Neither constant nor a fallacy

doi:10.1136/bmj.b1748

Published 29 April 2009, doi:10.1136/bmj.b1748
Cite this as: BMJ 2009;338:b1748

Letters

Standardised mortality ratios

Neither constant nor a fallacy

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

Mohammed and colleagues doubt the value of hospital standardisedmortality ratios.¹ They describe a possibility of increasedbias consequent on a "constant risk fallacy" occurring if theeffects of risk factors used in model adjustment process areassumed to be constant across hospitals in the subsequent hospitalgrouping, when they are not. Using data from four UK hospitals,Mohammed and colleagues showed effect modification of severalrisk factors across hospitals after adjustment for risk factorsat the population level.

We tested the generalisability of their finding, applying anidentical procedure to two groups of large Australian hospitals:those with hospital standardised mortality ratios close to orabove 100 and those with ratios resembling the hospitals studiedby Mohammed and colleagues.² In both cases, the effect of theCharlson comorbidity index did not differ across hospitals.³Variability in the risk for this important variable is not auniversal phenomenon. The effects . . . [Full text of this article]

David I Ben-Tovim, director, clinical epidemiology unit¹, Richard J Woodman, senior lecturer (biostatistics)², Paul Hakendorf, senior epidemiologist¹, James Harrison, director, AIHW national injury surveillance unit²

¹ Flinders Medical Centre, Bedford Park, SA 5042, Australia, ² Faculty of Health Sciences, Flinders University, Bedford Park, SA 5042

david.ben-tovim@health.sa.gov.au

CiteULike

Complore

Connotea

Del.icio.us Add to Digg

Digg

StumbleUpon Add to Technorati

Technorati What's this?

This article has been cited by other articles:

Mohammed, M. A, Deeks, J. J, Girling, A., Rudge, G., Carmalt, M., Stevens, A. J, Lilford, R. J (2009). Authors' reply. BMJ 338: b1750-b1750 [Full text]