Comparison of Weight-Gain-Based Prediction Models for Retinopathy of Prematurity in an Australian Population

Table 2

Predictions of type 1 ROP, type 2 ROP, and treated ROP for CHOPROP, WINROP, CO-ROP, and G-ROP.

	CHOPROP	WINROP	CO-ROP	G-ROP
Type 1 ROP	n = 24	n = 24	n = 24	n = 24
Sensitivity, % (95% CI)	100	83.3	100	100
	(86.2–100)	(64.1–93.3)	(86.2–100)	(86.2–100)
Specificity, % (95% CI)	46.0	52.3	32.0	28.2
	(41.7–50.3)	(47.8–56.7)	(28.0–36.1)	(24.5–32.3)
Positive predictive value, % (95% CI)	8.1	8.0	6.5	6.2
	(5.5–11.7)	(5.2–12.0)	(4.4–9.5)	(4.2–9.0)
Negative predictive value, % (95% CI)	100	98.4	100	100
	(98.4–100)	(96.1–99.4)	(98.3–100)	(97.4–100)
Positive likelihood ratio (95% CI)	1.85	1.75	1.47	1.39
	(1.64–2.00)	(1.43–2.14)	(1.60–1.94)	(1.26–1.48)
Negative likelihood ratio	0.00	0.32	0.00	0.00
	(0.00–0.69)	(0.13–0.78)	(0.00–0.72)	(0.00–1.13)
Type 2 ROP	n = 40	n = 40	n = 40	n = 40
Sensitivity (95% CI)	80.0	67.5	92.5	90.0
	(65.0–90.0)	(52.0–79.9)	(80.1–97.4)	(76.9–96.0)
Specificity (95% CI)	45.8	52.0	32.4	28.3
	(41.6–50.2)	(47.5–56.5)	(28.4–36.6)	(24.5–32.5)
Positive predictive value, % (95% CI)	10.7	10.6	10.0	9.3
	(7.7–14.8)	(7.4–15.0)	(7.4–13.5)	(6.8–12.6)
Negative predictive value, % (95% CI)	96.6	95.0	98.1	97.2
	(93.4–98.3)	(91.6–97.0)	(94.7–99.4)	(93.0–98.9)
Positive likelihood ratio	1.48	1.41	1.37	1.26
	(1.24–1.76)	(1.11–1.78)	(1.23–1.52)	(1.12–1.41)
Negative likelihood ratio	0.44	0.62	0.23	0.35
	(0.23–082)	(0.40–0.98)	(0.08–0.69)	(0.14–0.90)
Treated ROP	n = 29	n = 29	n = 29	n = 29
Sensitivity (95% CI)	93.1	82.8	100%	96.6
	(78.0–98.1)	(65.5–92.4)	(88.3–100)	(82.8–99.4)
Specificity (95% CI)	46.0	52.6	32.3	28.3
	(41.7–50.4)	(48.1–57.0)	(28.3–36.5)	(24.5–32.4)
Positive predictive value, % (95% CI)	9.1	9.6	7.9	7.2
	(6.3–12.9)	(6.5–13.8)	(5.5–11.1)	(5.0–10.2)
Negative predictive value, % (95% CI)	99.1	98.1	100	99.3
	(96.9–99.8)	(95.5–99.2)	(97.7–100.0)	(96.1–99.9)
Positive likelihood ratio (95% CI)	1.72	1.75	1.48	1.35
	(1.52–1.96)	(1.44–2.11)	(1.34–1.57)	(1.23–1.47)
Negative likelihood ratio	0.15	0.33	0.00	0.12
	(0.04–0.57)	(0.15–0.73)	(0.00–0.91)	(0.02–0.84)
Any ROP	n = 356	n = 350^†	n = 356	n = 356
Sensitivity (95% CI)	69.9	54.9	78.4	78.4
	(65.0–74.5)	(49.6–60.0)	73.8–82.3)	(73.8–82.3)
Specificity (95% CI)	72.0	63.2	48.6	37.7
	(4.9–78.1)	(55.6–70.2)	41.3–55.9)	(30.9–45.1)
Positive predictive value, % (95% CI)	83.6	76.2	75.6	71.9
	(78.9–87.3)	(70.6–81.0)	(71.0–79.7)	(67.2–76.1)
Negative predictive value, % (95% CI)	54.1	39.5	52.5	46.2
	(47.7–60.4)	(33.7–45.5)	(44.8–60.0)	(3.82–54.3)
Positive likelihood ratio	2.50	1.49	1.52	1.26
	(1.95–3.20)	(1.19–1.86)	(1.31–1.78)	(1.11–1.43)
Negative likelihood ratio	0.42	0.71	0.45	0.57
	(0.35–0.50)	(0.61–0.84)	(0.35–0.57)	(0.44–0.75)

^†Six infants GA >32 cannot be entered into WINROP.