Skip navigation

Bell's Palsy

Bandolier is more and more frequently being asked by GPs about particular problems. Not all can be answered, but we thought we could try for the question "Should steroids be used to treat Bell's palsy?", and whether it is important that treatment should be started within the first 24 hours for it to be effective. Little did we realise!

The textbook answer is straightforward. Cecil (17th edition, 1985) states: "Most authorities recommend treatment with prednisone" and "it is claimed that prednisone treatment should be instituted as soon as possible". The Oxford Textbook of Medicine (1996, 3rd edition) says "There is some evidence that corticosteroids may be advantageous by reducing oedema in the nerve", and "it is justifiable to treat all cases with corticosteroids if seen within a few days of onset, providing no contraindications to such treatment exists."

Not much equivocation here. But neither source uses references in support of their statements - and, indeed, why should they? Textbooks are not original articles, but condensates of knowledge, or general maps of a subject. If we want to know more, then we should consult maps with more miles to the inch - and do some searching of our own, even though here is a problem which is not rare (one case per GP per year) and for which there are clear textbook statements.

The searching strategy to find relevant studies used the free-text terms "Bell's" and "palsy" between the years 1966 and 1995. We found three reports since 1990, only one of which was relevant. This is reported in detail below, and is followed by a commentary about the quality of the study and the level of evidence that it represents.

Prednisolone in Bell's palsy

The study [1] was relatively recent, being published in November 1994 from the University of Alexandria in Egypt. It was not randomised because it was considered unethical to withhold steroids from patients with no contraindications. It was an open study, and blinding of treatments was not used.


One hundred and sixty patients with acute non recurrent unilateral Bell's palsy of no more than six days duration were studied. All had complete or nearly complete facial paralysis. None had severe hypertension, glaucoma, peptic ulcer, cardiac disease, diabetes or were pregnant.

The study group of 97 patients received prednisolone tablets at a dose of 1 mg/kg body weight (maximum 70 mg) in divided doses with meals for six days, and the dose was then reduced gradually over the next four days.

The control group of 67 patients were not given prednisolone or any other treatment other than oral paracetamol for pain. This group comprised patients who had refused prednisolone because of fears of complications or who had a relative contraindication like heartburn or moderate hypertension.

All patients received superficial muscle heating, massage and electrical muscle stimulation for all the paralysed facial muscles, three sessions a week until complete recovery or up to six months.


Initial clinical gradings and later evaluations were made by a doctor who was blind to the patients' grouping. Electrophysiological measurements were also made of nerve function. Four categories of clinical recovery were defined:-
  • Excellent recovery - patients with no evidence of facial muscle denervation and no residual facial asymmetry within six months without complication.
  • Good recovery - patients with minimal or no facial muscle denervation, but had negligible residual facial asymmetry only on close inspection during maximum effort on smiling and within six months of onset.
  • Fair recovery - patients with partial facial muscle denervation and mild residual facial muscle weakness on maximal effort within one year of disease onset.
  • Poor recovery - patients with facial nerve denervation and incomplete clinical recovery (obvious facial weakness) in spite of physical therapy for one year after disease onset.

A satisfactory recovery was defined as excellent or good, and an unsatisfactory recovery was defined as fair or poor.


The results are shown in the table and figure. Without steroid treatment there was a high rate of satisfactory recovery (by 46 of 67 patients, 69%).

The rate of satisfactory recovery in all steroid treated patients was higher (79 of 93 patients, 85%). This produced an NNT of 6; that is, for every six patients treated with prednisolone within six days of onset of the paralysis, one extra patient had a satisfactory recovery compared with no treatment.
When sub groups were examined, the extra benefits conferred by steroid treatment were seen only with treatment within the first 24 hours, where satisfactory recovery was 100%. Here the odds ratio lower confidence interval was 1.9 (well above 1), and the NNT was 3.2. That is, for every three patients treated with prednisolone within 24 hours of onset of the paralysis, one extra patient had a satisfactory recovery compared with no treatment.

Treatment after 24 hours failed to show a statistically significant difference from the recovery rate in the control group. There was a higher recovery rate in all treatment subgroups than in the control group, the lowest rate of satisfactory recovery being 76%.


There are very serious problems with this paper. Foremost is the fact that the study was not random (perhaps for good reasons), and, while the evaluations were said to be blind, it is very difficult to limit leakage of blinding in a study which must have taken at least two years. Absence of randomisation and blinding will lead to overestimation of treatment effects. Neither is the design of the study a classical case-control design.

The numbers in the subgroups were not large. With larger numbers of patients (or more patients included in a meta-analysis), beneficial effects which may be present with corticosteroid administration later than 24 hours after the onset of the facial paralysis. A case perhaps of absence of evidence not being the same as evidence of absence [2].

However, it is also the case that there are at least two randomised controlled trials [3,4], published over 20 years ago, one of which [3] examined large numbers of patients treated with steroid and placebo. They both came to a negative conclusion, that treatment with corticosteroid did not increase overall cure rates. Neither, however, looked specifically at the effects of early treatment, and the preponderance of patients in these studies began steroid treatment more than 24 hours after the onset of facial paralysis.

A number of non-randomised studies, also in the 1970s, showed significant benefit with steroid. In Bandolier 17, the way in which inadequate randomisation and blinding led to over-estimation of treatment effects was examined, and steroid treatment of Bell's palsy seems to be an example of that.

Only one report in our search mentioned the effects of early treatment. This study in an Indian journal is still awaited, but there was no indication that it was randomised.

What should GPs do?

Make up their own minds based on local guidelines and their experience, just as now.

The evidence that corticosteroid treatment of Bell's palsy is effective more than 24 hours after the onset of facial paralysis, and the (weak) evidence that early treatment may have some beneficial effects, come from non-randomised trials. We know that these are likely to exaggerate the effects of treatment; two randomised trials were negative.

This is a classic problem, where textbooks appear to be regurgitating the advice from non-randomised studies and ignoring the randomised. What is needed for Bell's palsy is a thorough systematic review of the literature, and we hope that friends of Bandolier will undertake it.

Even then, this may not be enough. To answer the original questions, especially that of the benefits of early administration of steroids, a new randomised controlled trial may yet be needed.


  1. TS Shafshak, AY Essa, FA Bakey. The possible contributing factors for the success of steroid therapy in Bell's palsy: a clinical and electrophysiological study. Journal of Laryngology and Otology 1994 108: 940-3.
  2. DG Altman, J M Bland. Absence of evidence is not evidence of absence. British Medical Journal 1995 311: 485.
  3. SM Wolf, JH Wagner, S Davidson, A Forsythe. Treatment of Bell palsy with prednisone: a prospective randomized study. Neurology 1978 28: 158-61.
  4. M May, R Wette, WB Hardin, The use of steroids in Bell's palsy: a prospective controlled study. Laryngoscope 1976 86:1111-2.

previous or next story in this issue