Data Management and Biostastics Core

点击翻译按钮获取中文摘要

This Data Management and Biostatistics Core will support and strengthen projects on the chemotherapy, immunology, and ecology of leishmaniasis. As a randomized clinical trial, the chemotherapy project has more stringent requirements in terms of data management and analysis, so receives greater emphasis. The general objective is to ensure that the data management and biostatistics capacities of CIDEIM (the Major Foreign Collaborator) reach high international standards, and are sustainable. Relative to biostatistics, the current data management capacity is lower, so requires more strengthening. In order to reach the general objective, we have set five specific aims, and planned the following activities. 1) To attain the required standards of accuracy and security in the data management of the proposed projects. We will develop custom data management systems using the TrialDB package developed in Yale, 2) To enable CIDEIM to sustain these standards for future projects. CIDEIM personnel will be trained in TrialDB, and will run the custom systems when they have been transferred from Yale to CIDEIM. 3) To attain the required standards in statistical aspects of design, planning and execution of the proposed intervention trial. Experienced statisticians at CIDEIM and Yale have, together with the trial investigators, constructed the trial design to give adequate statistical power; to minimize bias; and, by means of a planned interim analysis, to allow for early stopping. 4) To devise and implement a Data and Safety Monitoring Plan, and convene a Data Safety and Monitoring Board (DSMB). Although the trial is Phase II, rather than III, due to the multisite nature of the trial, and inclusion of children, we have decided to convene a DSMB to monitor progress and advise the investigators. 5) To devise and execute a plan for statistical analysis of the trial. The trial has a well defined binary endpoint, and the planned analysis will yields valid statistical inference, taking into account factors such as the prior interim analysis, and the possible need to use exact, rather than asymptotic, logistic regression.