Hacker News Reader

Comment by pc2g4d | Hacker News Reader

It's not just a waste of funders' money; it's not just an ethical problem because subjects expected to be advancing science; in my view the real problem with this low disclosure rate is that it skews the publicly available literature, almost universally in favor of the proposed new treatment. The low disclosure rate is the flip side of publication bias. It has the tendency of favoring the interests of corporations selling treatments at the expense of patients who end up wasting their money on ineffective (or less-effective-than-expected) new drugs/methods/procedures.

View on HN

This problem goes far beyond clinical trials to science in general. What kind of journal wants to publish results that are negative?

The only time I've ever seen negative results published is if someone tries to repeat "known" science, fails to do so, but finds something interesting in the mean time.

I used to hear these complaints about negative results in chemistry. The problem is, most scientific experiments (including clinical trials) are negative. If published, you'd need to increase the number of papers by ~500% to include all these failed experiments.

This is a pretty big problem. I've given it some thought, but haven't cone across a good answer. Incentivizing this is hard right now.

The best idea I've had is to minimize effort for negative experiments. Like arxiv, but perhaps a even less formal. Allow it to be easily searchable. It's still not incentivized, but it's a fuckton easier to put it out there.

I could also see dedicated data analysts publishing negative data of open data sets.

Journals requiring trials to be preregistered could help, if the registration included a requirement to report at least informally on the outcome of the trial.

Yes. This is something that Ben Goldacre has been railing on, as it's one of the biggest problems in medicine. https://www.ted.com/talks/ben_goldacre_what_doctors_don_t_kn... <-- Relevant talk.

http://www.alltrials.net/ <-- Relevant site

Reply on news.ycombinator.com