Unintentional injuries: prevention strategies for under 15s

Evidence statement 1.1

Three (+) international comparison studies show a lack of comparable in-depth information on exposure to risk to help in analysis of the relative impact of different legislative, regulatory, enforcement and compliance interventions.

Evidence Statement 1.2

Two ecological studies (one [+] and one [-]) in high income countries were unable to associate variations in child morbidity and/or mortality rates across countries to differences in legislation, regulation, enforcement and compliance for road environment modification, road design, home and leisure environment interventions. However for road safety, evidence from two ecological studies (one [+] and one [++]), suggest a weak trend towards better performing countries (in terms of child fatality rates) having more road environment modification and road design measures in place.

Evidence Statement 1.3

Evidence from one (++) ecological study indicates that differences in the distribution of exposure in the road environment for child pedestrians (in particular relating to time spent near busy main roads) can explain some of the difference in severe child injury and fatality rates between Great Britain and two other northern European countries, France and the Netherlands.

Evidence statement 2.2

There is evidence from 10 studies (one UK). There is evidence of a strong association (that is, relative risk equivalent of greater than 2.0) of injuries being associated with travelling in a car driven by a non-sibling teenager. There is evidence of weak to moderate association (that is, relative risk equivalent of greater than 1.0 to less than 2.0) of injuries with lower parental income, employment status, educational status, socioeconomic status, and with travelling in a car with a female driver (when the injured child was appropriately restrained). The increased risk in females may well reflect their longer periods of time in the presence of children. There is mixed evidence regarding the association of injuries with ethnicity.

Evidence statement 2.3

There is evidence from 18 studies (five UK). There is evidence of a strong association between the lowest socioeconomic quintiles, being of Native American descent (for pedestrians), having parents who were migrants, hyperactivity, behavioural difficulties, or bicycle riding (riding slowly or only on the pavement) and injuries. There is evidence of weak to moderate association of injuries with membership of the second socioeconomic quintile, social deprivation, non-professional parental occupation, rural and mixed-urban environments, being male, or behavioural disorders. There was no statistical evidence of injuries being associated with social fragmentation or ethnicity (for cyclists).

Evidence statement 2.4

There is evidence from seven studies (one UK). There is evidence of weak to moderate association of injuries with socioeconomic deprivation and being African-American. There is mixed evidence regarding the association of socioeconomic status (measured by parental occupation) with injuries. There was no statistical evidence of injuries being associated with autism.

Evidence statement 2.5

There is evidence from six studies (one UK) on burns and fire in the home of a strong association between child's age (less than 1 year), low mother education and age, and areas of concentrated poverty (and high numbers of African-American population) and injuries. There is evidence of weak to moderate association of burn injuries with children being male, from an ethnic minority, having behavioural problems and a poor reading score, low parental education, lower home income, a larger number of children in the home, and rural location. There was no statistical evidence of burn injuries being associated with type of home ownership.

Evidence statement 2.7

There is evidence from three studies (none UK) on falls in the home of a strong association between greater child's age (older than 1 year) and injuries. There is evidence of weak to moderate association of injuries with: being male, of African-American descent, families being in receipt of social welfare benefits, lower educational status of parents, lower income, single parent households, lower mother's age at childbirth, non-owner housing occupancy, living in a flat or farmhouse, older housing and being a migrant. Being lone parent status, neighbourhood poverty and living in cities were not statistically associated with falls.

Evidence statement 2.8

There is evidence from seven studies (one UK) on poisoning in the home of a strong association between child's age (from 1 to 4 years), behavioural problems, and autism and injuries. There is evidence of weak to moderate association of injuries being associated with: being male, having a lower reading score, lower educational status of parents, lower income, larger families, being in receipt of social welfare benefits, younger age of mother at childbirth, being of Native American descent, living in the country, and the birth of a sibling within 12 months (for iron tablet poisoning). There was no statistical evidence of injuries beingassociated with single parent households, family size, overcrowding, or house type.

Evidence statement 2.9

There is evidence from two studies (one UK) on undefined causes of injury in the home of weak to moderate association of injuries with lower educational status of parents and lower family income. There was no statistical evidence of injuries being associated with parental marital status or of being in receipt of social welfare benefits.

Evidence statement 2.10

There is evidence from four studies (none UK). There is evidence of a strong association between the use of public playgrounds or being of African-American descent and injuries. There is evidence of weak to moderate association of injuries being with being of Latin American descent, location of a school within an urban area, schools with larger numbers of classes (greater than or equal to 24), longer school hours, and the levels of physical activity engaged in outside of school. There was no statistical evidence of injuries being associated with the levels of physical activity engaged in within school.

Evidence statement 2.11

There is evidence from six studies (one UK) on burns and fire in all environments of a strong association between the most socioeconomically deprived families, living in a house with one to three or more bedrooms, attention deficit hyperactivity disorder (ADHD), and being of Native American descent and injuries. There was no statistical evidence of injuries being associated with autism, having previously endured an unintentional burn/fire injury, parental employment status, entitlement to Medicaid, or order of sibling birth.

Evidence statement 2.12

There is evidence from three studies (none UK). There is evidence of weak to moderate association of injuries with entitlement to Medicaid (in children aged 5 to 14 years) and with non-entitlement to Medicaid (in infants aged 0 to 4 years). There was no statistical evidence of injuries being associated with being of Native-American descent or the presence of behavioural disorders.

Evidence statement 2.14a

There is evidence from 12 studies (four UK) on all injury types in all environments of a strong association (compared with newborns aged up to 6 weeks) between children aged 7–24 months and injuries. There is evidence of weak to moderate association of injuries with increasing age (4 years or older versus younger than 4 years), children aged 15–54 months (versus younger than 6 months), and increasing age among children with a disability. There was no statistical evidence of injuries being associated with increasing age in the case of head injuries.

Evidence statement 2.14b

There is evidence from 16 studies (four UK). There is evidence of weak to moderate association of injuries (of all severities, including fatalities) with being male.

Evidence statement 2.14c

There is mixed evidence from eight studies (one UK) on ethnicity in all injury types in all environments regarding the association of child ethnicity with injuries. There is evidence of weak to moderate association of injuries with being of black or Native American descent. There was no statistical evidence of injuries being associated with being of Asian descent or a wide range of other ethnicities.

Evidence statement 2.14f

There is evidence from 27 studies (six UK) on family's socioeconomic status in all injury types in all environments of weak to moderate association of injuries with socioeconomic deprivation. There is no statistical evidence of injuries (reported in some studies) being associated with socioeconomic deprivation within certain age categories. There is mixed evidence regarding the association of parental educational attainment and household income with injuries.

Evidence statement 2.14i

There is evidence from eight studies (four UK). There is evidence of weak to moderate association of injuries with socioeconomic deprivation, but no evidence of association between other indicators of neighbourhood disadvantage and the occurrence of unintentional injuries.

Evidence Statement 3.1

This evidence statement differs from the one in the report submitted to NICE. It has been amended to include findings from 1 (++) systematic review that was included in the original report and has since been updated. The updated review is Wilson C et al. 2010. Speed cameras for the prevention of road traffic injuries and deaths (review). Cochrane Library: 10.

There is moderate evidence from three recent systematic reviews (one [++] and two [+]) that road speed enforcement devices (cameras, lasers or radar) reduce road injuries, and serious/fatal injury crashes/collisions in the vicinity of the devices. One systematic review (+) also concluded that similar size of speed reduction effects were observed over wider geographical areas around the enforcement device sites. The size of the observed reductions in different studies, and in different localities within studies, varies considerably. Similarly, one systematic review (++) found that in those studies where enforcement devices were temporarily placed at certain locations, the duration of speed reductions after removal of the devices (the 'time halo') varied from 1 day to 8 weeks. However, only one of the systematic reviews (++) was able to identify any factor which was consistently associated with higher injury or crash reductions – this was that the effect on urban roads was greater than that on rural roads. There was insufficient consistency between studies to enable the detection of the effects of other factors (such as different roads user groups, automated versus non-automated detection, mobile versus fixed, covert versus overt, or other roads versus motorway.). The greater effect on urban roads where children are more likely to be pedestrians is relevant. Included studies did not consistently state what the penalties or fines would be for detected speeding, although one systematic review (++) implied there was a relationship between size of pre- and post-reduction in speeding vehicles and the speed threshold set.

This evidence is judged as directly applicable to the UK as the results from the UK studies were generally consistent with the studies from other developed countries.

Evidence Statement 3.2

There is weak evidence from three controlled before-and-after studies (in Australia, Israel and California) that increased or rationalised police enforcement of traffic speeds reduces injury crashes (two [+] and one [-]). There is also weak evidence from three multivariate analyses of longitudinal road accident/injury data (in New Zealand, California and Greece) that increased levels of police enforcement of traffic speeds reduces injury crashes and all injuries (two [+] and one [-]). There is also moderate evidence from one (+) controlled before-and-after study, on motorways in the Netherlands, that increasing the intensity of enforcement – from apprehending 1 in 100 speeding offenders, to 1 in 25, to 1 in 6 – produces statistically significant (p less than 0.05) reductions in mean speed (1 km per hour for 1:25 versus 1:100; and 3.5 km per hour for 1:6 versus 1:25).

This evidence is judged as partially applicable to road safety policy in the UK. This is because in the included studies there are a number of differences in the way police forces are organised and contribute to speed enforcement. Also, in the role of the police in enforcing speed limits through speed traps and mobile cameras/radar needs to be considered in the context of the widespread use of fixed site automated cameras around the UK road network.

Evidence statement 4.1

There is evidence from one controlled before-and-after study (+) in the USA that law requiring the installation of smoke detectors, increases the number of houses which have at least one functioning smoke detector and that this may reduce fatalities related to fires in targeted properties.

Knowledge of the law and the penalty for non-compliance may be associated with greater smoke detector installation than knowledge of the law only.

The law assessed required smoke detectors in all bedroom areas of one-, two- and multi-family dwellings, applied retrospectively to homes built prior to the law, and can be enforced by a fine or jail time. In addition, sale of a property is contingent on appropriate smoke detectors being present.

Given the differences in legal systems, responsibilities and enforcement between the USA and the UK, and the high socioeconomic status of the studies communities, the applicability of this finding has been assessed as poor. However, the observations that systems of enforcement which involve regular inspection, with a system of warnings prior to prosecution are effective; that laws which reflect societal laws are effective and that media campaigns to support the introduction of new laws may be important, may be applicable across other settings.

Evidence statement 4.2

There is evidence from one comparative study in the USA (+) that window guard legislation in New York City reduces child injury related to falls from buildings by about half, despite greater numbers at risk as residents of multiple-family dwellings (1.5 per 100,000 children aged 0–18 years compared with an average of 2.81 per 100,000 in 27 other US states without legislation, and 3 per 100,000 in Massachusetts which introduced interventions without legislation). The law required owners of multiple-family dwellings to provide window guards in apartments where children aged 10 or under lived (half the injuries recorded in NYC were in those aged 11–18). Compliance was subject to annual enforcement. The introduction of the law was accompanied by a coordinated education and advertising programme ('Children can't fly') which involved outreach, dissemination of literature, a media campaign and the distribution of free window guards.

Given the differences in legal systems, responsibilities and enforcement between the USA and the UK, and the differences in housing stock and management, the applicability of this finding has been assessed as poor. However, the observation that effective enforcement is a key element of legislative success may be applicable across a range of settings.

Evidence statement 4.3

There is mixed evidence from four uncontrolled before-and-after studies (all [+], two from the US and two from Australia) about hot water tap temperature legislation. Two studies (one US and one Australia) reported that the annual incidence of burn injuries in children aged 4–13 years increased after the introduction of legislation, and a US study found that injury rates were raised compared to the period immediately prior to legislation being introduced but fell in relation to an earlier comparator time-period. Only one Australian study (+) reported p-values, but this was a significant increase (p = 0.01).

One study (Australia) suggested there may be a decrease in the number of scald injuries in children aged 0–4 years, however, the reported differences were non-significant (p=0.57).

Given the differences in legal systems, responsibilities and enforcement between the USA and Australia and the UK, and the differences in housing stock and management, the applicability of these findings have been assessed as poor. However, the observation that legislation aimed at safety in the home may be limited in its effectiveness where it is implemented only in that housing stock where access and enforcement is easier (such as in rented or newly built accommodation only), may be applicable across a range of settings.

Evidence statement 4.4

There is mixed evidence from four studies (two case control, and two comparative) about swimming pool fencing legislation (two [+] one from USA and one from Australia and two [-] one from New Zealand and one from Australia).

Two studies (both [+], one USA and one Australia) suggest that legislation is ineffective where it only requires three-sided fencing. The US study suggests no impact of such legislation on drowning in children aged younger than 10 years compared to no legislation (odds ratio [OR] 1.27, 95% confidence interval [CI] 0.72 to 2.25). The Australian study found the incident rate ratio of drowning in children aged younger than 5 years living in houses with three-sided rather than four-sided pool fencing was 1.78 (95% CI 1.14 to 1.79).

Three studies, two (-) and one (+) (two Australia, one New Zealand) report on outcomes related to legislative management and compliance.

The New South Wales study (-) found that a more structured and comprehensive approach to inspection (including a register of owners, annual inspections, and enforcement of the act including fines) resulted in twice the level of compliance as those with less structured or detailed approaches. Key informant interviews also suggest that lack of clarity in the Fencing Act, and failure to detail how councils should ensure compliance, including how it should be funded, hampered effective implementation.

The Western Australia study (+) suggests that compliance is highest immediately after legislation is introduced, and falls off thereafter, although regular inspection enhances compliance. The New Zealand study (-) found no association with compliance rates and: local authorities having written policies about locating and inspecting pools; a re-inspection programme; or advertising of pool owners' obligations under the relevant act.

Given the differences in legal systems, responsibilities and enforcement between the USA, Australia, New Zealand and the UK, and the low level of private swimming pool ownership in the UK, the applicability of these findings have been assessed as poor. However, some key lessons from these studies may be applicable across a range of settings, such as: the importance of adequate legal requirements in order to glean maximum benefit (as illustrated by three- versus four-sided fencing here); the need for regular inspection regimes which are consistently enforced, and the related need for clear lines of responsibility and sufficient funding for these; the need for concurrent education to help owners comply with the spirit as well as the letter of the law (for example, the need for maintenance of equipment, and the valuing of safety over convenience) and finally the need for legislation which does not contradict or confuse other existing rulings.

Evidence statement 5.3

There is moderate-to-weak evidence from two controlled before-and-after studies (one [+] and one [-]) to show that mass-media campaigns, employed as part of a broader non-legislative strategy (that involved educational programmes and purchase subsidies) were effective in increasing compliance with bicycle helmet use. There was also moderate evidence from uncontrolled before-and-after data from one of the studies (-) that the programmes helped to reduce the rates of bicycle-related head injuries in the intervention area.

In the US study (+), the sales of one brand of a youth helmet in the Seattle area (intervention area) rose from 1,500 to 22,000 over a 3-year period (no figures stated for the control area) while observed helmet usage rate among school-age children increased from 5% to 16% compared with a rise of only 1% to 3% in a control community, Portland, Oregon, over the same period.

In the UK study (-) self-reported helmet use among young people aged 11–15 years living in the campaign area increased from 11% at the start of the campaign to 31% after 5 years (p < 0.001), with no significant change in the control group. Hospital casualty figures in the campaign area (Reading) for cycle-related head injuries in the under 16 years age group, fell from 112.5 per 100,000 to 60.8 per 100,000 (from 21.6% of all cycle injuries to 11.7%; p < 0.005). No injury data were provided for Basingstoke, the control. Applicability: The evidence is judged to be directly applicable to the UK – one of the studies was carried out in the UK and although the other was carried out in the US, it was embarked upon and completed before the introduction of a bicycle helmet legislation, so in a sense the settings reflected what is currently obtainable in the UK, a country without mandatory helmet wearing legislation. Furthermore, both countries are similar in terms of living standards and economic development.

Evidence statement 5.4

There is mixed evidence from two controlled before-and-after studies (both [-], one from Canada and one from the UK) that removal and replacement of unsafe equipment to comply with regulatory standards is an effective strategy for preventing playground injuries. The Canadian study demonstrated statistically non-significant reduction in equipment-related injury rate in the intervention schools after replacement of equipment using the new Canadian Standards Association standards (relative risk [RR] = 0.82 to 0.66 to 1.03). This translated into 177 equipment-related injuries avoided during the study period. The comparable equipment-related injury rate in the non-intervention schools increased by about 15% after the study period, although not statistically significant (RR = 1.15; 95% CI 0.96 to 1.37). The overall injury rate reduced in the intervention schools (RR = 0.70; 95% CI 0.62 to 0.78) and increased in the non-intervention schools (RR = 1.40; 95% CI 1.07 to 2.53) after the study period. However, in the UK study, injury rate per observed child was significantly reduced in the five playgrounds where changes (use of greater depth of bark and replacement of overhead horizontal ladders with rope climbing frame) had been made compared to the control playgrounds without changes.

Applicability: The non‑UK study is only partially applicable to the current UK context due to similarities in level of economic development, nature of the playgrounds, as well as targeted populations. The UK study findings are directly applicable.

Evidence statement 5.5

There is weak evidence from two before-and-after studies (one [-] and one [+], from UK and Italy) and one retrospective time series (one [+] from UK) on the effect of fireworks legislation and enforcement activities on firework-related injuries.

One study in Italy (+) reported that a comprehensive, multifaceted programme, comprising the combination of enforcement of fireworks law, media campaign and education, reduced the rate of fireworks-related injury from 10 per 100,000 before the intervention programme to 6.1 per 100,000 after it was implemented, and a time-series based study found that amendments to restrictive fireworks legislation led to a reduction of firework-related injury in children.

The study from Northern Ireland (-) did not find a significant increase in fireworks-related injuries requiring hospital admission following liberalisation of the law on fireworks sale (incidence of admissions before: 0.38 per 100,000; after: 0.43 per 100,000). However, the annual number of injuries in this study was already very small relative to annual variations.

Applicability: The Italian study is partially applicable to current UK context while the UK findings are directly applicable. However, the Northern Ireland study may not be directly applicable to the rest of UK because of the civil unrest reported in that part of the kingdom.

Evidence statement 6.5

There were two cost-benefit analyses which assessed the impact of speed enforcement programmes. The photo radar programme in British Columbia was estimated to produce net benefits to society of about C$114 million (in 2001), and still produced substantial net savings of C$38 million if only considered from the provincial insurance corporation's perspective.

Similarly, the 420 automated speed camera sites in the UK in 1995/6 were estimated to have a positive net present value of over £26 million, even after 1 year, rising to £241 million after 10 years. This is because annualised fixed costs of £5.3 million plus annual recurrent costs of £3.6 million, would be offset not just by the £6.7 million in fine income, but also the over £30 million in the estimated annual value to society of accidents avoided. In all ten police force areas there was a positive net present value (that is, benefits exceeded costs) within a year of the programme starting.

These older findings should be seen as having been superseded by the more recent study for the Department for Transport, which evaluated the national safety camera programme. (This study was added to the review after the original report was submitted to NICE.)  In this study, it was estimated that there would be 4230 fewer personal injury collisions (any road collision which results in at least one casualty, whether fatal, serious or slight) annually as a result of the safety cameras across all 38 safety camera partnerships.  At an estimated value of £61,120 per collision avoided (using Department for Transport standard estimates for 2004) this means an annual estimated economic benefit of £258 million. This compares with the total annual cost of the programme of £96 million. Comparing only the revenue costs per collision prevented (£61,120) with the corresponding economic benefit per collision due to injuries prevented (£22,653), over the four years, gives a cost–benefit ratio of approximately 2.7:1. They also use data from both speed and red light camera sites, although at speed camera sites the reductions in personal injury collisions were associated with reductions in speeds.

Expert testimony

Expert testimony 1: 'Child road safety' (including 'Child casualties in road accidents: 2007. Road accidents factsheet number 5 [2009]' [Department for Transport])

Expert testimony 3: 'Inequities in child injuries'

Expert testimony 6: 'Monitoring and surveillance issues – A&E pilot'.