Firearm Storage in Homes With Children With Self-Harm Risk Factors

We used data from a Web-based survey designed by the authors [D.A., M.M.] and conducted by the survey firm Growth for Knowledge (GfK) to assess firearm ownership, storage practices, and use among a nationally representative sample of US adults. Respondents were drawn from GfK’s KnowledgePanel (KP), a group of ∼55 000 US adults selected (on an ongoing basis) with an equal probability of selection. All panel members, except those currently serving in the US Armed Forces, were eligible to participate. Gun owners and veterans were oversampled from the KP; sampling weights supplied by GfK were applied such that estimates from the survey are representative of US adults (aged ≥18 years) in 2015.

Survey pretesting occurred in March 2015, with administration of the final survey in April 2015. Potentially eligible panel members received a notification e-mail letting them know that a new survey was available for them to take. No description of the survey content accompanied the invitation. This e-mail notification contained a link that sent them to the survey questionnaire. No login name or password was required. After 3 days, automatic e-mail reminders were sent to all nonresponding panel members in the sample. Participants completed the main survey within a median of 14 minutes.

Unique identifiers linked respondents in our survey to routine profile data collected by GfK on all KP panel members to ascertain self-reported mental and behavioral health conditions for respondents and their children. Response to GfK surveys is motivated through a point system by which responders accrue points that are redeemable for merchandise, cash or participation in sweepstakes. All panel members were eligible for survey participation except those serving on active duty in the US Armed Forces. The Institutional Review Board of Northeastern University approved the study.

GfK structures recruitment for the KP with the goal of having the resulting panel represent the adult population of the United States with respect to a broad set of geodemographic distributions as well as particular subgroups of hard-to-reach adults (for example, those without a landline telephone or those who primarily speak Spanish). For selection of general population samples from KP, GfK uses an equal probability of selection method design by weighting the entire KP to the benchmarks from the latest March supplement of the US Census Current Population Survey (see Supplemental Information for details). Sampling weights supplied by GfK were applied such that estimates made from the survey are representative of US adults (≥18 years). We used weighted percentages and 95% confidence intervals (CIs) to describe outcomes.

Household gun ownership was assessed by asking respondents, “Do you or does anyone else you live with currently own any type of gun?” Personal gun ownership was assessed with a follow-up question: “Do you personally own a gun?” Gun owners were then asked follow-up questions about various storage practices for their household guns (eg, the number of guns stored loaded and unlocked, loaded and locked, unloaded and unlocked, and locked and unloaded).

We classified household firearms based on the least safely stored firearm. After having ascertained how many handguns and how many long guns a respondent owned, respondents were asked separately about the number of handguns and then about the number of long guns stored (1) loaded and unlocked, (2) unloaded and locked, and (3) in some other manner. For each type of gun we asked, “how many are stored locked and unloaded?”; the respondent would answer with an integer. Storage was not ascertained about guns individually (unless the respondent owned only 1 handgun and/or 1 long gun). Responses were sorted into 1 of 3 mutually exclusive, hierarchical, and collectively exhaustive categories: Category A (at least 1 gun loaded and unlocked); Category B (no guns in Category A but at least 1 gun either loaded and locked, or unlocked and unloaded), and Category C (all guns stored locked and unloaded). We also characterized household firearm storage more broadly, into 2 potentially overlapping categories: (1) any firearms stored unlocked, and (2) any stored loaded.

Basic demographic information was also elicited, as was information about the number of children in the household and whether any of these children fit into 3 mutually exclusive age categories (between 0–5 years, 6–12 years but none younger, and 13–17 years of age but none younger).

Because not all adults who live in homes with children are the child’s parent or guardian, we further ascertained the relationship between adult respondents who live in homes with children by asking respondents if they were “a caregiver or health care decision-maker for a child under the age of 18?” Hereafter we refer to respondents who responded affirmatively to the caretaker and guardian question as “parents.” Members of the GfK panel were also asked: “Are you a caregiver for a child under 18 with any of the following medical conditions (eg, depression).” Respondents could answer “yes” or “no” to the following 3 categories of conditions: attention-deficit/hyperactivity disorder (ADHD) or attention-deficit disorder, depression, and mental health conditions other than depression. Respondents who answered “yes” to any of these 3 categories of established self-harm risk factors^22,–24 were said to have a child with a self-harm risk factor. Respondents were not asked more specifically about these conditions (eg, we do not know whether children said to have these conditions ever received a formal diagnosis, or if respondents were asked whether their child had other mental health or behavioral health conditions more specifically).

We conducted all analyses by using Stata IC 14 (StataCorp, College Station, TX), with use of appropriate weighting commands (using the weight variable provided by GfK) to generate national estimates and following the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for reporting.²⁵ 

Of the 7318 invited panel members who received the survey, 4165 began the survey and 3949 completed it (excluding 48 active-duty military personnel who began the survey but were ineligible to complete it). This yielded a survey completion proportion of 54.6% based on the formula recommended for calculating response proportions for Web panels.²⁶ Characteristics of the underlying sample are presented in Table 1.

Approximately 1 in 3 US households contained firearms (34.8%; 95% CI: 32.2–36.8), irrespective of whether the household contains children (33.4%; 95% CI: 29.1–38.0), or not (34.9%; 95% CI: 29.1–38.0) (Fig 1). Among the subset of adults who self-identified as a caregiver or health care decision-maker for a child under the age of 18, ∼2 in 5 households contained firearms (42.4%; 95% CI: 35.7–49.4), whether the children in the household had a history of self-harm risk factors (43.5%; 95% CI: 24.0–65.2), or not (42.3%; 95% CI: 35.2–49.7) (Fig 1).

Among parents who own guns, only 1 in 3 stored all household firearms locked and unloaded, regardless of whether their child had or did not have a history of depression, mental health conditions other than depression, or ADHD (34.9% [95% CI: 20.2–53.2] vs 31.8% [95% CI: 25.9–38.3], respectively) (Fig 1). The proportion of gun-owning parents who store at least 1 household firearm both loaded and unlocked is lower, but not significantly so, for parents whose children have a history of these self-harm risk factors, compared with parents none of whose children have such a history (11.6% [95% CI: 4.3–27.6] vs 20.3% [95% CI: 15.3–26.4], respectively) (Fig 1). Gun-owning parents who have children with a history of depression, mental health conditions other than depression, or ADHD are also not significantly more likely to store all household firearms locked, compared with those whose children do not have such a history of self-harm risk factors (62.0% [95% CI: 44.8–76.6] vs 52.2% [95% CI: 45.3–58.9], respectively) (Table 2). Likewise, the proportion of gun-owning parents who store all household firearms unloaded does not does not depend on whether their child has or does not have a history of mental health conditions or ADHD (58.7% [95% CI: 41.2–74.3] vs 57.9% [95% CI: 51.0–64.5], respectively) (Table 2).

In homes that contain firearms, parents whose children have a history of mental health conditions or ADHD tend to have more children and a greater proportion of children in older age groups. For example, approximately one-third (36.1%; 95% CI: 30.5–42.1) of parents, none of whose children have a history of mental health conditions or ADHD, have children 0 to 5 years of age compared with 7% (95% CI: 2.3–19.2) of parents who have at least 1 child with such a history (Table 2). Whether a child has a history these self-harm risk factors does not appear to appreciably affect the distribution of the other measured parental and household characteristics.

To our knowledge, in the current study, we are the first to use a nationally representative sample to examine whether having a child with a history of mental health conditions or ADHD is associated with parental reports of (1) household firearm prevalence, and (2) how household firearms are stored in homes with guns and children. Based on our findings, it appears that having a child with self-harm risk factors does not, on average (1) factor into parents’ decision to have firearms in their homes, or (2) materially affect whether parents store all household guns in accordance with AAP recommendations (locked and unloaded).² 

Our finding that self-harm risk factors among US children are not associated with household firearm prevalence mirrors findings from 4 previous US studies of adults from the early 2000s, in each of which authors found that the prevalence of mental health disorders was fairly balanced across households with firearms versus without firearms.^21,27,–29 Similarly, our finding that parent’s report of their child’s history of self-harm risk factors did not predict whether parents stored household firearms locked and unloaded is consistent with findings from a nationally representative study of adults that failed to find an association between an adult’s mental health history and the likelihood of living in a home in which firearms were stored loaded and unlocked.²⁸ Moreover, our findings regarding storage practices are in line with a nationally representative survey of 13 to 18 year old adolescents who were interviewed between 2001 and 2003 and asked a related, but distinct, question about self-reported access to firearms in the home. In that study, adolescents with self-reported mental health risk factors for suicide were as likely to report easy access to household firearms as were adolescents without risk factors for suicide.²⁰ 

Our findings that household firearm prevalence in homes with children is modestly higher, albeit with overlapping tails of the 95% CIs, when prevalence estimates are based on respondents who self-identify as a child’s caregiver or health care decision-maker (eg, parents) compared with estimates based on a randomly chosen adult living in a home with children (eg, an adult sibling, a grandparent, a related or unrelated adult boarder) has not previously been reported (nor to our knowledge, assessed). Other survey-based firearm-related reporting gaps have been well documented in the literature, especially with respect to sex.^30,–33 The authors of these previous studies all suggest that household prevalence estimates for a variety of firearm related attributes (eg, presence of guns, number of guns, how guns are stored in the home) depend in nontrivial ways on which household member contributes information to estimates. Historically, estimates based on reports of respondents who are less familiar with guns (eg, females, nongun owners) appear to underestimate household prevalence and overestimate how safely guns are stored (the latter is why, by design, our study ascertained storage practices from the gun owner only). The extent to which prevalence discrepancies in our study, or in previous work, reflect the role of chance, respondent-dependent differential knowledge about household firearm status, or differential reluctance to report accurate knowledge of firearm status is unknown.

Several additional considerations should be kept in mind when interpreting findings from our survey. First, our null finding with respect to a child’s history of self-harm risk factors affecting the proportion of parents reporting that they store all household firearms in accordance with the AAP recommendations may, at least in part, be confounded by the fact that older children are more likely to have a history of mental health and behavioral conditions (as we observe in our study). In previous studies, authors have shown that although the risk of firearm injury is greater for adolescents than for younger children, parents tend to adopt safer firearm storage practices when they have young children, compared to when they have only older children.‍^33,34 Our sample size is not large enough to generate precise estimates of the relationship between firearm storage practices and a child’s mental and behavioral health history while adjusting for the age of children in the home or, for that matter, by other potentially relevant subgroups, such as type of firearm (ie, handgun versus long gun). Our sample size also precludes stratifying our measure of self-harm risk factors, and therefore precludes examining effect modification by more granular measures of mental or behavioral health. Lastly, our study is not powered to know whether the nonsignificant difference by self-harm risk factor status that we observed for the least safe storage practice measured (ie, 11% vs 20% loaded and unlocked) is due to the play of chance, or to real differences in this particular storage practice. Even so, our findings nevertheless suggest that too many children in every age category and regardless of their history of self-harm risk factors live in homes in which firearms are stored in less than an ideal fashion.

Second, we categorize children as having a mental or behavioral health condition based on parents’ response to survey items that have not been validated. This could result in misclassification if parents are not aware of their child’s full condition or withhold relevant information. To the extent that such misclassification is relevant, extrapolations from our findings would likely underestimate the number of children with self-harm risk factors living in homes with firearms, but it might also obscure differences in storage practices when children have conditions not reported or if storage is related to the severity of a mental or behavioral health condition or its recency of onset. Authors of larger studies that use validated measures of mental health and behavioral conditions, capture the severity of underlying conditions and report recency of disorder onset could add meaningfully to the findings from the current study.

Third, as with findings from all self-reported surveys, our findings are subject to potential inaccuracies because of a social desirability bias that might lead parents to report spuriously safe storage practices, especially if their children have risk factors for self-harm. However, with our findings that one-third of all parents who own guns report storing household firearms unloaded and locked up, independent of whether they had reported having a child with self-harm risk factors, we suggest that social desirability bias may not play a dominant role in our findings. With regard to social desirability bias more generally, online panel surveys, such as ours, have been shown to reduce social desirability bias and yield more accurate estimates of respondent characteristics compared with alternative methods used to elicit opinions, such as telephone surveys.³⁵ Reduced social desirability bias, along with our decision to ascertain storage information from gun owners only, and secular trends favoring handgun over long gun ownership (and protection rather than recreation as a primary reason to own guns),³⁶ may help explain why we find less optimal storage practices in homes with children and guns compared with previous studies.^33,34,37 Another advantage of online panels is high completion rates for those who begin the survey and the availability of information about panelists who do not elect to take the survey in the first place.²⁶ In our study, 99% of respondents completed the survey, fewer than 1% declined to answer our stem question about household gun ownership, no one declined to answer the subsequent question regarding whether they personally owned a gun, and fewer than 3% declined to answer the questions about storage. Finally, our survey completion rate (55%) is higher than the rates for typical nonprobability, opt-in, online surveys, which are 2% to 16%³⁸; higher than those of previous national injury surveys that included questions about firearm ownership^39,40; and similar to those from other surveys conducted by GfK.³⁷ Nevertheless, panel members who chose not to participate in our survey may have differed in important ways related to their storage practices compared with panel members who chose to participate.

Despite these limitations, our findings indicate that parents’ decision to store firearms in their home does not materially depend on whether their child has self-harm risk factors, nor, moreover, does having a child with a history of self-harm risk factors appear to increase the likelihood that parents in homes with firearms will store their household guns as safely as possible. Indeed, for homes with children and guns, the odds are roughly 2 to 1 that firearms are not stored in accordance with recommendations promulgated by the AAP, regardless of whether children in the home have a history of self-harm risk factors. Given the prevalence of household firearms in the United States, our findings suggest that millions of US children are placed at substantially higher risk of fatal firearm injury, especially suicide, than would be the case were parents to follow guidelines first put forward by the AAP more than a quarter century ago.

AAP

American Academy of Pediatrics

ADHD

attention-deficit/hyperactivity disorder

CI

confidence interval

GfK

Growth for Knowledge

KP

KnowledgePanel