Pages Menu
TwitterRssFacebook
Categories Menu

Posted by on May 14, 2013 in Statistics 101 |

Relative Risk

This post is still under construction.

Disregard its content, if any, as it only contains raw reference material for the future post.

Relative vs Absolute Risk

Relative Risk – Mathematically correct but very deceptive

[iframe src=”http://www.nutridesk.com.au/content/images/c3efedf7f0e9ac61cf18d7d511931902.jpg”]

Why use Relative Risk instead of Absolute Risk

  • Relative risk provides a big number and this makes for big headlines or a big impact on the person receiving the information.
  • In the example above, looking at mortality rates of 10,000 individuals, a 50 percent reduction could either mean a decreased mortality from 200 individuals to 100¬†individuals a very significant result or it could just as well mean a decreased mortality from a much less significant, 2 patients to 1 patient1.

Why absolute risk?

  • They take into account background rates
  • Given the absolute risk, the relative risk can be derived but not vice versa.

An example of increased risk of pancreatitis with tetracyclines

  • A large population-based case controlled study has shown a 60% increased risk for acute pancreatitis among current users adujsted for relevant confounding factors.
  • So if we know that the incidence in any given country for acute pancreatitis is say 100 cases per 100,000 people (1 per 1,000) then taking tetracyclines may increase the risk for acute pancreatitis from 1 per 1,000 to 1.6 per 1,000 and this then puts this increased risk into perspective.
  • The 60% increase which seems like a highly significant increased risk turns out to have an increased risk of 0.6 per 1,000 people taking these medications.