[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: RE: pooling incidence rates

From   "Lachenbruch, Peter" <[email protected]>
To   "[email protected]" <[email protected]>
Subject   st: RE: pooling incidence rates
Date   Sat, 23 Jan 2010 13:58:38 -0800

An alternative is to use each study as a block and input the number of cases and the study size.  Then a logistic regression with study as a factor would be a simple approach.  Or you could use a Mantel-Haenszel analysis.


From: [email protected] [[email protected]] On Behalf Of Sripal Kumar [[email protected]]
Sent: Saturday, January 23, 2010 10:07 AM
To: statalist
Subject: st: pooling incidence rates


I am trying to estimate the incidence of a particular adverse effect
from RCTs published so far.  The way I have pooled the incidence rate
is to weigh each trial by the inverse of the variance just like a
standard meta analysis.  My question is what do I do if the studies
have different lengths of follow up?  Should I calculate the incidence
rate for each study per person-year observation and then pool them by
weighting each study by the inverse of their variance? For clinical
interpretation and to compare this against published data, I need to
report the data as number per 100 patients exposed rather than number
per 100 patient-years of observation.

thanks in advance,
*   For searches and help try:
*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index