Table 3. Diagnosis of COPD by spirometry vs. by outcome predicted by the model.

Discriminant by model	Diagnosis by spirometry*		Total
	COPD	Non-COPD
Training Set†
Predicted COPD	228	12	340
Predicted non-COPD	15	100	115
Total	243	112	355
Validation Set‡
Predicted COPD	89	9	98
Predicted non-COPD	11	41	52
Total	100	50	50

*Diagnosis by spirometry was performed using the criteria of FEV1/FVC <70%; ^†In Training Set that were used to establish the model by stepwise discriminant analysis: sensitivity =93.83%, specificity =89.29%, accuracy =92.4%, PPV =95%, NPV =86.96%, +LR =9.23, -LR =0.07; ^‡In Validation Set that were used to check the model: sensitivity =89.00%, specificity =82.00%, accuracy =86.66%, PPV =90.82%, NPV =78.85%, +LR =4.94, -LR =0.13.