3.79 In a recent follow-up study, Putman et al (2004) suggest that road traffic accidents involving deer may be expected to increase due to the distribution and abundance of deer in Scotland combined with increasing traffic and higher speeds associated with road improvements. They point to research from Europe and the USA which supports this suggestion. The main focus of their study is into the costs and cost-effectiveness of various measures to reduce deer accidents.
3.80 Putman et al (2004) recommend that deer fencing is the most appropriate mitigation measure for motorways and high speed trunk roads. For more minor roads, or where deer fencing is not a feasible option, mitigation measures should be targeted at reducing driver speeds in areas of known high deer collision risk.
Vulnerable road users
3.81 It has long been recognised that young drivers are over-represented in accident statistics. The OECD highlighted the fact in its 1975 report ( OECD, 1975). In Scotland, young drivers/riders and passengers in the 16-22 age group have the highest killed and seriously injured casualty rate per head of population (Scottish Executive, 2004a) out of all age groups. However, the published statistics give no indication of the split between urban and rural areas.
3.82 As mentioned in paragraph 3.47, Barker et al (1998) found that on rural single carriageways, accidents involving 'faster' manoeuvres such as going ahead and overtaking were more likely to involve younger drivers, particularly male drivers. This was also the finding of Hughes and Amis (1996).
3.83 A large body of work has tried to establish whether young drivers are more at risk because of immaturity or lack of experience. The reader is directed to Grayson and Sexton (2002) and Williamson (2003) who give summaries of the issues.
3.84 Motorcyclists accounted for 15% of all killed and seriously injured casualties on Scottish non built-up roads in 2003 (Scottish Executive, 2004a). However, they constitute less than 1% of traffic on Scottish rural roads (Scottish Executive, 2004c).
3.85 Recent research (Sexton et al, 2004) has shown that whilst most motorcyclist accidents in Scotland occur on built-up roads, the majority of fatal and serious injuries to motorcyclists occur on non built-up roads. These accidents tended to be the fault of the motorcyclist, resulted from a 'loss of control', involved 'sports bikes' and occurred on single carriageways with 60mph speed limits. Accidents on non built-up roads tended to occur at weekends reflecting recreational use. The research found that 20% of accidents occurred on bends and most of these involved the rider losing control.
3.86 Sexton et al (2004) also found that there has been a sharp increase in the numbers of killed and seriously injured motorcyclists on non built-up roads in recent years and that this rise coincided with a doubling in the numbers of licensed motorcyclists and corresponding rises in motorcycle traffic.
3.87 Clark et al (2003, 2004) report on a study of over 1,000 accidents involving motorcycles in the Midlands. Their findings are broadly in line with Sexton et al (2004). They found that motorcyclists and drivers were approximately evenly to blame for all accidents involving motorcyclists. However, other drivers (ie not the motorcyclist) were found to be responsible for the majority of those accidents involving two or more vehicles. The authors suggest that any initiatives in motorcycle safety should address the behaviours of both riders and other road users. However, this suggestion is based on addressing all motorcycle accidents - not just those on rural roads.
3.88 As mentioned previously, the contribution of motorcyclists to fatalities and serious injuries on some rural roads was highlighted in the most recent EuroRAP data for Great Britain ( AA Foundation, 2004).
3.89 Road Accidents Scotland 2003 does not give any indication of the relative proportions of child casualties on built-up and non built-up roads.
3.90 However, Christie et al (2002) examined STATS19 data in Great Britain to establish the main features of accidents involving children in rural areas. They found that there were considerably fewer accidents involving children in non built-up areas compared to built-up areas. In those which did occur, the child casualties tended to be car passengers and the children tended to have a lower severity ratio than adult car occupants.
3.91 They also found much fewer child pedestrian and cyclist casualties in non built-up areas than in built-up areas although they highlighted issues with children walking with their back to traffic and child cyclists being vulnerable at private driveways.
3.92 Christie et al (2002) examined various sources of exposure data and concluded that children in rural areas may be more exposed to accident risk as car passengers because of the greater car ownership, longer trip lengths and higher levels of travel to school by car. They concluded that in-car safety interventions and educational interventions which improve restraint use and focus on driver behaviour, especially regarding speed and alcohol, may be particularly important.
3.93 As shown in Chapter Two, whilst cyclists account for just 3% of fatalities on non built-up roads, one quarter of all killed and seriously injured cyclist casualties occur on non built-up roads. On average in Scotland, nearly two-thirds of all cyclist fatalities occur on non built-up roads (Scottish Executive, 2004a).
3.94 Gardner and Gray (1998) examined the issues affecting cycling on rural roads. They pointed out that, whilst all fatalities and a substantial proportion of serious accidents are reported, cycling accidents generally (and single vehicle accidents in particular) tend to be under-reported. Their main findings are summarised below:
- The rate of fatal cycling accidents per 100 million veh. km. on non built-up roads is almost three times that of built-up roads.
- Accidents on rural roads tend not to be clustered making spot treatment and even route treatment difficult and expensive
- The severity of cycling accidents increases with the speed limit of the road.
- Rural traffic growth could increase the danger to cyclists and will be a disincentive to cycling.
- At the time of the report, local authorities had only just begun to tackle rural traffic management
- They suggest that a rural road hierarchy is defined on the basis of the character and function of roads to provide a network where all roads are used optimally for road safety, movement and environmental requirements.
3.95 The question of whether tourists and visitors to rural Scotland have a higher road accident risk than local drivers was examined by Sharples and Fletcher (2001). They pointed to previous research which indicated that visitors, especially foreign visitors, have a higher road accident risk than local people. In particular, some research has shown that visitors can be unfamiliar with the roads and potentially the driving rules and conventions, and may be distracted (reading maps or viewing scenery).
3.96 Briganti and Hoel (1994) devised design guidance for roads which are promoted as 'scenic byways' (tourist routes) in Virginia, USA. They indicated that particular problems were experienced by unfamiliar drivers on single carriageway roads because these roads are less likely to have consistent traffic/information signage and geometric design standards.
3.97 Sharples and Fletcher (2001) examined STATS19 postcode data to establish the 'localness' of drivers involved in crashes in 1999. However, this exercise could only be completed for UK drivers. They found that over half of drivers were within 5km of their home address at the time of the accident. However, in the more rural police areas of Northern Constabulary and Dumfries & Galloway, a markedly greater proportion of drivers were more than 240km from their home address at the time of the accident.
3.98 Sharples and Fletcher's (2001) main conclusions were:
- Tourist activity does significantly boost road accident numbers in rural tourist areas of Scotland
- The overall rate of road accidents per vehicle mile is not increased significantly during tourist high season
- The exposure data for foreign drivers is not adequate to establish whether they are at greater risk of an accident than local drivers
3.99 A recent study in New Zealand (Nind et al, 2004) examined tourists' attitudes, knowledge and experience of driving in the Otago and Southland regions. Whilst many of the findings are specific to that local area, there were a few interesting results. They found that tourists who drive are mainly experienced drivers who drive everyday in their home country - only a small minority of tourists risk driving in New Zealand with limited driving experience. Tourists perceived New Zealand roads as good with most negative perceptions relating to weather, terrain, remembering to keep to the left and keeping below the speed limit. These concerns could well apply to many rural parts of Scotland and it may be beneficial to carry out a similar survey in Scotland to help identify areas where better information could be targeted.
Emergency Medical Service Response
3.100 Emergency medical service ( EMS) response time is known to be a critical factor in the mortality associated with road traffic crashes, especially in rural areas. Several studies have suggested that the higher fatality rate on rural roads can, at least partly, be explained by the emergency service response time. Unpublished research carried out by the University of Leeds in 1994, cited by MVA (1997) suggested that factors such as light and weather conditions, response time by emergency services and size and age of vehicles may play a role in different severity rates on Scotland's rural roads.
3.101 Williams et al (1991) examined geographical distributions of male and female road traffic fatalities in Scotland between 1979 and 1988. They found significantly higher mortality rates in areas with low population densities such as the rural Highlands, Borders and Dumfries & Galloway. This effect was more profound in Scotland than in other rural parts of the UK and the researchers suggested that this may in large part be due to the population and geographical characteristics of Scotland affecting EMS response times.
3.102 The OECD points to research which identifies three clear time periods in which trauma death can occur. It states that 50% of trauma deaths occur within a few minutes of injury, that only a few of such casualties can be successfully treated, and then only in large urban areas where rapid treatment is available. The second period, often called 'the golden hour', is the period in which early treatment could make a significant difference to survival. The third period occurs several days or even weeks after the initial injury and the OECD asserts that early treatment may not have a significant effect on the outcome ( OECD, 1999). Therefore, strategies to improve EMS response within the 'golden hour' would appear to have the most merit.
3.103 There appears to be a fairly large body of research on EMS response times and associated effects on road crash survivability although the findings are sometimes conflicting. For example, Grossman et al (1997) compared differences in response times, scene times and transport times by paramedics to trauma incidents in both urban and rural locations in Washington State. Mean response times and transport times (from incident to hospital) in rural areas were found to be nearly double those in urban areas. They found that rural victims were more than seven times more likely to die before arrival at hospital if the response time was more than 30 minutes. By contrast, Jones and Bentham (1995) reported that ambulance response times made no difference to mortality rates amongst road traffic casualties. It is difficult to judge how applicable this is to the rural situation is Scotland but it is clear that longer response times could lead to an increased risk of death.
3.104 Estochen et al (1998) point out that providing EMS in rural areas requires a different approach to urban areas due to lower population densities. In rural areas, the service must be provided over a large geographical area with limited resources. In urban areas, services can be consolidated to a smaller number of locations.
3.105 Since response time is potentially such a significant factor, quick accurate identification of the accident location is crucial. In rural areas, this can present an additional challenge as there may be fewer identifying landmarks. In addition, on many lightly-trafficked rural roads, a significant time might pass before the accident is discovered and reported.
3.106 Automatic crash notification systems, also known as 'Mayday' systems, consist of vehicle crash sensors combined with a geographic positioning system and can automatically alert EMS controllers to the occurrence and location of a crash using mobile phone technology. Systems with some of this functionality are available in the UK (see http://www.ractrackstar.com/). Research in the USA (Evanco, 1999 and Clark & Bushing, 2002) suggests that such systems could reduce road traffic fatalities by up to 11% and would be most useful in rural areas. The European Commission intends that, by 2009, all new cars sold in the European Union will be fitted with a 'Mayday' system known as 'eCall' (European Commission, 2005).
3.107 The OECD points out that there is a role for publicity in improving accident reporting and response. The public could be made aware of the steps to take in the aftermath of an accident and the information required by the emergency services ( OECD, 1999).
3.108 In addition, the OECD suggests that the number of road users able to administer first-aid could be increased by increasing the availability of first-aid and resuscitation training in rural areas ( OECD, 1999). High risk groups and professional drivers can also be targeted and recent initiatives in the rural West Highlands of Scotland have targeted first-aid training at motorcyclists (Scottish Ambulance Service, 2004).
3.109 Several studies from abroad have examined the issues associated with the use of farm vehicles on public roads. For example, in a USA study, Luginbuhl et al (2003) highlighted that the speed differential between farm vehicles (slow speeds) and normal traffic was a major concern. Another issue highlighted by Luginbuhl et al (2003) was use of proper lighting on farm vehicles. It is not clear how these findings would relate to the Scottish situation.
3.110 Published accident statistics do not separately identify agricultural vehicles - these are included in 'other vehicles' along with emergency vehicles, refuse collection vehicles and road sweepers amongst others. Also, some farm vehicles will be classed as cars or goods vehicles. However, 'other vehicles' accounted for less than 1.5% of all vehicles involved in Scottish road accidents in 2003 (Scottish Executive, 2004a).
3.111 Even though they account for a small proportion of accident-involved vehicles, the issue of agricultural vehicle use on public roads is a concern in Scotland. In 2002, the Perthshire Machinery Ring, supported by Perth & Kinross Council published a code of practice for using agricultural machinery on the public roads. (Perthshire Machinery Ring, 2002).
3.112 In a study by Knight (2001) into fatal accidents involving 'other motor vehicles', drivers of agricultural vehicles were far less likely to be responsible for the accident than other parties. The study also highlighted that tractors may pose a specific risk to motorcyclists (presumably related to the high speed differential) and that there was a possible issue of agricultural vehicles not always being roadworthy. As in the US research, proper vehicle lighting was highlighted as an issue and it was estimated that, in the accidents studied, improved lighting or conspicuity of the agricultural vehicle could have saved the lives of 15% of car occupants, 15% of motorcyclists and 15% of tractor occupants themselves.