The Dark Side of Early Diagnosis | Hacker News Reader

The Dark Side of Early Diagnosis

prospectmagazine.co.uk

33 pointsDanBC7 comments

There are two main statistical errors touched on in the article. One is that metrics like 5-year survivability are heavily skewed by early detection, even in the placebo case. The other is that false positives in early detection can skew the numbers on effectiveness of treatment, even when the treatment is a placebo.

In cases where the treatment or the screening itself has negative effects, this is thrown into even more confusion, because the screening/treatment combo may actually have net negative effects, but those effects are impossible to measure without experimental data -- i.e. giving the treatment (blindly) to some people who have not been screened or who passed the screening, and withholding it (blindly) from some who failed the screening.

Ethical obligations of medical practitioners essentially prevent us from leaving things untreated when the risk of harm is considered reasonable, so those experiments are very hard to carry out.

Recently (4-5 years ago?) there was a change to the CPR protocols used by US paramedics and ERs. How they managed this is beyond me. Since CPR is a procedure that people perceive as having a higher rate of success than it does in reality, how would you like to be the paramedic that says "your spouse is dead, and also, by the way, on an completely unrelated note, we used an alternative method of CPR, but that probably had nothing to do with their death".

I'd never looked at it this way before but it makes sense. You'd have thought that accounting for the extent of a cancer at diagnosis would be a requirement for determining the success of any subsequent treatment.

Is it cynical to assume that a lot of medical research is next to worthless, done almost as a box ticking exercise. Do research to get a paper published to put on your resume. And a lot more is done by/for pharmaceutical companies who have an interest in their drug been shown in a good light.

I'd call that overly cynical, yes. Not the point about the pharm companies - seems that a big portion of physicians I've talked to pretty much outright disregard results from trials run by them. Even in unbiased trials, however, the struggle between sample sizes/statistical robustness and getting new potentially life-saving treatments out into the real world in a timely manner is real.

You could be right. The reason I said it was that as part of the pathway to becoming a senior hospital doctor (apart from moving a lot, working hard and studying for post-graduate exams) is the requirement to get stuff published. This makes a lot of doctors do research who aren't interested in a career in research. If that was me I'd be tempted to do something that was as easy as possible to get that box ticked.

There's been a lot of interest recently in the concept of diagnostic stewardship around this issue in antimicrobial resistance.

In the "negative" example (non curable cancer) there is a huge difference:

1) living from 62 to 66 blissfully unaware of it (and thus at least living peacefully these 4 years)

2) living from 62 to 66 in a extremely preoccupied status and possibly with the side effets of the (ending up as ineffective) cures

Though in many cases of non-early diagnosis there is the "regret" about "if it had been diagnosed earlier then it would have been probably curable" which may (or may not) be an aggravation of the quality of life in the last 4 years (66-70).

Tough to say what is better for the unfortunate patient.

This was eye opening. I was always under the impression that early detection would mean a different stage of cancer.

Reply on news.ycombinator.com