Table 6. Performance characteristics for first-biopsy (First) and three-month (All) biopsy models.^a .

	CAV						SEVERE CAV						GRAFT FAILURE DUE TO CAV
	At 1 Year		At 5 Years		At 10 Years		At 1 Year		At 5 Years		At 10 Years		At 1 Year		At 5 Years		At 10 Years
Measure	First	All	First	All	First	All	First	All	First	All	First	All	First	All	First	All	First	All
C (ROC Area)b	0.64	0.76	0.72	0.79	0.78	0.81	0.89	0.95	0.77	0.84	0.80	0.78	0.84	0.84	0.92	0.95	0.86	0.88
Sensitivity	0.69	0.81	0.79	0.78	0.62	0.76	0.90	1.00	0.74	0.83	0.74	0.82	0.86	1.00	0.83	0.94	0.87	0.77
Specificity	0.60	0.68	0.58	0.70	0.83	0.75	0.78	0.78	0.72	0.78	0.72	0.70	0.68	0.61	0.85	0.82	0.74	0.89
PPV	0.28	0.37	0.65	0.72	0.86	0.85	0.20	0.22	0.41	0.49	0.57	0.57	0.10	0.10	0.39	0.39	0.43	0.60
NPV	0.89	0.94	0.74	0.76	0.55	0.64	0.99	1.00	0.92	0.95	0.85	0.89	0.99	1.00	0.98	0.99	0.96	0.95
Cutoff Valuec	0.11	0.21	0.27	0.44	0.68	0.58	0.01	0.03	0.17	0.22	0.31	0.30	0.01	0.02	0.04	0.05	0.12	0.23
Pct Correctd	0.62	0.70	0.69	0.74	0.70	0.75	0.79	0.79	0.72	0.79	0.73	0.74	0.69	0.63	0.85	0.83	0.76	0.87e
Prevalencef	0.19		0.50		0.63		0.06		0.20		0.33		0.04		0.10		0.18

Abbreviations: PPV: positive predictive value; NPV: negative predictive value.

^aFirst-biopsy (First) and three-month (All) models show similar discriminative and predictive accuracy, particularly with respect to the prediction of severe CAV and graft failure. Negative predictive values (NPV) are particularly high for both first-biopsy and three-month models, indicating that it is possible, using only information from the first biopsy, to identify a patient subgroup at substantially reduced risk of developing long-term CAV or graft failure.

^bC: C-statistics: A measure of the area under the receiver operating characteristic (ROC) curve. Equivalently, it is the proportion of all case versus non-case pairs that were correctly classified by the model.

^cCutoff value: The expected value from the logistic regression model that serves as the threshold for predicting the event in question. Patients with expected values that exceed the cutoff are predicted to experience the event.

^dPct Correct is the percent correctly classified by the model and includes both positive and negative classifications.

^eThree-month model (All) is significantly better (p<.02) than the first-biopsy model (First) in classifying failure due to CAV at 10 years.

^fPrevalence: the proportion of patients that experienced the indicated event by the indicated time point.