Factor analysis of self-reported symptoms: does it identify a Gulf War syndrome?

Active duty US Naval mobile construction battalion personnel (Seabees) were surveyed in 1994 for the presence of a variety of symptoms. Questions were drawn from the Hopkins Symptom Checklist and from a collection of symptoms either defining clinical depression or commonly reported by Persian Gulf War veterans. Of those surveyed, 524 were Gulf War veterans and 935 were nondeployed Gulf War-era veterans. Factor analysis applied to Gulf War veterans yielded five factors, three deriving from the Hopkins Symptom Checklist, one suggesting clinical depression, and one containing symptoms commonly reported by Gulf War veterans. Factor analysis applied to nondeployed veterans yielded five similar factors. Three of the factors yielded statistically significantly greater standardized factor scores for Gulf War veterans than for nondeployed veterans. Four of the factors resembled factors resulting from a previous analysis on a sample of similar Gulf War veterans. Gulf War veterans and nondeployed era veterans reported similar clusters of symptoms and illnesses. However, Gulf War veterans reported these same clusters with greater frequencies than did nondeployed veterans. The authors conclude that, in contrast to a previous report, factor analysis did not identify a unique Gulf War syndrome.

Is systemic lupus erythematosus, amyotrophic lateral sclerosis, or fibromyalgia associated with Persian Gulf War service? An examination of Department of Defense hospitalization data.

Since the Persian Gulf War ended in 1991, veterans have reported diverse, unexplained symptoms. Some have wondered if their development of systemic lupus erythematosus, amyotrophic lateral sclerosis, or fibromyalgia might be related to Gulf War service. The authors used Cox proportional hazard modeling to determine whether regular, active-duty service personnel deployed to the Persian Gulf War (n = 551,841) were at increased risk of postwar hospitalization with the three conditions compared with nondeployed Gulf War era service personnel (n = 1,478,704). All hospitalizations in Department of Defense facilities from October 1, 1988, through July 31, 1997, were examined. With removal of personnel diagnosed with any of the three diseases before August 1, 1991, and adjustment for multiple covariates, Gulf War veterans were not at increased risk of postwar hospitalization due to systemic lupus erythematosus (risk ratio (RR) = 0.94, 95% confidence interval (CI): 0.65, 1.35). Because of the small number of cases and wide confidence limits, the data regarding amyotrophic lateral sclerosis were inconclusive. Gulf War veterans were slightly at risk of postwar hospitalization for fibromyalgia (RR = 1.23, 95% Cl: 1.05, 1.43); however, this risk difference was probably due to the Gulf War veteran clinical evaluation program beginning in 1994. These data do not support Gulf War service and disease associations.

Are Gulf War veterans suffering war-related illnesses? Federal and civilian hospitalizations examined, June 1991 to December 1994.

A previous epidemiologic study demonstrated no unexplained increase in risk for postwar hospitalization among Gulf War veterans who had remained on active duty. The authors sought to expand this study to include Reserve and separated military personnel. They examined hospitalization data from the Department of Defense, the Department of Veterans Affairs (VA), and the California Office of Statewide Health Planning and Development hospital systems for the years 1991-1994. Since denominator data were not available, the authors compared the proportional morbidity ratios (PMRs) of hospitalization discharge diagnoses (both large categories and specific diagnoses) between Gulf War veterans and other veterans of the same era. There were no indications that Gulf War veterans were suffering increased PMRs for infectious diseases; neoplasms; endocrine diseases; blood diseases; skin conditions; or diseases of the nervous system, circulatory system, or musculoskeletal system. However, these veterans did experience proportionally more hospitalizations for various specific diagnoses, namely, fractures and bone and soft-tissue injuries (Department of Defense and California Office of Statewide Health Planning and Development), various diseases of the respiratory (including asthma) and digestive systems (VA), and diverse symptom diagnoses (VA). While these findings may be influenced by chance or by a number of potential confounders, including health registry participation, they merit further examination using other study designs.

Risk factors for mental disorder hospitalization after the Persian Gulf War: U.S. Armed Forces, June 1, 1991-September 30, 1993.

Effects of Persian Gulf War (August 2, 1990-July 31, 1991) and Gulf War occupation on post-War hospitalization risk were evaluated through Cox proportional hazards modeling. Active-duty men (n = 1,775,236) and women (n = 209,760) in the Army, Air Force, Navy, and Marine Corps had 30,539 initial postwar hospitalizations for mental disorders between June 1, 1991 and September 30, 1993. Principal diagnoses in the Defense Manpower Data Center hospitalization database were grouped into 10 categories of ICD-9-CM codes. Gulf War service was associated with significantly greater risk for acute reactions to stress and lower risk for personality disorders and adjustment reactions among men. Personnel who served in ground war support occupations (men and women) were at greater risk for postwar drug-related disorders. Men who served in ground war combat occupations were at higher risk for alcohol-related disorders. Longitudinal studies of health, hospitalization, and exposure beginning at recruitment, are needed to better understand how exposure to combat affects the mental health of military personnel.

The postwar hospitalization experience of Gulf War Veterans possibly exposed to chemical munitions destruction at Khamisiyah, Iraq.

Using Department of Defense hospital data, the authors examined the postwar hospitalization experience from March 1991 through September 1995 of US Gulf War veterans who were near Khamisiyah, Iraq, during nerve agent munition destruction in March 1991. Multiple sources of meteorologic, munition, and toxicology data were used to circumscribe geographic areas of low level, vaporized nerve agent for 4 days after the destruction. Plume estimates were overlaid on military unit positions, and exposure was estimated for the 349,291 US Army Gulf War veterans. Exposure was classified as not exposed (n = 224,804), uncertain low dose exposure (n = 75,717), and specific estimated subclinical exposure (n = 48,770) categorized into three groups for dose-response evaluation. Using Cox proportional hazard modeling, the authors compared the postwar experiences of these exposure groups for hospitalization due to any cause, for diagnoses in 15 unique categories, and for specific diagnoses an expert panel proposed as most likely to reflect latent disease from such subclinical exposure. There was little evidence that veterans possibly exposed to the nerve agent plumes experienced unusual postwar morbidity. While there were several differences in hospitalization risk, none of the models suggested a dose-response relation or neurologic sequelae. These data, having a number of limitations, do not support the hypothesis that Gulf War veterans are suffering postwar morbidity from subclinical nerve agent exposure.

Testicular cancer and Persian Gulf War service.

We studied whether regular, active-duty servicemen deployed to the Persian Gulf War were at increased risk of testicular cancer compared with nondeployed Gulf War-era servicemen from August 1991 through March 31, 1996, using a Cox proportional hazards model for survival analysis with covariates. Race was an important predictor of hospitalization for testicular cancer [rate ratio (RR) = 0.19; 95% confidence interval (CI) = 0.12-0.29 for blacks, and RR = 0.59; 95% CI = 0.39-0.91 for Hispanics, other, and unknown (combined), relative to whites]. Age effects were modest (RR = 1.19; 95% CI = 0.91-1.56 for those of ages 22-25 years, and RR = 1.24; 95% CI = 0.96-1.59 for those of ages 26-31 years, compared with those of ages 17-21 years). Risk also varied with occupation (RR = 1.56; 95% CI = 1.23-2.00 for those in electronic equipment repair; RR = 1.26; 95% CI = 1.01-1.58 for those in electrical/mechanical repair; and RR = 1.42; 95% CI = 0.93-2.17 for those in construction-related trades, compared with those in other occupations). Deployment status was not important (RR = 1.05; 95% CI = 0.86-1.29 for the deployed compared with the nondeployed). There was an increase in testicular cancer in the deployed group in the immediate postwar period that was consistent with a previous report of a standardized RR of 2.12; 95% CI = 1.11-4.02 (compared with the nondeployed group) in the last 5 months of 1991, but by 4 years after the end of deployment, the cumulative risks for the two groups were not different. An additional analysis suggested that the immediate postwar increase in the deployed was likely due to regression to the mean after a healthy serviceman selection effect for deployment and the deferment of care during deployment.

Gulf War Veterans' Health Registries. Who is most likely to seek evaluation?

Since the Persian Gulf War ended in 1991, many veterans have sought medical evaluation in the Department of Veterans Affairs Persian Gulf Veterans' Health Registry (VA registry) or the Department of Defense's Comprehensive Clinical Evaluation Program (DoD registry). Using combined data collected from 1993 to 1997 from the VA and DoD registries, the authors compared the characteristics of registry participants (n=74,653) with those of all Gulf War veterans (n=696,531) to determine the personnel most likely to seek medical evaluation. Using multiple logistic regression, the authors found that service branch and type were strongly associated with registry participation, with Army (adjusted odds ratio (OR)=4.7, 95% confidence interval (CI) 4.6-4.9) and National Guard (OR=2.6, 95% CI 2.5-2.6) personnel at highest odds compared with reference category personnel. Registry participants also were more likely to have been stationed in the Gulf War theater during the fighting (OR=2.2), to be older (>31 years/<22 years OR=2.1), to have been an enlisted person (OR=2.0), to have been construction workers (OR=1.3), to be female (OR=1.3), and to have been hospitalized during the 12-month period before the war (OR=1.2). These findings are useful in generating hypotheses regarding postwar morbidity. They also suggest that subpopulations of Gulf War veterans have a higher prevalence of symptoms and merit further study.

Hospitalizations for unexplained illnesses among U.S. veterans of the Persian Gulf War.

Persian Gulf War veterans have reported a variety of symptoms, many of which have not led to conventional diagnoses. We ascertained all active-duty U.S. military personnel deployed to the Persian Gulf War (552,111) and all Gulf War era military personnel not deployed (1,479,751) and compared their postwar hospitalization records (until 1 April 1996) for one or more of 77 diagnoses under the International Classification of Diseases (ICD-9) system. The diagnoses were assembled by the Emerging Infections Program, Centers for Disease Control and Prevention, and are here termed "unexplained illnesses." Deployed veterans were found to have a slightly higher risk of hospitalization for unexplained illness than the nondeployed. Most of the excess hospitalizations for the deployed were due to the diagnosis "illness of unknown cause" (ICD-9 code 799.9), and most occurred in participants of the Comprehensive Clinical Evaluation Program who were admitted for evaluation only. When the effect of participation in this program was removed, the deployed had a slightly lower risk than the nondeployed. These findings suggest that active-duty Gulf War veterans did not have excess unexplained illnesses resulting in hospitalization in the 4.67-year period following deployment.

The postwar hospitalization experience of U.S. veterans of the Persian Gulf War.

BACKGROUND: Since the Persian Gulf War ended in 1991, many veterans of that conflict have reported diverse, unexplained symptoms. To evaluate the health of Gulf War veterans, we studied their postwar hospitalization experience and compared it with that of other military personnel serving at the same time who did not go to the Persian Gulf. METHODS: Using a retrospective cohort approach and data from Department of Defense hospitals, we studied hospitalizations of 547,076 veterans of the Gulf War who were serving in the Army, Navy, Marine Corps, and Air Force and 618,335 other veterans from the same era who did not serve in the Persian Gulf. Using multivariate logistic-regression models, we analyzed risk factors for hospitalization both overall and in 14 broad diagnostic categories during three periods from August 1991 through September 1993 (a total of 45 specific comparisons). RESULTS: After the war, the overall odds ratio for hospitalization of the Gulf War veterans was not higher than that of the other veterans, even after adjustment for selection effects related to deployment. In 16 of the 42 comparisons involving specific diagnoses, the risk of hospitalization among Gulf War veterans differed significantly from that among other veterans. Among these 16 comparisons, Gulf War veterans were at higher risk in 5: neoplasms (largely benign) during 1991, diseases of the genitourinary system during 1991, diseases of the blood and blood-forming organs (mostly forms of anemia) during 1992, and mental disorders during both 1992 and 1993. The differences were not consistent over time and could be accounted for by deferred care, postwar pregnancies, and postwar stress. CONCLUSIONS: During the two years after the Persian Gulf War, there was no excess of unexplained hospitalization among Americans who remained on active duty after serving in that conflict.

Nonparametric analysis of covariance for comparing change in randomized studies with baseline values subject to error.

Change from baseline to a follow-up examination can be compared among two or more randomly assigned treatment groups by using analysis of variance on the change scores. However, a generally more sensitive (powerful) test can be performed using analysis of covariance (ANOVA) on the follow-up data with the baseline data as a covariate. This approach is not without potential problems, though. The assumption of ordinary ANCOVA of normally distributed errors is speculative for many variables employed in biomedical research. Furthermore, the baseline values are inevitably random variables and often are measured with error. This report investigates, in this situation, the validity and relative power of the ordinary ANCOVA test and two asymptotically distribution-free alternative tests, one based on the rank transformation and the other based on the normal scores transformation. The procedures are illustrated with data from a clinical trial. Normal and several nonnormal distributions, as well as varying degree of variable error, are studied by Monte Carlo methods. The normal scores test is generally recommended for statistical practice.

A comparative study of two statistical models for the analysis of binary data from longitudinal studies.

This study extensively compares two statistical models for the analysis of binary data from longitudinal studies. The first model was proposed by Zeger, Liang, and Self, which was abbreviated as ZLS model and another model was proposed by Origasa. The comparison focuses on both analytical and statistical view-points. The first discusses a type of the models and the second evaluates the effect from model misspecification by stimulation, assuming that the ZLS model is true.

Estimation of error rates in discriminant analysis with selection of variables.

Accurate estimation of misclassification rates in discriminant analysis with selection of variables by, for example, a stepwise algorithm, is complicated by the large optimistic bias inherent in standard estimators such as those obtained by the resubstitution method. Application of a bootstrap adjustment can reduce the bias of the resubstitution method; however, the bootstrap technique requires the variable selection procedure to be repeated many times and is therefore difficult to compute. In this paper we propose a smoothed estimator that requires relatively little computation and which, on the basis of a Monte Carlo sampling study, is found to perform generally at least as well as the bootstrap method.

High-density lipoprotein cholesterol and cardiovascular disease. Four prospective American studies.

The British Regional Heart Study (BRHS) reported in 1986 that much of the inverse relation of high-density lipoprotein cholesterol (HDLC) and incidence of coronary heart disease was eliminated by covariance adjustment. Using the proportional hazards model and adjusting for age, blood pressure, smoking, body mass index, and low-density lipoprotein cholesterol, we analyzed this relation separately in the Framingham Heart Study (FHS), Lipid Research Clinics Prevalence Mortality Follow-up Study (LRCF) and Coronary Primary Prevention Trial (CPPT), and Multiple Risk Factor Intervention Trial (MRFIT). In CPPT and MRFIT (both randomized trials in middle-age high-risk men), only the control groups were analyzed. A 1-mg/dl (0.026 mM) increment in HDLC was associated with a significant coronary heart disease risk decrement of 2% in men (FHS, CPPT, and MRFIT) and 3% in women (FHS). In LRCF, where only fatal outcomes were documented, a 1-mg/dl increment in HDLC was associated with significant 3.7% (men) and 4.7% (women) decrements in cardiovascular disease mortality rates. The 95% confidence intervals for these decrements in coronary heart and cardiovascular disease risk in the four studies overlapped considerably, and all contained the range 1.9-2.9%. HDLC levels were essentially unrelated to non-cardiovascular disease mortality. When differences in analytic methodology were eliminated, a consistent inverse relation of HDLC levels and coronary heart disease event rates was apparent in BRHS as well as in the four American studies.

Physiological markers of smoking and their relation to coronary heart disease. The Lipid Research Clinics Coronary Primary Prevention Trial.

Several physiological variables which have previously been found to be associated with extent of cigarette smoking were investigated in a population selected to be free of existing coronary heart disease yet having high circulating cholesterol levels. Of these variables, white blood cell count had the strongest association with the extent of smoking. The addition of hematocrit and heart rate to white blood cell count strengthened the association. Given these three variables, however, other physiological variables had a minor effect on the association. Finally, the combination of these three physiological variables was more strongly predictive of subsequent coronary heart disease than was self-reported smoking level.

Estimating baseline values of the variable of intervention in a clinical trial.

Parametric empirical Bayes methodology is suggested for determining estimators of individual baseline values of the variable of intervention in a clinical trial, when the variable is measured twice--once for subject selection, and again, without selection, just before randomization. The resulting compromise estimator is seen to have more precision than the baseline estimator employing only the second value and less bias than the estimator that simply averages the two values. Construction of such an estimator is illustrated using data from the recruitment phase of the Lipid Research Clinics Coronary Primary Prevention Trial. Generalizations to other designs are also suggested. In all cases, however, an estimate of the intraindividual variance of the variable of intervention is required.

Survival after recovery from acute myocardial infarction. Two and five year prognostic indices.

A prognostic index for two year survival following recovery from acute myocardial infarction has been verified in an independent group of 105 patients. Five variables comprise the index: systolic blood pressure level on admission, highest blood urea nitrogen level in the cardiac care unit; atrial arrhythmias in the cardiac care unit; angina pectoris for more than three months or a previous myocardial infarction; and more than one ventricular ectopic beat per hour recorded on an 8 hour dynamic electrocardiogram during convalescence just prior to hospital discharge. One hundred twenty-six patients have also been followed for five or more years, and we now report a five year prognostic index. Discriminant analysis indicates that the same five variables, although weighted differently, continue to be significant for prognostic assessment and may be utilized in the identification of patients at high and lower risk.

Clinical estimation of gestational age: rules for avoiding preterm delivery.

Reliable knowledge of the duration of pregnancy prior to birth is often of crucial importance in making obstetric care decisions. Laboratory methods for estimating fetal maturity have received considerable attention, but the usefulness of historical information has only rarely been addressed. In order to examine the value of clinical estimators of fetal gestational age (GA) in 690 pregnancies, the correlations of menstrual history (LMP), first unamplified audible fetal heart tones (FFH), and quickening (Q), with GA, based on the modified Dubowitz examination at birth, were examined. Evaluation of each of the data sets used alone reveals that in order to be 90% certain that an infant will be mature at delivery (greater than or equal to 38 weeks), a reliable LMP must have been noted for 42 weeks prior to birth, the FFH heard for 21 weeks, and Q felt for 25 weeks. These findings suggest that carefully obtained historical and physical examination information remains a cornerstone of appropriate obstetric care.