Diagnostic Utilitity Statistics 1.0

Diagnostic Utilitity Statistics is an application that calculates diagnostic efficiency statistics: sensitivity, specificity, false positive rate, false negative rate, positive and negative predictive power, odds ratio, Kappa, etc.

This program calculates diagnostic efficiency statistics as recommended by Kessel and Zimmerman (1993) and Katz (2001).

Sensitivity
Proportion of subjects with a target disorder who are identified by a positive test finding. Tests ability to correctly identify individuals with the disorder. Tests capacity to detect disease when it is truly present. Probability of a test being positive given that the disorder is present. Also called true positive rate or hit rate.

Specificity
Proportion of subjects free of the disorder who are correctly identified by a negative test result. Tests ability to correctly identify individuals without the disorder. Tests capacity to exclude disease when it is truly absent. Also called true negative rate or correct rejection rate.

False Positive Rate
Proportion of subjects identified by a positive test score who truly do not have the target disorder.

False Negative Rate
Proportion of subjects identified by a negative test score who truly have the target disorder.

Positive Predictive Power
Proportion of subjects identified by a positive test score who truly have the target disorder. Frequency with which a positive test result signifies presence of the disorder. Probability that a disorder is present when the test scores indicate that it is present. Proportion of those with a positive test result that truly has the disease.

Proportion Misleading Positives
The proportion of positive test results that are incorrect (1 - Positive Predictive Power).

Incremental Validity of Positive Test Diagnoses
Is a positive test diagnosis more likely to be correct than random diagnosis? If the IPPP is negative, no matter how large the PPP, random assignment of diagnosis is more accurate than use of the test. On the other hand, if the IPPP is positive then a positive test diagnosis is more likely to indicate presence of the disorder than a randomly assigned positive diagnosis. IPPP = PPP - Prevalence.

Quality of the Positive Predictive Power
Rescales the IPPP to a range of 0.00 to 1.00. The ratio is 0.00 when the test adds nothing, but the maximum value of 1.00 results when there are no diagnostic errors. A QPPP of .25 indicates that there is a 25% increase in diagnostic value by using the test.

Negative Predictive Power
Proportion of subjects identified by a negative test score who truly do not have the target disorder. Frequency with which a negative test result signifies absence of the disorder. Probability that a disorder is absent when the test scores indicate that it is absent. Proportion of those with negative test results that is truly free of disease.

Proportion Misleading Negatives
The proportion of negative test results that are incorrect (1 - Negative Predictive Power).

Incremental Validity of Positive Test Diagnoses
Negative values of INPP means that negative test diagnoses are less likely to indicate absence of the disorder than negative diagnoses based on randomness.
INPP = NPP - (1 - Prevalence).

Quality of the Negative Predictive Power
Rescales the INPP to a range of 0.00 to 1.00. The ratio is 0.00 when the test adds nothing, but the maximum value of 1.00 results when there are no diagnostic errors. A QNPP of .67 indicates that there is a 67% increase in diagnostic value by using the test.

Odds Ratio
Tends to be used as an index of relative risk. The likelihood that a positive test result is true rather than false, relative to the likelihood that a negative test result is false rather than true. The odds that a person with disease will have a positive test result relative to the odds of such a result in a person free of disease. How probable a correct test result is relative to an incorrect test result. The higher the better the test is performing (near 50 for good performance according to Katz, 2001). Can be interpreted as a measure of the magnitude of association between the two raters.

Likelihood Ratio(+)
The odds that a positive test result has come from a person who has the disorder. A LR+ of 1.00 does not contribute to making a diagnosis. A LR+ of 2.00 indicates that a positive test result is twice as likely for those with the disorder as for those who do not have the disorder. LR+ = Sensitivity ÷ (1 - Specificity)

Likelihood Ratio(–)
The odds that a negative test result has come from a person who does not have the disorder. For example, a LR- of 3.00 indicates that a negative test score is three times as likely to have come from a person who does not have the disorder as from a person who does have the disorder. LR- = Specificity ÷ (1 - Sensitivity)

Observed Agreement
Hit rate. Observed proportion of agreement. Proportion of subjects with and without the target disorder who were correctly classified by the test.

Kappa
Proportion of agreement between the test and actual condition of the subjects (disordered vs. non-disordered) beyond that accounted for by chance.