Re: st: Re: compound outcome variables (not specifically Stata-related)

[Diego Navarro <the.electric.me@gmail.com>]

I recently finished a project for a private consulting shop where I  
used principal component analysis to compress a number of  
socioeconomic outcome measurements (GDP per capita, unemployment,  
diverse measures of education, Gini and Theil indexes, etc.) into a  
single State Development Index that was computed for states in Brazil.  
(It's like states in the USA). We're actually getting it published  
into the mainstream press!

Anyway, that's a datapoint for people treating multiple outcome issues  
with Stata.

--
Diego Navarro
(21) 2559-5620

"Writing is not really about meaning, but about land-surveying and  
cartography, particularly of lands yet to come" (Mille Plateaux #1.11)

“The first step is to measure what can be easily measured.  This is  
okay as far as it goes.  The second step is to disregard that which  
cannot be measured, or give it an arbitrary quantitative value.  This  
is artificial and misleading.  The third step is to presume that what  
cannot be measured really is not very important.  This is blindness.   
The fourth step is to say that what cannot be measured does not really  
exist.  This is suicide.” (Daniel Yankelovich, 1973)




On 24/01/2009, at 02:03, Joseph Coveney wrote:

> A few years ago, Prof. Gary G. Koch (University of North Carolina at  
> Chapel
> Hill) gave a talk here in Tokyo on his researches into the statistical
> handling of "multiple endpoints" (the term he used for what you  
> describe) in
> clinical trials.  One of his articles might be a good entry point  
> into the
> literature on the matter.  He gave his talk under the auspices of  
> the Drug
> Information Association, an organization that is largely focused on  
> industrial
> product-development activities (viz., pharmaceuticals and  
> biologics), but the
> approaches he described can be applied more generally, e.g.,  
> evaluating
> benefits of medical or surgical treatments, or life-style changes.   
> Tony
> Lachenbruch already mentioned the positions that the U.S. Food and  
> Drug
> Administration takes for so-called co-primary outcomes and multiple
> primary outcomes in the context of granting or denying sales  
> authorizations
> for medical products.  Professor Koch's talk covered those, but also  
> broached
> the topic of using the endpoints in a compound, or joint, manner (as  
> if
> combining them into a single index of treatment response), and  
> addressed some
> of the various approaches of doing that and their associated  
> problems.  The
> U.S. Food and Drug Administration allows these latter approaches in  
> selected
> cases, for example, I believe that there is such a compounded index  
> for
> rheumatoid arthritis that is widely and frequently used.
>
> Joseph Coveney
>
> Christopher W. Ryan wrote:
>
> > I would like to try to do a journal club session for my family  
> > practice
> > residents on the uses, and especailly the pitfalls, of compound  
> > outcome
> > variables, like "stroke or myocardial infarction or coronary bypass
> > surgery." They appear frequently in clinical trial literature.  Can
> > anyone recommend any good articles discussing the issues?  Thanks.
>
>
> *
> *   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/