Glaxo Smith Klyne fined $4.2 BILL for killing kids., page-15

Agree to a point. Ibuprofen is known to produce secondary effects after long term use, including gastric bleeds, renal problems, and blood clotting problems.

Paracetamol is no different, only with a host of other symptoms of long term use, liver dysfunction, clotting issues and ascities.

Both of these drugs are well known to ICU and ED staff, being part of the "a bit is good, so a lot must be better" regular overdose presentations.

The second link in that article is a link to a biased blog, and not to the Cochrane Society papers (which are freely available if you look)

The point they are actually making is that the key to VALID research (as distinct from fraudulent by intent) is that there must be the same framework around what is being studied (apples for apples) for it to be repeatable and therefore valid. If research techniques change some variables outside the ones that are being examined then the research, although "valid" in that single instance, is not valid for comparison.

This is why looking at your one-source articles and cut/pasting them as "valid" research rubs me up the wrong way. Dig deeper, and examine what is being studied at the grass roots, raw, numerical level (not what someone else is spoon feeding you), find at least three supporting studies (done with similar thoroughness to research design) and THEN you have a case.