Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: comparing rates

From   Michel Camus <>
Subject   Re: st: comparing rates
Date   Mon, 08 Sep 2003 15:22:39 -0400

Matching design and variable (patient) seems confused with clustering factor (physician)... (see below)
Chris Rohlfs wrote:

don't worry about controlling for practice.  you've already controlled for
it with the doctor id (i.e., they're perfectly multicollinear).  so just
run:	clogit referred, group(doctorid)
and any possible confounding effects of "practice" will be captured & controlled for by doctorid.

On Mon, 8 Sep 2003, Ricardo Ovaldia wrote:

Can I simply take the number of referrals at one
practice and divide by the number of patients seen at
that practice (referral rate for practice 1) and
compare it to the rate of the other practice without
taking in account that some physicians within practice
may refer more than others? Or, should I compute the
referral rate for each physician, and then combine
these rates within practice by weighting them some
how? If so how can I do this in Stata? what kind of
weights should I use?
Ricardo. --- Ricardo Ovaldia <> wrote:

We are interested in comparing the referral rates
for two physician practices. The fist group has 20
physicians and the second group has 22 physicians.
Each physician saw a variable number of similar
patients some saw as few as 4 and others as many as
43. The physician either referred the patient to a
specialist or did not. If I compute the referral rate for each physician
group and compare them, I concerned because the
observations are not independent. (...) I though that I could use clogit- to group by
doctor ID: . clogit referred practice, group(doctorid)
This did not work because the practice-variable does
not vary within doctor ID, so it is dropped from the
model (omitted due to no within-group variance).
Any suggestions will be greatly appreciated.
Ricardo Ovaldia, MS, Statistician, Oklahoma City, OK

Study question: do referral rates (possibly patterns) differ between practices (evidently composed of different physicians...)?
Study design? ...panel or cohort study of 2 independent groups of physicians.
The relation between pratice (explanatory variable) and referral (outcome variable) is not matched/conditioned on physician since you do not have the same physicians in both practices. Therefore physician is not a matching variable. However, you may have matched/conditioned on (? Ricardo said "similar"...) patients to study the "practice effect" conditionally on (independently of) patient characteristics.
A) If so, you can condition on "patient", not on physician... (This will exclude a patient seen in one practice but not seen in the other.)
. clogit referred practice, group(patientid)

B) If not matched on patients, use ordinary unconditional -logit- or -logistic- but with the cluster option for physician.
The following should correct for the lack of independence of referrals by a same physician.

. logit referred practice, cluster(doctorid)

I tested this with a simulated dataset of 18 observations, which I attach. I hope this helps.
(Please, correct me if I am wrong.)

Michel Camus, PhD, epidemiologist
Health Canada

Attachment: ricardo ovaldia 8-09-2003.dta
Description: Binary data

© Copyright 1996–2021 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index