Conversion rates under 2% most likely false

All the same, serial testing still best strategy

The tuberculin skin test (TST) results in more false positives than true positives in most hospital settings in the United States, says John Bass, MD, chairman of the department of internal medicine at the University of Alabama at Birmingham.

That’s not a slam at the test, either, adds Bass. It’s just an observation about the limits of any diagnostic test, coupled with the fact that the test is generally employed in a low-prevalence setting.

If your facility is typical for the United States, you’re probably well below that benchmark, and you can assume that whatever apparent conversions you’re looking at are probably false positives, he says, because "most places today are down below 1% now, even in New York."

Here’s Bass’ logic. "For most tests, the false-positive rate is about 2.5%, since 2.5% of results are high and 2.5% are low, for a standard [statistical] deviation of 95%," he says.

There’s no biological correlate yet to stack up against the TST, so in a sense, any such line of reasoning is, by necessity, circular, he concedes. "We don’t have an independent test to tell us whether the TST is accurate or not. So all we can do is try to estimate the test’s sensitivity and specificity. Then you’ve got what you need to predict the operating characteristics of the test in various populations."

At 2%, you’d better look harder

Or, you can do what Bass does, which is conduct a posterior analysis of the results of your skin-testing program. "For example, if you have a hospital that skin-tests 1,000 employees a year, and you have seven conversions, that’s a conversion rate of 0.7%," he says. If Bass is right about the TST producing a false-positive result about 1% of the time, then "all those [apparent conversions] are false positives," he concludes.

Bass adds that as the number goes up, he starts looking at skin-testing data in a different light. The point at which apparent conversions are likely to represent trouble is about 2%, he says. "If your conversion rate stands at 2% or higher, you’re probably transmitting TB. If your rate is less than 1%, you’re probably not. And if you’re between 1% and 2%, things are sort of iffy."

Most places fall well below that "iffy" zone, he says. "The likelihood of hospital workers becoming infected, especially if you’re not in Harlem or Jackson Memorial [in Miami], or else in an unknown outbreak situation, is much less than 1%. So mathematically, at least, almost all the conversions in a serial skin-testing program, where people are tested every year, are going to be false positives."

Debating the need for a federal standard

In his slow, patrician drawl, Bass recently ticked off the same points to members of an Institute of Medicine committee charged with looking at whether a need exists for a federal TB standard. Opponents of the federal standard were quick to seize on Bass’ testimony as proof that the Occupational Safety and Healthy Administration (OSHA) is mistaken in its estimates of the TB infection risk to health care workers because, most of the time, under most settings, individuals are probably showing false-positive results.

Because it’s bound by precedent and by statutory obligation to show a "reasonable risk" to worker health exists, OSHA is off-base when it adds up skin-test conversions to quantify that risk, they argue.

How to handle those positive results

Whatever the implications for OSHA, Bass says there’s no reason for hospitals not to keep using the TST in serial fashion. "The purpose of a skin-testing program is to detect an outbreak of TB in your hospital," he asserts. "It’s most valuable for that. The only problem lies in how to interpret the test for a given individual [who tests positive]. What do you do?"

What Bass does first is explain how and why the test result is more likely to be false than true. "Then we give them a chest X-ray to rule out active disease. And then I tell them that isoniazid is relatively innocuous and that they can take it if they like, or they can do nothing."

If the person falls sick with respiratory illness in the next year or so, he adds, "I tell them they should be aware it could be TB, so they should be evaluated."