Skip to main content

Table 3 Threshold effect analysis of HDL-C and hemoglobin using two-precise linear regression

From: The relationship between high-density lipoprotein cholesterol (HDL-C) and glycosylated hemoglobin in diabetic patients aged 20 or above: a cross-sectional study

Glycohemoglobin Adjustedβ(95% CI), p
Age < 40 years
 Fitting by a standard linear model −0.000 (− 0.017, 0.016) 0.9675
 Fitting by two precise linear model
  Inflection point 60
  HDL-C < 60 mg/dL 0.034 (0.015, 0.053) 0.0030
  HDL-C > 60 mg/dL −0.082 (− 0.120, − 0.044) 0.0006
  Log-likelihood ratio < 0.001
Age > =60 years
 Fitting by a standard linear model −0.016 (− 0.029, − 0.002) 0.0268
 Fitting by two precise linear model
  Inflection point 60
  HDL-C < 60 mg/dL −0.043 (− 0.066, − 0.020) 0.0006
  HDL-C > 60 mg/dL 0.012 (− 0.011, 0.035) 0.3178
  Log-likelihood ratio 0.001
Other race/ethnicity
 Fitting by a standard linear model 0.007 (−0.012, 0.027) 0.4770
 Fitting by two precise linear model
  Inflection point 60
  HDL-C < 60 mg/dL −0.002 (− 0.040, 0.036) 0.9181
  HDL-C > 60 mg/dL 0.013 (−0.015, 0.041) 0.3823
  Log-likelihood ratio 0.378
  1. Sex, age, race/ethnicity, BMI, waist circumference, systolic blood pressure, diastolic blood pressure, ALT, Cr, TG, TC, LDL, FDG, γGT, uric acid, insulin and lipid-lowering medications and anti-diabetic medications were adjusted.