Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


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

Re: st: analysis of cluster of fungal infection in an ICU-unit


From   Austin Nichols <austinnichols@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: analysis of cluster of fungal infection in an ICU-unit
Date   Wed, 28 Dec 2011 08:25:51 -0500

roland andersson <rolandersson@gmail.com>:
Why not instead construct a discrete-time competing risks model?  Just
use mlogit and regress on time since last infection of each type, and
time in ICU (capturing baseline hazard).  If a given type has a large
positive impact on the hazard of infection of that type, and other
type has no impact, you have evidence of clustering, and that type of
model can admit covariates that can constitute alternative
explanations, i.e. you may be able to falsify the model with another
kind of explanation than simple clustering in time.  In this case,
individual patients at each point in time (each day perhaps?) would be
the units of observation and no infection would not really be an
outcome type since the patient would remain in the risk pool; leaving
the ICU would be a censoring of data, i.e. the data ends when either
an infection is observed or the patient leaves the ICU.  You can also
run a bunch of separate logits for each type of risk regressed on time
since last infection of that type, but the mlogit allows some kinds of
infection to alter the hazard for some other type.  My concern about
alternative types of explanation is that there are always unmodeled
pathways; if the transmission cannot be affected by the building or
equipment etc. still perhaps the susceptibility to infection can be
(that is, the fungus is able to get a foothold when a patients immune
system is depressed by a chemical on the equipment).

On Thu, Dec 22, 2011 at 7:09 PM, roland andersson
<rolandersson@gmail.com> wrote:
> Austin
> Thank you for this nice example. Your response helps me in my own
> process. I also leran more about Stata.
>
> Your example only takes the order into consideration, whereas we are
> interested in the distance in time between the infected patients, ie
> during a weeks interval there may have been many non-infected patients
> as well as patients with many different fungal clones. We want to have
> a "moving window" in time were we can compare the clones of all the
> patients that were in the ICU unit within that window and identify the
> number of patients with identical clones among all patients with
> infection of all variable clones. If such patients are more common
> than by chance it may indicate a transmission.
>
> My plan is to create a dataset of all possible pairs of patients. I
> create a variable sameclone that identify pairs infected with same
> clone. I create dichotome variables that define a timeperiod that is
> close in time (2,3,4 days and so in) and tabulate sameclone against
> closeintime. From the margins I can calculate the expected number of
> sameclone and closeintime pairs and compare the expected with the
> observed. This will show if there is clustering in time.
> What do you think?
>
> About fungal infection. Many of us carry some fungal spores at times
> on our bodies (mostly candida). There are many different clones of
> these fungus species. So if many patients that are at the same time in
> an ICU unit are found to have the same clone we suspect that a
> transmission may have occurred. We need to find out if this is only
> occurring only by chance. The human fungus do not come from the
> building.
>
> Greetings and Merry Christmas
> Roland
>
>
>
> 2011/12/22 Austin Nichols <austinnichols@gmail.com>:
>> roland andersson <rolandersson@gmail.com>:
>>
>> I meant that if you just want a test of whether a given type of
>> infection is more likely after the same type, which you have already
>> said you observed in a graph, you could run a simple mlogit.  No
>> infection could also be a category modeled, and you could include all
>> the negative results.
>>
>> For example, here is a case where the null is true (no clustering
>> implied by the DGP):
>>
>> clear
>> range id 1 1000 1000
>> g type=ceil(uniform()*6)
>> tsset id
>> g lasttype=l.type
>> mlogit type i.lasttype
>>
>> A more sensible analysis might use duration with exact times of tests
>> and entry into into the ICU, as opposed to simple order of test for
>> infection, and try to isolate the actual mechanism causing the
>> observed clustering, perhaps using a competing risks analysis on time
>> to infection (with leaving the ICU being a censoring event). A good
>> model should incorporate a deep understanding of the science and
>> setting, which I do not have for fungal infections in an ICU.  I would
>> suspect ceiling tiles before staff, for example, but you clearly have
>> a reason for suspecting the staff of transmitting the infections.
>>
>> On Wed, Dec 21, 2011 at 5:50 PM, roland andersson
>> <rolandersson@gmail.com> wrote:
>>> Austin
>> <snip>
>>> I do not understand what you mean by  "You could just run an -mlogit-
>>> of type on last type"?

*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index