Researchers Question Efficacy of Antibiotics for Rhinosinusitis

It's a familiar scenario to most family physicians: A patient presents with symptoms of rhinosinusitis, but there is no way to know from that presentation whether the infection is viral or bacterial. Recently published guidelines on prescribing antibiotics for this condition are more stringent than those presented in the past, but many physicians continue to prescribe these antimicrobials in the absence of definitive evidence of the infection's etiology.

Now, a meta-analysis in the March 15 issue of The Lancet questions the efficacy of antibiotics in almost all cases of adult rhinosinusitis. (The article abstract is available online for free; nonsubscribers pay a fee to view the entire article.) However, Theodore Ganiats, M.D., professor and interim chair of the department of family medicine at the University of California, San Diego, says he thinks the conclusions reached by the researchers are faulty. Although he agrees that some points raised in the study are valid, he says family physicians must continue to use their clinical judgment on when to prescribe antibiotics.

The study, a meta-analysis of randomized trials, was designed to assess whether common signs or symptoms or other specific patient characteristics can be used to identify a subgroup of patients who would benefit from antibiotic treatment. Researchers compared trials from the Cochrane Central Register of Controlled Trials, Medline and Embase. All together, the trials selected included 2,640 patients; the final analysis included data from 2,547 patients in nine trials. To be included in the analysis, patients had to present with clinical signs or symptoms such as purulent nasal discharge, unilateral facial pain, toothache, pain when chewing, purulent discharge in the pharynx or pain when bending.

Overall, the researchers found that a mean of 15 patients who had rhinosinusitis would have to be treated with antibiotics before one additional patient was cured (number needed to treat, or NNT, of 15 for 10,000 simulated new patients). Cure was defined in most of the trials as patient-reported symptom resolution. For patients with purulent discharge in the pharynx, the mean NNT was 8.

The researchers also found that treatment effect was almost independent of the underlying risk. Patients in the trial who had the lowest cure rates had an estimated 32 percent probability of cure without treatment; new patients simulated with this degree of risk had a mean NNT of 13. Conversely, patients with the highest cure rates had approximately a 71 percent probability of cure without treatment; new patients simulated with this degree of risk had a mean NNT of 18.

Patients who had a longer duration of symptoms took longer to cure, but they were no more likely to respond to treatment than were other patients; similar results were reported among patients who had symptoms for six days or more, seven days or more, or 10 days or more. In addition, patients whose symptoms were more severe took longer to cure, but they were no more likely to benefit from antibiotic treatment than were other patients. Researchers also determined that moderately severe symptoms do not automatically distinguish a viral infection from a bacterial one.

Based on the results of their meta-analysis, the researchers in this study concluded that, "Common clinical signs and symptoms cannot identify a subgroup for whom antibiotic treatment is clearly justified, given the cost, adverse events and bacterial resistance associated with antibiotic use." Although purulent pharyngeal discharge appears to have some prognostic value, they noted, the fact that eight patients must be treated before one additional patient is cured nullifies the recommendation for prescribing antibiotics for patients who present with this symptom.

The researchers also concluded that patients who present with symptoms that have lasted for seven to 10 days do not automatically warrant an antibiotic prescription. They advocate a strategy of watchful waiting for these patients. "The implication for primary care is that antibiotics offer little benefit for patients with acute rhinosinusitis-like complaints," they wrote.

Ganiats says that although the problem The Lancet study researchers identify -- overprescription of antibiotics for rhinosinusitis -- is a valid one, the researchers go too far when they recommend that almost no one should receive antibiotics when presenting with rhinosinusitis-like symptoms. In all the trials the researchers studied, says Ganiats, "antibiotics are shown to be beneficial," but the researchers conclude that they shouldn't be used.

Ganiats says he worked on some of the guidelines referenced in this study, and he contends that those guidelines are less stringent than the conclusions reached by these researchers. He points out that, for many statins, the mean NNT is 30. Compared with that, he says, these mean NNT findings of 8 or 15 are "not bad." "We don't question the efficacy of statins," says Ganiats, "but we question the use of antibiotics."

In addition, Ganiats faults the researchers for not including a statement that clarifies who decides when antibiotics should be prescribed. That should be a decision made by the physician and the patient, based on the physician's clinical experience and the patient's history, he says. After all, if a patient presents with these symptoms and has had multiple instances of rhinosinusitis that responded to antibiotics in the past, that patient may well need antibiotics, according to Ganiats.

However, Ganiats says that this study does raise one very valid point. When patients present with symptoms of rhinosinusitis that have lasted seven to 10 days, many physicians automatically prescribe an antibiotic. Ganiats points out that this article shows that such a strategy "may not be any better than watchful waiting."