|
|
||||||||||||||||||||||||||||||||||||||||||||||
 |
 |
Literature review > Issue_3 > Review Black et al. |
||||||||||||||||||||||||||||||||||||||||||||
|
|
This is a highly informative paper even though the title of the paper gives readers the impression that its only objective is to compare the performance of NAT with NAATs for chlamydia testing in women. With a large sample size of 3,551 women from 5 sites, this paper provides substantial data towards answering 4 important questions regarding the diagnosis of genital chlamydial infections: 1. Should labs currently using NAT switch
to NAAT? What is the incremental gain? The answer to the first question is clear. NAATs detect 17-28% more infections than NAT while maintaining the high specificity generally expected of nucleic acid based tests. It is curious that in the abstract, the final sentence states that NAATs improved the detection of infected women by 17-38% but I cannot find any where in the paper a reference to the 38%. Is it a misprint? This second question is often asked, especially by laboratory or programme directors, who would consider switching from NAT or EIAs to NAATs. In this large multicentre study in women, the authors found that although LCR is consistently more sensitive than PCR, the difference is not statistically significant. On the flip side of its higher sensitivity, LCR was shown to have a lower specificity compared to PCR. When one considers that NAATs can detect between 1-10 elementary bodies, unless there is a substantial difference in sensitivity and specificity, the realities of choosing a NAAT often boil down to issues such as: how much does the test cost ( duplex vs single test formats) , how reliable is the technical service/support, is having an internal control to monitor false negatives an asset, how well does the test fit into the workflow of the lab etc. LCR has now been withdrawn from the market. Two other NAATs have become commercially available recently for the diagnosis of C. trachomatis. I suspect the same questions will be asked regarding differences in test performances and the answers will not be different from those presented here for PCR and LCR. In looking for the answer to the third question, I found it useful to pull out data from Table 1 to show a more direct comparison of the sensitivity of detection between cervical and urine samples for the 2 NAATs, PCR and LCR: Number of women (% of all tested) who tested positive:
From the above comparison, cervical swabs allow more women in 3 of 5 centres to test positive by LCR than urine specimens but the reverse is true of PCR (% positive urine higher in 3 centers than cervical swabs). It is not clear why there is so much variation and only 2 of the 5 centres were consistent for the 2 NAATs with regard to whether urine or cervical swab did better. It is good news that urine can be used instead of cervical swabs. The authors recommended doing both but perhaps this is only possible in a well-funded research study. Last but not least, the issue of how to evaluate a new diagnostic test without bias is well explained in the paper and illustrates the difference between using a specimen reference standard versus a patient or 2 specimen reference standard to estimate sensitivity and specificity of a test. This will set a new paradigm for diagnostic test evaluation, although I worry that the cost of using such an approach may be prohibitively expensive for most developing country settings |
|||||||||||||||||||||||||||||||||||||||||||||
|
about SDI | newsletters | grants | publications | literature reviews WHO
Home -
WHO
Search - TDR Home - SDI Home -
SDI Contact us
|
||||||||||||||||||||||||||||||||||||||||||||||
.gif){kind=small}