It was on 04 Jul 2017 (in RED BOX WARNING AGAINST KAPLAN-MEIER GRAPHING) that I first pointed out how profoundly misleading Kaplan-Meier survival graphs seem to be, as is recapped by the graph immediately below, and by the RED BOX WARNING just below that, where the incongruity is laid bare upon recognizing that "No. at Risk" means "No. Subjects Remaining in the Experiment" while also noticing that at Months=30 there are zero subjects remaining in both the Bortezomib and Control Groups.
It is important to keep in mind that many researchers report having performed a Kaplan-Meier analysis only because they have used software that carries the Kaplan-Meier name, but without realizing that in the absence of subjects who are fled/shed=censored, such software delivers regular and unobjectionable survival graphs. The accusation that Kaplan-Meier misleads does not apply to studies which lack fled/shed=censored, subjects, because then the graphs are not Kaplan-Meier graphs even though researchers, persuaded by the name on the box the software came in, may call them that. (In my own terminology, "fled" subjects terminate participation seemingly of their own volition, "shed" subjects have participation terminated by the researchers, and "censored" is the conventional term covering subjects vanished for either reason.)
San Miguel et al (2008)
Figure 1. Kaplan–Meier Curves for Overall Survival.
RED BOX WARNING
is not yet an
Black Box Warning, but should be:
Cancer patients are advised that estimates they may be given concerning survival times are likely to be based on Kaplan-Meier graphs which are capable of so radically distorting research data as to affirm that, for example, after 27 months of treatment, 286 patients are still alive, while the very same data simultaneously confesses that the total number of patients researchers have actually seen alive after 27 months is only 4.
During the more than five months since that 04 July 2017 publication, I have been expecting to have pointed out to me some error in my complaint, and thereby being prompted to modify my position. As it turns out, however, my complaint has received not a word of criticism or correction, and neither have my complaints of the same misleading effect of Kaplan-Meier in Mateos (2010) and San Miguel (2013) and Palumbo (2016) and Dimopoulos (2016).
As my challenge to Kaplan-Meier continues unanswered, the silence is being taken to mean that the challenge is valid and that Kaplan-Meier is indeed misleading, in which case the interests of the patient require that you, and the pharmaceutical industry, and the FDA, and the medical profession, and so on, begin to implement the following measures:
- Retract all estimates of survival in all publications relying on Kaplan-Meier.
- Block publication of all submitted papers which rely on Kaplan-Meier survival graphs.
- Assign the work of reviewing and editing and authoring only to those who have scored high on scientific-methodology examinations.
- Inform all patients in therapy, and all subjects in clinical trials, that the Kaplan-Meier estimates of survival that they may be relying on have been retracted.
- Settle all law suits of plaintiffs seeking compensation for injuries suffered as a result of trusting survival data that had been inflated by Kaplan-Meier.
To avoid having to take such disruptive and demoralizing measures as the five above, any rebuttal that is possible should be delivered immediately, and you are among the best-equipped of all people to compose that rebuttal and make that delivery. For one thing, you have a close acquaintance with Kaplan-Meier, as evidenced by your co-authoring innumerable research reports that rely on it, as for example
San Miguel (2008) VISTA,
San Miguel (2013), and
Dimopoulos (2016) POLLUX. And I came across one report which relies on Kaplan-Meier of which you are lead author, namely Richardson (2005), whose date testifies that you have now had at least a dozen years to contemplate and comprehend how Kaplan-Meier works, and to ponder whether its product stands up to scrutiny.
And another reason why you can be expected to be among the very best people in the world to explain Kaplan-Meier is the drug industry having entrusted you to handle $19.6 million of their dollars during 2013-2016 . In comparison, from the day I was born until today, the drug industry has entrusted me to handle zero of their dollars, and so it is clear that if there is going to be any enlightenment disseminated concerning the quality of pharmaceutical research generally, or the trustworthiness of Kaplan-Meier in particular, that enlightenment will flow from you to me, and which enlightenment I patiently await.
So, please send me your answer, which you can conveniently do by clicking on the envelope at the top of the instant page. I will immediately publish your answer in its entirety, and will also make all changes to my writing that your answer reveals to be necessary.
And one more thing — I would suggest that not much mileage can be gotten from the hypothesis that my complaint pertains to only the far-right edges of the Overall Survival graphs that I prefer to focus my attention on. In fact, Kaplan-Meier misleads no matter what the label on the Y-axis, and also well before reaching the extreme right edge of the graph.
Take, for example, the Progression-Free Survival (PFS) graph which is Figure 1 in Moreau et al (Oral Ixazomib, Lenalidomide, and Dexamethasone for Multiple Myeloma, New England Journal of Medicine, 2016;374:1621-1634), with which graph you may be taken to be acquainted, given that you are one of that publication's co-authors, and to which graph I have added some lines and numbers in red, and borders in yellow, serving to draw attention to facts I rely on immediately below.
What the graph tells us, in picture and in word, is that at Months=20.6, half of the 360 subjects in the Ixazomib Group are in a state of PFS, which is to say, are alive and disease-progression-free, which is precisely 360*0.5 = 180 subjects in PFS. ("Half" and "0.5" are marked in yellow in the preceding sentence because they correspond to "Median" being marked in yellow in the graph.)
However, the number of actually-countable Ixazomib subjects remaining in the experiment at Months=20.6 happens to be not 180 but only 22, which I calculate by interpolating between the 26 to its left and the 14 to its right.
In other words, no one has actually seen any but a small fraction of these incorporeal 180 PFS subjects, nor is able to name them, nor can display photographs of them, nor can take their pulse. They are only guessed to exist, mostly among the mass of subjects who are fled/shed=censored, which is to say subjects removed from the laboratory to a world lying beyond the ken of the researchers.
Moreau et al (2016)
Figure 1. Progression-free Survival in the Intention-to-Treat Population.