Skip navigation

Editorial and Spinal cord stimulators [Jan 1997; 35-1]

Looking Ahead

The new year is a good time to take stock and try, as best we can, to map out our way forward for 1997. Bandolier has a number of good intentions for the year ahead.

Evidence of effectiveness

Bandolier plans to continue carrying information on systematic reviews of evidence of effectiveness. Because more are appearing, it should be easier in future than it has been in the past. Regular searches of MEDLINE, for instance, can help, as well as other sources like the Cochrane Library and DARE.

There will be problems, for instance where reviews fail to find randomised trials, so that even though you may have asked for evidence on a topic, the highest quality evidence just isn't available.

Spinal cord stimulators

Take spinal cord stimulators (SCS). Bandolier has been asked for evidence on this. A review exists [1], but failed to find any randomised trials. Turner, Loeser and Bell conducted a thorough literature search in an attempt to determine the place of SCS in chronic pain treatment.

They included 39 studies, but none was randomised: all were case series. A majority (82%) did not have planned study protocols. The review does not give the total number of patients treated or reported. Few studies define patient inclusion or exclusion criteria, or give demographic information on patients treated. The source of follow-up data was unclear in most of the studies.

Twenty-nine studies had at least 50% pain relief as an outcome. Across these studies the mean (probably unweighted, but not stated) was 59% of patients achieving this outcome at some follow-up point, with a range of 15% to 100%. Fourteen studies reporting one year follow up with success defined as stimulator in use with at least 50% pain relief. In these, the mean across study success rate was 62%, again with a range of 15% to 100%.

Fewer studies reported success at later follow up times. At two years five trials reported a mean of 64% success (range 55% to 74%). At five years three studies reported a mean of 53% success (range 50% to 55%). At ten years one study reported 35% success.

Complications were common. In 13 studies any complication occurred in 42% of patients (range 20% top 75%). These were predominantly stimulator or electrode problems (mean 30% and 24% respectively, range 0% to 75%). Infection was less common, occurring in 5% of patients in 20 trials that reported it (range 0% to 12%). Most complications appeared to be minor.

So we have interesting information, but no real certainty that it is right.


  1. JA Turner, JD Loeser, KG Bell. Spinal cord stimulation for chronic back pain: a systematic literature synthesis. Neurosurgery 1995 37: 1088-96.

Diagnostic tests

Bandolier seeks reviews and insight on these. Few exist, though one is reported this month. We continue, though, to be intrigued by the triumph of simple clinical scoring over laboratory technology. One example (that of alcoholic cirrhosis ) this month is obvious, but the underwhelming ability of the biochemistry to make the diagnosis is fascinating, especially when highly publicised court cases make life-turning decisions based on some laboratory estimations.

Help wanted

Bandolier continues to be impressed by the searching questions posed by its readers, but is struck by how few of you see reviews (or other sources of evidence) you feel sufficiently important to ask Bandolier to carry so that others can benefit. Your help is requested in 1997 to bring to Bandolier's attention evidence worthy of trumpeting abroad.

Using reviews

One reader, recently awakened to Bandolier's usefulness, has asked for us to revisit NNT calculations, and how to make the best of a systematic review. This year we will try to find a review which will enable us to 'walk you through' the process and calculations.

next story in this issue