A Web-based tool designed to help parents and caregivers decide whether their children are sick enough with an influenzalike illness to warrant a trip to the emergency department showed woeful specificity in a pilot test of 294 adults.
"Our findings present a cautionary tale regarding the potential effects of self-triage tools," wrote Rebecca Anhang Price, Ph.D. and her associates in an article published online on Dec. 12.
"Safety was the chief concern" of the working group set up by the Centers for Disease Control and Prevention and the American Academy of Pediatrics that developed the Web-based triage tool in 2009 for parents and caregivers to use at home during that season’s novel influenza A (H1N1) pandemic. But the results of the current pilot study, which the CDC and AAP decided was necessary before making the triage tool available, showed an "unintended consequence" of the cautious approach taken by the working group: "Had it been made available to the public in its current form, it might have led more, rather than fewer, parents to bring their children to an ED [emergency department]," the authors wrote.
"Our goal was to reduce a massive surge of anxious parents" bringing children with flu-like illness to the ED, said Dr. Arthur L. Kellermann, senior investigator on the study. The web-based triage tool would ideally "help people who would otherwise opt to go to the ED recognize that their child was not at high risk and could instead be seen by a primary-care physician or walk-in clinic. We know that a number of kids who are brought to the ED don’t need to be there," he said in an interview.
The challenge is to devise a triage tool simple enough for most parents to use that will wind up directing essentially all critically ill children with influenza to an ED while maximizing the number who are not in immediate danger away from emergency services. "The goal is to provide help that enables that to happen more consistently," said Dr. Kellermann, an emergency medicine physician, senior policy analyst, and director of RAND Health in Arlington, Va.
The results of the triage tool’s pilot test showed how hard a goal that is. The study involved parents and caregivers aged 18 year or older who had brought a child aged 18 years or younger with a flu-like presentation during Feb.-April 2012 to either of two EDs, at Children’s National Medical Center in Washington at Inova Fairfax Hospital in Virginia. Parents worked through a Web-based questionnaire, called Strategy for Off-Site Rapid Triage (SORT) for Kids, in an average of just over 4 minutes. The tool used questions that included: Is it hard for the child to breath or is she breathing fast? Is she confused? Or having trouble staying awake?
To assess the accuracy of the triage tool’s decisions, the researchers matched the SORT for Kids recommendations against the actual outcomes of each case using electronic health records available for 286 children brought in by the 294 participants (Arch. Pediatr. Adolesc. Med. 2012 [doi:10.1001/jamapediatrics.2013.1573]).
Based on the health records, 15 of the ED visits (5%) were deemed clinically necessary. In addition, 8 of 165 parents (5%) reached for follow-up information reported bringing their child back to the ED for flu-like symptoms within a week of the index visit. None of these visits involved children whose cases had initially been classified as "necessary" according to explicit clinical criteria.
Of the 15 children who met explicit criteria for initial clinical necessity, 14 were flagged as high risk by SORT for Kids, a sensitivity of 93%. The one false negative was a 4-year old described by a parent as having a cough but not fever; the child subsequently needed intravenous fluids for dehydration and was discharged with a prescription for antibiotics and chest radiograph–confirmed diagnosis of pneumonia.
Among the 271 visits initially judged by clinical criteria as not medical emergencies, SORT for Kids identified 28 as low risk and 7 as intermediate risk, for a specificity of 13%. "The main reasons the algorithm classified so many of these children as high risk were that survey reports that the child had not urinated in the last 8 hours, was ‘fussy or cranky,’ was ‘much sleepier or more tired than usual,’ or was confused," the researchers said.
Sort for Kids initially flagged all eight children who required a return ED visit as high risk, a sensitivity of 100%. Of the 157 children with follow-up data who did not need later ED care, SORT for Kids had identified 13 as low risk and 4 as intermediate risk, a sensitivity of 11%.