gpraj

View Original

Diagnostic accuracy and screening

Key Equations

Equation 1 Odds = probability /(1-probability)

Equation 2 Probability = odds/ (1+ odds)

Equation 3 Post-test odds = Pre-test odds x Likelihood ratio

Need to convert pre-test probability of disease (prevalence) into pre-test odds (using equation 1).
Then, multiply pre-test odds by the positive likelihood ratio to generate post-test odds (using equation 3).
Then, convert post-test odds to post-test probability (using equation 2).


specificity, sensitivity,

Predictive value

Negative predictive value (NPV)

Positive predictive value (PPV)



SpPin (High specificity, positive result, rules in disorder)

When a sign, test or symptom has an extremely high specificity (say, over 95%), a positive result tends to rule in the diagnosis. For example, testing positive (≥3 positive answers on CAGE questionnaire) rules in the diagnosis of alcohol dependency.

Conversely, testing for Antinuclear antibodies (ANAs) in order to diagnose rheumatoid arthritis, has a low specificity owing to a high FP rate. ANAs are found in patients with other connective tissue disorders.

Low False Positive rate creates a high specificity and high Positive Predictive Value (PPV)


SnNout (High Sensitivity, negative result, rules out disorder)

When a sign, test or symptom has a high sensitivity, a negative result rules out the diagnosis.
For example, the sensitivity of the loss of retinal vein pulsation in diagnosing high intracranial pressure is 100 per cent. Therefore, if a person tests negative (no loss of retinal vein pulsation), it rules out disorder of increased intracranial pressure.

Low False Negative rate creates a high sensitivity and high Negative Predictive Value (PPV)