• Open Access

Increase in best practice child car restraint use for children aged 2–5 years in low socioeconomic areas after introduction of mandatory child restraint laws


Correspondence to: Dr Julie Brown, Neuroscience Research Australia, Barker St, Randwick, NSW 2031; e-mail: j.brown@neura.edu.au


Objectives : To examine changes in child car restraint practices in low socioeconomic areas following the introduction of mandatory child car restraint legislation in New South Wales (NSW), Australia.

Methods : Data from two cross-sectional studies of child car restraint use at pre-schools, early childhood centres and primary schools before and after the introduction of legislating mandatory age-appropriate car restraint use for children up to the age of seven years was used in this analysis. All included observations were from local government areas with socioeconomic status in the lowest 30% of urban Sydney. Children aged 2–5 years were observed in their vehicles as they arrived at observation sites (107 pre-legislation, 360 post-legislation). Multilevel logistic regression was used to examine changes in observed age-appropriate and correct use of car restraints.

Results : Age-appropriate car restraint use was higher post-legislation than pre-legislation. After controlling for child's age, parental income, language spoken at home and adjusting for clustering, the odds of children being appropriately restrained post-legislation were 2.3 times higher than in the pre-legislation sample, and the odds of them being correctly restrained were 1.6 times greater.

Conclusions : Results indicate an improvement in car restraint practices among children aged 2–5 in low socioeconomic areas after introduction of child restraint laws.

Implications : Despite improvements observed with enhanced legislation, further efforts are required to increase optimal child car restraint use.

Injury is the leading cause of death among Australian children aged 1–14 years and accounts for more than 10% of all Australian childhood hospital separations.1 In 2006/07 almost three-quarters of injury-related deaths involved children who were car occupants,1 and about 1,500 child occupants were seriously injured.2 The use of safety restraints when travelling in a car, regardless of the type of restraint, is known to significantly reduce the risk of death and serious injury.3 In 2006, Brown et al. reported sub-optimal child restraint practices were widespread among restrained crash-involved child occupants aged 2–8 years in New South Wales (NSW), and this was associated with a higher risk of serious injury.4 There is considerable scope to prevent these injuries by encouraging best practice child car restraint use.

Optimal child car restraint use requires that a child is restrained in a system designed for their age and size (appropriate restraint use) used in accordance with the manufacturer's instructions (correct use). North American studies have quantified the reduced risk of death and injury in children using car restraints appropriate to their age and size.5–7 Errors in the way restraints are used reduce the effectiveness of a restraint system in providing protection to an occupant in a crash.8–11 In 2008, about half of all restrained children aged 0–12 years were not using size-appropriate restraints12 and similar levels of inappropriate restraint use were reported in other states.13,14 The NSW study also examined correct use and found less than a quarter of restrained children (0–12 years) were correctly using size-appropriate car restraints.12

A number of family-level characteristics have been reported to be associated with an increased likelihood of sub-optimal child car restraint use. Low parental education, low income status, large family size and a non-English speaking background have been highlighted as specific indicators of the risk of not correctly using size-appropriate restraints in Australia and elsewhere.13–19 Additionally, the laws at the time of the 2008 Australian studies did not require age-appropriate restraint use and this likely contributed to high rates of inappropriate restraint use. Australian children have been required by law to use a seat belt or child restraint when travelling in a car since the late 1970s.20 For children over 12 months, however, the law did not stipulate the type of restraint to be used. This was addressed when the Australian Road Rules were revised21 and the current laws were enacted in all states and territories except the Northern Territory in 2009/10 to specify the use of age-appropriate restraints for children up to seven years of age. The new laws were implemented in the Northern Territory in 2013. While increases in appropriate restraint use have been observed after the introduction of legislation in other jurisdictions for children aged less than four years,22–24 this has not always been the case.24,25 A US study found self-reported child restraint use did not change significantly in a low socioeconomic sample of families after the introduction of legislation specifying age-appropriate restraint use. Furthermore, they found a significant increase in premature graduation to booster seats from forward-facing child restraints.25

Legislation requiring the use of specific types of car restraints by children of specific ages or sizes does not necessarily affect correct use of those restraints. Brown et al.12 noted that incorrect use was higher among Australian children using size-appropriate restraints and suggested this was due to the opportunity for more errors in child restraint systems than in seat belts. However, others have reported improvements in correct use with strengthened legislation.22–24

We hypothesised that recent amendments to Australian legislation would have increased the use of age-appropriate car restraints across all sectors of the community. As families from low educational, low-income and non-English-speaking backgrounds were previously seen to have the lowest compliance with optimal child restraint practices, we focused our analysis on this sector of the community. This study, therefore, aimed to examine changes in child car restraint practices in low socioeconomic areas following the introduction of mandatory child restraint legislation in NSW, Australia.


Child restraint practices among children aged 2–5 years from areas with Socio-Economic Index for Areas (SEIFA) lower than 1,010 were compared before and after introduction of the new Australian legislation mandating age-appropriate restraint use up to age seven. A SEIFA less than 1,010 indicates an area in the lowest 30% SEIFA of metropolitan Sydney.26 The total population of NSW is just under seven million and children under nine years of age represent about 13% of the population. Almost two-thirds (64%) of the NSW population resides in the greater Sydney area.27


The pre-legislation sample was drawn from data collected during a 2008 cross-sectional population-representative observational study of child restraint practices across NSW.12 Sample composition is described in detail elsewhere.12 In brief, children aged 0–12 years were observed in vehicles arriving at randomly selected early childhood health clinics, pre-schools, day care centres and primary schools within randomly selected local government areas (LGAs) across NSW. During that study, data for 501 children were collected from 40 observation sites. For the current analysis, all observations for children aged 2–5 years from sites within LGAs with SEIFA <1,010 were selected.

Post-legislation data were obtained from a cluster randomised trial of an intervention designed to improve child restraint practices in pre-school-aged children across Sydney during 2010.28 This study involved observation of 689 children aged 2–5 years at 27 community-based pre-schools and long day care centres within LGAs with SEIFA <1010. The intervention was delivered at 13 sites, and 14 sites acted as controls. For this analysis, all observations for children attending the control centres were selected.

Data collection and observation methods

The data collection and observation methods used by Brown et al.12 (pre-legislation sample) and Keay et al.28 (post-legislation sample) were identical. Trained researchers attended the observation sites during the morning drop-off period and potential participants were approached as they arrived in their vehicles. If the driver agreed to participate, initial observations of the child in the vehicle were made using a structured method. If more than one child met the age criteria, the child with the most recent birthday was selected. Parents were interviewed and the restraint system was inspected following a detailed pro-forma. Data collected included the child's age and gender, type of restraint, whether or not any errors in use were present and the nature of any errors. Demographic data collected from the parent included family income, education level and whether English was spoken at home.

Data collection methods were approved by the UNSW Human Research Ethics Committee12 and the University of Sydney Human Research Ethics Committee.28

Variable descriptions and definitions

Raw data from both studies were re-coded and categorised as follows. Age was truncated and grouped as 2–3 or 4–5 years old. In the pre-legislation sample, data were collected using three parental income categories (in Australian dollars) of <$50,000, $50,000–$100,000 and >$100,000. In the post-legislation sample, categories in increments of $20,000 – from less than $20,000 to more than $100,000 – were used. For the purposes of these analyses, income was dichotomised as low (<$50,000 pre-legislation and <$60,000 post-legislation groups) and high. Categories used for the ‘language’ and ‘education’ variables were identical in the two samples. Participants who spoke English at home were coded as ‘English’, and the rest were coded as ‘other’. For education, participants were grouped into those who had completed secondary school or less, and those who had completed some post-secondary education.

Restraint type was coded as: unrestrained; forward facing restraint; booster seat (used with a seat belt or accessory harness); accessory harness (used without a booster); or seat belt (includes both lap only and lap/shoulder belts).

Two outcome variables were constructed. Restraint use was coded as appropriate or inappropriate according to the current Australian Road Rules. Restraint use was judged appropriate if children aged two to three years used a forward facing restraint, and children aged four years or older used a forward facing restraint or a booster seat. Restraint use was also coded as correct or incorrect. Correct use was defined as use and installation as per the manufacturer's instructions, but minor errors that were unlikely to influence the level of crash protection were ignored. Error severity ratings were based on definitions described by Brown et al.29 Examples of minor errors are having less than 25 mm of slack in the restraint and/or anchorage system. Unrestrained children were coded as inappropriately and incorrectly restrained.

Data analysis

Child age, family income, education level and English language in the pre- and post-legislation samples were compared using the chi-square test as these have been previously found these to be associated with the outcome variables.15,31,32 We then used multilevel logistic regression (PROC GLIMMIX, SAS 9.1, SAS Institute Inc. 2008, Cary, NC) to examine the difference between the pre-and post-legislation proportions of appropriate restraint use and correct restraint use while controlling for those potential confounders found to be significantly different in the pre- and post-legislation samples. We adjusted for clustering of children by site using a random effect. Non-response and refusals in the family education level and English language variables were treated as missing data. As income was missing for more than 10% of the sample, this was coded as an ‘unknown’ category and included in modelling. Restraint type was not included in the models because it was closely associated with child age. Changes in restraint type distribution among those using forward facing child restraints, boosters and seat belts between the pre-and post samples were examined using the chi-square test.


There were 107 children from 15 pre-legislation sites and 360 from 14 post-legislation sites (Table 1). The pre-legislation sample had younger children, fewer families who spoke a language other than English at home and fewer low-income families.

Table 1.  Comparison of characteristics of the pre-and post-legislation samples.
N=107 (%)
N=360 (%)
p value
Age (years)2–3
59 (55)
48 (45)
136 (38)
224 (62)
70 (65)
27 (25)
10 (9)
161 (45)
142 (39)
57 (16)
103 (96)
1 (1)
3 (3)
282 (78)
142 (39)
19 (5)
62 (58)
40 (37)
5 (5)
218 (61)
118 (33)
24 (7)
Observed restraint use     
54 (51)
53 (50)
266 (74)
94 (26)
39 (36)
68 (68)
168 (47)
192 (53)

The proportion of children using age-appropriate restraints was greater in the post-legislation sample, as was the proportion observed to be using their restraint correctly (Table 1).

After controlling for child age, income, language spoken at home and adjusting for clustering of observations, the odds of children being appropriately restrained were 2.3 (95%CI 1.1–5.0) times greater in the post-legislation sample than in the pre-legislation sample, and the odds of them being correctly restrained were 1.6 (95%CI 1.5–4.1) times greater in the post-legislation sample (See Table 2). As there was a cell with one observation in the language variable, the model was re-run with this variable removed. This made little difference to the estimates (OR 2.7, 95%CI 1.3–5.3 for appropriate restraint, and OR 1.5, 95%CI 1.0–2.2 for correct restraint).

Table 2.  Adjusted and unadjusted odds ratios and 95% confidence intervals for appropriate versus inappropriate and correct versus incorrect restraint use. Random effect was used to account for clustering (n=22 for missing data).
VariableOdds ratio for appropriate restraint (95%CI)Odds ratio for correct restraint (95%CI)
  1. * Adjusted for all other variables shown

Post- legislation (vs pre-legislation)
3.6 (1.8–6.9)
2.3 (1.1–5.0)
1.6 (1.0–2.5)
1.6 (1.5–4.1)
English (vs other) spoken at home
0.8 (0.4–1.8)
1.0 (0.4–2.5)
2.3 (1.2–4.4)
2.5 (1.5–4.1)
Low income (vs high)
1.2 (0.5–2.8)
1.8 (0.8–4.0)
0.9 (0.5–1.5)
0.6 (0.3–1.3)
Unknown income (vs high)
1.0 (0.5–2.1)
1.8 (0.9–3.7)
0.9 (0.5–1.5)
0.7 (0.4–1.5)
Older −4–5 years (vs younger – 2–3 years)
4.3 (2.4–7.7)
3.6 (1.9–6.8)
2.0 (1.4–3.0)
1.9 (1.4–2.7)

There were differences in the distribution of restraint type by age between the pre-legislation and post-legislation samples (Table 3), particularly among 4–5 years olds, where substantially fewer children were using adult seat belts and more children were using forward facing child restraints and boosters post-legislation. While the difference was not significant among the 2–3 year olds, fewer were using adult belts in the post-legislation sample but more were using booster seats, which is inappropriate in this age group.

Table 3.  Comparison of distribution of restraint types in pre-and post-legislation samples.
Age (years)RestraintPre-legislation N=107Post-legislation N=360 p value
  1. *accessory harness used alone and unrestrained categories were omitted for chi-square test due to low numbers.

2–3Forward facing child restraint
Accessory harness used alone*
Seat belt
30 (51)
20 (34)
2 (3)
6 (10)
1 (2)
73 (54)
56 (41)
0 (0)
6 (6)
1 (1)
4–5Forward facing child restraint
Accessory harness used alone*
Seat belt
2 (4)
22 (46)
0 (0)
24 (50)
0 (0)
48 (21)
145 (65)
0 (0)
30 (13)
1 (0)


This analysis demonstrates a significant increase in best practice restraint use among children aged 2–5 years in low socioeconomic areas following the introduction of mandatory restraint type legislation for children up to age seven in NSW in 2010. Children in low socioeconomic areas is an important sector of the Australian community, as appropriate restraint use has previously been reported to be lower than in wealthier sectors.12 Keeping this in mind, the observations made here are encouraging. According to estimates by Du et al.,30 increases in appropriate restraint use from about 50% to 74%, as observed here, could equate to an expected reduction in fatal injuries of about 10% and a reduction in non-fatal injuries of about 25%.

While there is no record of the nature or number of campaigns encouraging best practice restraint conducted locally or nationally in the intervening years, there was definitely a distinct increase in public education campaigns preceding and accompanying the introduction of the legislative change in all jurisdictions. The change in restraint practice reported here likely demonstrates the effect of the legislation, enforcement and public education campaigns accompanying introduction of new legislation, in the short term. The level of appropriate restraint use seen in this post-legislation sample aligns with the number of children aged 0–6 years observed to be appropriately restrained during post-legislation roadside observations in Queensland where child age was estimated.34 However, to see full benefit, the proportion of children appropriately restrained needs to move closer to 100%, and these practices must be sustained over time. The fact that only 74% of children were using appropriate restraints in the post-legislation sample highlights the need for further efforts to reach those not complying with the legislation. In parallel work in our recent cluster randomised trial (CRT) of a multifaceted intervention in pre-schools in low socioeconomic areas of Sydney, which included education and restraint subsidies, our intervention boosted appropriate restraint use by 10%.28

An important part of the success of the CRT28 was providing subsidised restraints that were appropriate to the age and size of children 2–5 years. Currently, there are few hire or loan programs operating in Australia that would facilitate low financial access to these types of restraints. There are also few programs that facilitate subsidised access to these types of restraints. There are, however, numerous infant restraints schemes that provide rental options for parents of children less than 12 months of age. Interestingly, the early work by Bilston et al.15 had found appropriate use to be highest (>95%) for this age group of children.

This type of additional targeted effort is probably needed to sustain high levels of appropriate restraint use. This was demonstrated by events following the introduction of laws requiring children to use some sort of restraint in Australia in the 1970s, which saw restraint use among children in NSW increase from 30% to 60% immediately following introduction of the legislation; but without ongoing educational campaigns, restraint use dropped back to 40% eight months later.35,36

In this analysis, we saw more children aged 2–3 years in booster seats post-legislation, and this form of inappropriate restraint use might reflect confusion over messages being conveyed in public education campaigns. Brixey et al.25 made a similar observation following the introduction of appropriate restraint legislation in North America. This confusion likely reflects parental difficulties in distinguishing between forward facing restraints and booster seats and/or distinguishing the safety benefit of the two types of restraint for children in this age range.32 Communicating this level of detail in mass education campaigns is difficult. However, community-based interventions may have more capacity for delivering the detailed information required.

While the changes to Australian legislation per se would not theoretically be expected to have an effect on correct use of restraints, this analysis did demonstrate a significant increase in correct use in the post-legislation sample (from 36% to 47% correctly restrained). As the post-legislation sample was drawn from the control arm of a CRT, it is possible that survey activity at the observation sites undertaken earlier in the year heightened awareness of correct restraint use. However, as these were self-report surveys and no additional information about restraint use was disseminated, the impact was minimal. Also, increases in correct use of restraints with the introduction of mandatory appropriate use laws have been observed in similar studies in North America.22–24 It is possible our result reflects an increased awareness of the importance of optimal child restraint practices that might occur with the introduction of the legislation. Understanding the factors leading to the increased correct use is worthy of further study, as the greatest reductions in casualty numbers require widespread uptake of both appropriate and correct use of restraints.33

Currently, the primary countermeasure to incorrect child restraint use employed in Australia is the provision of restraint fitting services through commercially operated restraint fitting station networks and free restraint fitting checks provided by motoring clubs and other child safety advocate organisations. The use of these services has previously been shown to be associated with an increased likelihood of correct use.30 We were unable to control for any differences in use of these services between the pre- and post-legislation samples. It is worth noting that an internationally implemented countermeasure to incorrect installation – the use of universal anchorage systems such as ISOFIX or LATCH – has not yet been implemented in Australia.

This study was not able to examine ethnicity as a factor in child restraint practices because the relevant data was collected in terms of language spoken at home, rather than any specific measure of ethnicity. In North American studies, ethnicity has been reported to be a factor in sub-optimal child restraint practices.18 In our previous Australian studies, we have used language spoken at home as a measure to explore this factor.15,31,32 Notably, we did not see a significant association between language spoken at home and size appropriate restraint with or without adjustment for other factors, but we did see a significant association with correct use of restraints. The odds of being correctly restrained were 2.5 times greater in children from families who reported speaking English at home than those from families who reported speaking another language at home (95%CI 1.5–4.1, Table 2).

The primary limitation of this study is the use of a historical control sample, and disparity in factors known to be associated with child restraint practices between the pre- and post-legislation samples. While the analysis controlled for these factors, it is worth noting that the post-legislation sample contained more low-income families, more families who did not speak English at home and more older children (aged 4–5 years) than the pre-legislation sample. As each of these factors has previously been reported to be associated with an increased likelihood of sub-optimal child restraint use,15,31,32 any influence due to these variations likely resulted in a conservative estimate of an increase in optimal practices. We note that the results for those who did not report their income were very similar to those with low income (Table 2). The pre-legislation sample also included children who were recruited from a wider variety of recruitment sites i.e. early childhood health clinics, pre-schools, day care centres and primary schools, than the post- legislation sample. However most of the pre-legislation sample (71%) were recruited from the same type of site as the children in the post-legislation sample. Further, as there was no significant difference in any factor of interest between children in the pre-impact sample depending on recruitment site type, we believe this is unlikely to have had any adverse effect on this analysis.

A strength of this study is that we used parent interview to determine child age and direct observation of restraint practices of children in cars by trained technicians, rather than relying on self report25 or visual estimates of child age.34 This approach allowed robust assessment of age-appropriateness and correct use of the restraint.

While we cannot rule out the influence of other factors, we found an increase in best practice child restraint use following new legislation and associated activities. Further effort is required to gain full benefits for reductions in casualties for child passengers. Greater emphasis on the provision of community-level interventions to accompany legislative change and mass education campaigns may be needed to ensure detailed levels of information and practical support are given to those who need this to ensure widespread compliance with best practice. Finally, our analysis demonstrates that further work is required to find effective mechanisms for improving correct use of child restraints.


Data used in this study were collected from two larger studies supported by an Australian Research Council (ARC) Linkage Grant with partner funding from the NSW Centre for Road Safety and a National Health and Medical Research (NHMRC) Project Grant. JB, LB and RI are supported by NHMRC research fellowships and LK is supported by an ARC research fellowship.