Guarantor: William F. Page, PhD Contributors: Clare M. Mahan, PhD*; William F. Page, PhD ; Tim A. Bullman, MS*; Han K. Kang, DrPH*

Transcription

1 MILITARY MEDICINE, 170, 11:935, 2005 Helth Effects in Army Gulf Wr Veterns Possily Exposed to Chemicl Munitions Destruction t Khmisiyh, Irq: Prt I. Moridity Associted with Potentil Exposure Gurntor: Willim F. Pge, PhD Contriutors: Clre M. Mhn, PhD*; Willim F. Pge, PhD ; Tim A. Bullmn, MS*; Hn K. Kng, DrPH* In Mrch 1991, U.S. troops detonted the Khmisiyh, Irq, mmunition depot, possily relesing two chemicl wrfre gents, srin nd cyclosrin. The long-term helth effects ssocited with possile exposure to these chemicl wrfre gents re unknown. This study ws undertken to investigte whether possile exposure ws ssocited with moridity mong Army Gulf Wr veterns using moridity dt for 5,555 Army veterns who were deployed to the Gulf region. Responses to 86 self-ssessed helth mesures, s reported in the 1995 Deprtment of Veterns Affirs Ntionl Helth Survey of Gulf Wr Er Veterns, were evluted. We found little ssocition etween potentil exposure nd helth, fter djustment for demogrphic vriles, nd conclude tht potentil exposure to srin or cyclosrin t Khmisiyh does not seem to hve dversely ffected self-perceived helth sttus, s evidenced y wide rnge of helth mesures. Introduction mmeditely fter the Gulf Wr, demolition ws crried out in I Mrch 1991 t the Khmisiyh mmunition depot in southestern Irq. Troops who were possily exposed to chemicl wrfre gents were identified susequently y environmentl nd climtologicl modeling, in conjunction with unit loction dt for the dys of the demolition. 1 In this rticle, we compre the moridity outcomes in the group of Army veterns possily exposed to low levels of chemicl wrfre gents with those of similr group of unexposed Army personnel. The moridity dt were collected s prt of the Deprtment of Veterns Affirs (VA) Ntionl Helth Survey of Gulf Wr Er Veterns (NHS). 2 Two other rticles exmine deths ssocited with possile exposure 3 nd moridity ssocited with notifiction of possile exposure. 4 On Mrch 4 nd 10, 1991, comt engineer nd explosive ordnnce disposl units of the U.S. Army XVIII Corps (Airorne) destroyed two lrge cches of 122-mm rockets, one in unker nd the other in nery pit, t the Khmisiyh mmunition supply point, 350 km southest of Bghdd, Irq. In Octoer 1991, Mrch 1992, My 1992, nd My 1998, representtives from the United Ntions Specil Commission inspected Khmisiyh nd detected the existence of srin nd cyclosrin in oth intct nd dmged rockets in the unker nd pit. 1 *Veterns Helth Administrtion, Deprtment of Veterns Affirs, Wshington, DC Medicl Follow-up Agency, Institute of Medicine, Wshington, DC The opinions nd ssertions contined herein re those of the uthors nd re not to e construed s necessrily reflecting the views or positions of the Ntionl Acdemy of Sciences, the Institute of Medicine, the Ntionl Reserch Council, or the Deprtment of Veterns Affirs. This mnuscript ws received for review in June The revised mnuscript ws ccepted for puliction in Decemer Reprint & Copyright y Assocition of Militry Surgeons of U.S., Approximtely contemporneously, concerns incresed out postwr moridity mong Gulf Wr veterns On June 21, 1996, the Deprtment of Defense (DoD) relesed sttement confirming tht U.S. soldiers hd destroyed mmunition unkers t Khmisiyh, Irq, nd tht one of these unkers contined chemicl wrfre gents. 13 Following this, the DoD mde efforts to determine who ws possily exposed to chemicl gents (see elow) nd lso mde efforts to notify veterns of possile exposure; the effects of these notifiction letters re the suject of nother rticle. 4 Toxicology of Nerve Agents Srin nd Cyclosrin Srin, n orgnophosphorus ester, is highly toxic nerve gent. Exposure to cutely toxic concentrtions cn produce excessive ronchil, slivry, oculr, nd intestinl secretions, s well s sweting, miosis, ronchospsm, rdycrdi, muscle fscicultions, prlysis, convulsions, nd deth. 14 Miniml effects oserved t low concentrtions include miosis, chest tightness, rhinorrhe, nd dyspne. 14 There is limited evidence ssociting srin exposure t level sufficient to produce cute cholinergic signs with susequent long-term helth effects, such s ftigue, hedche, lurred vision, post-trumtic stress disorder (PTSD), nd norml test results of unknown clinicl significnce. 15 At doses too low to produce cute cholinergic effects, there is insufficient evidence to determine whether there is n ssocition with susequent long-term helth effects, in prt ecuse of lck of well-controlled studies. 15 Cyclosrin is similr in composition to srin, lthough less voltile. Its mechnism of ction is similr to tht of srin, lthough less is known out its toxicity. 15 A recent study of self-reported, long-term (25 45 yers), helth effects mong 1,339 veterns experimentlly exposed to nticholinesterse gents (including 287 exposed to srin) included neurologicl nd psychologicl outcomes such s peripherl nerve disese, vestiulr dysfunction, sleep disorders, nxiety, nd depression. There were only two sttisticlly significnt differences, i.e., sujects exposed to nticholinesterse gents hd fewer ttention prolems thn sujects in one control group nd greter sleep disturnce prolems thn sujects in nother control group. In contrst, self-reported exposure to hzrdous chemicls outside the experimentl testing progrm ws significntly ssocited with ll primry study outcomes. 16 Methods Study Popultion The cohort for this study ws selected in collortion with the Office of the Specil Assistnt for Gulf Wr Illnesses, the 935

2 936 Exposure nd Moridity Deployment Environmentl Surveillnce Progrm of the U.S. Army Center for Helth Promotion nd Preventive Medicine, nd the VA Environmentl Epidemiology Service. Eligiility for entry into the cohort ws sed on the vetern hving served in the Gulf theter of opertions. Individuls identified s hving een within nd outside the modeled potentil hzrd re were eligile for inclusion. The cohort ws further defined y hving prticipted in Phse I of the VA NHS, conducted in The NHS ws designed s retrospective cohort study in which helth fctors of popultion-sed smple of 15,000 troops deployed into the Gulf re were compred with those of 15,000 troops serving in the militry during the period of the Gulf Wr ut not in the Gulf re. Phse I of the survey ws performed in , efore troop notifiction of possile chemicl gent exposure t Khmisiyh, Irq, y the DoD. A totl of 11,441 militry personnel, who represented four rnches of service deployed to the Gulf region during the Gulf Wr, responded to either Phse I, the postl printed questionnire survey in , or Phse II, the telephone interview survey in The suset of veterns who prticipted in the postl questionnire survey nd who served in the Army numered 5,555. Determining Possile Exposure The risk fctors ssocited with the demolition in the Khmisiyh pit in Mrch 1991 re possile exposures to chemicl wrfre gents, including srin nd cyclosrin. 1 For completeness, we lso exmined the dt using exposure defined with the 50-km model, n erly exposure model tht declred Gulf Wr veterns who were within circle with rdius of 50 km, centered t Khmisiyh, Irq, to hve een possily exposed. 1 The history of DoD s exposure determintion efforts is given elow in rief. Determining the possile risk of chemicl gent exposure to U.S. troops in the vicinity of Khmisiyh egn s joint effort y the Centrl Intelligence Agency nd the DoD in lte It quickly ecme pprent tht the pit demolition posed numer of chllenges requiring expertise in modeling the physicl chrcteristics of open-ir demolition, s well s environmentl nd meteorologicl conditions t the site. The DoD-Centrl Intelligence Agency tem used interviews with troops who hd een t the site nd test demolitions nd other experiments t the Dugwy Proving Grounds nd Edgewood Lortories to reduce uncertinties ssocited with the physicl nd environmentl conditions t the site. Becuse of reltively scrce meteorologicl dt for Irq, the tem used stte-of-the-rt mesoscle meteorologicl models to simulte previling wether conditions over the region. Dispersion models were then used to predict the trnsport nd spred of chemicl wrfre gents, sed on these simulted meteorologicl conditions. To ccount for uncertinty, conservtive ssumption ws mde to define the potentil hzrd re s the union of the hzrd res given y ech of the vrious comintions of meteorologicl nd dispersion models. The result ws the genertion of potentil hzrd re tht vried in size nd shpe from Mrch 10 to Mrch 13, From this, the tem ws le to determine which units of troops were presumed to hve een within the potentil hzrd re over the course of the 4-dy period. The result of this effort is known s the 1997 hzrd re (see Ref. 1 for dditionl detils). Severl fctors contriuted to the development of susequent exposure model, the remodeling effort or the 2000 hzrd re. Meteorologicl modeling methods were further refined. The ssumed mount of nerve gent relesed ws reduced, sed on more-recent intelligence nlyses. The toxicity effects of cyclosrin were dded to those of srin, which ws used lone in the 1997 nlysis. Atmospheric removl mechnisms, such s dry deposition nd degrdtion, were lso considered. Finlly, nd possily of most importnce, unit loction informtion ws significntly improved in the 2000 model. Both models considered soldier to hve een exposed if his or her unit ws in the hzrd re ut in 1997 the unit ws generlly t the ttlion level ( 1,000 soldiers), wheres in 2000 the unit ws generlly t the compny level ( 200 soldiers). Therefore, on verge, n individul s unit ws locted with greter precision in the 2000 hzrd re. In this study, n individul is deemed presumed exposed if, during ny of the four 24-hour periods from Mrch 10 to Mrch 13, 1991, his or her unit ws t loction tht ws exposed to level of chemicl wrfre gent higher tht the generl popultion limit, djusted for short-term exposure. 1 The generl popultion limit is defined s the limit elow which ny memer of the generl popultion could e exposed dily, for lifetime, nd not experience ny relted dverse helth effects. The modeling process lso provided dt on the numer of dys (0 or 1 through 4) of potentil exposure. Although this does not constitute true mesure of intensity of exposure, we used numer of dys of exposure in some of our nlyses s limited proxy for intensity of exposure. We dopt the terms exposed to men possile exposure under the 2000 hzrd re, unexposed to men no exposure under the 2000 hzrd re, nd missing exposure to men tht there ws no 2000 hzrd re exposure sttus ssigned to n individul, ecuse of missing or incorrect informtion identifying the individul or his or her militry unit. Through use of the 2000 hzrd re nd loction dt for soldiers t the compny level in the Gulf region, exposure determintion ws provided for the 351,121 deployed Army personnel y the Office of the Specil Assistnt for Gulf Wr Illnesses. Record linkge of this chemicl exposure dt with the file of respondents to Phse I of the VA NHS yielded totl of 5,555 Army Gulf Wr veterns, who were clssified into one of three ctegories of exposure to chemicl gents, minly srin or cyclosrin, s follows: exposed Army Gulf Wr veterns, n 1,898; unexposed Army Gulf Wr veterns, n 3,336; missing exposure Army Gulf Wr veterns, n 321. Helth Outcomes Helth perception dt, efore notifiction, were derived for ll 5,555 Army Gulf Wr veterns who responded during Phse I of the VA NHS. The outcomes of interest re the self-ssessments tht were reported on the postl questionnire developed for the VA NHS. The self-dministered questionnire ws used to otin informtion on the presence of vrious medicl conditions nd symptoms, mesures of functionl impirment, limittion of ctivity, nd helth cre utiliztion. A self-report symptom inventory composed of 48 items tht were representtive of the symptom configurtion commonly

3 Exposure nd Moridity 937 TABLE I PREVALENCE OF SELECTED SEVERE SYMPTOMS DURING THE PAST 12 MONTHS ACCORDING TO EXPOSURE STATUS FOR 5,555 ARMY GULF WAR VETERANS Symptoms Grouped y Orgn System Exposed (n 1,898) Prevlence (%) Unexposed (n 3,336) Adjusted Risk Rtio (95% CI) Generl Hedches ( ) Runny nose ( ) Generl muscle ches/crmps ( ) Excessive ftigue ( ) Ftigue 24 hours fter exertion ( ) Sleep difficulty ( ) Sleepiness during dytime ( ) Awken tired or worn out ( ) Fever or chills ( ) Sweting not ttriutle to exercise ( ) Skin, hir, nd nils Skin rsh ( ) Hir loss ( ) Ers Hering loss ( ) Mouth nd throt Mouth, teeth, or gum prolems ( ) Sore throt or horse voice ( ) Troule swllowing ( ) Respirtory Wheezing ( ) Coughing ( ) Brething or shortness of reth ( ) Crdic Irregulr hertet ( ) Gstrointestinl Nuse ( ) Vomiting ( ) Stomch or dominl pin ( ) Reflux, herturn, or indigestion ( ) Dirrhe ( ) Constiption ( ) Genitourinry/reproductive Frequent/pinful urintion ( ) Pinful sexul intercourse ( ) Impotence or other sexul prolems ( ) Musculoskeletl Bck pin/spsms ( ) Joint ches or pin ( ) Swelling in ny joints ( ) Hemtologicl Bruise or leed esily ( ) Sensory/neurologicl Loss of lnce/dizziness ( ) Blurred vision ( ) Speech difficulty ( ) Sudden loss of strength ( ) Concentrtion/memory prolems ( ) Numness in hnds/feet ( ) Tremor/shking ( ) Psychitric Anxious, irritle, or upset ( ) Been depressed or lue ( ) Immunologicl Sensitive to chemicls ( ) Lymphtic Swollen glnds ( ) Other (symptom meets the criteri for more thn one ctegory) Tightness in chest ( ) Swelling of feet/nkles ( ) Wound slow to hel ( ) Adjusted risk rtios (nd 95% confidence intervls [CIs]) from Cochrn-Mntel-Henszel nlysis, with djustment for ge, gender, rce, rnk, mritl sttus, nd unit component.

4 938 Exposure nd Moridity oserved mong outptients ws used to ssess the prevlence of somtic nd psychologicl symptoms. Complete lists of symptoms (Tle I), grouped ccording to orgn system, nd conditions (Tle II) re presented. Questionnire items were used to evlute limittion of ctivity, prevlence of chronic conditions, self-ssessed helth sttus, nd use of medicl services, including physicin contcts relted to illness (excluding routine visits for vccintions nd physicl exmintions) nd hospitliztions overnight. The PTSD Checklist, which ws prt of the survey instrument, contined 17 questions concerning PTSD symptoms experienced in the pst 1 month. 17 A cutoff score of 50 points ws used to clssify veterns s hving PTSD. 18 Prticipnts were clssified regrding the presence of chronic ftigue syndrome (CFS). The cse definition tht hd een coordinted y the Centers for Disese Control nd Prevention in ws modified in terms of the time period for durtion of symptoms tht ws ville on the NHS survey instrument. This modifiction ws descried nd evluted elsewhere nd is termed CFS-like illness. 20 Questions on smoking history nd lcohol use were included, s well s weight chnge in the pst 6 months. All of these outcomes were fully descried elsewhere for the entire cohort of 11,441 Gulf Wr respondents nd 9,476 non-gulf Wr respondents. 2 Demogrphic Chrcteristics nd Militry Vriles Bsic demogrphic dt nd militry vriles (dte of irth, gender, mritl sttus, rce, rnch, rnk, militry occuptionl specilty code, nd unit component) were derived from Defense Mnpower Dt Center (DMDC) nd VA records. Dte of irth, gender, nd mritl sttus were lso investigted with the survey instrument. Gender, inry vrile, ws coded s mle vs. femle. Age t the time of the Gulf Wr (1991) ws recoded from the DMDC dtse s inry vrile, i.e., 30 yers vs. 30 yers. Rce/ethnicity ws coded into three strt, i.e., (1) Cucsin, (2) Africn Americn, or (3) Hispnic, Americn Indin, Alskn Ntive, Asin, Pcific Islnder, ny other, or missing. Mritl sttus t the time of the Gulf Wr, from the DMDC file, ws converted to inry vrile, i.e., single vs. ever mrried. Rnk t the time of the Gulf Wr ws coded s inry vrile, i.e., enlisted vs. officer or wrrnt officer. Unit component ws coded into three strt, i.e., Army ctive, Army Ntionl Gurd, or Army Reserve. Sttisticl Methods Anlyses mesured the ssocition etween presumed exposure to chemicl gents nd the vrious helth outcomes descried in the survey instrument. We used contingency tle TABLE II PREVALENCE OF SELECTED SELF-REPORTED MEDICAL CONDITIONS DURING THE PAST 12 MONTHS ACCORDING TO EXPOSURE STATUS FOR 5,555 ARMY GULF WAR VETERANS Conditions Prevlence (%) Exposed (n 1,898) Unexposed (n 3,336) Adjusted Risk Rtio (95% CI) Arthritis ( ) Lumgo ( ) Diseses of muscles or tendons ( ) Skin cncer ( ) Other cncer ( ) Eczem or psorisis ( ) Other dermtitis ( ) Diseses of the hir or sclp, hir loss ( ) Cirrhosis of liver ( ) Heptitis ( ) Other liver disese ( ) Gstritis ( ) Enteritis ( ) Colitis ( ) Frequent dirrhe ( ) Dietes mellitus ( ) Other endocrine disorder ( ) Repeted seizures ( ) Recurrent hedches ( ) Migrines ( ) Neurlgi or neuritis ( ) Any disese of genitl orgns ( ) Hert disese ( ) Hypertension ( ) Stroke ( ) Tchycrdi ( ) Sinusitis ( ) Bronchitis ( ) Asthm ( ) Other lung condition ( ) Bldder infection ( ) Adjusted risk rtios (nd 95% confidence intervls [CIs]) from Cochrn-Mntel-Henszel nlysis, with djustment for ge, gender, rce, rnk, mritl sttus, nd unit component.

5 Exposure nd Moridity 939 TABLE III PERCENTAGE DISTRIBUTION OF SELECTED DEMOGRAPHIC/MILITARY SERVICE CHARACTERISTICS AMONG 5,555 ARMY GULF WAR VETERANS ACCORDING TO POSSIBLE EXPOSURE TO CHEMICAL AGENTS Exposed (n 1,898) Unexposed (n 3,336) Missing Exposure (n 321) Chrcteristics No. % No. % No. % Gender Mle 1, , Femle Age (yers) (men in 1991) Interqurtile rnge (yers) Rce Cucsin 1, , Africn Americn Hispnic Other Missing Mritl sttus Mrried 1, , Single , Other Rnk Enlisted 1, , Officer Wrrnt Unit component Active Ntionl Gurd , Reserve , Totl (n 5,555) is composed of possily exposed (n 1,898) plus unexposed (presumed not in hzrd re) (n 3,336) plus missing exposure sttus (n 321). Possily exposed is sed on the DoD revised 2000 hzrd re. Rce other includes Americn Indin, Alskn Ntive, Asin, Pcific Islnder, nd ny other. nlysis to exmine differences with respect to demogrphic nd militry fctors etween exposed nd unexposed Army Gulf Wr veterns. Undjusted risk rtios (reltive risk) were clculted directly from the rw tle entries. Adjusted risk rtios included djustments for ge in 1991 ( 30 yers vs. 30 yers), gender, rce/ethnicity (Cucsin, Africn Americn, or ll other, including Hispnic), mritl sttus (single vs. ll other), rnk (enlisted vs. officer or wrrnt officer), nd Army ctive duty vs. Army Reserve or Ntionl Gurd service. These djustment fctors were chosen sed on their potentil ssocition with helth outcomes. We djusted for these fctors using propensity score, 21 which, divided into quintiles, ws used to strtify the Cochrn-Mntel-Henszel nlyses. Sttisticl significnce ws scertined y exmining the coverge of 95% confidence intervls, nd no djustments were mde for multiple comprisons. Computtions were crried out with stndrd softwre. 22 Imputtion To void losing oservtions ecuse of missing demogrphic nd militry dt in the DMDC dt file, we extrcted the corresponding items from the NHS Phse I postl questionnire, when possile. For the few cses of missing vlues for the rce vrile for which this pproch did not work (n 10), we creted ctegory of missing for the crude rtes ut pooled the 10 missing oservtions with the other rce strtum for the clcultion of djusted risk rtios. No imputtion ws performed for item nonresponses regrding outcome vriles. Sensitivity Anlyses Becuse 5.7% of the Army Gulf Wr veterns hd missing exposure dt, sensitivity nlysis ws conducted for the min outcomes, to ssess the potentil effect of these missing dt. First, sujects in the unknown exposure group were dded to the exposed group nd compred with unexposed sujects. Second, sujects in the missing exposure group were dded to the unexposed group nd compred with exposed sujects. Institutionl Review Bord Approvl This project ws sumitted for institutionl review ord review, nd pprovl ws otined from oth the Deprtment of Veterns Affirs nd the Ntionl Acdemy of Sciences. Results Chrcteristics of Exposed nd Unexposed Army Gulf Wr Veterns The demogrphic nd militry chrcteristics of the 5,555 Army Gulf Wr veterns re presented in Tle III, ctegorized y the three levels of exposure to chemicl gents. The exposed group hd greter proportion of femle veterns thn did the unexposed group (24.9% vs. 21.1%; p 0.01) nd differed y rce, hving fewer Africn Americn veterns (19.8% vs. 21.5%) nd greter proportion of other rces (Hispnic, Americn Indin/Alskn, or Asin/Pcific Islnder) (8.3% vs. 6.3%) thn the unexposed group (p 0.013). Regrding unit component,

6 940 Exposure nd Moridity there ws greter proportion of reservists in the exposed group thn in the unexposed group (35.0% vs. 31.6%; p 0.02). There were no significnt differences etween the exposed nd unexposed groups for the vriles ge, mritl sttus, nd rnk. Functionl Impirment, Limittion of Work, nd Medicl Cre Utiliztion Attriutle to Illness Approximtely the sme percentges of exposed Army Gulf Wr veterns nd unexposed Army Gulf Wr veterns reported stying home ll or prt of dy ecuse they did not feel well or s result of illness or injury within the 2 weeks efore the survey response (31.0% vs. 31.9%) (Tle IV). The percentge distriution of numer of dys t home during the pst 2 weeks is shown for oth groups; there ws no difference etween these two distriutions (p 0.15). Similrly, 20.9% of exposed Gulf Wr veterns nd 22.0% of unexposed Gulf Wr veterns reported tht they were limited in employment or in the kind of work they could do round the house ecuse of n impirment or helth prolem (p 0.36). Less thn 2% of the respondents did not nswer either of these two questions. There ws no difference etween the exposed nd unexposed groups in the numer of clinic or doctor visits mde during the pst 12 months ecuse of illness (p 0.62). Approximtely 6% of ech group did not nswer this question. Among the exposed veterns, 8.9% reported hving een hospitlized overnight or longer for illness during the pst 12 months, wheres the corresponding percentge mong unexposed veterns ws 9.2%. The percentge distriutions of those hospitlized were not different etween the two groups (p 0.88). Item nonresponse rtes for this question were 3.4% for the two groups. Sensitivity Anlyses In Tle V, we ssumed tht ll Army Gulf Wr veterns with missing exposure dt were ctully exposed. We consequently found no sttisticlly significnt differences for functionl impirment (ed dys), limittion of employment, numer of clinic or doctor visits, or numer of hospitliztions. Conversely, when we ssumed tht the veterns with missing exposure dt were ctully unexposed (Tle VI), we reched the sme conclusion. Becuse these results re the sme s those in Tle IV, we excluded sujects with missing exposure dt from the reminder of the nlyses in Tles I, II, VII, VIII, nd IX. Perception of Generl Helth There ws no difference in perception of generl helth sttus etween exposed nd unexposed veterns (p 0.72; Tle VII). The item nonresponse rte for this question ws 14% for ech group, which is higher thn rtes for other items in the survey ecuse the physicl plcement of this item on the survey form ment tht it ws indvertently skipped y prticipnts. TABLE IV PERCENTAGE DISTRIBUTION OF FUNCTIONAL IMPAIRMENT, LIMITATION OF EMPLOYMENT, AND MEDICAL CARE UTILIZATION ATTRIBUTABLE TO ILLNESS AMONG 5,555 ARMY GULF WAR VETERANS ACCORDING TO EXPOSURE STATUS Exposed (n 1,898) Unexposed (n 3,336) Missing Exposure (n 321) Conditions No. % No. % No. % p Functionl impirment Bed dys 0 1, , Not nswered Limittion of employment No 1, , Yes Not nswered Clinic visit during pst 12 months None , Not nswered Hospitliztion during pst 12 months None 1, , Not nswered p, significnce proility y 2 test of independence etween exposure ctegories (known exposed vs. known unexposed) nd helth condition indictor. Not nswered ctegory is presented for interest ut is not included in the denomintor in clcultion of percentges.

7 Exposure nd Moridity 941 TABLE V SENSITIVITY ANALYSIS WITH 321 VETERANS WITH MISSING EXPOSURE STATUS INCLUDED IN THE CATEGORY OF EXPOSED: PERCENTAGE DISTRIBUTION OF FUNCTIONAL IMPAIRMENT, LIMITATION OF EMPLOYMENT, AND MEDICAL CARE UTILIZATION ATTRIBUTABLE TO ILLNESS Exposed Missing Exposure (n 1, ) Unexposed (n 3,336) Conditions No. % No. % p Functionl impirment Bed dys 0 1, , Not nswered Limittion of employment No 1, , Yes Not nswered Clinic visit during pst 12 months None , Not nswered Hospitliztion during pst 12 months None 1, , Not nswered p, significnce proility y 2 test of independence etween exposure ctegory nd helth condition indictor. Not nswered ctegory is presented for interest ut is not included in the denomintor in clcultion of percentges. Prevlence of Severe Symptoms On the survey instrument, the vetern ws given the guideline for reporting symptom experienced during the pst yer s severe if its presence ws sufficient to seek medicl dvice, tke prescription drugs, lose work, or limit routine ctivities. The prevlences of severe symptoms for exposed nd unexposed veterns re presented in Tle I for 47 symptoms (one symptom pertined only to mle veterns nd ws excluded). The djusted risk rtios showed two sttisticlly significnt ssocitions etween exposure nd self-reported severe symptoms, i.e., the rtes of wheezing nd of swelling in ny joints were lower in the exposed group. Becuse one would expect 1 of 20 sttisticl tests performed t the p 0.05 level to e sttisticlly significnt y chnce, the finding of two significnt risk rtios in 47 tests could well e ttriutle to chnce lone. Prevlence of Mild or Severe Symptoms On the survey instrument, the vetern ws given the guideline for reporting symptom experienced during the pst yer s mild if its presence cused the vetern to e just wre ut not slowed down y symptoms, or sufficient to tke nonprescription drugs to relieve the symptoms (spirin, Tums, etc.). The prevlence of either mild or severe symptom ws computed for ech of the 47 symptoms, nd results were compred for exposed nd unexposed veterns. The djusted risk rtios were not mterilly different from those for severe symptoms. Prevlence of Chronic Medicl Conditions The prevlences of 31 medicl conditions during the pst 12 months re shown in Tle II for exposed nd unexposed veterns. Of the 31 medicl conditions in Tle II, ll except two showed the sme rtes in the exposed nd unexposed groups; enteritis nd colitis hd significntly higher rtes in the exposed group. Agin, ecuse one would expect 1 of 20 sttisticl tests performed t the p 0.05 level to e sttisticlly significnt y chnce, the finding of two significnt risk rtios in 31 tests could well e ttriutle to chnce lone. In Tle VIII, the prevlences of two symptom-sed medicl conditions, PTSD nd CFS-like illness, re tulted. There ws no significnt difference in the prevlence of either medicl condition ccording to exposure sttus. Alterntive Mesures of Exposure Next, we exmined the reltionship etween numer of dys of possile exposure to chemicl gents from Mrch 10 to 13, 1991, coded s 0, 1, or 2 dys, nd generl helth sttus, numer of clinic or doctor visits, numer of hospitliztions, PTSD, nd CFS-like illness (Tle IX). The 2 tests showed no reltionships etween the numer of dys exposed (coded s 0, 1, or 2) nd ny of the helth sttus indices (p 0.50). Finlly, using n lterntive definition of exposure sed on the 50-km proximity, i.e., within the 50-km rdius from Khmisiyh from Mrch 10 to 13, 1991, we ssessed the frequency distriutions of the sme helth indices (Tle X). The 2 tests showed no

8 942 Exposure nd Moridity TABLE VI SENSITIVITY ANALYSIS WITH 321 VETERANS WITH MISSING EXPOSURE STATUS INCLUDED IN THE CATEGORY OF UNEXPOSED: PERCENTAGE DISTRIBUTION OF FUNCTIONAL IMPAIRMENT, LIMITATION OF EMPLOYMENT, AND MEDICAL CARE UTILIZATION ATTRIBUTABLE TO ILLNESS Conditions No. % No. % p Functionl impirment Bed dys 0 1, , Not nswered Limittion of employment No 1, , Yes Not nswered Clinic visit during pst 12 months None , , Not nswered Hospitliztion during pst 12 months None 1, , Not nswered ssocition etween 50-km exposure defined nd ny of the helth sttus indices (p 0.45). Discussion Exposed (n 1,898) A popultion-sed smple of Army Gulf Wr veterns who prticipted in the NHS nd who were presumed to hve een exposed to chemicl gents in Khmisiyh, Irq, did not report higher rtes of dys spent in ed within the pst 2 weeks, helth limittion, or medicl cre utiliztion, s mesured y outptient office visits or inptient hospitliztions within the pst Unexposed Missing Exposure (n 3, ) p, significnce proility y 2 test of independence etween exposure ctegory nd helth condition indictor. Not nswered ctegory is presented for interest ut is not included in the denomintor in clcultion of percentges. TABLE VII PERCENTAGE DISTRIBUTION OF PERCEPTION OF GENERAL HEALTH AS REPORTED BY VETERANS ACCORDING TO EXPOSURE TO NERVE AGENTS SARIN OR CYCLOSARIN Exposed (n 1,898) Unexposed (n 3,336) Generl Helth No. % No. % Excellent Very good Good Fir Poor Not nswered Significnce proility y 2 test of independence etween exposure nd generl helth sttus, p 0.72, Not nswered ctegory is presented for interest ut is not included in the denomintor in clcultion of percentges. The excess item nonresponse for this question is ttriutle to the position of the question on the survey instrument. yer, thn did similr unexposed veterns. Similrly, the two groups perceived their helth sttus to e the sme when sked to clssify themselves into one of five ctegories, rnging from excellent to very good, good, fir, or poor. After djustment for possile differences in demogrphic/militry chrcteristics, the sujects in the exposed nd unexposed groups reported equl prevlences for 45 of 47 severe symptoms nd for 29 of 31 medicl conditions. We lso note tht, sed on wht is known out helth effects following exposure to nticholinesterse gents, we would hve expected differences in neurologicl nd ehviorl symptoms, which we did not oserve.

9 Exposure nd Moridity 943 TABLE VIII PERCENTAGE OF SYMPTOM-BASED MEDICAL CONDITIONS ACCORDING TO EXPOSURE STATUS Medicl Condition Exposed (n 1,898) Prevlence (%) Unexposed (n 3,336) PTSD CFS-like illness p, significnce proility y 2 test of independence etween exposure nd symptom-sed medicl condition. PTSD, 50 points y self-report on PTSD Checklist. The ccurcy of exposure mesurement is lwys n issue in this kind of study. The fct tht there hve een three officil exposure models to dte shows tht the science nd technology hve een evolving. Even so, there re some who dout the ccurcy of the process. 23 We were le to scertin tht the results we otined were roust with respect to the choice of model, y performing nlyses sed on the most recent dt, the 2000 hzrd re, s well s the erliest dt, the 50-km model. We lso determined tht our results were not likely to hve een ffected y missing exposure dt; including veterns with missing exposure dt in either the exposed group or the unexposed group hd no noticele effect on the study s results. Regrding endpoints, there re known limittions to self-reporting. First, there is the issue of reporting is. We note tht the rtes of self-reported illness we sw in this study re higher thn those in the NHS. 2 This my e ttriutle to the fct our study popultion differed in some respects from the NHS. We TABLE IX DOSE-RESPONSE RELATIONSHIP BETWEEN NUMBER OF DAYS OF EXPOSURE AND GENERAL HEALTH, CLINIC VISITS, HOSPITALIZATIONS, PTSD, AND CFS-LIKE ILLNESS Helth Sttus Rte (%) with Dys of Exposure 0 (n 3,336) 1 (n 1,740) 2 (n 158) Generl helth Excellent Very good Good Fir Poor Clinic visits None Hospitliztions None PTSD present CFS present Missing outcomes were excluded from denomintors when percentges were clculted. p, significnce proility y 2 test of independence etween numer of dys of exposure nd helth sttus indictor. p p TABLE X PERCENTAGE DISTRIBUTION OF GENERAL HEALTH, CLINIC VISITS, HOSPITALIZATIONS, PTSD, AND CFS-LIKE ILLNESS BY EXPOSURE STATUS ACCORDING TO THE MODEL OF 50-km RADIUS AROUND KHAMISIYAH Helth Sttus Exposed (n 199) included only deployed Army Gulf Wr veterns, nd our smple hd slightly greter proportions of femle veterns, non-cucsin veterns, enlisted personnel, nd Ntionl Gurd or Reserve personnel. Nonetheless, the rtes in our study resemle more closely those for deployed personnel thn the lower rtes seen for nondeployed personnel. 2 Second, there re limittions in the type of helth outcomes tht cn e scertined y questionnire. We note tht some of the long-term helth effects seen mong sujects with srin exposure t levels high enough to cuse cute cholinergic rections 15 were mnifested s norml test results (e.g., digitl symol test of psychomotor performnce nd encephlogrms during sleep) of unknown clinicl significnce, which might not hve een identified if they did not correlte strongly with specific questionnire items. Therefore, ecuse of the limittions imposed y self-report, the lck of differences in helth etween exposed nd unexposed sujects should not e over-interpreted. In summry, we conclude tht self-reported helth mong deployed Army Gulf Wr veterns 5 yers fter potentil exposure to low levels of chemicl gents t Khmisiyh, Irq, did not differ ccording to exposure sttus. These results were the sme regrdless of which of two exposure models were used nd were not ffected y missing exposure dt. Acknowledgments Unexposed (n 5,356) p Generl helth Excellent Very good Good Fir Poor Clinic visits None Hospitliztions None PTSD present CFS present See text for detils on exposure. Missing outcomes were excluded from denomintors when percentges were clculted. p, significnce proility y 2 test of independence etween exposure nd helth sttus indictor. We cknowledge the vlule help of memers of the expert pnel constituted to dvise us on the conduct of the study, s follows: Brr S. Hulk, MD, MPH (chir), University of North Crolin; Dn G. Blzer, MD, PhD, Duke University; Evelyn J. Bromet, PhD, Stte University of

10 944 Exposure nd Moridity New York, Stony Brook University; Germine Buck, PhD, Ntionl Institute of Child Helth nd Humn Development; Dniel H. Freemn, Jr., PhD, University of Texs Medicl Brnch; Richrd T. Johnson, MD, Johns Hopkins University; nd Peter S.J. Lees, PhD, Johns Hopkins University. We cknowledge Michel E. Kilptrick, MD, nd the stff of the Deployment Helth Support Directorte, Office of the Assistnt Secretry of Defense for Helth Affirs, for providing the exposure dt nd for useful discussions. We lso thnk Hrriet Crwford nd stff memers for ssistnce in dt collection. This study ws funded y Army grnt DAMD , nd the contrcting office technicl representtive ws COL Roert F. DeFrites, Army Surgeon Generl s Office. References 1. Office of the Specil Assistnt for Gulf Wr Illnesses: Cse nrrtive: U.S. Demolition Opertions t Khmisiyh, April 16, Aville t Kng HK, Mhn CM, Lee KY, Mgee CA, Murphy FM: Illnesses mong United Sttes veterns of the Gulf Wr: popultion-sed survey of 30,000 veterns. J Occup Environ Med 2000; 42: Bullmn TA, Mhn CM, Kng HK, Pge WF: Mortlity in U.S. Army Gulf Wr veterns exposed to 1991 Khmisiyh chemicl munitions destruction. Am J Pulic Helth 2005; 95: Pge WF, Mhn CM, Kng HK, Bullmn TA: Helth effects in Army Gulf Wr veterns possily exposed to chemicl munitions destruction t Khmisiyh, Irq: II: Moridity ssocited with notifiction of potentil exposure. Milit Med 2005; 170: Centers for Disese Control nd Prevention: Unexplined illness mong Persin Gulf Wr veterns in n Air Ntionl Gurd unit: preliminry report: August 1990 Mrch MMWR 1995; 44: Nicholson GL, Hymn E, Korenyi-Both A, et l: Progress on Persin Gulf Wr illnesses: relity nd hypotheses. Int J Occup Med Toxicol 1995; 4: Persin Gulf Veterns Coordinting Bord: Unexplined illnesses mong Desert Storm veterns: serch for cuses, tretment, nd coopertion. Arch Intern Med 1995; 155: Presidentil Advisory Committee on Gulf Wr Veterns Illnesses: Interim Report. Wshington, DC, U.S. Government Printing Office, Presidentil Advisory Committee on Gulf Wr Veterns Illnesses: Finl Report. Wshington, DC, U.S. Government Printing Office, Gry GC, Smith TC, Knoke JD, Heller JM: The postwr hospitliztion experience of Gulf Wr veterns possily exposed to chemicl munitions destruction t Khmisiyh, Irq. Am J Epidemiol 2000; 150: McCuley LA, Rischitelli G, Lmert WE, Lsrev M, Sticker DL, Spencer PS: Symptoms of Gulf Wr veterns possily exposed to orgnophosphte chemicl wrfre gents t Khmisiyh, Irq. Int J Occup Environ Helth 2001; 7: McCuley LA, Joos SK, Spencer PS, et l: Strtegies to ssess vlidity of self reported exposures during the Persin Gulf Wr. Environ Res 1999; 81: Office of the Specil Assistnt for Gulf Wr Illnesses: Cse nrrtive: U.S. Demolition Opertions t Khmisiyh, April 16, 2002: section III, nrrtive H: Deprtment of Defense nnounced possile chemicl wepons t Khmisiyh: Jnury Septemer Aville t khmisiyh iii s05.htm. 14. Ntionl Reserch Council, Bord on Environmentl Studies nd Toxicology, Committee on Toxicology, Sucommittee on Acute Exposure Guideline Levels: Acute Exposure Guideline Levels for Selected Airorne Chemicls, Vol 3, pp Wshington, DC, Ntionl Acdemy Press, Institute of Medicine, Division of Helth Promotion nd Disese Prevention, Committee on Helth Effects Associted with Exposures during the Gulf Wr: Gulf Wr nd Helth, Vol 1, Depleted Urnium, Pyridostigmine Bromide, Srin, Vccines. Wshington, DC, Institute of Medicine, Ntionl Acdemy Press, Pge WF: Long-term helth effects of exposure to srin nd other nticholinesterse chemicl wrfre gents. Milit Med 2003; 168: Blnchrd EB, Jones-Alexnder J, Buckley TC, Forneris CA: Psychometric properties of the PTSD Checklist (PCL). Behv Res Ther 1996; 34: Wethers FW, Litz BT, Hermn DS, Husk JA, Kene TM: The PTSD Checklist (PCL): reliility, vlidity, nd dignostic utility. Presented t the Ninth Annul Meeting of the Interntionl Society for Trumtic Stress Studies, Octoer 24 27, 1993, Sn Antonio, TX. 19. Fukud K, Strus SE, Hickie I, Shrpe MC, Doins JG, Komroff A: Interntionl Chronic Ftigue Syndrome Study Group: the chronic ftigue syndrome: comprehensive pproch nd its definition nd study. Ann Intern Med 1994; 121: Kng HK, Ntelson BH, Mhn CM, Lee KY, Murphy FM: Post-trumtic stress disorder nd chronic ftigue syndrome-like illness mong Gulf Wr veterns: popultion sed survey of 30,000 veterns. Am J Epidemiol 2003; 157: D Agostino RB Jr: Propensity score methods for is reduction in the comprison of tretment to non-rndomized control group. Stt Med 1998; 17: SAS Institute: SAS Online Doc, Version 8. Cry, NC, SAS Institute, Rhodes K: Gulf Wr Illnesses: Preliminry Assessment of Dodd Plume Modeling for U.S. Troops Exposure to Chemicl Agents. Report GAO T. Wshington, DC, U.S. Generl Accounting Office, 2003.