Re: st: stcox after psmatch2

[Joerg Luedicke <joerg.luedicke@gmail.com>]

Eduardo,

I don't think this advice is very helpful. Why exactly would you want
to leave Stata for this kind of stuff? Getting propensity scores and
creating a desired weight (e.g., an ATT weight) is all very easy and
straightforward in Stata. Moreover, there are lots of options for
survival analyses or other longitudinal methods. If I miss anything
here you need to tell us what the R package you mention can do in
particular that Stata cannot.

George,

Several thoughts:

- For predicting the propensity scores and thus, for the variables you
want to use for matching, it is important to choose pre-treatment
variables. If survival times or whatever differ after enrollment/ no
enrollment then this could possibly regarded as a treatment effect.

- I am not sure what this -pscore- (which btw is user-written and you
are asked to state where you got it from) is doing. But as far as I
understand you, you want to do a 1-1 matching and run some survival
analyses using the resulting sample and so I doubt that you need that.
I would just do:

logit Status age i.gender_coded i.Race_coded i.Hospital_coded
i.Patient_class_coded i.Asthma i.copd i.dm i.ht i.Stroke i.chd i.hf
i.Dyslipidaemia i.Obesity if PrimaryFirst1==1
predict double pscore

in order to obtain propensity scores.

WRT: "I used one matched control per treated, I selected paircount=2
and did stcox regression. Am i doing it the right way?"

Does not look wrong from what I can see here. However, as always, the
devil is in the details (e.g., how did you derive your caliper?) and
there are many options available for what you may want (e.g.,
different matching methods, propensity score weighting (instead of
matching), different kinds of treatment effects). Here is an
introductory textbook on the matter which outlines a lot of the
different flavors you can get, including hands-on examples:

Guo S & Fraser MW (2010): Propensity Score Analysis. Thousand Oaks: SAGE.

HTH,

J.


On Fri, May 11, 2012 at 5:26 AM, Eduardo Nunez <enunezb@gmail.com> wrote:
> George,
> You should look at ipw, an R package that estimate a propensity score
> and use it as inverse probability weighting. The software allows
> longitudinal data with survival endpoint.
> Best wishes,
> Eduardo
>
>
> On Fri, May 11, 2012 at 5:35 AM, George Gunapal Pradeep Paul (NHGHQ)
> <pradeep_paul_g_gunapal@nhg.com.sg> wrote:
>>
>> Thanks Ariel, Steve and Joerg for your suggestions and references. To
>> clarify more,
>>
>> I am comparing a group of Chronic Obstructive Pulmonary Disease
>> (COPD)patients enrolled in a disease management program and a control group
>> selected from a administrative database. There is an enrolment criteria for
>> enrolling patients into the program, but however not all COPD patients are
>> enrolled.
>>
>> I sampled control cases from hospital administrative database, i.e. COPD
>> patients who fulfilled the COPD disease management program's enrolment
>> criteria but where not enrolled into the program. Now that I have program
>> patients and controls, I did matching to adjust for the difference in
>> characteristics on who gets enrolled into the program. For eg. If younger,
>> more healthier patients are more often enrolled in the COPD disease
>> management program, then we can expect them to have better outcomes (lower
>> mortality) than the control group.
>>
>> My hypothesis: No difference is mortality rate (expressed in person years)
>> between the patients enrolled in a COPD program and controls who were not
>> enrolled in the program.
>>
>> Other things to consider, multiple records per patient and the follow time
>> is different between the program patients and controls.
>>
>>
>> Steve & Joerg: Since follow up period is different for these two groups,
>> hence i thought of doing cox regression to adjust for the effect of follow
>> up time on mortality on the matched program patients and controls. Kindly
>> let me know your thoughts.
>>
>> Ariel:
>> I used the following code for generating pscore and matching
>>
>> xi:pscore Status age i.gender_coded i.Race_coded i.Hospital_coded
>> i.Patient_class_coded i.Asthma i.copd i.dm i.ht i.Stroke i.chd i.hf
>> i.Dyslipidaemia i.Obesity if PrimaryFirst1==1, pscore(mypscore)
>> blockid(myblock) comsup
>>
>> xi:psmatch2 Status if PrimaryFirst1==1, pscore(mypscore)
>> outcome(i.Mortality_resp_system) caliper(.001) noreplace neighbor(1)
>>
>> And after matching, I identified pair count using the following code
>> gen pair = _id if _treated==0
>> replace pair = _n1 if _treated==1
>> bysort pair: egen paircount = count(pair)
>>
>> I used one matched control per treated, I selected paircount=2 and did
>> stcox regression. Am i doing it the right way?
>>
>>
>> Thank you for your time and greatly appreciate your help.
>>
>> Best regards,
>> Pradeep
>>
>>
>>
>> -----Original Message-----
>> From: owner-statalist@hsphsun2.harvard.edu
>> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Ariel Linden,
>> DrPH
>> Sent: Thursday, May 10, 2012 11:52 PM
>> To: statalist@hsphsun2.harvard.edu
>> Subject: re: st: stcox after psmatch2
>>
>> I am not sure what problem your facing? -psmatch2 (a user written program
>> found on ssc) will gives you a field called _nn that tells you who the
>> matched control is. BTW, you didn't give details of which matching approach
>> you used or how many controls. If you used more than one matched control per
>> treated, you'll need the _weight variable as well.
>>
>> You then run your stcox model on only those patients and their matched
>> controls (1:1), and if you have 1:k matches, you also include the _weight in
>> the model...
>>
>> I hope this helps
>>
>> Ariel
>>
>> Date: Thu, 10 May 2012 05:33:22 +0000
>> From: "George Gunapal Pradeep Paul (NHGHQ)"
>> <pradeep_paul_g_gunapal@nhg.com.sg>
>> Subject: st: stcox after psmatch2
>>
>> Dear all,
>> I am comparing a group of Chronic Obstructive Pulmonary Disease
>> (COPD)patients enrolled in a disease management program and a control group.
>> I calculated pscore and did psmatch2 using c:\ado\plus\p\psmatch2.ado *!
>> version 4.0.5 18apr2012 E. Leuven, B. Sianesi. I used psmatch2 to adjust
>> for their baseline differences and now I would like to do a cox regression
>> on the matched program patients and controls to see if there is a difference
>> in time to death between the two groups.
>>
>> I would like to know how to run stcox after psmatch2 , thereby identifying
>> the matching sequence for the cox regression. I tried looking for answers
>> for this but found only one unanswered relevant question which could be
>> found at http://www.stata.com/statalist/archive/2009-11/msg00709.html
>>
>> Kindly help.
>>
>> Thank you,
>>
>> Best regards,
>> Pradeep
>>
>>
>> *
>> *   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/
>>
>>
>> ---------------------------------------------------------------------------------
>> The information contained in this e-mail and the attachments (if any) may
>> be privileged and confidential and is intended solely for the named
>> addressee.  If you are not the intended recipient, please do not print,
>> retain copy, disseminate, distribute, or use this e-mail or any part
>> thereof.  Please notify the sender immediately by replying to this e-mail
>> and delete all copies of this e-mail and the attachments.
>>
>> *
>> *   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/

*
*   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/