search this site.

080223P - ANALYSIS OF CLINICAL DATA ON DIABETES

Print Friendly and PDFPrint Friendly

Discussed at Kuala Belait on 22-23 February 2008 by Professor Omar Hasan Kasule


ABSTRACT
Over the past several years Dr Soad has been collating data about diabetic patients she sees in her clinics (diabetic foot & erectile dysfunction).  In addition to screening examinations carried out at government departments, schools, peripheral areas, and villages. The screening data covered the following items of information: (a) history: age, sex, race, smoking, family history of diabetes, and family history of hypertension. (b) clinical examination: height, weight, BMI, waist circumference, SBP, DBP,   (c) laboratory measurements: lipid profile, sugar, and cholesterol. Diabetic foot data is 200+ subjects. Screening data has 3800+ subjects.  Dr Soad wants to analyze the data and publish it in a high impact journal.

PAPERS PLANNED
It is expected that 3-5 papers may be generated from the data. The first paper will deal with the process of data collection and will include analyses of sampling, response, consistency and validity of data points. The second paper will deal with study of variance from the mean with discussion of what the normal range means for this group of patients. The third paper will discuss the covariance of the variables and will try to establish whether the ‘metabolic syndrome’ is an organic entity. The next 3 papers will each deal with a hypothesis about diabetes. The hypotheses will be generated after reviewing the most recent publications on the epidemiology of diabetes mellitus.  The vision is that the analysis may establish facts about diabetes that re peculiar to the KB local environment.

ANALYSIS STRATEGY
Much background reading and thinking will have to do into this analysis to make it worthwhile. The data collected is quite routine and unless new hypotheses are generated it is difficult to think of originality. Thus the analysis team will focus on generating new hypotheses and testing them in innovative statistical ways. For example instead of using BMI as an anthropometric measure we can use a combination of waist circumference and height. More data will also be needed to be collected from the clinical records about outcome measures. Most of those screened end up visiting KB hospital if they develop any complications and it should be easy to trace their records.