Skip to main content
. 2015 May 12;53:227–236. doi: 10.1007/s00592-015-0769-1

Table 3.

Association of FSH with FPG and HbA1c: linear regression

Dependent variables Standardized β P value R 2
Log FPG (model 1) −0.138 <0.001 0.11
Log FPG (model 2) −0.152 <0.001 0.11
Log FPG (model 3) −0.092 0.007 0.31
Log FPG (model 4) −0.087 0.011 0.31
Log HbA1c (model 1) −0.138 <0.001 0.03
Log HbA1c (model 2) −0.135 0.001 0.03
Log HbA1c (model 3) −0.091 0.022 0.06
Log HbA1c (model 4) −0.097 0.014 0.09

R 2 represented the coefficient of determination

Since FPG and HbA1c were non-normally distributed, they were log-transformed

Model 1 included terms for age, residence area, economic development and luteinizing hormone

Model 2 included terms for model 1 and E2

Model 3 included terms for model 2, waist circumference and HOMA-IR

Model 4 was a fully adjusted model including all covariates in model 3, metabolic factors [low-density lipoprotein, high-density lipoprotein, triglycerides and systolic blood pressure] and current smoker

FPG fasting plasma glucose, HbA1c glycated hemoglobin