How to use it
- Enter TP, FP, FN and TN from your 2×2 table (reference standard against test result).
- Read sensitivity and specificity at the top, then the predictive values, likelihood ratios and summary measures.
- To see how the test would perform where the condition is more or less common, enter that prevalence in percent.
Formulas
- Sensitivity = TP/(TP + FN), specificity = TN/(TN + FP)
- PPV = TP/(TP + FP), NPV = TN/(TN + FN)
- LR+ = sensitivity/(1 − specificity), LR− = (1 − sensitivity)/specificity
- Accuracy = (TP + TN)/total, Youden’s J = sensitivity + specificity − 1, DOR = (TP·TN)/(FP·FN)
- At prevalence P: PPV = sens·P / (sens·P + (1 − spec)(1 − P)), NPV = spec·(1 − P) / (spec·(1 − P) + (1 − sens)·P)
Worked example
A test is checked on 1000 people: TP = 90, FP = 40, FN = 10, TN = 860.
- Sensitivity 90/100 = 90% (95 % CI 82.56% to 94.48%), specificity 860/900 = 95.56%.
- PPV 90/130 = 69.23%, NPV 860/870 = 98.85%.
- LR+ = 0.9/0.044444 = 20.25, LR− = 0.104651, accuracy 95%, DOR 193.5.
The table’s prevalence is 10%. In a screening population with 1% prevalence the same test has a PPV of only 16.98% and an NPV of 99.89%.
Tips and common mistakes
- Make sure rows and columns are not swapped: FP are healthy people with a positive test, FN are ill people with a negative test.
- Sensitivity and specificity describe the test; PPV and NPV describe a test result in a given population.
- High accuracy can hide a poor test when the condition is rare: a test that is always negative is 99 % accurate at 1 % prevalence.