Advertisement
Research Article

Long-Term Risk of Incident Type 2 Diabetes and Measures of Overall and Regional Obesity: The EPIC-InterAct Case-Cohort Study

  • Published: June 05, 2012
  • DOI: 10.1371/journal.pmed.1001230

Reader Comments (1)

Post a new comment on this article

Why categorise continuous variables..

Posted by nshephard on 06 Jun 2012 at 14:30 GMT

Its unfortunate, from a statisticians perspective, that the authors chose to categorise continuous variables as there are a number of reasons not to do so, it reduces power, loss of information/precision about the underlying variation, assumes constant risk across categories, categories are often derived based on data (percentiles of differing types) etc.1 2

We have sophisticated statistical methods and software capable of handling and analysing large data sets with continuous outcomes, and these should be utilised when analysing and presenting data to avoid the many pitfalls

1 http://biostat.mc.vanderb...
2 http://www.biomedcentral....

No competing interests declared.