Probability of fatty liver-onset and effects of decreased body weight in patients with multiple risk syndrome using logistic regression analysis

Haruo Yamada, Yoshitaka Hukuzawa, Rumi Seko, Hironobu Kakuta, Hideo Hukasawa, Shinichi Kakumu

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)

Abstract

We studied the probability of fatty liver (FL) onset by subjecting FL risk factors to logistic regression analysis and clarifying the probability of FL disappearance due to weight loss. Subjects were 611 people with an average age of 50.6±9 years who underwent thorough medical checkups. Of risk factors for FL, we analyzed BMI, HDL cholesterol(HDL), triglyceride(TG), HbA1C(A1C), and uric acid(UA) to investigate FL onset probability. Results indicated that probability in terms of the average risk factor score for people without FL was estimated at 15% for men and 1% for women. At a A1C of 6.5%, probability was 75.5% for men and 22% for women. At a TG of 200 mg/dl, probability was almost 100% for men and 85% for women. Regarding the effects of weight loss from a multiple risk syndrome state, the probability of FL onset at a BMI of 26 kg/m2. A1C of 5.8%, UA of 7.0 mg/d/, TG of 150 mg/d/, and HDL of 39 mg/d/ was 88% for men and 46% for women. With weight loss of 1 kg/m2 of BMI, the probability decreased to 59% for men and 9.3% for women, and with weight loss of 2 kg/m2 of BMI, it decreased to 20% for men and 7.6% for women. These findings suggest that weight loss of approximately 10% for men and 5% for women is sufficient for reducing the FL risk.

Original languageEnglish
Pages (from-to)9-15
Number of pages7
JournalJournal of the Japan Diabetes Society
Volume50
Issue number1
Publication statusPublished - 2007

All Science Journal Classification (ASJC) codes

  • Internal Medicine
  • Endocrinology, Diabetes and Metabolism
  • Endocrinology

Fingerprint

Dive into the research topics of 'Probability of fatty liver-onset and effects of decreased body weight in patients with multiple risk syndrome using logistic regression analysis'. Together they form a unique fingerprint.

Cite this