Video Abstract

Video Abstract

Close modal
BACKGROUND AND OBJECTIVES:

The American Academy of Pediatrics designed The Injury Prevention Program (TIPP) in 1983 to help pediatricians prevent unintentional injuries, but TIPP’s effectiveness has never been formally evaluated. We sought to evaluate the impact of TIPP on reported injuries in the first 2 years of life.

METHODS:

We conducted a stratified, cluster-randomized trial at 4 academic medical centers: 2 centers trained their pediatric residents and implemented TIPP screening and counseling materials at all well-child checks (WCCs) for ages 2 to 24 months, and 2 centers implemented obesity prevention. At each WCC, parents reported the number of child injuries since the previous WCC. Proportional odds logistic regression analyses with generalized estimating equation examined the extent to which the number of injuries reported were reduced at TIPP intervention sites compared with control sites, adjusting for baseline child, parent, and household factors.

RESULTS:

A total of 781 parent–infant dyads (349 TIPP; 432 control) were enrolled and had sufficient data to qualify for analyses: 51% Hispanic, 28% non-Hispanic Black, and 87% insured by Medicaid. Those at TIPP sites had significant reduction in the adjusted odds of reported injuries compared with non-TIPP sites throughout the follow-up (P = .005), with adjusted odds ratios (95% CI) of 0.77 (0.66–0.91), 0.60 (0.44–0.82), 0.32 (0.16–0.62), 0.26 (0.12–0.53), and 0.27 (0.14–0.52) at 4, 6, 12, 18, and 24 months, respectively.

CONCLUSIONS:

In this cluster-randomized trial with predominantly low-income, Hispanic, and non-Hispanic Black families, TIPP resulted in a significant reduction in parent-reported injuries. Our study provides evidence for implementing the American Academy of Pediatrics’ TIPP in routine well-child care.

What’s Known on This Subject:

Injury in young children is a major cause of morbidity and mortality. Although the American Academy of Pediatrics’ primary care-based The Injury Prevention Program was developed in 1983, its effectiveness for injury prevention has never been tested in a randomized trial.

What This Study Adds:

The Injury Prevention Program (when implemented with support of resident training and tangible tools) reduced parent-reported injuries throughout the first 2 years of life, supporting effectiveness of primary care-based injury prevention approaches. Most reported injuries did not seek medical attention.

Unintentional injury is prevalent during childhood, accounting for ∼10% of visits to ambulatory care.1,3 Unintentional injury may or may not require medical attention, and injuries in young children range in severity from cuts, burns, and falls, to choking, unintentional ingestions, drownings, motor vehicle crashes, and injuries from firearms. For the past 20 years, injuries have been a leading cause of death for young children, and many of those injury-related deaths are preventable.4 Unintentional injuries are the fourth leading cause of death and morbidity for children <1 year of age and the leading cause of death among children in the second year of life.4 Less-serious injuries can also have a significant impact on children’s overall well-being and may help identify environmental factors that place young children at risk for more serious injury. Many of these factors may be addressed in pediatric primary care, especially through parent education and anticipatory guidance.5 

To date, most studies of injury prevention programs in primary care have had positive findings, though findings have been limited to effectiveness at reducing certain injuries or improving knowledge and behavior rather than injury reduction. A 25-year review for office-based injury prevention in young children showed positive effects on knowledge and/or behaviors for certain outcomes like falls, poisoning, burns, traffic injury, and drowning prevention.6 Most, though not all,7 studies have found that injury prevention programs are associated with improved knowledge and better home safety practices without concomitant reductions in injury.8,9 

Because injury has been an important cause of morbidity and mortality, The Injury Prevention Program (TIPP) was designed by the American Academy of Pediatrics’ (AAP’s) Committee on Injury, Violence, and Poison Prevention in 1983 to help pediatricians identify parental “at-risk behavior,” support provider counseling regarding important injury-related anticipatory guidance using a developmentally-based schedule, and provide a standardized set of written information about injury prevention for parents and caregivers to take home as a resource.10 Unfortunately, most pediatrics providers spend little time counseling about injury prevention,11,12 despite the known national burden of childhood injury, evidence of the efficacy of clinical counseling for some injury prevention behaviors,13,14 national guidelines that promote injury preventive service counseling about injury prevention in general5,15 (and on car seat safety16 and choking17 in particular), and longstanding models for clinical prevention (eg, the TIPP anticipatory guidance framework).8 

Significant disparities in pediatric injury exist, with members of minoritized racial or ethnic groups and those living in poverty experiencing an increased risk of injury.18,22 Although most studies of pediatric injury lack sufficient inclusion of social factors, such as differences in language, literacy, and socioeconomic status, there are factors known to increase the risk for unintentional child injury. These include being male, living with a single parent,23 having a young parent,24 having less adult supervision,12 and having parents from lower socioeconomic status backgrounds.25,26 

Despite the widespread availability of the AAP’s TIPP, there have been few studies that have researched it. Simon et al showed that infants who had received less than the TIPP-recommended injury guidance were more likely to have subsequent injury visits, though it was not a randomized trial.27 Miller and Galbraith used separate analyses to demonstrate that office-based counseling saved substantial costs per child.28 Studies comparing TIPP to other injury prevention interventions have been equivocal,29,30 suggesting that basic education may be just as good as enhanced counseling when it comes to injury prevention anticipatory guidance. To our knowledge, however, there have been no previous experimental studies examining TIPP’s effectiveness.

We implemented a stratified, cluster-randomized controlled trial, known as the Greenlight Study,31 in a large, diverse sample of young children to compare children who received an obesity prevention intervention to a group that focused on injury prevention using TIPP. This unique design allowed us to evaluate the impact of the obesity intervention,32 but also to evaluate the impact of TIPP compared with the obesity intervention (with injury prevention usual care). We sought to examine the extent to which the number of injuries reported was reduced at TIPP intervention sites compared with non-TIPP sites. We hypothesized that those at TIPP intervention sites would have fewer reported injuries than those at non-TIPP sites.

The Greenlight Study was a pragmatic, stratified, cluster-randomized controlled trial of an obesity prevention intervention for children in their first 2 years of life that has been previously described.31 The Greenlight intervention (registered with the national Clinical Trials Registry [clinicaltrials.gov] study #NCT0104087) sought to decrease overweight at age 24 months and decrease interval weight trajectories, and the main-outcome results from this obesity prevention trial were recently reported.32 One of the arms of the registered trial was behavioral change with respect to injury prevention, which we report here. Two of 4 university-affiliated pediatric continuity clinics (University of North Carolina [UNC] at Chapel Hill, New York University [NYU]/Bellevue Hospital Center, Vanderbilt University/Vanderbilt University Medical Center [Vanderbilt], and University of Miami/Jackson Memorial Medical Center [Miami]) were randomly assigned to address injury prevention through the AAP’s TIPP, whereas 2 of 4 sites were randomly assigned to use a literacy and numeracy-sensitive approach to obesity prevention (Greenlight). To avoid contamination, randomization occurred at the site level, and we stratified sites on the basis of whether they were located in a predominantly urban (NYU and Miami) or suburban (UNC and Vanderbilt) city, and then randomized within strata. Each site was assigned to intervention or active control using the Stata (Stata Corp, College Park, Texas) random number generator by a blinded statistician. For this study, we assessed the TIPP intervention and assigned Greenlight as the active control comparator.

We enrolled healthy infants and their caregivers (mostly mothers, aged ≥18 years old) at their infants’ 2-month-old preventive services visit. Parents provided consent after procedures delineated within institutional review board-approved protocols at all 4 sites. Once consent was obtained, parents completed questionnaires with trained research assistants fluent in the language of the parents’ choice (English or Spanish). Data were managed through Research Electronic Data Capture, a secure, Web-based application to support data capture for research studies hosted at Vanderbilt University.33 The study was registered with the national Clinical Trials Registry (#NCT01040897 at clinicaltrials.gov).

We used the AAP’s TIPP program for early child injury prevention. This program includes a developmentally based safety counseling schedule that guides what materials (safety sheets and an age-appropriate Framingham safety survey) to ask about risk behaviors. For the age group relevant here, there are pediatric patient handouts for parents of children who are aged 0 to 6 months, 6 to 12 months, and 1 to 2 years, and they review safety for falls, motor vehicles, firearms, drowning, poisoning, choking, and burns. The AAP provided permission to use TIPP materials available at the time, and 4 native Spanish-language speakers from 4 nations of origin in Latin America translated portions that were not already available in Spanish. In attention equipoise with the other arm of the trial, we designed and provided TIPP “tangible tools,”13 and we provided training and certification in injury prevention and TIPP use to the residents throughout the study. Training took place via modules designed to teach about injury prevention, and certification consisted of observation of developmentally appropriate injury prevention counseling in encounters with patients by a trained research assistant. We found that >90% of residents were certified in the first observation, and all were certified after the second. In the comparator group, residents were provided with training in obesity prevention, as well as health communication. Greenlight obesity prevention materials were readily available and encouraged to be reviewed with families and distributed. Injury prevention was usual care in this comparator group. Families were provided with packets of handouts on a range of anticipatory guidance topics, and residents had templates that reminded them to do counseling on a variety of topics including injury prevention, but no formal standardized injury prevention counseling was done using TIPP or any other program.

Over a 4-year period, we assessed interval injuries since the last visit at the successive AAP-recommended preventive care visits (4, 6, 9, 12, 15 and/or 18, and 24 months).

Measures

Independent Variables

For regression models, we included the following independent variables measured at the baseline, 2-month well-child check (WCC) visit: Child sex, injuries (yes/no) before baseline, annual household income, caregiver reading language, self-reported caregiver race and ethnicity, caregiver age, caregiver education, caregiver health literacy as measured by the Short Test of Functional Health Literacy in Adults (S-TOFHLA), number of children in the household, and number of adults in the household. To capture the site and, ultimately, the treatment (TIPP versus Greenlight) association with injuries over time, we included: Site (Miami, UNC, NYU, Vanderbilt), age, and site-by-age interaction. See the Analyses section for details.

Dependent Variables

Our primary outcome was based on reported injuries since the age of the preceding planned WCC visit, and injuries were defined for parents broadly. At each WCC, parents reported the number of injuries since the age at the preceding planned WCC with answer choices of 0, 1, 2, or 3 or more. So, for example, at the 4-month WCC, a parent was asked: “Since the child was 2 months old, how many times has he/she been injured? An injury includes anything that causes harm or pain, such as a cut, a burn, a bruise, or other harm that might result from a fall, a car crash, heat, hot water, fire or electricity, a choking episode, swallowing or tasting a poison, a gun, or nearly drowning.” Families were also asked, “What were the causes of these injuries?” and asked to check all that apply among the following categories: Fall; motor vehicle crash; heat, hot water, fire, or electricity; and choking, poisoning, firearm, nearly drowning, or other. The need for medical attention for injuries was assessed with the following questions: “Were any of these injuries serious enough that you had to go to a doctor?” to which parents replied with a yes or no.

We summarized baseline characteristics, stratified by treatment group and sites separately, using 10th, 50th, and 90th percentiles for continuous variables (to summarize a larger percentage of the sample than the standard 25th, 50th, and 75th percentiles do) and proportions with counts for categorical variables.

We report the prevalence of the number of injuries and injuries requiring medical attention across WCC time points graphically. To examine the extent to which the TIPP intervention was associated with reductions in injury rates across the follow-up period, we fit a proportional odds logistic regression model. To appropriately acknowledge the within-child correlation induced by taking repeated measurements, we fit the model with a generalized estimating equation and a working exchangeable dependence structure.34,35 We estimate SEs with the robust, sandwich form.34,36,37 To capture the intervention association with injuries over time, we included in the regression model a site factor variable with Vanderbilt as the reference site, a flexible age functional with restricted cubic splines to permit nonlinear changes over ages, a site by flexible age interaction, and the baseline covariates described above. Because the main effect of site represents a preintervention association (ie, the site association at baseline) that cannot be attributed to the intervention itself, we characterize site associations with injuries using the site by age interaction. To estimate the TIPP versus Greenlight intervention association with injuries, we computed a linear contrast between the 2 TIPP sites (UNC and Miami) and the 2 Greenlight sites (Vanderbilt and NYU), and report estimates and pointwise confidence intervals (CIs) across the follow-up period. Further, we conduct the corresponding Wald-based test as a single test of the difference between the TIPP and Greenlight interventions. Because we used a proportional odds model, exponentiated parameter estimates and CIs capture the odds ratio (OR) of more injuries associated with being at a TIPP intervention site as opposed to a Greenlight site. ORs >1 are consistent with more injuries.

Although this was a cluster-randomized clinical trial, there were only 4 sites and so we built models that control for a robust list of variables (see Independent Variables subsection) that represent potential confounders.

We were interested in the marginal risk difference of at least 1, at least 2, and at least 3 injuries during each follow-up period between WCC visits between TIPP and Greenlight intervention sites. We captured estimated risk difference and CIs with a simulation-based alternative to the delta method. To describe our approach, we replicated the following algorithm 2000 times. At replication (m), we sampled a realization from the estimated sampling distribution of our parameter estimates using the proportional odds logistic regression model fit, β(m)N(β^,Cov̂(β^)). For each site, we calculated predicted risk for all participants in the study and took the average value. We then calculated the difference in the predicted risks between the average of the 2 TIPP sites and the average of the 2 Greenlight sites, which provides a single estimate of the marginal risk difference. We then summarized the distribution of 2000 replicates to obtain the point estimate (the average) and 95% CI (2.5th and 97.5th percentiles) for the marginal risk difference.

When fitting our model, to address missing data, we conducted multiple imputation with chained equations and used predictive mean matching to fill in the missing values. We generated 25 imputed data sets, analyzed them, and then combined results using Rubin’s rules.38,39 All analysis were conducted using the R Language version 3.6.3 (R project, https://www.r-project.org) and the rms library.40,41 

We enrolled 865 parent–child dyads into the study from 2010 to 2014, of whom 406 were enrolled at a TIPP intervention site and 459 at a Greenlight site. Among those enrolled, 781 (90%) were retained for the present analysis because they completed questions about injuries at baseline and at least 1 follow-up WCC after enrollment (Fig 1). Supplemental Table 3 compares baseline characteristics of those included and excluded from analyses. There were a few differences between the groups, and those with lower education and those who identified as Hispanic were more likely to be retained for analyses. Table 1 and Supplemental Table 4 summarize the demographics and baseline characteristics of the analysis sample. Caregivers reported on their own race and ethnicity as follows: 51% Hispanic/Latino, 28% non-Hispanic Black, 17% non-Hispanic white, and 4% non-Hispanic other. Thirty-six percent of caregivers reported reading mostly in Spanish, 59% of households had annual incomes <$20 000 per year, and 27% of caregivers reported having less than a high school education. Median S-TOFHLA (health literacy) scores were 34, with an interdecile range from 32 to 36, and 85% percent of families reported receiving Special Supplemental Nutrition Program for Women, Infants, and Children benefits at baseline.

FIGURE 1

Consort flowchart.

FIGURE 1

Consort flowchart.

Close modal
TABLE 1

Child and Caregiver (Parent) Characteristics at Baseline

NVanderbiltNYUUNCMiamiTotal
(N = 212)(N = 220)(N = 238)(N = 111)(N = 781)
Child characteristics 
 Age, d 781 62 [48–76] 63 [52–80] 68 [56–86] 62 [45–80] 63 [50–81] 
 Female 781 111 (52%) 120 (55%) 119 (50%) 55 (50%) 405 (52%) 
Caregiver characteristics 
 Age at baseline, y 777 25 [20–33] 28 [21–37] 27 [21–37] 28 [21–37] 27 [20–36] 
 Female  208 (98%) 214 (97%) 227 (95%) 103 (93%) 752 (96%) 
 Race and ethnicity 781      
  Hispanic  77 (36%) 172 (78%) 95 (40%) 56 (50%) 400 (51%) 
  White, non-Hispanic  58 (27%) 15 (7%) 55 (23%) 7 (6%) 135 (17%) 
  Black, non-Hispanic  72 (34%) 22 (10%) 81 (34%) 40 (36%) 215 (28%) 
  All other, non-Hispanica  5 (2%) 11 (5%) 7 (3%) 8 (7%) 31 (4%) 
 Reading language 781      
  English  155 (73%) 105 (48%) 156 (66%) 84 (76%) 500 (64%) 
  Spanish  57 (27%) 115 (52%) 82 (34%) 27 (24%) 281 (36%) 
 BMI, kg/m2 694 28 [22–36] 26 [22–35] 29 [22–39] 27 [21–34] 27 [22–37] 
 Education 781      
  Less than high school  49 (23%) 81 (37%) 63 (26%) 17 (15%) 210 (27%) 
  High school or some college  140 (66%) 98 (45%) 136 (57%) 60 (54%) 434 (56%) 
  College  22 (10%) 41 (19%) 39 (16%) 34 (31%) 136 (17%) 
  Missing  1 (0%) 0 (0%) 0 (0%) 0 (0%) 1 (0%) 
 Annual household income 778      
  <$10 000  53 (25%) 88 (40%) 57 (24%) 40 (37%) 238 (31%) 
  $10 000–$19 999  75 (36%) 53 (24%) 65 (27%) 21 (19%) 214 (28%) 
  $20 000–$39 999  56 (27%) 55 (25%) 56 (24%) 17 (16%) 184 (24%) 
  $40 000–$59 999  14 (7%) 18 (8%) 21 (9%) 10 (9%) 63 (8%) 
  $60 000 or more  13 (6%) 5 (2%) 21 (9%) 18 (17%) 57 (7%) 
  Caregiver did not know  0 (0%) 1 (0%) 18 (8%) 3 (3%) 22 (3%) 
 Number of children at home 780 2 [2–4] 2 [2–5] 2 [1–3] 2 [2–4] 2 [2–4] 
 Number of adults at home 780 2 [1–4] 2 [1–4] 2 [1–4] 2 [1–4] 2 [1–4] 
 Health literacy       
  S-TOFHLA score (0–36) 778 35 [31–36] 34 [16–36] 35 [29–36] 34 [19–36] 34 [22–36] 
NVanderbiltNYUUNCMiamiTotal
(N = 212)(N = 220)(N = 238)(N = 111)(N = 781)
Child characteristics 
 Age, d 781 62 [48–76] 63 [52–80] 68 [56–86] 62 [45–80] 63 [50–81] 
 Female 781 111 (52%) 120 (55%) 119 (50%) 55 (50%) 405 (52%) 
Caregiver characteristics 
 Age at baseline, y 777 25 [20–33] 28 [21–37] 27 [21–37] 28 [21–37] 27 [20–36] 
 Female  208 (98%) 214 (97%) 227 (95%) 103 (93%) 752 (96%) 
 Race and ethnicity 781      
  Hispanic  77 (36%) 172 (78%) 95 (40%) 56 (50%) 400 (51%) 
  White, non-Hispanic  58 (27%) 15 (7%) 55 (23%) 7 (6%) 135 (17%) 
  Black, non-Hispanic  72 (34%) 22 (10%) 81 (34%) 40 (36%) 215 (28%) 
  All other, non-Hispanica  5 (2%) 11 (5%) 7 (3%) 8 (7%) 31 (4%) 
 Reading language 781      
  English  155 (73%) 105 (48%) 156 (66%) 84 (76%) 500 (64%) 
  Spanish  57 (27%) 115 (52%) 82 (34%) 27 (24%) 281 (36%) 
 BMI, kg/m2 694 28 [22–36] 26 [22–35] 29 [22–39] 27 [21–34] 27 [22–37] 
 Education 781      
  Less than high school  49 (23%) 81 (37%) 63 (26%) 17 (15%) 210 (27%) 
  High school or some college  140 (66%) 98 (45%) 136 (57%) 60 (54%) 434 (56%) 
  College  22 (10%) 41 (19%) 39 (16%) 34 (31%) 136 (17%) 
  Missing  1 (0%) 0 (0%) 0 (0%) 0 (0%) 1 (0%) 
 Annual household income 778      
  <$10 000  53 (25%) 88 (40%) 57 (24%) 40 (37%) 238 (31%) 
  $10 000–$19 999  75 (36%) 53 (24%) 65 (27%) 21 (19%) 214 (28%) 
  $20 000–$39 999  56 (27%) 55 (25%) 56 (24%) 17 (16%) 184 (24%) 
  $40 000–$59 999  14 (7%) 18 (8%) 21 (9%) 10 (9%) 63 (8%) 
  $60 000 or more  13 (6%) 5 (2%) 21 (9%) 18 (17%) 57 (7%) 
  Caregiver did not know  0 (0%) 1 (0%) 18 (8%) 3 (3%) 22 (3%) 
 Number of children at home 780 2 [2–4] 2 [2–5] 2 [1–3] 2 [2–4] 2 [2–4] 
 Number of adults at home 780 2 [1–4] 2 [1–4] 2 [1–4] 2 [1–4] 2 [1–4] 
 Health literacy       
  S-TOFHLA score (0–36) 778 35 [31–36] 34 [16–36] 35 [29–36] 34 [19–36] 34 [22–36] 

Values are n (%) for categorical variables and median [10th–90th] percentiles for continuous variables.

a

All other, non-Hispanic included: 19 non-Hispanic and Asian American, 9 non-Hispanic and other, 2 non-Hispanic and mixed American Indian or Alaskan Native and white, and 1 non-Hispanic and American Indian or Alaskan Native.

As expected, the proportion of visits with injuries, and injuries requiring medical attention, increased with age (Fig 2). For example, at the 2-, 6-, 12-, 18-, and 24-month WCCs, parents recorded at least 1 injury in 3%, 9%, 25%, 31%, and 40% of children, respectively, and 0.4%, 3%, 11%, 16%, and 22% reported at least 2 injuries. Further, at these same respective WCCs, 0%, 1%, 2%, 4%, and 6% recorded at least 1 injury requiring medical attention.

FIGURE 2

First panel: Proportion of WCC visits with injuries. Second panel: Proportion of WCC visits with injuries requiring medical attention.

FIGURE 2

First panel: Proportion of WCC visits with injuries. Second panel: Proportion of WCC visits with injuries requiring medical attention.

Close modal

In answer to the question about the type of injuries since the last WCC (What were the causes of these injuries?), when parents were asked to check all that apply among the categories provided (fall; motor vehicle crash; heat, hot water, fire, or electricity; choking, poisoning, firearm, nearly drowning, or other), parents checked fall 626 times. The next most common injury checked was other, which occurred 163 times. The category heat, hot water, fire, or electricity was reported 31 times. Checks for all other categories (choking, poisoning/ingestion, motor vehicle crashes, and nonfatal drownings) were very low (n = 6, 6, 6, and 1, respectively).

Figure 3 shows the results from the primary regression analysis. The adjusted OR (95% CI) associated with more reported injuries is 0.77 (0.66–0.91), 0.60 (0.44–0.82), 0.32 (0.16–0.62), 0.26 (0.12–0.53), and 0.27 (0.14–0.52) at 4, 6, 12, 18, and 24 months, respectively, when comparing TIPP sites to Greenlight (active control) sites. Because the ORs are <1, these results are consistent with fewer reported injuries at TIPP sites compared with the Greenlight sites (P = .005). In an exploratory analysis (not shown), we conducted a binary logistic regression analysis (with generalized estimating equation) of falls at WCCs. OR estimates were consistent with those shown for the ordinal model. For example, the OR (95% CI) associated with falls (TIPP sites versus Greenlight sites) was estimated to be 0.33 (0.15–0.71) and 0.22 (0.09–0.53) at 12 and 24 months, respectively.

FIGURE 3

Results from the primary regression analysis. Adjusted ORs (95% CI) at each WCC associated with more reported injuries when comparing TIPP sites to Greenlight (active control) sites.

FIGURE 3

Results from the primary regression analysis. Adjusted ORs (95% CI) at each WCC associated with more reported injuries when comparing TIPP sites to Greenlight (active control) sites.

Close modal

Table 2 summarizes model results further to display absolute risks and risk differences of at least 1, at least 2, and at least 3 injuries across all time points for TIPP and Greenlight interventions. We estimate that the proportion of visits with at least 1 injury reported was 0.30 (95% CI) (0.23–0.39) and 0.14 (0.08–0.24) at the Greenlight and TIPP sites, respectively, with an absolute risk reduction equal to 0.16 (0.08–0.23). Similarly, the TIPP intervention was associated with an absolute risk reduction of 0.10 (0.05–0.14) for at least 2 injuries reported and 0.05 (0.03–0.08) for at least 3 injuries reported.

TABLE 2

Estimated Per-Visit Injury Proportions for the TIPP and Greenlight Intervention Arms Averaged Across WCC Visits and Participant Characteristics From 4 To 24 Months and the Risk Reduction Associated With the TIPP Arm

Number of Injuries ReportedTIPPGreenlightRisk Reduction
At least 1 injury 0.14 (0.08–0.24) 0.30 (0.23–0.39) 0.16 (0.08–0.23) 
At least 2 injuries 0.06 (0.03–0.11) 0.16 (0.11–0.22) 0.10 (0.05–0.14) 
At least 3 injuries 0.03 (0.01–0.05) 0.08 (0.05–0.11) 0.05 (0.03–0.08) 
Number of Injuries ReportedTIPPGreenlightRisk Reduction
At least 1 injury 0.14 (0.08–0.24) 0.30 (0.23–0.39) 0.16 (0.08–0.23) 
At least 2 injuries 0.06 (0.03–0.11) 0.16 (0.11–0.22) 0.10 (0.05–0.14) 
At least 3 injuries 0.03 (0.01–0.05) 0.08 (0.05–0.11) 0.05 (0.03–0.08) 

We provide a graphical display of the entire analysis model in Supplemental Fig 4 and point out several observations here. Compared with households with 1 child, those with at least 4 children exhibited lower odds of having more injuries (OR [95% CI] 0.63 [0.44–0.91]). We observed that Spanish-speaking Latino caregivers reported fewer injuries than English-speaking caregivers. Specifically, the ORs (95% CI) associated with more reported injuries were 1.84 (1.25–2.69), 1.48 (1.00–2.19), 1.46 (1.02–2.08), and 1.41 (0.81–2.46) when comparing non-Hispanic, white English speakers; Hispanic English speakers; non-Hispanic, Black English speakers; and non-Hispanic other race English speakers, respectively, to Spanish-speaking Hispanic caregivers. Finally, lower S-TOFHLA literacy scores were associated with fewer reported injuries. For example, estimated ORs (95% CI) were 0.80 (0.68–0.94), 0.61 (0.43–0.85), and 0.57 (0.40–0.81) when comparing S-TOFHLA scores of 34, 30, and 20 (low) to a score of 36 (high).

There were no significant adverse events or side effects in either intervention or control groups.

TIPP, implemented with pediatric resident support and tangible tools for parents, reduced parent-reported injuries throughout the first 24 months of life. This trial was conducted at 4 resident clinic sites with patient populations that have sociodemographic factors usually associated with higher injury incidence. Our study findings demonstrate that primary care-based interventions can effectively reduce reported injury and specifically provide evidence for implementing TIPP as part of resident-provided routine well-child care. Our study also sheds light on a previously understudied phenomenon of injuries that do not come to medical attention, and found that most injuries assessed by interval parental recall do not come to medical attention.42 

Although this study is the first randomized trial of TIPP, our findings align with previous research that has shown the benefits of injury prevention counseling, including the use of TIPP. A review of injury prevention interventions identified 18 studies that reported positive effects of injury prevention counseling on outcomes such as knowledge, behaviors, and number of injuries, with 15 of those studies involving physician counseling as part of the intervention strategy.8 One study of TIPP pediatrician injury counseling sessions showed large-estimated savings per child and even per visit, and a cost reduction of $230 million by 1995 monetary standards28; though, because our study examined injuries that did and did not come to medical attention, it is hard to know how much the TIPP program sessions would have saved in this case.

A strength of our study was the repeated query for injury at each standard health preventive services visit, which allowed us to track injuries that did not necessarily come to medical attention. In previous studies, injury visits are prevalent in ambulatory care, and half of the injury visits were to primary care physicians.1,2 However, rates of all injuries (including those that do not come to medical attention) are likely much higher than what was found in our study. One study conducted at local day care centers found 0.33 incidents per child over 120 hours of observation.42 

Unlike previous literature,3,24,26 our study did not find common associations of numbers of injuries with parental education, parental income, or child sex. Also, contrary to previous research,43,44 having more children in the household was associated with lower odds of reporting more injuries. Perhaps having 4 or more children in the household reduces the likelihood of reporting minor injuries for any 1 child. Or, perhaps, older children are helping with the supervision of younger children, resulting in fewer injuries. Spanish-speaking Latino caregivers reported fewer injuries than non-Hispanic, English-speaking caregivers, similar to previous reports3,45 and counter to others.46 Finally, also unexpectedly, having a lower S-TOFHLA (or health literacy) score was associated with having fewer reported injuries, despite previous research showing that low parent health literacy was related to injury risk behaviors.47,48 Further research is needed to better understand why those with lower health literacy scores reported fewer injuries. These findings may be related to how the question was understood by those with lower literacy, and/or differences in the threshold for reporting minor injuries among those with lower literacy.

Our study had several limitations. There may have been social desirability bias, because parents who received TIPP materials and counseling may have wanted to report fewer injuries than those who did not. However, it is typical for WCCs to include both injury and obesity prevention counseling, and our study included numerous questions about both injury and obesity prevention topics, which should have curbed the likelihood of this bias. Our findings also may be subject to recall bias, because we relied on parent reports of injuries, and, although we carefully defined injuries for parents and included minor injuries that are rarely captured in research studies, we suspect that parents are still underreporting injuries. Our study relied on a “missing at random” assumption, and we recognize that the missingness mechanism could have been informative (with reasons for missingness that were related to the outcome), which could potentially alter study results. Although we asked about injuries requiring medical attention, we were not powered to detect whether TIPP use decreased serious injury. In addition, our findings may not be generalizable to some settings because this study was conducted in 4 academic medical center pediatric resident physicians’ continuity clinics. However, these clinics have a disproportionate number of patients thought to be at higher risk for injury and are important venues for injury prevention. This was a cluster-randomized trial at only 4 sites, which has the potential to increase the type-1 error rate. It is important to note that, because of potential unmeasured confounders, residual confounding may exist. We included each of the nonintervention variables in the model to control for confounding of intervention–injury associations. We did not rigorously consider potential confounding for each of the nonintervention variable associations with injuries. Observed associations may undercontrol or overcontrol for confounding factors. Finally, our study included research enhancements of TIPP, including tangible tools and training of residents to certification. Although low cost and easy to implement, this study does not necessarily fully reflect how TIPP would operate in the “real world” without such enhancements.

Despite the limitations, there are public health implications of our findings. The rate of reported injury reduction that we saw here, an absolute risk reduction of 0.16, could have a significant effect if this intervention were implemented widely. Anticipatory guidance is a cornerstone of pediatrics, and TIPP supports provision of standardized and structured age-specific, developmentally appropriate guidance at each health preventive services visit. Unfortunately, few pediatric providers routinely provide injury prevention counseling in well-child care, with barriers of time, confidence in counseling, perceived attending expectations, competing priorities, lack of skills in behavioral change, and continuity cited.29,49,50 Quality improvement efforts, such as that reported by Gittelman et al,51 may be 1 way to overcome barriers to achieving adequate preventive service counseling at pediatric visits noted in other studies, such as the recent AAP periodic survey in which 69% of pediatricians reported lack of time, and a third reported lack of staff and electronic medical record prompts.52 

Further research is needed to determine if the use of TIPP results in reduction of serious injury and to examine the mechanisms by which TIPP leads to a reduction in injuries, including the role of parent attitudes, knowledge, and behaviors. Further research is also needed to determine the best TIPP implementation strategies, because implementation by pediatric providers or primary care teams is often difficult in busy practice settings. Finally, it would be good to conduct and implement a similar larger study in nontraining settings to verify the generalizability of these findings and continue to gather evidence on best practices to reduce injury morbidity for our nation’s children. Nevertheless, our research study provides important evidence of the efficacy of TIPP in resident pediatric clinics for reducing parent-reported injuries.

In a cluster-randomized trial that enrolled a large proportion of low-income and non-Hispanic Black and Hispanic families, TIPP reduced parent-reported injuries throughout the children’s first 24 months of life. In fact, there was an immediate and overall enhanced treatment effect from 4 months through 24 months, with those at TIPP sites reporting fewer injuries. Although further research is needed to determine if TIPP prevents serious injuries and prevents injuries in nontraining settings, our study provides important evidence that a primary care-based intervention can be effective in reducing injury.

We thank the many research coordinators of the Greenlight and TIPP projects who contributed to the work here, especially Joanne Propst, Sophie Ravanbakht, Janna Howard, Andrea Bronaugh, Lucila Bloise, Evelyn Cruzatte, Maria Cerra, Adriana Guzman, and Daniela Quesada. We thank Margaret Kihlstrom, Christine Connor, Lourdes Forster, Audrey Ofir, Barron Patterson, Seth Scholer, Cynthia Osman, and Mary Jo Messito who served as clinic champions as the study was being implemented. We also thank Carol Runyan, Marianna Garrettson, Andrea Gielen, and Judy Schaechter who advised us on this project. Finally, we thank Lauren Klein, Shafkat Meraj, and Divya Konduru for their literature searches and administrative support of this article.

Dr Perrin conceptualized and designed the study, codesigned and implemented the parent Greenlight Study, and drafted the initial manuscript; Dr Skinner conducted initial analyses of the study; Drs Sanders, Rothman, and Yin codesigned and implemented the Greenlight Study; Dr Schildcrout and Ms Bian conducted data analyses, and drafted portions of the initial manuscript; Drs Barkin, Coyne-Beasley, and Steiner made substantial contributions to conception and design, including creating the injury questions upon which this study is based; Drs Heerman, Flower, and Delamater contributed to the acquisition of data; and all authors reviewed and revised this manuscript critically for important intellectual content, approved the final manuscript as submitted, and agree to be accountable for all aspects of the work.

This study is registered at ClinicalTrials.gov, #NCT01040897, https://clinicaltrials.gov/ct2/show/NCT01040897. We will make data available to the scientific community with as few restrictions as feasible while retaining exclusive use until the publication of major outputs. Data are maintained at Vanderbilt University Medical Center. This includes individual participant data that underlie the results reported in this article. Requests for data, study protocol, analysis plan, and/or analytic code should include a methodologically sound proposal that does not preclude ongoing or planned analyses by the Greenlight Study team. Research data will be shared in a deidentified data set to protect subject privacy.

AAP

American Academy of Pediatrics

CI

confidence interval

Miami

University of Miami/Jackson Memorial Hospital

NYU

New York University

OR

odds ratio

S-TOFHLA

Short Test of Functional Health Literacy in Adults

TIPP

The Injury Prevention Program

UNC

University of North Carolina at Chapel Hill

Vanderbilt

Vanderbilt University Medical Center

WCC

well-child check

1
Freid
VM
,
Makuc
DM
,
Rooks
RN
.
Ambulatory health care visits by children: principal diagnosis and place of visit
.
Vital Health Stat 13
.
1998
;(
137
):
1
23
2
Cherry
DK
,
Hing
E
,
Woodwell
DA
,
Rechtsteiner
EA
.
National ambulatory medical care survey: 2006 summary
.
Natl Health Stat Report
.
2008
;(
3
):
1
39
3
Hambidge
SJ
,
Davidson
AJ
,
Gonzales
R
,
Steiner
JF
.
Epidemiology of pediatric injury-related primary care office visits in the United States
.
Pediatrics
.
2002
;
109
(
4
):
559
565
4
Centers for Disease Control and Prevention
.
Web-based injury statistics query and reporting system (WISQARS)
. Available at: https://www.cdc.gov/injury/wisqars/index.html. Accessed March 2, 2024
5
Hagan
JF
,
Shaw
JS
,
Duncan
PM
.
Bright Futures Guidelines for Health Supervision of Infants, Children, and Adolescents
, 4th ed.
Elk Grove Village
:
American Academy of Pediatrics
;
2017
:
235
256
6
Zonfrillo
MR
,
Gittelman
MA
,
Quinlan
KP
,
Pomerantz
WJ
.
Outcomes after injury prevention counselling in a pediatric office setting: a 25-year review
.
BMJ Paediatr Open
.
2018
;
2
(
1
):
e000300
7
Kendrick
D
,
Marsh
P
,
Fielding
K
,
Miller
P
.
Preventing injuries in children: cluster-randomized controlled trial in primary care
.
BMJ
.
1999
;
318
(
7189
):
980
983
8
Bass
JL
,
Christoffel
KK
,
Widome
M
, et al
.
Childhood injury prevention counseling in primary care settings: a critical review of the literature
.
Pediatrics
.
1993
;
92
(
4
):
544
550
9
Nansel
TR
,
Weaver
N
,
Donlin
M
,
Jacobsen
H
,
Kreuter
MW
,
Simons-Morton
B
.
Baby, be safe: the effect of tailored communications for pediatric injury prevention provided in a primary care setting
.
Patient Educ Couns
.
2002
;
46
(
3
):
175
190
10
Krassner
L
.
TIPP usage
.
Pediatrics
.
1984
;
74
(
5 Pt 2
):
976
980
11
Gielen
AC
,
McDonald
EM
,
Forrest
CB
,
Harvilchuck
JD
,
Wissow
L
.
Injury prevention counseling in an urban pediatric clinic. Analysis of audiotaped visits
.
Arch Pediatr Adolesc Med
.
1997
;
151
(
2
):
146
151
12
Schnitzer
PG
,
Dowd
MD
,
Kruse
RL
,
Morrongiello
BA
.
Supervision and risk of unintentional injury in young children
.
Inj Prev
.
2015
;
21
(
e1
):
e63
e70
13
Barkin
SL
,
Finch
SA
,
Ip
EH
, et al
.
Is office-based counseling about media use, timeouts, and firearm storage effective? Results from a cluster-randomized, controlled trial
.
Pediatrics
.
2008
;
122
(
1
):
e15
e25
14
Colson
ER
,
Willinger
M
,
Rybin
D
, et al
.
Trends and factors associated with infant bed sharing, 1993–2010: the National Infant Sleep Position Study
.
JAMA Pediatr
.
2013
;
167
(
11
):
1032
1037
15
Gardner
HG
.
American Academy of Pediatrics Committee on Injury, Violence, and Poison Prevention
.
Office-based counseling for unintentional injury prevention
.
Pediatrics
.
2007
;
119
(
1
):
202
206
16
Durbin
DR
;
Committee on Injury, Violence, and Poison Prevention
.
Child passenger safety
.
Pediatrics
.
2011
;
127
(
4
):
788
793
17
Committee on Injury, Violence, and Poison Prevention
.
Prevention of choking among children
.
Pediatrics (Evanston)
.
2010
;
125
(
3
):
601
607
18
Reading
R
,
Pressley
JC
,
Barlow
B
,
Kendig
T
,
Paneth-Pollak
R
.
Twenty-year trends in fatal injuries to very young children: the persistence of racial disparities
.
Child Care Health Dev
.
2007
;
33
(
5
):
647
19
Shenassa
ED
,
Stubbendick
A
,
Brown
MJ
.
Social disparities in housing and related pediatric injury: a multilevel study
.
Am J of Public Health
.
2004
;
94
(
4
):
633
639
20
Brown
RL
.
Epidemiology of injury and the impact of health disparities
.
Curr Opin Pediatr
.
2010
;
22
(
3
):
321
325
21
Moore
M
,
Conrick
KM
,
Fuentes
M
, et al
.
Research on injury disparities: a scoping review
.
Health Equity
.
2019
;
3
(
1
):
504
511
22
Tomashek
KM
,
Hsia
J
,
Iyasu
S
.
Trends in postneonatal mortality attributable to injury, United States, 1988–1998
.
Pediatrics
.
2003
;
111
(
5 Pt 2
):
1219
1225
23
Larson
CP
,
Pless
IB
.
Risk factors for injury in a 3-year-old birth cohort
.
Am J Dis Child
.
1988
;
142
(
10
):
1052
1057
24
Orton
E
,
Kendrick
D
,
West
J
,
Tata
LJ
.
Independent risk factors for injury in preschool children: 3 population-based nested case-control studies using routine primary care data
.
PLoS One
.
2012
;
7
(
4
):
e35193
25
Schwebel
DC
,
Gaines
J
.
Pediatric unintentional injury: behavioral risk factors and implications for prevention
.
J Dev Behav Pediatr
.
2007
;
28
(
3
):
245
254
26
Laflamme
L
,
Hasselberg
M
,
Burrows
S
.
Twenty years of research on socioeconomic inequality and children’s unintentional injuries understanding the cause-specific evidence at hand
.
Int J Pediatr
.
2010
;
2010
:
1
23
27
Simon
TD
,
Phibbs
S
,
Dickinson
LM
, et al
.
Less anticipatory guidance is associated with more subsequent injury visits among infants
.
Ambul Pediatr
.
2006
;
6
(
6
):
318
325
28
Miller
TR
,
Galbraith
M
.
Injury prevention counseling by pediatricians: a benefit-cost comparison
.
Pediatrics
.
1995
;
96
(
1 Pt 1
):
1
4
29
Gielen
AC
,
Wilson
MEH
,
McDonald
EM
, et al
.
Randomized trial of enhanced anticipatory guidance for injury prevention
.
Arch Pediatr Adolesc Med
.
2001
;
155
(
1
):
42
49
30
Powell
EC
,
Malanchinski
J
,
Sheehan
KM
.
A randomized trial of a home safety education intervention using a safe home model
.
J Trauma
.
2010
;
69
(
4
Suppl
):
S233
S236
31
Sanders
LM
,
Perrin
EM
,
Yin
HS
,
Bronaugh
A
,
Rothman
RL
.
Greenlight Study Team
.
“Greenlight study”: a controlled trial of low-literacy, early childhood obesity prevention
.
Pediatrics
.
2014
;
133
(
6
):
e1724
e1737
32
Sanders
LM
,
Perrin
EM
,
Yin
HS
, et al
.
Greenlight Study Team
.
A health-literacy intervention for early childhood obesity prevention: a cluster-randomized controlled trial
.
Pediatrics
.
2021
;
147
(
5
):
e2020049866
33
Harris
PA
,
Taylor
R
,
Thielke
R
,
Payne
J
,
Gonzalez
N
,
Conde
JG
.
Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support
.
J Biomed Inform
.
2009
;
42
(
2
):
377
381
34
Liang
K
,
Zeger
SL
.
Longitudinal data analysis using generalized linear models
.
Biometrika
.
1986
;
73
(
1
):
13
22
35
Zeger
SL
,
Liang
KY
.
Longitudinal data analysis for discrete and continuous outcomes
.
Biometrics
.
1986
;
42
(
1
):
121
130
36
White
H
.
A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity
.
Econometrica
.
1980
;
48
(
4
):
817
838
37
Huber
PJ
.
The behavior of maximum likelihood estimates under nonstandard conditions
.
Berkeley Symp. on Math. Statist. and Prob
.
1967
;
5
(
1
)
221
233
38
Raghunathan
TE
,
Lepkowski
JM
,
Hoewyk
JV
,
Solenberger
PW
.
A multivariate technique for multiply imputing missing values using a sequence of regression models
.
Surv Methodol
.
2001
;
27
:
85
95
39
Rubin
DB
.
Multiple Imputation for Nonresponse in Surveys
.
New York
:
John Wiley & Sons
;
1987
40
Harrell
Jr
FE
.
Rms: regression modeling strategies
. Available at: https://CRAN.R-project.org/package=rms. Accessed January 17, 2024
41
R Core Team
.
R: a language and environment for statistical computing
. Available at: https://www.R-project.org/. Accessed January 17, 2024
42
Fearon
I
,
McGrath
JP
,
Achat
H
.
‘Booboos’: the study of everyday pain among young children
.
Pain
.
1996
;
68
(
1
):
55
62
43
Otters
H
,
Schellevis
FG
,
Damen
J
,
van der Wouden
JC
,
van Suijlekom-Smit
LWA
,
Koes
BW
.
Epidemiology of unintentional injuries in childhood: a population-based survey in general practice
.
Br J Gen Pract
.
2005
;
55
(
517
):
630
633
44
Morrongiello
BA
,
House
K
.
Measuring parent attributes and supervision behaviors relevant to child injury risk: examining the usefulness of questionnaire measures
.
Inj Prev
.
2004
;
10
(
2
):
114
118
45
Bernard
SJ
,
Paulozzi
LJ
,
Wallace
DL
.
Centers for Disease Control and Prevention (CDC)
.
Fatal injuries among children by race and ethnicity–United States, 1999–2002
.
MMWR Surveill Summ
.
2007
;
56
(
5
):
1
16
46
Aagran
PF
,
Winn
DG
,
Anderson
CL
,
Del Valle
CP
.
Pediatric injury hospitalization in Hispanic children and non-Hispanic white children in southern California
.
Arch Pediatr Adolesc Med
.
1996
;
150
(
4
):
400
406
47
Heerman
WJ
,
Perrin
EM
,
Yin
HS
, et al
.
Health literacy and injury prevention behaviors among caregivers of infants
.
Am J Prev Med
.
2014
;
46
(
5
):
449
456
48
Cheng
ER
,
Bauer
NS
,
Downs
SM
,
Sanders
LM
.
Parent health literacy, depression, and risk for pediatric injury
.
Pediatrics
.
2016
;
138
(
1
):
e20160025
49
Cohen
LR
,
Runyan
CW
.
Barriers to pediatric injury prevention counseling
.
Inj Prev
.
1999
;
5
(
1
):
36
40
50
Schuster
MA
,
Duan
N
,
Regalado
M
,
Klein
DJ
.
Anticipatory guidance: what information do parents receive? What information do they want?
Arch Pediatr Adolesc Med
.
2000
;
154
(
12
):
1191
1198
51
Gittelman
MA
,
Carle
AC
,
Denny
S
,
Anzeljc
S
,
Arnold
MW
.
A quality improvement program in pediatric practices to increase tailored injury prevention counseling and assess self-reported changes made by families
.
Inj Epidemiol
.
2018
;
5
(
Suppl 1
):
17
70
52
Burr
WH
,
Lee
LK
,
Hoffman
BD
,
Somberg
C
,
Zonfrillo
MR
.
Pediatrician-reported injury prevention anticipatory guidance by patient age group
.
Acad Pediatr
.
2023
;
23
(
3
):
610
615

Competing Interests

CONFLICT OF INTEREST DISCLOSURES: The authors have indicated they have no conflicts of interest relevant to this article to disclose.

Supplementary data