Unfortunately there has been a bug in the TSA program when calculating the required information size for continuous outcomes and we have now published a new amended version. In the former version the required information size was calculated twice as large as it should be in a fixed- as well as in a random-effects TSA. We apologize for the inconvenience this may cause and urge you to download the new version from the TSA homepage and redo your analyses.

The required information size for dichotomous outcomes has been calculated correct in the old version so there is no need to re-analyse dichotomous meta-analyses.

Nowadays, meta-analyses are typically updated every 1 to 3 years as data from new clinical trials emerge. Repeated updating of meta-analyses force authors to repeatedly employ the same statistical hypothesis testing on accumulating meta-analysis data. This conduct has long been known to inflate the risk of falsely inferring a statistically significant difference between two interventions under investigation. Similarly, repeated statistical hypothesis testing also inflates the risk of inferring no difference exists between two interventions, when the truth is that there is a difference.

In clinical trials where interim analyses (interim hypothesis tests) are performed over the accumulation of patients (a scenario akin to meta-analysis updates over the accumulation of trials), advanced sequential hypothesis testing techniques have long been used to control the risks of false statistical inferences. Empirical studies suggest that such sequential hypothesis testing techniques are equally effective in controlling the risks of false inferences in meta-analyses.

Interesting Links

The TSA manual ought to be referenced as:
Thorlund K, Engstrøm J, Wetterslev J, Brok J, Imberger G, Gluud C. User manual for trial sequential analysis (TSA). Copenhagen Trial Unit, Centre for Clinical Intervention Research, Copenhagen, Denmark. 2011. p. 1-115. Available from www.ctu.dk/tsa

Acknowledgements

We thank cand.jur. Ulla Terese Kræmer, Rigshospitalet, for the development of the Privacy Policy and the License Agreement