Editor's note: The author does not work for a testing manufacturer, and has no conflict of interest.
Advice for a new class of syphilis tests: get new spin doctors.
The public relations debacle for those tests, called enzyme or chemiluminescence (EIA/CIAs) tests, started on February 11. That’s when CDC’s weekly MMWR journal published a report on “reverse sequence” syphilis screening, in which laboratories use EIA/CIA tests to help diagnose syphilis.
According to the report, which evaluated “reverse sequence” screening at five U.S. laboratories from 2006 to 2010, nearly 18% of all positive EIA/CIA test results were not confirmed by other syphilis tests. That suggests that the EIA/CIA results were “false positives.”
The press pounced.
“Test Gets Almost One in Five Syphilis Cases Wrong,” headlined a widely published Associated Press article that CDC itself featured verbatim (headline included!) on its websites. “Reverse sequence” screening has “problems with false positives,” another article asserted, and can lead to “unnecessary anxiety and treatment.” Yet another ran under the headline “After syphilis scare, health experts advise retests.”
Not so. The problem was, really, one of bad spin. By highlighting the unimpressive performance of false-positive EIA/CIA tests, and not the high accuracy of “reverse sequence” algorithms overall, the MMWR report – and, in turn, the press – missed the forest for the trees.
Understanding why starts with the crucial fact that confirming a laboratory diagnosis of syphilis always requires more than a single test.
The traditional screening algorithm involves two tests. If the first is negative, the laboratory can stop. The patient doesn’t have syphilis. If it’s positive, the laboratory does a second, confirmatory test.
“Reverse sequence” algorithms start with EIA/CIA tests. EIA/CIA tests are closely related to the second type of test done in the traditional algorithm, which is why algorithms that start with EIA/CIA tests are called “reverse sequence.” algorithms. If the EIA/CIA test is positive, the laboratory does a second test, which, if positive, confirms a syphilis diagnosis.
But If the second test is negative, the laboratory must break the tie with a third test. A positive tiebreaker confirms syphilis.
If the tiebreaker is negative, then syphilis is unlikely. In those cases, according to recommendations provided in a prior MMWR report and reinforced in the more recent report, syphilis should not necessarily be diagnosed or treated. Rather, the patient typically should be retested for syphilis in several weeks. (Ultimately, as in any other disease, clinical management decisions when syphilis is suspected should rest not on laboratory results alone but on an examination of all relevant clinical and epidemiologic data.)
Unfortunately, both the MMWR report’s title (which begins, ominously, with the words “discordant results”) and text mostly focused on EIA/CIA tests, rather than “reverse sequence” algorithms overall. It’s no surprise that the press followed suit. (The press further assumed, incorrectly, that patients were unnecessarily diagnosed with and treated for syphilis solely on the basis of positive EIA/CIA results.)
The accuracy of EIA/CIA tests is important, in and of itself, to laboratories, because it dictates how often they’ll have to run tiebreaker tests. But to clinicians and patients, the performance of any single test in a laboratory screening algorithm for syphilis – EIA/CIA tests included – doesn’t much matter. What really counts is the performance of the algorithm overall.
Looked at that way, it’s a different story entirely. Of 140,176 specimens analyzed in the MMWR report, in fact, only 866, or 0.6%, had results suggesting that the EIA/CIA result was a false positive. Sure, the EIA/CIA test itself got almost one in five results wrong – but the overall “reverse sequence” algorithm got it right more than 99 times out of 100.
Not too shabby.
Nevertheless, why not stick with the gold standard for the laboratory diagnosis of syphilis –the traditional algorithm – which, after all, CDC continues to recommend?
The answer comes down to economics. EIA/CIA tests can be automated. That means they can be done relatively cheaply. Other types of syphilis tests, meanwhile, must be done manually, and, hence, more expensively. Because about 95% of specimens (based on data in a prior MMWR report and the more recent report) require only the relatively cheap EIA/CIA test, it’s more efficient for high-volume laboratories to start (and in most cases stop) after a single, cheap, automated test.
Furthermore, CDC does not recommend against using “reverse sequence” screening, and, in fact, has made recommendations regarding interpretation results of “reverse sequence” algorithms. And the FDA has approved a number of EIA/CIA tests for use in the laboratory diagnosis of syphilis.
There are potential downsides to “reverse sequence” algorithms. They might help the bottom line of laboratories. But accounting for all costs associated with false-positives suggests that “reverse sequence” screening can in fact be more expensive, compared with traditional algorithms, for the health-care system as a whole. Additionally, the anxiety likely experienced by the small percentage of patients told they need to retest for syphilis cannot be discounted.
Whether “reverse sequence” syphilis screening truly contributes to public health efforts to prevent and control syphilis, then, is still an open question. With syphilis rates rising in the United States, it’s an important one. But answering it requires a more meaningful framing of the pros and cons of different approaches to diagnosing syphilis in the laboratory than occurred in this case.
Next time, in other words, the spin doctors should set their sights on the forest of syphilis testing, not its trees.
 For the syphiliterati, two caveats. First, the exception to this rule is persons who have had syphilis in the past, for whom only one test should be performed. But that test is not an EIA/CIA or related treponemal test, so the accuracy of EIA/CIA tests is not relevant to those cases. Second, this article only refers to the laboratory diagnosis of syphilis on the basis of serologic tests, not other laboratory tests used in some cases (e.g., darkfield microscopy, histopathologic examination, or polymerase chain reaction-based testing).
 Determining whether it’s a new case or a previously treated cases of syphilis requires additional laboratory and clinical data.
 The same caveat noted in footnote 2 also applies here.
 Again, the same caveat noted in footnote 2 applies here.