Bell Palsy: What the Steroid and Antiviral Evidence Shows

What does the evidence say about treating Bell palsy?

The American Academy of Neurology's evidence-based guideline update on Bell palsy reaches two clear and unequal conclusions. Corticosteroids earn a Level A recommendation, the guideline's strongest tier, meaning they should be offered to people with new-onset Bell palsy to raise the chance that facial nerve function recovers. Antivirals earn only a Level C recommendation, meaning the benefit is uncertain and, if it exists, is likely small. The gap between those two verdicts is the whole story, and it comes down to the quality and consistency of the trials behind each drug. This article is general education, not medical advice.

What Bell palsy is, and why timing matters

Bell palsy is a sudden weakness or paralysis of the muscles on one side of the face, caused by dysfunction of the facial nerve. It typically comes on over hours to a couple of days, and the cause is not fully understood, though inflammation and swelling of the nerve as it passes through a narrow bony canal are central to most explanations. Many people recover well on their own. That fact is important, because any treatment has to prove it improves on a natural history that is already fairly favorable for a large share of patients.

The recovery question is what the guideline set out to answer. Its authors, Gary Gronseth and Remia Paduga, reviewed the controlled-trial evidence and graded each study using the AAN's standard scheme, which sorts trials from Class I (least likely to be biased) down to Class IV, then translates the body of evidence into a recommendation level from A to U. Understanding that grading system is the key to reading the conclusions correctly.

Why steroids earned the strongest recommendation

The steroid recommendation rests on two Class I trials, meaning two well-designed, low-bias randomized studies pointed in the same direction. When high-quality trials agree, the AAN scheme allows its top recommendation level. In this case, the trials found that people treated with corticosteroids were meaningfully more likely to regain facial function than people who were not. The guideline described a risk difference in the range of roughly 12.8 to 15 percent in favor of steroid treatment.

It helps to translate what a figure like that means. A risk difference in that range implies that out of every hundred or so treated patients, somewhere around a dozen additional people recover facial function who might not have otherwise. That is a genuine, clinically relevant effect, and the consistency across two strong trials is what lets the guideline say steroids should be offered rather than merely might be considered. The language of a guideline is deliberate. "Should be offered" signals confidence; softer verbs signal doubt.

Why antivirals landed in a weaker tier

Antivirals tell a different story, and the difference is instructive for anyone trying to read medical evidence. Herpes simplex virus reactivation has long been proposed as a trigger for Bell palsy, which made antiviral drugs a biologically reasonable thing to test. Reasonable hypotheses, though, still have to survive trials, and here the trials did not deliver a convincing benefit.

The guideline concluded that antiviral agents added to steroids do not increase the probability of facial functional recovery by more than about 7 percent. Read that carefully, because the framing matters. It is not a flat statement that antivirals do nothing. It is a statement bounding how large any added benefit could plausibly be, and that ceiling is low. On that basis the guideline assigned antivirals a Level C recommendation and offered nuanced guidance: patients might be offered antivirals in addition to steroids, but they should be counseled that a benefit has not been established and that, if one exists, it is likely modest at best.

That is a careful, honest way to handle uncertainty. It neither oversells a drug the evidence does not support nor forbids a treatment that some patients and clinicians may still reasonably choose after an informed conversation. Notice also the structure of the comparison. The strong evidence is for steroids, and the antiviral question is framed as whether adding an antiviral on top of steroids buys anything extra. Steroids are the anchor; antivirals are the open question layered on top.

How to read recommendation levels in general

The Bell palsy guideline is a useful worked example of something that applies far beyond this one condition. A recommendation level is a summary of how much the evidence can bear, not a measure of how enthusiastic anyone feels. Level A means multiple strong, consistent trials support the action. Level C means the support is thinner, more mixed, or capped at a small effect. When you see a Level A recommendation sitting next to a Level C one for the same disease, the difference usually reflects the underlying trials, not the plausibility of the biology.

A few habits help when reading any guideline. Look at the class and number of studies behind each claim, because two Class I trials carry very different weight than a handful of small, biased ones. Look at the effect size, not only the direction, because a real but tiny benefit and a large benefit are both technically "positive." And watch the verbs, because "should be offered," "might be offered," and "should not be offered" encode the strength of the evidence in plain language.

What this means in practice

For a person facing a new diagnosis of Bell palsy, the practical shape of the evidence is straightforward. The best-supported step is early corticosteroid treatment, backed by strong trials and a clinically meaningful effect on recovery. Antivirals occupy a more optional space, reasonable to discuss but not something the evidence firmly endorses. Individual decisions still belong in a conversation with a treating clinician, who weighs the specifics of a given case, including timing, severity, and other health conditions that steroids can affect. The guideline sets the evidentiary backdrop for that conversation rather than replacing it.