General considerations of model-based meta-analysis / 中国临床药理学与治疗学

With the increasing cost of drug development and clinical trials, it is of great value to make full use of all kinds of data to improve the efficiency of drug development and to provide valid information for medication guidelines. Model-based meta-analysis (MBMA) combines mathematical models with meta-analysis to integrate information from multiple sources (preclinical and clinical data, etc.) and multiple dimensions (targets/mechanisms, pharmacokinetics/pharmacodynamics, diseases/indications, populations, regimens, biomarkers/efficacy/safety, etc.), which not only provides decision-making for all key points of drug development, but also provides effective information for rational drug use and cost-effectiveness analysis. The classical meta-analysis requires high homogeneity of the data, while MBMA can combine and analyze the heterogeneous data of different doses, different time courses, and different populations through modeling, so as to quantify the dose-effect relationship, time-effect relationship, and the relevant impact factors, and thus the efficacy or safety features at the level of dose, time and covariable that have not been involved in previous studies. Although the modeling and simulation methods of MBMA are similar to population pharmacokinetics/pharmacodynamics (Pop PK/PD), compared with Pop PK/PD, the advantage of MBMA is that it can make full use of literature data, which not only improves the strength of evidence, but also can answer the questions that have not been proved or can not be answered by a single study. At present, MBMA has become one of the important methods in the strategy of model-informed drug development (MIDD). This paper will focus on the application value, data analysis plan, data acquisition and processing, data analysis and reporting of MBMA, in order to provide reference for the application of MBMA in drug development and clinical practice.

Mental health status and internet-surfing behavior among rural adolescents / 中国心理卫生杂志

Objective:To investigate the prevalence and influencing factors of mental health status and internet-surfing behavior among rural adolescents in Sichuan province,and explore the mutual effects between mental health status and internet-surfing behavior.Methods:Totally 2745 junior and senior high school students of grade seven and grade ten from two rural schools were selected.Mental health status,self-esteem and social support of students were assessed with Mental Health Inventory of Middle-school students (MMHI),Rosenberg self-esteem scale (SES) and social support rating scale (SSRS) respectively.Demographic characteristics,internet-surfing behavior were obtained using cross-sectional survey.Non-recursive structural equation model was applied to analyze the effects of other variables on mental health status and internet-surfing behavior and the mutual effects between them.Results:The mean score of MMHIwas (2.1 ±0.7),and the dimensions including academic stress (2.4 ±0.9),emotional instability (2.4 ±0.8) and anxiety (2.4 ± 1.0) got the top three.The total prevalence of long-time internet-surfing was 32.8％ (899/2745).The structural equation model showed that female and increasing age had positive effects on score of MMHI (β =0.058,0.058,P ＜ 0.001),and male and increasing age positively influenced internet-surfing behavior (β =-0.171,0.149,P ＜ 0.001).The scores of SES and S SRS were directly negatively related to the score of MMHI (β =-0.300,-0.263,P ＜ 0.001),and indirectly negatively affect internetsurfing behavior through the mediating effect of mental health (βi =-0.074,-0.065,P ＜ 0.010).The score of MMHI had positively effects on long-time internet-surfing behavior (β =0.246,P ＜ 0.001),and long-time internetsurfing behavior had positively effects on the score of MMHI in reverse (β =0.008,P ＜ 0.001),but much weaker.Conclusion:There are mild mental health problem among rural adolescents,and internet-surfing behaviors are prevalent among this population.Poor mental health and long-time internet-surfing behavior are risk factors mutually.

The therapeutic drug monitoring network server of tacrolimus for Chinese renal transplant patients / 药学学报

This study is to develop a therapeutic drug monitoring (TDM) network server of tacrolimus for Chinese renal transplant patients, which can facilitate doctor to manage patients' information and provide three levels of predictions. Database management system MySQL was employed to build and manage the database of patients and doctors' information, and hypertext mark-up language (HTML) and Java server pages (JSP) technology were employed to construct network server for database management. Based on the population pharmacokinetic model of tacrolimus for Chinese renal transplant patients, above program languages were used to construct the population prediction and subpopulation prediction modules. Based on Bayesian principle and maximization of the posterior probability function, an objective function was established, and minimized by an optimization algorithm to estimate patient's individual pharmacokinetic parameters. It is proved that the network server has the basic functions for database management and three levels of prediction to aid doctor to optimize the regimen of tacrolimus for Chinese renal transplant patients.

Comparison study of model evaluation methods: normalized prediction distribution errors vs. visual predictive check / 药学学报

The objective of this study is to compare the normalized prediction distribution errors (NPDE) and the visual predictive check (VPC) on model evaluation under different study designs. In this study, simulation method was utilized to investigate the capability of NPDE and VPC to evaluate the models. Data from the false models were generated by biased parameter typical value or inaccurate parameter inter-individual variability after single or multiple doses with the same sampling time or multiple doses with varied sampling time, respectively. The results showed that there was no clear statistic test for VPC and it was difficult to make sense of VPC under the multiple doses with varied sampling time. However, there were corresponding statistic tests for NPDE and the factor of study design did not affect NPDE significantly. It suggested that the clinical data and model which VPC was not fit for could be evaluated by NPDE.

Application of pharmacometrics in drug development and therapeutic drug monitoring / 中国药科大学学报

Pharmacometrics,developed from the conventional pharmacokinetics,is the science of applying mathe-matical and statistical methods to characterize,understand,and predict a drug's pharmacokinetic,phannacodyna-mic,and biomarker-outcome behaviors.Pharmacometrics has been widely valued for its utility of modeling and simulation in drug research and development,therapeutic drug monitoring and individualized therapy.This paper reviewed the advances of pharmacometrics employed in new drug research and development and therapeutic drug monitoring both at home and abroad.