Judicious strep testing can reduce unnecessary antibiotic use in children

Testing every child who presents with a sore throat for group A streptococcal (GAS) pharyngitis can lead to unnecessary antibiotic treatment, particularly when a positive result reflects bacterial carriage rather than an active infection, according to Rana Hamdy, MD, MPH, MSCE.

Hamdy, a pediatric infectious diseases physician at Children's National Hospital in Washington, DC, discussed strategies for improving GAS testing and antibiotic prescribing during the American Academy of Pediatrics (AAP) 2026 National Conference & Exhibition.

In healthy children without symptoms, approximately 12% to 20% can test positive for GAS, Hamdy said. That background prevalence makes selecting the right patients for testing particularly important.

"It's really important that we're judicious about which patients we test in, so that we don't end up overtesting and then leading to overprescription of antibiotics," Hamdy said.

Which children should not be tested for group A strep?

Children presenting with symptoms that strongly suggest a viral infection generally should not undergo GAS testing, according to Hamdy. These symptoms include rhinorrhea, cough, hoarseness, conjunctivitis, and oral ulcers.

In these patients, she explained, a positive test is more likely to represent GAS carriage than the cause of the child's current illness.

Hamdy also advised against routinely testing children younger than 3 years because true GAS pharyngitis and associated complications are less likely in this age group.

Avoiding unnecessary tests is important because once a positive result is available, clinicians may face pressure to treat it even when they would not have initially suspected GAS infection.

Which antibiotics should pediatricians use?

When GAS pharyngitis is diagnosed, Hamdy emphasized that penicillin and amoxicillin remain the preferred first-line therapies.

"We have not shown any resistance to penicillin or amoxicillin, so that should be first line," she said.

For patients with a nonanaphylactic penicillin allergy, Hamdy recommended cephalexin. Clindamycin or azithromycin can be considered when there is a true anaphylactic allergy. However, she cautioned against routinely turning to these alternatives.

"We see a lot of treatment with azithromycin for group A strep pharyngitis, but it's actually far inferior," Hamdy said. "So 30% of group A strep isolates are going to be resistant to azithromycin and clindamycin, so it's really important to remember that it's a far inferior treatment."

The issue also underscores the importance of accurately identifying penicillin allergy. Hamdy said many children carrying a penicillin allergy label are not truly allergic, potentially steering them toward less effective treatment.

How can practices reduce unnecessary strep testing?

Improving testing requires more than changing an individual pediatrician's behavior, Hamdy said. Clinical algorithms can establish criteria for when testing is appropriate, but those protocols should be incorporated throughout the practice workflow.

That means including nurses, medical assistants, receptionists, and other staff who may interact with the patient before the pediatrician enters the examination room.

"So many times a pediatrician will walk in the door, and the patient already has been tested, has been swabbed, and has a positive test," Hamdy said.

Patient education can also set expectations before the visit. Hamdy suggested providing materials explaining why every sore throat does not require a strep test. Practices could additionally use a checklist before testing to confirm that a child meets appropriate criteria, such as being older than 3 years and not having a cough.

Penicillin allergy delabeling workflows represent another potential practice-level intervention, she added.

Could machine learning help identify low-risk children?

Hamdy is also working with machine-learning engineers at Children's National on a multi-institutional study evaluating a tool called Strep App.

Approximately 6000 children presenting for sore throat evaluation have been enrolled. Researchers are combining clinical information with photographs of the back of the throat to develop a machine-learning classifier that can distinguish GAS from non-GAS presentations.

Preliminary analyses have demonstrated approximately 83% accuracy and a high negative predictive value, according to Hamdy.

She said the eventual goal could be to use the technology in telemedicine to identify children at sufficiently low risk that they do not need an in-person visit or strep test.

"We envision that this tool, which is called Strep App, is something that a physician, perhaps in the telemedicine setting, can prescribe to their patient to take and identify those patients at low risk for group A strep who don't even need to come into the clinic," Hamdy said.

References

Hamdy R. Group A streptococcal pharyngitis—are we testing and treating appropriately? Presented at: American Academy of Pediatrics 2026 National Conference & Exhibition; October 2, 2026; San Diego, CA.