A recent study by Wang et al. searched the available databases and conducted a systematic review and meta-analysis to investigate the optimal parameters of a tai chi intervention to improve balance performance of older adults, which was published in the recent issue of Journal of the American Geriatrics Society.1 They came to the conclusion that “Tai chi is effective at improving the balance ability of adults over 60 years of age and a medium duration and high frequency of 24-form tai chi may be the optimal program for improving balance.” Although the results are of great importance, we would like to underline some issues as to registration of the studies, quality assessment, and data analysis that we feel ought to be considered when interpreting these findings.

First, there was no published protocol or registration for this study. Registering a meta-analysis protocol is important as it enables the promotion of transparency and avoidance of potential biases including both selection and selective outcome reporting biases.2

Second, the authors have clearly mentioned that “The methodological quality of included studies was assessed using the Physiotherapy Evidence Database (PEDro) scale” in the Methods section. However, the results of some evaluations may be inaccurate. For example, the studies of Lu et al.3 and Li et al.4 did not explain the specific methods of randomization. Furthermore, the Kappa score, which measures agreement between the reviewers, should also be provided in the article.

Finally, according to the Cochrane Handbook for Systematic Reviews of Interventions 5.0,5 a high inter-study heterogeneity makes definitive conclusions hard to draw, thus high heterogeneity in the outcomes is our biggest concern. The authors attempted to explain the source of heterogeneity by adequate meta-regression and subgroup analyses, but these did not guarantee that the outcomes were sufficiently stable. Compared with the random effects model used by the authors in the article, a model called inverse variance heterogeneity (IVhet) could have been used as an improved alternative to the random effects model. It was shown that the known issues of underestimation of the statistical error and spuriously overconfident estimates with the random effects model could be resolved by the use of an estimator under the fixed effect model assumption with a quasi-likelihood based variance structure.6 Therefore, we recommend using the IVhet model to re-analyze the outcomes in this meta-analysis.

We respectfully appreciate that Wang et al. provided us with an important meta-analysis which can provide a guide for clinical decision-making. However, more studies with large sample size and good scientific design should be carried out to clarify this issue. We would welcome some comments by the authors as this would help us to further support the findings of this important clinical trial.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

Xin Tian contributed to conceptualization, methodology, writing - review and editing, Aoran Yang contributed to investigation and writing the original draft.

SPONSOR'S ROLE

The sponsor had no role in the design, methods, data collections, analysis, and preparation of this paper.