Strength
- Clear defined primary outcome variables
- Maximum Excursion – This measures the limits of self-initiated movements as patients shift or lean their centre of gravity, without falling, toward the theoretical limit (100%) in each of the eight target directions.
- Directional Control – This is calculated by comparing the amount of movement toward the target with the amount of extraneous movement.
Defining primary outcome variables in advance reduces the risk of bias. If primary outcomes are not pre-selected, there is a possibility of choosing the most favourable outcome after the fact (post hoc).
Jacqueline French, Lost in a jungle of evidence: we need a compass
- Concealed and randomised allocation
- Participants were randomly assigned to one of the three groups in a 1:1:1 ratio using permuted-block randomisation
- The process is also concealed to ensure true randomisation.
This randomisation process minimizes selection bias.
- Presence of concurrent control group
The Tai Chi group is compared to the resistance training and stretching group as control group in the study, to reduce the risk of bias.Using non-concurrent controls, such as solely comparing the same patients before and after treatment or using external controls (e.g., historical data), increases the risk of bias.
Jacqueline French, Lost in a jungle of evidence: we need a compass - Reasonable control group choices backed by literature
The Resistance Training (ref. 10-12) and Stretching (ref. 7, 8, 9) groups were chosen as control groups because both forms of exercise are supported by evidence in the literature for their beneficial effects in improving posture, balance, and gait. - Well-defined detailed inclusion/exclusion criteria
The study applied strict criteria, using well-known standards like the Hoehn and Yahr scale and the Unified Parkinson’s Disease Rating Scale (UPDRS III), to ensure consistency and reliability. - Objective measure
The study used mainly objective measures with quantifiable measurements, such as computerised dynamic posturography for balance, isokinetic dynamometry for strength, and gait velocity measurements, adding to the credibility of the results. - Reasonable statistical analysis
The study was adequately powered with a sample size calculation that ensured enough participants to detect clinically significant differences between groups. Using statistical analyses, such as repeated measures Analysis of Variance (ANOVA) and negative binomial regression for falls, further strengthens the reliability of the results.For post-hoc comparisons, the study uses independent-samples t-tests to compare group means, which is suitable for pairwise comparisons after ANOVA indicates a significant difference. - Masked (Blinding)
The outcome investigator is unaware of the patient’s original treatment assignment to prevent bias based on the hypothesis.- All study assessors collecting outcome measures will be blinded to the study's hypotheses and group assignments.
- Blinding will be strictly enforced by emphasizing the importance of minimizing assessment bias and regularly checking the assessors' blinding status.
- Efforts will be made to keep study assessors separate
- Research assistants handle administrative tasks and monitor class safety.
- Class instructors delivering the interventions.
- Participants will be instructed not to reveal their group assignment to the assessors at any point.
- Both the investigators and data analysts will be blinded to group assignments and will review coded data, with participant names replaced by numbers.
- High Completeness of Follow-up
90.26% of participants completed the assigned interventions, with an average attendance rate of 77% across the 24-week intervention period. The study adhered to the 'intention-to-treat' principle when dropouts occurred.When patients drop out, results are analysed based on the group to which they were originally assigned (treated or controlled). This approach preserves the similarity in patient characteristics between groups, ensuring more reliable and unbiased results.However, a limitation of the intention-to-treat analysis is that it can introduce bias, as the last recorded outcome for patients who dropped out is compared with those who remained in the study. This creates an unfair comparison due to the time difference between assessments. Nonetheless, given the study’s large initial sample size and high completion rate, the impact of this bias is likely negligible. - Large and diverse sample size
The study utilizes a large sample size of 195 participants, encompassing a diverse range of gender, age, and severity of Parkinson's disease. This diversity enhances the generalizability of the results to the broader population. - Interpretation of Results Based on Observed Cases Analysis
All conclusions and discussions in this article are grounded in the observed data collected throughout the study.
Weakness
- Lacking of non-exercise control as baseline measure
Without a non-exercise control group as a baseline, it is difficult to compare and evaluate the benefits gained from practising Tai Chi. - Lack of Uncertainty Analysis
According to the definition of uncertainties from the Guide to the Expression of Uncertainty in Measurement (GUM).The article primarily discusses Type A uncertainties evaluated using statistical methods related to the analysis of variance. However, it does not address Type B uncertainties, especially those associated with the measuring equipment. This is especially significant when multiple inputs from different devices are needed to produce a final measurement.To enhance the evaluation of the results' reliability, a more comprehensive analysis of uncertainties should be conducted and included in the article.Examples of sources of Type B uncertainties:- Human error in the calibration procedures for the measuring instruments.
- Errors associated with the measuring instruments, as specified by the manufacturer.
- Limitations in the presentation of digital measurements caused by rounding up or down.