Any variability in outcome comparisons adjusted for case mix must be accounted for

BMJ 1999;318:128 ( 9 January )

Letters

Any variability in outcome comparisons adjusted for case mix must be accounted for

The first 150 words of the full text of this article appear below.

EDITOR $---$ Parry et al draw attention to the difficulties faced by those wishing to use comparative outcome data to indicate performance.¹They clearly show the importance of adjusting for differencesin case mix and allowing for random variation by establishing95% confidence intervals for estimates of adjusted outcome. Inaddition to the uncertainty in the observed mortality, however,there is uncertainty in the predicted mortality. The overall lackof clarity in the rankings of the neonatal intensive care unitsmight therefore be even greater if this additional uncertaintywere acknowledged, which would reinforce the reservations expressedabout decision making with these kinds ofdata.

Predictive models are only approximations to reality. They must be estimated from previous data and thus are themselves proneto noise and random fluctuation. Both the size of the originaldataset and the predictive ability of the variables used determinethe precision of the predicted . . . [Full text of this article]

CiteULike

Complore

Connotea

Del.icio.us Add to Digg

Digg

StumbleUpon Add to Technorati

Technorati What's this?

Relevant Article

Annual league tables of mortality in neonatal intensive care units: longitudinal study: Gareth J Parry, Craig R Gould, Chris J McCabe, and William O Tarnow-Mordi
BMJ 1998 316: 1931-1935. [Abstract] [Full Text] [PDF]