Visceral fat area and red-blood-cell-related indices in adults with type 2 diabetes: Cross-sectional evidence of nonlinear associations

Table 3 Threshold effect analysis of the association between visceral fat area and red blood cell indices

	Adjusted β (95%CI)	P value
RBC100
VFA (model I)	0.20 (0.11-0.33)	< 0.0001
VFA (model II)
Inflection point	45 (43-62)
VFA < 45	1.62 (1.15-2.10)	< 0.0001
VFA ≥ 45	0.13 (0.06-0.19)	0.0002
Difference (segment 2-1)	-1.50 (-1.99 to -1.00)	< 0.0001
Log likelihood ratio		< 0.001
HCT100
VFA (model I)	0.02 (0.01-0.03)	< 0.0001
VFA (model II)
Inflection point	45 (43-62)
VFA < 45	0.12 (0.08-0.16)	< 0.0001
VFA ≥ 45	0.01 (0.01-0.02)	0.0415
Difference (segment 2-1)	-0.11 (-0.15 to -0.07)	< 0.0001
Log likelihood ratio		< 0.001
HGB
VFA (model I)	0.05 (0.03-0.09)	< 0.0001
VFA (model II)
Inflection point	45 (43-62)
VFA < 45	0.42 (0.28-0.55)	< 0.0001
VFA ≥ 45	0.03 (0.02-0.05)	0.0004
Difference (segment 2-1)	-0.38 (-0.52 to -0.24)	< 0.0001
Log likelihood ratio		< 0.001
WBC
VFA (model I)	0.01 (0.00-0.01)	< 0.0001
VFA (model II)
Inflection point	66 (50-74.78)
VFA < 66	0.00 (-0.01 to 0.02)	0.5506
VFA ≥ 66	0.01 (0.00-0.01)	< 0.0001
Difference (segment 2-1)	0.00 (-0.01 to 0.02)	0.5512
Log likelihood ratio		0.549

Data are adjusted regression coefficients (β) and 95% confidence intervals from linear (model I) and two-piecewise linear (model II) regression models. Model I assumes a single linear association between visceral fat area (cm²) and each red blood cell index. Model II allows for an inflection point and estimates separate slopes for visceral fat area below and above this point. All models were adjusted for age, sex, estimated glomerular filtration rate, current smoking, current drinking, total cholesterol, triglycerides and glycated hemoglobin. The P value for the log likelihood ratio test compares model II with model I. CI: Confidence interval; VFA: Visceral fat area; RBC: Red blood cell; HGB: Hemoglobin; WBC: White blood cell; HCT: Hematocrit.