Which laboratory serology test for H pylori? 

Metaanalysis of laboratory tests remains a rare bird, so it is
particularly pleasing to see one which is of considerable relevance given the
growing recognition of the importance of Helicobacter pylori infection for peptic
ulcers and other disorders. Les Irwig and colleagues from Sydney [1] searched the
literature for full papers and abstracts which examined the sensitivity and
specificity of laboratory serology tests for the bacterium  but did not include
nearpatient blood tests.
FindingsThey found 11 full papers and 10 abstracts which reported on 12 different tests  11 of which were ELISA tests and one latex agglutination. Most were based on European subjects and predominantly used culture (with histology and urease testing) as the reference method.Quality of studies was mixed. For instance, only eight of the 21 studies excluded patients recently treated with antibiotics (which can affect culture, but not serology tests because antibodies in the blood remain elevated for up to six months), and only three stated specifically that the study was appropriately blinded. Eight studies of 20 contained arithmetical errors and only four included results for every individual patient. ResultsThere was no real difference in test accuracy between any of the methods. At a mean sensitivity of 85%, the overall specificity was 79%.What do the results mean?If you send a blood sample to a laboratory for detection of Helicobacter pylori antibodies, it really doesn't matter what test is used to detect the antibodies. There have been many circumstances in pathology where this hasn't been the case.If you know the prevalence of Helicobacter infection in your population you can make a judgement about the predictive value of a positive or negative test from the table. 
Predictive value of a test with 85% sensitivity and 79% specificity  

Prevalence of disease (%)  Probability of disease with a positive result (%)  Probability of disease with a negative result (%) 
10  31  2 
50  80  16 
90  97  63 
From these sensitivity and specificity data we can calculate the Likelihood Ratio (LR) for a positive and a negative test ( Bandolier 28 ). 
The LR for a positive test is 4 and for a negative test it is
0.2. These points are marked on the nomogram. For an individual patient,
therefore, an assessment on clinical findings for the pretest probability can be
converted into the posttest probability of disease based on the result of the
test.
If the test is positive run a line through the pretest probability and the X marking 4 on the LR line to find the posttest probability. If it is negative run the line from the pretest probability through the X marking 0.2. Of course what is missing here is some guide as to clinical findings and ultimate diagnosis to help you. Since we don't have that guide, this is where clinical expertise and experience take over from evidence  or provide the best evidence available. This is important  a simple look at the nomogram will show that determining the pretest probability of the disease is at least as important as the test result in determining the posttest probability. Reference:

previous or next story in this issue