Table 3.

Clinical performance of carcinogenic HPV detection by Linear Array (LA) and Hybrid Capture 2 (hc2) for two-year cumulative ≥CIN3 as diagnosed by QC pathology (n = 3,289) and ≥CIN2 as diagnosed by clinical center pathologists (n = 2,189, restricted to women in the IC and HPV arms) referred into ALTS for an ASCUS Pap. Differences in sensitivity, specificity, and referral were tested for statistical significance using an exact McNemar's χ² test. Statistical differences in positive (PPV) and negative (NPV) predictive values were determined by Statistical differences in positive (PPV) and negative (NPV) predictive values were a method developed by Leisenring et al. (36), a score statistic derived from a marginal regression model and bears some relation to McNemar's statistic.

	Linear Array		hc2
	Estimate	95%CI	Estimate	95%CI	P
QC Pathology ≥CIN3 (n = 285)
Sensitivity	93.3%	89.8%-95.9%	92.6%	89.0%-95.4%	1
Specificity	48.1%	46.3%-49.9%	50.6%	48.8%-52.4%	0.05
PPV	14.6%	13.0%-16.3%	15.1%	13.4%-16.9%	<0.0001
NPV	98.70%	97.98%-99.22%	98.64%	97.92%-99.15%	0.4
Referral	55.5%	53.7%-57.2%	53.2%	51.5%-54.9%	0.001
CC Pathology ≥CIN2 (n = 354)
Sensitivity	87.8%	83.9%-91.1%	91.0%	87.5%-93.8%	0.09
Specificity	50.4%	48.1%-52.7%	53.1%	50.8%-55.4%	0.007
PPV	24.8%	22.4%-27.3%	26.5%	24.0%-29.2%	<0.0001
NPV	95.68%	94.20%- 96.87%	96.93%	95.67%-97.91%	0.8
Referral	55.6%	53.5%-57.7%	53.9%	51.7%-56.0%	0.04