Skip navigation

Bell's palsy systematic reviews


Clinical bottom line

Treatment of Bell's palsy (idiopathic facial nerve paralysis) with adequate doses of oral steroids for about a week soon after onset will help between 10% and 20% of affected persons to complete cure over four months. About half will recover completely without treatment. Acyclovir and surgery for facial nerve decompression are unproven.


M J Ramsey et al. Corticosteroid treatment for idiopathic facial nerve paralysis: a meta-analysis. Laryngoscope 2000 110: 335-341.

PM Grogan & GS Grosneth. Practice parameter: steroids, acyclovir, and surgery for Bell's palsy (an evidence-based review). Neurology 2001 56: 830-836.


Bell's palsy affects about 20 persons per 100,000 every year. In about half it will resolve spontaneously within a few months. Treatments are aimed at getting more people better, and quicker. These two systematic reviews and meta-analysis are useful in examining how good the evidence is.


Both reviews examine steroids, but do it in different ways. The most useful is Ramsey et al, because it is more definite about what it will accept or not accept in the way of trials, and because it gives raw data fro the number of people with complete cure at four months.

The Ramsey review used a wide search strategy from 1966 to 1998. For inclusion there were eight inclusion criteria, summarised as prospective with controls, definite diagnosis of idiopathic facial paralysis which was unilateral and non-recurrent, treatment started within seven days of total dose of prednisolone of 400 mg or more, no other interventions, and complete outcomes for all patients over four months or more.

Three randomised trials met all the criteria, and six additional trials failed on one or two criteria. Figure 1 shows a scatter plot of the cure rates for steroid and control (with traisl meeting all criteria as red symbols and those not meeting all criteria as blue symbols). Table 1 gives the actual numbers, together with relative risks and NNTs.

Figure 1: L'Abbé plot of steroids and control for four months cure rates

Table 1: Analysis of steroid effectiveness by trial quality


Quality of study

Number of trials



Relative benefit (95% CI)

NNT (95%CI)



62/124 (50)

44/106 (42)

1.4 (1.03 to 1.8)

12 (5 to no benefit)



324/405 (80)

143/256 (56)

1.5 (1.3 to 1.7)

4.1 (3.2 to 5.9)

All trials


385/529 (73)

187/62 (52)

1.5 (1.3 to 1.6)

4.7 (3.6 to 6.7)

Clearly trials with a lower quality were more optimistic than trials of higher quality. One very small trial apart (with 24 patients, the smallest of the nine), the trials all lay above the line of equality demonstrating efficacy for steroids.

The response rate with control, placebo in this case, was variable, even within tight definitions for what constitutes Bell's palsy. This is probably simply a reflection of the small size of most of the trials. As Figure 2 shows, the overall control recovery rate was 52% in the nine trials, and the largest trial with 105 patients given control was close to the overall average.

Figure 2: Recovery rate without steroid

The review of Grogan & Gronseth uses mostly the same trials, but includes one additional small trial from 1954.


Grogan & Gronseth found three studies prospectively examining the effect of acyclovir on Bell's palsy compared with patients not so treated. The number of patients was 230, and no trial showed any significant benefit for complete recovery. Recovery rates without acyclovir were 76% to 100%, making benefits of treatment very hard to find.

Facial nerve decompression

Grogan & Gronseth found four studies comparing outcomes in patients treated with surgery with those not so treated. The number of patients treated was 227, and recovery rates without surgery were between 15% and 47%. One trial, using mid cranial fossa and meatal foramen approach showed benefit for complete recovery.


The weight of evidence is not overwhelming. Despite differing about their conclusions about steroids to some degree, both reviews think that early use of oral steroids is probably effective in improving outcomes in patients with Bell's palsy. The extent is perhaps between 1 in 10 to 1 in 5 patients benefitting. The evidence for other treatments is inadequate.