Childhood Injury Prevention Interventions

Skills Training Programs

Background

Street crossing is a complex series of actions, with as many as 26 tasks needed to negotiate traffic safely. Pedestrian skills training programs have long been the mainstay of interventions to reduce child pedestrian injuries. These training programs are of variable length and intensity and have occurred both in school settings as well as community settings. One study found 159 different programs intended for children.2 Evaluations of these programs which we have considered here consist of both changes in injury rates, as well as changes in pedestrian behavior. We have not included evaluations based solely on changes in knowledge, attitudes, or reported behavior change, since these correlate poorly with actual behavior change. Our 1997 review included studies which evaluated educational programs using simulated or model traffic situations. For this 2001 update we excluded studies using these "artificial" outcomes. (Renaud 1989, Nishioka 1991, Thomson 1998).


Review of pedestrian skills training programs:

Author

Roberts et al., 1995

Study design and target population

Case control study.

Auckland New Zealand

Cases: (n=54) children killed or hospitalized by pedestrian injury occurring on journey home from school

Controls: (n=157) Random sample of children who walk to and from school.

Intervention

Adult accompaniment on school-home journey for children ages 5-15.

Outcomes

Deaths and hospitalized injuries.

Study period January 1992 to March 1994.

Results

Reduced risk of injury for children who had an adult walk them home from school.

OR=0.31, (0.07-1.49).

Adjusted for age, sex and SES.

Study quality and conclusions

Having adults accompany children appears to reduce risk of injury. Presence of adults should be accounted for when evaluating pedestrian injury rates on basis of exposure.


Author

Rivara et al., 1991

Study design and target population

Controlled trial

229 children aged 5-9 years (kindergarten to Grade 4) in 3 public elementary schools in Seattle

Intervention

6 session curriculum based on modeling and positive reinforcement, training in real traffic environment, parent booklet to reinforce school program.

Outcomes

Pre-test and post-test observations of children as they walked home from school evaluating if they walked on sidewalk, stopped before crossing, looked before crossing, and kept on looking.

Results

Not Walking on sidewalk: OR=0.39, 0 .04-2.45

Not Stopping before crossing: OR=0.53, 0.27-1.03

Not Looking before crossing: OR=0.34, 0.18-0.66

Not Keep on looking: OR=0.24, 0 .11-0.54.

Study quality and conclusions

The addition of the parent component helped to improve behavior. However, even after the intervention, approximately 50% of K-1 children and 50-70% of Grade 2-4 children still had unsafe behavior.


Author

Young and Lee, 1987

Study design and target population

Before-after design

5 year old children, Edinburgh. Group 1, n=23; group 2, n=15

Intervention

Group 1 received series of training sessions on 1-way street; group 2 series of sessions on 2-way street

Outcomes

Percent of subjects crossing with too short of gap ("tight fits") between self and vehicle; percent of subjects with missed opportunities for crossing

Results

Single lane roads: No effects on "tight fits"; reduction in number of missed opportunities (OR=0.69, 0.18-2.67)

Two lanes: Reduction in "tight fits" (OR=0.76, 0.14-4.11); no effect on missed opportunities.

Study quality and conclusions

Artificial training and testing situation. Results indicate potentially modest effect, although sample size very small.


Author

Rothengatter, 1984

Study design and target population

Non-randomized controlled trial: parent-trained group, research assistant-trained group, control group

307 children 4-6 years old in the Netherlands

Intervention

Practical training in the real-world environment, supplemented with social learning and behavior modification techniques. Utilized parents and kindergarten teachers.

Outcomes

Traffic behavior at end of intervention and 4 months later

Results

Children <5 yrs: Crossing incorrectly reduced by 98% with assistant training (OR=0.02, 0.00-0.19); reduced by 63% with parent training (OR=0.37, 0.11-1.22).

Children>5 yrs: Crossing correctly at junctions reduced by 57% with assistant training (OR=0.43, 0 .14-1.27) and by 58% with parent training (OR=0.42, 0.41-1.26).

Study quality and conclusions

Training appeared to improve crossing behavior, with training by parents almost as good as training by research assistants.

Children knew they were being observed so that this should be viewed as optimal behavior.



Author

Van der Molen et al., 1983

Study design and target population

Before-after design

63 pre-school children in the Netherlands

Intervention

Training in crossing streets, crossing near parked cars, and crossing at intersections.

Consisted of four 10-15 minute training sessions per task.

Outcomes

Observations of children crossing real streets.

Proportion of children stopping at the curb, looking left, right, and backward.

Results

Stopping increased from 46% to 63% (OR for not stopping=0.49, 0.23-1.06); looking to left increased from 35% to 46% (OR for not looking left=0.63, 0.29,-1.37); looking right increased from 41% to 60% (OR for not looking right=0.46, 0.21-1.00); looking backward increased from 3% to 18% (OR for not looking backward=0.15, 0.02-0.77). Greatest improvement was on streets with low density traffic.

Study quality and conclusions

This is a strong intervention from the Gronigen group, involving parents teaching their children as well. It appears to have been somewhat effective, although after the intervention many children still had poor skills. There was no long term follow-up to see the decay of the intervention effects.


Author

Fortenberry and Brown, 1982

Study design and target population

Tine series design

6-7 year old children in 4 cities in Alabama

Intervention

School-based educational program

Outcomes

Ratio of the number of pedestrian accidents in 6-7 year olds to all age groups for 2 years before and 2 years after the training program.

Results

12% of the pedestrian accidents before the intervention were to 6-7 year olds compared to 8% after (p<0.01). Pedestrian injuries decreased by 33.8% in the 4 intervention cities compared to no change in rest of the state (OR=0.66, 0 .45-0.96)

Study quality and conclusions

One of the few studies to show a reduction in pedestrian injuries from a school program. The data quality appears reasonable. Unfortunately, this program has never been replicated.


Author

Limbourg and Gerber, 1981

Study design and target population

Controlled trial

658 children 3-6 years old in Germany

Intervention

Behavioral training of parents consisting of film and video who in turn trained children; control group shown other film or no treatment

Outcomes

Pre- and post-intervention observations of pedestrian crossing skills of the children in real streets, including distraction by ball going into street. Two outcomes were: stopping at the curb and looking both ways before crossing

Results

Experimental group less likely not to stop at the curb than the control group both with distraction (OR=0.31, 0 .19-0.48 ) and without distraction (OR=0.36, 0.21-0.68). In other words, training reduced risk of not stopping at curb by 69% and 64% respectively.

Experimental group less likely not to look both with distraction (OR=0.55, 0.36-0.84) and without distraction (OR=0.57, 0.35-0.94). In other words, training reduced the risk of not looking before crossing by 45% and 43% respectively.

Study quality and conclusions

At baseline only 8-15% stopped at curb and only 20-25% looked before crossing. Significant placebo effect, especially for non-distracted crossing.

Study results appear to be valid, although probably represent optimal behavior.



Author

Yeaton and Bailey, 1978

Study design and target population

Before and after design.

12 children in grades K-3 (ages 5-9) at each of 2 schools

Intervention

Training session (average of 16 minutes), using modeling and practice, each day for 3 days.

Outcomes

Observation of children crossing "generalization" street.

Results

Wide variation from day-to-day on pedestrian behavior. No discernible immediate effect in 2nd-3rd grade, but some effect for kindergarten-1st grade students.

Study quality and conclusions

No control groups. Behavior on training street much better than on test street. Effects faded over time; one of the only studies to have a 1 year f/u.

Summary of pedestrian skills safety training programs

There were no randomized controlled trials of the effect of pedestrian training programs. With the exception of the Fortenberry study7, and the Roberts 1995 study, all of the studies examined pedestrian behavior as their outcome and did not examine the effect of the programs on pedestrian injuries. Some of the studies used more artificial situations to evaluate the impact of the training program than others. The effectiveness of the interventions varied considerably, from achieving near perfect pedestrian crossing skills to others in which about one-half of the children even after the intervention did not cross the streets correctly. The effectiveness of the interventions varied considerably, from achieving near perfect pedestrian crossing skills to others in which about one-half of the children even after the intervention did not cross the streets correctly.

An example of a comprehensive (multi-modal) program is the Children's Village, a life safety education facility for children evaluated by Gielen et al 1996. Programs included a broad array of educational approaches, such as individual instruction, audiovisual materials, behavior modification techniques, modeling, simulations, training in real street situations, and mass media campaigns. The evaluations suggest that: 1) educational approaches can increase children's knowledge of pedestrian safety; 2) training in real street situations has potential for changing unsafe behaviors; and parent e ducation component needed to maximize success. Unfortunately the evaluation used self-reported outcomes, not observed behaviors. This approach, while promising, requires the commitment of substantial resources. The authors point out that neither their study nor the published literature has linked improved knowledge to increased safety skills which should be measured by observed behavior or changes in injuries.

Child pedestrian injuries and deaths have decreased over the past 20 years. Several studies h ave indicated that a change in travel patterns (i.e. less walking and biking) is a major contributor to reduction in deaths.(DiGuiseppi 1997, Roberts 1995 , Roberts 1993, MacPherson 1998) These changes in exposure patterns need to be tak e n into account in evaluations of pedestrian training programs which use changes in injuries as an outcome measure. Changes in transport policy to encourage walking and cycling for short trips and bus, train, air trips for longer distances should be consi dered, (Roberts 1995). This would include environmental changes to make walking safer and a philosophical shift to value pedestrian and cyclists' convenience over that of motor vehicle drivers.

Recommendations on pedestrian skills safety programs

The programs do appear to improve child pedestrian crossing skills and can therefore be recommended. Programs which are multi-modal and programs which use parents are likely to be most successful in altering behavior. However, they are not panaceas and are not the complete answer to the pedestrian injury problem.

Recommendations for future research

There is a need for a rigorous randomized controlled trial to test the effectiveness of pedestrian skills training programs. Ideally, these should be done in large communities in which the outcome of interest could be pedestrian-motor vehicle collisions. At the least, the outcome should be observed pedestrian behavior and street crossing skills in the real world environment, in which the child is unaware he or she is being observed.