Article Text


Updating the evidence. A systematic review of what works in preventing childhood unintentional injuries: Part 2
  1. E Towner,
  2. T Dowswell,
  3. S Jarvis
  1. Department of Child Health, University of Newcastle, UK
  1. Correspondence to:
 Dr E Towner, Community Child Health, 13 Walker Terrace, Gateshead NE8 1EB, UK

Statistics from

Injuries in the home environment

Four recent studies focus on the prevention of home accidents at a general level.1–4 These papers suggest that, while educational campaigns and equipment loan schemes may be potentially effective in terms of promoting behavioural change, there is little evidence that injury reductions are achieved by these means (table 1).

Table 1

The home environment: prevention of general home accidents

There is increasing evidence from the United States of the positive effect of campaigns promoting the use of smoke alarms (table 2).5–9 A smoke alarm giveaway programme in the central area of Oklahoma City, where there was a high fire risk, showed an 80% annual injury rate decline from 15.3/100 000 to 3.1/100 000, compared with a slight increase in the rest of the city.5 The authors point out that, “interventions that target areas with high rates of fires may be especially efficient ways to lower the incidence of injuries and deaths from residential fires”. This is the first community based study to demonstrate the effectiveness of a smoke alarm promotion programme on health outcomes. There is also evidence that smoke alarm promotion programmes lead to changes in behaviour which are sustained over a long period of time (3–4 years), resulting in greater numbers of households with functioning smoke alarms.7

Table 2

The home environment: prevention of burns and scalds

Additional evidence related to the prevention of childhood poisoning comes from a long term follow up of legislation in the United States introduced in 1974, which required child resistant packaging for all prescription drugs.10 For children aged 0–4 years the mortality rate for oral prescription drugs declined from 3.5 per million in the late 1960s (before legislation), to fewer than 2 deaths per million in the early 1990s. This represented 460 fewer deaths for the period 1974 to 1992 (table 3).10,11

Table 3

The home environment: prevention of poisoning

Injuries in the leisure environment

There remains very little evidence regarding the effectiveness of health promotion in preventing injuries in the leisure environment (table 4).12–16

Table 4

The leisure environment: prevention of leisure injuries

An Australian study of an educational intervention, directed at parents and teachers, aimed to increase compliance with safety standards in school playgrounds. There was a small improvement in the number of hazards observed in school playgrounds after the intervention.14 A study conducted in Wales evaluated the effect of environmental changes to playgrounds, including increasing the depth of bark surfaces in some playgrounds and changes in playground equipment. Changes in injury rates and in rates of fractures were noted.16

Two papers describing interventions, set within the leisure environment, by D'Argenio et al12 and Malinowska-Cieslik and Borne13 provide examples of innovative content. These two interventions highlight the importance of the cultural setting of some interventions.

Community-wide studies

Five recent papers evaluate community based interventions (table 5).17–21 The findings of these studies are not simple to interpret. While there is some evidence that such programmes can result in health gain, the quality of the evidence is limited.

Table 5

Community based studies

Mass media interventions

Table 6 summarises findings relating to mass media and more general training events.22–24 Exhibitions and public information campaigns can increase knowledge, but there is no new evidence that such approaches have any impact on injury rates.

Table 6

Mass media general interventions

New target groups and implementation strategies

In our first systematic review of childhood injury prevention studies, published in 1993, very few intervention studies targeted deprived groups in society, despite the strong associations between social deprivation and childhood injury.25 Our second review, published in 1996, contained more studies which targeted social deprivation.26 This trend has continued in the current review. Between 1995 and 1998 studies which have targeted deprived groups or communities include Thomson and Whelan's evaluation of practical roadside training of children in a deprived area of Glasgow.27 Bicycle helmet educational campaigns targeting more deprived groups include programmes evaluated in Canada28 and the United States.29,30 Economic barriers to helmet purchase were recognised in several programmes where helmets were either distributed free or discount vouchers provided. Mallonee et al's evaluation of a smoke alarm giveaway programme in a high risk, central city location5 and four other studies in the home safety field targeted low income families: Bablouzian et al in the USA who evaluated a community based home hazard reduction programme,1 Clamp and Kendrick's study of general practitioner safety advice and provision of safety equipment,2 Kendrick et al's package of home safety interventions,4 and Thompson et al's study of home safety equipment loan schemes.3

We found more evidence of educational interventions which targeted very young children. For example Britt et al targeted children aged 3–4 years in a classroom intervention to increase bicycle helmet use.29 McConnell et al evaluated the effect of a fire safety programme in the classroom on the safety knowledge of 3–5 year old children.6 Interestingly what emerged in this programme was that knowledge changes were greatest in the youngest age group. What has not been demonstrated, however, is whether such knowledge change leads to changes in behaviour and, at a more general level, doubts remain about the benefits of education in such young children.

Other programmes have employed innovative approaches to deliver programmes. These include Thomson and Whelan's pedestrian training programme in Glasgow where parent volunteers have been recruited to train children (other than their own) in developing safer pedestrian skills.27 In some interventions the focus has not been children or parents. Targeting interventions at teachers and childcare staff (along with parents) achieved modest reductions in playground hazards in an Australian study.14 Professionals were also targeted in a UK study in an intervention aiming to increase knowledge among health care staff.24

Additions to the way interventions have been evaluated

The study by Ni et al illustrates how a greater use of data collection techniques has been employed in evaluating interventions.31 This utilised statewide observation of bicycle helmet use, local observations of use, and self report of helmet use by both children and their parents. This range of sources of data allowed greater confidence in the results. Another study of bicycle helmet promotion, this time targeted at preschool children, used home visits to observe young children playing on bicycles and wearing their helmets.29 Observations of helmet wearing are difficult to conduct in this age group and this study represents an innovative approach to attempt to capture this information. A study in the UK evaluating the effects of changes to playgrounds on childhood injuries, included the use of exposure data.16 This allowed the rates of injury to be calculated based on a unit of exposure. More specific data on injury type (fractures) were also included in this study. A community based programme from Motala, Sweden, provides the first example in the literature of a study using injury severity data as an outcome.20 In the UK, Kendrick et al's trial of a package of interventions delivered in a primary care setting, used both frequency and severity of medically attended injuries as an outcome measure.4

Hazinski et al's study of a Children's Traffic Safety Programme made an attempt to assess the level of programme implementation (that is, the degree to which teachers delivered the programme as it had been planned) and compared this with the outcomes achieved.32 Schools with good programme implementation achieved better results.

Discussion and conclusions

The review drew on the world literature on child injury prevention. The search strategy attempted to include a range of databases, including the Transport Research Laboratory road safety database and also included the reference lists of a range of recently published articles. However, some areas, such as product safety and engineering, may be under-represented, leading to some bias in the results we have reported.

However, the studies published are not evenly spread across the various injury types and do not reflect the injury burden. The studies reviewed included a relatively large number (10/42) relating to the use of bicycle helmets. The prevention of child pedestrian injuries was addressed in a relatively small number of studies (and exclusively in only one study) despite the fact that it remains the main cause of child injury death. This bias in the literature may reflect the fact that some injury areas can be researched relatively easily compared with others. Injuries where there is a simple, single intervention (such as a bicycle helmet) that can be evaluated relatively simply within closed systems (for example, schools) are more likely to receive research attention.26

Recent studies have included a larger concentration of studies focusing on high risk or socially deprived groups. These studies represent an important addition to the literature. In one such study, parent volunteers carried out the child pedestrian training programme. Using locally available resources may be more likely to result in low cost and sustainable child injury prevention programmes. As in our earlier review, the cost of the interventions was rarely considered in the studies included here.

The quality of the evidence was very mixed. Fewer than a third of the studies used research designs where the evidence was rated as good/reasonable. The remaining studies had weaker designs and it was therefore more difficult to interpret and have confidence in the results. Experimental methods are not always appropriate to evaluate injury prevention programmes, especially where more than one injury type is targeted or where interventions are aimed a large groups. Nevertheless, the inclusion of appropriate control groups, well defined target groups, and an adequate sample size increases the strength of the evidence. Few studies included process information; such information is useful to understand issues such as programme reach and impact and assists in study replication in different settings.

Given the complexity of the injury problem, there are unlikely to be simple solutions which result in dramatic changes in injury rates. Over the past decade, our knowledge has increased incrementally and the last four years has seen the publication of further evidence on strategies to reduce the injury burden. There remains a need for further research. A concerted attempt is needed, however, to implement established interventions, both nationally and locally.

The results are summarised in table 7.

Table 7

What interventions work? (that is, reduce injury or change behaviour). Bold type indicates areas where there is new evidence or increased evidence, italic = injury reduction, and roman = behavioural change


This project has been funded by England's NHS Executive National R&D Programme in Mother and Child Health (CH 10–21).


View Abstract


  • Part 1 of this paper was published in June (Inj Prev 2001;7:161–4)

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.