UGA study finds limited evidence to diagnose bacterial sinusitis

Sinus infections remain a leading reason why patients are prescribed antibiotics they don’t need. But that may be because there isn’t a lot of guidance for clinicians on which signs and symptoms accurately point to a bacterial sinus infection.

That’s according to a new study from the University of Georgia, which analyzed the existing medical literature on diagnosing sinus infection.

“The data on how to clinically diagnose sinusitis is not great,” said Dr. Mark Ebell, epidemiology and biostatistics professor at UGA’s College of Public Health and study lead author.

“We found that there just wasn’t a whole lot of data out there for individual signs and symptoms. There were some clinical prediction rules that combine signs and symptom to improve diagnosis, but none of them had been prospectively evaluated, where you develop the rule in one group of patients and then validate it in another group of patients.”

Only one in three sinus infections are caused by bacteria, but almost three in four patients are prescribed antibiotics to treat their illness, said Ebell, and in an age of growing antibiotic resistance, this discrepancy between prescription and need is a problem.

In the study appearing in the Annals of Family Medicine, Ebell and his co-authors sought to identify whether certain symptoms, such as cough, headache, or discolored mucus, best predicted the presence of a bacterial sinus infection.

They found that the clinician’s overall impression of the patient’s symptoms, which relies on the doctor’s experience and judgment, was the most accurate method of diagnosis. A patient complaint of a foul odor and pain in the upper teeth were the second and third, respectively, most accurate symptoms of sinus infection.

Ebell said it is reassuring that the overall clinical impression often pointed toward a correct diagnosis.

“The vast majority of the time, primary care clinicians use their overall clinical impression as their primary tool. They don’t count the number of symptoms or use diagnostic scores. They examine and talk to the patient,” he said, “so it was important to see that it’s the most accurate clinical variable that we found.”

This study is one of over 30 systematic reviews and meta-analyses Ebell has conducted over his career, and he sees them as an important part of advancing the field as well as training the next generation of health professionals who are interested in questions concerning medical practice and public health.

Ebell’s co-authors were either master’s and doctoral students at the UGA College of Public. Each were very involved in reviewing and selecting the studies to include in the meta-analysis, which Ebell says is good experience for public health students at the early stages of their careers. They also participated in the analysis, and helped to write the final article.

“The meta-analysis is used very widely in many fields, including in public health,” said Ebell. “The same methods we apply in this meta-analysis also would apply to any topic where you want to synthesize the data, and it’s a great skill for students working on their doctoral thesis.”

Co-authors were Brian McKay, Ariella Dale, Ryan Guilbault and Yokabed Ermias.

A full version of the paper can be found online here: https://www.annfammed.org/content/17/2/164.full.

Additional coverage at March 21, 2019 ASPPH Friday Letter, Healio and AAFP.

– Lauren Baggett

Posted on March 20, 2019.