Statistical Method For Safety Assessment In The Clinical Researches Of New Drugs

Safety assessment in the clinical research of new drugs is one of core problems before the new drugs come into the market and it is the important guarantee for new drugs to be used safely. The purpose of safety assessment is to decide whether or not the new drug should be put in the market, and help regulate the administration of new drugs in clinical practice. No drug is 100% safe for all the people in all the circumstances. The past century, witnessed many reports on the intense toxicity of certain new drugs such as France organic tin event in 1950s and Germany thalidomide tragedy in 1960s. These events taught us a big lesson. As there are more and more toxicity reactions in the different stages of clinical researches, there is a great need for a more comprehensive and objective safety assessment for the coming new drugs. However the existing safety assessment system in China is not well-organised and lack of a systematic guideline so that it can't meet the need of the clinical medical workers. Therefore we should itemize safety assessment and set up a normative analysis framework. In addition, there are still many other problems about safety assessment in the clinical researches on new drugs. For example,we analyze adverse events by the crude rate, which is equal to the total number of people treated divided by the number of people who suffer from adverse event. The crude rate index has several disadvantages. For example, it ignores frequency of adverse events and some factors which may affect the occurrence of adverse events. With the development of medicine and medical ethics, more and more people are beginning to attach great importance to the objective and comprehensive results of safety assessment based on correct statistical methods.According to Statistical Principles for Clinical Trials of ICH and Good Clinical Practice of SFDA, we summed up four aspects of safety assessment: adverse events, vital signs, extent of exposure and laboratory data and attempted to set up the elementary system of safety assessment based on an extensive review of the safety assessment abroad and at home in the four aspects in the clinical researches of new drugs. We analyzed the problems in the existing safety assessment and attempted to employ negative binomial regression to solve them, find some effective methods and widen our insight into safety assessment. Based on the data obtained in the clinical researches on new drugs, the present study designed to employ SAS software in all analysis procedures about safety assessment. The main works and results of the study are as follows:1. Four aspects of the safety assessment were reviewed extensively: adverse events, vital signs, extent of exposure and laboratory data. Their necessity in the safety assessment and corresponding statistical methods were emphasized.The first was statistical evaluation about adverse events. Pearson chi-square test was employed to compare the crude rates of adverse events of two or more groups. We described and compared the occurrence of adverseevents according to the body system categories, the event severity categories and the causality categories, so as to gain a deep insight into safety. If the occurrence of adverse events satisfied the requirement of negative binomial distribution, we employed negative binomial regression to analyze the frequency of adverse events and some affecting factors.The second was statistical evaluation about vital signs which include shrink blood pressure, diastolic blood pressure, breath rate, heart rate and body temperature. We can use some statistics such as the number of cases, mean, standard variance, median, minimum, maximum and etc. to describe the changes of vital signs values before and after treatment and the difference between these values and baseline. We employed Paired-Samples T Test or Wilcoxon signed-rank test to analyze the changes before and after treatment and apply Independent-Samples T Test or Wilcoxon rank sum test to the analysis of the difference of two groups.The third was statistical evaluation about extent of exposure which is often specified by the number of patients exposed, the duration of exposure, the duration of research, the obedience and the dose to which they were exposed. We applied statistical methods for quantitative data such as Independent-Samples T Test to the analysis of the duration of research and the duration of exposure. We classify the obedience into three categories: <80%, 80%-120% and >120%. The number and percent of patients in every category were computed and Nonparametric Test was employed to compare the difference among the different groups.The fourth was statistical evaluation about laboratory data. Cross tables were used to show the clinical significance of different laboratory parameters before and after treatment. If the laboratory data obtained after treatment was abnormal, we listed the detailed information about it. The statistical methods...