While sensitivity and specificity provide useful
information, examination of ROC curves and positive
and negative predictive values can be used to evaluate
the diagnostic accuracy of a test. Both positive andnegative predictive values are dependent on the prevalence
of the disease, the quality of the gold standard
used, and the decision threshold that separates positive
from negative results. Given the low prevalence of
dogs in this study with a positive bone marrow aspirate
(24%), positive and negative predictive values are not
adequate representations of the diagnostic accuracy.
Instead, AUC for the ROC curves was evaluated. A test
with excellent diagnostic accuracy has an AUC ≥ 0.90,
while an AUC of ~0.50 does not discriminate between
experimental groups.17 In this study, individual evaluation
of a positive blood smear (AUC = 0.75) or thrombocytopenia
(AUC = 0.73) had fair diagnostic
accuracy, while evaluation of both variables together
resulted in good diagnostic accuracy (AUC = 0.83).
While sensitivity and specificity provide usefulinformation, examination of ROC curves and positiveand negative predictive values can be used to evaluatethe diagnostic accuracy of a test. Both positive andnegative predictive values are dependent on the prevalenceof the disease, the quality of the gold standardused, and the decision threshold that separates positivefrom negative results. Given the low prevalence ofdogs in this study with a positive bone marrow aspirate(24%), positive and negative predictive values are notadequate representations of the diagnostic accuracy.Instead, AUC for the ROC curves was evaluated. A testwith excellent diagnostic accuracy has an AUC ≥ 0.90,while an AUC of ~0.50 does not discriminate betweenexperimental groups.17 In this study, individual evaluationof a positive blood smear (AUC = 0.75) or thrombocytopenia(AUC = 0.73) had fair diagnosticaccuracy, while evaluation of both variables togetherresulted in good diagnostic accuracy (AUC = 0.83).
การแปล กรุณารอสักครู่..
