CLINICAL RESEARCH STUDY
Trends and Burden of Firearm-related Hospitalizations in the United States Across 2001-2011 Shikhar Agarwal, MD, MPH Department of Cardiovascular Medicine, Cleveland Clinic, Ohio.
ABSTRACT BACKGROUND: Firearm-related hospitalizations are a major burden to the current health care infrastructure. We examined the trends in the incidence and case-fatality rates of firearm-related hospitalizations over the past decade. We also hypothesized that major national economic perturbations would be partly responsible and correlate temporally with national firearm-related hospitalization trends. METHODS: We used the 2001-2011 Nationwide Inpatient Sample for analysis. Firearm-related hospitalizations were identified using International Classification of Diseases, 9th Revision codes. In addition, we examined the relationship between the US stock market performance (Dow Jones Industrial Average) and the annual firearm-related hospitalization incidence rates. RESULTS: In the last decade, there has been a modest decline in firearm-related hospitalizations, interrupted by spikes in the annual incidence that closely corresponded to periods of national economic instability. In addition, the overall case-fatality rate following firearm-related hospitalization has been stable at w8%; the highest rates being present among those who attempted suicide using firearms. Also, there has been an increase in the prevalence of mental health disorders among individuals admitted with firearm-related injuries. Moreover, there was an increase in the length of stay and the cost/charges associated with hospitalization over the last decade. CONCLUSION: Over 2001-2011, the national incidence of firearm-related hospitalizations has closely tracked the national stock market performance, suggesting that economic perturbations and resultant insecurities might underlie the perpetuation of firearm-related injuries. Although the case-fatality rates have remained stable, the length of stay and hospitalization costs have increased, imposing additional burden on existing health care resources. Ó 2015 Elsevier Inc. All rights reserved. The American Journal of Medicine (2015) 128, 484-492 KEYWORDS: Case fatality; Cost of illness; Firearms; In-hospital mortality; Length of stay
Firearms are the second leading cause of injury-related deaths after motor vehicle accidents in the US.1 Although the rates of firearm-related injuries have decreased over the last 2 decades, mortality resulting from firearms in the US remains the highest in the world.2 The mass shooting incidents that have occurred recently in Newtown, Funding: None. Conflicts of Interest: None. Authorship: The author is solely responsible for writing this manuscript. Requests for reprints should be addressed to Shikhar Agarwal, MD, MPH, Department of Cardiovascular Medicine, Cleveland Clinic, 9500 Euclid Avenue, Desk J2-3, Cleveland, OH 44195. E-mail address: [email protected] 0002-9343/$ -see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjmed.2014.12.008
Connecticut; Tucson, Arizona; Virginia Tech University; Columbine High School; and at the Washington Navy Yard have brought firearm-related violence to the forefront of national discussion.3 Although these mass shootings killed several people, injured numerous others, and stirred up a major national debate about gun policies, these shootings represent only the tip of the iceberg. Approximately 88 people are believed to die every day due to a direct firearmrelated injury, including suicides, homicides, unintentional injuries, or accidents.3 Homicides and suicides by firearms result in 11,000 and 20,000 deaths, respectively, each year.4 In addition, the number of nonfatal firearm injuries is roughly 40 times higher than the number of fatal firearm injuries.5
Agarwal
Firearm-related Hospitalizations in the United States
485
It appears that firearm-related injuries are a major burden 2001-2009. The number of diagnoses coded in the database on our health care system and consume a large portion of was expanded to 25 for the years 2010-2011. All these have already-constrained health care resources. Besides being a been coded using the standard International Classification of criminal justice issue, firearm-related injuries have become a Diseases, 9th edition, Clinical Modification (ICD-9 CM) major public health challenge facing the nation. Although codes. In addition, we used the HCUP Clinical Classificathere is a large amount of literature detailing the vital station Software (CCS) to identify patient comorbidities and tistics of firearm-related injuries, specific procedures.6,7 CCS has there is a conspicuous paucity of been developed by the AHRQ for CLINICAL SIGNIFICANCE literature exploring the burden clustering patient diagnoses and on health care resources imposed procedures into a manageable Over the last decade, the incidence of by firearm injuries. To that end, number of clinically meaningful firearm-related hospitalizations has we conducted a detailed analysis categories.6,7 Information on closely tracked the national stock market of trends in the incidence rates firearm-related hospitalizations performance, suggesting that economic and in-hospital case-fatality rates was derived using the E codes of perturbations and resultant insecurities of firearm-related hospitalizations the ICD-9 as well as the CCS. All might underlie the perpetuation of over the last decade using a large, firearm-related hospitalizations firearm-related injuries. well-validated nationwide datawere identified using the CCS base. Based on the understanding code 2605, in addition to the Although the case-fatality rates have of the socioeconomic factors that standard ICD-9 codes. Types of remained stable, the length of stay and contribute to the use of firearms in firearm-related hospitalization, as hospitalization costs have increased, the current society, one could defined by the cause, included imposing additional burden on existing surmise that there would be a suicide (E950-E959), assault health care resources. strong relationship between the (E960-E969), and others; innational economic situation and cluding accidents (E922.0-E922.3, national firearm-related hospitaliE922.8, E922.9), legal intervenzation rates. This led us to hypothesize that major national tion (E970), undetermined event (E985.0-E985.3), and economic perturbations partly would be responsible and war (E991). correlate temporally with national firearm-related hospitaliThe entire study duration of 2001-2011 was analyzed zation trends. as 2 distinct intervals: an early period consisting of years 2001-2006, and a later period consisting of years 2007-2011. Demographic variables available for analysis METHODS included age, sex, race (white, black, other), primary source of payment, weekday vs weekend admission, along Data Source with relevant comorbidities including smoking, alcohol Data were obtained from the Nationwide Inpatient Sample use, substance use, and mental health disorders. The his(NIS) database from 2001-2011. The NIS is sponsored by the tory of smoking, alcohol use, substance use, and mental Agency for Healthcare Research and Quality (AHRQ) as a health disorders was reliably coded starting in 2007 in the part of Healthcare Cost and Utilization Project (HCUP). The NIS database and hence, analysis of these variables was number of states that contribute the discharge-level data to restricted to the time period 2007-2011 only. The mental the NIS has grown from 33 in 2001, covering 81% of the health disorders included adjustment disorders (CCS code entire US population, to 44 in 2011, covering over 90% of the 650), anxiety disorders (CCS code 651), attention-deficit, entire US population. Currently, the NIS contains dischargeconduct, and disruptive behavior disorders (CCS code level data from approximately 8 million hospitalizations 652), delirium, dementia, cognitive, and amnestic disorannually from about 1000 hospitals across the US. This ders (CCS code 653), impulse control disorders (CCS database is designed to represent a 20% stratified sample of code 656), mood disorders (CCS code 657), personality all hospitals in the country. Criteria used for stratified samdisorders (CCS code 658), and psychotic disorders (CCS pling of hospitals into the NIS include location (urban or code 659). Hospital characteristics such as region rural), teaching status, geographic region, patient volume, (Northeast, Midwest, South, West), bed size (small, and hospital ownership. All data available from the HCUP medium, large), location (rural, urban), hospital control have been de-identified and hence, the analysis is exempt (government, private), and teaching status were also from the federal regulations for the protection of human included. In addition, NIS has classified the residential zip research participants. The dataset was obtained from the code of each patient into quartiles based on median AHRQ after completing the data use agreement with HCUP. household income of each zip code, which was utilized as a surrogate for socioeconomic status of each firearmStudy Population related hospitalization. Several prior studies have validated this approach for imputing individual socioecoThe NIS database provides up to 15 diagnoses and 15 nomic status in epidemiologic settings.8,9 procedures for each hospitalization record for the years
486
The American Journal of Medicine, Vol 128, No 5, May 2015
Study Outcomes
was estimated using the method of Taylor series (linearization) variance estimation. Continuous variables are presented as mean standard error (SE) and categorical variables are presented as proportion SE. Student’s t test was utilized for comparing means of continuous variables between 2 categories. For comparing the means of continuous variables among 3 or more categories, we utilized oneway analysis of variance. In cases of significant differences detected using analysis of variance, pairwise comparisons were performed utilizing Bonferroni correction for multiple comparisons. The chi-squared test was utilized for comparison of categorical variables. To evaluate for trends across the time periods or consecutive years, a test for linear trend across the years was performed using a linear contrast of coefficients derived using logistic regression analysis or linear regression analysis incorporating the calendar year as an indicator variable with multiple predictor levels.12 Also, multivariable logistic regression analysis was utilized to understand the independent demographic predictors of in-hospital case fatality among those admitted with firearmrelated injuries. All statistical analyses were performed using the statistical software Stata v 13.1 (StataCorp, College Station, TX). All statistical tests were 2-tailed; a P-value < .05 was considered significant.
The incidence of firearm-related hospitalizations, expressed as a proportion of all hospitalizations as well as all injuryrelated hospitalizations each year, was the primary outcome of this study. We also examined the trends in inhospital case fatality, patient disposition after hospitalization, as well as resource utilization across the 2 time periods. The resource utilization was evaluated using the total length of stay as well as the mean cost of hospitalization. The NIS database provides the total charges associated with each hospital stay claimed by the respective hospital for all years. The total charges of each hospital stay were converted to cost estimates using the group average all-payer in-hospital cost and charge information from the detailed reports by hospitals to the Centers of Medicare and Medicaid Services. All costs and charges were converted to projected estimates for the year 2011, after accounting for annual inflation rates based on consumer price index data available from the Bureau of Labor Statistics.10 Based on the understanding of the socioeconomic factors that contribute to the use of firearms in current society, we hypothesized that there would be a strong relationship between the national economic situation and national firearmrelated hospitalization rates. To test this hypothesis, we qualitatively examined the relationship between the US stock market performance (assessed using the Dow Jones Industrial Average) and the annual national firearm-related hospitalization rates.11
RESULTS Trends in Firearm-related Hospitalizations From a total of 87,195,470 hospitalizations recorded in the NIS during 2001-2011, there were 11,221,897 injury-related hospitalizations designated using E codes. Of these, we identified 70,974 hospitalizations that were related to firearmrelated injuries. Table 1 demonstrates the annual national incidence of firearm-related hospitalizations across the entire study duration. Compared with the year 2001, there was a modest increase in the annual incidence of firearm-related hospitalizations noted during 2002 and 2004, followed by a
Statistical Analysis Survey statistics traditionally used to analyze complex semirandom survey designs were employed to analyze these data. Data obtained from NIS were pooled using standard methods. Subsequently, appropriate NIS discharge and hospital weights were utilized to weight the entire sample to account for the clustered structure of the data according to the NIS stratum as well as the treating hospital. Variance
Table 1
Firearm-related Hospitalizations in the US
Year
Total Number
Injury-related Hospitalizations
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
7,452,727 7,853,982 7,977,728 8,004,571 7,995,048 8,074,825 8,043,415 8,158,381 7,810,762 7,800,441 8,023,590
776,139 856,807 901,686 943,640 935,656 977,988 1,003,639 1,197,503 1,136,493 1,215,018 1,277,328
CI ¼ confidence interval.
Firearm-related Hospitalizations
Incidence (95% CI) of Firearmrelated Hospitalizations (Per 100,000 Hospitalizations)
Incidence (95% CI) of Firearm-related Hospitalizations (Per 100,000 Injury-related Hospitalizations)
4769 6943 6328 7159 6579 6599 6647 5841 5977 7968 6164
65.9 88.4 79.9 89.2 81.4 82.3 83.2 70.6 75.6 105.2 77.4
630.4 806.5 704.5 756.7 695.9 677.6 663.3 480.4 519.8 673.4 486.7
(52.3-83.0) (67.4-115.8) (65.4-97.6) (73.5-108.2) (64.0-103.6) (64.5-105.0) (65.2-106.2) (57.3-87.0) (61.2-93.3) (85.2-129.9) (62.2-95.8)
(505.7-785.6) (620.6-1047.5) (582.0-852.7) (628.1-911.4) (551.4-878.0) (536.4-855.6) (524.4-838.7) (391.9-588.7) (423.9-637.4) (548.8-826.1) (399.1-593.4)
Agarwal
Firearm-related Hospitalizations in the United States
slow decline in the annual incidence until 2008. The years 2009 and 2010 demonstrated a sharp increase in the annual incidence rate of firearm-related hospitalizations, followed by a sharp decline in annual rates in 2011 (Figure 1).
Baseline Characteristics of Firearm-related Hospitalizations Table 2 demonstrates the baseline characteristics of all firearm-related hospitalizations stratified according to the time period. Compared with the early period, there was a small but statistically significant increase in the age of patients admitted with firearm injuries (P < .001). Although the proportion of blacks admitted with firearm-related injuries was significantly higher than whites or other races, there were no significant changes in relative proportions across the study duration. A majority of the patients admitted with firearm-related injuries were either uninsured or had Medicaid as primary payor. Compared with the early time period, there was a small increase in the proportion of patients with Medicare and Medicaid, with a corresponding decrease in the proportion of patients with private insurance in the later period. Stratifying the entire population according to socioeconomic quartiles also demonstrated no significant changes over the 2 time periods. Most of the firearm-related hospitalizations were from the lowest socioeconomic quartiles. Stratifying the entire population of all firearm-related hospitalizations by the type of injury demonstrated that a majority of patients were admitted with firearm-related assaults in both time periods (Table 2). Notably, there was a significant increase in the proportion of all firearm-related hospitalizations due to suicide, from 8.8% in 2001-2006 to 10.3% in 2007-2011 (P ¼ .01). Table 3 demonstrates the differences in the baseline characteristics of the firearmrelated hospitalizations during the most recent time period spanning 2007-2011. Although there was a significant
Figure 1 Annual national incidence rate of firearm-related hospitalizations, expressed per 100,000 hospitalizations. The bars represent the incidence rate and the lines above each bar represent the standard error of the effect estimate.
487 Table 2 Characteristics of Firearm-related Hospitalizations Stratified by the Time Period of Admission Characteristics
2001-2006 2007-2011 P-Value
N Mean (SE) age, years Females Weekend admission Race White Black Others Type of injury Suicide Assault Others (accident, unintentional, legal intervention, undetermined) Primary payer Medicare Medicaid Private insurance Uninsured Other Socioeconomic status quartiles Quartile 1 Quartile 2 Quartile 3 Quartile 4
38,369 29.8 (0.2) 10.7 (0.3) 36.1 (0.3)
32,597 30.7 (0.2)
Trends and Burden of Firearm-related Hospitalizations in the United States Across 2001-2011 Shikhar Agarwal, MD, MPH Department of Cardiovascular Medicine, Cleveland Clinic, Ohio.
ABSTRACT BACKGROUND: Firearm-related hospitalizations are a major burden to the current health care infrastructure. We examined the trends in the incidence and case-fatality rates of firearm-related hospitalizations over the past decade. We also hypothesized that major national economic perturbations would be partly responsible and correlate temporally with national firearm-related hospitalization trends. METHODS: We used the 2001-2011 Nationwide Inpatient Sample for analysis. Firearm-related hospitalizations were identified using International Classification of Diseases, 9th Revision codes. In addition, we examined the relationship between the US stock market performance (Dow Jones Industrial Average) and the annual firearm-related hospitalization incidence rates. RESULTS: In the last decade, there has been a modest decline in firearm-related hospitalizations, interrupted by spikes in the annual incidence that closely corresponded to periods of national economic instability. In addition, the overall case-fatality rate following firearm-related hospitalization has been stable at w8%; the highest rates being present among those who attempted suicide using firearms. Also, there has been an increase in the prevalence of mental health disorders among individuals admitted with firearm-related injuries. Moreover, there was an increase in the length of stay and the cost/charges associated with hospitalization over the last decade. CONCLUSION: Over 2001-2011, the national incidence of firearm-related hospitalizations has closely tracked the national stock market performance, suggesting that economic perturbations and resultant insecurities might underlie the perpetuation of firearm-related injuries. Although the case-fatality rates have remained stable, the length of stay and hospitalization costs have increased, imposing additional burden on existing health care resources. Ó 2015 Elsevier Inc. All rights reserved. The American Journal of Medicine (2015) 128, 484-492 KEYWORDS: Case fatality; Cost of illness; Firearms; In-hospital mortality; Length of stay
Firearms are the second leading cause of injury-related deaths after motor vehicle accidents in the US.1 Although the rates of firearm-related injuries have decreased over the last 2 decades, mortality resulting from firearms in the US remains the highest in the world.2 The mass shooting incidents that have occurred recently in Newtown, Funding: None. Conflicts of Interest: None. Authorship: The author is solely responsible for writing this manuscript. Requests for reprints should be addressed to Shikhar Agarwal, MD, MPH, Department of Cardiovascular Medicine, Cleveland Clinic, 9500 Euclid Avenue, Desk J2-3, Cleveland, OH 44195. E-mail address: [email protected] 0002-9343/$ -see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjmed.2014.12.008
Connecticut; Tucson, Arizona; Virginia Tech University; Columbine High School; and at the Washington Navy Yard have brought firearm-related violence to the forefront of national discussion.3 Although these mass shootings killed several people, injured numerous others, and stirred up a major national debate about gun policies, these shootings represent only the tip of the iceberg. Approximately 88 people are believed to die every day due to a direct firearmrelated injury, including suicides, homicides, unintentional injuries, or accidents.3 Homicides and suicides by firearms result in 11,000 and 20,000 deaths, respectively, each year.4 In addition, the number of nonfatal firearm injuries is roughly 40 times higher than the number of fatal firearm injuries.5
Agarwal
Firearm-related Hospitalizations in the United States
485
It appears that firearm-related injuries are a major burden 2001-2009. The number of diagnoses coded in the database on our health care system and consume a large portion of was expanded to 25 for the years 2010-2011. All these have already-constrained health care resources. Besides being a been coded using the standard International Classification of criminal justice issue, firearm-related injuries have become a Diseases, 9th edition, Clinical Modification (ICD-9 CM) major public health challenge facing the nation. Although codes. In addition, we used the HCUP Clinical Classificathere is a large amount of literature detailing the vital station Software (CCS) to identify patient comorbidities and tistics of firearm-related injuries, specific procedures.6,7 CCS has there is a conspicuous paucity of been developed by the AHRQ for CLINICAL SIGNIFICANCE literature exploring the burden clustering patient diagnoses and on health care resources imposed procedures into a manageable Over the last decade, the incidence of by firearm injuries. To that end, number of clinically meaningful firearm-related hospitalizations has we conducted a detailed analysis categories.6,7 Information on closely tracked the national stock market of trends in the incidence rates firearm-related hospitalizations performance, suggesting that economic and in-hospital case-fatality rates was derived using the E codes of perturbations and resultant insecurities of firearm-related hospitalizations the ICD-9 as well as the CCS. All might underlie the perpetuation of over the last decade using a large, firearm-related hospitalizations firearm-related injuries. well-validated nationwide datawere identified using the CCS base. Based on the understanding code 2605, in addition to the Although the case-fatality rates have of the socioeconomic factors that standard ICD-9 codes. Types of remained stable, the length of stay and contribute to the use of firearms in firearm-related hospitalization, as hospitalization costs have increased, the current society, one could defined by the cause, included imposing additional burden on existing surmise that there would be a suicide (E950-E959), assault health care resources. strong relationship between the (E960-E969), and others; innational economic situation and cluding accidents (E922.0-E922.3, national firearm-related hospitaliE922.8, E922.9), legal intervenzation rates. This led us to hypothesize that major national tion (E970), undetermined event (E985.0-E985.3), and economic perturbations partly would be responsible and war (E991). correlate temporally with national firearm-related hospitaliThe entire study duration of 2001-2011 was analyzed zation trends. as 2 distinct intervals: an early period consisting of years 2001-2006, and a later period consisting of years 2007-2011. Demographic variables available for analysis METHODS included age, sex, race (white, black, other), primary source of payment, weekday vs weekend admission, along Data Source with relevant comorbidities including smoking, alcohol Data were obtained from the Nationwide Inpatient Sample use, substance use, and mental health disorders. The his(NIS) database from 2001-2011. The NIS is sponsored by the tory of smoking, alcohol use, substance use, and mental Agency for Healthcare Research and Quality (AHRQ) as a health disorders was reliably coded starting in 2007 in the part of Healthcare Cost and Utilization Project (HCUP). The NIS database and hence, analysis of these variables was number of states that contribute the discharge-level data to restricted to the time period 2007-2011 only. The mental the NIS has grown from 33 in 2001, covering 81% of the health disorders included adjustment disorders (CCS code entire US population, to 44 in 2011, covering over 90% of the 650), anxiety disorders (CCS code 651), attention-deficit, entire US population. Currently, the NIS contains dischargeconduct, and disruptive behavior disorders (CCS code level data from approximately 8 million hospitalizations 652), delirium, dementia, cognitive, and amnestic disorannually from about 1000 hospitals across the US. This ders (CCS code 653), impulse control disorders (CCS database is designed to represent a 20% stratified sample of code 656), mood disorders (CCS code 657), personality all hospitals in the country. Criteria used for stratified samdisorders (CCS code 658), and psychotic disorders (CCS pling of hospitals into the NIS include location (urban or code 659). Hospital characteristics such as region rural), teaching status, geographic region, patient volume, (Northeast, Midwest, South, West), bed size (small, and hospital ownership. All data available from the HCUP medium, large), location (rural, urban), hospital control have been de-identified and hence, the analysis is exempt (government, private), and teaching status were also from the federal regulations for the protection of human included. In addition, NIS has classified the residential zip research participants. The dataset was obtained from the code of each patient into quartiles based on median AHRQ after completing the data use agreement with HCUP. household income of each zip code, which was utilized as a surrogate for socioeconomic status of each firearmStudy Population related hospitalization. Several prior studies have validated this approach for imputing individual socioecoThe NIS database provides up to 15 diagnoses and 15 nomic status in epidemiologic settings.8,9 procedures for each hospitalization record for the years
486
The American Journal of Medicine, Vol 128, No 5, May 2015
Study Outcomes
was estimated using the method of Taylor series (linearization) variance estimation. Continuous variables are presented as mean standard error (SE) and categorical variables are presented as proportion SE. Student’s t test was utilized for comparing means of continuous variables between 2 categories. For comparing the means of continuous variables among 3 or more categories, we utilized oneway analysis of variance. In cases of significant differences detected using analysis of variance, pairwise comparisons were performed utilizing Bonferroni correction for multiple comparisons. The chi-squared test was utilized for comparison of categorical variables. To evaluate for trends across the time periods or consecutive years, a test for linear trend across the years was performed using a linear contrast of coefficients derived using logistic regression analysis or linear regression analysis incorporating the calendar year as an indicator variable with multiple predictor levels.12 Also, multivariable logistic regression analysis was utilized to understand the independent demographic predictors of in-hospital case fatality among those admitted with firearmrelated injuries. All statistical analyses were performed using the statistical software Stata v 13.1 (StataCorp, College Station, TX). All statistical tests were 2-tailed; a P-value < .05 was considered significant.
The incidence of firearm-related hospitalizations, expressed as a proportion of all hospitalizations as well as all injuryrelated hospitalizations each year, was the primary outcome of this study. We also examined the trends in inhospital case fatality, patient disposition after hospitalization, as well as resource utilization across the 2 time periods. The resource utilization was evaluated using the total length of stay as well as the mean cost of hospitalization. The NIS database provides the total charges associated with each hospital stay claimed by the respective hospital for all years. The total charges of each hospital stay were converted to cost estimates using the group average all-payer in-hospital cost and charge information from the detailed reports by hospitals to the Centers of Medicare and Medicaid Services. All costs and charges were converted to projected estimates for the year 2011, after accounting for annual inflation rates based on consumer price index data available from the Bureau of Labor Statistics.10 Based on the understanding of the socioeconomic factors that contribute to the use of firearms in current society, we hypothesized that there would be a strong relationship between the national economic situation and national firearmrelated hospitalization rates. To test this hypothesis, we qualitatively examined the relationship between the US stock market performance (assessed using the Dow Jones Industrial Average) and the annual national firearm-related hospitalization rates.11
RESULTS Trends in Firearm-related Hospitalizations From a total of 87,195,470 hospitalizations recorded in the NIS during 2001-2011, there were 11,221,897 injury-related hospitalizations designated using E codes. Of these, we identified 70,974 hospitalizations that were related to firearmrelated injuries. Table 1 demonstrates the annual national incidence of firearm-related hospitalizations across the entire study duration. Compared with the year 2001, there was a modest increase in the annual incidence of firearm-related hospitalizations noted during 2002 and 2004, followed by a
Statistical Analysis Survey statistics traditionally used to analyze complex semirandom survey designs were employed to analyze these data. Data obtained from NIS were pooled using standard methods. Subsequently, appropriate NIS discharge and hospital weights were utilized to weight the entire sample to account for the clustered structure of the data according to the NIS stratum as well as the treating hospital. Variance
Table 1
Firearm-related Hospitalizations in the US
Year
Total Number
Injury-related Hospitalizations
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
7,452,727 7,853,982 7,977,728 8,004,571 7,995,048 8,074,825 8,043,415 8,158,381 7,810,762 7,800,441 8,023,590
776,139 856,807 901,686 943,640 935,656 977,988 1,003,639 1,197,503 1,136,493 1,215,018 1,277,328
CI ¼ confidence interval.
Firearm-related Hospitalizations
Incidence (95% CI) of Firearmrelated Hospitalizations (Per 100,000 Hospitalizations)
Incidence (95% CI) of Firearm-related Hospitalizations (Per 100,000 Injury-related Hospitalizations)
4769 6943 6328 7159 6579 6599 6647 5841 5977 7968 6164
65.9 88.4 79.9 89.2 81.4 82.3 83.2 70.6 75.6 105.2 77.4
630.4 806.5 704.5 756.7 695.9 677.6 663.3 480.4 519.8 673.4 486.7
(52.3-83.0) (67.4-115.8) (65.4-97.6) (73.5-108.2) (64.0-103.6) (64.5-105.0) (65.2-106.2) (57.3-87.0) (61.2-93.3) (85.2-129.9) (62.2-95.8)
(505.7-785.6) (620.6-1047.5) (582.0-852.7) (628.1-911.4) (551.4-878.0) (536.4-855.6) (524.4-838.7) (391.9-588.7) (423.9-637.4) (548.8-826.1) (399.1-593.4)
Agarwal
Firearm-related Hospitalizations in the United States
slow decline in the annual incidence until 2008. The years 2009 and 2010 demonstrated a sharp increase in the annual incidence rate of firearm-related hospitalizations, followed by a sharp decline in annual rates in 2011 (Figure 1).
Baseline Characteristics of Firearm-related Hospitalizations Table 2 demonstrates the baseline characteristics of all firearm-related hospitalizations stratified according to the time period. Compared with the early period, there was a small but statistically significant increase in the age of patients admitted with firearm injuries (P < .001). Although the proportion of blacks admitted with firearm-related injuries was significantly higher than whites or other races, there were no significant changes in relative proportions across the study duration. A majority of the patients admitted with firearm-related injuries were either uninsured or had Medicaid as primary payor. Compared with the early time period, there was a small increase in the proportion of patients with Medicare and Medicaid, with a corresponding decrease in the proportion of patients with private insurance in the later period. Stratifying the entire population according to socioeconomic quartiles also demonstrated no significant changes over the 2 time periods. Most of the firearm-related hospitalizations were from the lowest socioeconomic quartiles. Stratifying the entire population of all firearm-related hospitalizations by the type of injury demonstrated that a majority of patients were admitted with firearm-related assaults in both time periods (Table 2). Notably, there was a significant increase in the proportion of all firearm-related hospitalizations due to suicide, from 8.8% in 2001-2006 to 10.3% in 2007-2011 (P ¼ .01). Table 3 demonstrates the differences in the baseline characteristics of the firearmrelated hospitalizations during the most recent time period spanning 2007-2011. Although there was a significant
Figure 1 Annual national incidence rate of firearm-related hospitalizations, expressed per 100,000 hospitalizations. The bars represent the incidence rate and the lines above each bar represent the standard error of the effect estimate.
487 Table 2 Characteristics of Firearm-related Hospitalizations Stratified by the Time Period of Admission Characteristics
2001-2006 2007-2011 P-Value
N Mean (SE) age, years Females Weekend admission Race White Black Others Type of injury Suicide Assault Others (accident, unintentional, legal intervention, undetermined) Primary payer Medicare Medicaid Private insurance Uninsured Other Socioeconomic status quartiles Quartile 1 Quartile 2 Quartile 3 Quartile 4
38,369 29.8 (0.2) 10.7 (0.3) 36.1 (0.3)
32,597 30.7 (0.2)
