August 4, 2010
School of Medicine, Kochi
When planning a research study, an important question often asked is, “What would be right sample size?”
Dr. K. R. Sundaram and his colleagues at the Department of Biostatistics at the Amrita School of Medicine help students and researchers arrive at the right answer.
“Sample size is not a magic figure; nor is it a universal figure,” explains the Professor, who is also the Head of the Department. “It depends upon many factors – design of the study, whether it is an estimation problem or a hypothesis testing problem and the type of variable(s) to be studied (measurable or categorical).”
“Too large a sample size may result in an unnecessary increase in the budget. It may only make a small difference in the parameter values between the statistically significant groups and this may not be clinically significant.”
“Too small a sample size also may not work. It would make it difficult to obtain a clinically important difference in the parameter values between statistically significant groups.”
Recently Dr. Sundaram traveled to Ljubljana, Slovenia to present a paper on this topic at the 8th International Conference on Teaching Statistics.
Titled Teaching–Estimation of Minimum Sample Size and the Impact of Effect Size and Altering the Type-I & II Errors in Clinical Research the paper focused on the estimation of sample size for clinical trials.
“We compared the efficacy of two different drugs, in case of both measurable and categorical variables, and the effect of altering the type I & II errors,” Dr. Sundaram informed.
Also invited to visit the SAS headquarters at Cary, North Carolina, USA, Dr. Sundaram attended a workshop on data mining organized for senior professors during July 19-23.
“My presentation at the workshop focused on the set up of SAS at the Amrita Institute of Medical Sciences and its utility with respect to our organization,” shared Dr. Sundaram.
SAS statistical software is extensively used at Amrita for teaching, especially in the PG Diploma course on Clinical Research and the M.Sc course on Medical Statistics.
It is also used for statistical analysis of research data at Amrita. It enables data mining, which is important when choosing the appropriate multivariate models in diagnosis, treatment and prognosis of diseases.
An expert in the field of Biostatistics, Dr. Sundaram has a PhD from the All India Institute of Medical Sciences, New Delhi. During his four decades of teaching and research, he has published over 150 papers in national and international journals.