This article is more than 1 year old
That Public Health study? No, it didn't say 'don't do chemo'
When big media goes clueless about big data
The dead don't offer a control group
There are a couple of other reasons the study does not tell us “chemotherapy kills”, as the ill-informed would have it – and yes, they're both about the data, not about the study.
The first is simple: the raw data doesn't tell us about any alternative outcomes, Dr Saunders explained. It's not like (for example) a high-quality clinical study of a new treatment, where you can double-blind the study design and say “the new treatment is better than a placebo”.
Regrettably, we can't take a patient back to before chemotherapy, and see what happens without it.
Dr Saunders also pointed out a more subtle problem: attributing outliers. Since another hot-button headline is that 19 hospitals have been asked to review their chemo practices, that's worth noting. In such a huge system as the NHS, across so many hospitals, only access to the source data could prove whether an outlier is the result of clinical management, or merely a difference in data management.
“Are differences between hospitals the result of a clinical decision, or are they gaps in the data?” he noted.
However: for its purpose, this big data exercise is perfectly appropriate, Dr Saunders believes. The aim was not to assess whether chemotherapy is safe (it's not, and nobody chooses it without a very good reason) or effective (which in a great many cases it is).
The study was undertaken to improve the clinical decision matrix: to identify which patients it offers the best future. ®
Bootnote disclosure
The author's wife has been undergoing chemotherapy for immune-suppression purposes since 2010. Nobody mainlines mustard compounds or their derivatives for fun, and doctors don't prescribe it lightly.