Page 87 - ISMCON souvenir 2021

ISMSCON - 2021

of variation is utilized in the weight function to select more consistent features in the final model. Further,
we include the process repetition number and the counts of each of the selected factors in the current
weights. This modified approach mainly aims to avoid the selection bias due to the choice of training and
validation data. It also provides an ordered list of consistent variables according to their importance in the
model. We use the R programming language to execute the essential analysis.
Keywords: Consistency, feature selection, weight function.

OS62: Association of Obesity and Erectile Dysfunction among

type 2 diabetes mellitus patients

Srigowtham Subramaniam, Anil C. Mathew, Benson Benny, Debin Sunny, Senthil Kumar R
PSG Institute of Medical Science & Research, Peelamedu, Coimbatore, Tamilnadu.
Email: srigowth@gmail.com
(Affiliated to The Tamilnadu Dr. MGR Medical University, Chennai)

Abstract
Background: Type 2 Diabetes mellitus and obesity are common co-morbidities of Erectile Dysfunction
(ED). However, the association of obesity on ED among type 2 diabetes is unclear. Besides, ED was
measured on an ordinal scale, and in most of the earlier studies, these scales were dichotomized and
analyzed using standard logistic regression analysis.

Objective: The aim of the study is to examine the association of obesity on ED among type 2 diabetes
mellitus patients.

Materials and methods: A total of 204 married men aged 20-60 years with a diagnosis of type 2 diabetes
at the outpatient unit of the Department of Endocrinology at PSG Hospitals during the months of May and
June 2019 were studied. We examined the association between obesity and erectile dysfunction using
binary, ordinal and polytomous regression models.

Results: Logistic regression model revealed that patients with diabetes who are obese, have a higher
odds of ED (odds ratio=2.508, p<0.05) after adjusting the effect of age and physical activity. In proportional
odds model, obesity independently increased the odds of erectile dysfunction (odds ratio=2.264, p<0.05)
across three severity categories. In the polytomous regression model, it was observed that mild ED has
an odds ratio of 3.136 with respect to normal ED (p<0.05). However, we could not find any significance
for moderate/severe ED when compared it with normal ED. It was observed that a cut off value of waist
circumference more than 92.71 centimeters would be suitable for the classification of ED in diabetic
patients.

Conclusions: The study considered the ordinal nature of the outcome variable investigated and
supported the association of obesity and ED among type 2 diabetes. In addition, we have also suggested
an optimum cut-off value of waist circumference among diabetic patients with high ED.
Keywords: Type 2 diabetes, obesity, erectile dysfunction, logistic regression, ordinal logistic
regression, polytomous regression.

CONFERENCE SOUVENIR 85