Tuesday, November 29, 2005

Studies suggesting prolonged survival in COPD by ICS may have been biased by "immortal time"

In discussing tiotropium (a recent posting)I mentioned that the best we have seen so far with medication in COPD is alleviating symptoms and decreasing exacerbations and hospitalizations and improving the nebulous quality of life.It was pointed out off line to me that ICS had been shown to prolong survival.However, the study that reached that conclusion may have been flawed by something called the "immortal time bias". Something that obscure (although with Google can anything be obscure for long?) and with as interesting a name as that needs to be talked about a bit. An explanation and example regarding COPD treatment with ICS can be found here.
The author explains that this bias refers to a cohart study in which the followup time during which a subject in one group cannot by definition develop the outcome of interest, such that during that time they are "immortal". A rate ratio based on that type division of treated and untreated will be biased.The literature regarding ICS treatment in COPD is interesting. A meta-analysis demonstrated improved pulmonary function but no difference in number of exacerbations while subsequent RCTs demonstrated no changes in pulmonary function but fewer exacerbations with ICS treatment.Once again the highly touted meta-analysis was trumped by RCTs. References to those studies are found in the introduction of the article referenced above.I take some sort of perverse pleasure in pointing to situations in which the two highest rungs on the epistomologic hiearchy of evidence based medicine when they investigate the same issue appear to contradict each other. This is not a rare occurence.

No comments: