David M. Zucker

An Analysis of Variance Pitfall: The Fixed Effects Analysis in a Nested Design:

This paper illustrates a major pitfall with fixed effects analysis of variance in the nested design. The pitfall is not clearly identified and discussed in standard texts.

In Defense of the Randomized Controlled Trial for Health Promotion Research

The overwhelming evidence about the role lifestyle plays in mortality, morbidity, and quality of life has pushed the young field of modern health promotion to center stage. The field is beset with intense debate about appropriate evaluation methodologies. Increasingly, randomized designs are considered inappropriate for health promotion research. We have reviewed criticisms against randomized trials that raise philosophical and practical issues, and we will show how most of these criticisms can be overcome with minor design modifications. By providing rebuttal to arguments against randomized trials, our work contributes to building a sound methodological base for health promotion research.

Statistical design of the child and adolescent trial for cardiovascular health (catch): Implications of cluster randomization☆

This paper describes some statistical considerations for the Child and Adolescent Trial for Cardiovascular Health (CATCH), a large-scale community health trial sponsored by the National Heart, Lung, and Blood Institute. The trial involves randomization of entire schools rather than individual students to the experimental arms. The paper discussed the implications of this form of randomization for the design and analysis of the trial. The power calculations and analysis plan for the trial are presented in detail. The handling of outmigrating and immigrating students is also discussed.

Restricted Mean Life with Covariates: Modification and Extension of a Useful Survival Analysis Method

Abstract Karrison has presented a method for covariate-adjusted comparison of two groups with respect to survival when the group effect is expected to not be of proportional hazards form. This article describes a simplified procedure for implementing Karrisons basic approach, along with an extended version designed to achieve robustness against misspecification in the underlying analytical model. A simulation study of these techniques is presented, and the techniques are illustrated with an example taken from Karrison.

Computerized cognitive testing battery identifies mild cognitive impairment and mild dementia even in the presence of depressive symptoms

Cognitive and depressive symptoms co-occur, complicating detection of mild cognitive impairment (MCI) and early dementia. In this study, discriminant validity of a novel computerized cognitive battery for MCI detection was evaluated after covariation for depressive symptom severity. In addition to the computerized battery, participants at two sites received the 30-item self-administered Geriatric Depression Scale (GDS; n = 72); those at two other centers received the observer-administered Cornell Scale for Depression in Dementia (CSDD; n = 88). In both cohorts, a Global Cognitive Score and memory, executive function, visual spatial, and verbal index scores discriminated among cognitively healthy, MCI, and mild dementia groups after covariation for GDS or CSDD, respectively (p < 0.05). Thus, the computerized battery for detection of mild impairment is robust to comorbid depressive symptoms, supporting its clinical utility in identifying neurodegenerative disease even in elderly with depression.

Research Design and Analysis Issues

The overall perspective of this article is the need for researchers of school-based health promotion to make more progressive use of existing statistical methods to improve both the rigor and the efficiency of school-based experiments. Investigative teams working in the school setting have an advantage over health education researchers working in communities or worksites in that they have greater choice of the experimental unit and usually easier access to large clusters of units. They are, therefore, in a position to make optimal use of a wide variety of experimental designs and observation strategies. In order to make full use of this advantage, however, researchers in this field need to assemble multidisciplinary teams, including researchers from other fields who are not restricted by the traditional approaches evident in many of the current school studies. Such teams should include statisticians and epidemiologists who have broadly-based experimental design and survey experience to work with the health educators and other behavioral and biomedical scientists.

A Pseudo–Partial Likelihood Method for Semiparametric Survival Regression With Covariate Errors

This article presents an estimator for the regression coefficient vector in the Cox proportional hazards model with covariate error. The estimator is obtained by maximizing a likelihood-type function similar to the Cox partial likelihood. The likelihood function involves the cumulative baseline hazard function, for which a simple estimator is substituted. The method is capable of handling general covariate error structures; it is not restricted to the independent additive error model. It can be applied to studies with either an external or internal validation sample, and also to studies with replicate measurements of the surrogate covariate. The estimator is shown to be consistent and asymptotically normal, and an estimate of the asymptotic covariance matrix is derived. Some extensions to general transformation survival models are indicated. Simulation studies are presented for a setup with a single error-prone binary covariate and a setup with a single error-prone normally distributed covariate. These simulation studies show that the method typically produces estimates with low bias and confidence intervals with accurate coverage rates. Efficiency results relative to fully parametric maximum likelihood are also presented. The method is applied to data from the Framingham Heart Study.

Analysis Issues in School-Based Health Promotion Studies

In school-based research, usually the nature of the intervention or other practical factors indicate that assignment of treatment be done by school or classroom rather than by individual student. In this situation, randomization of schools (or classrooms) and analysis by school means (or classroom means) provide a firm statistical basis for internal validity of the study. When the number of schools available is small, this approach is not practicable, and therefore the investigator must be both more creative in developing solutions and more cautious in interpreting the results. This article provides a number of suggestions which the authors hope will assist the field in dealing with such circumstances. The authors stress that the best approach to assessing treatment effect is a well-designed, properly analyzed randomized experiment. The suggestions in this article attempt to indicate how one might make the most that one can from more limited data.

Differences in quality of life among patients receiving dialysis replacement therapy at seven medical centers

The purpose of this study is to investigate the variations in quality of life (QOL) among patients with end-stage renal disease (ESRD) who are receiving replacement therapy in several dialysis centers. This observational study includes interviews with nurses and data extraction from medical charts for all 680 adults who had been on dialysis therapy for more than 4 weeks in seven dialysis centers. By using multivariate analysis, we generated a model to explain the variance in QOL as measured by the QL index score (developed by Spitzer et al., J Chronic Dis 1981; 34:585-597) among patients pooled from all centers. The expected mean QL index score and 95% confidence interval were computed for each dialysis center. Centers with observed mean QL index scores outside of the expected confidence range were marked as possible outliers. We found the following patient attributes to be independently associated with QOL: age, education, occupation, and certain comorbidities (e.g., diabetes, stroke). After adjustment for case mix, we could identify four outlier centers. After further adjustment for albumin in serum, a possible process indicator, two centers were no longer considered as outliers. These findings indicate that the variance in QOL of ESRD patients at different centers is not entirely explained by known case-mix factors. Further research should explore whether such variations are related to dissimilarity in the process of care at different centers.

Meta-analysis of Parental Protection of Children From Tobacco Smoke Exposure

BACKGROUND AND OBJECTIVE: Worldwide, roughly 40% of children are exposed to the damaging and sometimes deadly effects of tobacco smoke. Interventions aimed at reducing child tobacco smoke exposure (TSE) have shown mixed results. The objective of this study was to perform a systematic review and meta-analysis to quantify effects of interventions aimed at decreasing child TSE. METHODS: Data sources included Medline, PubMed, Web of Science, PsycNet, and Embase. Controlled trials that included parents of young children were selected. Two reviewers extracted TSE data, as assessed by parentally-reported exposure or protection (PREP) and biomarkers. Risk ratios and differences were calculated by using the DerSimonian and Laird random-effects model. Exploratory subgroup analyses were performed. RESULTS: Thirty studies were included. Improvements were observed from baseline to follow-up for parentally-reported and biomarker data in most intervention and control groups. Interventions demonstrated evidence of small benefit to intervention participants at follow-up (PREP: 17 studies, n = 6820, relative risk 1.12, confidence interval [CI] 1.07 to 1.18], P < .0001). Seven percent more children were protected in intervention groups relative to control groups. Intervention parents smoked fewer cigarettes around children at follow-up than did control parents (P = .03). Biomarkers (13 studies, n = 2601) at follow-up suggested lower child exposure among intervention participants (RD −0.05, CI −0.13 to 0.03, P = .20). CONCLUSIONS: Interventions to prevent child TSE are moderately beneficial at the individual level. Widespread child TSE suggests potential for significant population impact. More research is needed to improve intervention effectiveness and child TSE measurement.