Question 5: Is there evidence of statistical heterogeneity?
Yes, there is some evidence. Two of the 13 confidence intervals for individual studies do not overlap (Foster and Witt). This means that there is no common value of the treatment effect which is compatible with both studies, though these are only two out of the 13. The chi-squared heterogeneity test is on the cusp of significance at the 0.10 level. The evidence is quite weak.
Since there is clear clinical heterogeneity, there is no reason to think that these studies should all be estimating the same effect of acupuncture on pain.
Back to question.
Back to Systematic Reviews index.
To Martin Bland's home page.
This page maintained by Martin Bland.
Last updated: 17 February, 2009.
Back to top.