Anup Amatya

Building Protective Factors to Offset Sexually Risky Behaviors among Black Youths: A Randomized Control Trial

OBJECTIVES To test the effectiveness of the CHAMP among black South Africans in KwaZulu-Natal, South Africa. METHODS A randomized control trial was conducted in KwaDedangendlale, South Africa, among youths (ages 9-13) and their families (245 intervention families rearing 281 children and 233 control families rearing 298 children). The CHAMPSA intervention targeted HIV risk behaviors by strengthening family relationship processes as well as targeting peer influences through enhancing social problem solving and peer negotiation skills for youths. RESULTS Among caregivers in the control and experimental conditions, significant intervention group differences were revealed regarding HIV transmission knowledge, less stigma toward HIV-infected people, caregiver monitoring-family rules, caregiver communication comfort, caregiver communication frequency and social networks. Among youths, data revealed that control and experimental groups were significantly different for children in AIDS transmission knowledge and less stigma toward HIV-infected people. CONCLUSIONS CHAMPSA enhances a significant number individual, family and community protective factors that can help youths avoid risky behaviors leading to HIV-positive status.

Meta-Analysis of Rare Binary Adverse Event Data

We examine the use of fixed-effects and random-effects moment-based meta-analytic methods for analysis of binary adverse-event data. Special attention is paid to the case of rare adverse events that are commonly encountered in routine practice. We study estimation of model parameters and between-study heterogeneity. In addition, we examine traditional approaches to hypothesis testing of the average treatment effect and detection of the heterogeneity of treatment effect across studies. We derive three new methods, a simple (unweighted) average treatment effect estimator, a new heterogeneity estimator, and a parametric bootstrapping test for heterogeneity. We then study the statistical properties of both the traditional and the new methods via simulation. We find that in general, moment-based estimators of combined treatment effects and heterogeneity are biased and the degree of bias is proportional to the rarity of the event under study. The new methods eliminate much, but not all, of this bias. The various estimators and hypothesis testing methods are then compared and contrasted using an example dataset on treatment of stable coronary artery disease.

Post-Approval Drug Safety Surveillance

Following the drug-approval process, concerns remain regarding the safety of new drugs that are introduced into the marketplace. In the case of rare adverse events, the number of subjects that are treated in randomized controlled trials is invariably inadequate to determine the safety of the new pharmaceutical. Identifying safety signals for new and/or existing drugs is a major priority in the protection of public health. Unfortunately, design, analysis, and available data are often quite limited for detecting in a timely fashion any potentially harmful effects of drugs. In this review, we examine a variety of approaches for determining the possibility of adverse drug reactions. Our review includes spontaneous reports, meta-analysis of randomized controlled clinical trials, ecological studies, and analysis of medical claims data. We consider both experimental design and analytic problems as well as potential solutions. Many of these methodologies are then illustrated through application to data on the possible relationship between taking antidepressants and increased risk of suicidality.

Deficits in complex motor functions, despite no evidence of procedural learning deficits, among HIV+ individuals with history of substance dependence.

Human immunodeficiency virus (HIV) and drugs of abuse affect common neural systems underlying procedural memory, including the striatum. The authors compared performance of 48 HIV seropositive (HIV+) and 48 HIV seronegative (HIV-) participants with history of cocaine and/or heroin dependence across multiple Trial Blocks of three procedural learning (PL) tasks: Rotary Pursuit (RP), Mirror Star Tracing (MST), and Weather Prediction (WP). Groups were well matched on demographic, psychiatric, and substance use parameters, and all participants were verified abstinent from drugs. Mixed model analyses of variance revealed that the individuals in the HIV+ group performed more poorly across all tasks, with a significant main effect of HIV serostatus observed on the Mirror Star Tracing and a trend toward significance obtained for the Rotary Pursuit task. No significant differences were observed on the Weather Prediction task. Both groups demonstrated significant improvements in performance across all three procedural learning tasks. It is important to note that no significant Serostatus x Trial Block interactions were observed on any task. Thus, the individuals in the HIV+ group tended to perform worse than those in the HIV- group across all trial blocks of procedural learning tasks with motor demands, but showed no differences in their rate of improvement across all tasks. These findings are consistent with HIV--associated deficits in complex motor skills, but not in procedural learning.

Mixed-effects Poisson regression analysis of adverse event reports: The relationship between antidepressants and suicide

A new statistical methodology is developed for the analysis of spontaneous adverse event (AE) reports from post-marketing drug surveillance data. The method involves both empirical Bayes (EB) and fully Bayes estimation of rate multipliers for each drug within a class of drugs, for a particular AE, based on a mixed-effects Poisson regression model. Both parametric and semiparametric models for the random-effect distribution are examined. The method is applied to data from Food and Drug Administration (FDA)s Adverse Event Reporting System (AERS) on the relationship between antidepressants and suicide. We obtain point estimates and 95 per cent confidence (posterior) intervals for the rate multiplier for each drug (e.g. antidepressants), which can be used to determine whether a particular drug has an increased risk of association with a particular AE (e.g. suicide). Confidence (posterior) intervals that do not include 1.0 provide evidence for either significant protective or harmful associations of the drug and the adverse effect. We also examine EB, parametric Bayes, and semiparametric Bayes estimators of the rate multipliers and associated confidence (posterior) intervals. Results of our analysis of the FDA AERS data revealed that newer antidepressants are associated with lower rates of suicide adverse event reports compared with older antidepressants. We recommend improvements to the existing AERS system, which are likely to improve its public health value as an early warning system.

Sample size determination for clustered count data

We consider the problem of sample size determination for count data. Such data arise naturally in the context of multicenter (or cluster) randomized clinical trials, where patients are nested within research centers. We consider cluster-specific and population-averaged estimators (maximum likelihood based on generalized mixed-effect regression and generalized estimating equations, respectively) for subject-level and cluster-level randomized designs, respectively. We provide simple expressions for calculating the number of clusters when comparing event rates of two groups in cross-sectional studies. The expressions we derive have closed-form solutions and are based on either between-cluster variation or intercluster correlation for cross-sectional studies. We provide both theoretical and numerical comparisons of our methods with other existing methods. We specifically show that the performance of the proposed method is better for subject-level randomized designs, whereas the comparative performance depends on the rate ratio for the cluster-level randomized designs. We also provide a versatile method for longitudinal studies. Three real data examples illustrate the results.

Subcellular Localization of p27 and Prostate Cancer Recurrence: Automated Digital Microscopy Analysis of Tissue Microarrays

Previous investigations have linked decreased nuclear expression of the cell cycle inhibitor p27 with poor outcome in prostate cancer. However, these reports are inconsistent regarding the magnitude of that association and its independence from other predictors. Moreover, cytoplasmic translocation of p27 has been proposed as a negative prognostic sign. Given the cost and accuracy limitations of manual scoring, particularly of tissue microarrays, we determined if laser-based fluorescence microscopy could provide automated analysis of p27 in both nuclear and cytoplasmic locations and, thus, clarify its significance as a prognostic biomarker. We constructed tissue microarrays covering 202 recurrent cases (rising prostate-specific antigen) and 202 matched controls without recurrence. Quadruplicate tumor samples encompassed 5 slides and 1616 cancer histospots. Cases and controls matched on age, Gleason grade, stage, and hospital. We immunolabeled epithelial cytoplasm with Alexafluor 647, p27 with Alexafluor 488, and nuclei with 4c6-diamidino-2-phenylindole·2HCl. Slides were scanned on an iCys laser scanning cytometer (CompuCyte Corp, Cambridge, MA). Nuclear crowding required a stereological approach--random arrays of circles (phantoms) were layered on images and the content of each phantom was analyzed in scatter plots. Both nuclear and cytoplasmic p27 were significantly lower in cases versus controls (P = .014 and P = .004, respectively). Regression models controlling for matching variables plus prostate-specific antigen showed strong linear trends for increased risk of recurrence with lower p27 in both nucleus and cytoplasm (highest versus lowest quartile; odds ratio, 0.35; P = .006). Manual scoring identified an inverse association between p27 expression and tumor grade but no independent association with recurrence. In conclusion, we developed an automated method for subcellular scoring of p27 without the need to segment individual cells. Our method identified a strong relationship, independent of tumor grade, stage, and prostate-specific antigen, between p27 expression--regardless of subcellular location--and prostate cancer recurrence. This relationship was not observed with manual scoring.

Development of a Nuclear Morphometric Signature for Prostate Cancer Risk in Negative Biopsies

Background Our objective was to develop and validate a multi-feature nuclear score based on image analysis of direct DNA staining, and to test its association with field effects and subsequent detection of prostate cancer (PCa) in benign biopsies. Methods Tissue sections from 39 prostatectomies were Feulgen-stained and digitally scanned (400×), providing maps of DNA content per pixel. PCa and benign epithelial nuclei were randomly selected for measurement of 52 basic morphometric features. Logistic regression models discriminating benign from PCa nuclei, and benign from malignant nuclear populations, were built and cross-validated by AUC analysis. Nuclear populations were randomly collected <1 mm or >5 mm from cancer foci, and from cancer-free prostates, HGPIN, and PCa Gleason grade 3–5. Nuclei also were collected from negative biopsy subjects who had a subsequent diagnosis of PCa and age-matched cancer-free controls (20 pairs). Results A multi-feature nuclear score discriminated cancer from benign cell populations with AUCs of 0.91 and 0.79, respectively, in training and validation sets of patients. In prostatectomy samples, both nuclear- and population-level models revealed cancer-like features in benign nuclei adjacent to PCa, compared to nuclei that were more distant or from PCa-free glands. In negative biopsies, a validated model with 5 variance features yielded significantly higher scores in cases than controls (P = 0.026). Conclusions A multifeature nuclear morphometric score, obtained by automated digital analysis, was validated for discrimination of benign from cancer nuclei. This score demonstrated field effects in benign epithelial nuclei at varying distance from PCa lesions, and was associated with subsequent PCa detection in negative biopsies. Impact This nuclear score shows promise as a risk predictor among men with negative biopsies and as an intermediate biomarker in Phase II chemoprevention trials. The results also suggest that subvisual disturbances in nuclear structure precede the development of pre-neoplastic lesions.

Simultaneous generation of multivariate mixed data with Poisson and normal marginals

The present paper develops a procedure for simulating multivariate data with count and continuous variables with a pre-specified correlation matrix. The count and continuous variables are assumed to have Poisson and normal marginals, respectively. The data generation mechanism is a combination of the normal to anything principle and a newly established connection between Poisson and normal correlations in the mixture. A step-by-step algorithm is provided and its performance is evaluated using two simulated and one real-data scenarios.

Statistical Methods for Drug Safety

Introduction Randomized Clinical Trials Observational Studies The Problem of Multiple Comparisons The Evolution of Available Data Streams The Hierarchy of Scientific Evidence Statistical Significance Summary Basic Statistical Concepts Relative Risk Odds Ratio Statistical Power Maximum Likelihood Estimation Non-Linear Regression Models Causal Inference Multi-Level Models Introduction Issues Inherent in Longitudinal Data Historical Background Statistical Models for the Analysis of Longitudinal and/or Clustered Data Causal Inference Introduction Propensity Score Matching Marginal Structural Models Instrumental Variables Differential Effects Analysis of Spontaneous Reports Proportional Reporting Ratio Bayesian Confidence Propagation Neural Network (BCPNN) Empirical Bayes Screening Multi-Item Gamma Poisson Shrinker Bayesian Lasso Logistic Regression Random-Effect Poisson Regression Discussion Meta-Analysis Fixed-Effect Meta-Analysis Random-Effect Meta-Analysis Maximum Marginal Likelihood/Empirical Bayes Method Bayesian Meta-Analysis Confidence Distribution Framework for Meta-Analysis Discussion Ecological Methods Time Series Methods State Space Model Change Point Analysis Mixed-Effects Poisson Regression Model Discrete-Time Survival Models Introduction Discrete-Time Ordinal Regression Model Discrete-Time Ordinal Regression Frailty Model Illustration Competing Risk Models Illustration Research Synthesis Introduction Three-Level Mixed-Effects Regression Models Analysis of Medical Claims Data Introduction Administrative Claims Observational Data Experimental Strategies Statistical Strategies Illustrations Conclusion Methods to Be Avoided Introduction Spontaneous Reports Vote Counting Simple Pooling of Studies Including Randomized and Non-Randomized Trials in Meta-Analysis Multiple Comparisons and Biased Reporting of Results Immortality Time Bias Summary and Conclusions Final Thoughts Bibliography Index