Effects of a Booster Seat Education and Distribution Program in Child Care Centers on Child Restraint Use Among Children Aged 4 to 8 Years

Objective  To compare child care center–based booster seat education and distribution with no intervention when implemented immediately after booster seat legislation.

Design  Cluster randomized controlled trial.

Setting  Thirty-nine urban child care centers.

Participants  Eight hundred fifty-four parents and 1010 children aged 4 to 8 years in vehicles leaving centers.

Intervention  We trained 168 staff members at 20 centers to give parents and children messages promoting booster seats and supplied lesson plans, children's activities, and free booster seats.

Main Outcome Measures  Observed booster seat use, “good practice” restraint use, and legal restraint use.

Results  Parents at intervention centers were more likely to report receiving restraint information from the center (adjusted odds ratio [AOR], 4.06; 95% confidence interval [CI], 2.48-6.67), speaking with staff about booster seats (AOR, 3.95; 95% CI, 2.26-6.88), and using fit to decide when to move children into seat belts (AOR, 3.39; 95% CI, 1.91-5.99). Groups did not differ in proportions using booster seats (44% vs 43%; AOR, 1.03; 95% CI, 0.62-1.73), good practice (42% vs 41%; AOR, 1.11; 95% CI, 0.70-1.74), or legal restraints (65% vs 65%; AOR, 0.79; 95% CI, 0.48-1.31). Results were similar for children aged 4 to 5 and 6 to 8 years. All outcomes were significantly less likely among children riding in pickup trucks or with Hispanic or black drivers.

Conclusions  The intervention increased parents' receipt of information from center staff and knowledge about booster seats but not booster seat use. Research is needed to identify methods and messages that will empower center providers to promote booster seats effectively and reach high-risk populations.

Among children aged 4 to 8 years, booster seats to improve the fit of lap/shoulder seat belts substantially reduce motor vehicle crash fatalities and injuries.^1,2 In 2007, only 37% of 4- to 7-year-old passengers in the United States were restrained in booster seats.³

Booster seat laws increase by 39% the likelihood that children aged 4 to 7 years will be appropriately restrained in booster or safety seats.⁴ Even with such laws in place, however, estimated use is less than 80% in children aged 4 to 5 years and less than 60% in children aged 6 to 7 years.⁴ Communitywide information campaigns, combined with enforcement strategies, enhance use of child safety seats beyond that achieved by legislation.⁵ This suggests that adding educational interventions to existing legislation might enhance booster seat use.

Programs in diverse settings that combine incentives or free booster seats with education have been shown to increase booster seat use.⁶ Child care centers may be appropriate venues for booster seat promotion programs because many children aged 4 to 8 years attend such centers. Parents interact with staff members daily and view them as useful information sources.⁷ Although one trial evaluating a booster seat promotion program in preschools and child care centers did not demonstrate a benefit, the study was underpowered because of high attrition.⁸ In contrast, a before/after study⁹ found that booster seat use increased 35% after an educational intervention targeting staff, parents, and children in child care centers, suggesting potential benefits from programs in this setting.

Colorado's child occupant safety law (covering car seat and seat belt use) allows drivers to be stopped if an unrestrained or improperly restrained child is observed in the vehicle (primary enforcement). The state booster seat law, implemented in August 2004, allows enforcement only when a driver is stopped for another offense (secondary enforcement). This study aimed to examine the effect of a booster seat education and distribution program targeting child care center staff and families, when conducted immediately after implementation of a secondary booster seat law.¹⁰ Since Colorado's booster seat law applies only to 4- to 5-year-olds, we evaluated whether program effects differed among 4- to 5-year-olds and 6- to 8-year-olds. We also evaluated the intervention's effect on center directors' knowledge, attitude, and behavior toward child passenger restraint use and on parents' knowledge about appropriate restraint use.

We conducted a cluster randomized controlled trial in child care centers. Researchers uninvolved in recruitment randomly assigned centers by minimization,¹¹ balancing groups for total center enrollment (in quartiles), minority enrollment (using cutoffs matching the 99% confidence interval [CI] for mean proportion of minority persons living in city census tracts), and percentage of families with children younger than 18 years living in poverty in the center's census tract (higher than vs at or lower than the city median) from the 2000 Census Summary File 3 (Table 1). Estimating 29 eligible children per center, 1.2 eligible children per vehicle,¹² 10% center attrition, intracluster correlation coefficient of 0.01,¹³ and 50% control booster seat use, we aimed to enroll 22 centers per group, providing 80% power at α = .05 to detect a 10% difference in booster seat use between groups.

The study was conducted in Colorado Springs, Colorado, where observed booster seat use among children aged 4 to 8 years was 37% in 2002.¹⁴ The sampling frame included all active, licensed child care centers providing care to 15 or more children aged 4 to 8 years (Figure 1). We excluded preschools (which serve children <6 years) and home-based and drop-in centers. Between August and October 2004, we recruited centers by letter and telephone. We obtained written consent from the owner or director. Participating centers received $50 gift certificates.

We collected data in autumn 2004 (baseline) and spring 2006 (postintervention). Research assistants (RAs), blinded to experimental group, observed restraint use of children aged 4 to 8 years in vehicles leaving center parking lots. The RAs were trained by a National Highway Traffic Safety Administration–certified Child Passenger Safety Instructor and achieved 90% agreement with the instructor during practice observation. The RAs asked drivers about demographics (including race/ethnicity), child height and weight, receipt of information about child restraints from the center in the past year, and whether center staff talked to them about booster seats. Drivers' knowledge was assessed by asking how they decided when children should move from car or booster seats to seat belts.

Data were collected on one afternoon per center by 2 to 5 RAs, with centers from both groups observed each afternoon. The RAs recruited adult drivers (or a responsible adult passenger) with 4- to 8-year-old children in the vehicle. Drivers were given brief information sheets and gave verbal consent. Participants received brochures with information about local health department services (including safe child transportation) and free pens; children received small toys. The RAs counted vehicles that left without contact and drivers that were ineligible or refused screening or participation.

Before and after intervention, we surveyed center directors to assess whether and where they received booster seat training, knowledge about Colorado child occupant restraint laws, attitudes about compliance with the law and talking to parents about restraints, and center practices and policies regarding child restraints.

Using baseline survey results from parents and center staff, we developed the intervention using the PRECEDE-PROCEED model.¹⁵ We addressed predisposing factors by promoting awareness and using consistent guidelines about booster seat use, enabling factors by increasing staff skills and self-efficacy and providing educational resources and booster seats, and reinforcing factors by emphasizing existing law, making repeated contacts with staff, and promoting center transportation policies.

During the 18-month intervention, we trained 168 center staff, providing them with motivation, skills, and self-efficacy to give parents and children clear messages about state law and the safety advantages of booster seats. In collaboration with trained center staff, we coordinated one parent education event per center, modeling techniques for discussing booster seats with parents and distributing 573 booster seats directly to parents. Centers received resource kits (model policy, lesson plans, children's activities, brochures) and booster seats for use in center vehicles (n = 133). Over the course of the 18-month program, we gave 378 additional seats to the centers, on request, to distribute to parents who missed the education event, whose children were newly enrolled or transitioning into booster seats as they aged, or who requested a seat for a second car. All centers participated in 3 or 4 holiday-themed booster seat promotion activities. Some wrote newsletter articles and/or distributed a recommended family transportation policy. We contacted centers every 1 to 2 months to motivate staff, provide additional resource materials and booster seats, and track center activities.

Primary outcomes were postintervention booster seat, legal restraint, and “good practice” restraint use. We could not evaluate center-specific changes in outcomes between baseline and follow-up because the restraint use survey was revised after baseline data collection to improve its accuracy, so outcomes before and after the intervention were not directly comparable. Booster seat use included any booster seat with proper lap/shoulder belt use, regardless of height, weight, or age. Legal restraint use was consistent with Colorado law, incorporating various combinations of age, height, weight, and restraint type.¹⁰ Good practice was defined as correctly using the appropriate restraint for height and weight (based on National Highway Traffic Safety Administration and manufacturers' guidelines) and sitting in the rear seat.^16,17 If weight or height alone was missing, we assumed average weight for height. Outcomes were classified as missing when recorded data were insufficient to enable determination. Booster seat use was missing for 5% of children, good practice for 12%, and legal restraint for 11%. Missing outcomes did not differ by experimental group.

Secondary outcomes included center policies and practices and center directors' knowledge, attitude, and behavior regarding child passenger occupant protection, as reported in the center directors' survey, and drivers' self-reported knowledge about child restraint use.

Center, vehicle, and driver characteristics at baseline and follow-up are described. All participating centers were analyzed regardless of the extent to which they implemented the program. Relative risks and 95% CIs were calculated to compare groups on center transportation policies, center director attitudes, behavior, and knowledge and process outcomes postintervention. Odds ratios (ORs) and 95% CIs for child restraint and driver outcomes were calculated using logistic binomial analysis models for distinguishable data, including a random effect to account for clustering within centers and, for child restraint outcomes, a random effect to account for clustering within vehicles. We adjusted each model for the centers' relevant pretest outcomes. We assessed potential confounding by vehicle type, driver, and child characteristics. To examine confounding by child characteristics when vehicle was the analysis unit, we randomly selected one child in vehicles containing multiple eligible children, after excluding children for whom outcome data were missing. Covariates that significantly improved fit at P < .10 or changed the estimated intervention effect by 10% or more were retained.

We examined outcomes by age group (4-5 years and 6-8 years) using methods described earlier. We were unable to model both vehicle and center as random effects because of smaller numbers in each subgroup. We therefore used vehicle as the analytic unit, randomly selecting one child when there were multiple eligible children per vehicle. For each outcome, we retained any covariate that met retention parameters for either age group in the models for both age groups, to make results comparable.

Institutional review boards at the Colorado Department of Public Health and Environment and Centers for Disease Control and Prevention approved this study.

At baseline, intervention and control groups had similar center minimization factors; characteristics of vehicles, drivers, and children; and restraint use (Table 1). Figure 2 shows drivers contacted, screened, and enrolled at follow-up. Most exclusions (99%) were because no eligible children were in the vehicle. “Lack of time” explained most refusals (65%). Including vehicles not contacted or screened, we estimate that at least 75% of all eligible drivers participated in postintervention surveys.

We observed a mean of 25.9 children per center and 1.2 children per car at follow-up. As at baseline, intervention and control groups did not differ in vehicle, driver, or child characteristics postintervention. Most vehicles were cars (42.6%), sport utility vehicles (33.6%), or minivans (14.1%). Pickup trucks accounted for 80% of “other” vehicles. Typical drivers were mothers (61.7%), of white race (71.2%), and wearing seat belts (78.3%). Most children were aged 4 to 5 years (61.7%) and attended the center where observed (81.3%).

Common restraint types at follow-up were high-back booster seats (27.4%), no-back booster seats (26.3%), lap/shoulder belts (19.3%), and forward-facing car seats (13.4%). Less common were built-in seats (2.4%), other restraints (eg, lap belts, shield booster seats) (6.7%), and no restraint (4.5%). Of children in restraints, 4.1% were actually unrestrained (car seat not secured to car, 3.4%; booster seat without seat belt, 0.7%). Another 9.2% of restrained children used unsafe practices, including lap/shoulder belt tucked behind back (5.2%), booster seat with lap belt (3.7%), and shared seat belt (0.3%).

Most respondents were directors or owners who had been at the center throughout the intervention period (Table 2). Intervention respondents were significantly more likely to have attended any booster seat training class and one at their own center. Larger proportions of intervention respondents responded correctly to questions about Colorado's child restraint laws, reported talking to parents at least monthly about child restraints, and complied with Colorado law by providing appropriate restraints when transporting children, but CIs were wide. There was no apparent effect on center policies or directors' comfort with talking to parents about restraints.

Intervention center drivers were more likely to report receiving information from the center about child restraints (52% vs 22%; adjusted OR [AOR], 4.06; 95% CI, 2.48-6.67) and talking with center staff about booster seats (43% vs 17%; AOR, 3.95; 95% CI, 2.26-6.88). They were also more likely to report appropriately using seat fit, rather than age, law, etc, to decide when to move children from safety seats to seat belts (14% vs 6%; AOR, 3.39; 95% CI, 1.91-5.99).

There were no apparent differences between intervention and control centers in the proportions of children using booster seats (43% vs 44%), good practice restraint use (41% vs 42%), and legal restraint use (65% vs 65%). Adjustment for pretest center restraint use and child, driver, and vehicle characteristics did not materially influence these results (Table 3).

In the subgroup of intervention centers, we observed higher booster seat use (48% vs 40%), good practice use (50% vs 36%), and legal restraint use (69% vs 60%) in vehicles where drivers reported having received any information on child restraints from the center, compared with those where drivers did not. Only the relationship with good practice use was statistically significant after adjusting for differences in driver, vehicle, and child characteristics (AOR, 1.89; 95% CI, 1.11-3.23). Within intervention centers, driver report that staff had spoken to them about booster seats was not associated with any outcomes (data not shown).

Intervention effects on booster seat, good practice, and legal restraint use were similar for vehicles with selected children aged 4 to 5 years (legally required to use booster seats) and 6 to 8 years (legally restrained by seat belts) (Table 3).

Children aged 6 to 8 years were less likely than children aged 4 to 5 years to be restrained using good practice or in booster seats (Table 3), but more likely to be legally restrained. Children were significantly less likely to use good practice restraints, booster seats, or legal restraints if the driver was of Hispanic ethnicity or black race or driving a pickup truck or other vehicle, both overall and in the 2 age groups (Table 3).

We evaluated a comprehensive booster seat program in child care centers emphasizing skills training and educational resources for staff, distribution of free booster seats, children's activities, and parent education. Parents of children enrolled at centers that received the program were more likely to report receiving information from the center about child restraints, talking with center staff about booster seats, and correctly using seat fit to decide when to move a child from a car safety seat to a seat belt. However, getting the booster seat message and free booster seats to parents did not significantly increase the likelihood of children using booster seats or of being restrained legally or with good practice.

Because parents had access to free booster seats from intervention centers, factors other than cost probably influenced parental decisions not to use booster seats. The high turnover rate typical of child care center staff (estimated 25%-40% nationwide)¹⁸ means that parents may not have received consistent, repeated booster seat promotion messages. A local child care resource/referral agency estimated that 65% of children enrolled in a given center will be new each year. This turnover is likely to have attenuated program effects, despite the fact that we continued to distribute seats and promote staff efforts throughout the 18-month intervention period. Contamination may also have been a factor since both staff and families may move between centers. Most studies that found a beneficial effect of education and promotion on booster seat use had low baseline rates,⁶ whereas in our study, more than 40% of children were using booster seats at baseline, immediately after a secondary booster seat law was implemented. We may not have identified the right combination of training and activities to enable child care providers to influence parents who had not begun using booster seats despite a law requiring their use for children aged 4 to 5 years.

Our emphasis on good practice did not lead to higher booster seat use among 6- to 8-year-olds, who are not covered by the state booster seat law, at intervention centers compared with control centers. After the intervention, this age group continued to have significantly lower booster seat and good practice restraint use than did 4- to 5-year-olds. Convincing parents of older children that booster seat use is necessary may be difficult when the law does not require such use, especially if the children are already using seat belts.⁹

Booster seat, good practice, and legal restraint use were significantly less likely in vehicles with Hispanic or black drivers. Such families may have been less likely to understand and act on the promotion messages, or staff may have been less likely to implement the messages for these families or in these neighborhoods. Our intervention did not specifically address how center staff could best interact with Hispanic families; some Spanish-speaking families may not have been adequately reached by our messages. Other studies have indicated the need to design messages tailored to target groups.^19,20 A trial in progress⁶ evaluating culturally tailored intervention materials to increase booster seat use among Latino families may inform this issue.

We used a cluster randomized controlled trial design to evaluate a program delivered to child care centers, with adequate allocation concealment, masking of observers to experimental group, and enrollment of an estimated 75% of eligible vehicles, supporting the internal validity of our results. Sample size was calculated to identify a difference between groups of 10%, large enough to be of public health significance, whereas observed differences were 1% or less, suggesting that lack of power is unlikely to fully explain our results.

We enrolled 67% of eligible centers and followed up more than 90% of enrolled centers, which allowed us to examine the intervention in diverse centers where staff and center interest and participation varied widely. This more closely approximates the true effects of implementing such programs than do studies where programs are implemented in a few locations chosen based on interest. This may account for differing results between our study and Apsler et al.⁹ Our findings were similar to those of another randomized controlled trial in child care centers.⁸

Our study was conducted in a single urban Colorado city with a large proportion of highly mobile military families, and results may not be generalizable to different types of communities. We surveyed only the center owner or director; program effects on other staff members may have been different. Our results might overestimate true restraint use if the presence of RAs in the parking lots prompted more child restraint use. Finally, we could not determine restraint properties such as height and weight restrictions and whether the restraints were attached properly, and child height and weight may have been inaccurately reported by drivers. Therefore, good practice and legal restraint use may have been assigned incorrectly, although the booster seat outcome would not have been affected.

Communitywide booster seat promotion programs have been successful in other contexts. Although we did not find a similar beneficial effect in child care centers, our study does provide insight into potentially important target populations for booster seat promotion programs, where limited resources might be best directed. These include children aged 6 to 8 years, pickup truck drivers (who are less likely to use restraints themselves²¹), and minority populations. In communities where overall restraint use is high among 4- to 8-year-olds, programs might emphasize persuading parents to move children who are inappropriately restrained into booster seats. Programs in communities with low restraint use in the target group may need to stress the value of any restraint use, emphasizing booster seats as the best choice.

Child care centers varied in their willingness to approach parents about signing a pledge to transport their children using good practice restraint use. Several center directors indicated that they would be more likely to implement such agreements if required by licensing agencies. Pediatricians, health departments, and other child health advocates may wish to work with state child care center licensing agencies to expand or update child passenger safety provisions.

We found child care center directors and staff receptive to working with researchers on a program potentially benefiting both the center (by providing free booster seats, staff training, and educational resources) and enrolled families (by providing free booster seats and education). Many centers responded enthusiastically to the program by sending multiple staff for training, distributing seats, and implementing recommended lesson plans and activities. Child care centers might therefore be effective venues for booster seat promotion using a different type of program or for programs targeting other conditions or behaviors.

Further studies are needed to identify methods and messages that will empower child care center providers to promote booster seats in ways that will change behavior. Identifying ways to help parents move 6- to 8-year-olds back into child safety restraints once they have begun using seat belts may be necessary. Studies are also needed to understand how to reach black and Hispanic families and pickup truck drivers.

Correspondence: Sallie Thoreson, MS, Colorado Department of Public Health and Environment, 222 S 6th St, Room 232, Grand Junction, CO 81501 (sallie.thoreson@state.co.us).

Accepted for Publication: September 20, 2008.

Author Contributions: Dr DiGuiseppi had full access to all study data and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Thoreson, Myers, and DiGuiseppi. Acquisition of data: Myers and Goss. Analysis and interpretation of data: Thoreson, Goss, and DiGuiseppi. Drafting of the manuscript: Thoreson and DiGuiseppi. Critical revision of the manuscript for important intellectual content: Thoreson, Myers, Goss, and DiGuiseppi. Statistical analysis: DiGuiseppi. Obtained funding: Thoreson. Administrative, technical, and material support: Thoreson, Myers, and Goss. Study supervision: Thoreson.

Financial Disclosure: None reported.

Funding/Support: The Centers for Disease Control and Prevention (CDC) funded this research through grant U17/CCU823429. Lorann Stallones, PhD, Colorado Injury Control Research Center, assisted early program development, funded by CDC grant U17/CCU818943.

Role of the Sponsor: The CDC was not involved in study design or conduct; data collection, management, analysis, or interpretation; or manuscript preparation, review, or approval.

Disclaimer: The contents are solely the responsibility of the authors and do not necessarily represent official views of the CDC.

Additional Contributions: We thank participating child care centers, volunteer research assistants, and Colorado Springs Safe Kids Coalition.