Advances in Sampling and Screening for Chlamydia

Jane S Hocking; Rebecca Guy; Jennifer Walker; Sepehr N Tabrizi


Future Microbiol. 2013;8(3):367-386. 

In This Article

Nucleic Acid Amplification Tests & Their Performance

The conventional diagnostic assays, such as culture and enzyme immunoassay previously used for the detection of chlamydia lacked sensitivity, required viable organisms, sterile equipment, fastidious transport conditions and invasive sampling, including speculum examination in women and urethral swabs in men, necessitating a trained healthcare provider and a clinical setting with an examination room.[49] With the advent of nucleic acid amplification test (NAAT) technologies for chlamydia testing, we now have access to more sensitive tests that can generally allow the use of self-collected samples from material collected further away from the original site of infection (e.g., urine for a cervical infection), which may consequently contain fewer organisms than in clinician-collected swabs. In addition, transport conditions are less critical for test performance, which means that these samples can be collected outside a traditional clinic setting and, in some cases, can even be mailed to the diagnostic laboratory.

While the advent of NAAT tests has revolutionized chlamydia testing, there are a number of factors that should be considered when interpreting the results of a study or selecting a test for use. This includes understanding how the test was originally evaluated for use, being aware of what a positive or negative result means and how reliable it is, and appreciating how test performance may differ in other population groups.

How was the Test Evaluated for Use?

Sensitivity and specificity are the two key statistics used to assess the performance of a screening test as compared with a gold-standard method. Sensitivity is the probability that a diagnostic test will be positive, given that the true diagnosis is positive; and specificity is the probability that the diagnostic test will be negative, given that the true diagnosis is negative.[50] The best screening tests have high sensitivity and high specificity. Low sensitivity leads to high false-negative results and low specificity leads to high false-positive results. Concern has been expressed regarding the validity of the published sensitivity and specificity of NAATs for chlamydia as there is no current gold- standard test. Traditionally, chlamydia culture has served as the gold standard and is believed to have nearly 100% specificity; however, when compared with NAATs, culture has been shown to have a much lower sensitivity, as it only detects live organisms.[51–53] As a result, estimating sensitivity and specificity of a new test by comparing its performance with culture, can result in biased sensitivity and specificity estimates of the new test. In an attempt to minimize this bias, the discrepant analysis approach was initially proposed for evaluation of NAATs for chlamydia by performing another NAAT on the specimen to adjudicate the 'true' results. If this additional test was positive, then the positive result was considered to be a true positive.[53] Concerns were raised as this method can lead to overestimates of sensitivity and specificity; however, its use has declined.[53–56] In its place, another estimation approach has been proposed, patient-infected-status algorithm (PISA).[57] There have been several versions of PISA proposed, but essentially, PISA involves using multiple tests to define the 'gold standard,' and the sensitivity and specificity of a new test is then compared with this.[57] However, it has been shown that PISA can also produce biased test-performance parameter estimates. In a series of simulated scenarios conducted by Hadgu and colleagues, none of the 95% CIs for PISA-based estimates of sensitivity and prevalence contained the true values.[57] In addition, the PISA-based estimates of sensitivity and specificity change markedly as the true prevalence changes. More advanced statistical modeling techniques, such as latent class modeling, have been proposed as an alternative for evaluating test performance.[53] While PISA is endorsed by the US FDA for evaluating chlamydia diagnostic tests,[58] both PISA and latent class modeling are now being used for test evaluation.[59]

How Reliable is a Positive Result?

Reproducibility of NAATs, a measure of the degree to which the results of a test remain consistent over repeated tests of the same specimen under the same conditions of a diagnostic or screening assay, is of prime importance. NAATs have had considerable problems with reproducibility in the past.[60–64] Approximately a decade ago, the Abbott Laboratories (IL, USA) ligase chain reaction had much publicized problems that eventually led to the test being discontinued from the market.[51] In a review of chlamydia NAAT reproducibility, Hadgu et al. found that, for studies using PCR, between 37.5 and 96.7% of positive results were confirmed; and for strand displacement amplification (SDA), between 79.3 and 93.7% of positive results were confirmed.[51] Another study by Schachter and colleagues assessed reproducibility and found that 96.7% of positive results using PCR, 83.8% of SDA and 97.7% of TMA were confirmed.[65] Hadgu et al. observed that many of the reproducibility problems in NAATs appear to occur in specimens of low-level positivity[51] and, as a result, some authors recommend the retesting of samples of low-level positive results as a means of reducing the number of potential false-positive results.[61] Others treat these low-level positive specimens as true positives and do not feel retesting is necessary, because they argue that 'failure to repeat a positive test does not mean the initial test result was a false positive'.[65] However, given the potential impact of a positive chlamydia test on an individual and the potential impact a positive diagnosis may have on relationships, positive results should be interpreted taking the individual's sexual history and clinical picture into consideration and, when there is any doubt, a retest should be considered.

It should also be remembered that, when interpreting a positive result, NAATs do not require viable organisms. NAATs have superior analytical sensitivity as they can potentially produce a positive signal from as little as a single copy of the target DNA or RNA in the reaction. However, this high level of sensitivity can create problems because, although a positive NAAT result could reflect clinical infection, the test could also be amplifying dead organisms[66] or be positive as a result of surface contamination (including contamination of the container in the case of self-collected specimens)[67] and hence, produce a false-positive diagnosis. Subsequently, it is important that an individual is not retested too quickly after treatment for infection. Studies have investigated the dynamics of chlamydia DNA or RNA clearance after treatment for chlamydia infection. One study of 115 women diagnosed with chlamydia found that, 7 and 14 days after treatment, 54 and 21% of women, respectively, still had detectable rRNA in their self-collected vaginal swabs.[68] Another recent study prospectively followed 59 treated cervical and/or rectal infections in 52 women and men, and assessed the presence of chlamydia plasmid DNA and rRNA systematically by multiple-time sequential measurements over an 8-week period.[69] This study found that a high proportion (42%) of chlamydia infections tested positive on at least one of the samples taken after 3 weeks and frequent intermittent positive chlamydia results over time. While the possibility of a new infection cannot be excluded in either of these studies, their results suggest that retesting too soon following treatment may lead to a false-positive diagnosis. The current CDC sexually transmitted disease treatment guidelines recommend against a test of cure at 3–4 weeks after treatment, and now recommend that nonpregnant women and men diagnosed with chlamydia should be retested 3 months following a positive diagnosis.[42]

How Reliable is a Negative Result?

False-negative results owing to mutations in the target gene of commercial assays is another issue that can impact the test performance and highlights the importance of continued surveillance to ensure any drop in chlamydia diagnosis rates is not due to the introduction of mutations in the strains currently circulating in the population. An example of this is the chlamydia variant identified in Sweden in 2006, with a 377 base-pair deletion in the cryptic plasmid.[70] This finding came after the unexpected 25% decrease in chlamydia infections between November 2005 and August 2006 in Halland County, southwest Sweden.[70] This had important ramifications in the diagnoses of chlamydia infections, given that several commercially available chlamydia assays were affected.[71,72]

Does Test Performance Change in Different Population Groups?

Another important statistic used to describe the performance of a diagnostic test is positive predictive value (PPV). The PPV is the proportion of people who test positive for the disease and who actually have the disease. It is related to the prevalence of the disease in the population and, as the prevalence increases, the PPV of the test increases. At low prevalence, the PPV decreases and the proportion of false-positive test results increases. Figure 2 below shows the association between PPV and prevalence for a NAAT with 90% sensitivity and different levels of specificity. It is worth noting that at low prevalence estimates, there is a steep decrease in the PPV, meaning high rates of false-positive diagnoses. This figure shows that, at 4% prevalence and 97% specificity, the PPV is approximately 50%, which means that half the positive NAAT test results will represent false-positives, whereas at 99.5% specificity, the proportion of positive tests that are falsely positive falls to 12%.

Figure 2.

Association between prevalence of a condition and the positive predictive value of a test by different levels of test specificity.

There are significant public health implications of false-positive chlamydia results. First, false-positive diagnoses can have considerable social and psychological harm on an individual, as well as his or her relationship, a factor that should not be underestimated.[73–76] Second, incidence and prevalence estimates will be overestimated, leading to biased surveillance monitoring and research outcomes; this also has implications for cost–effectiveness analyses. Finally, false-positive results will inevitably lead to overtreatment of individuals.