Specificity and sensitivity, near- patient testing

Why are medical tests performed?

To reduce uncertainty, guide investigations/treatment and prevent disease.

What is point-of-care testing (POCT)?

Any analytical test performed outside the conventional laboratory setting, usually by non-laboratory personnel.

What are the main uses of medical tests?

Predisposition, screening, diagnosis, staging, prognosis, stratification, efficacy, monitoring and recurrence.

What is test accuracy?

The ability of a test to correctly identify people with and without disease.

What is the target condition in a test accuracy study?

The disease or condition being tested for.

What is the target population in a test accuracy study?

The group of patients being investigated.

What is the index test?

The test being evaluated.

What is the reference standard?

The best available method for determining the true disease status.

What is the basic design of a diagnostic accuracy study?

A series of patients receive both the index test and reference standard, then results are compared.

What does TP stand for?

True positive = disease present and test positive.

What does FP stand for?

False positive = disease absent but test positive.

What does FN stand for?

False negative = disease present but test negative.

What does TN stand for?

True negative = disease absent and test negative.

What is sensitivity?

The proportion of people with disease who test positive.

What is the formula for sensitivity?

TP/(TP+FN).

What is sensitivity based on?

People with disease.

What makes a highly sensitive test useful?

Ruling out disease when the result is negative.

How do you remember sensitivity?

SnNout = Sensitive test + Negative result = rule OUT disease.

What is specificity?

The proportion of people without disease who test negative.

What is the formula for specificity?

TN/(TN+FP).

What is specificity based on?

People without disease.

What makes a highly specific test useful?

Ruling in disease when the result is positive.

How do you remember specificity?

SpPin = Specific test + Positive result = rule IN disease.

What is a positive predictive value (PPV)?

The probability that someone with a positive test truly has the disease.

What is the formula for PPV?

TP/(TP+FP).

What is PPV based on?

People with positive test results.

Why is PPV important clinically?

It tells you how likely a positive test represents true disease.

What is a negative predictive value (NPV)?

The probability that someone with a negative test truly does not have the disease.

What is the formula for NPV?

TN/(TN+FN).

What is NPV based on?

People with negative test results.

Why is NPV important clinically?

It tells you how likely a negative test represents absence of disease.

What does a false positive result mean?

The test is positive but the patient does not have the disease.

What does a false negative result mean?

The test is negative but the patient does have the disease.

Which measures describe the intrinsic performance of a test?

Sensitivity and specificity.

Which measures depend on disease prevalence?

PPV and NPV.

What is prevalence?

The proportion of the population that has the disease = (TP+FN)/(TP+FP+FN+TN).

How does decreasing prevalence affect PPV?

PPV decreases.

How does decreasing prevalence affect NPV?

NPV increases.

How does increasing prevalence affect PPV?

PPV increases.

How does increasing prevalence affect NPV?

NPV decreases.

Does prevalence affect predictive values?

Yes, PPV and NPV depend directly on prevalence.

Can prevalence influence sensitivity and specificity?

Yes, through effects on case mix and disease spectrum.

What factors should be considered when interpreting a test result?

The clinical pathway and consequences of false positives and false negatives.

In the UTI example, what was the prevalence?

10%.

In the UTI example, what was the sensitivity?

90%.

In the UTI example, what was the false positive rate?

5%.

If 10,000 children are tested and prevalence is 10%, how many truly have UTI?

1000 children.

If sensitivity is 90% and 1000 children have UTI, how many are true positives?

900.

If 9000 children do not have UTI and the false positive rate is 5%, how many are false positives?

450.

How many positive test results were there in the UTI example?

1350 (900 TP + 450 FP).

Calculate the PPV in the UTI example.

PPV = 900/1350 = 67%.

If a child tests positive in the UTI example, what is the probability they truly have UTI?

67%.

How should test results be communicated to patients?

Using natural frequencies, simple sentences and visual aids.

How can test results be explained using natural frequencies?

E.g. "67 out of 100 people with a positive test truly have the disease."

What is the key take-home message about test accuracy?

Test accuracy is not fixed; prevalence affects predictive values and can influence sensitivity and specificity.