Environmental factors and child pedestrian injuries Ian Roberts and Robyn Norton

Roger Dunn

Ian Hassall

Injury Preventicni Research Centre, University of Auckland

Department of Engineering, University of Auckland

Commissimr for Children, Wellingtmi

Trevor Lee-Joe Injury Prevenlion Research Gntre, University of Auckland

Abstract: The aim of this study was to provide information on modifiable environnlental factors at sites of child pedestrian injuries which would be useful for the design of preventive strategies. A total of 103 child pedestrian injuries was identified over a one-year surveillance period. Of these, 71 per cent occurred on public roads, 24 per cent occurred in residential driveways and 4 per cent occurred in car parks. For roadway injury sites, the median traffic flow was 635 vehicles per hour and the median speed was 49.0 kph. For 3 1 (42 per cent) roadway injury sites the median vehicle speed was in excess of the posted speed limit. The majority of roadway injuries occurred on residential streets. For driveway injuries there was no separation of the driveway from the children's play area in 75 per cent. Changes to the urban traffic environment which reduce traffic flows and vehicle speeds have the potential to prevent child pedestrian injuries. Controlled epidemiologic studies which examine the niagnitudc of the risks associated with modifiable environmental factors arc required. (Awt J Public Healfh 1994; 18: 43-6)

edestrian injuries are a leading cause of child- hood mortality in Australasia. Each year over P 110 Australian children a re killed in

pedestrian-motor vehicle collisions. I Similarly, ped- estrian injuries are an important child health prob- lem for New Zealand, which has a child pedestrian mortality rate significantly higher than that of com- parable countries.' For the children who survive a pedestrian-motor vehicle collision, the injuries are often particularly severe. Pedestrian injuries are a leading cause o f admission of children to intensive care facilities, most often for the treatment of severe brain injuries.' About 75 per cent of these children will still have to contend with disability one year after injury.4

William Haddon described injury causation in terms of the epidemiologic triad of host, agent and environment, suggesting that modification of any of these components has the potential to interrupt the causal sequence.5 In the prevention of child ped- estrian injuries, interventions directed at the host, by way of pedestrian skills training programs, have proven to be of limited value, with even large efforts t o improve child pedestrian behaviour being rewarded by only small gains6 Similarly, there is evi- dence that interventions directed at the agent, in this case the driver and vehicle, are also likely to be inef- fective. Negligent driver behaviour has been shown to be particularly refractory to change and vehicle design features have been shown to have little effect on injury severity in children.'." As a result, changes to urban I raffic environments are considered to hold

Correspondence t o Dr Ian Roberts, Injury Prevention Research Centre, Drpanment of Community Health, IJniversity of Auckland, Private Bag92 019, Auckland. New Zealand. Fax 9 373 7503.

the greatest potential for prevention.9 Despite this, there is surprisingly little information on the charac- teristics of the traffic environment at sites of child pedestrian injuries. To date the studies which have examined the traffic environment have been based on police accident reports."' However, since these data are collected primarily for litigation purposes, information useful for prevention is often absent. More importantly, child pedestrian injuries are sig- nificantly underreported in police accident databases and the validity of the environmental data in these reports has been shown to be low.".'2 The aim of this study therefore, was to provide valid information on modifiable aspects of the traffic environment (at sites of child pedestrian injuries) which would be useful for the d e s i p of preventive strategies.

Methods All children under fifteen years of age, killed or hos- pitalised as a result of a pedestrian injury in the Auck- land region between 15 January 1992 and 15 January 1993 were identified through a surveillance system established at hospitals which admit injured children and with the Auckland coroner. For the purposes of this study a 'pedestrian injury' was defined as an injury sustained in a collision with a motor vehicle by a person who was not at the time of injury riding in o r on a motor vehicle, train, animal-drawn or other vehicle or on a bicycle or animal. Both traffic and non-traffic pedestrian injuries were therefore included. The Auckland region has a predominantly urban population of 936 981, of whom 213 174 are under fifteen years of age.IY Information on the time, date and exact location of the pedestrian-motor vehicle collision was obtained during parental inter-

AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1994 VOL 18 NO. 1 43

ROBERTS E T AI

view, or if this information was unknown by the parent, from police records.

Data on the environmental characteristics of the injury sites were collected by a civil engineer who visited each site on the same day of the week and at the same time of day as the injury, usually one week later. For injuries occurring in continuous traffic flow, a 24-hour profile of traffic volume and vehicle speed were obtained at the injury site using vehicle recorders positioned as near as possible to the injury site. From these 24-hour profiles, the mean traffic volume and mean vehicle speed for the three-hour period around the time of injury were obtained. For injuries at intersections, we measured the traffic flow on the street that the colliding vehicle was travelling on immediately before the collision. Information was also collected on: roadway function, number of lanes, the speed limit in force, the presence or absence of traffic-calming measures and the distances between the children’s homes and the injury sites.

In order to ascertain whether traffic flow and vehicle speed measured one week following an injury would accurately reflect flow and speed at the time of injury, repeat measurements were made for a ran- dom sample of ten injury sites, both one and two weeks following the injury, If these measurements corresponded closely it would suggest that the speed and flow characteristics of a particular site are stable characteristics and that the data collection instru- ment is reliable. Differences in the means of the two flow and speed measurements were examined with both parametric (paired t test) and nonparametric (sign test) significance tests. As a further test of reliability, the regression equations describing the relationship between the Week 1 and Week 2 flow and speed measurements respectively were obtained, with regression t tests used to test the hypotheses that the intercepts were zero and that the gradients were unity.

For driveway injuries, site visits were conducted (with the permission of the householder) and infor- mation was collected on type and tenure of dwelling and the presence or absence of fences separating the children’s play area from the driveway.

The study was approved by the Auckland Area Health Board Ethics Committee.

Results Over the one-year period, a total of 103 injured child pedestrians were identified by the surveillance sys- tem, three of whom were identified by surveillance at the coroner’s office. The median age of the injured children was five years. There were 65 boys and 38 girls, a male to female ratio of 1.7 to 1. The site of injury was ascertained for all cases. For 99 cases this information was provided by the parent or caregiver; for four cases the information was obtained from the Ministry of Transport.

Of the 103 injury sites, 73 were public roads, 25 were residential driveways, four were car parks and one was a public park.

Roadway injuries Victim characteristics: For the 73 children injured on public roads the median age was seven years (range 1 to 14 years). There were 44 boys and 29 girls, a male to female ratio of 1.5 to 1.

Temporal characteristics: The modal time for injury was 15.30 hours (range 08.00 hours to 23.30 hours). More than half (59 per cent) of the injuries occurred between 1500 and 1800 hours, with 30 per cent occurring between 1500 and 1600. The peak day for injury was Thursday (21 per cent). Environmental characteristics: The volume and speed data for the ten injury sites for which repeat measurements were made are shown in Table 1. There was no significant difference between the traffic flows (t test P= 0.59, sign test P = 0.85) or the vehicle speeds (t test P = 0.73, sign test P = 1.00) between the two measures. In a plot of the traffic flow data, with they-axis being the traffic flow at week 1 and the x-axis the traffic flow at week 2, the fitted model was y=-19.26 + 1.00309~. The test of intercept = 0 gave P= 0.68, while the test of gradient = 1 gave P = 0.92. Similarly, for the speed data, the fitted model was y = -0.59 + 1.00943~. The test of intercept = 0 gave P = 0.66, while the test for gradient = 1 gave P = 0.71. These tests therefore suggest that the study data provide an accurate indi- cation of the conditions at the time of injury.

Of the 73 roadway injuries, 12 (16 per cent) occurred on a pedestrian crossing, 16 (22 per cent) occurred at an intersection (not on a crossing) and 45 (62 per cent) occurred mid-block (not on a crossing).

Data on the environmental characteristics are pre- sented in Table 2. The median traffic flow at injury sites was 635 vehicles per hour with a median speed for the three-hour period around the time of injury of 49.0 kph. Of the 73 injury sites, the posted speed limit was 50 kph or less in 69 (95 per cent). The

Table 1 : Reliability of speed (kph) and flow (vehicles per hour) measurements a t injury sites

Injury site Week 1 Week 2 Difference Oh change

Site 7 Mean flow Mean speed Site 2 Mean flaw Mean speed Site 3 Mean flow Mean speed Site 4 Mean flow Mean speed Site 5 Mean flow Mean speed Site 6 Mean flow Mean speed Site 7 Mean flow Mean speed Site 8 Mean flow Mean speed Site 9 Mean flow Mean speed Site 10 Mean flaw

3039 64

1078 65

609 38

892 56

51 42

822 60

355 41

283 32

178 59

3413

301 2 65

1348 64

663 39

799 57

39 41

796 59

385 42

278 32

175 59

3384 Mean speed 66 66

+27 1 -1 2

-270 25 +1 2

-54 9 -1 3

+ 93 10 -1 2

+12 24 +1 2

+ 26 3 +1 2

-30 8 -1 2

+ 5 2 0 0

+ 3 2 0 0

+ 29 1 0 0

44 AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1994 VOL 18 NO. 1

CHILD PEDESTRIAN INJURIES

Table 2: Environmental characteristics for 73 roadway injury sites

Characteristic

Traffic volume (vehicles per hour) 0 to 500 500 to 1000 Over 1000 Site mean speed (kph) <50 kph > 50 kph

Roadway function Arterial Residential through rood Residential road Cul de soc Other (motorway)

Number of lanes (each side) 1 lane 2 lanes 3 lanes

Speed limit in force 50 kph 70 kph 100 kph

Traffic calming measures N o Yes

Home-injury site distance 0 krn

<0.5 km <1.0 km < 2.0 km

Number ~ - _ _

31 23 19

36 37

13 19 36 4 1

57 12 4

69 2 2

69 4

20 44 51 58

010

42 32 26

49 51

18 26 49 6 1

78 16 6

95 3 3

94 6

27 60 70 80

median vehicle speed was greater than the posted speed limit at 31 (42 per cent) of injury sites. The majority of injury sites (81 per cent) were residential roads with only 18 per cent occurring on urban arterials (Table 2). Twenty-seven per cent of children were injured immediately outside their homes, 60 per cent within 0.5 kilometres and 70 per cent within one kilometre. For the children injured immediately outside their homes, the median traffic flow was 297 vehicles per hour with a median vehicle speed of 48.0 kph.

hiveway injuries Victim characteristics: For the 25 children injured in residential driveways the median age was 1.5 years (range 0 to 7 years). There were 15 males and 10 females, a male to female ratio of 1.5 to 1. Temporal characteristics: The injury times were evenly distributed through the daylight hours. The peak day for injury was Sunday (25 per cent). Environmental characteristics: Of the 25 driveway sites, site visits were completed in all but one case, where the parents refused permission. Of the remaining 24 sites, 19 (79 per cent) were associated with single dwellings with private driveways and five (21 per cent) were associated with multiple dwellings (flats) with shared driveways. Thirteen (54 per cent) of the homes were rented, seven of which were rented from the state, seven (29 per cent) were owner-occupied homes and for the remaining four cases, the parents of the injured child were living with

relatives. There was a fence separating the driveway from the main children’s play area in only six (25 per cent) of the homes.

Discussion The extent to which descriptive injury studies are able to contribute to the development of injury countermeasures is often limited. However, studies which provide new information on injury problems which are of high frequency and high severity are of value. Pedestrian injuries, in particular, have been flagged as an area deserving greater attention.’” In this study, collaboration between the disciplines o f public health and civil engineering facilitated the provision of accurate information on a dimension of the child pedestrian injury problem which to date has received scant attention. Indeed, the lack of infor- mation on environmental factors in child pedestrian- motor vehicle collisions may, in part, be responsible for the overemphasis on the role of human factors in the causation of child pedestrian injuries.Ig

The results show that child pedestrian injuries occur in situations of high traffic flow where a large proportion of vehicles are travelling at speeds in excess of the posted speed limit. These roads are predominantly local residential roads and not urban arterials. Even for the children injured immediately outside their own homes there were high traffic vol- umes and average speeds close t o the posted speed limit. It can be estimated that with a median traffic flow of 635 vehicles per hour, assuming a steady flow, the time available for a child to cross is approximately six seconds.

The observation that most child pedestrian injuries occur on local residential streets was also made in a study in Perth.’” However, a North American study found that residential streets were rarely the site of injury.16 The importance of this finding is that traffic- calming measures (for example, speed humps and road narrowing), which are an effective method o f reducing speeds and have been associated with size- able reductions in injury rates, are likely to be a feas- ible and cost-effective countermeasure in Australasian cities. l i Giving priority to traffic calming schemes to urban areas with particularly high child pedestrian injury rates (such as the most socio- economically disadvantaged areas), would further increase their cost-effectiveness. I ” However, despite a large body of evidence supporting the use of environmental approaches to speed control, detec- tion and enforcement continue as the mainstay of speed control strategies in New Zealand.I9

The residential driveway was the site of pedestrian injury in 24 per cent of pedestrian injuries in this study compared with only 4 per cent of pedestrian injuries in Perth. The Perth study, however, was based on police accident reports and the difference is more likely to reflect underreporting of driveway injuries in police accident databases than a real dif- ference in injury occurrence.2o Seventy-five per cent of driveway injuries occurred at homes where there was no physical separation between the driveway and the children’s playing area. As yet, there is no infor- mation from controlled epidemiologic studies on the magnitude of the risks associated with unfenced driveways. As 24 per cent of all pedestrian injuries were driveway-related this information would be of

AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1994 VOL 18 NO. 1 45

ROBERTS E T AL

considerable value for the development of appropri- ate countermeasures.

To date, the major thrust of preventive strategy in Australasia has been effort to change child ped- estrian behaviour through pedestrian skills edu- cation programs. This study provides some insight into why these programs have proven to be of limited value. Our data suggest that the road crossing task for many children involves sprinting through gaps in rapidly moving traffic, with errors ofjudgment likely to result in serious injury. Nevertheless, 'don't run across the road' has been advocated as a road cross- ing principle for many years, even though the only basis for this statement is the observation that a large number of children injured crossing roads were run- ning at the time.2'

In contrast, there is some evidence to suggest that limiting traffic volumes in urban areas may be a more effective way of reducing child pedestrian mortality rates. In New Zealand, government policies to dis- courage car use in the aftermath of the 1973 energy crisis were associated with a 46.4 per cent reduction in the child pedestrian mortality rate." A case- control study in Washington State found that chil- dren living in neighbourhoods with the highest traffic volumes had a risk of pedestrian injury over three times that of children living in the least busy areas.gg However, past and current transport policies have encouraged car use. In particular, the high capi- tal investment in roads compared with other trans-

proudly present the

GRAND FINALE TO A SERIES OF TRANS-AUSTRALIAN

CONSENSUS MEETINGS

SAFE AND EFFECTIVE LIFTING CONFERENCE

Sydney, 21 - 22 September, 1994

This 2 day Grand Finale will propose a solution that is 'a set of principles' for lifting training, test the principles for adaptability and suggest 'how' these should be taught. Experts in the field have been invited to challenge or support the 'solution'.

This conference is for People with professional interests in lifting ie, those in Occupational Health and Safety, Medicine, Physiotherapy, Occupational Therapy, Nursing, Ergonomics, Rehabilitation, Exercise and Sports Science, Fitness, etc. Previous participants are strongly urged to attend.

For brochures phone (02)646-6343 Faculty o f Health Sciences, The Univcrsity o f Sydney

portation modes, cheap car parking and running costs and cuts in public transport subsidies have exacerbated the trend towards increasing car traveLg4 Despite this, children continue to be held responsible for the problem of child pedestrian injuries.gs Public health practitioners now have an obligation t o draw attention to the responsibility of government.

Acknowledgments This study was funded by the Health Research Coun- cil of New Zealand.

References 1 .

2.

3.

4.

Australian Bureau of Statistics. Road traflic accidents involving casualties, Awtralza. Cat. no. 940503. Canberra: ABS, 3987. Roberts I , Norton R, Hassall 1. Child pedestrian injury 1978- 1987. NZ Med J 1992; 105: 51-2. Roberts lG, Streat SJ. Judson JA. Norton R N . Critical injur- ies in paediatric pedestrians. NZ Med J 1991; 104: 247-8. Harris BH, Schwaitzberg SD, Sernan T M , Herrmann C. The hidden morbidity o f paediatric trauma. J Pediatr Surg 1989; 24: 103-6. Haddon W. Advances in the epidemiology of injuries as a basis for public policy. Public Health Rep 1980; 95: 4 1 1-2 1 . Rivara FP. Booth CL, Bergman AB, Rogers 1.W. Wiess J. Pre- vention of pedestrian injuries to children: effectiveness of a school training program. Pedidrics 1991; 88: 770-5.

7. Baker SP. Robertson LS, O'Neill B. Fatal pedestrian collisions. Am J Public Health 1974; 64: 318-25.

8. Pitt R, Cuyer B, Hsieh C, Malek M. The severity of pedestrian injuries in children: an analysis of the pedestrian injury caus- ation study. Accid Anal Preu 1990; 22: 549-59. Sibert JR. Accidents t o children: the dortur's role. Education or environmental change? Arch Dis Child 199 I ; 66: 890-3. Stevenson MR, 1.0 SK, Laing BA, Jamrwik KD. Childhood pedestrian injuries in the Perth metropolitan area. MedJAust 1992; 156: 234-8.

1 1 . Maas MW, Harris S. Police recording of road accident in-patients. Acczd Anal Preu 1984; 16: 167-84.

12. Shinar D. Treat JR. McDonald ST. The validity of police reported accident data. Accid Anal Prev 1983; 15: 175-91.

13. Department of Statistics. C m w o /popu ld im and dwellingv. March 1991. Wellington: Department of Statistics. 1991.

14. Rivara FP, Wolf ME. injury research: where should we go from here? Pediatno 1989; 84: 180-1.

15. Carsten OM, Tight MR, Southwell MT. Plows B. Urbun acci- &nf.s: why do they happen? Basingstoke: AA Foundation for Road Safety Research, 1989. Rivara FP, Reay DT, Bergman AB. Analysis of fatal ped- estrian injuries in King County, WA, and prospects for pre- vention. Public Health Repods 1989; 104: 293-7. Faure A, Neuville A. Safety in urban areas: The French pro- gramme 'safer city. accident free districts'. Accid Anal Preu 1992; 24: 1031-4. Roberts I , Marshall R, Norton R. Borman B. An area analysis of child injury morbidity in Auckland. J Paedratr Child Health 1992; 28: 438-4 1.

19. National Road Sa/ety Plan. Wellington: The Officials' Com- mittee on Road Safety, 1991.

20. Bull JP, Roberts BJ. Road accident statistics: a comparison uf police and hospital information. Accid Anal freu 1973; 5: 45-53. Crayson GB. The identification of training objectives: what shall we tell the children? Accid Anal Preu 1981; 13: 169-73. Roberts I. Marshall R, Norton R. Child pedestrian mortality and traffic volume in New Zealand. BMJ 1992; 305: 283. Mueller BA, Rivara FP, Shyh-Mine L, Weiss NS. Environ- mental factors and the risk for childhood pedestrian motor vehicle collision occurrence. Am J Epidemiol 1990; 132: 550-60. Roberts 1. Norton R. Transport and public health. NZ Med J

Hillman M, Adams J, Whitelegg J . Oncfalse move: study of children's ind.t?endeflt mobility. London: Policy Studies Insti- tute, 1991.

5.

6.

9.

10.

16.

17.

18.

21.

22.

23.

24.

25. 1993; 106: 26-7.

46 AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1994 VOL 18 NO. 1