A clinical pharmacology trial, which examines the safety and pharmacodynamics of an investigational drug, is typically the first time that the drug is administered to humans. We are, therefore, often forced to maintain some restrictions on the trial conditions; for example, incremental doses in succeeding stages may be necessary when safety is concerned, and the repetition of the treatment on the same subject may be restricted in terms of the imposition on and convenience of subjects. The present paper investigated optimal trial designs under such restrictions, adopting Ds-optimality (D-optimality for subset) as the criterion. In order to identify the optimal design, all admissible designs that satisfy the restrictions were listed in a lexical order and their optimality was compared. As a result, it was revealed that a relatively high number of subjects were allocated lower doses in the optimal design when the increment restriction was regarded as relevant, whereas more reasonable designs were identified as optimal when the restrictions were modified.
In the later phases of selecting candidate chemicals for drug development, pharmacological experiments are conducted using animal organs or human peripheral blood cells. In these experiments, data analysis is often performed on the basis of mixed-effect models so that the individuality effect can be incorporated in the dose-response relationship. This paper studied the influence of the incorporation of random effects on parameters in a logistic model intended for analysis, with the assumption that the dose-response relationship is really described by a four-parameter logistic model. Using Monte-Carlo simulation experiments, we compared the performances of eight models in which various mixed-effects were incorporated. In each of the eight analysis models, five methods of calculation, namely, the standard two-stage method (STS), first-order approximation method (FOA), Laplacian approximation method (LAP), Monte Carlo integration method (MCI), and Gaussian quadrature method (GAU), were applied to the simulation data. The eight analysis models and five estimation methods were compared, using estimability and the deviation of estimates from the true value as the criteria. The results revealed the analysis model incorporating the random effect on only the maximum response to be the best. The results also indicated that the FOA, LAP, MCI, and GAU methods had almost the same performances for this analysis model. The authors recommend LAP as the preferred method because of the simplicity of its calculation.
The intraclass and concordance correlation coefficients (ICC and CCC) are popular indices of reliability or agreement for continuous variables. While various versions of ICC are used according to study design, the CCC is solely used as an index of agreement among fixed raters. However, we considered other versions of CCC in connection with all types of ICC, as classified by McGraw and Wong (Psychological Methods1: 30-46, 1996), and examined the similarities and differences between ICCs and CCCs. Because they were found to be compatibly similar to each other, it is considered that CCC has a wider range of applicability and ICC shares the framework of CCC including factors of precision and accuracy as well as applicability to pairwise sub-analysis among multiple raters. Although it is said that some version of ICC should selectively be used according to study purpose and design, useful information is added by the combined use with the other ICCs and CCCs and by pairwise sub-analysis. For a better understanding of the study results, additional information concerning the possible factors which could lower or enhance precision and/or accuracy will be necessary.
The means and correlation structures of four basic tastes data for Japanese and Chinese men and women, are compared. Tasting data were measured by a filter paper disk method, and recorded as missing values when subjects were not able to recognize a highest concentration of a taste substance solution. The missing values are recorded as counts, so the tasting data is expressed as censored data. A few statistics, i.e. mean vectors, variance-covariance matirices, and correlation matirices, are estimated by a multivariate statistical model for censored data to avoid biases on the estimation. Moreover, a correlative map among four tastes for men and women of Japan and China is drawn by using a principal components analysis. These results suggest that: (i) An appropriate statistical model is needed for analysing data with missing values. (ii) Positive correlations among the four basic tastes were estimated. (iii) The differences in experience and custom of eating appear as taste sensitivities and mean and correlation structures, such that the correlation between men and women is positive, while between Japanese and Chinese is negative. (iv) The correlation structures suggest a kind of an ordering of the tastes for Japanese and Chinese, and for men and women.