Understanding the elevated suicide risk of female soldiers during deployments.

BACKGROUND: The Army Study to Assess Risk and Resilience in Servicemembers (Army STARRS) has found that the proportional elevation in the US Army enlisted soldier suicide rate during deployment (compared with the never-deployed or previously deployed) is significantly higher among women than men, raising the possibility of gender differences in the adverse psychological effects of deployment. METHOD: Person-month survival models based on a consolidated administrative database for active duty enlisted Regular Army soldiers in 2004-2009 (n = 975,057) were used to characterize the gender × deployment interaction predicting suicide. Four explanatory hypotheses were explored involving the proportion of females in each soldier's occupation, the proportion of same-gender soldiers in each soldier's unit, whether the soldier reported sexual assault victimization in the previous 12 months, and the soldier's pre-deployment history of treated mental/behavioral disorders. RESULTS: The suicide rate of currently deployed women (14.0/100,000 person-years) was 3.1-3.5 times the rates of other (i.e. never-deployed/previously deployed) women. The suicide rate of currently deployed men (22.6/100,000 person-years) was 0.9-1.2 times the rates of other men. The adjusted (for time trends, sociodemographics, and Army career variables) female:male odds ratio comparing the suicide rates of currently deployed v. other women v. men was 2.8 (95% confidence interval 1.1-6.8), became 2.4 after excluding soldiers with Direct Combat Arms occupations, and remained elevated (in the range 1.9-2.8) after adjusting for the hypothesized explanatory variables. CONCLUSIONS: These results are valuable in excluding otherwise plausible hypotheses for the elevated suicide rate of deployed women and point to the importance of expanding future research on the psychological challenges of deployment for women.

Perinatal bisphenol A exposure beginning before gestation enhances allergen sensitization, but not pulmonary inflammation, in adult mice.

Bisphenol A (BPA), a monomer of polycarbonate plastics and epoxide resin, is a high-production-volume chemical implicated in asthma pathogenesis when exposure occurs to the developing fetus. However, few studies have directly examined the effect of in utero and early-life BPA exposure on the pathogenesis of asthma in adulthood. This study examines the influence of perinatal BPA exposure through maternal diet on allergen sensitization and pulmonary inflammation in adult offspring. Two weeks before mating, BALB/c dams were randomly assigned to a control diet or diets containing 50 ng, 50 µg or 50 mg BPA/kg of rodent chow. Dams remained on the assigned diet throughout gestation and lactation until postnatal day (PND) 21 when offspring were weaned onto the control diet. Twelve-week-old offspring were sensitized to ovalbumin (OVA) and subsequently challenged with aerosolized OVA. Sera, splenocytes, bronchoalveolar lavage fluid and whole lungs were harvested to assess allergen sensitization and pulmonary inflammation after OVA challenge. Serum anti-OVA IgE levels were increased two-fold in offspring exposed to 50 µg and 50 mg BPA/kg diet, compared with control animals. In addition, production of interleukin-13 and interferon-γ were increased in OVA-stimulated splenocytes recovered from BPA-exposed mice. Pulmonary inflammation, as indicated by total and differential leukocyte counts, cytokines, chemokines and pulmonary histopathology inflammatory scores, however, was either not different or was reduced in offspring exposed to BPA. Although these data suggest that perinatal BPA exposure beginning before gestation enhances allergen sensitization by increasing serum IgE and splenocyte cytokine production, a substantial impact of BPA on OVA-induced pulmonary inflammation in adulthood was not observed.

The design and conduct of clinical trials to limit missing data.

This article summarizes recommendations on the design and conduct of clinical trials of a National Research Council study on missing data in clinical trials. Key findings of the study are that (a) substantial missing data is a serious problem that undermines the scientific credibility of causal conclusions from clinical trials; (b) the assumption that analysis methods can compensate for substantial missing data is not justified; hence (c) clinical trial design, including the choice of key causal estimands, the target population, and the length of the study, should include limiting missing data as one of its goals; (d) missing-data procedures should be discussed explicitly in the clinical trial protocol; (e) clinical trial conduct should take steps to limit the extent of missing data; (f) there is no universal method for handling missing data in the analysis of clinical trials - methods should be justified on the plausibility of the underlying scientific assumptions; and (g) when alternative assumptions are plausible, sensitivity analysis should be conducted to assess robustness of findings to these alternatives. This article focuses on the panel's recommendations on the design and conduct of clinical trials to limit missing data. A companion paper addresses the panel's findings on analysis methods.

Life on the farm-children at risk.

BACKGROUND: More than 1 million children live, play, and work on farms, surrounded by animals and machinery. This symbiotic relationship between work and home exposes children to unique risks. METHODS: Children presenting with a farm-related injury (November 1994 to August 2001, 82 months) were included. Trauma registry parameters included injury severity score (ISS); Glascow Coma Scale (GCS); time to presentation; season and day of injury; emergency room, intensive care unit, and total length of stay type; and mechanism of injury; and operations. RESULTS: A total of 1,832 pediatric trauma patients were evaluated. Ninety-four children were identified with farm-related injuries. Mean age was 10.75 years. Mean ISS was 7.38. Three children died. Four children wore protective equipment. Forty-four percent of injuries occurred during summer, 31% during spring, and 55% on weekends. Average time to initial presentation was 39 minutes. A total of 177 minutes elapsed before transfer to regional trauma center. Seventy-two children required admission. LOS was 0 to 28 days, mean, 2.76 days. Twenty-six children (28%) required operations. Injuries included dislocations/fractures (52%), lacerations/avulsions (38%), concussions (31%), contusions (30%), and burns (14%). Mechanism included animals (41%), falls (34%), motor vehicles (28%), all-terrain vehicles (20%), and firearms (4%). CONCLUSIONS: Farm injuries occur most commonly during weekends, summer, and spring months, resulting in significant morbidity. Most injuries required hospitalization. Unless unstable, initial transfer to a regional pediatric trauma center should result in the most cost-effective, prompt, and highest quality of care.

Patterns of neuropsychological performance in multiple system atrophy compared to sporadic and hereditary olivopontocerebellar atrophy.

Although neuropsychological symptoms are associated with multiple system atrophy (MSA), sporadic olivopontocerebellar atrophy (sOPCA), and dominantly inherited olivopontocerebellar atrophy (dOPCA), the differences between these groups have not been explored. We compared 28 MSA patients on psychiatric rating scales and neuropsychological measures to 67 sOPCA patients, 42 dOPCA patients, and 30 normal controls. Patients with dOPCA, sOPCA, and MSA all exhibited significant deficits on motor-related tasks, as well as relatively mild deficits in cognitive functioning. Patients with MSA had greater neuropsychological dysfunction, particularly in memory and other "higher order" cognitive processes, than patients with either sOPCA or dOPCA.

A profile conditional likelihood approach for the semiparametric transformation regression model with missing covariates.

We propose a profile conditional likelihood approach to handle missing covariates in the general semiparametric transformation regression model. The method estimates the marginal survival function by the Kaplan-Meier estimator, and then estimates the parameters of the survival model and the covariate distribution from a conditional likelihood, substituting the Kaplan-Meier estimator for the marginal survival function in the conditional likelihood. This method is simpler than full maximum likelihood approaches, and yields consistent and asymptotically normally distributed estimator of the regression parameter when censoring is independent of the covariates. The estimator demonstrates very high relative efficiency in simulations. When compared with complete-case analysis, the proposed estimator can be more efficient when the missing data are missing completely at random and can correct bias when the missing data are missing at random. The potential application of the proposed method to the generalized probit model with missing continuous covariates is also outlined.

Changes in young adult offense and crash patterns over time.

A study of 13,809 young adult drivers in Michigan examined offenses and crashes ('incidents') for an average of 7 years after their original license date. During this period, 73% of subjects committed an offense that resulted in a conviction and 58% had a crash that was reported to the police. Forty-two percent had committed an offense classified as 'serious,' and 21% had an 'at-fault' crash. The odds of an offense being serious decreased approximately 8% per year of licensure, independent of gender or age at licensure. Similarly, the odds of a crash being at-fault decreased overall about 6% per year of licensure, but the decline was more than twice as fast for women as for men. Examining the empirical rates directly, it was found that the rate for minor offenses increased somewhat with time and then stabilized, while the rate for serious offenses declined. Also, offenses were less likely to be serious the later they occurred in the sequence of offenses for an individual. For crashes, the risk of having an at-fault crash declined more rapidly than the risk of a not-at-fault crash, although the rate of decrease began to equalize after approximately 5 years of licensure. The proportion of crashes that were at-fault did not decline over the sequence of crashes for an individual. Although crashes and offenses are positively correlated, they follow different trajectories over the early years of licensure.

Causal effects in clinical and epidemiological studies via potential outcomes: concepts and analytical approaches.

A central problem in public health studies is how to make inferences about the causal effects of treatments or agents. In this article we review an approach to making such inferences via potential outcomes. In this approach, the causal effect is defined as a comparison of results from two or more alternative treatments, with only one of the results actually observed. We discuss the application of this approach to a number of data collection designs and associated problems commonly encountered in clinical research and epidemiology. Topics considered include the fundamental role of the assignment mechanism, in particular the importance of randomization as an unconfounded method of assignment; randomization-based and model-based methods of statistical inference for causal effects; methods for handling noncompliance and missing data; and methods for limiting bias in the analysis of observational data, including propensity score matching and sensitivity analysis.

Balanced centralized and distributed database design in a clinical research environment.

Clinical research databases can meet both research and clinical needs, but this ideal is seldom achieved. Priorities often differ for those who collect and ultimately use the data and those who develop data systems. Traditional database designs also create logistical barriers that hamper communication. The Michigan Alzheimer's Disease Research Center has developed a secure, distributed data system with centralized data entry that provides an intuitive, individually customized interface for investigators in their clinics, laboratories and offices. Data are kept in a form that can be readily understood without reference to a code-book. Investigators can modify and query their own copies of the database without knowledge of programming languages. Balancing centralized and distributed designs for research databases enhance the accuracy and completeness of data collection and increases the use of data for research and clinical care.

A comparison of subset selection and analysis of covariance for the adjustment of confounders.

Two common methods for adjusting group comparisons for differences in the distribution of confounders, namely analysis of covariance (ANCOVA) and subset selection, are compared using real examples from neuropsychology, theory, and simulations. ANCOVA has potential pitfalls, but the blanket rejection of the method in some areas of empirical psychology is not justified. Assumptions of the methods are reviewed, with issues of selection bias, nonlinearity, and interaction emphasized. Advantages of ANCOVA include better power, improved ability to detect and estimate interactions, and the availability of extensions to deal with measurement error in the covariates. Forms of ANCOVA are advocated that relax the standard assumption of linearity between the outcome and covariates. Specifically, a version of ANCOVA that models the relationship between the covariate and the outcome through cubic spline with fixed knots outperforms other methods in simulations.

On summary measures analysis of the linear mixed effects model for repeated measures when data are not missing completely at random.

Subjects often drop out of longitudinal studies prematurely, yielding unbalanced data with unequal numbers of measures for each subject. A simple and convenient approach to analysis is to develop summary measures for each individual and then regress the summary measures on between-subject covariates. We examine properties of this approach in the context of the linear mixed effects model when the data are not missing completely at random, in the sense that drop-out depends on the values of the repeated measures after conditioning on fixed covariates. The approach is compared with likelihood-based approaches that model the vector of repeated measures for each individual. Methods are compared by simulation for the case where repeated measures over time are linear and can be summarized by a slope and intercept for each individual. Our simulations suggest that summary measures analysis based on the slopes alone is comparable to full maximum likelihood when the data are missing completely at random but is markedly inferior when the data are not missing completely at random. Analysis discarding the incomplete cases is even worse, with large biases and very poor confidence coverage.

Soft drugs based on hydrocortisone: the inactive metabolite approach and its application to steroidal antiinflammatory agents.

PURPOSE: The soft drug approach was applied to the design of analogs of highly potent synthetic steroids but with a metabolically labile ester group which at the same time served as an activating group. METHODS: Several structural modifications of soft antiinflammatory steroids were synthesized and tested in several assays of biological activity. The hydrolytic stability of the compounds was also determined. RESULTS: One of the compounds synthesized was determined to be a very potent steroid and had a highly significant separation of topical from systemic activity. However, the compound demonstrated greater than expected stability in the hydrolysis studies. CONCLUSIONS: The goal of the soft drug approach has been achieved with the development of a highly potent drug which displays little or no systemic activity as measured in the tests presented here. The anticipated hydrolytic instability of the compounds was not corroborated; however, in view of other results, the interpretation is allowed that the rapid hydrolysis of the unbound fraction of the drug is an important factor in its lack of systemic effects.

Assessment of weighting methodology for the National Comorbidity Survey.

The authors studied weighting adjustments for the National Comorbidity Survey (1990-1992), a large-scale national epidemiologic investigation of the prevalence, risk factors, and consequences of psychiatric morbidity and comorbidity in the United States. Weighting adjustments for differential selection within households, new construction, unit nonresponse, and poststratification were examined separately and in combination. Specific issues addressed included the magnitude of the bias incurred from ignoring the weights, the added variance from weighting and how well this was predicted by simple formulae, and the performance of methods for trimming the weights. Weights had quite modest effects on point estimates of prevalences but resulted in major increases in variance unless trimmed. The weights after trimming and poststratification appeared to work well. It is suggested that the added variance from weighting be carefully monitored in similar surveys. Alternatives to the use of trimming for controlling variance are worth exploring.

Cognitive, behavioral, and motor effects of the NMDA antagonist ketamine in Huntington's disease.

BACKGROUND: Excitotoxicity may contribute to neuronal degeneration in Huntington's disease (HD). N-methyl-D-aspartate (NMDA) receptor antagonists can prevent neuronal degeneration caused by excitotoxicity, but their effects in HD patients are not known. METHODS: We investigated the acute cognitive, behavioral, and motor effects of the NMDA-receptor antagonist ketamine in HD patients. Double-blind infusions of 0.10, 0.40, and 0.60 mg/kg/hr ketamine were given to 10 HD patients on one test day and compared with placebo infusions on a second, identical testing day. Linear mixed-effects models and randomization tests were used to identify whether, and at which dose, a significant change from baseline occurred in outcome variables. RESULTS: We demonstrated that ketamine is well tolerated at low and intermediate subanesthetic doses. Intermediate ketamine doses produced specific decline in memory and verbal fluency. Higher subanesthetic doses caused a significant increase in psychiatric symptoms and impairment of eye movements. CONCLUSIONS: These results describe the spectrum of clinical effects produced by increasing NMDA receptor blockade in HD patients. The clinical effects appearing with higher levels of NMDA receptor blockade can identify the range of doses used in clinical trials of NMDA receptor antagonists.

Inference for phylogenies under a hybrid parsimony method: evolutionary-symmetric transversion parsimony.

A new method is proposed for inferring topology for evolutionary trees. Existing methods have complementary strengths and weaknesses. Maximum and transversion parsimony are powerful methods, but they lack statistical consistency, that is, they do not always infer the correct tree as the sequence length becomes very large. Evolutionary parsimony overcomes this deficiency, but it may lack sufficient power when sequence length is small (less than 1000 aligned nucleotides; Sinsheimer, Lake, and Little, 1996, Biometrics 52, 193-210). Our proposed method, evolutionary-symmetric transversion parsimony, is a hybrid that retains the consistency of evolutionary parsimony, while increasing power by incorporating a modified form of transversion parsimony within a statistical model. The method requires choice of a parameter gamma that represents the prior probability that symmetric transversion parsimony yields consistent results. Properties of the method are assessed for a variety of choices of gamma in a large simulation study. In general, inference under the evolutionary-symmetric transversion parsimony has more discriminating power than inference under evolutionary parsimony and is better calibrated than inference under symmetric transversion parsimony. The results are quite robust to the choice of gamma, indicating a value of 0.90 as a reasonable overall choice when the true value of gamma ranges between 0.85 to 1.00. Our method is, like evolutionary parsimony and maximum parsimony, computationally straightforward. The same statistical approach can be applied to combine evolutionary parsimony with other inconsistent methods, such as maximum parsimony, but at the expense of more difficult computations.

Patterns of DSM-III-R alcohol dependence symptom progression in a general population survey.

Age of onset reports obtained retrospectively for each symptom of DSM-III-R alcohol dependence (AD) are used to study patterns of lifetime symptom progression in a large general-population survey of people in the United States. It is shown that symptom progression among a substantial majority of respondents can be summarized as movement across three clusters. Cluster A is defined by symptoms of role impairment/hazardous use (A4), use despite social, psychological or physical problems (A6), and drinking larger amounts or over a longer period of time than intended (A1). Cluster B is defined by tolerance (A7) and impaired control (A2, A3). Cluster C is defined by withdrawal (A8, A9) and giving up activities in order to drink (A5). Clusters are shown to follow a time sequence, with at least one symptom in Cluster A usually occurring first, followed by symptoms in Clusters B and C. In all, 83.4% of the symptom cluster transitions estimated from retrospective age of onset reports are consistent with this progression. Progression to AD is differentially predicted by symptom profiles reported at the age of first symptom onset, with persons reporting Cluster C symptoms most likely to progress subsequently to AD. Furthermore, profiles of AD defined by the highest symptom cluster present at AD onset are differentially predicted by prior personal and parental histories of psychopathology and, among men, are predictive of diagnosis persistence.

Pattern-mixture models for multivariate incomplete data with covariates.

Pattern-mixture models stratify incomplete data by the pattern of missing values and formulate distinct models within each stratum. Pattern-mixture models are developed for analyzing a random sample on continuous variables y(1), y(2) when values of y(2) are nonrandomly missing. Methods for scalar y(1) and y(2) are here generalized to vector y(1) and y(2) with additional fixed covariates x. Parameters in these models are identified by alternative assumptions about the missing-data mechanism. Models may be underidentified (in which case additional assumptions are needed), just-identified, or overidentified. Maximum likelihood and Bayesian methods are developed for the latter two situations, using the EM and SEM algorithms, direct and interactive simulation methods. The methods are illustrated on a data set involving alternative dosage regimens for the treatment of schizophrenia using haloperidol and on a regression example. Sensitivity to alternative assumptions about the missing-data mechanism is assessed, and the new methods are compared with complete-case analysis and maximum likelihood for a probit selection model.

Bayesian hypothesis testing of four-taxon topologies using molecular sequence data.

The reconstruction of phylogenetic trees from molecular sequences presents unusual problems for statistical inference. For example, three possible alternatives must be considered for four taxa when inferring the correct unrooted tree (referred to as a topology). In our view, classical hypothesis testing is poorly suited to this triangular set of alternative hypotheses. In this article, we develop Bayesian inference to determine the posterior probability that a four-taxon topology is correct given the sequence data and the evolutionary parsimony algorithm for phylogenetic reconstruction. We assess the frequency properties of our models in a large simulation study. Bayesian inference under the principles of evolutionary parsimony is shown to be well calibrated with reasonable discriminating power for a wide range of realistic conditions, including conditions that violate the assumptions of evolutionary parsimony.

Assessing HIV vaccine effects.

It is unlikely that human immunodeficiency virus (HIV) vaccines will create impenetrable barriers to infection. When the barriers to infection are broken, however, vaccine effects on the progression of infection to disease and on the contagiousness of infection could be considerable. The usual outcomes of vaccine trials are either infection or disease. The authors argue that for HIV vaccines, the alternative outcome of contagiousness may be more important. Because of the long incubation period to acquired immunodeficiency syndrome (AIDS), a vaccine trial with AIDS as the outcome would be a long and costly undertaking. Because contagiousness is concentrated into the period of primary infection, vaccine trials assessing contagiousness would not take as long. An approach to assessing vaccine effects on the contagiousness of primary infection while simultaneously assessing protection against infection is presented. It involves randomizing vaccination of couples in whom both individuals are uninfected and one or both have a risk of infection outside the couple. In such a study, the vaccine effect on susceptibility to infection can be estimated from the proportions of vaccinated and unvaccinated couples in whom neither partner is infected. Estimation of the contagiousness effect also uses information on the frequency with which both partners are infected. In areas of the world where heterosexual epidemics are emerging within the context of concurrent partnerships, the randomization of vaccination of couples could increase the efficiency and decrease the costs of vaccine trials.