I read with interest the article by Mancardi et al. [1] A few remarks should be emphasized regarding the number of enrolled patients, the statistical analysis, and the quality of data:
1. The minimum number of patients for the application of Wilcoxon Rank-Sum test (one-tailed, significance level 95%, statistical power 0.80, effect-size 0.5) should be 106 (calculated with G-Power software), instead of 30 as seen in the article. Moreover, the authors used data of only 17 patients (25 scans for group MTX, 26 scans for group AHSCT).
2. The Last Observation Carried Forward (LOCF) rule was adopted by the authors although it was formerly contested for too simplicistic of an approach. [2, 3]
3. No reference was provided on the distribution symmetry of the data difference around their median value.
4. The supposed negative binomial distribution of data was not supported by any statistical fitting test, such as the Kolmogorov-Smirnov test. [4] Also, it was not demonstrated that such distribution could be considered suitable for each yearly data collection.
5. No reference was provided on the outliers test adopted (2 data eliminated).
6. No reference was provided on the repeatability and the competence of the sole operator involved in RMI analysis.
7. No reference was provided concerning the quality accreditation ISO of the laboratories involved in RMI analysis.
Due to these concerns, the level of phase trial should be reconsidered.
1. Mancardi GL, Sormani MP, Gualandi F, et al. Autologous hematopoietic stem cell transplantation in multiple sclerosis: a phase II trial. Neurology 2015;84:981-988.
2. Little RJ, D'Agostino R, Cohen ML, et al. The prevention and treatment of missing data in clinical trials. N Engl J Med 2012;367:1355-1360.
3. Streiner DL. Missing data and the trouble with LOCF. Evid Based Ment Health 2008;11:3-5.
4. Hazra A. An Exact Kolmogorov-Smirnov Test for the Negative Binomial Distribution with Unknown Probability of Success. RR Journal of Stat 2013;2:1-13.
For disclosures, please contact the editorial office at journal@neurology.org.
I read with interest the article by Mancardi et al. [1] A few remarks should be emphasized regarding the number of enrolled patients, the statistical analysis, and the quality of data:
Due to these concerns, the level of phase trial should be reconsidered.
1. Mancardi GL, Sormani MP, Gualandi F, et al. Autologous hematopoietic stem cell transplantation in multiple sclerosis: a phase II trial. Neurology 2015;84:981-988.
2. Little RJ, D'Agostino R, Cohen ML, et al. The prevention and treatment of missing data in clinical trials. N Engl J Med 2012;367:1355-1360.
3. Streiner DL. Missing data and the trouble with LOCF. Evid Based Ment Health 2008;11:3-5.
4. Hazra A. An Exact Kolmogorov-Smirnov Test for the Negative Binomial Distribution with Unknown Probability of Success. RR Journal of Stat 2013;2:1-13.
For disclosures, please contact the editorial office at journal@neurology.org.