Explore the vast range of titles available.

Procedures for controlling the false positive rate when performing many hypothesis tests are commonplace in health and medical studies. Such procedures, most notably the Bonferroni adjustment, suffer from the problem that error rate control cannot be localized to individual tests, and that these procedures do not distinguish between exploratory and/or data-driven testing vs. hypothesis-driven testing. Instead, procedures derived from limiting false discovery rates may be a more appealing method to control error rates in multiple tests. Controlling the false positive rate can lead to philosophical inconsistencies that can negatively impact the practice of reporting statistically significant findings. We demonstrate that the false discovery rate approach can overcome these inconsistencies and illustrate its benefit through an application to two recent health studies. The false discovery rate approach is more powerful than methods like the Bonferroni procedure that control false positive rates. Controlling the false discovery rate in a study that arguably consisted of scientifically driven hypotheses found nearly as many significant results as without any adjustment, whereas the Bonferroni procedure found no significant results. Although still unfamiliar to many health researchers, the use of false discovery rate control in the context of multiple testing can provide a solid basis for drawing conclusions about statistical significance.

Background Limited evidence exists regarding the association of pre-existing mental health conditions in patients with stroke and stroke outcomes such as rehospitalization, mortality, and function. We examined the association between mental health conditions and rehospitalization, mortality, and functional outcomes in patients with stroke following inpatient rehabilitation. Methods Our observational study used the 2001 VA Integrated Stroke Outcomes database of 2162 patients with stroke who underwent rehabilitation at a Veterans Affairs Medical Center. Separate models were fit to our outcome measures that included 6-month rehospitalization or death, 6-month mortality post-discharge, and functional outcomes post inpatient rehabilitation as a function of number and type of mental health conditions. The models controlled for patient socio-demographics, length of stay, functional status, and rehabilitation setting. Results Patients had an average age of 68 years. Patients with stroke and two or more mental health conditions were more likely to be readmitted or die compared to patients with no conditions (OR: 1.44, p = 0.04). Depression and anxiety were associated with a greater likelihood of rehospitalization or death (OR: 1.33, p = 0.04; OR:1.47, p = 0.03). Patients with anxiety were more likely to die at six months (OR: 2.49, p = 0.001). Conclusions Patients with stroke with pre-existing mental health conditions may need additional psychotherapy interventions, which may potentially improve stroke outcomes post-hospitalization.

Financial models for asset and derivatives pricing, risk management, portfolio optimization, and asset allocation rely on volatility forecasts. Time-varying volatility models, such as generalized autoregressive conditional heteroscedasticity and stochastic volatility (SVOL), have been successful in improving forecasts over constant volatility models. We develop a new multivariate SVOL framework for modeling financial data that assumes covariance matrices stochastically varying through a Wishart process. In our formulation, scalar variances naturally extend to covariance matrices rather than to vectors of variances as in traditional SVOL models. Model fitting is performed using Markov chain Monte Carlo simulation from the posterior distribution. Because of the model's complexity, an efficiently designed Gibbs sampler is described that produces inferences with a manageable amount of computation. Our approach is illustrated on a multivariate time series of monthly industry portfolio returns. A test of the economic value of our model found that minimum-variance portfolios based on our SVOL covariance forecasts outperformed out-of-sample portfolios based on alternative covariance models, such as dynamic conditional correlations and factor-based covariances.

Paired comparison data in which the abilities or merits of the objects being compared may be changing over time can be modelled as a non-linear state space model. When the population of objects being compared is large, likelihood-based analyses can be too computationally cumbersome to carry out regularly. This presents a problem for rating populations of chess players and other large groups which often consist of tens of thousands of competitors. This problem is overcome through a computationally simple non-iterative algorithm for fitting a particular dynamic paired comparison model. The algorithm, which improves over the commonly used algorithm of Elo by incorporating the variability in parameter estimates, can be performed regularly even for large populations of competitors. The method is evaluated on simulated data and is applied to ranking the best chess players of all time, and to ranking the top current tennis-players.

In a medical setting, observational studies commonly involve patients who initiate a particular treatment (e.g., medication therapy) and others who do not, and the goal is to draw causal inferences about the effect of treatment on a time-to-event outcome. A difficulty with such studies is that the notion of a treatment initiation time is not well-defined for the control group. In this paper, we propose a Bayesian approach to estimate treatment effects in longitudinal observational studies where treatment is given by indication and thereby the exact timing of treatment is only observed for treated units. We present a framework for conceptualizing an underlying randomized experiment in this setting based on separating the time of indication for treatment, which we model using a latent state-space process, from the mechanism that determines assignment to treatment versus control. Next, we develop a two-step inferential approach that uses Markov Chain Monte Carlo (MCMC) posterior sampling to (1) infer the unobserved indication times for units in the control group, and (2) estimate treatment effects based on inferential conclusions from Step 1. This approach allows us to incorporate uncertainty about the unobserved indication times which induces uncertainty in both the selection of the control group and the measurement of time-to-event outcomes for these controls. We demonstrate our approach to study the effects on mortality of inappropriately prescribing phosphodiesterase type 5 inhibitors (PDE5Is), a medication contraindicated for certain types of pulmonary hypertension, using data from the Veterans Affairs (VA) health care system.

Randomized trials of pulmonary vasodilators in pulmonary hypertension due to left heart disease (Group 2) and lung disease (Group 3) have demonstrated potential for harm. Yet these therapies are commonly used in practice. Little is known of the effects of treatment outside of clinical trials. We aimed to establish outcomes of vasodilator treatment for Groups 2/3 pulmonary hypertension in real-world practice. We conducted a retrospective cohort study of 132,552 Medicare-eligible Veterans with incident Groups 2/3 pulmonary hypertension between 2006 and 2016, and a secondary nested case–control study. Our primary outcome was a composite of death by any cause or selected acute organ failures. In our cohort analysis, we calculated adjusted risks of time to our outcome using Cox proportional hazards models with facility-specific random effects. In our case–control analysis, we used logistic mixed-effects models to estimate the effect of any past, recent, and cumulative exposure on our outcome. From our cohort study, 3249 (2.5%) Veterans were exposed to pulmonary vasodilators. Exposure to vasodilators was associated with increased risk of our primary outcome, in both Group 3 (HR: 1.58 (95% CI: 1.37–1.82)) and Group 2 (HR: 1.26 (95% CI: 1.12–1.41)) pulmonary hypertension patients. The case–control study determined odds of our outcome increased by 11% per year of exposure (OR: 1.11 (95% CI: 1.07–1.16)). Treating Groups 2/3 pulmonary hypertension with vasodilators in clinical practice is associated with increased risk of harm. This extension of trial findings to a real-world setting offers further evidence to limit use of vasodilators in Groups 2/3 pulmonary hypertension outside of clinical trials.

Background The U.S. Food and Drug Administration (FDA) recently linked antiepileptic drug (AED) exposure to suicide-related behaviors based on meta-analysis of randomized clinical trials. We examined the relationship between suicide-related behaviors and different AEDs in older veterans receiving new AED monotherapy from the Veterans Health Administration (VA), controlling for potential confounders. Methods VA and Medicare databases were used to identify veterans 66 years and older, who received a) care from the VA between 1999 and 2004, and b) an incident AED (monotherapy) prescription. Previously validated ICD-9-CM codes were used to identify suicidal ideation or behavior (suicide-related behaviors cases), epilepsy, and other conditions previously associated with suicide-related behaviors. Each case was matched to controls based on prior history of suicide-related behaviors, year of AED prescription, and epilepsy status. Results The strongest predictor of suicide-related behaviors (N = 64; Controls N = 768) based on conditional logistic regression analysis was affective disorder (depression, anxiety, or post-traumatic stress disorder (PTSD); Odds Ratio 4.42, 95% CI 2.30 to 8.49) diagnosed before AED treatment. Increased suicide-related behaviors were not associated with individual AEDs, including the most commonly prescribed AED in the US - phenytoin. Conclusion Our extensive diagnostic and treatment data demonstrated that the strongest predictor of suicide-related behaviors for older patients newly treated with AED monotherapy was a previous diagnosis of affective disorder. Additional, research using a larger sample is needed to clearly determine the risk of suicide-related behaviors among less commonly used AEDs.

Objectives. We examined (1) mental and physical health symptoms and functioning in US veterans within 1 year of returning from deployment, and (2) differences by gender, service component (Active, National Guard, other Reserve), service branch (Army, Navy, Air Force, Marines), and deployment operation (Operation Enduring Freedom/Operation Iraqi Freedom [OEF/OIF]). Methods. We surveyed a national sample of 596 OEF/OIF veterans, oversampling women to make up 50% of the total, and National Guard and Reserve components to each make up 25%. Weights were applied to account for stratification and nonresponse bias. Results. Mental health functioning was significantly worse compared with the general population; 13.9% screened positive for probable posttraumatic stress disorder, 39% for probable alcohol abuse, and 3% for probable drug abuse. Men reported more alcohol and drug use than did women, but there were no gender differences in posttraumatic stress disorder or other mental health domains. OIF veterans reported more depression or functioning problems and alcohol and drug use than did OEF veterans. Army and Marine veterans reported worse mental and physical health than did Air Force or Navy veterans. Conclusions. Continuing identification of veterans at risk for mental health and substance use problems is important for evidence-based interventions intended to increase resilience and enhance treatment.

We present a method of ranking U.S. undergraduate programs based on students’ revealed preferences. When a student chooses a college among those that have admitted him, that college “wins” his “tournament.” Our method efficiently integrates the information from thousands of such tournaments. We implement the method using data from a national sample of high-achieving students. We demonstrate that this ranking method has strong theoretical properties, eliminating incentives for colleges to adopt strategic, inefficient admissions policies to improve their rankings. We also show empirically that our ranking is (1) not vulnerable to strategic manipulation; (2) similar regardless of whether we control for variables, such as net cost, that vary among a college’s admits; (3) similar regardless of whether we account for students selecting where to apply, including Early Decision. We exemplify multiple rankings for different types of students who have preferences that vary systematically.

Recent public debate on costs and effectiveness of health care in the United States has generated a growing emphasis on \"profiling\" of medical care providers. The process of profiling involves comparing resource use and quality of care among medical providers to a community or a normative standard. This is valuable for targeting quality improvement strategies. For example, hospital profiles may be used to determine whether institutions deviate in important ways in the process of care they deliver. In this article we propose a class of performance indices to profile providers. These indices are based on posterior tail probabilities of relevant model parameters that indicate the degree of poor performance by a provider. We apply our performance indices to profile hospitals on the basis of 30-day mortality rates for a cohort of elderly heart attack patients. The analysis used data from 96 acute care hospitals located in one state and accounted for patient and hospital characteristics using a hierarchical logistic regression model. We used Markov chain Monte Carlo methods to fit the model and to obtain performance indices of interest. In particular, we estimated the posterior probability that mortality at the ith hospital is 1-1/2 times the median mortality rate over all the hospitals in the state. We also calculated the posterior probability that the deviation in average risk-adjusted and \"standardized\" mortality at the ith hospital is \"large.\" We compare the results of evaluating hospitals based on our performance indices to those obtained using conventional measures. With 30-day risk-adjusted mortality rates ranging from 12% to 14%, one-quarter of the hospitals had posterior probabilities that hospital-specific mortality was 1-1/2 times the median mortality rate greater than 15%. The posterior probability of a large difference between risk-adjusted and standardized mortality rates was less than 6% for three-quarters of the hospitals we examined. Although there were differences in the evaluation of each hospital by the various criteria, one hospital consistently emerged as having the worst performance by all criteria.