Evidence for non-random sampling in randomised, controlled trials by Yuhji Saitoh.

Area of Science:

• Medical research integrity
• Statistical analysis in clinical trials

Background:

• Numerous randomized trials by Yoshitaka Fujii were retracted due to improbable random sampling.
• Dr. Yuhji Saitoh co-authored 34 of these trials, serving as corresponding author for eight.
• This study examines additional randomized trials where Saitoh was the corresponding author, but Fujii was not involved.

Purpose of the Study:

• To statistically analyze baseline data and outcomes from randomized controlled trials associated with Dr. Yuhji Saitoh.
• To investigate potential non-random sampling and data irregularities in Saitoh's publications.
• To assess the reproducibility and reliability of findings in selected trials.

Main Methods:

• Monte Carlo simulations were employed to analyze baseline data from 32 randomized controlled trials.
• Statistical analysis of muscle twitch recovery ratios and graphical data from Saitoh's publications.
• Comparison of baseline data p-values and homogeneity of outcome ratios across trials.

Main Results:

• 14 out of 32 trials showed baseline data with p < 0.01, seven with p < 0.001.
• Homogeneous distributions of muscle twitch recovery ratios yielded p values below 0.05 for observed Q statistics.
• Graphical analysis revealed coincident curves across multiple publications, inconsistent with random sampling.
• Combined probability of random sampling across 32 trials was extremely low (p = 1.27 × 10⁻⁸).

Conclusions:

• The analysis indicates a high probability of non-random sampling in the examined trials.
• Repetitive graphical patterns suggest potential data fabrication or manipulation.
• Further scrutiny of Dr. Saitoh's research is strongly warranted due to these statistical anomalies.