Guideline 68: Dyspepsia

Dyspepsia
Annex 2

Selection of Diagnostic Tests

When considering how useful diagnostic tests are in differentiating people with disease from healthy people reference is frequently made to the sensitivity, specificity, positive predictive value, and negative predictive value of a test. These terms are defined as follows:

	Test Positive	Test Negative
Patient with disease	True Positive (TP)	False Negative (FN)
Normal patient	False Positive (FP)	True Negative (TN)

From this the following can be calculated:

Value	Calculation
Sensitivity	TP/(TP+FN)
Specificity	TN/(TN+FP)
Positive predictive value (PPV)	TP/(TP+FP)
Negative predictive value (NPV)	TN/(TN+FN)

Positive Predictive value (PPV) is a function of the True and False positive values. If there were no false positives, the PPV would be TP/TP or 100%. In the real world this is rarely the case as laboratory results usually fall on a continuous scale and a cut off has to be selected to separate positive from negative results. As such there is some degree of overlap between results from normal and affected patients.

It is theoretically possible that everyone in the population has a disease. In this case every positive result would be a true positive and the PPV would be 100%. Conversely, if no one in the population had the disease, every positive result would be a false positive. There could be no true positives, and the PPV would be 0%. This makes it clear that the prevalence of a disease in the population has an important influence on the positive predictive value of a diagnostic test. With decreasing disease prevalence, the less likely it becomes that a person with a positive test result has the disease, and the more likely it becomes that the positive result is a false positive. The diagnostic accuracy of a test therefore depends upon the prevalence of the disease in the population. The table below outlines how the positive and negative predictive values of all commonly used diagnostic tests vary by prevalence, using the published sensitivity and specificities quoted in the text (see section 4) for each test.

Prevalence of H. pylori	Serology tests Primary Care		Laboratory based serology tests		CUBT		Faecal antigen tests
	PPV	NPV	PPV	NPV	PPV	NPV	PPV	NPV
0.1%	15%	99%	0.5%	99%	2%	99%	1%	00%
5%	43%	99%	21%	99%	55%	99%	33%	99%
10%	61%	99%	37%	97%	73%	99%	51%	99%
20%	78%	97%	57%	94%	85%	99%	70%	98%
30%	86%	95%	69%	90%	91%	98%	80%	97%
40%	90%	92%	78%	85%	94%	97%	86%	96%
50%	93%	88%	84%	80%	96%	96%	90%	94%
60%	96%	83%	88%	73%	97%	93%	93%	91%

Relating this to testing for H. pylori infection, young people are less likely to have the disease and in Scotland the prevalence is likely to be less than 20%, in this group the best serological test only has a PPV of 78%, but the negative predictive value remains high. In this group therefore serology, even assuming the worst sensitivity and specificity could be used to reliably exclude H. pylori infection. In an older patient group where between 50-60% could have the disease, CUBT or faecal antigen testing would be a more appropriate choice of test as both have a PPV of over 90%. The best performing serology in this group has a PPV of over 90% but many studies using serology show much poorer results in this group.