Waist-to-Height Ratio Is a Better Anthropometric Index than Waist Circumference and BMI in Predicting Metabolic Syndrome among Obese Mexican Adolescents

Table 4

Area under the ROC curves and 95% confidence intervals for the selected anthropometric indices, with cut-offs for the sensitivity, specificity, balanced accuracy, PPV, and PNV of an MS diagnosis.

Variable	Area under the ROC curve (95% C.I.)	Significance^a	Optimal cut-off point^b	Sensitivity	Specificity	Balanced accuracy	PPV^c	PNV^d
BMI percentile
Overall	0.651 (0.547–0.755)	= 0.008^e	99	61.8%	66.7%	64.3%	75.0%	51.9%
Female	0.606 (0.441–0.771)	= 0.217	99	65.5%	68.4%	67.0%	76.0%	56.5%
Male	0.677 (0.540–0.814)	= 0.021^e	98	76.9%	56.5%	66.7%	75.0%	59.1%
WC (cm)
Overall	0.704 (0.604–0.804)	< 0.001^e	90	67.6%	66.7%	67.2%	76.7%	56.0%
Female	0.708 (0.561–0.854)	= 0.016^e	95	55.2%	84.2%	69.7%	84.2%	55.2%
Male	0.696 (0.556–0.836)	= 0.011^e	90	74.4%	65.2%	69.8%	78.4%	60.0%
WHtR
Overall	0.652 (0.544–0.759)	= 0.008^e	0.60	69.1%	57.1%	63.1%	72.3%	53.3%
Female	0.682 (0.528–0.836)	= 0.034^e	0.60	69.0%	63.2%	66.1%	74.1%	57.1%
Male	0.627 (0.477–0.776)	= 0.098	0.63	53.8%	69.6%	61.7%	75.0%	47.1%

In all cases, 110 subjects (48 females and 62 males) were considered.
^aNull hypothesis: area = 0.5.
^bPositive if assessment is more than or equal to the optimal cut-off point; it was calculated as the minimum value of the square root of the following: [(1 − sensitivity)² + (1 − specificity)²], and greater accuracy is reflected by a smaller distance to point (0, 1) in the ROC curve.
^cPPV: predictive positive value.
^dPNV: predictive negative value.
^eSignificant values.