Household Firearm Ownership Levels and Homicide Rates Across U.S. Regions and States, 1988-1997

Rates of Household Firearin Owiiership and HoinicideAcross US Regions and States, 1988-1997I Matthew Miiier, MD, MPH, ScD, Deborah Azraei, MS, PhD, and David iHemenway, PhD

The United States has higher rates of firearmownership than do other developed nations,and higher rates of homicide.''^ Of the233 251 people who were homicide victimsin the United States between 1988 and1997, 68% were killed with guns, of whichthe large majorify were handguns."*

Case-control studies suggest that the pres-ence of a gun in the home is a risk factor forhomicide in the home,^ that fhe purchase of ahandgun from a licensed dealer is associatedwith becoming a homicide victim," and thatgim ownership may be a risk factor for com-mitting homicide^ (although other studiesfound no association with homicide perpetra-tion^). Most, but not all,"''* cross-sectionalstudies have found a positive association be-tween various measures of firearm availabilityand overall rates of homicide, a trend thatholds across regions," states,' "'"* cities,'^'^and coimties,'^

Nationally representative studies of the ef-fect of fircami prevalence on rates of homi-cide have been hampered by the lack of di-rect measures of firearm ownership withinareas smaller than the 9 US Census regionsand hy uncertain^ regarding the validity offirearm ownership proxies. Our study extendsprevious work by using recent data, lookingacross both regions and all 50 states, disag-gregating victims by age, and adjusting for po-tential confounders, including poverty, urban-ization, unemployment, alcohoi consumption,aggravated assault, forcible rape, and robbery.In addition, we used the 2 best currentlyavailable measures of rates of householdfirearm ownership—direct survey-based mea-sures for regional analyses^" and a rigorouslyvalidated proxy of household gun owner-ship''' for region- and state-level analyses,

METHODS

We used pooled cross-sectional time-seriesdata for tbe 9 census regions and the 50

Objectives, in this study we explored the association between rates of househoidfirearm ownership and homicide across the United States, by age groups.

Methods. We used cross-sectional time-series data (1988-1997) to estimate the as-sociation between rates of househoid firearm ownership and homicide.

Results, In region- and state-ievei anaiyses, a robust association between rates ofhousehoid firearm ownership and hoinicide was found. Regionaiiy, the association ex-ists for victims aged 5 to 14 years and those 35 years and oider. At the state ievei,theassociation exists for every age group over age 5, even after controiiing for poverty, ur-banization, unempioyment, aicohoi consumption, and nonlethai vioient crime.

Conclusions. Aithough our study cannot determine causation, we found that in areaswhere househoid firearm ownership rates were higher, a disproportionately large num-ber of peopie died from homicide. (Am J Public Heaith. 2002;92:1988-1993)

states for the period 1988 to 1997 to exam-ine the association between rates of house-hold firearm ownership and rates of overallhomicide, gun-related homicide, and non-gun-related homidde. State-, age-, and year-specific mortality data came from the Na-tional Center for Health Statistics (NCHS)moriaiity files. " Gun-related deaths of imde-termined intention constitute less than 3% ofall gun-related deaths and were excludedfrom analyses. Region-specific population andmoriaiity data were derived by aggregatingstate-level data.

At the regional level, survey-based mea-sures of the percentage of households withguns and handguns came from the GeneralSocial Survey (GSS),' The GSS, conductedby the National Opinion Research Center(Chicago, IU) in most years from 1972 to1993 and biennially since 1994, is the goldstandard for national surveys of gun owner-ship. In its current form, the GSS is con-ducted by personal interview with a nationalarea probability sample of 3000 noninstitu-tionalized adults. The sample is chosen to berepresentative of each of the 9 census regionsand of the nation as a whole, but not of indi-vidual states. At the state level, publisheddata on reported hoijsehold gun ownershiparc available lor only a iionrandom sample of21 states,' ^ To analyze all 50 states, we useda proxy for household fireann ownership: the

fraction of all suicides in a state that involve afirearm, referred to hereafter as FS/S.

FS/S, which measures the distribution offirearm vs nonfirearm methods used in sui-cide rather than the rate of suicide, has beenvalidated against survey-based measures ofhousehold gun ownership. A recent study de-termined that FS/S is the best proxy forhousehold firearm ownership rates of thehalf-dozen or more proxies that have beenLLsed in the literature, ^ FS/S is highly corre-lated with the percentage of households re-porting fireann ownership in studies of 16 de-veloped nations (r=0,91),^^ the 9 US censusregions (r=0.93),^^ 21 US states (r=0.90),^^170 US cities (r-().86),^'* and 12 areaswithin a single state (r=0,87),"'

Regressions allowed each region and stateto assume a distinct firearm ownership ratefor each of the 10 years in the study. Becausean area's homicide rate in a given year is de-pendent on its rate in other years, standardcrmrs were comxled hy clustering observa-tions by region or state. Distributions of deathrates were skewed, and variances weregreater than means. Consequently, negativebinomial rather than Poisson models wereused.

Primary analyses use incidence rate ratios(lRRs). obtained by exponentiating p coeffi-cients in the negative binomial regressions, toexpress the association between firearm own-

1988 I Research and Practice I Peer Reviewed I Miller et ai. American Journal of Public Heaith I December 2002, Vol 92, No, 12

TABLE 1-Crude Incidence Rate Ratios of Regional Homicide in the United States by

Region-Level Proxies of Firearm Prevalence, 1988-1997

Victim Age

0-4 years

FS/S

GSSANYGUN

GSS HANDGUN

5-14 years

FS/S

GSSANYGUN

GSS HANDGUN

15-24 years

FS/S

GSSANVGLJN

GSS HANDGUN

25-34 years

FS/S

GSSANYGUN

GSS HANDGUN

35-44 years

FS/S

GSSANYGUN

GSS HANDGUN

45-54 years

FS/S

GSS ANY GUN

GSS HANDGUN

55-64 years

FS/S

GSSANYGUN

GSS HANDGUN

> 65 years

FS/S

GSSANYGUN

GSS HANDGUN

All ages

FS/S

GSSANYGUN

GSS HANDGUN

Homicide

(95% CI)

1.11(0.98,1.24)*

1,08 (0.98,1.20)

1.07 (0.97,1.18)

1,15 (1,02,1,29)**

1,11 (0.98,1.26)

1.14(1.02,1.26)"

1.10(0.90,1,34)

1.04(0,87,1,24)

1,10(0,92,1,31)

1,24 (1.00,1.53)*

1,16(0,96,1,40)

1,21 (1.01,1,47)**

1.35(1.09,1.66)"'

1.26(1.05,1.53)"

1.31(1.09,1.58)"*

1,32 (1.10,1.59)"'

1.25(1.05,1.49)"

1.30(1.10,1.54)"*

1.35 (1.19,1.53)t

1,27(1.11,1.45)"*

1.32(1.17,1.50)1

1,38 (1.20,1.59)t

1.31(1.13,1.51)1

1.35(1.18,1.56)1

1.23(1.02,1.49)**

1.16 (0,98,1,38)*

1.21(1,02,1,44)**

Gun-Related Homicide

(95% CI)

1.22 (1.10,1.35)t

1.24 (1.13,1.36)t

1.20 (1.07,1,34)***

1.21(1.03,1.42)**

1.15 (0.94,1.39)

1.20 (1,02,1,41)**

1,11(0,89,1,39)

1.04 (0.85,1.27)

1.11 (0.91,1.35)

1.26 (0.99,1,61)*

1.18 (0.95,1,46)

1.24(1.00,1.54)**

1.44(1,13,1,84)***

1,34(1,07,1.67)**

1.39(1.13,1.73)***

1.43 (1.18,1.73)t

1,35(1.12,1.63)***

1,40(1,17,1,67)1

1,48 (1.29, l,70)t

1,39(1.16,1.63)1

1.44(1.26,1.65)1

1,63(1,44,1,85)1

1.51(1,28,1,80)1

1,60(1.36,1.88)1

1.27(1.02,1.58)**

1.19(0.97,1.45)*

1.25(1.03,1.52)**

Nor-gun-Related Homicide

(95% CI)

1.09(0.97,1.24)

1,06(0.96,1.18)

1,06(0.95,1.17)

1,05(0.98,1.12)

1,05(0.97,1.11)

1,04(0.98,1.11)

1.07 (0.96,1.20)

1,02(0,93,1.12)

1,07(0.96,1.20)

1,17(1.01,1.37)*

1,11(0,98,126)*

1,15(1.01,1.32)**

1.19(1.02,1.39)1

1.14(0.99,1.30)*

1,17(1,02,1,34)**

1.16(0.97,1.38)*

1.10(0.95,1.28)

1.14 (0.98,1.33)*

1.21(1.07,1.36)***

1.14(1.02,1.27)**

1,19(1,06,1,33)***

1.26(1.09,1.45)***

1.21(1.05,1.38)**

1.23(1,07,1,40)*'*

1,16(1,01,1.33)**

1.11(0.98,1.25)*

1.14(1,01,1.29)**

Note. 01 - confidence interval. Regional household firearm ownership prevalence was estimated using 3 measures: (1) a proxyrepresenting the percentage of sjicides that are firearm suicides (FS/S), (2) household gm ownership rates as reported inthe General Social Survey (GSS ANY GUN), and (3) household handgm ownership rates as reported in the General SocialSurvey (GSS HANDGUN), All measures are standardized at the regional level so that their mean equals 0 and their standarddeviation equals 1. Incidence rate ratios correspond to tfie standardized proxies. Over the 10-year stjdy period, on average,our proxies vaiy across regions by 3.1,3.4, and 3.4 standard delations for fS/S, GSS ANY GUN, and GSS HANDGUN,respectively. Longitudinal variation is an order of magnitude smaller: over time, FS/S, GSS ANY GUN, and GSS HANDGUNmeasures fall within 0.35,0.33, and 0,34 standard deviations of one another. Overall, when each region assumes a distinctprevalence estimate for eacli year, proxies span 3,6,4,5, and 4.7 standard deviations for FS/S, GSS ANY GUN, and GSSHANDGUN, respectively.*P<O.l: * *P<0.05: * * * P < 0 , 0 1 ; tP<0 .001 . Ail P values are two tailed.

ership rates and rates of overall homicide,gun-related homicide, and non—gun-relatedhomicide. To make our measures of firearmownership easier to compare with one an-other, we standardized all measures to have amean of 0 and a standard deviation of 1, Be-cause the standard deviation of each firearmownership measure was equal to 1, the re-ported IRRs represent the percentage changein the homicide rate for each standard devia-tion change in firearm ownership rate.

State-level analyses controlled for charac-teristics linked to homicide in the literature,including rates of violent crimes other thanhomicide (i.e., the Federal Bureau of Investi-gation "index crimes" of aggravated assault,forcible rape, and rohbery), percentage of theadult population who are unemployed, per-centage of the population who live in poverly(as defined by the poverty index developedby the Social Security Administration in 1961and revised by the Federal Interagency Com-mittees in 1980, wilh thresholds updatedyearly to reflect changes in the ConsimierPrice Index), urbanization (percentage of thepopulation living in metropolitan areas), andper capita alcohol consumption. " ' Alcoholconsumption data came from the National In-stitute on Alcohol Abuse and Alcoholism,''and data for other control variables camefrom the Statistical Abstract of the United

States.^' Because of the small number of ob-servations, regional analyses did not includecontrol variables.

RESULTS

At the regional level, we found a positiveand statistically significant relationship be-tween rates of household gun ownership andhomicide victimization for the entire popula-tion, for victims aged 5 to 14 years, and forvictims 35 years and older (Table 1). Theseresults were attributed primarily to highergun-related homicide rates in regions withhigher rates of firearm ownership; non-gun-related homicide rates were also elevated inregions where there were more guns, but to alesser extent. Homicide victimization rates forthose aged 0 to 4 years and aged 15 to 34years were higher in regions with higher ratesof gun ownership, but the association did notreach statistical significance. Results obtained

December 2002, Vol 92, No, 12 I American Journal of Public Health Milter et al. | Peer Reviewed | Research and Practice | 1989

^SEARCH AND PRACTICE

using survey (GSS) and proxy (FS/S) mea-sures of firearm prevalence were nearly iden-tical. Rates of household handgun ownershipwere somewhat more likely to be significantlyassociated with homicide rates than weremeasures of ownership of all householdfirearms.

At the state level, multivariate resultsshowed a positive and significant relationshipbetween rates of household gun ownership

and homicide victimization, for the entirepopulation and for each age gi oup aged 5years and older (Table 2). As in the regionalanalyses, state-level results were attributedprincipally to substantially elevated gun-re-lated homicide rates in states with higherrates of firearm ownership, although corre-sponding non-gun-related homicide rateswere also somewhat elevated. The associa-tion between household gun ownership and

TABLE 2-Crude and Multivariate Adjusted Incidence Rate Ratios of State-Level Homicide

by State-Level Measures of Firearm Prevalence, 1988-1997

Victim Age

0-4 years

FS/S (bivariate)

FS/S (muitivariate)

5-14 years

FS/S (bivariate)

FS/S (riiuitivaiiate)

15-24 years

FS/S (bivariate)

FS/S (muitivariate)

25-34 years

FS/S (bivariate)

FS/S (multivariate)

35-44 years

FS/S (bivariate)

FS/S (muitivariate)

45-54 years

FS/S (bivariate)

FS/S (mjltivariate)

55-64 years

FS/S (bivariate)

FS/S (multivariate)

> 65 years

FS/S (bivariate)

FS/S (multivariate)

Plages

FS/S (bivariate)

FS/S (muitivariate)

Homicide

(95% Ci)

1,09 (1,02,1.17)"

1,07(0.98,1,17)

1.13 (1,03,1.24)*'*

1,15(1,03,1,28)**

1,10(0.94,1,27)

1,23 (1.12, l,34)t

1.23 (1,08,1.41)**

1,27(1,10,1,49)**

1.32 (1.17,1.49)t

1.35 (1,23, l,48)t

1.32 (1.16,1.49)t

1,33 (1,19, l,48)t

1,34 (1.21, l,49)t

1,36 (1,24, l,50)t

1.36 (1.22,1.52)t

1.39 (1.27,1.53)t

1.22(1,08,1,37)***

1.27(1.16,1.39)1

Gun-Related Homicide

(95% Ci)

1.26 (1.13, l,41)t

1,35 (1,18,1.53)t

1,20 (1,07,1.36)"*

1,23(1,07,1.41)***

1,12 (0,93,1,34)

1,32(1,18,1.47)1

1,28(1.09,1.52)"

1,39 (1,23,1.57)t

1.45(1.24,1.71)1

1,52(1,35,1,77)1

1,45(1.25,1.68)1

1,48(1,30,1,68)1

1.51(1,32,1.73)1

1.54 (1,37, l,72)t

1.64 (1,46,1.84)t

1.71 (1.53,1.91)t

1.28(1,10,1,50)***

1.41 (1.27,1.57)t

Noi-gui-Related Homicide

(95% Ci)

1.07 (1,00,1.15)*

1,05(0,96,1,15)

1,04(0.97,1,11)

1,05(0,97,1,13)

1,03(0,94,1,11)

1,02 (0,95,1,10)

1.13(1.03,1,25)**

1.10(1.02,1,19)**

1.15 (1.04,1,26)"*

1,15(1,06,1.24)1

1.14 {1.02,1,27)"

1,12(1,02,1,24)**

1.17 (1.06,1.29)***

1,18(1,07,1,38)*'*

1,24 (1,11,1.38)t

1,24 (1,13.1.37)t

1,11(1,03,1,21)**

1.10 (1.02,1.19)'*

Wore, Ci - confidence inten/ai. Adjusted araiyses controi for rates of vioient index crimes other tliari homidde (aggravatedassault, forcible rape, arid robbery), percentage unemployed of the adult population, percentage of tiie population living inpoverty, percentage of the population living in metropoiitan areas, and per capita aicohoi consumption. Household firearmownership prevalence was estimated for aii 50 states using the percentage of suicides that are firearm suicides (FS/S) as aproxy, FS/S is standardized so that its mean equals 0 and its standard deviation equals 1, incidence rate ratios correspondto standardized values for FS/S. Over the 10-year study period, on average, FS/S spans 4.1 standard deviations across the50 states (FS/S ranges from 0,29 to 0,80); within-state variation is far smaller, ranging, on average, across 0,28 standarddeviation (FS/S ranges from 0.58 to 0,61), Overali, when each state assumes a distinct value for each year, FS/S spans5,1 standard deviations (FS/S ranges from 0.21 to 0,85),*P<0,\; •*P<0.05; • ' •P<0 .01 ; tP<0,001. Ail Pvaiues aretwo-taiied.

homicide victimization was strongest for vic-tims 25 years and older.

Firearm ownership rates varied far moreacross states and regions (i.e., over time) thanwithin states and regions. Not surprisingly,nearly identical results were obtained whenfirearm ownership rates were calculated using(1) distinct values for each state- (2) region-year or the average GSS or FS/S value foreach geographic area over the 10-year studyperiod, or (3) when 5-year rolling averageswere used (not shown). Regressions acrossstates in any given year yielded point esti-mates that were within 8"/(i of the point esti-mate obtained when all 10 years of data wereanalyzed (not shown).

Table 3 compares the actual numher ol'homicide victims between 1988 and 1997 inthe states with the 4 lowest and 6 highestfirearm ownership rates, ITiese 10 states werechosen on the basis of their extreme firearmownership rates, not on the basis of theirhomicide rates. The number of states in eachgroup was selected so that the 2 groups' totalperson-years over the 10-year period wereapproximately the same (158 million vs 160million).

In the "high gun states," 21 148 individualswere homidde victims, compared with 7266in the "low gun states" (Table 3), For everyage group of at least 5 years minimum age,people living in the high-gun states weremore than 2,5 times more likely than thoseiii the low-gun states to become homiddevictims, 'ITiese results were largely driven hyhigher rates of gun-related homicide, al-though rates of non-gun-related homiddewere also somewhat higher- in high-gunstates. For all age groups, people living inhigh-gun states were 2,9 times more likely todie in a homicide; they were 4.2 times morelikely to die in a gun-related homicide and 1,6times more likely to die in a non-gun-relatedhomidde.

State firearm ownership rates in Table 3were determined using our proxy, FS/S, forall 50 states. Direct measures of firearm own-ership rates are available fi"Oin the BehavioralRisk Factor Surveillance System for 3 of the 6high-gun states, in which aji average of 53%of households contain firearms (range; 51 "/o-55^k>). The system provides direct estimatesof fireann ownership for 2 of the 4 low-gun

1990 i Research and Practice | Peer Reviewed 1 M7/er et ai. American Journal of Public Heaith I December 2002, Vol 92, No, 12

TABLE 3-Homici(le Deaths in States With the Highest vs the Lowest Average GunOwnership Prevalence Index, 1988-1997

Victim Age High Gun States Low Gun StatesMortality Rate Ratio(High Gun:Low Gun)

Total population, all ages

0-4 years

Giin-related homicide

Noi-gm-related homicide

Total

5-14 years

Gji-related homicide

Non-gun-related homicide

Total

15-24 years

Gun-related homicide

Non-gji-related homicide

Total

25-34 years

Gun-related homicide

Nor-gji-related homicide

Total

35-44 years

Gun-related homicide

Non-gun-relatad homicide

Total

45-54 years

Gun-related homicide

Noi-gun-related homicide

Total

55-64 years

Gui-related homicide

Non-gin-related homicide

Total

> 65 years

Gun-related iiomicide

Nor-gun-related homicide

Total

All ages

Gur-related homicide

M on-gun-related homicide

Total

158 million

67

437

504

302

149

451

5157

963

6120

4397

1445

5842

2825

1168

3993

1316

544

1860

609

402

1011

602

745

1347

15283

5865

21148

160 million

17

293

310

80

104

184

1539

697

2236

1078

920

1998

495

684

1179

264

331

595

106

216

322

80

331

411

3668

3598

7266

4.0

1.5

1.6

3.8

1.5

2.5

3.4

1.4

2.8

4.1

1.6

3.0

5.8

1,7

3.4

5,0

1,7

3.2

5.8

1.9

3.2

7,6

2.3

3,3

4.2

1.6

2,9

tote. For ease of comparison, similar popjiations were obtained by comparing the 4 states with the lowest gun ownershiprates ("low gjn states") and the 6 states with the highest gm ownership rates ("high gin states"). The 6 states with thehighest average gjn ownership rates for 1988 to 1997 were Louisiana, Aiabama, Mississippi, Wyoming, West Virginia, andArkansas,The 4 states with the lowest average gun ownership rates for 1988 to 1997 were Hawaii, Massachusetts, RhodeIsland, and New Jersey. Overall mortality rate ratios and ratios for each age stratum use strata-specific populations asdenominators.

states, in which 13% of households containfirearms (range: 12' /o-14%). The correspon-ding FS/S measures for the 6 high-gun states

and the 4 low-gun states are 76% (range:750/0-80%) and 33% (range:respectively.

Introducing a lagged firearm ownershipmeasure in regressions did not alter our find-ings. Homicide rates for the period 1988 to1997 were associated with firearm ownershiprates regardless of whether ownership datacame from contemporaneous years or fromthe preceding decade (i.e., the averagefirearm ownership rates between 1979 and1987).

Overall homicide rates were significantlyhigher in states with higher rates of nonlethalviolent crime, poverty, and urbanization. Percapita alcohol consumption and tmemploy-ment were not significantly associated withhomicide rates in multivariate regressions,

DISCUSSION

In the United States, regions and stateswith higher rates of firearm ownership havesignificantly higher homicide victimizationrates. This result is driven primarily by gun-related homicide victimization rates, althoughnon—gun-related victimization rates were alsohigher in states with higher rates of firearmownership. The close correspondence be-tween our proxy [FS/S) and survey-based(GSS) measures of household firearm owner-ship is readOy apparent in Table 1, in whichresults obtained with survey and proxy mea-sures are nearly identical.

The association between higher householdgun ownership rates and higher overall homi-cide rates is robust. Regressions were drivenneither by either the most populous states norby the states with the most extreme rates ofgun ownership. Overall, the results obtainedwhen we analyzed all 50 states and the 40least and 40 most populous states wereequivalent to those obtained when analysesexcluded the 10 states most extreme in FS/S(i.e., the 5 states with the highest FS/S andthe 5 states with the lowest FS/S). Thefirearm—homicide association remained signif-icant even when state-level analyses con-trolled for rates of poverty, urbanization, un-emplojonent, per capita alcohol consumption,and violent crimes other than homicide (i.e.,aggravated assault, forcible rape, and rob-bery). In fact, the cross-sectional associationbetween rates of firearm ownership andhomicide victimization was so stable overtime that regressions across states in any

December 2002, Vol 92, No, 12 I American Journal ot Public Health M/7(er et al. I Peer Reviewed I Research and Practice I 1991

given year produced point estimates that werewithin 8% of the point estimate obtainedwhen all 10 years of data were analyzed.

The association between household firearmownership rates and homicide rates held forvirtually all age groups and was particularlystrong for adults aged 25 years and older. Anexample is the category of homicide victimsaged 35 to 44 years. Table 2 indicates that ina comparison of states that differed by 1 stan-dard deviation in our firearm proxy (FS/S),the homidde rate was on average 35%higher in the states with the higher FS/S (i.e.,multivariate IRR=1.35). Given that FS/S was4-fold higher in states with the lowest relativeto those with the highest gun ownership rates,our multivariate model suggested that thehomidde rate in the high-gun states would be3.3 times that in the low-gun states (35%compounded 4-fold), and our bivariate modelsuggested a 3-fold difference (32% com-poimded 4-fold). Table 3 presents the corre-sponding bivariate comparison of the actualnumber of homidde victims in the states withthe 4 lowest and the 6 highest gun ownershiprates; for victims aged 35 to 44 years, homi-cide rates were 3,4 times higher in the high-gun states.

One reason that FS/S may be such a goodproxy for household firearm ownership isthat guns used for suidde appear typically tobe household guns. However, guns used inhomidde, especially homicides committed byadolescents and young adults, may often heobtained on the street. If, as has been re-ported,^^"^'' it is relatively easy for adoles-cents and young adults to acquire illegal gunson the street, the assodation between house-hold gun ownership inddence and rates ofhomidde committed by this age group mightbe diluted by this alternative source offirearms. Because individuals murdered by15- to 24-year-olds tend to be other 15- to24-year-olds,^'' this may explain, in part, ourfinding that the association between house-hold firearm ownership and the rate of homi-dde was stronger among adults 25 years andolder than it was among younger adults andadolescents. Consistent with this possibility,others have found that in areas with fewguns and strict gun control laws, criminaladolescents and young adults appear to ob-tain their fireamis via gun runners who pur-

chase the weapons in states with more per-missive gun laws.' ^

Neither survey estimates of householdfireann ownership nor our proxy is an idealmeasure of fireann availability. Surveys typi-cally imderrepresent poor people, and womenliving in 2-adult households with guns do notalways have accurate information aboutwhether a gun is present in their home. ' "''In addition, household firearm ownershiprates indicate nothing about the number ofguns per household, storage practices, or theease with which high-risk individuals can oh-tain firearms in secondary market transfers.Given that household firearm ownership ratesare likely to be only a crude measure offirearm availability, the robust assodation wereport between measures of firearm preva-lence and rates of homicide is striking.

Our study included only a limited numberof potential confounders—poverty, urbaniza-tion, xmemployment, alcohol consumption,and violent crimes (aggravated assault,forcible rape, and robbery)—and then only instate-level analyses. We found, consistent withprevious work, that homicide rates werehigher in areas with higher rates of urbaniza-tion, poverty, and nonlethai violent crime (notshown), ''" ** but many other factors may af-fect homidde rates. It is not clear, however,whether accotmdng for these or other area-wide characteristics would increase or reducethe magnitude or significance of the associa-tion between rates of household firearm own-ership and homicide.

Our study did not provide informationabout causation. One approach to evaluatingcausal direction is to iose a lagged measure ofthe key independent variable. Our findingthat a lagged measure of fireann ownershipyielded results similar to results obtEiined withcontemporaneous ownership and homiddemeasures is consistent with higher gun own-ership rates leading to higher homidde rates.However, this result does not rule out thepossibility that reverse causation or a non-causal explanation accounts for the assoda-tion between rates of firearm ownership andhomidde. It is possible, for example, that lo-cally elevated homicide rates may have led toincreased local gun acquisition. Unfortunately,we were unable to resolve this issue, in partbecause cross-sectional pattems of gun own-

ership rates across the United States are sostable over time.'"

The current study adds to previous workby using recent data, looking across both re-gions and all 50 states, disaggregating victimsby age, and adjusting for several potentialconfounders not previously accounted for innationally representative studies. We foundthat across US regions and states, and for vir-tually every age group, higher rates of house-hold firearm ownership were assodated withhigher rates of homicide. Our findings heldregardless of the following: M hether fii'earmownership rates were survey-based or derivedfrom a validated proxy, whether states mostextreme in ownership rates were excludedfrom analyses, whether the most and the leastpopulous states were excluded, and whetherregressions controlled for rates of poverty, ur-banization, unemployment, alcohol consump-tion, and violent crimes other than homidde.In areas with more firearms, people of allages were more likely to be murdered, espe-dally with handguns. •

About the AuthorsMatthew Miller. Deborah Azraei. and David Hemenwayare all from the Harvard School of Public Health, Boston.Mass.

Bequests for reprints should be sent to Matthew Miller,MD, MPH, ScD, Harvard School of Public Health, 677Huntington Ave, Boston, MA 02115 (e-mail: mmiller@hsph harvard.edu).

77JIS article was accepted February 18. 2002.

ContributorsEach author contributpd siibstantially to the conceptual-ization, design, and interpretation of the study, M, Miller\\Tote the article and analyzed the data. D, Azinel andD, Hemenway contnhuted to the writing of the articleand the analysis of the data, and made critical revisionsto various iterations of the aiticle.

AcknowledgmentsThis research was supported in part by graiits froni theJoyce Foundation, the Robert Wood johnsoii Fouiida-tion, artd the Open Society Institute,

Human Participant ProtectionNo protocol approval was needed for this study.

References1. Killias M. Gun ownership, suicide and homidde;an international perspective. In: del F'iate A, Zvekic U,van Dijk JM, eds. Understanding Crime: Experiences ofCrime and Crime Control. Rome, Italy: United NationsInterregional Crime and Justice Research Institute,1993:289-302,

2, Hemenway D, Miller M, Firearm availability

1992 I Research and Practice I Peer Reviewed I Miiier et al. American Journal of Publio Health I December 2002, Vol 92, No. 12

and homicide rates across 26 high-income countnes.J Trauma. 2000;49;985-988.

3. Nationai Center tbr Injury Prevention and Con-trol. Centers for Disease Control and Prevention. Injurymortality statistics. Available at; http://wonder.cdc.gov/mortICD9J.shtnil. Accessed January 10, 2002,

4. Sourcebook of CriminatJustice Statistics 2000.28th ed. Washington, DC: Bureau of Justice Statistics.US Dept of Justice; 2001: Table 3.125. Available at;http;//www.albany.edu/sourceb()ok/1995/pdf/t3125.pdf. Accessed January 10, 2002.

5. Kellermann AL, Rivara FP, Rushfoitb NB, et alGun ownership as a risk factor for homicide in thehome. NEnglJMed 1993;329:1084-lU9t.

6. Cummings P. Koepseli YD, Grossman DC, et al.The association between the purchase of a handgunand homicide or suicide. Am J Pubtic Health. 1997;87:974-978.

7. Rowland J, Holtzhauer F. Homicide involvingfirearms between family, relatives and friends in Ohio;an offender based case-conti-ol study. Am f Epidemiol.1989;130:825.

8. Kleck G. Hogan M. A national case-control studyof homicide offending and gun ownership, Soc Probl.1999;46;275-293.

9. Kleck G. Patterson EB. 'ITie impact of gun controland gun ownership levels on violence rates. / QuantCriminol 1993;9(3):249-287.

10. Kapian M, Geling 0. Firearm suicides and homi-cides in the United States: regional variations and pat-terns of gun ownership. Soc Sci Med. 1998;46(9):1227-1233.

11. Lester D. Firearm availability and the incidence ofsuicide and homicide. Acta Psychiatr Betg. 1988;88;387-393-

12. Brearley UC. Homicide in the United States. ChapelHill. NC; University of North Carolina Press, 1932.

13. Seitz ST. Firearms, homicide, and gun control ef-fectiveness. LawSocRev. 1972;6i595-614.

14. Lester D. Relationship between firearm availabil-ity and primary and secondary murder. Psychot Rep.1990;67:490.

15. Cook P. The effect of gun availability on robberyand rohbery murder. In: Havemian R. Zellner B. eds.Poticy Studies R^iew Annual. Tlioitsand Oaks, Calif;Sage PubUcations; 1979:743-781.

Itt. Brill S. Firearm Abuse: A Research andPoiiq/Re-port. Washington, DC; Police Foundation. 1977.

17. Duggan M. More gtms, more crime. J Potit Econ.2002;109;1086-m4.

18. Davis JA, Smith TW. General Social Surveys,1972-1998 [madiine-readable data file]. Chicago, IU;National Opinion Research Center Iproducer]; Storrs.Conn; The Roper Center for Public Opinion Research,University of Connecticut Idistributor]; 1998.

19. Azrael D. Cook P, Miller M. State and Local Preva-lence of Firearms Ownership: Measurement, Structure, andTrends. National Bureau of Economic Research Work-ing Paper W8570. October 2001. Available at; http;//www.nber.org/papers/w8570. Accessed January 10,2002.

20. National Center for Health Statisdcs. Hyattsville,

Md; National Center for Health Statistics, Centers forDisease Control and Prevention; 1998, Available at:http://wonder.cdc.gov/mDrlsql.shtml, Accessed Janu-ary 10, 2002,

21. Powell KE. Jacklin BC, Nelson DE, Biand S. Stateestimates of household exposure to firearms, loadedfirearms, and handguns. 1991 through 1995. AmfPubtic Health. l998;88;969-972.

22. Killias M. Intemationai correlations between gunownership and rates of homicide and suicide. CMAJ.1993;148;1721-1725.

23. Miller M, Azrael D, Hemenway D, Firearm avail-ability and unintentional firearm deaths, suicides andhomicides among 5-14 year olds. / trauma. 2002;52;267-275.

24. Kleck G. Targeting Guns: Firearms and Their Con-troi. New York, NY; Aldine de Gruyter, 1997.

25. Baker S, O'Neill B. Ginsburg M, Li G. 77ie InjuryFact Book. 2nd ed. New York, NY; Oxford Universi^Press. 1992.

26. Fingerhut LA, Ingram DD. Feldman JJ. Firearmand nonfirearm homicide among persons 15 through19 years of age: differences by level of urbanization.United States, 1979-1989.//1AM. 1992;267;3048-3053.

27. Hsieh CC, Pugh MD. Poverty, income inequality,and violent crime; a meta-analysis of recent data stud-ies. Criminal Justice Rev. 1993;18;182-202.

28. Parker KF. Pruitt MV. Poverty, poverty concentra-tion, and homicide. Soc Sd Q. 2000;81:555-57O.

29. Goodman RA, Mercy JA. Loya F, et al. Alcoholuse and interpersonal violence. Amj Public Heatth.1986;76;144-]49.

30. National Institute on Aicohoi Abuse and Alco-holism. Per capita ethanol consumption for states, cen-sus regions, and the United States, 1970-98 [table].Available at; http;//www.niaaa.nih,gov/databases/consum03,txt. Accessed Janttary 10, 2002.

31. Statistical Abstracts of the tJnited States. 1988-1998. Washington, DC; US Bureau of the Census.

32. Cnme Gun Trace Reports, 1999: 77ie Youth CrimeGun Interdiction Initiative. Washington. DC; Bureau ofAlcohol, Ibbacco. and Firearms; Department of theTreasury. November 2000.

33. Ber^tein JM. Hemenway D. Kennedy B, QuadayS, Ander R. Guns in young hands; a survey of urbanteenagers' attitudes and behaviors related tn handgunviolence, / 7hjuma 1996;41:794-798.

34. Page RM, Hammermeister J. Weapon carryingand youth violence. Adolescence. 1997;32(127);505-513.

35. Snyder H, Finnegan T, Wan Y, Kang W Easy Ac-cess to the FBI's Supptementary Homicide Reports:1980-1999. 2001. National Center for Juvenile Jus-tice. Available at; http;//ojjdp.ncjrs.org/ojstatbb/ezashr.Accessed January 10, 2002.

36. Ludwig J, Cook PJ, Smith TW. The gender gap inreporting household gun ownership. Amj PublicHeatth. 1998;88;I7I5-1718.

37. Azrael D. Hemenway D. In the safety of yourown home; results from a nationai survey on gun useat home. SocSdMed. 2000;50;285-291.

Compendium of Methods for the

MicrobiologicalExamination of FoodsEdited by Frances Pouch Downes

and Keith Ito

The fourth edition of the Compendiumdocuments the latest, state-of-the-art

technologies in microbiology and molecu-lar biology that play a key role—from pro-duction to consumption—in the safety offood.

This book brings together more than200 experts from business, governmentand academia to address key issues suchas quality assurance, common pathogens,laboratory procedures, shipment practicesand the safety of specific foods.

This book is vital to all those involvedin I Food processing and manufacturing IFood testing for safety and quality IPathogen surveillance I Laboratory scienceI Public health safety

ISBN 0-87553-175-X2001 I 704 pages I Hardcover

$87,50 APHA Members$125,00 Nonmembersplus shipping and handling

Bulk Order Discounts andExamination Copies are available.

American Public Health AssociationPublication SalesWeb: www.apha.orgE-mail: APHA@TASC01 .comTel: (301) 893-1894FAX: (301) 843-0159 FDOIJ?

December 2002, Vol 92, No. 12 I American Journal of Public Health tvtilter et al. I Peer Reviewed I Research and Practice ! 1993