Search

Objective: Researchers in Health Sciences and Medicine often use cohort designs to study treatment effects and changes of outcome variables over time period. The costs of these studies can be reduced by choosing an optimal number of repeated measurements over time and by selecting cohorts of subjects more efficiently with optimal design procedures. The objective of this study is to provide evidence on how to design large-scale cohort studies with budget constraints as efficiently as possible., Study Design and Setting: A linear cost function for repeated measurements is proposed, and this cost function is used in the optimization procedure. For a given budget/cost, different designs for linear mixed-effects models are compared by means of their efficiency., Results: We found that adding more repeated measures is only beneficiary if the costs of selecting and measuring a new subject are much higher than the costs of obtaining an additional measurement for an already recruited subject. However, this gain in efficiency and power is not very large., Conclusion: Adding more cohorts or repeated measurements do not necessarily lead to a gain in efficiency of the estimated model parameters. A general guideline for the optimal choice of a cohort design in practice is required and we offer this guideline., (Copyright © 2011 Elsevier Inc. All rights reserved.)

Repeated measures (RM) and ANCOVA models are compared with respect to treatment effect estimation in randomized clinical trials with a pre- and a post-treatment measure. The covariance matrices of repeated measures are assumed to be I) homogeneous or II) heterogeneous across groups. In situation I, ANCOVA is preferred to RM, because the estimated variance of the treatment effect estimator is unbiased for ANCOVA and biased downwards for RM. In situation II, RM with Kenward and Roger's adjustment is preferred to ANCOVA, because the ANCOVA variance estimator does not correct for unknown pre-treatment expectation. The results are illustrated with an example.

The effect of number of repeated measures on the variance of the generalized least squares (GLS) treatment effect estimator is considered assuming a linearly divergent treatment effect, equidistant time-points and either a fixed number of subjects or a fixed study budget. The optimal combination of group sizes and number of repeated measures is calculated by minimizing this variance subject to a linear cost function. For a fixed number of subjects, the variance of the GLS treatment effect estimator can be decreased by adding intermediate measures per subject. This decrease is relatively large if a) the covariance structure is compound symmetric or b) the structure approaches compound symmetry and the correlation between two repeated measures does not exceed 0.80, or c) the correlation between two repeated measures does not exceed 0.60 if the time-lag goes to zero. In case the sample sizes and number of repeated measures are limited by budget constraints and the covariance structure includes a first-order auto-regression part, two repeated measures per subject yield highly efficient treatment effect estimators. Otherwise, it is more efficient to have more than two repeated measures. If the covariance structure is unknown, the optimal design based on a first-order auto-regressive structure with measurement error is preferable in terms of robustness against misspecification of the covariance structure. The numerical results are illustrated by three examples.

This article reviews three traditional methods for the analysis of multicenter trials with persons nested within clusters, i.e., centers, namely naïve regression (persons as units of analysis), fixed effects regression, and the use of summary measures (clusters as units of analysis), and compares these methods with multilevel regression. The comparison is made for continuous (quantitative) outcomes, and is based on the estimator of the treatment effect and its standard error, because these usually are of main interest in intervention studies. When the results of the experiment have to be valid for some larger population of centers, the centers in the intervention study have to present a random sample from this population and multilevel regression may be used. It is shown that the treatment effect and especially its standard error, are generally incorrectly estimated by the traditional methods, which should, therefore, not in general be used as an alternative to multilevel regression.

Finger and foot movement related potentials (MRPs) were recorded over the frontal, central and parietal areas of both hemispheres in 20 left-handed subjects. A unilateral flexion of the index finger and a plantar flexion of the foot were studied on either side. MRPs were larger preceding foot movements than preceding finger movements, their onset being earlier also. Prior to a finger flexion amplitudes were larger over the hemisphere contralateral to the movement than over the ipsilateral hemisphere. Preceding a foot movement, however, amplitudes were larger over the ipsilateral hemisphere. These results indicate differently localized sources of the MRPs in the two kinds of movement, in accordance with data obtained in right-handed subjects. No indication of a hemisphere effect, possibly related to motor dominance, was found in left-handers. This is in contrast to a slight hemisphere effect found with foot movements in right-handed subjects in the former study.