Table 4.

Accuracy analyses of width-ratio-based clinical cutoffs at differentiating severities of the Erlenmeyer flask deformity, with the best performing ratio in italics

Ratio	Severity cutoff	Accuracy (95% CI)
$\frac{\mathit{\text{Width}}6\mathit{cm}}{\mathit{\text{Width Physis}}}$	Normal ≥ mild (cutoff = 0.47)	Sensitivity: 80.7% (62.5–92.6%)
		Specificity: 79.0% (54.4–94.0%)
		PPV: 86.2% (72.0–93.8%)
		NPV: 71.4% (54.0–84.2%)
		Accuracy: 80.0% (66.3–90.0%)
	Mild ≥ Severe (cutoff = 0.61)	Sensitivity: 73.5% (55.6–87.1%)
		Specificity: 93.6% (78.6–99.2%)
		PPV: 92.6% (76.3–98.0%)
		NPV: 76.3% (64.6–85.0%)
		Accuracy: 83.1% (71.7–91.2%)
$\frac{\text{Width}4\mathrm{cm}}{\text{Width}2\mathrm{cm}}$	Normal ≥ mild (cutoff = 0.75)	Sensitivity: 77.4% (58.9–90.4%)
		Specificity: 79.0% (54.4–94.0%)
		PPV: 85.7% (71.1–93.6%)
		NPV: 68.2% (51.8–81.1%)
		Accuracy: 78.0% (64.0–88.5%)
	Mild ≥ severe (cutoff = 0.82)	Sensitivity: 76.5% (58.8–89.3%)
		Specificity: 83.9% (66.3–94.6%)
		PPV: 83.9% (69.5–92.2%)
		NPV: 76.5% (63.5–85.9%)
		Accuracy: 80.0% (68.2–88.9%)
$\frac{\text{Width}6\mathrm{cm}}{\text{Width}2\mathrm{cm}}$	Normal ≥ mild (cutoff = 0.61)	Sensitivity: 83.9% (66.3–94.6%)
		Specificity: 73.7% (48.8–90.9%)
		PPV: 83.9% (70.7–91.8%)
		NPV: 73.7% (54.6–86.7%)
		Accuracy: 80.0% (66.3–90.0%)
	Mild ≥ severe (cutoff = 0.72)	Sensitivity: 67.7% (49.5–82.6%)
		Specificity: 74.2% (55.4–88.1%)
		PPV: 74.2% (60.2–84.5%)
		NPV: 67.7% (55.2–78.0%)
		Accuracy: 70.8% (58.2–81.4%)

PPV positive predictive value, NPV negative predictive value