Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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


From   "Carlo Lazzaro" <>
To   <>
Date   Fri, 3 Jun 2011 13:31:08 +0200

Cornelius may want to perform a logistic regression in -logit-
mode(dependent variable: changed status to CA):

set obs 100
g First_Test="Positive" in 1/70
replace  First_Test="Negative" in 71/100
g Second_Test="Positive" in 1/80
replace   Second_Test="Negative" in 81/100
g changed status to CA =1 if  First_Test=="Negative" ///
&  Second_Test=="Positive"
replace  changed status to CA =0 if  changed status to CA ==.
Logit changed status to CA 1st_ind_var 2nd_ind_var 3rd_ind_var

Kindest Regards,
-----Messaggio originale-----
[] Per conto di Cornelius Nattey
Inviato: venerdì 3 giugno 2011 11.15

Dear All,
I am involve in a study "Longitudinal variability of skin prick test results
to six common aeroallergens (CA) and three soybean allergens" 

The dependant variable is "changed status to CA" YES or NO. I.e. started off
negative and became positive at some point in the survey to at least one CA
or started off positive and became negative at some point in the survey to
at least one CA = changes status YES; and if stayed the same throughout( ie
negative  negative or positive positive)   

independent variables are: (1) the number of positive SPTs to CA at the
initial test; (2) the size of the largest wheal at the initial test; and (3)
the number of testings done.

What type of analysis can one do with this type of dataset?

Below are some considerations so please advice on what to do with the

The dataset with 110 people with 2 testings. At the initial testing we have
+ve or -ve. At the second testing we have remained  +ve, remained  -ve,
converted (-ve to +ve) or reverted.   (1) We will examine determinants of
conversion. But only people who were initially negative can convert. Does
this mean that we drop from this analysis the subjects who were initially
positive (they have no chance of getting the outcome conversion)?
 I think we do not have to drop them as they still have the range of
measurements across the independent variables and so can be used as a
reference group for these variables. But I ask this question because for
example in a study of the determinants of prostate cancer  you would not
include subjects who have had a prostatectomy (for non-cancer reasons). (2)
Same consideration for the analysis of reversion.   

The dataset with 4 testings does not have the same problem because
conversion and reversion can occur along the time course of the study
irrespective of initial status, and so all subjects are at risk of the
outcome at some point (e.g. initially positive can revert and then convert
so even if you are initially positive you have a risk of conversion at some
point). I think.

Cornelius Nattey

Medical Scientist: Epidemiology and Surveillance
National Health Laboratory Service
National Institute for Occupational Health
Office: 011 712 6438
Fax: 086 604 1214
Cell: 079 631 5857

The views expressed in this email are, unless otherwise stated, those of the
author and not those of the National Health Laboratory Services or its
management. The information in this e-mail is confidential and is intended
solely for the addressee.
Access to this e-mail by anyone else is unauthorized. If you are not the
intended recipient, any disclosure, copying, distribution or any action
taken or omitted in reliance on this, is prohibited and may be unlawful.
Whilst all reasonable steps are taken to ensure the accuracy and integrity
of information and data transmitted electronically and to preserve the
confidentiality thereof, no liability or responsibility whatsoever is
accepted if information or data is, for whatever reason, corrupted or does
not reach its intended destination.

*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index