Jump to comment:

• Experience versus Evidence
  Sidha S Sambandan
  Published on: 9 March 2005
• Caution: under-powering a study can lead to overstated conclusions
  Anne-Maree Keenan
  Published on: 29 November 2004

• Published on: 9 March 2005

  Experience versus Evidence

  • Sidha S Sambandan, GP

  Dear Editor,

  Over quarter of a century of experience with injecting steroids has led me to believe in the value of steroid injections for OA of small and large joints. I have used Depot-Medrone with lidocaine for the CMC Joint, especially in the elderly who have disabling CMC joint OA, leading to inability to lift a pot of Tea, an extremely important activity of daily living for the elderly at least in UK! Many...

  Show More

  Dear Editor,

  Over quarter of a century of experience with injecting steroids has led me to believe in the value of steroid injections for OA of small and large joints. I have used Depot-Medrone with lidocaine for the CMC Joint, especially in the elderly who have disabling CMC joint OA, leading to inability to lift a pot of Tea, an extremely important activity of daily living for the elderly at least in UK! Many of these patients cannot tolerate many analgesics and anti inflammatory drugs, compounded by co-morbidities. In my experience the pain relief has been as long as 3 years, but could be as short as 4 weeks for very few. The pain and functional improvement has been longest with injections for OA in CMC Joints and first Metartarsophalageal joints without large osteophytes. A community based study on a larger number of patients will be illuminating! CMC Joint OA is very common in primary care.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 29 November 2004

  Caution: under-powering a study can lead to overstated conclusions

  • Anne-Maree Keenan, Research Fellow
  • Other Contributors:
    • Ai Lyn Tan, Paul Emery on behalf of the Rheumatology Journal Club.

  Dear Editor,

  We wish to comment on the conclusions of a study recently published in Annals: A randomized controlled trial of intra-articular injection of the carpometacarpal joint of the thumb in osteoarthritis (Ann Rheum Dis 2004: 63:1260-1263).

  This study was designed as a randomized control trial, which was powered to detect a difference "between the placebo and steroid injection which was likely t...

  Show More

  Dear Editor,

  We wish to comment on the conclusions of a study recently published in Annals: A randomized controlled trial of intra-articular injection of the carpometacarpal joint of the thumb in osteoarthritis (Ann Rheum Dis 2004: 63:1260-1263).

  This study was designed as a randomized control trial, which was powered to detect a difference "between the placebo and steroid injection which was likely to be clinically significant was 20%" (pg. 1261). Non parametric prospective power calculations established that 45 patients were required in each group in order to detect a difference with an alpha level of 5% and power of 80%. The authors are to be commended for planning their study with this in mind. However, the authors reported difficulty in recruitment which, rightly so, resulted in the early termination of the project.

  While the authors report the difficulty in recruitment to achieve appropriate power, it appears that they have not considered the impact of the under-recruitment on their results and therefore the conclusions that they can draw from this study. The purpose of appropriate a-priori powering of a study is to ensure that the chance of incorrectly accepting a null hypothesis (i.e. a type II error) is sufficiently small. It is a way of ensuring that a non-significant result is probabilistically likely to be due to no difference between two groups.

  Failure to reach the intended recruitment leads to difficulty in interpreting studies with non-significant results. Two possible explanations may be drawn from such non significant results:

  1) There may be no actual difference between the group or
  2) There may indeed be a difference, however the sample size was insufficiently large to detect this.

  Either way, this conundrum must be reflected in the discussion and conclusions of the study.

  As such, the authors of this article cannot claim, as they do in their conclusions that "no clinical benefit was gained from inta-articular steroids into the CMCJ...". At best, they can claim that there study, due to the unfortunate, but very real problems associated with recruitment for hand osteoarthritis trials, was unable to provide conclusive evidence to evaluate whether steroids were effective.

  Reference

  (1). Baskin L. Statistical interpretation can also bias research evidence. BMJ 2003; Jun 28;326(7404):1453-5.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top