Dr Scott Richardson recently described his answer to this problem. He suggested that there are three levels of personal involvement with the evidence that a clinician may have:

'Applying': Where the clinician applies the recommendations, clinical routines and practice policies that others formulate, after these others find/appraise/synthesise the evidence. The emphasis is then on individualising care to fit the patient's particular circumstances (biologic factors, psychological state, social situation,etc)

'Using': Where the clinician finds up-to-date summaries of evidence, that others have searched/appraised/synthesized, and uses these summaries to build clinical routines and practice policies themselves. The clinician then applies these policies to the patient's predicament.

'Appraising': Where the clinician searches for original research evidence, appraises it, synthesizes it in some fashion, and then uses it to build clinical policies, which then get applied to the clinical situation.

He says that he has just finished a month auditing the admissions to check on this question. Early results suggest that they did much more ‘applying’ clinical routines/practice policies than the other approaches.

He would like to hear of others who have completed similar audits of their own clinical practices at his e-mail address:  wscottr@verdict.uthscsa.edu

Bandolier in November 1998 had a section on painkillers for painful periods. Dysmenorrhoea affects many women of reproductive age, and is a frequent cause of time lost from work or school as well as interfering with daily living. A systematic review tells us how effective painkillers are for primary dysmenorrhoea.

The results for pain relief show that compared with placebo, naproxen (550 or 275 mg four times daily), ibuprofen (400 mg four times daily) and mefenamic acid (250-500 mg four times daily) had numbers needed to treat of between 2.4 and 3.0, with overlapping 95% confidence intervals, indicating no real difference between them. Five trials of aspirin (650 mg four times daily) had a much higher NNT of 9.2, with no overlap of confidence intervals with the NSAIDS. One comparison between paracetamol (650 mg four times daily) and placebo showed no difference between them.

Women taking naproxen or ibuprofen were less likely to have restrictions of daily living. The NNTs were 3.8 (3.2 to 4.6) for naproxen and 2.4 (1.9 to 3.2) for ibuprofen. Aspirin did not have this beneficial effect, and the point estimate for the NNT was 8. There were no data on mefenamic acid.

This systematic review demonstrated that naproxen, ibuprofen and mefenamic acid are effective. Aspirin was less effective and paracetamol 650 mg was not effective in a single study. Ibuprofen is probably the treatment of choice.

Some of the Community health council readers of Bandolier have asked for an explanation of the term ‘Numbers Needed to Treat (NNTs).’ Put simply NNTs tell one the likely number of cases who need to receive a particular treatment for one to receive benefit. The lower the number, the more effective the treatment.

But the NNT is only an average. The evidence for the use of a drug may be from a small sample. To take this into account we use ‘confidence intervals’ or CIs, usually set at 95%. An NNT of 2.4 (1.9 - 3.2), for ibuprofen in allowing women to carry out normal activities of daily living was quoted in the previous article. This means that in similar circumstances to the original trials after treating (on average) 2.4 women one would expect to see one receive benefit. It would be unusual for that number to be lower than 1.9 or higher than 3.2 women treated. The odds against this happening are 20:1, a pretty long shot.

An NNT of about 2 means that the drug is very effective. The confidence intervals are close to the original number. This reflects the fact that there were a good number of patients treated in the original trials and one can therefore have a lot of confidence in the NNT figure.

Ref to Bandolier generally

Bandolier in December had an article worth repeating on urinary catheters.

Apparently about one hospital patient in four has an indwelling urinary catheter. Urinary catheters are associated with urinary tract infection in about 5% of those with an indwelling catheter. Therefore out of every 1000 patients in hospital, about 12 will have urinary tract infection because of their indwelling urinary catheter.

In terms of effectiveness a meta-analysis of silver catheters says yes and no, depending on the type of silver coating employed.

For silver alloy catheters, 11% of patients had bacteriuria, compared with 32% with uncoated catheters . The NNT with a silver-alloy coated catheter to prevent one case of bacteriuria was 4.6.

For silver oxide catheters, 12% of patients had bacteriuria, compared with 14% with uncoated catheters. The NNT with a silver-oxide coated catheter to prevent one case of bacteriuria was 51, obviously very poor compared with the silver alloy coated.