This chapter provides an overview of recent developments in the design and analysis of cross-over trials. We first consider the analysis of the trial that compares two treatments, A and B, over two periods and where the subjects are randomized to the treatment sequences AB and BA. We make the distinction between fixed and random effects models and show how these models can easily be fitted using modern software. Issues with fitting and testing for a difference in carry-over effects are described and the use of baseline measurements is discussed. Simple methods for testing for a treatment difference when the data are binary are also described. Various designs with two or more treatments but with three or four periods are then described and compared. These include the balanced and partially balanced designs for three or more treatments and designs for factorial treatment combinations. Also described are nearly balanced and nearly strongly balanced designs. Random subject-effects models for the designs with two or more treatments are described and methods for analysing non-normal data are also given. The chapter concludes with a description of the use of cross-over designs in the testing of bioequivalence.

/pdf/design-and-analysis-of-cross-over-trials-4ootbxi8hy.pdf

Design and Analysis of Cross-Over Trials

Therapeutic drug monitoring (TDM) is the quantification and interpretation of drug concentrations in blood to optimize pharmacotherapy. It considers the interindividual variability of pharmacokinetics and thus enables personalized pharmacotherapy. In psychiatry and neurology, patient populations that may particularly benefit from TDM are children and adolescents, pregnant women, elderly patients, individuals with intellectual disabilities, patients with substance abuse disorders, forensic psychiatric patients or patients with known or suspected pharmacokinetic abnormalities. Non-response at therapeutic doses, uncertain drug adherence, suboptimal tolerability, or pharmacokinetic drug-drug interactions are typical indications for TDM. However, the potential benefits of TDM to optimize pharmacotherapy can only be obtained if the method is adequately integrated in the clinical treatment process. To supply treating physicians and laboratories with valid information on TDM, the TDM task force of the Arbeitsgemeinschaft fur Neuropsychopharmakologie und Pharmakopsychiatrie (AGNP) issued their first guidelines for TDM in psychiatry in 2004. After an update in 2011, it was time for the next update. Following the new guidelines holds the potential to improve neuropsychopharmacotherapy, accelerate the recovery of many patients, and reduce health care costs.

/pdf/consensus-guidelines-for-therapeutic-drug-monitoring-in-18b8y5mu39.pdf

Consensus Guidelines for Therapeutic Drug Monitoring in Neuropsychopharmacology: Update 2017.

INTRODUCTION Regulatory Requirement Basic Considerations Procedures for Sample Size Calculation Aims and Structure of the Book CONSIDERATIONS PRIOR TO SAMPLE SIZE CALCULATION Confounding and Interaction One-Sided Test Versus Two-Sided Test Crossover Design Versus Parallel Design Subgroup/Interim Analyses Data Transformation Practical Issues COMPARING MEANS One-Sample Design Two-Sample Parallel Design Two-Sample Crossover Design Multiple-Sample One-Way ANOVA Multiple-Sample Williams Design Practical Issues LARGE SAMPLE TESTS FOR PROPORTIONS One-Sample Design Two-Sample Parallel Design Two-Sample Crossover Design One-Way Analysis of Variance Williams Design Relative Risk - Parallel Design Relative Risk - Crossover Design Practical Issues EXACT TESTS FOR PROPORTIONS Binomial Test Fisher's Exact Test Optimal Multiple-Stage Designs for Single Arm Trials Flexible Designs for Multiple-Arm Trials Remarks TESTS FOR GOODNESS-OF-FIT AND CONTINGENCY TABLES Tests for Goodness-of-Fit Test for Independence -Single Stratum Test for Independence -Multiple Strata Test for Categorical Shift Carry-Over Effect Test Practical Issues COMPARING TIME-TO-EVENT DATA Basic Concepts Exponential Model Cox's Proportional Hazards Model Weighted Log-Rank Test Practical Issues GROUP SEQUENTIAL METHODS Pocock's Test O'Brien and Fleming's Test Wang and Tsiatis' Test Inner Wedge Test Binary Variables Time-to-Event Data Alpha Spending Function Sample Size Re-Estimation Conditional Power Practical Issues COMPARING VARIABILITIES Comparing Intra-Subject Variabilities Comparing Intra-Subject CVs Comparing Inter-Subject Variabilities Comparing Total Variabilities Practical Issues BIOEQUIVALENCE TESTING Bioequivalence Criteria Average Bioequivalence Population Bioequivalence Individual Bioequivalence In Vitro Bioequivalence NONPARAMETRICS Violation of Assumptions One-Sample Location Problem Two-Sample Location Problem Test for Independence Practical Issues SAMPLE SIZE CALCULATION IN OTHER AREAS Dose Response Studies ANOVA with Repeated Measures Quality of Life Bridging Studies Vaccine Clinical Trials Appendix: Tables of Quantiles References Index

Sample Size Calculations in Clinical Research

Therapeutic drug monitoring (TDM), i. e., the quantification of serum or plasma concentrations of medications for dose optimization, has proven a valuable tool for the patient-matched psychopharmacotherapy. Uncertain drug adherence, suboptimal tolerability, non-response at therapeutic doses, or pharmacokinetic drug-drug interactions are typical situations when measurement of medication concentrations is helpful. Patient populations that may predominantly benefit from TDM in psychiatry are children, pregnant women, elderly patients, individuals with intelligence disabilities, forensic patients, patients with known or suspected genetically determined pharmacokinetic abnormalities or individuals with pharmacokinetically relevant comorbidities. However, the potential benefits of TDM for optimization of pharmacotherapy can only be obtained if the method is adequately integrated into the clinical treatment process. To promote an appropriate use of TDM, the TDM expert group of the Arbeitsgemeinschaft fur Neuropsychopharmakologie und Pharmakopsychiatrie (AGNP) issued guidelines for TDM in psychiatry in 2004. Since then, knowledge has advanced significantly, and new psychopharmacologic agents have been introduced that are also candidates for TDM. Therefore the TDM consensus guidelines were updated and extended to 128 neuropsychiatric drugs. 4 levels of recommendation for using TDM were defined ranging from “strongly recommended” to “potentially useful”. Evidence-based “therapeutic reference ranges” and “dose related reference ranges” were elaborated after an extensive literature search and a structured internal review process. A “laboratory alert level” was introduced, i. e., a plasma level at or above which the laboratory should immediately inform the treating physician. Supportive information such as cytochrome P450 substrate- and inhibitor properties of medications, normal ranges of ratios of concentrations of drug metabolite to parent drug and recommendations for the interpretative services are given. Recommendations when to combine TDM with pharmacogenetic tests are also provided. Following the guidelines will help to improve the outcomes of psychopharmacotherapy of many patients especially in case of pharmacokinetic problems. Thereby, one should never forget that TDM is an interdisciplinary task that sometimes requires the respectful discussion of apparently discrepant data so that, ultimately, the patient can profit from such a joint effort.

/pdf/agnp-consensus-guidelines-for-therapeutic-drug-monitoring-in-18q2ciintb.pdf

AGNP Consensus Guidelines for Therapeutic Drug Monitoring in Psychiatry: Update 2011

The Hill equation was first introduced by A.V. Hill to describe the equilibrium relationship between oxygen tension and the saturation of haemoglobin. In pharmacology, the Hill equation has been extensively used to analyse quantitative drug-receptor relationships. Many pharmacokinetic-pharmacodynamic models have used the Hill equation to describe nonlinear drug dose-response relationships. Although the Hill equation is widely used, its many properties are not all well known. This article aims at reviewing the various properties of the Hill equation. The descriptive aspects of the Hill equation, in particular mathematical and graphical properties, are examined, and related to Hill's original work. The mechanistic aspect of the Hill equation, involving a strong connection with the Guldberg and Waage law of mass action, is also described. Finally, a probabilistic view of the Hill equation is examined. Here, we provide some new calculation results, such as Fisher information and Shannon entropy, and we introduce multivariate probabilistic Hill equations. The main features and potential applications of this probabilistic approach are also discussed. Thus, within the same formalism, the Hill equation has many different properties which can be of great interest for those interested in mathematical modelling in pharmacology and biosciences.

The Hill equation: a review of its capabilities in pharmacological modelling

We study a class of selection problems for which the 'more sampling-more information' principle is true, but not obvious. The case examined, which includes sampling from a multinomial distribution and from a hypergeometric distribution as special cases, concerns a one-sample procedure to select an m-subset of thef largest denominations in a multivariate Polya urn. For a given sample size n, the selection procedure takes the m denominations with highest frequencies, where ties are broken by randomization. Let +(n) be the probability of making a correct selection when n is the sample size. According to the 'more sampling-more information' principle, we expect 0(n + 1) > 0(n) for all n > 0. This inequality is proved.

A class of selection problems for which more sampling is more informative

Ridge regression is often discussed as an estimation procedure for producing estimators which are biased but with a smaller mean squared error than the usual least square estimators. In this paper we show that this procedure can also be used to reflect the nature of dependency among a set of highly collinear regressor variables. In particular, we prove that, when data are severely multicollinear, the ridge estimators can be made very close to the principal component estimators. Examples are given to illustrate the point.

Ridge regression from principal component point of view

In medical practice, it has been realized that noncompliance may have an impact on the therapeutic effect of a drug therapy, regardless of race, gender, and education of patients. Therefore, it is of interest to study the impact of noncompliance on drug absorption through in vivo testing. Efron and Feldman examined dose-response relationship when noncompliance is an issue. In this paper, we study bias and variation induced by noncompliance for pharmacokinetic parameters such as the area under the curve (AUC). We use bioequivalence testing as an example to demonstrate that a false conclusion could be drawn if one ignores the effect of noncompliance. We propose a new test for the assessment of bioequivalence in multiple doses. The proposed test appears to have a substantial improvement over the usual two one-sided tests based on a simulation study.

An adjusted two one-sided t-test for the assessment of bioequivalence with multiple doses

The overparameterized and cell means models are alternative formulations of the conventional linear model that underlies designed experiments with fixed effects. Each has its relative advantages and adherents. Statistical software packages generally adopt just one of these formulations, and transition between the two is not routinely undertaken. In this paper we present a modified sweep algorithm for undertaking translation in either direction. The algorithm is illustrated with a published data set from Milliken and Johnson (1984). We demonstrate how to generate the cell means model hypothesis corresponding to any testable hypothesis in the overparameterized model, and indicate that the interpretability of the corresponding sums of squares depends heavily on the selected model.

A modified sweep algorithm for interchanging between overparameterized and cell means linear models

We consider a logistic transition model for analyzing strings of correlated binary (0,1) data, which occur in behavioral teratology experiments targeted at testing learning impairment in rat pups. Estimating the learning parameter in the model has led us to the following question: How to find the joint distribution of the counts of overlapping occurrences of {00}, {01}, {10}, and {11} in a string, when the string is ended when k, k>2, successes in a row are observed? Similar questions have been tackled in the literature using a variety of methods. Here, we use Markov potential theory to find the joint moment generating function, and derive an explicit formula for the joint probability distribution. This result is then used to find the asymptotic standard error of a consistent estimator for the learning parameter in the logistic transition model. We illustrate our results with a real behavioral teratology dataset.

Francis Hsuan

Papers

A class of selection problems for which more sampling is more informative

Ridge regression from principal component point of view

An adjusted two one-sided t-test for the assessment of bioequivalence with multiple doses

A modified sweep algorithm for interchanging between overparameterized and cell means linear models

The Distribution of Pattern Counts in Markov Chains With Stopping Rules: With Application to Behavioral Teratology Experiments