Skip navigation

Anyone for tennis elbow?



Tennis elbow is common, 4-7 per thousand per year in general practice, according to a Dutch systematic review of injection treatment [1]. Peak incidence is between 35 and 54 years, and only 5% are due to wielding a racquet in anger. They estimate that on average the problem lasts for between 6 months and 2 years, and in Holland between 10 and 30% of patients take time off work with their tennis elbow, and the average time-off is 12 weeks.

Lack of elbow grease

The three commonest remedies used by Dutch GPs are analgesics (18-35%), steroid injections (14-38%) and physiotherapy (28-30%).

The obvious advantages of injection are that the patient doesn't have to take tablets, the injection is easy to do, and no referral is necessary. Incidentally Bandolier really likes the `how-to-do' pictures of various injection procedures in Dr Bull's book [2]. Do the injections work?

Systematic review

Twelve randomised trials were unearthed, five comparing steroid plus local anaesthetic injection with injection of local anaesthetic or saline, two comparing different steroid doses or formulations, and five comparing steroid with other treatment. The Dutch are tough reviewers, and found fault with most of the trials. They did go on to try to pool the results from the various trials.

Bandolier struggled with the validity of their analysis

  1. the outcome of `treatment success' taken from nine trials was necessarily crude because the different trials used different ways of measuring success
  2. the real problem is that they pooled trials which compared steroid versus local anaesthetic or saline with trials which compared steroid with another active treatment, such as NSAID, wrist brace or ultrasound. Not surprisingly pooling apples with pears produced evidence of heterogeneity in the results
  3. within the trials which compared steroid versus local anaesthetic or saline there was variation in the number of injections studied

Results


Bandolier has tried to tease out the results for the trials comparing steroid with local anaesthetic or saline. The best we can do is to count votes. The disadvantage of vote-counting is that it takes no account of whether trials are big or small, no account of the extent or size of the effect found in the various trials, and no account of the relative validity of the trial designs.
The 'does steroid injection for tennis elbow work?' vote-counting Table
Trial
dichotomous data
odds ratio significant at authors' most important follow-up time
odds ratio significant for 2-6 weeks
odds ratio significant beyond 6 weeks
Price et al 1991
Y
Y
Y
N
Day et al 1978
Y
Y
Y
n/a
Murley 1954
Y
Y
Y
n/a
Freeland & Gribble 1954
Y
N
n/a
N
Halle et al 1986
N
n/a
n/a
n/a
n/a means no data available. References are in the systematic review. Y = yes, N = no.


So, three of four trials with available data showed an effect of steroid injection compared with local anaesthetic alone or saline at authors' most important follow-up time. Three of three showed significant short term (2-6 weeks) advantage, and none showed a long term advantage (more than 6 weeks; two trials with data).

What evidence there is, and some of it is pretty ancient, does support an effect of steroid injections, but this is not a clean answer, and the effect doesn't last for more than six weeks.

All that glisters isn't gold


The report shows how difficult it is for all of us to pick out the problems. When you read you need to have your critical faculties honed. Bandolier does not believe that the trials in the review should have been pooled in the way they were, and the data presented is not robust enough for us to dignify it with an NNT.

Shoulder steroid versus elbow steroid


In Bandolier 32 five studies comparing shoulder joint injection of steroid with local anaesthetic or saline also suggested that the steroid injections could produce no useful long term effect (NNT for success beyond four weeks from injection of 17 compared with saline, with a confidence limit which included no benefit to any patient, and for steroid versus local anaesthetic the NNT was 33).

Elbow joint injections may do better short term but don't look different from shoulder injections beyond six weeks. Once again the WD40 approach of injecting steroid is not a long term solution to the problem.

Adverse effects


The adverse effects of elbow injection in the Price trial were pain after injection in 58 of the 116 injected with steroid plus local compared with 9 of 29 with local alone, and skin atrophy in 31 of 116 compared with 5 of 29. These seem a bit high.

References:

  1. WJ Assendelft, EM Hay, R Adshead, LM Bouter. Corticosteroid injections for lateral epicondylitis: a systematic overview. British Journal of General Practice 1996; 46:209-16.
  2. Bull MJV, Gardiner P. Surgical procedures in primary care. Oxford: Oxford University Press, 1995.



previous or next story in this issue