A Randomized Controlled Trial for Parents of Hospitalized Children: Keeping Kids Safe From Guns

This is a randomized controlled, 3-arm preintervention and postintervention study investigating the effect of an educational intervention by using the original 5-minute, 20-second BeSMART video²⁴  and written material¹³  on parental and/or legal guardian behaviors, beliefs, and knowledge regarding firearm-related injury prevention. As tobacco smoke exposure interventions have been studied in hospitalized children, we used a video and written materials on tobacco smoke exposure education as our control.²³  We used survey techniques (see Supplemental Information) for 3 study groups receiving the same study questions on firearms and tobacco smoke exposure to assess behaviors, beliefs, and knowledge in parents and/or legal guardians of hospitalized children aged <20 years preintervention, postintervention, and 1 month after intervention.

The study was done at a single-site quaternary care children’s hospital in a major city with a diverse population, in an area with relatively strict gun laws.

One parent and/or legal guardian of hospitalized children aged <20 years. The same parent and/or guardian was surveyed at 3 study time points: preintervention, postintervention, and 1 month after intervention.

Parents and/or guardians who reside with a hospitalized child aged <20 years.

Parent and/or guardian who was previously enrolled in the study, with preferred language other than English or Spanish (for consent), or with a child in the ICU and/or who is critically ill or acutely distressed.

Study personnel approached a convenience sample on the basis of recruiter availability (March 2017 to July 2019) of parents and/or guardians of hospitalized children aged <20 years and assessed eligibility. After informed consent, participants were randomly assigned to 1 of 3 study groups: (1) BeSMART alone, (2) BeSMART + physician review, or (3) control. Participants were randomly assigned by using a computer-generated block randomization scheme (www.randomization.com) (1:1:1 allocation; blocks of 15). Researchers proceeded to verbally administer the initial parent and/or guardian survey (preintervention time point) to all study groups. After the initial survey, subjects received the intervention by study group: (1) the BeSMART alone group received the BeSMART video and BeSMART written material; (2) the BeSMART + physician review group received the BeSMART video and BeSMART written material + standardized reinforcement and/or review with a physician; and (3) the control group received the Kids and Smoke Don’t Mix²³  video and written materials from our state’s Smokers Quitline. For patients in the BeSMART + physician review group, a physician verbally reviewed all elements of BeSMART with participants via a checklist and answered participant questions. Immediately after interventions (postintervention time point), researchers conducted a brief, verbally administered postintervention parent and/or guardian survey, including a question evaluating their likelihood of asking if there is a gun in others’ homes before visiting. One month later (1-month time point), study personnel verbally administered, via telephone, the 1-month follow-up parent and/or guardian survey to assess actual behavior over the last month in asking if there is a gun in others’ homes. Questions in surveys were adapted from previously published studies in which youth firearm exposure and safety^8,25–28  and tobacco exposure are addressed.²³  All study groups received the same set of questions on surveys. Survey questions were pilot tested with a sample of parents of hospitalized children before the study (see Supplemental Information for surveys).

Our primary outcome was the change in parents’ and/or guardians’ behavior between preintervention and 1 month after in asking if there were guns in others’ homes when their child(ren) visited, as indicated by parent-reported Likert scale assignment of an ordinal value (1 = never, 2 = rarely, 3 = sometimes, 4 = most of the time, and 5 = always). We compared both within- and between-group changes. We chose this as our outcome measure because “asking about guns where children visit” is a gun safety measure applicable to gun-owning and non–gun-owning families alike. Additionally, pilot data revealed low reported gun ownership in our hospital population, therefore making safe gun storage a difficult outcome to study.

Secondary outcomes included (1) comparison of mean Likert score (1 = very unlikely, 2 = unlikely, 3 = not sure, 4 = likely, and 5 = very likely) of likelihood of asking about firearms in others’ homes immediately postintervention between study groups (in contrast to the primary outcome of parent-reported actual behavior, this outcome focuses on parent-reported likelihood or intention of asking); (2) change in parental beliefs about firearm safety preintervention and postintervention within groups, measured by mean Likert score of agreement with belief statements; (3) change in parental general knowledge about firearm-related injury prevention preintervention and postintervention within groups, measured by correct responses to factual questions; and (4) parental beliefs regarding physicians counseling on gun safety.

For our primary outcome, we assessed within each study group if parent-reported behavior of asking about firearms in the homes of others when their children go to visit increased after the intervention. Additionally, we performed a head-to-head comparison of the magnitude of the intervention effects between groups. Pilot data revealed the proportion of the population who either always (Likert scale score = 5) or most of the time (Likert scale score = 4) asks about guns in others’ homes was 50%. A sample size of 60 subjects each in the 2 intervention groups, BeSMART alone and BeSMART + physician review, would yield at least 80% power to detect an absolute within-group increase of 20% between the preintervention and 1-month postintervention responses (ie, raising the proportion of those who most of the time or always ask about guns in others’ homes from 50% to 70%) by using the McNemar’s test with a 2-sided α level of .05. We expected minimal change in the preintervention and 1-month postintervention responses in the control group. We anticipated most subjects enrolled would complete the inpatient portion of the study; however, we estimated up to 20% would not complete the 1-month postintervention call. Given our primary outcome was based on behavior assessed at 1-month postintervention, we enrolled an additional 15 subjects in each study group for a total sample size of 225 subjects.

We summarized characteristics by study group and compared using analysis of variance (ANOVA) or Kruskal–Wallis tests for continuous variables and χ² or Fisher’s exact test for categorical values, as appropriate. For the primary outcome, we analyzed preintervention and 1-month postintervention changes measured by mean Likert score within groups using the paired t test. We compared preintervention and 1-month postintervention changes between groups using generalized estimating equations (GEEs) models fit to the data. As a sensitivity analysis for the primary outcome, we also examined the change in the proportion of respondents endorsing most of the time or always asking about guns in others’ homes, repeating the process above using McNemar’s test for within-group comparisons, and GEEs for between-group comparisons.

For our secondary outcomes, we used the following analyses: (1) ANOVA for likelihood of asking, (2) paired t tests for change in parental beliefs, (3) McNemar’s test to assess changes in the proportion of correct responses to knowledge questions, and (4) t test for parental belief about physician counseling.

Analyses were conducted by using SAS 9.4 (SAS Institute, Inc, Cary, NC).

This study was approved by our college of medicine’s institutional review board.

A total of 225 parents and/or guardians enrolled, with 76, 75, and 74 participants in each of 3 study arms. Sixty-two percent of BeSMART, 60% of BeSMART + physician review, and 58% of control group participants completed the 1-month follow-up (Fig 1). Groups were similar in demographic characteristics except for child sex (Table 1). Median child age was 5 years (interquartile range [IQR] 1.0–11.0). Fifty-six percent of children were Hispanic and 30.2% were non-Hispanic Black; 68.2% were insured by Medicaid. Median parent and/or guardian age was 33 years, and 87.1% of parent and/or guardian participants were female. Groups were similar with respect to baseline questions about exposure to and concerns about gun violence (Table 1). Sixty percent of parents reported having heard gunshots in their neighborhood, and >40% said a friend or relative had been shot, with similar numbers across study groups, revealing the relatively common exposure to gun violence in our study population.

Within each of the 3 study groups, there was a significant increase between preintervention and 1-month postintervention mean Likert score frequency of asking about guns in others’ homes (BeSMART: 1.5 to 2.3, P = .04; BeSMART + physician review: 1.4 to 1.9, P = .03; control: 1.4 to 2.3, P = .01). Between-group differences in preintervention and 1-month postintervention mean Likert scores were not significant (P = .81). The changes in proportion of respondents endorsing most of the time or always within each study group was nonsignificant (Tables 2–4).

1. Postintervention, there was a significant difference between the 3 study groups in parent-reported likelihood of asking about guns in others’ homes before visiting, with mean Likert scale scores out of 5 as follows: BeSMART = 4.3 (SD1.1), BeSMART + physician review = 4.4 (SD 1.0), and control group = 3.7 (SD 1.4) (P < .001) (Fig 2). This outcome differs from the primary outcome by assessing likelihood or intentions of asking at 1 month postintervention versus parent-reported actual behavior (belief).

2. When asked whether “people who have guns should use gun locks,” the mean Likert score response was ≥4.5 (4 = agree and 5 = strongly agree) both preintervention and postintervention for all study groups (belief).

3. There was significant preintervention and postintervention increase in knowledge of the most effective way to keep kids safe from guns in the home in the 2 BeSMART groups that was not seen in the control group (BeSMART: 74.7% to 92.0%, P = .003; BeSMART + physician review: 74.7% to 96.0%, P ≤ .001; control: 74.3% to 79.2%, P = .26) (knowledge).

   Participant responses to questions about beliefs and knowledge about guns are reported in Tables 5 and 6.

4. At 1 month postintervention, parents agreed with the statement “Doctors who take care of children should talk to parents/guardians about safe gun storage,” with a mean Likert score of ≥4.2 across all study groups: BeSMART = 4.2 (SD 0.9), BeSMART + physician review = 4.3 (SD 0.8) and control group = 4.3 (SD 0.8) (belief).

In this 3-arm, randomized controlled study, researchers found that participants in all study groups reported an increase in asking about guns in others’ homes when their children went to visit 1 month after the study. Although the proportion of parents who always or almost always ask about guns in others’ homes at 1-month after intervention did not reveal a statistically significant increase, the implication of the increase in the mean Likert score of asking in all 3 groups is important. This BeSMART firearm injury prevention intervention delivered in a hospital setting significantly and strikingly increased parental knowledge regarding firearm safety, including safe storage knowledge, when compared with the control group. As previous studies have revealed, actual behavior change may lag behind knowledge change.²⁹  For all study groups, the needle moved from asking “never to rarely” to “rarely to sometimes” at 1-month after intervention, which may be an important practical shift or starting point for changing behavior.

This study also revealed postintervention that the educational interventions resulted in parents and/or guardians in the 2 BeSMART groups reporting significantly higher likelihood of asking about guns in others’ homes when their children visit, as compared with the control group. Both BeSMART groups, compared with the control group, revealed statistically stronger increase in agreement postintervention with the statement “Stricter gun laws will decrease gun-related violence,” although all 3 study groups revealed agreement with the statement. All 3 study groups also reported agreement with the statement “Doctors who take care of children should talk to parents/guardians about safe gun storage.” Because these beliefs are held across the study population, a practical take-home point for pediatricians is that parents in this study population are receptive to discussing safe gun storage.

Our study revealed that the BeSMART video is a feasible intervention in the inpatient setting to discuss firearm-related injury prevention. Improvements were seen in all groups, which may be due to the idea that raising the topic with parents may alter behavior, given that all participants received both tobacco smoke exposure questions and firearm-related questions; or, because of the Hawthorne effect,³⁰  with a verbally administered survey, participants might know what they are “supposed to say.” It is possible that other techniques revealing positive behavior change related to gun safety interventions, such as motivational interviewing, might provide more benefit than counseling and a device alone.^7,31  One discrepancy highlighted was the difference between beliefs and knowledge and behavior; most parents agreed it is important to ask about guns in other’s homes; however, only a minority of parents always or almost always ask. There are several possibilities why the discrepancy exists, which may relate to theoretical models of behavior change, including unconscious barriers beyond the scope of this article, but may be analogous to studies around seat belt use and behavior change.²⁹ 

Our findings add to the literature around educational interventions on gun safety, supporting findings of other studies in outpatient or emergency department settings.^6–8,32,33  In this study, we convincingly demonstrate that the inpatient setting is also an area in which physicians can provide anticipatory guidance to parents on gun-related injury prevention and safety; thus, we add significantly to what is known about this subject. Additionally, providing this education while admitted can be an opportunity to deliver potentially life-saving guidance to families and patients, including those who may have a gap in consistent primary care, children with chronic illness who may view their subspecialists as their primary doctor,¹⁹  or those whose primary physicians, for many reasons, may not have discussed gun-related injury prevention.^10,34  BeSMART is feasible and well-received, increases knowledge about safe gun storage, and supports other findings of parents’ receptivity to counseling about safe storage.^35,36 

There are several limitations to this study. First, given limited funding, enrollment was via convenience sample on the basis of study personnel availability. Although certain beliefs about gun violence may be influenced by timeliness related to current events, this factor should affect all groups equally and likely be distributed over the study period. Additionally, because of limited study personnel availability on the weekends, it is theoretically possible that short-stay, weekend-only subjects may have differed from the overall population had they been included; however, this seems unlikely. Similarly, because of lack of optimal funding, the survey was most feasible by verbal administration (face-to-face and/or phone). Answers to sensitive questions might be more honest if the survey was read independently at an optimized reading level or read to the participant via technology assistance via electronic tablet or another device.

One of the main strengths of this study was researchers’ establishing parental willingness to discuss gun-related injury prevention and to be part of such a study, thereby supporting future studies with optimized technology and enrollment strategies. This study was limited by the small number of paired samples, fewer than the calculated sample size of n = 60, because of difficulty reaching many subjects at the 1-month follow-up and the numerous participants who reported not applicable for the primary outcome (their child had not been to another’s house in the previous month). This limited the power for analysis of our primary outcome more than anticipated. The number of parents reporting not applicable may help inform our approach to providing gun safety advice. Parents often do not consider going to a family member’s or close friend’s house as a time to ask about unsecured guns; however, tragedies do occur with guns accessed from the homes of family members and close friends.

Additionally, as a single-center study, the application of our findings may not be generalizable to other regions and settings given variation in local gun culture, beliefs, and behaviors. In this study, we used the original BeSMART video, which is limited in diversity and, therefore, may be less relatable for various populations. Despite these limitations, parents in our diverse study population with substantial exposure to gun violence, were receptive to the intervention and discussions around gun violence prevention. It is unclear what impact the high exposure to gun violence reported by parents has on the results of this intervention, but it would be an interesting aspect for future study. Finally, this was not a blinded study and not feasible when delivering a viewable educational intervention; therefore, the Hawthorne effect may have altered parental responses by being observed and/or spoken to by researchers.

The BeSMART firearm injury prevention intervention (brief video and written materials) delivered in a hospital setting is feasible and increases parental knowledge regarding firearm safety. At 1 month after intervention, all groups reported increased frequency of asking about guns. Most parents are receptive to doctors talking to them about safe gun storage. Future multicentered studies using the newer, shorter (3 minutes, 10 seconds), more diverse, and more broadly representative video that more directly related to the BeSMART tenets would be beneficial to examine these findings in different settings, given the diversity of opinions and/or gun culture and rates of gun ownership in different regions. Additionally, in future studies, researchers could consider other techniques that may help with behavior change and include longer-term follow-up to see if findings persist (eg, 6 months). In addition, given the discrepancy between parental receptiveness to physician counseling on safe gun storage and low report that their child’s doctors have provided counseling, in future studies with qualitative techniques both with parents and providers, researchers may investigate barriers to and ways to increase pediatricians providing gun safety guidance to parents and similarly examine educational interventions for trainees to increase pediatrician counseling on this topic.

We thank the patients and families who participated in this study. We also thank Henna Boolchandani, Shana Burstein, Christine Knauer Costa, Karolina Mieczkowska, Casey Pitts, Jessica Tugetman, and Grace Ye for their assistance in enrolling participants in the study and assistance with data collection and data entry.