Biostatistics is undoubtedly an ethics-laden undertaking, as illustrated by [1,2] among others. But I’m convinced that there are important experiential (not merely theoretical) aspects to this perspective.
Following up on this tweet referencing [3] and the AACR’s Scientist↔︎Survivor Program, I’d like to solicit biostatisticians’ and trialists’ stories of experiences (whether as a patient, a caregiver, or a professional engaged formally with patients or patient advocates) that altered their perspectives on ethical dimensions of the practice of biostatistics.
I wish to recuse myself from the rest of this thread, so as to leave it open for others’ contributions.
-
Altman DG. Statistics and ethics in medical research. Misuse of statistics is unethical. Br Med J. 1980;281(6249):1182-1184. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1714517/pdf/brmedj00045-0026.pdf
-
Berry DA. Bayesian Statistics and the Efficiency and Ethics of Clinical Trials. Statist Sci. 2004;19(1):175-187. doi:10.1214/088342304000000044
-
Ciccarella A, Staley AC, Franco AT. Transforming research: engaging patient advocates at all stages of cancer research. Annals of Translational Medicine. 2018;6(9). doi:10.21037/atm.2018.04.46
the first clinical trial i worked on, 20 years ago, was a factorial design randomising people in northern queensland (australia) to sunscreen/no sunscreen and beta-carotene/ placebo. We counted new lesions (nonmelanoma skin cancer, SCC, BCC, and solar keratoses), over 4.5 years. This is a high risk group. I’m from queensland, my family has scc, bcc, melanoma. The study reported the highest rates of nonmelanoma recorded at the time.
To this day I remain ambivalent about the merits of the study. Obviously compliance is going to be an issue and may dilute the treatment effect to such an extent that our efforts become futile: people randomised to the no sunscreen group modify their behavior (they stay indoors or wear long sleeves). Fair enough too!
It feels like scientific scepticism is misplaced, and a clinical trial appears to be a rarefied toy for academics who seek publications. After all the results won’t affect anything: in school we weren’t allowed outside without a hat on, you cannot be more stringent than that. Thus a positive result retrospectively justifies behaviour and a null result is ignored.
And what would a bayesian think about all this? There is no direct evidence that sunscreen prevents nonmelanoma skin cancer in people, but it had been shown in animals, and we do know sunscreen prevents sunburn and there is a link between no. of bad sunburns and future nonmelanoma skin cancer.
At that point maybe we can be pragmatic and the detached and costly tools of science and the academics who espouse them should be scoffed at?
edit: they even considered using a placebo sunscreen but learned that it enhanced the effects of UV on the skin
3 Likes