Statistics from Altmetric.com
In the article by García-Arumí and associates,1 the authors reported the results of several techniques used for purification of triamcinolone acetonide suspension for intravitreal injection (two filter and two non-filter techniques). I very much appreciate that they finally recommended our method (centrifugation for 5 minutes at 3000 rpm with extraction of 0.9 ml of the supernatant and pellet resuspension with 0.9 ml of balanced salt solution, BSS) among those compared.2
This could have been an interesting report but I want to raise some important issues mentioned in their paper. The source of the evaluated techniques was not clearly specified by the authors. Are they previously published techniques? Are they techniques developed entirely by García-Arumí et al? As far as I know, the centrifugation technique described by García-Arumí et al was first reported by Hernaez-Ortega in 20033 and it was first published by Hernaez-Ortega and myself in the 2004 July/August issue of Ophthalmic Surgery, Lasers & Imaging.2 In fact, the term “purification” was first applied to removal of most of the vehicle from a commercially available triamcinolone acetonide suspension by us.
Parametric tests tend to be more powerful than non-parametric tests if their distributional assumptions are met. However, non-parametric methods require fewer assumptions, and they are generally robust against problems like outliers and non-constant variances. The authors tested statistical significance by the Kruskal-Wallis test, a non-parametric procedure that is a K-sample generalisation of the two sample rank sum test (also called the Mann-Whitney U test). It tests the null hypothesis of identical group medians, rather than means. With the data available in this report, the authors should have noticed that figures 1 and 2 show means and standard deviations; they should instead have shown group medians. Moreover, their results should have been expressed as medians.4 It also remains unclear how many assays were performed. By looking at figures 1 and 2, it seems that three assays were done each time but somehow they were done in duplicate. This is important to show the real sample size that was used to compute the estimates.
The pellet resuspension with 0.9 ml of BSS in the centrifugation technique appeared in their final recommendation at the end of the paper. However, the authors stated in the methods section that the pellet was resuspended with 1 ml of BSS. This point needs to be explained.
I recommend that the authors clarify the above mentioned issues. I also suggest that they learn the golden rule of fair use of another author’s protected material: take from someone else only what you wouldn’t mind someone taking from you.
The author does not have commercial interest in any product mentioned in the manuscript.