The role of a biostatistician in clinical research: from the Protocol to the Statistical Analysis Report - A dose finding case study -
Andrea Nizzardo | Università degli Studi di Milano-Bicocca (22/09/2010)
Corso di Laurea SGI (Triennale)
Relatore: prof. Giovanni Corrao
The principal purpose of this thesis was to find out the competence of the biostatistician in clinical research, from the protocol writing to the statistical analysis.
The ICH guidelines are reported in order to clarify how a clinical study has to be conducted. Guidelines consider every aspect of clinical research, from ethical standards to statistical principles. Session 1 of this thesis is dedicated to the illustration of regulatory aspects and legislation, good clinical practice, statistical principles to analyze the data in order to demonstrate the effectiveness of a drug. In Session 2 a practical example is reported. The clinical study BKOS-02 on drug development in osteoarthritis is considered. The study protocol in section 2.2 illustrates the study design, the primary and secondary endpoints of the study, considering also inclusion and exclusion criteria and safety assessment. Section 2.3 shows the aspects of data management.
It illustrates how the data are collected starting from the construction of the Case Report Form (CRF), where informations about the patients are collected, to the description of the IVRS, which is a computerized phone-vocal system customized on the specific research project to perform several clinical trial activities like visits recording and randomization. Section 2.4 considers the statistical analysis how it is explained in the Statistical Analysis Plan (SAP). The whole statistical analysis is performed programming in SAS and using a dummy variable for treatment. It starts from descriptive statistics and model checks to verify the normal distribution of variables as well as autocorrelation of errors and homogeneity of the variance. Then for the efficacy analysis, a mixed model is applied, containing both fixed and random effects.
The primary efficacy analysis compared the improvement versus baseline of the index of knee pain for four different doses of the drug under study. The primary variable is analyzed applying a repeated measures analysis of covariance (ANCOVA) using a mixed model in order to test differences in treatment efficacy as quantified by WOMAC pain score. Statistical Analysis Report (SAR) is reported from descriptive statistics to safety analysis. In appendix, SAS syntax is also reported.