Archived Comments for:
Evaluation of a structured goal planning and tailored follow-up programme in rehabilitation for patients with rheumatic diseases: protocol for a pragmatic, stepped-wedge cluster randomized trial
Power calculation in the article "Evaluation of a structured goal planning and tailored follow-up programme in rehabilitation for patients with rheumatic diseases: protocol for a pragmatic, stepped-wedge cluster randomized trial".
Ingvild Kjeken, Diakonhjemmet Hospital
9 October 2014
As the authors of this article, we would like to inform that we have discovered an error related to the sample size and power calculation in the protocol. The intra- cluster correlation coefficient is < 0.01, not 0.3 as cited in the article. To be able to detect a 10% difference (10 points) in the primary outcome the Patient Generated Index (PGI) with a 0.05 level of significance and a power of 0.80, when the standard deviation is 14 units in the baseline PGI scores, the mean number of patients in each cluster is 52 and the drop-out rate is 20%, we will need a total of 124 participants, not 312 as cited in the article. As we discovered this error after the inclusion period was finished, we have already included 312 participants. A consequence is that the included sample will give us a possibility to detect a 10% difference in the PGI score with a 0.01 level of significance and a power of 95%.
Power calculation in the article "Evaluation of a structured goal planning and tailored follow-up programme in rehabilitation for patients with rheumatic diseases: protocol for a pragmatic, stepped-wedge cluster randomized trial".
9 October 2014
As the authors of this article, we would like to inform that we have discovered an error related to the sample size and power calculation in the protocol. The intra- cluster correlation coefficient is < 0.01, not 0.3 as cited in the article. To be able to detect a 10% difference (10 points) in the primary outcome the Patient Generated Index (PGI) with a 0.05 level of significance and a power of 0.80, when the standard deviation is 14 units in the baseline PGI scores, the mean number of patients in each cluster is 52 and the drop-out rate is 20%, we will need a total of 124 participants, not 312 as cited in the article. As we discovered this error after the inclusion period was finished, we have already included 312 participants. A consequence is that the included sample will give us a possibility to detect a 10% difference in the PGI score with a 0.01 level of significance and a power of 95%.
Competing interests
None declared