Mental Health Outcomes in Children after Parental Firearm Injury

Abstract

Background

Every year, approximately 20,000 youth lose a parent to firearm injury. Many more have parents who sustain nonfatal firearm injuries. The effect of parental firearm injuries on their children’s health and health care is poorly understood.

Methods

Using 2007–2022 commercial health insurance claims data, we identified youth, aged 1–19 years, whose parents received treatment for firearm injury (exposure). Each exposed youth was matched to up to five control youths based on year, month, youth sex, metropolitan statistical area, state, insurance plan type, and prescription drug coverage, while balancing the means of age and a risk score predicting future health care use (a proxy for health status). Our primary outcome was youth psychiatric diagnosis rates, defined as the number of youths with at least one related insurance claim in a given month divided by total youths. Additional outcomes included substance use disorder diagnosis rates, health care use, and spending. After matching, we estimated the difference-in-differences in outcomes between the exposed and control groups 12 months before through 12 months after parental injury using a least squares regression model adjusted for age and risk score.

Results

We examined 3,790 exposed youth and 18,535 matched controls. Mean age was 10.7 years, and 51.5% were male. Parental firearm injury was associated with 8.4 (95% CI: 4.8–12.0) additional psychiatric diagnoses and 23.1 (95% CI: 8.2–38.1) additional mental health visits per 1000 youth averaged over the year. This associated increase was largest for trauma/posttraumatic stress disorder diagnoses, with an additional 8.5 (95% CI: 6.0–10.9) diagnoses per 1000 youth averaged over the year. There were no apparent changes in other diagnoses, medical encounters, procedures, services, or spending.

Conclusions

Parental firearm injury was associated with increases in psychiatric disorders and mental health visits among their children.

Introduction

Firearm injuries are the leading cause of death among children and adolescents (aged 1–19), with 4,470 killed in 2023 alone.^1,2 Direct injury, however, is only one way that firearm violence affects individuals, families, and communities. A recent study³ estimated that between 1999 and 2020, 434,000 youth (aged 0–18) lost a parent to firearm injury–13 times the number of children killed by firearm injuries over this period.² Because nonfatal firearm injuries are thought to be 2–3 times more frequent than deaths, these deaths capture only a fraction of the harm caused by firearm violence.⁴

Youth firearm injury survivors face increased psychiatric, substance use, and pain-related morbidity with accompanying increases in health care utilization and spending.^5–10 Firearm injury is followed by increases in psychiatric diagnoses and mental health care use in other family members,¹¹ especially among parents of firearm-injured children.⁹ Two recent single-city studies found increased prevalence of mental health diagnoses—stress, anxiety, and mood disorders in particular—of family members of fatally¹² and nonfatally¹³ firearm-injured individuals, with the greatest increases among the children of injured adults. Other studies have investigated the relationship between self-reported psychiatric symptoms and exposure to family and friend homicide, but have been limited by cross-sectional study designs, limited geographic coverage, or relatively small sample sizes, and most tend to focus on fatal injuries.^14–18

Studies using methods that enhance a causal understanding of the effects of parental firearm injury on their children are lacking. We conducted a matched cohort analysis of children and parents with commercial health insurance using a difference-in-differences design to estimate associations between parental firearm injury and the health, health care use, and medical spending of their children.

Methods

Data

We analyzed the 2007–2022 Merative MarketScan commercial claims databases. These comprise a nationwide convenience sample of enrollees with employer-sponsored insurance with >40 million individuals covered annually.¹⁹

Study population

Included individuals were youth (aged 1–19) who were the children of adults who received treatment for incident firearm injury between January 2008 and January 2022 and comparable youth without such exposure. We excluded youth who received treatment for firearm injuries themselves. We identified firearm injuries using International Classification of Diseases (ICD) 9^th and 10^th revision codes (Methods S1-3). We identified children as the non-spouse dependents sharing a plan number. For each youth, we required a minimum enrollment of 12 months before and after parental firearm injury, defined as active health insurance coverage by an employer-sponsored plan reporting data to MarketScan. We also required a minimum of 300 days of youth enrollment in the calendar year preceding parental injury in order to calculate the Diagnostic Cost Related Group (DxCG) risk score, which served as a proxy for youth baseline health status. This score reflects expected future health care spending by calendar year based on current age, sex, and diagnoses.²⁰

We did not require a minimum enrollment duration for parents; therefore, they may have disenrolled post-injury. Because it is not possible to ascertain death as the cause of disenrollment, this cohort may include exposure to fatal in addition to nonfatal parental firearm injuries. We calculated the proportion of firearm-injured parents enrolled for at least 3 months following firearm injury, suggesting that they survived initial injury. Parents who disenrolled earlier may have done so for reasons other than a fatal injury, including losing health insurance, changing insurance plans, separation from a spouse providing them with insurance, or incarceration.

Matching

We matched each exposed youth to up to 5 unexposed control youths. We used risk-set matching without replacement, with exact matching on year, month, youth sex, metropolitan statistical area, state, insurance plan type, and prescription drug coverage.²¹ We balanced the means of two continuous variables, youth age and risk score, using a mathematical optimization approach to maximize the number of matches and reduce the covariate distance between matched pairs.^22–24 Controls received matching weights of 1/N where N was the number of controls matched to an exposed youth (e.g., if 5 controls were matched to a specific exposed youth, each control received a weight of 1/5). To assess match quality, we calculated absolute standardized mean differences between the exposed and control cohorts before and after matching.

Outcomes

Our primary outcome was the rate of psychiatric diagnoses per 1000 youth by month, and averaged over the year. We examined these in aggregate and by subcategory (anxiety, mood, neuropsychiatric, psychotic, trauma/posttraumatic stress disorder [PTSD], other). We examined attention deficit hyperactivity disorder (ADHD)-related diagnoses separately given the high baseline prevalence of ADHD. We examined substance use disorder rates in aggregate and by substance-specific subcategories (Methods S4). Other outcomes included frequency of mental health visits, general medical visits, emergency department visits, hospitalizations, imaging tests, lab tests, other services, medical spending, and prescription drug use. We analyzed the rate of chronic pain disorders as a negative control (i.e., an outcome we would not expect to change following exposure).²⁵

Diagnoses were identified using ICD codes (Methods S1). Diagnosis rates were calculated as the number of youth with at least one related diagnosis on a claim in a given month divided by total youths, and reported per 1000 youth. If in a single month a youth had diagnoses in multiple subcategories (e.g., PTSD and anxiety disorders), these counted as one at the aggregate category level (psychiatric disorders) but counted individually in their subcategories. Therefore, the sum of the component rates could exceed the aggregate rate.

We used procedure code fields to classify outpatient visits as mental health or general office visits and to identify emergency department visits, hospitalizations, tests, and other services (Methods S1). The method for calculating medical spending and prescription drug use are reported in the Supplement (Methods S1).

Statistical analysis

We examined youth psychiatric diagnoses 12 months before and after parental firearm injury. We calculated the unadjusted psychiatric diagnosis rate for each month pre- and post-injury. We then calculated the change in psychiatric diagnoses as the difference between the pre- and post-injury means (i.e., the difference between the mean of monthly rates in the 12 months before and the mean of monthly rates 12 months after exposure) within the exposed and control groups. To calculate the unadjusted differential change associated with parental firearm injury, we subtracted the difference in means for the control group from the difference in means for the exposed group. We used least squares regression to adjust the differential change for residual differences in age and risk score, which were not exactly matched (Methods S5). The same method was used to estimate the change in other diagnostic categories, healthcare utilization, prescription drug use, and medical spending.

We also calculated the differential change in the proportion of youth who had at least one claim with a psychiatric diagnosis at any point in the 12 months before and the 12 months after parental firearm injury, again adjusting for age and risk score. The same method was used to estimate the change in proportions in other diagnostic categories.

Total missingness of any matching variable was <5%. We conducted a complete case analysis with the assumption that data missingness was not associated with the outcome variables after controlling for the observed variables (see Methods S6 for additional details on missing data).²⁶ The low rate of missingness attenuates potential biases if this assumption does not hold.

Confidence intervals were not adjusted for multiple comparisons and should not be used in place of hypothesis testing. Analyses were conducted in R version 4.4.0 and Stata version 18.0. The Harvard Medical School Institutional Review Board approved this research.

Subgroup and sensitivity analyses

We conducted subgroup analyses stratifying by injury severity, defined as the parent receiving ICU care during their index hospitalization following injury, as higher injury severity is associated with worse health outcomes and higher spending.^9,11 We also conducted subgroup analyses stratified by youth sex, given sex differences in behavioral responses to stress, frequency of pediatric psychiatric conditions, and likelihood of diagnosis.^27,28

We repeated our analyses using non-parametric permutational t-tests to assess the sensitivity of our estimates to the parametric assumptions of linear regression. We also used Rosenbaum bounds to assess sensitivity to unmeasured confounding (Methods S7).²⁹

Results

Cohort characteristics

4,053 youth met inclusion criteria (Figure S1). 202 (4.98%) were excluded due to missing data for one or more matching variables and 61 (1.5%) were excluded due to an inability to find at least one match. After matching, the cohort comprised 3,790 exposed youth and 18,535 controls (Table 1). Mean youth age was 10.7 years and 51.5% were male. Characteristics were closely balanced between exposed and control groups (Table S1). The average age of the firearm-injured parents was 40.6 years, 84.1% were male, and 96.7% were enrolled in their insurance plan for at least 3 months following injury, suggesting that they survived the initial injury.

Table 1.

Characteristics of youth exposed to parental firearm injury and matched controls

Characteristics	Exposed	Control
Individuals, n	3,790	18,535
Characteristics
Age, y (SD)	10.9 (5.2)	10.6 (5.4)
Risk Score*	0.35 (1.95)	0.32 (1.30)
Male, %	51.5	51.5
Drug Coverage, %	86.7	86.7
Plan type, %
EPO	0.8	0.8
HMO	14.7	14.7
PPO	58.8	58.8
Basic/Comprehensive	3.1	3.1
POS	5.4	5.4
CDHP/HDHP	17.2	17.2

EPO: Exclusive Provider Organization; HMO: Health Maintenance Organization; PPO: Preferred Provider Organization; CDHP: Consumer Directed Health Plan; HDHP: High Deductible Health Plan.

^*The Diagnostic Cost Related Group (DCxG) risk score is measure of health status and expected spending with higher scores indicating worse health and greater expected spending.

Diagnoses and prescriptions

Unadjusted analysis demonstrated comparable rates of diagnoses pre-injury, with trends that did not clearly deviate between the exposed and unexposed groups. After injury, psychiatric diagnoses demonstrated a sharp and persistent increase in the exposed cohort (Figure 1), most pronounced in the trauma/PTSD subcategory (Figure 2).

Figure 1. Youth diagnoses before and after parental firearm injury.

Unadjusted diagnosis rates are displayed by month for exposed youth along with their matched controls. The vertical line indicates the month immediately prior to parental firearm injury.

Figure 2. Youth psychiatric diagnoses by clinical category before and after parental firearm injury.

Unadjusted diagnosis rates are displayed by month for exposed youth along with their matched controls. The vertical line indicates the month immediately prior to parental firearm injury.

In adjusted analysis (Table 2), parental firearm injury was associated with a differential increase of 8.4 (95% CI: 4.8–12.0) psychiatric diagnoses per 1000 youth averaged over the year. There was a differential increase per 1000 youth over the year of 8.5 (95% CI: 6.0–10.9) for trauma/PTSD diagnoses and 2.2 (95% CI: 0.5–3.9) for mood disorder diagnoses. There were no apparent differential changes in ADHD-related diagnoses, other psychiatric diagnosis subcategories, substance use diagnoses, or in drug prescriptions. There was no apparent difference in chronic pain diagnoses, which was included as a negative control.

Table 2.

Changes in diagnoses, health care use, prescriptions, and medical spending in youth exposed to parental firearm injury

Study Outcome	Exposed Group		Control Group		Differential change between groups
	12 Months Preinjury	12 Months Postinjury	12 Months Preinjury	12 Months Postinjury	Adjusted	(95% CI)	Percentage Change	(95% CI)
Diagnoses, per 1000 youth	(n = 3,790)		(n = 18,535)
Psychiatric	20.0	33.0	22.9	27.6	8.4	(4.8, 12.0)	42.0	(23.8, 60.1)
Anxiety	6.3	9.3	8.4	11.1	0.3	(−1.3, 2.0)	5.1	(−21.4, 31.5)
Mood	5.0	8.8	7.3	8.9	2.2	(0.5, 3.9)	43.5	(9.2, 77.8)
Neuropsychiatric	0.5	0.4	0.6	0.5	−0.1	(−0.5, 0.3)	−17.9	(−96.2, 60.3)
Psychotic	0.2	0.4	0.3	0.3	0.2	(−0.1,0.5)	84.0	(−57.1, 225.1)
Trauma/PTSD	4.7	13.9	5.1	5.8	8.5	(6.0, 10.9)	178.3	(126.4, 230.3)
Other	6.7	7.8	6.4	7.5	0.0	(−1.7, 1.7)	−0.4	(−25.9, 25.1)
ADHD	11.7	14.1	10.3	11.7	1.0	(−1.2, 3.2)	8.7	(−9.8, 27.2)
Substance use	0.7	1.2	0.7	1.1	0.1	(−0.4, 0.7)	18.3	(−59.8, 96.5)
Chronic pain	17.3	18.2	16.0	16.6	0.2	(−1.9, 2.4)	1.4	(−11.1, 13.9)
Prescription drugs, psychiatric, per youth
Number of prescriptions	0.1	0.1	0.1	0.1	0.0	(0.0, 0.0)	4.6	(−3.9, 13.0)
Days covered	1.6	2.0	1.8	2.1	0.0	(−0.1, 0.2)	2.6	(−6.5, 11.8)
Healthcare utilization, per 1000 youth
Mental health visits	38.3	69.7	56.1	64.3	23.1	(8.2, 38.1)	60.4	(21.4, 99.3)
General office visits	244.9	241.8	253.6	243.7	6.7	(−7.2, 20.5)	2.7	(−2.9, 8.4)
Emergency department visits	28.7	29.7	18.1	18.1	1.1	(−2.7, 4.8)	3.7	(−9.4, 16.9)
Hospitalizations	1.4	1.9	1.6	1.4	0.7	(0.0, 1.3)	47.4	(2.6, 92.2)
Medical spending, US dollars, per youth
Total	170.0	206.0	177.0	192.0	20.0	(−27.0, 67.0)	11.9	(−16.0, 40.0)
Out of pocket	30.0	29.0	28.0	31.0	−3.0	(−6.0, -0.0)	−10.7	(−20.0, -1.0)

Unadjusted differential change was calculated as the difference between the pre- and post-injury means of the exposed group (youth exposed to parental firearm injury) minus the difference between the pre- and post-injury means of the control group (youth with no exposure to parental firearm injury). Means were calculated by averaging monthly rates over the 12 months before and after parental firearm injury. The differential change was then adjusted for age and risk score, which were not exactly matched.

Confidence intervals were not adjusted for multiple comparisons and should not be used in place of hypothesis testing. If in a single month a youth had diagnoses in multiple diagnostic subcategories (e.g., PTSD and anxiety disorders), these counted as one at the aggregate category level (psychiatric disorders) but counted individually in their subcategories. Therefore, the sum of the component rates could exceed the aggregate rate.

Adjusted analysis of the change in the proportion of youth with at least one psychiatric diagnosis claim any time in the 12 months before compared with any time 12 months after parental firearm injury demonstrated a differential increase per 1000 youth of 13.5 (95% CI: 0.1–26.9) for any psychiatric diagnosis and 23.1 (95% CI: 15.4–30.9) for any trauma/PTSD diagnosis (Table S6).

Health care use and spending

Parental firearm injury was associated with a differential increase of 23.1 (95% CI: 8.2–38.1) mental health visits per 1000 youth (Table 2). There were no apparent differential changes in spending (Table 2), other types of medical encounters, or other procedures or services (Table S3).

Subgroup analyses

There were 380 youth exposed to severe parental firearm injury, defined by ICU care after injury. Severe injury was associated with larger differences (Table 3; Figure S2 and S3) with differential increases per 1000 youth over the year of 28.2 (95% CI: 9.8–46.7) in aggregate psychiatric diagnoses, 23.1 (95% CI: 9.8–36.3) in trauma/PTSD diagnoses, and 80.1 (95% CI: 25.3–134.8) in mental health visits.

Table 3.

Changes in diagnoses, health care use, prescriptions, and medical spending in youth exposed to severe parental firearm injury

Study Outcome	Exposed Group		Control Group		Differential change between groups
	12 Months Preinjury	12 Months Postinjury	12 Months Preinjury	12 Months Postinjury	Adjusted	(95% CI)	Percentage Change	(95% CI)
Diagnoses, per 1000 youth	(n = 380)		(n = 1,850)
Psychiatric	24.1	55.7	29.5	32.9	28.2	(9.8, 46.7)	117	(40.6, 193.4)
Anxiety	10.3	17.3	9.6	14.6	2.0	(−7.1, 11.1)	19.3	(−69.0, 107.6)
Mood	7.2	11.8	11.2	10.5	5.4	(−0.4, 11.1)	74.1	(−5.7, 154.0)
Neuropsychiatric	0.4	0.7	0.3	0.3	0.3	(−0.4, 1.0)	66.7	(−89.0, 222.3)
Psychotic	0.2	0.7	0.3	0.3	0.4	(−0.4, 1.2)	180	(−169.6, 529.6)
Trauma/PTSD	4.8	28.5	7.7	8.3	23.1	(9.8, 36.3)	478.2	(202.9, 753.4)
Other	6.8	8.3	10.1	11	0.6	(−5.5, 6.6)	8.2	(−81.4, 97.7)
ADHD	13.4	19.1	12.6	12.6	5.7	(−1.4, 12.9)	43	(−10.5, 96.4)
Substance use	0.2	1.5	0.5	1	0.9	(−1.9, 3.7)	400	(−869.0, 1669.0)
Chronic pain	18.6	17.5	15.7	14.2	0.4	(−5.9, 6.7)	2.4	(−31.4, 36.2)
Prescription drugs, psychiatric, per youth
Number of prescriptions	0.1	0.1	0.1	0.1	0.0	(0.0, 0.0)	13.1	(−4.3, 30.5)
Days covered	3.4	4.1	2.2	2.5	0.5	(−0.2, 1.2)	14.2	(−6.6, 35.0)
Healthcare utilization, per 1000 youth
Mental health visits	32.2	115.6	67.2	70.5	80.1	(25.3, 134.8)	248.4	(78.5, 418.2)
General office visits	251.5	251.8	258.1	237	21.3	(−9.5, 52.2)	8.5	(−3.8, 20.7)
Emergency department visits	34.9	31.8	16.4	15.1	−1.8	(−11.8, 8.2)	−5.1	(−33.8, 23.6)
Hospitalizations	2.2	2.4	1.8	1.8	0.2	(−1.8, 2.3)	10	(−83.3, 103.3)
Medical spending, US dollars, per youth
Total	162	221	163	162	61	(−53.0, 175.0)	37.8	(−33.0, 108.0)
Out of pocket	30	30	29	29	0.0	(−8.0, 7.0)	−0.8	(−26.0, 25.0)

Unadjusted differential change was calculated as the difference between the pre- and post-injury means of the exposed group minus the difference between the pre- and post-injury means of the exposed group. Means were calculated by averaging monthly rates over the 12 months before and after parental firearm injury. The differential change was then adjusted for differences by age and risk score, which were not exactly matched. Confidence intervals were not adjusted for multiple comparisons and should not be used in place of hypothesis testing. If in a single month a youth had diagnoses in multiple diagnostic subcategories (e.g., PTSD and anxiety disorders), these counted as one at the aggregate category level (psychiatric disorders) but counted individually in their subcategories. Therefore, the sum of the component rates could exceed the aggregate rate.

We found larger associated differences on female youth (Tables S4 and S5, Figures S4) with differential increases over the year in psychiatric diagnoses of 13.1 (95% CI: 6.9–19.4) per 1000 female youth and 3.9 (95% CI: −1.0–8.8) per 1000 male youth.

Subgroup analyses of the differential changes in proportions of youth with at least one diagnosis before and after parental firearm are presented in Tables S7-9.

Sensitivity analyses

Non-parametric estimation with permutational T-tests produced very similar results (Table S10). Sensitivity analysis with Rosenbaum bounds showed that after matching an unmeasured confounder—differing across both exposure and time—would need to increase odds of exposure to parental firearm injury in the main analysis by 1.27 for aggregate psychiatric diagnoses and 2.19 for trauma/PTSD diagnoses for the confidence interval of these estimates to include the null (see Methods S7 for additional details). This is consistent with results that are insensitive to small-to-moderate unmeasured biases, but sensitive to large biases.

Discussion

A parent’s firearm injury was associated with substantial increases in psychiatric disorders and mental health visits in their children, compared with closely matched youth who were not exposed to parental firearm injury. These changes were larger among children whose parents suffered more severe injuries and among female youth.

Severe parental firearm injury was associated with greater increases in mental health diagnoses for youth. Given that ICU care for the parent is an observable marker of injury severity, it can help identify these particularly vulnerable children for assistance. Severe injuries are associated with longer hospital stays and provide the opportunity for more extensive interactions with clinical and social service providers, including hospital-based violence intervention programs (HVIPS), both during hospitalization and post-discharge.³⁰ Critically, however, mental health consequences are not limited to the children of severely injured parents, yet children of less severely injured parents may receive less support services because they are less readily identified and engaged.

The amplified association of parental firearm injury on female children is also notable. In general, girls’ and women’s experience with firearm violence is underappreciated because boys and men comprise more than 85% of firearm deaths annually.² Yet, the reverberations of parental firearm injury are at least as likely to affect female children as male children, as evidenced by the even sex distribution in our study, and caregiving responsibilities following firearm injury fall disproportionately on the female relatives of male survivors.³¹ The sex differences we observe should, however, be interpreted in light of reported sex differences in the underlying prevalence of mental health conditions, behavioral responses to stress, care seeking, clinician diagnostic biases and tendencies, and the distribution of specific psychiatric diagnoses that may lead to underdiagnosis among male children.^27,28

While we did not observe apparent changes in youth substance use disorder diagnoses in the year following parental firearm injury, prior research has found associations between substance use and personal and indirect exposure to firearm violence.^9,11,32 The absence of such a finding may be related to the young age in our cohort as substance use disorders are less common in younger children, and may develop more slowly and take longer to come to medical attention, if they do at all, than our observation period allowed us to detect.

For youth with severely injured parents, the increase in psychiatric disorders we found (28.2 per 1000) exceeded prior estimates—from data using identical outcome definitions and similar methods—of the effect of children’s firearm injuries on their own mental health (18.8 per 1000) though it was below the increase observed following severe firearm injury to the child themselves (99.4 per 1000).⁹ While cross-cohort comparisons should be made with caution, this finding is consistent with a prior study that found that PTSD prevalence among children following parental trauma surpassed the prevalence following injury to children themselves.³³

Taken together, our findings emphasize the substantial within-family spillover effects of firearm injury. Our findings extend research that has linked negative outcomes to a variety of youth exposures to firearm violence, including direct injury and community level exposures.^5–10,34 Recognition of the many ways that firearm violence can affect youth has motivated calls to formally classify firearm and community violence as a specific adverse childhood experience.^35,36 Such experiences, especially when severe, co-occurring, and persistent, have strong associations with changes in brain development and long-term effects on physical and mental health that persist into adulthood.^{37,32,38–40}

Our findings highlight the importance of robust supportive care for children following parental firearm injury and deliberate care coordination between teams taking care of firearm-injured adults and those caring for their children. Prior research on parental homicide death has linked worsening mental health in their children not only to the intense grief accompanying the loss of a parent but also to a loss of confidence in society, a sudden sense of vulnerability and insecurity, fear of additional violence to other family members or to themselves, worsening mental health in the surviving parent, and strained family dynamics.¹⁶ Studies of both survivors of firearm injury and their affected relatives, however, suggest that mental health services are difficult to access and underutilized.^41–44

Informing firearm injury survivors of the possible mental health consequences for their children and the benefit of intervention may facilitate broader recognition of such challenges and could potentially facilitate earlier access to care. Clinicians can also seek parental permission to coordinate with pediatricians for their children to monitor for emerging mental health needs. Hospital-based violence intervention programs can assist in coordinating care for injured adults and their affected children.⁴⁵ Teams caring for adult survivors can refer their children to supportive programs designed to serve youth survivors of firearm injury to leverage existing resources. They can also refer children to community-based programs that support survivors and their families.

Limitations

Our study has limitations. First, despite careful matching with closely balanced exposed and control groups, there may be less balanced unobserved variables, including socioeconomic status. The balanced trends in pre-exposure outcomes among the exposed and control cohorts and the visual break following injury in the exposed cohort alone, however, support our interpretation.

Second, we analyzed families with employer-sponsored health insurance from large employers, which may not generalize to other populations, particularly Medicaid, which is the other major source of insurance for children and more commonly represented among individuals injured by firearms. Third, we are not able to definitively identify fatal parental firearm injury, which could be associated with larger effects on the mental health of children. That more than 95% of the injured parents were still enrolled 3 months post-injury suggests that they survived their initial injury. Our findings, therefore, likely underestimate the effects on children of parents with fatal firearm injuries.

Fourth, our study excluded youth who were not continuously enrolled for 12 months following parental firearm injury. Firearm injury is associated with insurance disruption in survivors and their dependents, with greater effects seen following severe firearm injury.⁴⁶ Excluding those who experienced such a disruption may have also biased our estimates towards the null. Fifth, we did not analyze effects by firearm injury intent (i.e., unintentional, assault, self-harm) because ICD codes lack precision in distinguishing among these.⁴⁷

Fifth, our primary outcome was dependent on recorded diagnoses on claims and was therefore susceptible to factors related to care-seeking, screening, and diagnosis. Medical diagnoses may underestimate the full mental health impact of parental firearm injury on their children and therefore bias our results towards the null. This may be especially true in contexts where mental healthcare is especially difficult to access, such as in rural and inner-city areas, or when such care is not insurance-billable such as care provided by school counselors. On the other hand, it is possible that increased screening because of familial or clinician concern following parental firearm injury could have led to increased ascertainment of psychiatric disorders and related mental health visits, elevating the apparent impact.

Finally, confidence intervals were not adjusted for multiple comparisons and should not be used in place of hypothesis testing.

Conclusion

Parental firearm injury was associated with increases in psychiatric diagnoses and mental health visits in their children, driven by trauma/PTSD-related diagnoses. Severe injuries in parents were associated with larger mental health effects on children and female children experienced larger effects than male children.

(Funded by the National Institute for Health Care Management, and others.)