HMS ShieldHomeEducationResearchLinks HMS Division of Nutrition

 

Nutrition Obesity Research Center at Harvard

Core Facilities

Biostatististics

Dr. David Schoenfeld
Principle Investigator

Overview

Dr. David SchoenfeldThe Biostatistics support for the NORC is provided for the MGH Biostatistics Center which also provides statistical support for the Clinical Research Center, the Clinical Research Program, The Neurology Clinical Trials Unit, the Large Scale Collaboration: Injury and The Host Response, and several other groups within the hospital. The Center is expert in clinical trials, laboratory studies, and the analysis of micro-array data and is ready to help develop grant proposals, design studies, advise on data collection and help with data analysis.

Study Design

Nutrition investigators are encouraged to meet with the NORC biostatistician before they submit pilot feasibility studies or new grant proposals. The project biostatistician is an ad hoc member of the Executive Committee board, which will review every study to be performed in the NORC. This insures that every study is reviewed for proper statistical design and that every study has a sample size that is adequate to meet its objectives. Investigators wishing to submit studies to the Executive Committee can meet with the project statistician before submitting the study to discuss statistical, computational and data management issues.

Computer Systems

The Biostatistics Center has specialized statistical software, SAS, Matlab, Mathematica and R, to help investigators in data analysis. They can help investigators set up databases for their study. Investigators who keep their data on excel spreadsheets are encouraged to read the guide on preparing excel spreadsheets for statistical analysis (PDF download) if they plan to have the Biostatistics Center help with the data analysis. The Biostatistics maintains a web site on sample size to help investigators determine the number of patients required for clinical trials.

Study Conduct and Analysis

The Biostatistics Center will help investigators design case report forms and plan data collection. It is necessary for all human studies to have a Data and Safety Monitoring Plan to protect patient safety during the trial. Investigators should read the guidelines on safety monitoring to determine an appropriate plan for their study. The Biostatistics Center can provide statistical support for a Data and Safety Monitoring Committee if necessary. The Biostatistics Center also has the capabilities of performing sequential interim analyses for studies that require them. In these studies, there are planned interim analyses at specified times during the course of the study or after a certain number of events have occurred. At these analyses the accrued results are tested for statistical significance using criteria which take into account the fact that multiple analyses are being performed. The statistical center has software that implements Lan-DeMets methodology for monitoring clinical trials. (K.K. Godan Lan, David L. DeMets, Discrete sequential boundaries for clinical trials, Biometrika, 70:659-663, 1983).

The Biostatistics Center can help investigators perform final analysis of studies. For clinical trials the suggested sequence of events is as follows. The Biostatistics center prepares the randomization file which is sent to the pharmacy but not the investigator. The investigator finalizes the database without knowledge of the treatment codes. This involves making all decisions about eligibility, evaluability and the values of all measurements. The Biostatistics Center can provide a list of extreme values for each measurement to help in this process. In addition the Biostatistics Center in collaboration with the investigator will prepare an analysis plan. Then the data is locked, the treatment codes are merged with the database and the analysis plan is carried out. This procedure insures an unbiased analysis of the data.

Data Quality Control

The Biostatistics Center does not provide centralized data management. Each study will have a nurse/research assistant or "study coordinator" responsible for data management. The project biostatistician will develop a data quality control protocol with this person and help the person implement data quality control procedures. This model has been used successfully with the fish oil restenosis study, where forms, data base structure, and data query procedures were developed jointly by the study coordinator and the project biostatistician.

Development of Statistical Methods and Software

When necessary, new statistical methods or novel applications of existing methods will be developed for specific projects.

Biostatistic Courses

A representative course in biostatistics is provided annually by the Biostatistics center for laboratory and clinical investigators in the NORC.

For more information about this core, please contact:

David Schoenfeld
Department of Biostatistics
50 Staniford Street - Suite 560
Boston, MA 02114

Phone: 617-726-6111

Email: dschoenfeld@partners.org

Dr. Schoenfeld’s biographical sketch is linked here.



Home | Education | Research | Links

Copyright 2002-2008 by the President and Fellows of Harvard College
Updated 1/25/2008