Trends in Firearm-Related Injuries in Washington State, 1989–1995

Trends in Firearm-Related Injuries in Washington State, 1989–1995

INJURY PREVENTION/ORIGINAL CONTRIBUTION Trends in Firearm-Related Injuries in Washington State, 1989–1995 From the Harborview Injury Prevention and R...

76KB Sizes 1 Downloads 48 Views

INJURY PREVENTION/ORIGINAL CONTRIBUTION

Trends in Firearm-Related Injuries in Washington State, 1989–1995 From the Harborview Injury Prevention and Research Center, University of Washington, and the Department of Epidemiology, University of Washington School of Public Health and Community Medicine,* Seattle, and the Injury Prevention Program, Washington State Department of Health,‡ Olympia, WA.

Peter Cummings, MD, MPH* Mary LeMier, MPH‡ Douglas B Keck‡

Received for publication June 25, 1997. Revision received December 9, 1997. Accepted for publication December 27, 1997. Supported by a grant from the Centers for Disease Control and Prevention (U17/CCU011040-01), Atlanta, GA. Copyright © 1998 by the American College of Emergency Physicians.

See related editorials, p 77 and p 79. Study objectives: To describe the incidence and outcome of firearm-related injuries and to describe trends over time. Design: Information about firearm-related injuries in Washington state was extracted from computerized death certificate and hospital discharge files for the period 1989 through 1995. Cases seen only in emergency departments were estimated using initial results from a statewide firearm injury surveillance system. Results: An estimated 9,995 persons were shot during the 7year period (incidence 28.0/100,000 person-years); 2,944 persons (29%) required ED care only, 2,864 persons (29%) were hospitalized and survived, and 4,187 persons (42%) died. Survival was lowest for suicide attempts (11%), intermediate for assaults (68%), and greatest for unintentional shootings (96%). Most deaths (91%) occurred before hospital admission. From 1989 to 1995, the incidence of firearm-related injuries showed little change: an increase of 4.5%. However, suicidal shootings among persons younger than 35 years increased by 31.3%, and the incidence of firearmrelated assaults rose by 53.7% among persons of all ages. Survival among all persons who were shot and among hospitalized shooting victims showed little change over time. Conclusion: Firearm-related suicides among the young and assault-related shootings among all persons increased. The estimated proportion of shooting victims who survived did not change over time. [Cummings P, LeMier M, Keck DB: Trends in firearm-related injuries in Washington state, 1989–1995. Ann Emerg Med July 1998;32:37-43.]

JULY 1998

32:1

ANNALS OF EMERGENCY MEDICINE

3 7

FIREARM-RELATED INJURIES IN WASHINGTON STATE Cummings, LeMier & Keck

INTRODUCTION

In 1995, 35,957 persons were killed by firearms in the United States; 18,503 of these deaths were suicides, 15,835 were homicides, 1,225 were unintentional, and 394 were of undetermined intent.1 When injuries are classified by their etiologic mechanism, firearm-related mortality is second only to mortality resulting from traffic crashes as a cause of injury death.2,3 Several reports have described mortality related to firearms, including trends over time.4-8 Studies from Wisconsin and Connecticut have provided information about firearm-related hospitalizations and deaths.9,10 Two studies reported on all firearm injuries in a few cities, including persons who were treated and released from emergency departments,11,12 and similar information has been provided for Massachusetts.13 Firearm-related deaths and nonfatal injuries for the entire United States were estimated for mid-1992 through mid-1993 with data from a probability sample of EDs.14 However, trends over time in all firearm-related injuries have not been published. We estimated all deaths, hospitalizations, and ED visits as a result of firearm-related injuries among Washington state residents from 1989 through 1995. Trends in the incidence of injury, by intent of the shooting, were estimated for the 7-year period. By examining all injuries related to firearms, we were able to determine whether changes in mortality rates were attributable to changes in the survival of people who were shot. We also estimated trends in survival among hospitalized persons and we estimated total hospital days and charges associated with these injuries.

M AT E R I A L S A N D M E T H O D S

Mortality data were obtained from single cause of death files for 1989 through 1995, prepared by the Center for Health Statistics of the Washington State Department of Health. Deaths were identified using external cause of death codes: accident caused by a firearm missile (E922.0 to E922.9), firearm suicide (E955.0 to E955.4), firearm homicide (E965.0 to E965.4), shooting by a law officer (E970), and shooting of uncertain intent (E985.0 to E985.4).15 We also identified homicides and suicides not involving a firearm (E950 to E954, E956 to E958, E960 to E964, E966 to E968). Hospitalization data were obtained from a computerized file maintained by the Office of Hospital and Patient Data Systems of the Department of Health, which contains information on all inpatient discharges from acute care civilian

3 8

hospitals in Washington. E-codes based on the International Classification of Diseases, 9th Revision, Clinical Modification16 have been required in these hospital data since 1989. The completeness of E-coding for all discharges with an injury diagnosis code was 90.7% in 1989 and rose to 98.1% in 1995. The hospital data files contain a code that allows the records of individuals to be linked. During the 7-year interval, some people with gunshot trauma were admitted more than once because of transfers, additional surgeries, and complications and some were probably shot on more than one occasion. Being shot more than once in a 7-year period was probably uncommon compared with rehospitalization for the same gunshot injury. We therefore made the following assumptions: the first hospital discharge with a gunshot E-code was related to the initial gunshot injury, and all subsequent hospital visits with a gunshot E-code were attributed to the initial shooting. A hospitalization was considered to be fatal if there was a code for death at discharge; if a person had more than one hospitalization, we used the discharge status from the last hospitalization. EDs have been required to report firearm-related injuries to the Washington State Department of Health since May 1996. We used data from the first 454 reports of persons who survived a shooting to determine the proportion of patients released from the ED as opposed to being hospitalized; this was .62 for unintentional shootings, .44 for shootings as a result of assault, .21 for suicide attempts with a gun, .25 for persons shot by law enforcement officials, and .41 for shootings of uncertain intent. We used these proportions to estimate the number of persons released from Washington EDs with firearm-related injuries: number released = [proportion/(1–proportion)] × number hospitalized. Because conclusions about trends might be affected by our estimates of ED cases, we performed a sensitivity analysis by recalculating our estimates using information from the National Electronic Injury Surveillance System (NEISS) for June 1992 through May 1995.14 In NEISS, the proportion of gunshot victims discharged alive from an ED varied by category of intent: unintentional, .64; assault, .63; suicide attempt, .22; shot by law enforcement, .41; uncertain intent, .44 (personal communication, September 1996, from J Lee Annest. Data from the CDC Firearm Injury Surveillance Study: National Center for Injury Prevention and Control [K59], CDC, Atlanta, GA 30341). Population data were from the 1980 and 1990 decennial US census and from intercensal and postcensal estimates from the Office of Financial Management of Washington State. Analyses were limited to residents of Washington state. We used negative binomial regression and adjusted for changes in the age, sex distribution, and size of the state’s

ANNALS OF EMERGENCY MEDICINE

32:1 JULY 1998

FIREARM-RELATED INJURIES IN WASHINGTON STATE Cummings, LeMier & Keck

population to estimate linear trends in the incidence of injuries over time.17-19 This regression method is similar to Poisson regression, but allows for the possibility that the variance is greater than is permitted by Poisson models. By using a linear term for year of injury, we estimated the adjusted mean change in injury incidence per year; results were expressed as the overall mean change for the study period with appropriate 95% confidence intervals. Adjustments for age used 12 categories: 0–4, 5–9, 10–14, 15–19, 20–24, 25–34, 35–44, 45–54, 55–64, 65–74, 75–84, and 85 or more years. Within categories of intent, trends were examined for different age groups; to maximize statistical power, we used age groups that contained roughly one third of the gunshot injuries. Logistic regression was used to estimate trends over time in the survival of shooting victims.20 Survival of hospitalized persons was analyzed using Cox proportional hazard methods.21 Hospital charges were adjusted to 1995 dollars using the consumer price index. All analyses were performed using Stata software.22

times more likely to attempt suicide with a gun compared with a woman; age-adjusted rate ratio 6.5 (95% confidence interval [CI], 4.6 to 9.0). After middle age, attempted suicide with a gun was unusual for women, but increasingly common among men (Figure). For assaults, the age-adjusted rate ratio comparing men with women was 3.2 (95% CI, 2.5 to 4.2) and for unintentional shootings the ratio was 4.9 (95% CI, 3.7 to 6.5). For both of these outcomes, the peak incidence was among persons who were 15 to 24 years old; assault-related and unintentional shootings were uncommon after age 50 years. Outcome and Intent of Shootings

Among firearm fatalities, 68% were suicides, 26% were homicides, and 3% were unintentional shootings (Table 1). Among nonfatal hospital admissions, only 10% were suicide attempts, 45% were related to assault, and 36% were unintentional. Of persons discharged alive from EDs, 3% were suicide attempts, 35% were related to assault, and 57% were unintentional. Trends in Incidence Over Time

R E S U LT S

From 1989 to 1995, the population of Washington state increased from 4,728,077 to an estimated 5,429,900. During these 7 years, an estimated 9,995 residents were shot (incidence 28.0/100,000 person-years). The outcomes included 4,187 deaths (42%), 2,864 nonfatal hospitalizations (29%), and 2,944 live discharges (29%) from EDs (Table 1). Only 383 (9.1%) of the deaths occurred after hospital admission. Age and Sex of Victims

Shooting victims were male in 84% of shooting injuries; the rate of injury related to firearms was 47.5 per 100,000 person-years for men and 8.8 for women. A man was six

From 1989 through 1995, the rate of gunshot injury showed a modest increase that was not statistically significant (Table 2). Suicides by gunshot, however, increased 31% among persons younger than 35 years. Among all age groups, firearm injuries resulting from assault increased by 54%; the change was largest, 98%, among persons younger than 25 years. Unintentional gunshot injuries declined somewhat; this decline was not statistically significant; however, it is consistent with national trends. This decline was greatest, 46%, among persons younger that 20 years. Increases in firearm-related suicides and assaults could not be attributed to the substitution of one type of weapon for another. For example, among persons younger than 35

Table 1.

Deaths, nonfatal hospitalizations, and estimated nonfatal ED visits related to firearms among residents of Washington state, 1989–1995, by category of intent. Intent Category Outcome

Suicide No. (Rate)

Assault No. (Rate)

Unintentional No. (Rate)

All No. (Rate)

Death Nonfatal hospital admission Nonfatal ED treatment All

2,866 (8.0) 285 (.8) 74 (.2) 3,225 (9.0)

1,103 (3.1) 1,300 (3.6) 1,020 (2.9) 3,423 (9.6)

112 (.3) 1,018 (2.9) 1,683 (4.7) 2,813 (7.9)

4,187 (11.7) 2,864 (8.0) 2,944 (8.2) 9,995 (28.0)

Rate per 100,000 person-years. Shootings of undetermined intent (N=431) and by law enforcement (N=103) are included in the last column.

JULY 1998

32:1

ANNALS OF EMERGENCY MEDICINE

3 9

FIREARM-RELATED INJURIES IN WASHINGTON STATE Cummings, LeMier & Keck

years, suicides with a gun increased by 37% and suicides by other means increased 12%. Among persons of all ages, homicides with a gun increased 35%, whereas homicides by other methods declined by only 1%. Overall Survival Trends

The proportion of shooting victims who survived was lowest among suicide attempts (11%), intermediate for shootings by law enforcement officials (61%) and as a result of assaults (68%), greater for shootings of unknown intent (85%), and greatest for unintentional shootings (96%). During the 7-year period there was little change in the overall proportion of survival for all shootings. The odds of death, adjusted for sex and age category of the victim, and intent of the shooting, increased by 1% per year, a small change that was not statistically significant (P=.4).

4.5 for uncertain intent (95% CI, 2.8 to 7.2), 2.8 for assaults (95% CI 1.9 to 4.2), and 2.4 for shootings by law officers (95% CI, 0.8 to 6.8). There was little evidence of any linear trend in the proportion of survivors over the 7-year period; the hazard of death in the hospital increased by 2% in each year, a small change that was not statistically significant (P=.6); this estimate was adjusted for categories of intent, age, and sex. We redid our analyses of overall trends and trends in survival, using estimates of ED discharges from the NEISS data, to determine whether results changed substantially. In this analysis, there were 731 more assault-related injuries and the total number of shooting injuries was 10,862. All of the conclusions, however, about trends were essentially the same (details not shown). Hospital Length of Stay and Charges

Survival Trends Among Hospitalized Patients

The likelihood of survival was greater if a person survived long enough to be hospitalized. Hospital survival was strongly related to the intent of the shooting. Survival was least likely among persons who attempted suicide (60%), more common for shootings of uncertain intent (86%), those related to assault (91%), and shootings by law enforcement officials (92%), and highest for unintentional shootings (97%). Compared with unintentional firearm-related injuries, the hazard ratio for death, adjusted for sex and category of age, was 11.5 for suicide attempts (95% CI, 7.8 to 16.8),

Rate/100,000

Figure.

Incidence rates of firearmrelated injuries in Washington state, 1989– 1995, plotted by years of age and category of intent. Dotted line indicates male gunshot victims; solid line indicates female victims.

Hospitalized patients who survived usually had just one admission (85.9%) for their injury, but 10.4% were admitted twice during the 7-year period, and 3.7% had 3 or more admissions (up to 8). We summed total hospital days across all admissions for each survivor; median length of stay was 5 days (range 1 to 217 days). Patients who died in the hospital usually did so quickly; 82% died during the first hospital day and 93% died within the first week. Washington residents were hospitalized for a total of 28,686 days for gunshot trauma during the 1989–1995 period, regardless of survival status. Total hospital charges were

Suicide

Assault

Unintentional

All

200 150 100 50 0

200 150 100 50 0 <5 10 20 30 40 50 60 70 80 90

<5 10 20 30 40 50 60 70 80 90 Age

4 0

ANNALS OF EMERGENCY MEDICINE

32:1 JULY 1998

FIREARM-RELATED INJURIES IN WASHINGTON STATE Cummings, LeMier & Keck

to 1995, our estimates of trends over time could be in error. We have some evidence that any error is unlikely to affect our conclusions. First, mortality was probably ascertained with little change in accuracy during the 7 years, and the trend in mortality closely followed trends in the estimated incidence of all firearm injuries. Second, the lack of any important change in the survival of hospitalized patients, who could be ascertained throughout the 7 years, lends credence to the finding that overall survival from gunshot injuries did not change. Finally, managed care increased in Washington state during the period of this study and this might have influenced EDs to admit fewer patients with gunshot wounds; if this occurred, the severity of injury among hospitalized patients would tend to be greater over time, with a higher proportion of patients dying. But our analysis found only a very minimal increase in mortality among hospitalized patients, suggesting that this trend was unlikely to be very strong. We may have underestimated the total number of firearmrelated injuries. Residents who were shot and killed outside of Washington were included in our totals; 3.4% of the deaths occurred outside the state. However, we did not ascertain injuries to residents who were shot outside the state and survived. Furthermore, the state hospital discharge data do not include military hospitals that treated some per-

$61,733,652, the median hospital charge was $9,011, and the highest charge for a single injury was $464,566.

DISCUSSION

Nearly 10,000 people were shot in Washington state during 1989 through 1995. The proportion of victims who survived was lowest for suicide attempts (11%), greater for assaults (68%), and greatest for unintentional shootings (96%). There was little overall change in the incidence of injuries related to firearms. However, the rate of suicidal shootings among persons younger than 35 years increased by 31%, the incidence of firearm-related assaults rose by 54% among persons of all ages, and unintentional shooting rates declined 46% among persons younger than 20 years. The case-fatality rate among all persons who were shot and among persons hospitalized for gun injuries showed little change over time. We estimated ED visits for firearm-related injuries by assuming that among survivors, the proportion of persons released from EDs, rather than hospitalized, was the same during the 7-year study period as it was in cases from an ED reporting system initiated in Washington in 1996. If, however, this proportion changed substantially from 1989

Table 2.

Estimated number of injuries related to firearms, average incidence rate (per 100,000 person-years), and overall change in rate, by category of intent and age, Washington state, 1989–1995. Average Intent All

Suicide

Assault

Unintentional

Age

No.

Rate

Change (%)*

95% CI for Change Estimate

<25 years 25–34 years 35+ years All ages <35 years 35–54 years 55+ years All ages <25 years 25–34 years 35+ years All ages <20 years 20–34 years 35+ years All ages

3,795 2,514 3,686 9,995 1,261 937 1,027 3,225 1,535 1,000 887 3,423 884 1,240 688 2,813

29.5 42.2 21.8 28.0 6.7 9.4 14.9 9.0 11.9 16.8 5.3 9.6 8.5 14.7 4.1 7.9

+9.1 +24.6 –5.1 +4.5 +31.3 –7.9 –4.6 +6.3 +98.1 +50.3 +21.0 +53.7 –45.7 +14.3 –16.5 –19.7

–6.2 to +26.9 –.3 to +55.7 –16.9 to +8.4 –4.9 to +14.9 +11.2 to +55.0 –24.0 to +11.7 –20.7 to +14.6 –5.0 to +19.1 +39.6 to +181.1 –11.4 to +154.8 –14.3 to +70.7 +22.1 to +93.4 –66.1 to –13.1 –30.6 to +88.4 –44.9 to +26.7 –39.5 to +6.5

P† .1

.01

.2

.1

*Change †P

in rate estimates adjusted for changes in the age and sex distribution of the population. value for a likelihood ratio test that within each category of intent, the changes in rates for the three age categories were statistically different.

JULY 1998

32:1

ANNALS OF EMERGENCY MEDICINE

4 1

FIREARM-RELATED INJURIES IN WASHINGTON STATE Cummings, LeMier & Keck

sons with these injuries. Finally, a few patients with gunshot injuries may not seek care,12 or may obtain care outside EDs. We assumed that readmissions to a hospital did not represent a new shooting episode. This assumption may have caused us to underestimate the number of separate shooting episodes.23-25 However, hospitalization for more than one shooting episode may not be common in Washington. This state has an ED reporting system for firearm injuries: 1,075 persons were reported to this system during the 16 months from May of 1996 to September of 1997, and 517 were hospitalized. Although three persons were shot on two occasions, not one was hospitalized for both episodes. Our estimates of intent-specific case-fatality rates can be compared with those from other recent studies (Table 3). The general pattern is similar across all studies; suicidal bullet wounds are highly lethal, unintentional injuries are only occasionally fatal, and assault-related injuries fall between these extremes. Part of the difference between studies may be related to difficulties in the classification of intent; for example, if we had assigned all shootings of undetermined intent to assaults, the proportion who survived an assault-related shooting would have been .70. In this study, we used hospital discharge data and intentspecific estimates of the proportion of survivors who did not need hospitalization to estimate the number of firearmrelated injuries that only required emergency care. Because Washington now has a reporting system for firearm injuries it may be possible, in the future, to compare estimated ED cases with actual counts. States that have computerized hospital data with E-codes could make similar estimates. Data from NEISS could be used to generate proportions of ED department survivors that are not only intent specific, but are specific for sex and category of age.14 In Washington, 91% of the deaths occurred before hospital admission and 98% occurred within a day of the injury. This is consistent with the report that 97% of firearm deaths in 3 cities occurred within 24 hours of the shoot-

ing.12 In Wisconsin, 90% of gun-related deaths occurred outside a hospital.9 Washington state is covered by an organized trauma care network, and most patients who have a chance of surviving a shooting injury already receive trauma care. We found essentially no evidence of a change in the likelihood of survival among hospitalized victims during the 7-year period. Substantial reductions in shooting-related mortality are likely to occur only if shooting episodes are reduced. REFERENCES 1. Anderson RN, Kochanek KD, Murphy SL: Report of final mortality statistics, 1995. Monthly Vital Statistics Report, vol 45, no 11, suppl 2. Hyattsville, MD: National Center for Health Statistics, 1997:55. 2. Wintemute G: Motor vehicles or firearms; which takes a heavier toll? [letter]. JAMA 1993; 269:2213. 3. Deaths resulting from firearm- and motor-vehicle–related injuries—United States, 1968-1991. MMWR Morbid Mortal Wkly Rpt 1994;43:37-42. 4. Fingerhut LA, Ingram DD, Feldman JJ: Firearm and nonfirearm homicide among persons 15 through 19 years of age: Differences by level of urbanization, United States, 1979 through 1989. JAMA 1992;267:3048-53. 5. Fingerhut LA: Firearm mortality among children, youth, and young adults 1-34 years of age, trends and current status; United States, 1985-90. Advance data from vital and health statistics; no 231. Hyattsville, MD: National Center for Health Statistics, 1993. 6. Firearm-related years of potential life lost before age 65 years—United States, 1980–1991. MMWR Morbid Mortal Wkly Rpt 1994;43:609-611. 7. Trends in rates of homicide—United States, 1985–1994. MMWR Morbid Mortal Wkly Rpt 1996;45:460-464. 8. LeMier M, Keck D, Cummings P: Suicide—Washington, 1980–1995. MMWR Morbid Mortal Wkly Rpt 1997;46:502-506. 9. Hargarten S, Haskins L, Stahlsmith L, et al: Firearm-related deaths and hospitalizations— Wisconsin, 1994. MMWR Morbid Mortal Wkly Rpt 1996;45:757-760. 10. Bretsky PM, Blanc DC, Phelps S, et al: Epidemiology of firearm mortality and injury estimates: State of Connecticut, 1988-1993. Ann Emerg Med 1996;28:176-182. 11. Lee RK, Maxweiler RJ, Dobbins JG, et al: Incidence of firearm injuries in Galveston, Texas, 1979-1981. Am J Epidemiol 1991;134:511-512. 12. Kellermann AL, Rivara FP, Lee RK, et al: Injuries due to firearms in three cities. N Engl J Med 1996;335:1438-1444. 13. Barber CW, Ozonoff VA, Schuster M, et al: When bullets don’t kill. Public Health Rep 1996; 111:482-493. 14. Annest JL, Mercy JA, Gibson DR, et al: National estimates of nonfatal firearm-related injuries: Beyond the tip of the iceberg. JAMA 1995;273:1749-1754. 15. World Health Organization: Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death. Geneva: World Health Organization, 1977.

Table 3.

Proportion of victims who survived, by category of intent, from studies of all firearm-related injuries.

17. McCullagh P, Nelder JA: Generalized linear models. New York: Chapman & Hall, 1989:198-199.

Intent Category Study Population States12

United Memphis, Seattle, Galveston10 Massachusetts11 Washington state (current study)

4 2

16. National Center for Health Statistics: International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). DHHS Publication no. (PHS) 91-1260. Washington DC: National Center for Health Statistics, 1991.

Suicide

Assault

Unintentional

.23 .14 .16 .11

.77 .84 .84 .68

.93 .94 .96 .96

18. Gardner W, Mulvey EP, Shaw EC: Regression analyses of counts and rates: Poisson, overdispersed Poisson, and negative binomial models. Psychol Bull 1995;118:392-404. 19. Glynn RJ, Buring JE: Ways of measuring rates of recurrent events. BMJ 1996;312:364-367. 20. Breslow NE, Day NE: Statistical methods in cancer research, vol I: The analysis of case-control studies. Lyon, France: International Agency for Research on Cancer, 1980:192-242. 21. Breslow NE, Day NE: Statistical methods in cancer research, vol II: The design and analysis of cohort studies. Lyon, France: International Agency for Research on Cancer, 1987:178-229.

ANNALS OF EMERGENCY MEDICINE

32:1 JULY 1998

FIREARM-RELATED INJURIES IN WASHINGTON STATE Cummings, LeMier & Keck

22. StataCorp: Stata statistical software: release 5.0. College Station, TX: Stata Corporation, 1997. 23. Sims DW, Bivins BA, Obeid FN, et al: Urban trauma: A chronic recurrent disease. J Trauma 1989;29:940-947. 24. Morrissey TB, Byrd CR, Deitch EA: The incidence of recurrent penetrating trauma in an urban trauma center. J Trauma 1991;31:1536-1538. 25. Dowd MD, Langley J, Koepsell T, et al: Hospitalizations for injury in New Zealand: Prior injury as a risk factor for assaultive injury. Am J Public Health 1996;86:929-934.

Reprint no. 47/1/90453 Address for reprints: Peter Cummings, MD, MPH Harborview Injury Prevention and Research Center Box 359960 325 Ninth Avenue Seattle, WA 98104-2499 206-521-1549 Fax 206-521-1562 E-mail [email protected]

JULY 1998

32:1

ANNALS OF EMERGENCY MEDICINE

4 3