Table 3.

Estimation of the impact of matching accuracy on the screening accuracy^a.

Assumption of matching accuracy (%)			Screening sensitivity (%)			Screening specificity (%)			Positive predictive value (%)
Sensitivity	Specificity	True		Estimate	True		Estimate	True		Estimate
90	100	90		NAb	90		89.94	6.6		5.92
85	100	90		NA	90		89.91	6.6		5.59
50	100	90		NA	90		89.69	6.6		3.29
100	99.99	90		88.99	90		NA	6.6		6.58
100	99.90	90		80.93	90		NA	6.6		6.67
100	99.80	90		73.70	90		NA	6.6		6.76
88.71	99.80	90		90.00	90		89.89	6.6		6.02
82.26	100	90		72.09	90		89.93	6.6		5.41
90	100	60		NA	90		89.96	4.5		4.03
85	100	60		NA	90		89.94	4.5		3.81
50	100	60		NA	90		89.81	4.5		2.24
100	99.99	60		59.37	90		NA	4.5		4.49
100	99.90	60		54.33	90		NA	4.5		4.58
100	99.80	60		49.81	90		NA	4.5		4.67
88.71	99.80	60		48.81	90		89.96	4.5		4.17
82.26	100	60		60.00	90		89.68	4.5		3.18

^aThe table shows the impact of matching accuracy on cancer-screening accuracy estimates when the true sensitivity of cancer screening is set at 90% and 60%, and the true specificity is set at 90%. The cancer incidence rate is approximately 775.7 person per year, which is the national average in Japan.

^bNA: not affected. “NA” represents that no change occurred between the true and estimated values. The italicized values show the estimates obtained using the experimental data.