Geographical and socio-ecological variations of traffic accidents among children

Sec. Sci. Med. Vol. 33, No. 7, pp. 765-769, Printed in Great Britain

1991

0277-9536/9l 53.00 + 0.00 Pcrgamon Press plc

GEOGRAPHICAL AND SOCIO-ECOLOGICAL VARIATIONS OF TRAFFIC ACCIDENTS AMONG CHILDREN MARIE-FRANCE JOLY,” PETERM. FCGGIN*and I. BARRY PLESS’ ‘Department of Social and Preventive Medicine, Center for Research on Transportation, University of Montreal, C.P. 6128, Succ. A, Montreal, Quebec, Canada H3C 3J7, *Department of Geography, University of Montreal, Montreal, Quebec, Canada H3C 357 and )Department of Epidemiology, McGill University, Montreal Children’s Hospital (Community Pediatric Research), Montreal, Quebec, Canada H3A 2T5 Abstract-This paper deals with geographical and socio-ecological variations of pedestrian and cyclist accidents (n = 1233) among children (c 15 years) on the Island of Montreal. The model includes variables on each child and his behavior when the accident happened and other temporal and spatial characteristics; environmental and socio-ecological data on the areas in which accidents occurred were also recorded. Descriptive, spatial and comparative analyses show specific patterns of location and occurrence of accidents. Factor analysis identifies the structure of characteristics linked to high accident areas for children’s traffic accidents. A strong similarity between zone characteristics emerged from the factor analysis for both types of accident. Only a few census tracts (between 9% to 13%) are high accident areas, but they are very concentrated spatially, and for some of them (2% for pedestrians and 4% for cyclists) the rate is five to eight times higher than for the rest of the urban area. Population structure and density factors contribute 40% of the variation in accident rates. In terms of numbers, fewer children were injured as cyclists than as pedestrians, and more boys than girls are involved in these accidents. Accidents occur under good conditions of visibility and on straight streets. Parked cars in school areas are also a sign of danger. Difficulties in interpreting information on directions and speed of moving vehicles are associated with high accident frequencies, particularly for young pedestrians (6-7 years). Key words-traffic factors

accidents, children’s accidents, medical geography, pedestrian, cyclist, socio-ecological

Traffic accidents among children are, according to the World Health Organization [l], the most serious epidemic problem in the industrialized world. Accidents account for between one quarter and one half of all causes of death within the O-14 age-group, and of these, traffic accidents account for between one quarter and one half of the total. Pedestrian and cycling accidents are the major types of traffic accidents among children [2-51. Traditionally, mortality statistics provide the most reliable source of accident data. However, the importance of morbidity data cannot be overemphasized, particularly in the urban context [6-91. This study uses morbidity data exclusively on the premise that it is particularly important to the scope of this problem given that the majority of victims are not fatalities but survive the accidents in which they are involved. There has been little published on the geography of traffic accidents. Notable exceptions include Moellering [lo], one of the first American geographers to analyze the problem of traffic accidents from a spatial point of view; Whitelegg [l 11, who has discussed the geography of accident locations in England; Thomas [12], who studied all corporeal and material accidents at the level of administration districts in Belgium in 1983 and 1984; and Baker er al. [ 131, who have

*To whom all correspondence

should be addressed.

published a study of variation in mortality rates by county in the contiguous 48 United States for the period 1979-1981. In Quebec, Pampalon [14] has described the way in which highway mortality is unevenly distributed between urban and rural areas. Loslier [15] has studied the relationship between social areas and mortality by motor vehicle accident in the Montreal region.

METHODS From October 1980 through March 1982, a prospective survey was undertaken to identify ail traffic accidents among children living in Montreal. This was done by implementing a monitoring system in emergency wards of the two main children’s hospitals and of nine other hospitals. These hospitals were chosen for their coverage of children’s traffic trauma throughout the Island of Montreal. In addition, data from police accident reports for the same period were cross-checked with the identified cases. Some 40% of the accidents seen in the emergency wards had no corresponding police accident report and two-thirds of the cases recorded by the police did not match the emergency records. All cases were retained in this study, whether they were identified by the police, by the hospital records, or by both sources together. The total number of cases included 1006 pedestrians and 227 cyclists, all injured children Coroner data were also checked 765

aged 15 and less. but no accidental

MANE-FRANCEJOLY et al.

766

pedestrian or cyclist death was recorded on the Island during the study [16]. Accident occurrences were subjected to a descriptive analysis which was done by victim of accident (pedestrian or cyclist). Then a spatial analysis for identifying the geographical variations of accidents was carried out at the census tract level. Census tracts were chosen for two reasons: first, tracts in Montreal (n = 471) are small areas whose main characteristics are homogeneity of environmental and urban design, and socioeconomic and ethnic status of the population [17]. Generally, the accident took place in the same census tract in which the child lived. The analysis involved identifying the location of accidents on a map and by creating a Comparative Accident Index (C.A.I.) for every census tract. The C.A.I. is a ratio between the observed accident rates and the theoretical expected rates. The accident rate for the community as a whole was used to calculate expected rates. Comparative

Accident Index =

observed rate x 100 expected rate

Then, factor analysis was used for identifying socio-ecological variations among accident areas. The aim of factor analysis (in the case) is to show the socio-ecological structure of the study region. Also, in view of the multicollinearity of the variables, particularly socio-ecological ones (education, employment, income), factor analysis is an effective tool for identifying the structure of the variables linked to areas characterized by high rates of accidents among children [l&19]. The model included variables on the child (sex, age, and behavior at the time of the accident), location, time, place, and other environmental data pertaining to the accident. Socio-ecological characteristics of the areas were also recorded. These include unemployment rates, schooling levels, single-families status, ethnicity, value of the dwelling, average number of people living in a room (which is an indirect index of poverty), density of the built-up areas, and traffic flow [20]. RESULTS

In this study, the percentage of accidents involving boys as pedestrians- and cyclists, respectively were 62 and 82%. These rates reflect the predominance of males in traffic accidents in general. The highest frequency by age for pedestrians was among the 6-7-year-olds: for the cyclists it was 13-14 years.

Table 1. Geographical

variations of children accidents

Number of areas for pedestrians

Accident level

Highest 9 (2%) I.C.A. = 500-800 High 35 (7.3%) I.C.A. = 300-500 Medium 168 (35.7%) I.C.A. = 100-300 Low 137 (29.1%) I.C.A. = < IO0 No accident reported 122 (25.9%) Total number of arcas (census tracts): 471

Number of areas for cyclists 21 (4%) 41 (9.2%) 82 (17.4%) 17 (3.6%) 310 (65.8%)

More accidents occurred between May and August (75%) than at other times, and more accidents took place on weekdays (87%) than on weekends; 55% of accidents happened during the late afternoon as opposed to 6% when going to school in the morning, and 20% between noon and 2 o’clock when the children were away from school for lunch. Thirty percent of the children were injured when coming back from school, or playing in the street after school (25%). Over two-thirds of the injuries took place in clear sunny conditions when visibility was excellent. Ninety percent happened on straight sections of streets and three-quarters of the accident sites were located far from traffic signals. Seventy percent of accidents occurred in 50 km/hr speed zones, and more than two-thirds of all accidents took place when traffic was moving in both directions. Data on pedestrian accidents show that one-third of the children were injured getting in or out of a vehicle. Another third were injured when crossing the street in disobeyance of traffic rules. However, 10% were injured when crossing the street properly. A small portion (6%) were hurt while playing on the street. One out of three cyclists was reported on the police accident records as being ‘distracted’ when the accident happened. The geographical variation among accident areas shows a strong pattern. High accident level areas as measured by the C.A.I. cover between 9 and 13% of the metropolitan area (Table 1). The mapping of these data indicates a definite concentration of those areas in the central part of the Island. The highest accident areas are situated along the shores of the St Lawrence River near the downtown core. In striking contrast, the high-income residential neighbourhoods stand out as zones where few accidents were reported. The results of the factor analysis (Table 2) demonstrated that demographic structure contributed to

Table 2. Factor analysis Pedestrians Factors Demographic structure Socio-economic status Density Underpriviliged areas Mobility Ethnicity Single parents Suburbs

Eigcnvaltte 9.58 4.44 2.99 I .94 1.64 I .27 -

Cvclists

Variatton _

Cumulative __..._~_.. variation

Eigenvaluc

% Variation

Cumulative variation

33. I 15.3 10.3 6.7 5.7 4.4 -

33. I 48.4 58.7 65.4 71.1 75.5 -

9.48 4.76 3.35 I .85 I .I3 1.20

31.6 15.9 Il.2 6.2 5.8 4.0

31.6 47.5 58.7 64.9 70.7 74.7

Variations

of traffic accidents

among

children

767

Table 3. Pedestrian factor loadinas Factor loadings Variable Population Male O-4 Male 5-9 Male IO-14 Female O-4 Female 5-9 Female IO-14 Population density Density of children Dwelling density Traffic intensity index Family income Education <9th grade University degree British origin French origin Ethnic origin Unemployment rate Single parents Mobility Poverty index Old housing 1971 Housing needing repairs Housing value

I

2

3

4

5

6

0.85 0.96 0.95 0.96 0.95 0.95 0.96 -0.08 0.23 0.16 -0.27 0.17 -0.32 -0.05 0.35 -0.11 0.02 0.31 0.33 -0.00 0.16 -0.46 0.17 -0.22 0.10

0.15 0.06 0.10 0.08 0.06 0.09 0.08 -0.06 -0.27 0.02 -0.06 0.75 -0.68 0.90 0.67 -0.78 0.65 -0.49 -0.19 0.04 -0.64 -0.11 -0.08 -0.36 0.72

0.18 0.00 -0.08 -0.04

-0.23 -0.14 -0.11 -0.12 -0.17 -0.10 -0.14 -0.02 0.11 -0.07 0.65 -0.35 0.36 -0.01 0.05 0.07 -0.19 0.56 -0.03 0.22 -0.03 0.66 -0.27 0.68 -0.17

0.01 0.06 0.04 -0.05 0.06 0.06 0.02 0.06 -0.21 0.19 0.18 -0.16 -0.01 0.14 0.04 -0.04 0.04 0.17 0.04 0.18 0.02 -0.26 0.83 -0.15 -0.17

0.01 0.02 0.02 0.01 0.03 0.00 0.01 -0.09 -0.06 -0.09 0.00 -0.08 -0.18 -0.02 0.17 -0.10 -0.36 -0.06 0.74 -0.13 -0.10 -0.03 0.14 -0.02 -0.15

over 30% of the variability in the data. Factor loading for pedestrian and cyclist are on Tables 3 and 4. The second factor, representing socioeconomic status, contributed over 15% of the variability with high factor loadings for variables such as people with a university diploma, high income and high-valued dwellings; negative factor loadings were observed for variables such as people with low level of schooling and a high number of people living in a same room which is a proxy for a poverty index. The third factor contributed to 10 to 11% of the variability and was characterized by density variables: density of adult population; density of the population of less than 15 years old; high dwelling density. The fourth factor

-0.03 -0.06 -0.07 0.96 0.80 0.92 -0.25 -0.13 0.06 0.04 -0.16 0.22 -0.00 -0.00 0.03 0.16 -0.12 0.23 -0.12 0.05 -0.13

reflected low socioeconomic areas; it contributed from 6 to 7% of the variability. The variables which emerged were: dwelling in need of repair, dwellings built before 1946, so quite old by North American standards, high unemployment rate and high intensity of traffic. The last two factors, which each contributed between 4 to 6% of the variability, were-for the pedestrians-a mobility factor (people who moved recently into a new dwelling) and a singleparent factor. For the cyclists the factors corresponded to ethnicity (other origin than English or French), and to people living in suburbs built after 1971. The total amount of explained variability was 75%.

Table 4. Cvclist factor loadinns Factor loadings Variable Population Male O-4 Male 5-9 Male IO-14 Female O-4 Female 5-9 Female IO-14 Population density Density of children Dwelling density Traffic intensity index Family income Education ~9th grade University degree British origin French origin Ethnic origin Unemployment rate Single parents Mobility Poverty index Old housing -c 1946 New housing > 1971 Housing needing repairs __ Housrng value

1

2

3

4

5

6

0.88 0.95 0.96 0.93 0.94 0.96 0.94 -0.13 0.07 -0.18 -0.03 0.13 -0.20 -0.01 0.09 -0.24 0.27 -0.28 -0.26 -0.23 0.2 I -0.33 0.14 -0.24 _ __ 0.02

0.07 0.05 0.10 0.11 -0.00 0.12 0.13 -0.17 -0.21 -0.14 -0.00 0.89 -0.82 0.94 0.65 -0.70 0.40 -0.51 -0.41 -0.01 -0.73 -0.01 -0.15 -0.30 0.83

0.07 -0.00 -0.04 -0.08 -0.01 -0.06 -0.07 0.93 0.86 0.9 I -0.20 -0.11 0.06 0.09 -0.32 0.06 0.31 0.09 0.02 0.24 -0.07 0.33 -0.21 0.10 -0.11

-0.20 -0.11 -0.08 -0.10 -0.12 -0.08 -0.04 0.01 0.10 -0.02 0.71 -0.13 0.32 0.02 0.05 0.07 -0.14 0.38 0.39 0.29 -0.01 0.67 -0.22 0.77 -0.03

0.00 0.02 0.03 -0.03 0.03 0.02 -0.01 0.03 0.12 0.00 0.06 -0.02 0.01 0.08 0.09 -0.45 0.61 -0.07 0.07 0.11 0.38 -0.24 0.09 -0.05 0.16

-0.08 -0.02 0.04 0.06 -0.05 0.01 -0.02 -0.04 -0.15 0.05 0.00 -0.13 -0.03 0.04 0.07 -0.02 -0.03 0.29 0.15 0.69 0.15 -0.17 0.75 0.00 -0.01

768

MARIE-FRANCE JOLY et al. DISCUSSION

Fewer children were injured as cyclists than as pedestrians, but this difference is mainly a reflection of relative exposure. Montreal winters are harsh: snow and ice are on the streets from December to March, conditions which are not conducive to cycling. A recent study [21] confirmed that mobility of children between 5 to 14 years throughout Montreal is essentially achieved through walking. Unfortunately, at this scale, individual data on exposure-in term of miles or hours spent walking or cycling-are not available. Many studies have shown a direct correlation between accident rates and risk expressed in terms of time and length of exposure, distance covered, holidays, or school vacations [22-261. However we found that 6% of the children were injured on their way to school, compared with 30% when they were coming back home. We can reasonably consider that attitude, lack of attention and fatigue are some factors which explain these differences. Complex situations (66% of the accidents took place where traffic was moving in both directions) are associated with high accident rates, particularly for young children (6-7 years). This confirms the results of the O.C.D.E. [26], the Skandia Report (271, and the works of Sandels [28,29]. By far the greatest number of pedestrian and cyclist accidents took place on streets considered to be quite safe. They often occur near the child’s home [2,22-261. In our study area, 96% of all accidents took place in the census tract of the child’s residence. A strong similarity between zone characteristics appears in the factor analysis for both accident types, as is the case for the descriptive analysis of pedestrian and cyclist accidents. For a few census tracts (9 for pedestrians and 21 for cyclists) the accident level is evaluated as being five to eight times higher than that for all city census tracts taken as a whole. The demographic structure and density factors, which together account for 40% of the variability in the data, show an over-representation of accidents among areas of high population density. These factors are the expression of low socioeconomic status and they are also linked to high residential mobility (usually in the same area), as well as to presence of recent immigrants and single-parent families. It would seem to be the case of a ‘captive’ population. CONCLUSION

Geographic analysis is an alternative approach to the study of traffic accidents and their causes. The use of a comparative accident index in reference to spatial zones also shows the comparative magnitude of traffic accidents among children. The similarity of the characteristics of high accident level areas for pedestrian as for cyclists shows us that the families living in low socioeconomic areas have more than their share of juvenile injuries and interventions aimed at decreasing these rates should have high priority in those areas. Acknowledgements-This research was supported by a fellowship from F.C.A.R./Transport and the Social Sciences and Humanities Research Council of Canada.

REFERENCES

I

Deschamps

J. P. L.u Prevention

des Accidents

de ia

W.H.O. Rapports et etudes EURO 26, Copenhague, 1980. 2. Wailer J. A. Injury Control: A Guide to the Causes and Prevention of Trauma. Lexington Books, Lexington, 1984. 3. Baker S. P., O’Neill B. and Karpf R. S. The Injury Fact Book. Lexington Books, Lexmgton. 1984. _ _ 4. Lx Bourdais C., Laberae-Nadeau C. and Bourbeau R. Pedestrian Injuries ani Medical Treatment. American Association for Automotive Medicine, Proceedings, Morton Grove. Illinois (7-9 October 1980. Rochester. New York). R.A.A.Q. La s&urirP rouriere des pietons. Recherche et Statistique. R.A.A.Q., Qubbec, 1982. Pampalon R. Environnement et Sante. Elements d’une Problematique QuPbPcoise. M.A.S., Qutbec, 1980. Read J., Bradley E., Morison J., Lewall D. and Clarke D. The epidemiology and prevention of traffic accidents involving child Pedestrians. Can. Med. Ass. J. Circulation

89, 687-701,

Chez

les Enfonts.

1963.

8. Bisson A. La SPcuritP Routiere, Dossier sur les Pietons. R.A.A.Q., Quebec, 1984. 9. Rivara F. P. The causes, impact, and preventability of childhood iniuries in the United States: Childhood pedestrian injuries in the United States. Am. J. Dis. Child. 144, 692-696, 1990. 10. Moellering H. J. The Journey of Deafh: A Spatial Ann&is

of Fatal

Trafic

Crashes

in Michigan

1969.

Ph.D. Thesis, Department of Geography, University of Michigan, 1973. II. Whitelegg J. A geography of road traffic accidents. Trans. Inst. Br. Geog. 12, 161-176,

1987.

12. Thomas I. Les accidents de la circulation de la route en Belgique. BufIetin de la SociPtP Beige des Etudes Geogrophiques, pp. 103-l 11. Etudes Geographic, SOBEG, Bruxelles, 1987. 13. Baker S. P., Whitfield R. A. and O’Neil B. Geographic variation in mortality from motor vehicle crashes. New Engl. J. Med. 316(22), 1383-1387, 1987. 14. Pampalon R. Gtographie de la mortalite sur les routes au Quebec. Routes et transports 32, 18-21, 1982. 15. Loslier L. LQ Mortalire darts Ies sires sociales de la region mktropolitaine de Montreal. MAS, Quebec, Decembre 1976. 16. Pless I. B., Verrault R., Arsenault L., Frappier J. Y. and Stulginskas J. The epidemiology of road accidents in childhood. Am. J. Pub. Health 77, 358-360. 1987. 17. Foggin P. and Polese M. La GPographie Sociale de Montreal en 1971. INRS, Urbanisation, Etudes et Documents 1, Montreal, 1976. 18. Kitagawa E. and Hauser R. Differential Mortality in the &ted States; A Study of Socio-Economic EpidemiOIOPV. Harvard Universitv Press. Cambridge. 1973.

19. Fog&r P. and Bissonneite F. L’analyse in composantes. Revue de Geographic de Montreal. Les Presses de 1’Universite Lava& Quebec, 1976. 20. Joly Pomes Barr&e M-F. Geographic des Accidents de la Route chez les Enfants en Milieu Urbain. Ph.D. Thesis, Universite de Montreal, 1985. 21. Buss&e Y., Marcoux R. and Tessier M. Analyse Prospective de la Demande de Transport des Personnes dons la Region Metropolitaine de Montreal, 1981-1986.

INRS Urbanisation, Etudes et Documents Number 56, Montreal, 1988. 22. Jamieson K. G. and Tait I. A. Trafic Injury in Brisbane. National Health and Medical Research Council, Special Report, Series Number 13, Canberra, 1966. 23. Aldman B. and Thorson J. Motorization and Traffic Mortality in Sweden. Act. Anal. Prev. 3, 215-221. 1971.

Variations of traffic accidents among children 24. Williams A. F. Factors in the Initiation of BicycleMotor Vehicle Collisions. Am. J. Dir. Child. 130, 370-377,

1976.

25. Slatis P. Injury Finnish National 26. O.C.D.E. Tra#ic Research, Paris,

Patterns

in Road

Trafic

Accidents.

Board of Health, Helsinki, 1967. Safety of Children. Road Transport 1983.

769

27. Federation

Intemationale des Pittons. Les Enfanrs Dans les Accidents de la Circulation en SuPde. La Voie du Pi&on III. Skandia Report, Stockholm, 1973. 28. Sandels S. Why are children injured in traffic? The Skandia Report II, Skandia Insurance Company Limited, Stockholm, 1974. 29. Sandels S. Children in Trafic. Elek Books, London. 1975.