Researchers identify best predictors of transmission
A quantitative method to rate exposures
Alabama researchers are refining a tool they hope will streamline contact investigations by taking some of the guesswork out of deciding which exposures should take top priority.
"With contact investigations, people typically use terms like close contact’ and casual contact,’ but the trouble is that no one knows exactly what those terms mean," says Michael Kimerling, MD, MPH, assistant professor in the schools of medicine and public health at the University of Alabama in Birmingham. "We’re trying to take some of the guesswork out of it by quantifying exactly what constitutes a significant contact."
To do that, Kimerling and colleagues have analyzed more than 60 variables to see which are the best predictors that someone exposed to active TB will become infected. So far, six factors appear to be strongly predictive of skin-test positivity, he says. They are place (when "place" is the home); frequency of exposures; total hours of exposure; presence of cavitation on a chest radiograph; positive sputum culture; and the two highest categories of smear positivity (i.e., the presence of "moderate" or "numerous" numbers of bacilli).
Some factors, such as age, seem to be inversely correlated with skin-test positivity, Kimerling says. Children ages 0 to 5, for example, are less likely to be infected (though once infected, of course, are still more likely than adults to progress to active disease).
The exception is children ages 4 to 15, the period known as the "Golden Age of Childhood." Children that age appear to be just as likely to test positive as adults in the same circumstances but less likely to progress to active disease.
Other factors that haven’t proven to be strong predictors for infection include certain environmental variables, including room size and type of ventilation. Traits of the index case that haven’t shown predictive value include the patient’s race and ethnicity. (These factors were significant in a univariate analysis but not in a logistic regression analysis, Kimerling notes.)
Perhaps only due to small numbers, other factors don’t look as if they are predictive but may become so as more data come in. They include HIV status; a history of certain other diseases, including diabetes and renal failure; homelessness; and the presence of intravenous drug abuse.
At least one variable wasn’t put into the hopper at all, Kimerling notes. Although clinicians typically assume an index case who complains of a cough is more infectious than someone who does not complain of that symptom, researchers decided that the question of how much someone has been coughing probably resists being quantified objectively.
Help in assigning top priorities
In the course of the project so far, data have been analyzed from 291 cases and 2,856 exposures, says Kimerling. He plans to continue collecting more information until this spring, when he will perform a final analysis and compare it with the preliminary analysis. Once that’s complete, each variable will be assigned a weight on a scoring system.
Within the scoring system, certain variables — such as the presence of cavitation on chest X-ray — override others in importance. Contacts whose circumstances add up to a certain score will be categorized as "significant" and will be accorded a commensurate amount of time and attention, he says.
The findings and the scoring system, in turn, eventually will be incorporated into a piece of software that will be loaded onto the laptop programs that outreach workers in the state already use during contact investigations.
Findings will help clinicians assign priorities more easily and will aid field workers in knowing where to direct their attention, researchers hope. "It will help us know whether to go full-speed on an investigation or whether it can wait," Kimerling says. "It will also help focus the efforts of field workers by showing them what questions are most important to ask — and to keep asking." In the field setting, the tool also will help field workers decide how far to extend their investigation. Is it enough, for example, to investigate contacts in the home environment? Should they extend to the work setting? Investigate just one shift, or all three shifts at work?
The program also will lend a certain authority to health department officials seeking to reassure those who have been exposed to a case, he adds. "For example, we’ll be able to say with more certainty that in the absence of cavitation, or with the presence of only minimal bacilli on a smear, that a contact’s chances of being infected are very small."
The software, along with a manual for users, should be ready by year’s end. The state plans to share it at no cost with other programs.