Brief background: statins are drugs that lower cholesterol. Trials comparing statins with placebo show that statins lower mortality (in terms of hazard ratio) by almost 20% (hazard ratio 0.8). There are numerous trials conducted with similar findings. Mean follow-up is approx 4 years in these studies.

Now comes this study: http://bmjopen.bmj.com/content/5/9/e007118 - which is based on the statin trials and this study concludes that in terms of survival time, satins prolong life by 5 days.

This cannot be mathematically sound.

Could I go backwards and estimate the hazard ratio that these numbers would yield? Because the hazard ratio should be approimately equal to the studies it was based on. This is a way of prooving that the study results are incorrect.

  • $\begingroup$ I did not read the paper, if the statin can only prolong life for 5 days, I would prefer do not take these medicines for the cost and incovenient! $\endgroup$ – Deep North Nov 8 '17 at 11:55

Which mortality is reduced by statins? The overall mortality, or the specific mortality from heart disease? If the latter, this can certainly be consistent with extending lifespans by only five days.

Compare the effect that more frequent screening can detect cancers in early stages that can be treated and healed (increasing cancer survival rates) - but that these cancers would likely never have been fatal, so people will die of other causes anyway, with or without screening. So overall, screening artificially inflates cancer survival, but does little to nothing about life expectancy or mortality.

  • $\begingroup$ In these trials, mortality is primarily driven by cardiovascular disease (these are generally high risk persons with respect to cardiovasc dis) so the large reduction in cardiovasc mortality should be reflected upon overall survival. $\endgroup$ – Adam Robinsson Nov 8 '17 at 14:18
  • $\begingroup$ I just found this: bendixcarstensen.com/DMreg/YLL-handout.pdf which I presume explains what they did wrong? $\endgroup$ – Adam Robinsson Nov 8 '17 at 14:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.