Wednesday, March 22, 2006
Re: st: left-truncation of entry in survival analysis
On 22 Márta 2006, at 12:46, Sue Chinn wrote:
> > So, am I missing something, or are advocates of the first model > deluding themselves? Can left truncation be ignored with age as the > timescale?
One powerful argument for using age as the time variable in survival analysis has to do with the hazard function itself. In modelling risk factors or prognostic factors, the shape of the hazard function is often treated as a high-dimensional nuisance parameter (I love that expression and use it as often as I can). However, in risk estimation the hazard function represents the absolute risk, without which the hazard ratios cannot be interpreted in real-life terms.
Using age as the time variable allows the construction of hazard functions spanning the age range from the youngest entry to the oldest exit. It allowed us, in the SCORE project, to make cardiovascular risk charts showing ten-year risk from datasets many of which contained at most 8 years of follow up.
Since the publication of Ed Korn's paper, there have been a number of papers looking to see if the modelling of risk factor effects is affected by the choice of time metric and the short answer seems to be 'not so you would notice'. I haven't been keeping up with the literature on this, but I don't know of a case where someone got very different results - does anyone else?
Korn's proposal appeals to my sense of elegance, as well as being a gift to those of us who have to deal with absolute risks.
Ronán Conroy firstname.lastname@example.org
* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/
Links to this post: