Significance test for differences in standard mortality ratios
- From: "reaf" <drreaf@xxxxxx>
- Date: 1 Jul 2006 05:45:11 -0700
The table below gives a hypothetical example for standard mortality
ratios (SMR) with their 95% Poisson lower (LCL) and upper (UCL)
confidence limits for the number (obs) of deaths in three different
populations (A,B,C) and the expected number (exp) in their respective
appropriate reference population.
What is the exact test for the null that SMR-B = SMR_A and SMR-C =
SMR-A ?
Asymptotically I would use the (scaled) variances s^2 = obs, but I
would prefer an exact test.
Popl. | obs/ exp : ( LCL, SMR, UCL )
A | 55/ 75.34: (0.5500,0.7300,0.9502)
B | 44/ 81.73: (0.3912,0.5384,0.7227)
C | 33/ 61.28: (0.3707,0.5385,0.7563)
Many thanks in advance, Rainer Facius
.
- Follow-Ups:
- Prev by Date: Re: Are graded clinical signs more reliable than dichotomized?
- Next by Date: Re: varimax rotation code
- Previous by thread: Re: Are graded clinical signs more reliable than dichotomized?
- Next by thread: Re: Significance test for differences in standard mortality ratios
- Index(es):
