We read with great interest the systematic review by Mehrholz et al. We consider it as a comprehensive and innovative systematic review and meta-analysis, but we have a question about the data processing. By checking the data between the systematic review and included original studies, we found Mehrholz et al. [1] analyzed post-treatment outcomes rather than the difference between pre-treatment and post-treatment, which is at risk of generating an unreliable result. Not only when between-studies baselines are similar, but also when between-groups baselines are not significantly different, we can directly analyze post-treatment outcomes, and generate a stable result.

Considering the included studies’ district, enrollment criteria, and so on, we observed huge heterogeneity of baseline among partial studies. For instance, in Kapadia [2], subjects were of American Spinal Cord Injury Association Impairment Scale (AIS) C or D grade, while subjects in Sadeghi [3] were of AIS B and C grade.

Futhermore, we found a huge diversity of baseline between the intervention and control groups in Kapadia [2] and Sadeghi [3]. In Kapadia [2], in the 6-min walk test, subjects achieved 187.9 ± 123.4 m in the BWSTT plus FES group, while they achieved 79.4 ± 83.9 m in the group performing resistance and aerobic training. Meanwhile, in Sadeghi [3], the capacity of 6MWT is 340.00 ± 89.70 m in the BWSTT group, but 685.71 ± 176.51 m in the traditional therapy group. The tremendous diversity at baseline is non-negligible for directly analyzing post-treatment outcomes, because the posttest 6MWT value of the BWSTT group, 640.00 ± 97.99 m, is even lower than the pretest value of the control group, 685.71 ± 176.51 m. Calculating the difference between pretest and posttest, the efficacy of BWSTT is 300 m on average, which is higher than that of the control group (reaching ~130 m on average). Actually, a contrast result was generated by extracting posttest outcomes only (640.00 ± 97.99 versus 814.29 ± 180.51).

In addition, based on our experience and another systematic review published in Spinal Cord [4], the posttest outcome should be replaced by difference between pretest baselines and posttest outcomes when performing meta-analysis to compare the efficacy of merging physiotherapy interventions.