Cool new algorithm can triage contacts
Cool new algorithm can triage contacts
Same factors, but better measuring sticks
In Alabama, where TB experts have been laboring for years to perfect an electronic patient record, software whizzes are about to pull another wonder out of their hats.
This time, it’s an algorithm that can be used to prioritize case contacts. Outreach workers using it should be able to trim back the number of tuberculin skin tests they place and read by almost as much as a third without missing any conversions, says William Bailey, MD, professor of medicine at the University of Alabama in Birmingham (UAB) and director of UAB’s Lung Health Center.
Especially for jurisdictions strapped for resources, a way to trim fat from laborious contact investigations ought to be a help, Bailey adds. "We think such a thing would be a really useful tool," he says.
The algorithm, once it’s finished, will represent the culmination of a mammoth six-year effort on the part of state TB controllers to get all of their data into a readily accessible statewide software bank. Someday — and not far off, either, TB controllers in Alabama promise — the software will be completely debugged, and anyone with the necessary security clearance will be able to browse up-to-the-minute records of TB cases and their related contacts, says Nancy Brook, MPH, the Alabama TB program manager.
In the meantime, TB controllers decided they might as well sift through the mountains of data piling up to see what they could learn about contact investigations. "We decided we wanted to see if we could get contact investigations down to a science," says Brook.
But 3,941 contacts (and 292 patients) later, that study hasn’t turned up any big surprises, she admits. On the contrary, the project has confirmed most of the common wisdom about contact investigations. (You know: A patient with cavitary disease is apt to be more infectious than someone who’s smear-negative. Spend a few days with that patient in an airless closet, and you’ll probably be infected yourself.) The surprises have been finding the varying strengths displayed by the assorted factors working in combination with each other, she says.
When in doubt, simplify
On the other hand, one of the toughest parts of the study — and one of the most rewarding — has been trying to retool the contact data screen. The new format doesn’t change the rules of contact investigation so much as it refines them, say Brook and Bailey. That’s because creating the new format forced researchers to come up with simple ways for outreach workers to estimate the various factors at work, Bailey says. "It’s helped us come up with more exact ways to measure these factors, hopefully in ways that are easy for people to use," he says.
That’s meant devising common-sense terminology for judging things such as the size of an indoor space, ranging from "about the size of a phone booth" to "about the size of a car" to "about the size of a big meeting hall."
In the same way, investigators wanted to make sure outreach workers don’t need an engineering degree to be able to spot the salient features of ventilation systems. Here, choices include more simple selections, such as "open window," "fan," and "central heating and air."
Researchers also have devised simple ways to effectively tally frequency and duration of exposures and determined what each increment adds to the risk of infection.
"We’ve tried to measure the effects of variables in contacts we suspected might leave people more vulnerable, too, such as history of alcohol abuse or smoking," adds Brook. As for source-case variables, investigators looked at the varying effects of smear-positivity: "We knew, of course, that being smear-positive meant you were more apt to be infectious. But we wanted to find out what it meant to have few’ organisms on your slide, as opposed to moderate’ or numerous.’"
Extracting the information for the contact investigation study has been anything but easy, Brook and Bailey are quick to add.
Back in 1994, when the state TB program began equipping all its TB outreach workers with laptop computers, mountains of data on cases and contacts began pouring into 11 area "servers." But because the servers can’t "talk" to each other yet (a problem software jocks are working to fix), researchers have had to dig information out of each server separately.
Plus, entering a contact activates a clever cross-referencing system, which TB controllers quickly realized could jeopardize confidentiality. So for now, outreach workers have to pull over to the side of the road after their visits (or wait until they get back to the office) and enter data in privacy.
"One of the challenges of birthing a new paradigm like this is the technology keeps changing under your feet, so you’re always in catch-up mode," says Charles Woernle, MD, assistant state health officer for disease control and prevention. "You worry sometimes whether you’re ever going to get to the initial visions, or whether the visions will just keep changing."
More gain than pain
But everyone, Woernle included, agrees the electronic patient record has resulted in more benefits than headaches.
"The things we can already pull out of this system are amazing," says Brook. As soon as the de-bugging process is finished, the state plans to share its work, including the contact-investigation algorithm, with other programs, she adds. "There’s absolutely no reason other people should have to make our mistakes all over again."
The only piece of the contact-investigation puzzle left is where background rates of positivity begin to interfere with other parts of the process, says Bailey.
"After we’d already eliminated people we knew were previous reactors and people from what we call political’ contact investigations, we still found that as we got down to a lower level of transmissions, the people who were inexplicably positive tended to be older," he says. "We think eventually we’ll be able to develop a mathematical formula that predicts when you begin to get into these background rates."
Subscribe Now for Access
You have reached your article limit for the month. We hope you found our articles both enjoyable and insightful. For information on new subscriptions, product trials, alternative billing arrangements or group and site discounts please call 800-688-2421. We look forward to having you as a long-term member of the Relias Media community.