With the publication of a 11th article we complete our series of hopefully informative articles on statistical issues in haematopoietic cells transplant studies [1]. With the help of experts from many centers we tried to cover a range of topics of interest to clinical scientists and especially clinical trialists. The series, begun in December, 2018, was a 3 year effort. When we introduced the series we identified several challenges in interpreting data from clinical trials (Fig. 1), included in the Introduction, showed the correlation between people drowning after falling out of a fishing boat and marriage rate in Kentucky. We now add Fig. 2 correlating per capita cheese consumption and numbers of people who died by becoming entangled in their bedsheets. Correlations are not cause-and-effect! Although eating too much epoisses in bed (especially with cognac) might make you lightheaded and therefore more likely to get entangled in your bedsheets this seems an unlikely explanation. This issue is discussed elsewhere in a more mathematical context [2] (https://web.archive.org/web/20190925212058/http://www.burns.com/wcbspurcorl.htm).

Table 1 Articles all of which are or will soon be available as a NATURE Collection at https://www.nature.com/collections/ejhigdbeeh.

Several articles explain complexities of survival and cumulative incidence analyses including those by (Hu, Gale and Zhang/Othus, Zhang and Gale and/de Wreede, Schetelig and Putter/Wei and Peng). Another article provides a SAS macro readers can use to estimate direct adjusted survival functions for time-to-event data with or without left truncation (see the correction at [3]).

There are also articles on when and how to use spline plots (Gauthier, Wu and Gooley), on precision medicine (Moodie and Krakow) and on case control study design (Cai, Kim). Another article deals with the complexities of analyzing and interpreting results of measurable residual disease-testing (Othus, Gale, Hourigan and Walter).

There is considerable confusion regarding the correct interpretation of statistical power explained in the article by Fraser. For example, it is improper to calculate a study’s power retrospectively. A last article with what some might consider a philosophical bent deals with the fundamental issue of causal inference (Zheng, Dai, Gale and Zhang). It shows, for example, limitations in implying causation even from results of a randomized controlled trial. Readers may also be interested in two other relevant articles, one by us entitled What is a p value anyway [4] and a second on the role of observational databases in analyses of transplant outcomes [5].

We hope this collection of articles will be of interest to readers of Bone Marrow Transplantation and encourage you to read them carefully. Each article is preceded by a Note from us explaining the relevance of the content to clinician scientists. You may want to share some with statistical colleagues at your center. We have intentionally gone light on equations and the like, so you need not fear.

As we wrote in the series Introduction: We can be reached on Twitter at #BMTStats. Our operators are standing by.