...
First, the incidence is not events/population but events/person-years.
Typically with population counts of an event, the population is not
under direct observation and so the total person-time for everyone in
the cohort is unknown.
An estimate of the total person-years can however be obtained by using
the population size in the middle of the period of observation.
For example, if the total number of events for males aged 0-4 years are
counted in a calendar year, then the mid-year population size in that
year is an estimate of the total person-years accumulated by males aged
0-4 years.
You can analyse this with Poisson regression,
xi: poisson events i.sex i.age, exp(person_years)
using -testparm- after this will allow you to test the overall
association between age and the incidence rate.
If you want to test the trend with age, then you could try:
xi: poisson events i.sex age, exp(person_years)
The p-value for age is a test of the linear trend. Note: this may not
have much power if the trend is non-linear.
You might also consider,
xi: poisson events i.sex*i.age, exp(person_years)
and test the significance of the interaction terms using -testparm-.
This would tell you if the age association is different for males and
females.
For each model, you should also check that the residuals are Poisson
using -estat gof- or, alternatively, by refitting each model using
-nbreg- and comparing the results with the -poisson- results.
______________________________________________
Kieran McCaul MPH PhD
WA Centre for Health & Ageing (M573)
University of Western Australia
Level 6, Ainslie House
48 Murray St
Perth 6000
Phone: (08) 9224-2701
Fax: (08) 9224 8009
email: [email protected]
http://myprofile.cos.com/mccaul
http://www.researcherid.com/rid/B-8751-2008
______________________________________________
If you live to be one hundred, you've got it made.
Very few people die past that age - George Burns
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of David Winter
Sent: Friday, 28 August 2009 1:04 AM
To: [email protected]
Subject: st: Heterogeneity and trend in proportions (Chi-square)
Dear Colleagues,
I am an IT Manager with limited statistical experience trying to help a
Masters student with some statistical testing. We currently use Stata SE
version 10.
The data relates to the occurrence of meningitis in a specific
geographic area of the UK. I have the relevant population statistics
for the geographic area to calculate incidence. I would like to be able
to run a Chi-square test for heterogeneity and trend in proportions in
various factors. The following is an example of the data.
Num. Events Incidence (Events/Population)
=========== =============================
Factor
Sex:
Male 1210 17.31
Female 986 14.79
Age(yrs):
0-4 1474 46.20
5-9 271 7.89
10-14 191 5.26
15-19 260 7.65
In the above example, I should like to run the tests for heterogeneity
and trend (where applicable) on Sex and then Age.
What would be the best way to proceed please?
Dave Winter
School of Health & Population Sciences
University of Birmingham
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
