R and Shiny for Cost-Effectiveness Analyses: Why and When? A Hypothetical Case Study.

INTRODUCTION: Health economics models are typically built in Microsoft Excel® owing to its wide familiarity, accessibility and perceived transparency. However, given the increasingly rapid and analytically complex decision-making needs of both the pharmaceutical industry and the field of health economics and outcomes research (HEOR), the demands of cost-effectiveness analyses may be better met by the programming language R. OBJECTIVE: This case study provides an explicit comparison between Excel and R for contemporary cost-effectiveness analysis. METHODS: We constructed duplicate cost-effectiveness models using Excel and R (with a user interface built using the Shiny package) to address a hypothetical case study typical of contemporary health technology assessment. RESULTS: We compared R and Excel versions of the same model design to determine the advantages and limitations of the modelling platforms in terms of (i) analytical capability, (ii) data safety, (iii) building considerations, (iv) usability for technical and non-technical users and (v) model adaptability. CONCLUSIONS: The findings of this explicit comparison are used to produce recommendations for when R might be more suitable than Excel in contemporary cost-effectiveness analyses. We conclude that selection of appropriate modelling software needs to consider case-by-case modelling requirements, particularly (i) intended audience, (ii) complexity of analysis, (iii) nature and frequency of updates and (iv) anticipated model run time.

Application of Physiologically Based Pharmacokinetic (PBPK) Modeling Within a Bayesian Framework to Identify Poor Metabolizers of Efavirenz (PM), Using a Test Dose of Efavirenz.

Poor metabolisers of CYP2B6 (PM) require a lower dose of efavirenz because of serious adverse reactions resulting from the higher plasma concentrations associated with a standard dose. Treatment discontinuation is a common consequence in patients experiencing these adverse reactions. Such patients benefit from appropriate dose reduction, where efficacy can be achieved without the serious adverse reactions. PMs are usually identified by genotyping. However, in countries with limited resources genotyping is unaffordable. Alternative cost-effective methods of identifying a PM will be highly beneficial. This study was designed to determine whether a plasma concentration corresponding to a 600 mg test dose of efavirenz can be used to identify a PM. A physiologically based pharmacokinetic (PBPK) model was used to simulate the concentration-time profiles of a 600 mg dose of efavirenz in extensive metabolizers (EM), intermediate metabolizers (IM), and PM of CYP2B6. Simulated concentration-time data were used in a Bayesian framework to determine the probability of identifying a PM, based on plasma concentrations of efavirenz at a specific collection time. Results indicated that there was a high likelihood of differentiating a PM from other phenotypes by using a 24 h plasma concentration. The probability of correctly identifying a PM phenotype was 0.82 (true positive), while the probability of not identifying any other phenotype as a PM (false positive) was 0.87. A plasma concentration >1,000 ng/mL at 24 h post-dose is likely to be from a PM. Further verification of these findings using clinical studies is recommended.

Applications of linking PBPK and PD models to predict the impact of genotypic variability, formulation differences, differences in target binding capacity and target site drug concentrations on drug responses and variability.

This study aimed to demonstrate the added value of integrating prior in vitro data and knowledge-rich physiologically based pharmacokinetic (PBPK) models with pharmacodynamics (PDs) models. Four distinct applications that were developed and tested are presented here. PBPK models were developed for metoprolol using different CYP2D6 genotypes based on in vitro data. Application of the models for prediction of phenotypic differences in the pharmacokinetics (PKs) and PD compared favorably with clinical data, demonstrating that these differences can be predicted prior to the availability of such data from clinical trials. In the second case, PK and PD data for an immediate release formulation of nifedipine together with in vitro dissolution data for a controlled release (CR) formulation were used to predict the PK and PD of the CR. This approach can be useful to pharmaceutical scientists during formulation development. The operational model of agonism was used in the third application to describe the hypnotic effects of triazolam, and this was successfully extrapolated to zolpidem by changing only the drug related parameters from in vitro experiments. This PBPK modeling approach can be useful to developmental scientists who which to compare several drug candidates in the same therapeutic class. Finally, differences in QTc prolongation due to quinidine in Caucasian and Korean females were successfully predicted by the model using free heart concentrations as an input to the PD models. This PBPK linked PD model was used to demonstrate a higher sensitivity to free heart concentrations of quinidine in Caucasian females, thereby providing a mechanistic understanding of a clinical observation. In general, permutations of certain conditions which potentially change PK and hence PD may not be amenable to the conduct of clinical studies but linking PBPK with PD provides an alternative method of investigating the potential impact of PK changes on PD.

Deciding on success criteria for predictability of pharmacokinetic parameters from in vitro studies: an analysis based on in vivo observations.

Prediction accuracy of pharmacokinetic parameters is often assessed using prediction fold error, i.e., being within 2-, 3-, or n-fold of observed values. However, published studies disagree on which fold error represents an accurate prediction. In addition, "observed data" from only one clinical study are often used as the gold standard for in vitro to in vivo extrapolation (IVIVE) studies, despite data being subject to significant interstudy variability and subjective selection from various available reports. The current study involved analysis of published systemic clearance (CL) and volume of distribution at steady state (Vss) values taken from over 200 clinical studies. These parameters were obtained for 17 different drugs after intravenous administration. Data were analyzed with emphasis on the appropriateness to use a parameter value from one particular clinical study to judge the performance of IVIVE and the ability of CL and Vss values obtained from one clinical study to "predict" the same values obtained in a different clinical study using the n-fold criteria for prediction accuracy. The twofold criteria method was of interest because it is widely used in IVIVE predictions. The analysis shows that in some cases the twofold criteria method is an unreasonable expectation when the observed data are obtained from studies with small sample size. A more reasonable approach would allow prediction criteria to include clinical study information such as sample size and the variance of the parameter of interest. A method is proposed that allows the "success" criteria to be linked to the measure of variation in the observed value.