Chronic Multisymptom Illness Affecting Air Force Veterans of the Gulf War

Keiji Fukuda, MD, MPH; Rosane Nisenbaum, PhD; Geraldine Stewart, MA; William W. Thompson, PhD; Laura Robin, DO, MPH; Rita M. Washko, MD; Donald L. Noah, DVM, MPH; Drue H. Barrett, PhD, MS; Bonnie Randall, MCP; Barbara L. Herwaldt, MD, MPH; Alison C. Mawle, PhD; William C. Reeves, MD, MSPH

Context.—Gulf War (GW) veterans report nonspecific symptoms significantly was much lower than expected.9 No spe- more often than their nondeployed peers. However, no specific disorder has been cific illness is evident among the 18 598 identified, and the etiologic basis and clinical significance of their symptoms remain GW veterans, according to the Depart- unclear. ment of Defense Comprehensive Clini- 6 Objectives.—To organize symptoms reported by US Air Force GW veterans into cal Evaluation Program. Other investi- a case definition, to characterize clinical features, and to evaluate risk factors. gators have found no unusual increases in birth defects,10 unexplained illness,11 Design.—Cross-sectional population survey of individual characteristics and excess hospitalizations,12 or excess mor- symptoms and clinical evaluation (including a structured interview, the Medical tality13 among GW veterans. However, Outcomes Study Short Form 36, psychiatric screening, physical examination, clini- one study reported excess mortality due cal laboratory tests, and serologic assays for antibodies against viruses, rickettsia, to unintentional injuries rather than parasites, and bacteria) conducted in 1995. from disease.14 Evidence of an Iraqi Participants and Setting.—The cross-sectional questionnaire survey included chemical and biological weapons pro- 3723 currently active volunteers, irrespective of health status or GW participation, gram has been documented,15 and al- from 4 air force populations.The cross-sectional clinical evaluation included 158 though use of such weapons during the GW veterans from one unit, irrespective of health status. GW has not been confirmed,16 troops Main Outcome Measures.—Symptom-based case definition; case prevalence may have been exposed to chemical or rate for GW veterans and nondeployed personnel; clinical and laboratory findings biological warfare agents during the de- struction of storage bunkers. The long- among veterans who met the case definition. term effects of such chemical exposures Results.—We defined a case as having 1 or more chronic symptoms from at are uncertain, although some investiga- least 2 of 3 categories (fatigue, mood-cognition, and musculoskeletal). The preva- tors have suggested that wartime expo- lence of mild-to-moderate and severe cases was 39% and 6%, respectively, among sures may have contributed to chronic 1155 GW veterans compared with 14% and 0.7% among 2520 nondeployed per- neurotoxic syndromes.17,18 sonnel. Illness was not associated with time or place of deployment or with duties during the war. Fifty-nine clinically evaluated GW veterans (37%) were noncases, For editorial comment see p 1010. 86 (54%) mild-to-moderate cases, and 13 (8%) severe cases. Although no physical examination, laboratory, or serologic findings identified cases, veterans who met the case definition had significantly diminished functioning and well-being. In December 1994, the US Secretary Conclusions.—Among currently active members of 4 Air Force populations, a of Defense and the Secretary of Veter- ans Affairs, and the Commonwealth of chronic multisymptom condition was significantly associated with deployment to the Pennsylvania asked the Centers for Dis- GW. The condition was not associated with specific GW exposures and also ease Control and Prevention (CDC), At- affected nondeployed personnel. lanta, Ga, to investigate a “mystery ill- JAMA. 1998;280:981-988 ness” reported among GW veterans from an Air National Guard (ANG) unit SOON AFTER cessation of Gulf War to concerns about a “Gulf War syn- in Lebanon, Pa.19 In the first phase of the (GW) hostilities, anecdotal reports of ill- drome.”1-3 Subsequent studies have investigation,weinterviewedandexam- ness and speculation over environmen- shown that GW veterans report numer- inedillGWveteransandfoundthattheir tal, biological, and chemical hazards led ous chronic nonspecific symptoms, such major health problems consisted of per- asfatigue,neurocognitiveproblems,and sistent fatigue and other chronic symp- From the Divisions of Viral and Rickettsial Diseases musculoskeletal pain, significantly more toms that began during deployment or (Drs Fukuda, Noah, Mawle, and Reeves) and Parasitic oftenthantheirnondeployedpeers.4,5 No soon after returning from the GW, but Diseases (Dr Herwaldt), National Center for Infectious widespread disorder has been linked to we did not find associated physical signs Diseases, the Division of Environmental Hazards and 4 Health Effects, National Center for Environmental that conflict, and the etiologic basis and or clinical laboratory abnormalities. Pa- Health (Dr Barrett), and the Epidemic Intelligence Ser- clinical significance of veterans’ symp- tients’ illnesses resembled those re- vice, Epidemiology Program Office (Drs Robin and tomatology remain unclear.6 ported previously, and the classification Washko), Centers for Disease Control and Prevention, 2 Atlanta, Ga; Klemm Analysis Group, Atlanta, Ga (Dr The evidence for unique health prob- and etiology remain unclear. Nisenbaum); and Abt Associates Inc, Cambridge, lems among GW veterans is mixed. Al- This article reports the second phase Mass (Mss Stewart and Randall and Dr Thompson). though acute gastroenteritis was fre- of our survey of the index ANG unit and Reprints: William C. Reeves, MD, MSPH, Centers for 7,8 Disease Control and Prevention, Mail Stop A-15, 1600 quently reported among troops, the 3 comparison air force populations. Our Clifton Rd, Atlanta, GA 30333 (e-mail: [email protected]). overall occurrence of infectious diseases primary objectives were to determine

Downloaded From: by a Harvard University User on 03/12/2018 Table 1.—Factor Analysis Loadings (ϫ 100) consecutive unit training weekends factors. After rotation, the factor pattern Exploratory Sample and Confirmatory Sample* (held monthly) to maximize participa- was examined to determine whether an Loading Values tion rates. Unit C was surveyed during oblique rotation yielded high loadings in a 10-day period. only one factor. Symptoms with factor Factor Factor On each base, one of us (K.F., L.R., loadings of greater than 0.40 were kept Symptoms PC 1 1PC22PC3R.W., or D.N.) met with small groups, fortheconfirmatoryfactoranalysis.22 The Depression 76 71 −3 −17 −1 explained the study, and distributed exploratory analysis yielded 10 compo- Anxiety 73 73 −5 −20 1 questionnaires to subject volunteers nents with an eigenvalue greater than 1. Moodiness 71 62 2 0 −10 whose identities remained anonymous. Memory 59 61 6 6 0 The first (feeling depressed, feeling anx- problems The questionnaire took 15 to 45 minutes ious, feeling moody, difficulty remember- Fatigue 50 50 19 19 3 to complete and queried health status, ing or concentrating, trouble finding Difficulty with 48 61 −15 −9 4 demographic and military characteris- words,difficultysleeping,andfatigue)ac- words tics, and potential risk factors for illness. counted for 16.8% of the total common Difficulty 45 42 30 14 −2 We asked about the 35 symptoms that variance.Thesecond(jointstiffness,joint sleeping hadbeenidentifiedduringtheearlierex- pain, and muscle pain) accounted for Joint stiffness 2 −5 85 76 4 ploratory study4 including their sever- 11.9% of the variance. The third (wheez- Joint pain 3 −13 84 85 −1 ity (mild, moderate, or severe) and du- ing, shortness of breath, coughing, and Muscle pain 0 12 63 47 1 ration (Ͻ6 months or Ն6 months). chest pain) accounted for 10.4%. The re- Wheezing −14 19 0 8 75 maining components each contributed Shortness of 10 35 3 21 73 Case Definition breath less than 10% of the total variance. Cough −8 29 2 13 59 We used 2 conceptually different ap- Confirmatory Factor Analysis.— Chest pain 11 33 0 17 53 proaches to develop exploratory case Only symptoms from the first 3 princi- definitions from the symptom data. The pal analysis components were used in the *The exploratory sample represents the first 3 prin- 23 cipal components of 35 symptoms submitted to a pro- first exploratory case definition was de- confirmatory factor analysis. These max oblique rotation (n = 1631). The confirmation factor rived from clinical and epidemiological symptoms were submitted to an un- analysis represents symptoms from the first 3 principal analysis components that were submitted to an un- reasoning; the second used statistical weighted least squares factor analysis 24 weighted, least squares factor analysis with a Pro- methods. with a Procrustes oblique rotation. Con- crustes oblique rotation (n = 1624). Intercorrelations for Clinical Approach.—We reasoned firmatory factor analysis identified 2 fac- factors 1 and 2 were 42% and 69%, respectively, for exploratory and confirmatory samples. PC indicates that if an illness existed among GW vet- tors (mood-cognition-fatigue and mus- principal components. erans, case-defining symptoms would be culoskeletal) (Table 1). chronic,prominentamongGWveterans, Preliminary Working Case Defini- whether the prevalence of symptoms and more common among GW veterans tions.—The 10 symptoms from factor was increased among currently active than nondeployed personnel. Thus, we analysis included all 6 symptoms identi- members of the index unit and among decided that symptoms for the clinical fied by the clinical approach. We used GW veterans from other air force popu- case definition had to be reported for 6 these 10 symptoms to derive 2 possible lations in comparison with nondeployed months or longer, by at least 25% of GW working case definitions. First, we de- personnel and to develop a case defini- veterans and by GW veterans at least rived a case definition on the basis of tion that would allow further studies of 2.5 times more often than by nonde- factor scores. The confirmatory phase fac- etiology. We also report the third phase, ployed personnel. The clinical case defi- tor analysis model was fit to the partici- whichwasintendedtocharacterizeclini- nitionwasidentifiedasfatigue,difficulty pants’ symptom data and a total factor cal and laboratory findings among GW remembering or concentrating, moodi- score was calculated for each participant veterans who met our case definition. ness, difficulty sleeping, joint pain, and by adding the scores of the factors. A case METHODS joint stiffness. was defined as having a combined factor Statistical Approach.—In the statis- score in the top 25th percentile. Cross-sectional tical approach, we randomly divided the Second, we developed an alternative Questionnaire Survey 3255 participants into 2 subsamples of working case definition based on symp- Study Population and Survey Proce- 1631 and 1624 subjects (446 were ex- tom categories. This grouped the symp- dures.—We used study protocols that cluded from analysis because of incom- tomsinto3categories:fatigue,mood-cog- were approved by the Human Subjects plete symptom data). We conducted an nition, and musculoskeletal. We sepa- Committee of the CDC and followed the exploratoryprincipalcomponentsanaly- rated chronic fatigue even though it did human experimentation guidelines is- sis on the first subsample and a confir- not load as a separate factor because of sued by the US Department of Health matory factor analysis on the second. the central role of chronic fatigue in vir- and Human Services in this study. All Exploratory Principal Components tually all previous studies of GW veter- study participants provided informed Analysis.—All 35 symptoms were sub- ans.Wedefinedacaseashaving1ormore consent. mitted to a principal components analy- symptoms from 2 or more symptom cat- We surveyed the index and 3 compari- sis, an exploratory method that can help egories. son air force units between January and determine the probable number and na- Comparison of the Preliminary March 1995. Unit A, another ANG unit ture of common components (or factors) Working Case Definitions.—We com- in Pennsylvania, had similar demo- .20 We used a promax rotation because it pared the case definitions by determin- graphic features as the index unit but is a well-formulated procedure to extract ing the prevalence of each type of case had a different primary mission. Unit B correlated factors21 and provides inter- among GW veterans and nondeployed (US Air Force Reserve) and unit C (ac- factor correlations. The Kaiser-Guttman personnel and calculated their overall tive-duty air force) had similar primary rule (an eigenvalue Ͼ1) was used to de- agreementbya␬statistic.25 Forty-seven missions as the index unit’s mission but termine the number of factors to retain percent of GW veterans and 15% of non- were located in Florida. Any unit mem- forrotation.Inaddition,weexaminedthe deployed were classified as factor score ber who was on base was eligible for scree plot and verified that the slope con- cases. Of these, 25% had symptoms that the study. The index unit and units A necting the eigenvalues approached zero included fatigue, mood-cognition, and and B were each surveyed during 3 after retaining the number of selected musculoskeletal pain; 39.2% fatigue and

Downloaded From: by a Harvard University User on 03/12/2018 mood-cognition; 15.6% fatigue and mus- definition was used to classify cases of To test for other agents, serum culoskeletal pain; 6.6% mood-cognition CFS.29 A trained interviewer privately samples were tested for antibodies to and musculoskeletal pain. An additional administered selected modules on soma- Helicobacter pylori at the Division of 9.9% reported only mood-cognition symp- tization disorder, major depression, and Bacterial and Mycotic Diseases, CDC. toms, and 8.1% reported only musculo- panic disorder from the Diagnostic In- To screen for exposure (either by vacci- skeletal pain. Forty-five percent of GW terview Schedule version of Diagnostic nation or in combat) to 2 widely dis- veterans and 15% of nondeployed were and Statistical Manual of Mental Dis- cussed putative biologic warfare agents, classified as symptom-category cases. Of orders, Fourth Edition,(DSM-IV)30 The we tested serum samples for antibodies these, 41.8% reported symptoms of fa- Diagnostic Interview Schedule was to toxin produced by Clostridium botu- tigue, mood-cognition, and musculoskel- modified, so the onset of disorders could linum and Bacillus anthracis. Serum etal pain; 26.7% reported fatigue and be referenced to periods before, during, samples were screened at the Division of mood-cognition; 8.4% reported fatigue and after deployment to the GW. Each Bacterial and Mycotic Diseases, CDC, and musculoskeletal pain; and 23.1% re- subject next received a screening physi- for antibodies to type A botulinum toxin. ported mood-cognition and musculoskel- cal examination performed by a physi- Serum samples were assayed at the US etal pain. The ␬ statistic was 0.79, which cianassistant.Thisexaminationsystem- Army Medical Research Institute of In- indicated substantial overall agreement atically covered vital signs, height, fectious Diseases, Washington, DC, for between the working case definitions. weight, skin, ears (including gross hear- antibodies against anthrax protective Chronic Multisymptom Illness.— ing), eyes (including extraocular eye antigen and lethal factor. Since both case definitions were compa- movements and retina), nasal passages, Stool Specimens.—Fresh fecal speci- rable, we chose the symptom-category throat, neck (including thyroid), lungs, mens were examined for red and white approach because it was easier to apply heart, abdomen, lymph nodes (cervical, blood cells and were cultured for Salmo- in a clinical setting. We defined a case as axillary, and inguinal areas), rectal vault nella, Shigella, Yersinia, Campylobac- having 1 or more chronic symptoms (including testing of stool for occult ter, and Escherichia coli 0157:H7. Fecal (present for Ն6 months) from at least blood), reflexes, and tests of coordina- specimens were examined at the Divi- 2 of the following categories: fatigue; tion (finger to nose), strength, and sen- sion of Parasitic Diseases, CDC, for ova mood and cognition (symptoms of feel- sation (light touch, vibration, and pin and parasites,35,36 including but not lim- ing depressed, difficulty remembering prick). The personnel who performed ited to Cryptosporidium parvum, Cy- or concentrating, feeling moody, feeling these evaluations were not aware of the clospora cayetanensis, Isospora belli, anxious, trouble finding words, or diffi- subjects’ case status. and microsporidia.37 Fecal specimens culty sleeping); and musculoskeletal Case Definition.—We classified par- also were screened for enteroviruses at (symptoms of joint pain, joint stiffness, ticipants in the clinical study as cases or Southern General Hospital, Glasgow, or muscle pain). We subclassified a case noncases based on their responses to the Scotland.36 as severe if each case-defining symptom mailed clinical epidemiological question- was rated as severe; otherwise, we con- naire and the case definition derived in Statistical Analyses sidered the case to be mild-to-moderate. the cross-sectional survey. Weestimatedtheassociationbetween Blood and Urine Specimen Tests.— symptomsanddeploymenttotheGWby Cross-sectional Clinical Evaluation Blood specimen evaluation included a prevalence ratios. We assessed univari- Study Population and Enrollment complete blood cell count, erythrocyte ate associations of categorical variables Procedures.—We conducted a third- sedimentation rate, serum chemistry val- by odds ratios (OR) and 95% confidence phase, cross-sectional clinical study dur- ues (alanine aminotransferase, alkaline intervals (95% CI). We used the ␹2, ing April and May 1995. Subjects were phosphatase, total protein, albumin, Fisherexact,andMantel-Haenszeltests GW veterans currently in the index unit. globulin, calcium, phosphorus, glucose, to calculate P values. We used the Stu- During the preceding cross-sectional electrolytes, blood urea nitrogen, and cre- dent t test and 1-way, between-subjects survey,weinformedthe667participants atinine), and thyroid-stimulating hor- analysis of variance (ANOVA) to com- from the index unit of our intent to con- mone level. Urine specimens under- pare means of continuous variables and duct a follow-up clinical study and asked went routine urinalysis. compared pairs of means by the Scheffe´ for volunteers. We also recruited par- Serologic Testing.—To test for arbo- test. We used the PROC FACTOR pro- ticipants by posting notices throughout viruses and rickettsia, serum samples cedure to perform principal components the base. Our solicitations stressed the were tested for antibodies to yellow fe- and factor analyses. In the factor analy- need for both symptomatic and non- ver, dengue, Sindbis, West Nile, phlebo- sis, the proportion of total common vari- symptomatic GW veterans. tomus fever viruses (Naples and Sicil- ance explained by each factor was deter- Clinical Evaluation.—Each volun- ian), Toscana, Karimbad, and Isfahan mined by the direct contribution teer was mailed an informed consent viruses at the Division of Vector-Borne method.23 The PROC CATMOD proce- form, a clinical questionnaire, and the Infectious Diseases, CDC.31 Serologic dure was used to fit a polytomous logis- Medical Outcomes Study Short Form 36 testing for antibodies to Rickettsia tic regression model and simultaneously (MOSSF-36)26,27 tocompleteathome.The typhi, Rickettsia rickettsii, Coxiella assess associations between both mild- clinical questionnaire asked about the burnetii, and Ehrlichia chaffeensis was to-moderate and severe cases and pos- presence, duration, and intensity of 35 performed at the Division of Viral and sible risk factors. All calculations were symptoms and included questions about Rickettsial Diseases, CDC.32 performed using Statistical Analysis fatigue, activity levels, and diarrhea. To test for parasites, serum samples Systems (SAS) software.39 All statisti- Each subject was individually evalu- were tested at the Division of Parasitic cal tests were 2-tailed. ated by a CDC field team at the Veter- Diseases, CDC for reactivity to Leish- ans Affairs (VA) Medical Center in mania donovani and Leishmania tropica RESULTS Lebanon, Pa. Subjects completed the promastigote antigens,33 and also were Mississippi Post Traumatic Stress Dis- tested for antibodies to Toxoplasma gon- Cross-sectional order Scale (modified for use among GW dii, Schistosoma mansoni, Schisto- Questionnaire Survey veterans).28 The 1994 revised chronic fa- soma haematobium,34 and Strongyloi- Survey Participation Rates.—Ques- tigue syndrome (CFS) working case des stercoralis. tionnaires were completed by 4036 per-

Downloaded From: by a Harvard University User on 03/12/2018 Table 2.—Demographic, Military, and Health Characteristics of All Subjects Clinical Evaluation Study

% Participation Rate and Classifica- Index Unit A Unit B Unit C Combined tion of Cases.—Among the index unit’s Characteristics (n = 667) (n = 538) (n = 838) (n = 1680) (n = 3723) 1083 members, 490 (45%) were GW vet- Age, median, y 39 38 38 31 34 eransand173ofthose(35%)volunteered Male 85 85 82 88 86 to participate in the clinical evaluation White 94 93 85 81 86 study. Fifteen (8.7%) provided incom- Married 72 68 73 66 69 plete symptom data and were excluded Less than college education 36 26 18 35 30 from further analysis. Of the remaining Employed full-time 89 85 84 92 89 158 veterans, 13 (8%) were severe cases, Deployed to Persian Gulf 47 22 32 28 31 Current enlisted rank 84 85 85 88 86 86 (54%) were mild-to-moderate cases, Years on active duty and 59 (37%) were noncases. Ͻ1 17 17 10 0.1 8 General Characteristics and Cur- 1-10 78 78 78 59 69 rent Symptoms.—Most participants Ն11 5 5 11 42 23 were men, but 4 of the severe cases (31%) Years in guard or reserves occurred among women. The preva- Ͻ1 0.5 0.2 0.4 94 42 lence of all chronic symptoms (including 1-10 45 48 56 5 30 those not used to define cases) was high- Ն11 54 52 44 1 28 est among severe cases, followed by mild- Listed on Veterans Affairs/ 15 1 3 1 4 to-moderate cases, and noncases (data not Department of Defense registry shown). Diarrhea (Ն3 loose stools per Self-assessed health day) was reported by 3% (n = 2) of non- Excellent 27 29 28 22 25 cases, by 26% (n=22) of mild-to- Very good 43 43 43 40 42 moderate cases (OR, 10; 95% CI, 2.2- Good 25 25 25 33 29 44.2), and by 77% (n=10) of severe cases Fair 4 2 4 5 4 (OR, 95; 95% CI, 14.1-642.3). All but 3 sub- Poor 0.4 0.4 0.4 0.5 0.5 jects who reported diarrhea indicated it Current smoker 20 20 22 25 23 had been present for at least 6 months, Average No. of alcoholic drinks per week and all but 2 dated onset from the period None 37 35 42 34 37 during or immediately following their re- 1-3 37 39 36 36 36 turn from the Gulf region. Only 2 sub- Ն42626223027jects with diarrhea also reported weight loss. Medical History Prior to Deploy- sons. We excluded 313 subjects; 109 were reported more often by GW veter- ment.—Severe cases reported histories were not members of the units, and 204 ans from the index unit than those from of depression before deployment signifi- were younger than 17 years during the comparison units. Chronic diarrhea was cantly more often (15%) than noncases GW. Among 3723 remaining partici- reported substantially more often by (0%) (Fisher exact test PϽ.05) and also pants, 1163 (31.2%) were GW veterans GW veterans from the index unit. The were associated with self-reported si- and 2560 (68.8%) had not been deployed. prevalence of symptoms among nonde- nusitis prior to first deployment more After exclusions, unit participation ployed personnel was similar among all frequently (38%) than noncases (10%) rates were 62% (667/1083) for the index units (data not shown). (OR, 5.8; 95% CI, 1.4-24.5). Otherwise, unit, 35% (538/1520) for unit A, 73% (838/ Occurrence of Chronic Multisymp- there were no statistically significant 1141) for unit B, and 70% (1680/2407) for tomIllnessCases.—Inall,3675subjects differences referable to the year before unit C. provided complete data on the 10 case- deployment to the GW between severe Demographic, Military, and Health definingsymptoms.Amongthe1155GW or mild-to-moderate cases and noncases Characteristics.—The median age of veterans, 6% were classified as severe in terms of 11 allergy symptoms; 7 self- participants was 34 years (mean, 35 casesand39%asmild-to-moderatecases reported surgical, dental, or invasive years; range, 21-60 years); 86% were compared with 0.7% and 14%, respec- medical procedures (including blood or men; 86% were white; and 89% were em- tively, among the 2520 nondeployed per- blood product transfusions and mercury ployed full-time. Other demographic, sonnel. Deployment to the GW was the amalgam repair of dental cavities); or military, and general health character- most important risk factor for severe self-reportedlifetimeprevalence,before istics were similar in the 4 units (Table and mild-to-moderate illness (Table 4). deployment to the Persian Gulf, of 35 2),exceptthatactive-dutysubjects(unit Multivariate analyses (Table 5) showed medical and psychiatric conditions (in- C) were younger, less likely to be white, that severe cases were associated with cludingheartdisease,hypertension,dia- and less likely to describe their current GW service, enlisted rank, female sex, betes, alcohol or substance abuse, an- health as “excellent.” Demographic char- and smoking. Mild-to-moderate cases orexia or bulimia, migraine or severe acteristics of survey participants were wereassociatedwiththesamevariables, headaches, anxiety, diarrhea, irritable similar to overall unit demographics. and increasing age and current member- bowel syndrome, and impotence). Prevalence of Symptoms.—Overall, ship in the index unit. Illness was not PhysicalExamination.—Physicalex- 3701participants(99%)reportedatleast associated with the number of deploy- aminations were most notable for the 1 symptom as a current health problem. ments, month or season of deployment, generalpaucityofabnormalfindings.Se- Gulf War veterans reported all symp- duration of deployment, military occu- vere cases were associated with slightly toms, except hay fever and other al- pationalspecialty,directparticipationin higher body mass indices, a measure of lergies, significantly more often than combat, or self-reported locality in the weight in kilograms divided by the nondeployed (Table 3). All symptoms, Gulf region (most were in the Rhiyadh square of height in meters (BMI, 29.6), except headache and difficulty sleeping, area of Saudi Arabia). than noncases (BMI, 26.5) (PϽ.01). Rash

Downloaded From: by a Harvard University User on 03/12/2018 was present more often among mild- Table 3.—Prevalence (percent) of 35 Current and Long-term Symptoms Among Gulf War Veterans to-moderate (14%) and severe (15%) (n = 1163) and Nondeployed Military Personnel (n = 2538) chronicmultisymptomcasesthanamong Symptoms Lasting noncases (3%), but the differences were Current Symptoms More Than 6 mo not statistically significant. Several mi- Nondeployed Nondeployed nor abnormalities were found on lymph Symptoms Veterans Military Personnel Veterans Military Personnel node, liver, spleen, or neurologic exami- Sinus congestion 52 39 47 32 nation, but none was associated with Headache 50 41 45 32 cases. A summary of physical examina- Fatigue 43 17 41 13 tion data is available from the authors. Joint pain 36 13 33 11 FunctionalStatusandWell-being.— Difficulty remembering or concentrating 34 9 32 8 As measured by the MOS SF-36, severe Joint sitffness 30 11 28 9 and mild-to-moderate cases were asso- Difficulty sleeping 28 13 25 9 ciatedwithasignificantdecreaseinfunc- Gas, bloating, cramps, or abdominal pain 27 15 25 11 tioning and well-being. Severe cases Trouble ﬁnding words 26 9 24 8 were associated with a decrease on all of Moody or irritable 25 9 25 7 the 6 subscales (Table 6). Rash or sores 23 7 12 5 BloodandUrineTesting.—Themean Numbness or tingling 21 8 19 5 values of a few routine blood tests dif- Muscle pain 20 8 18 6 fered among cases and noncases, but the Hay fever or allergies* 19 19 18 17 differences were marginal and clinically Feeling depressed 18 10 16 8 unimportant. None of the urinalysis re- Diarrhea (Ն3 loose stool samples per 24 h) 18 4 16 3 sults differed between cases and non- Sore throat 17 10 15 6 cases. A summary of blood and urine Cough 17 11 14 6 data is available from the authors. Feeling anxious 17 7 15 5 Stool Specimen Testing.—No fecal Unintended weight gain Ն10 lb 17 8 15 6 specimen tested positive for occult Shortness or breath 16 6 14 5 blood, white blood cells, Salmonella, Chest pain 15 7 13 5 Shigella, Campylobacter, Yersinia, en- Decreased interest in sex 14 6 13 4 teropathogenic Escherichia coli, micro- Dizziness or trouble maintaining balance 14 4 12 2 sporidia,Cryptosporidiumparvum,Cy- Night sweats that soak bed sheets 13 4 12 2 closporacayetanensis,Isosporabelli,or Fatigue lasting 24 h after exertion 13 2 12 2 Entamoeba histolytica. Blastocystis Nasal sores 11 6 9 4 hominis was found in stool specimens Swollen lymph nodes 10 4 8 2 Milk intolerance 7574 from 7 noncases (12%) and from 6 mild- Episodes of disorientation 7151 to-moderate and chronic (7%) multi- Nausea or vomiting 6261 symptom cases. Giardia lamblia was Wheezing 6352 found in stool specimens from 1 noncase Chemical sensitivity 5252 (2%) and 1 mild-to-moderate case (1%). Fever 5241 Enteroviruses were found in stool speci- Unintended weight loss Ն10lb3121 mens from 6 noncases (10%) and 8 mild- to-moderate cases (9%). A summary of *Except hay fever or other allergies, all current and chronic symptoms were reported signiﬁcantly (PϽ.05) more stool specimen data is available from the often by Gulf War veterans than nondeployed military personnel. authors. Serologic Testing.—There was no as- rates were similar between cases and ment to the GW, and 19 also reported sociation between seropositivity to vari- noncases. Sixteen subjects had antibod- depression within the last year (1994- ous infectious agents and chronic multi- ies to dengue virus and rates were simi- 1995). Overall, major depression was symptom cases. Troops are routinely lar in cases and noncases. Similar, low, more common among GW veterans after vaccinated against yellow fever virus, and equally distributed (among noncase deployment than before deployment. and 83% of subjects had antibodies and cases) seroreactivity existed to The prevalence of current depression against this agent, but there were no dif- Coxiella burnetii (7 positive), R rickett- was significantly higher among severe ferences between cases and noncases. sii(6positive),Echaffeensis(5positive), chronic multisymptom cases (54%) and Ten subjects reacted to botulina toxin and R typhi (1 positive). No subject had mild-to-moderate cases (13%) than among and 14 to anthrax protective antigen, antibodies to Sindbis, West Nile, Tos- noncases (2%). but there were no differences between cana, Karimbad, or Isfahan viruses, Somatization and Panic Disor- cases and noncases. Eight participants Schistosoma species, or S stercoralis.A ders.—Foursubjects(allclassifiedasse- (4.9%) seroreacted to leishmanial anti- summary of serologic data is available vere chronic multisymptom cases) met gens. However, no severe case was se- from the authors. DSM-III-Rcriteriaforsomatizationdis- roreactive; reactivity rates were higher order. Three met criteria for panic dis- among noncases than mild-to-moderate Syndromic Disorders order (2 were mild-to-moderate cases cases; the difference was not significant; PosttraumaticStressDisorder.—One and 1 was a severe case). Panic disorder and titers were low (range, 1:16-1:64). subject screened positive for posttrau- was uncommon in the periods before and Nine subjects were seropositive to phle- matic stress disorder also was classified during deployment to the GW. Four of botomus fever viruses, but rates were as a severe chronic multisymptom case. the subjects met criteria for panic disor- similarlylowamongnoncases(2%),mild- Depression.—Twenty-two subjects der during their deployment to the war, to-moderate cases (9%), and severe met the Diagnostic Interview Schedule including 1 classified a as mild-to-mod- cases (8%). Thirty persons (19%) were of the DSM IV criteria for major depres- erate case and 1 classified as a severe seroreactive to Toxoplasma gondii, and sion occurring after their last deploy- case.

Downloaded From: by a Harvard University User on 03/12/2018 Table 4.—Univariate Associations With Mild-to-Moderate and Severe Cases Table 5.—Multivariate Associations With Mild-to- Moderate and Severe Cases Mild-to-Moderate Case Severe Case Noncase (n = 803) (n = 86) Mild-to- (n = 2786), Moderate Case Severe Case Variables No. No. OR (95% CI)* No. OR (95% CI) Gulf War veteran 638 449 4.3 (3.6-5.0) 68 12.7 (7.5-21.5) Variables OR (95% CI)* OR (95% CI) Enlisted rank Gulf War 4.08 (3.39-4.93) 16.18 (8.99-29.14) August 1990-July 1991 1930 663 1.5 (1.2-2.0) 82 15.5 (2.2-111.8) veteran Current enlisted rank 2350 701 1.3 (1.1-1.7) 85 14.9 (2.1-107.1) Current 1.71 (1.29-2.26) 17.38 (2.36-127.89) Age, y enlisted rank Ͻ30 879 153 1.0 17 1.0 Female 1.64 (1.27-2.12) 3.13 (1.65-5.94) 30-39 1109 296 1.5 (1.2-1.9) 39 1.8 (1.1-3.2) Age, y 40-49 535 254 2.7 (2.2-3.4) 19 1.8 (0.9-3.6) Ͻ30 1.00 1.00 Ն50 191 84 2.5 (1.9-3.4) 9 2.4 (1.1-5.6) 30-39 1.24 (0.98-1.57) 1.32 (0.72-2.44) Ն30 Combined 1835 634 2.0 (1.6-2.4) 67 1.9 (1.1-3.2) 40-49 2.14 (1.65-2.78) 1.47 (0.71-3.04) Current smoker 580 213 1.4 (1.1-1.6) 32 2.2 (1.4-3.5) Ն50 1.59 (1.11-2.27) 1.63 (0.65-4.08) Index unit member Current 1.54 (1.24-1.91) 1.49 (0.86-2.57) August 1990-July 1991 272 196 2.6 (2.1-3.2) 19 2.2 (1.3-3.8) index unit Current index unit member 414 222 2.2 (1.8-2.6) 23 2.1 (1.3-3.4) member Other deployment 567 236 1.6 (1.4-2.0) 27 1.8 (1.1-2.9) Current 1.27 (1.03-1.55) 1.72 (1.06-2.77) Less than college education 795 270 1.3 (1.1-1.5) 34 1.6 (1.1-2.5) smoker Female 396 112 1.0 (0.8-1.2) 17 1.5 (0.9-2.5) *OR indicates odds ratio; CI, conﬁdence interval. White 2339 713 1.5 (1.2-2.0) 69 0.7 (04-1.2) Married 1870 597 1.4 (1.2-1.7) 60 0.1 (0.7-1.8)

*OR indicates odds ratio; CI, conﬁdence interval. Our case definition represents a unique approach toward organizing Chronic Fatigue Syndrome.—Eight Factor analysis also uses rigorous em- symptom data. We intended it to pro- subjects met all criteria for CFS.29 Of piricalstatisticalmethodstoidentifyfac- videasummarymeasureofillnesstotest these, 7 (54%) also were classified as se- tors and associated factor scores. Thus, for associations with clinical abnormali- vere cases, and 1 (1%) was a mild-to- we used information from factor analy- ties and risk factors and not as a defini- moderate case. sis to derive the final case definition. We tivelabelforasingle,distinctillness.The first derived a factor score case defini- resulting case definition needs to be rep- COMMENT tion that included as cases those sub- licated by other studies of GW veterans. Our study of currently active mem- jects whose total factor scores were in However, in a civilian population, factor bers from 4 air force populations showed the upper 25th percentile of the study analysis methods identified fatigue, a substantially higher prevalence of population. We derived a second case mood, and cognition as the most impor- symptoms among GW veterans than definition by grouping the symptoms tant factors among symptoms similar to nondeployed. This result was similar to identified by factor analysis into 3 of the the ones in the current study.42 findings of an earlier and unrelated epi- following categories: fatigue, mood-cog- The key observation of this study was demiological study of Pennsylvanian nition, and musculoskeletal pain. We that air force GW veterans were signifi- and Hawaiian ANG personnel,40 a popu- separated chronic fatigue, even though cantly more likely to meet criteria for se- lation-based study from all branches of it did not load as a separate factor be- vere and mild-to-moderate illness than service,5 and is consistent with clinical cause of the central role of chronic fa- were nondeployed personnel. There was data from GW veteran registries.6 Thus, tigue in all studies of GW veterans. no association between the chronic mul- air force GW veterans likely experience The ␬ statistic measuring agreement tisymptom illness we defined and risk fac- an illness similar to that reported from between the 2 preliminary case defini- tors specific to combat in the GW (month other branches of the service. tions was 0.79, which represents sub- or season of deployment, duration of de- The number and diversity of symp- stantial agreement.41 Disagreement be- ployment, duties in the GW, direct par- toms reported by our study subjects and tweenthe2definitionsoccurredbecause ticipation in combat, or locality of GW by GW veterans in other studies make factor score–defined cases could have service). Our finding that 15% of nonde- analytic epidemiological study difficult. symptoms from only 1 factor and symp- ployed also met illness criteria was Therefore, a major objective was to de- tom score–cases could have symptoms equally important and suggests that the velop a case definition that captured the from2categoriesbutnotexceedthe25th multisymptom illness we observed in this chronic multisymptomatic nature of ill- percentile of factor scores. We chose the population is not unique to GW service. ness in GW veterans. We used 2 inde- symptom category approach for the fi- To characterize the clinical features of pendent methods to identify case-defin- nal working case definition because it GW veterans with this chronic multi- ing symptoms. An exploratory clinical generally identified the same cases as symptom illness, we evaluated veterans approach showed that veterans had a the factor score approach, better cap- from one unit. Neither mild-to-moder- condition defined by fatigue, neurocog- tured the multisymptom nature of ill- ate nor severe cases were associated nitivesymptoms,andjointpainandstiff- ness, was clinically more understand- with clinically significant physical ex- ness. Factor analysis identified 2 major able, and allowed us to subclassify cases amination or routine laboratory test factors that included and extended according to severity. To further evalu- abnormalities. However, GW veterans symptoms identified by the clinical ap- ate the validity of defining case subjects classified as having mild-to-moderate proach. based on symptom categories rather and severe illness had a significant de- Besides extending the number of cor- thanfactorscores,weexaminedunivari- crease in functioning and well-being related symptoms, factor analysis used ate and multivariate associations (vari- compared with noncases, as measured data collected from all participants and ables from Tables 4 and 5) with cases by the MOS SF-36. did not bias our study toward finding a defined by both methods and found no We tested participants for exposure higher case prevalence among veterans. meaningful differences. to several infectious agents that are im-

Downloaded From: by a Harvard University User on 03/12/2018 portant health problems in the Gulf re- Table 6.—Functional Disability Among Long-term Multisymptom Cases and Noncases as Measured by the gion, that may have been used in vac- Medical Outcomes Study Short Form 36 (MOS SF 36)* cines, and that might be associated with Noncase Mild-to-Moderate Severe Case Student t Test a chronic illness. Leishmaniasis is en- MOS SF 36 Subscales (n = 59) Case (n = 86) (n = 13) P Value demic to the region, and a syndrome General mental health 36.4 (3.7) 32.1 (6.3) 28.0 (8.8) .001 termed viscerotropic leishmaniasis has Vitality 17.8 (3.9) 13.4 (3.6) 8.5 (3.4) .001 been described in GW veterans.43,44 Ap- Physical functioning 29.5 (0.9) 27.9 (2.4) 24.5 (3.4) .001 proximately 5% of participants serore- General physical 30.8 (3.0) 25.1 (4.8) 17.0 (3.7) .001 acted to leishmanial antigens, but the ti- Physical limit 7.9 (0.5) 6.7 (1.4) 5.5 (1.6) .001 ters were low and seroreactivity rates Emotional limit 5.8 (0.6) 5.4 (1.0) 4.8 (1.4) Ͻ.001 were not significantly different between *Values followed by numbers in parentheses are the mean (SD). cases and noncases. No study subject had clinical manifestations of visceral or cutaneous leishmaniasis. However, different strategy first to document the These limitations notwithstanding, the manifestations of viscerotropic presence, magnitude, and duration of our results indicate that a substantial leishmaniasis can be nonspecific.43,44 The various symptoms reported by veterans proportion of currently active air force serologic assay we used is a sensitive di- and nondeployed personnel and then to GW veterans have a chronic multisymp- agnostic tool for classic visceral leish- determine if a meaningful construct tom illness, which is accompanied by sig- maniasis, but there is, as yet, no reliable could be identified. Although a report- nificant decreases in functioning and screening test for viscerotropic leish- ing bias may have contributed to the well-being. The illness was not associ- maniasis.43,44 As a surrogate for expo- moderately higher prevalence of symp- ated with physical examination or clini- sure to phlebotomine sand flies, which toms reported by GW veterans in the cal laboratory abnormalities or with in- transmit Leishmania parasites and sand index unit, the main findings of this fectionbyassortedagentsthatarefound fly fever viruses, we tested participants study remain valid for the comparison in the region or to which troops in com- for antibodies against Naples and Sicil- units. In addition, although the partici- bat may have been exposed. Poorly char- ian phlebotomus fever viruses. Seroprev- pationrateofoneofthecomparisonunits acterized illness, including fatigue, alence was similar to that noted with the (unit A) was low, the findings within that neurocognitive, and musculoskeletal com- Leishmania serologic assay. Clearly, nei- unit were consistent with those from the plaints, has affected veterans of many ther seroreactivity to leishmanial anti- other units. other wars.45 The empirically based ap- gens nor exposure to the phlebotomine Thesamelimitationsapplytotheclini- proach described here provides a method vectors was associated with the illness cal evaluation phase of the study. We for assessing the prevalence of multiple we defined. evaluated GW veterans from one unit in chronic symptoms and exploring the clini- Similarly, we found no association be- order to establish clinical parameters of cal basis of this condition. Other studies tween illness and antibody against the the illness we had defined and not to should use similarly rigorous analytic other viruses, rickettsiae, parasites, or conduct a case-control study. Partici- methods to classify cases and standard- bacteria for which we assayed. Chronic pantsunderwentasystematicstandard- ized measures of function and well- diarrhea was reported among 77% of se- ized physical examination by physician being to collect data and should enroll vere cases, but we found no associations assistants, who focused on documenting personnel from other branches of the mili- with any infectious agents. Although we obvious abnormalities. Thus, this study tary. did not evaluate subjects for noninfec- did not attempt to assess subtle physi- The elevated case prevalence among tious causes of diarrhea, such as mal- ologic differences. Furthermore, the num- GW veterans remains unexplained by our absorption syndromes, the descriptive ber of severe, chronic, multisymptom study. It is possible that the symptom com- stool sample characteristics and lack of cases was relatively small, which poten- plex is associated with GW-specific expo- weight loss suggest irritable bowel syn- tially limited our ability to identify sig- sures (eg, an as-yet-unidentified chemi- drome as the predominant reason for the nificant findings associated with those cal or biological agent).16 Although in our reported chronic diarrhea. cases. study population the risk of illness was not Our study has several limitations. In addition, all clinically evaluated associated with the dates, season, dura- Most important, the study involved cur- subjects were from a single ANG unit, tion, number of deployments, or military rently active air force personnel (prima- and only 35% of GW veterans from that occupational activities, we believe a more rily reservists) and cannot be general- unitparticipated.Thisparticipationrate plausible explanation is that key risk fac- ized to other branches of service or to may not adequately represent GW vet- tors present among both deployed and GW veterans who have left the service. erans. The extent to which ill GW vet- nondeployed populations were present at The primary objectives were to devise a erans were underrepresented or over- higher intensity or greater frequency case definition, evaluate the occurrence representedcouldinfluencethefindings. among GW veterans. It is clear that the of illness, and clinically describe the ill- A first interpretation may suggest that distribution of cases among GW veter- ness. We did not intend to test specific ill veterans were more likely to have vol- ans and nondeployed personnel in this hypotheses, many of which have arisen unteered, because 63% of the clinically study cannot easily be explained by risk subsequent to the study; rather, we in- evaluatedparticipantswereclassifiedas factors unique to Southwest Asia. tended to establish basic parameters to cases in contrast with the 45% preva- aid in future studies. lence of illness in GW veterans from the We gratefully thank Tara Strine, Carolyn The symptom data were self-reported cross-sectional survey. However, it was Sanchez, Drew Factor, MD, and Ewart Parris for and, thus, may be subject to reporting or our impression that military personnel their substantial help in conducting the study. We also thank Susan Killin and Theresa King for their recall bias. Other studies have at- with severe illness were less likely to help in conducting field work; Susanne P. Wahlquist, tempted to categorize GW veterans as participate for fear of identification and Craig Koniver, Marianna Wilson, Doris A. Ware, having illness in various predefined service-related medical consequences. FrancisJ.Steurer,andHenryS.Bishopfortheirhelp categories (eg, CFS, multiple chemical Moreover, it may be that the most seri- with laboratory testing; John Gow, MD, for entero- virus polymearse chain reaction assays; Fran Norris sensitivities, depression, and neurologic ously affected GW veterans are no longer for her help with assessing fatigue and stress; and abnormalities). We chose a completely in the military. Michele Reyes, PhD, for reviewing the study.

Downloaded From: by a Harvard University User on 03/12/2018 References 1. NationalInstitutesofHealthTechnologyAssess- for the Ongoing Monitoring and Verification of 32. Newhouse VF, Shepard CC, Redus MD, Tzi- ment Workshop Panel. The Persian Gulf experience Iraq’s Compliance with Relevant Parts of Section C anabos T, McDade JE. A comparison of the comple- and health. JAMA. 1994;272:391-396. of Security Council Resolution 687 (1991). New ment fixation, indirect immunofluorescent antibod- 2. Persian Gulf Veterans Coordinating Board. Un- York, NY: United Nations; 1995. ies, and microagglutination tests for the serologic explained illnesses among Desert Storm veterans: a 16. Presidential Advisory Committee on Gulf War diagnosis of rickettsial diseases. Am J Trop Med search for causes, treatment and cooperation. Arch Veterans’ Illnesses. Special Report. Washington, Hyg. 1979;28:387-395. Intern Med. 1995;155:262-268. DC: US Government Printing Office; October 1997. 33. Camargo ME. Fluorescent antibody test for the 3. Institute of Medicine. Health Consequences of 17. Haley RW, Kurt TL, Hom J. Is there a Gulf War serodiagnosis of American trypanosomiasis: techni- ServiceinthePersianGulfWar:Recommendations syndrome?searchingforsyndromesbyfactoranaly- cal modification employing preserved culture forms for Research and Information Systems. Washing- sis of symptoms. JAMA. 1997;277:215-222. of Trypanosoma cruzi in a slide test. Rev Inst Med ton, DC: National Academy Press; 1996. 18. Haley RW, Kurt TL. Self-reported exposure to Trop Sao Paulo. 1966;8:227-234. 4. Centers for Disease Control and Prevention. Un- neurotoxic chemical combinations in the Gulf War: a 34. Tsang VCW, Wilkens PP. Immunodiagnosis explained illness among Persian Gulf War veterans cross-sectional epidemiologic study. JAMA. 1997; ofschistosomiasis.ClinLabMed.1991;11:1029-1040. in an Air National Guard unit: preliminary report— 277;231-237. 35. Melvin DM, Brooke MM. Laboratory Proce- August 1990-March 1995. MMWR Morb Mortal 19. Murray-Leisure KA, Daniels MO. Visceral and dures for the Diagnosis of Intestinal Parasites. 3rd Wkly Rep. 1995;44:443-447. skin aspects of “Persian Gulf mystery disease” as- ed. Atlanta, Ga: US Dept of Health and Human Ser- 5. Iowa Persian Gulf Study Group. Self-reported sociated with sand exposure in a Pennsylvania mili- vices, Centers for Disease Control; 1982. Publica- illness and health status among Gulf War veterans: tary unit. In: Program and abstracts of the 34th In- tion CDC 85-8282. a population-based study. JAMA. 1997;277:238-245. terscience Conference on Antimicrobial Agents and 36. Ash LR, Orihel TC. Parasites: A Guide to Labo- 6. Department of Defense. Comprehensive Clini- Chemotherapy; October 1994; Orlando, Fla. Ab- ratory Procedures and Identification. Chicago, Ill: cal Evaluation Program for Gulf War Veterans: stract J159. American Society of Clinical Pathologists Press; Report on 18,598 Participants. Washington, DC: 20. Basilevski A. Statistical Factor Analysis and 1987:51. Dept of Defense; 1996. Related Methods. New York, NY: John Wiley & 37. Weber R, Bryan RT, Owen RL, Wilcox CM, 7. Hyams KC, Bourgeois AL, Merrell BR, et al. Di- Sons; 1994. Gorelkin L, Visvesvara GS. Improved light-micro- arrheal disease during Operation Desert Shield. 21. Floyd FJ, Widaman KF. Factor analysis in the scopical detection of microsporidia spores in stool N Engl J Med. 1991;325:1423-1428. development and refinement of clinical assessment and duodenal aspirates. N Engl J Med. 1992;326: 8. Hyams KC, Malone JD, Kapikian AZ, et al. Nor- instruments. Psych Assess. 1995;7:286-299. 161-166. walk virus infection among Desert Storm troops. 22. Tabachnick BG, Fidell LS. Using Multivariate 38. Gow JW, Behan WMH, Clements GB, Woodall J Infect Dis. 1993;167:986-987. Statistics. New York: Harper & Row; 1983. C, Riding M, Behan PO. Enteroviral RNA se- 9. Hyams KC, Hanson K, Wignall FS, Escamilla J, 23. Harman HH. Modern Factor Analysis. Chi- quences detected by polymerase chain reaction in Oldfield EC. The impact of infectious diseases on the cago, Ill: University of Chicago Press; 1976:268-270. muscle of patients with postviral fatigue syndrome. health of U.S. troops deployed to the Persian Gulf 24. Gorsuch RL. Factor Analysis. Hillsdale, NJ: BMJ. 1991;302:692-696. during operations Desert Shield and Desert Storm. Erlbaum; 1983. 39. SAS/Statistical User’s Guide. Vols 1 and 2. Clin Infect Dis. 1995;20:1497-1504. 25. Agresti A. Categorical Data Analysis. New Cary, NC: Statistical Analysis Systems Institute 10. Penman AD, Tarver RS, Currier MM. No evi- York: John Wiley & Sons; 1990. Inc; 1990. dence of increase in birth defects and health prob- 26. Ware JE, Sherbourne CD. The MOS 36-item 40. Stretch RH, Bliese PD, Marlowe DH. Physical lems among children born to Persian Gulf War vet- short-formhealthsurvey(SF-36):conceptualframe- symptomatology of Gulf War-era service personnel erans in Mississippi. Mil Med. 1996;161:1-6. work and item selection. Med Care. 1992;30:473-483. from the states of Pennsylvania and Hawaii. Mil 11. DeFraitesRF,WanantER,NorwoodAE,Willi- 27. McHorney CA, Ware JE, Rogers W, Raczek Med. 1995;160:131-136. ams S, Cowan D, Callahan T. Investigation of a Sus- AE, Lu JFR. The validity and relative precision of 41. Landis JR, Koch GG. The measurement of ob- pected Outbreak of an Unknown Disease Among MOS short- and long-form health status scales and server agreement for categorical data. Biometrics. Veterans of Operations Desert Shield/Storm, 123rd Dartmouth Coop charts: results from the Medical 1977;33:159-174. Army Reserve Command, Fort Benjamin Harri- Outcomes study. Med Care. 1992;30(suppl 5): 42. Nisenbaum R, Reyes M, Mawle AC, Reeves son, Indiana, April 1992. Washington DC: Walter MS253-MS265. WC. Factor analysis of unexplained severe fatigue Reed Army Institute of Research; 1992. 28. Keane TM, Caddell JM, Taylor KL. Mississippi and interrelated symptoms: overlap with criteria 12. Gray GC, Coate BD, Anderson CM, et al. The scale for combat related posttraumatic stress dis- for chronic fatigue syndrome. Am J Epidemiol. postwar hospitalization experience of US veterans order:threestudiesinreliabilityandvalidity. JCon- 1998;148:72-77. of the Persian Gulf War. N Engl J Med. 1996;335: sult Clin Psychol. 1988;56:85-90. 43. Centers for Disease Control and Prevention. 1505-1513. 29. Fukuda K, Straus SE, Hickie I, Sharpe MC, Viscerotropic leishmaniasis in persons returning 13. Writer JV, DeFraites RF, Brundage JF. Com- Dobbins JG, Komaroff A. The chronic fatigue syn- from Operation Desert Storm, 1990-1991. MMWR parative mortality among US military personnel in drome: a comprehensive approach to its definition Morb Mortal Wkly Rep. 1992;41:131-134. the Persian Gulf region and worldwide during op- and study. Ann Intern Med. 1994;21:953-959. 44. Magill AJ, Grogl M, Gasser RA, Sun W, Oster erations Desert Shield and Desert Storm. JAMA. 30. American Psychiatric Association. Diagnostic CN. Visceral infection caused by Leishmania 1996;275:118-121. and Statistical Manual of Mental Disorders.4th ed. tropica in veterans of Operation Desert Storm. 14. Kang HK, Bullman TA. Mortality among US Washington,DC:AmericanPsychiatricPress;1994. N Engl J Med. 1993;328:1383-1387. veterans of the Persian Gulf War. N Engl J Med. 31. Tsai TF. Arboviruses. In: Murray PR, Baron 45. Hyams RI, Wignall FS, Roswell R. War syn- 1996;335:1498-1504. EJ, Pfaller MA, Tenover FC, Yolken RH, eds. dromes and their evaluation: from the U.S. Civil 15. UnitedNationsSpecialCommissiononIraq.Re- Manual of Clinical Microbiology. 6th ed. Washing- War to the Persian Gulf War. Ann Intern Med. 1996; port of the Secretary-General on the Status of the ton, DC: American Society of Microbiology Press; 125:398-405. Implementation of the Special Commission’s Plan 1995:980-996.

Downloaded From: by a Harvard University User on 03/12/2018