There have been a lot of interesting comments in this thread. Let me just
add my two cents.
A lot of the doctors and nurses that I work with get hung up over the actual
statistical analysis in medical journal articles, and they get intimidated
by the statistical jargon. Why anyone who uses words like
"hyperbilirubinemia" has the right to complain about jargon is an
interesting question, but that's a different story.
Anyway, what I tell them is that nine times out of ten, the mistake was not
in how the data was analyzed, but in how it was collected. After all, if you
collect the wrong data, it doesn't matter how sophisticated the analysis is,
does it?
Furthermore, you don't need to be an expert in statistics to critically
assess the process of data collection.
There's an interesting review article that appeared in the British Medical
Journal (BMJ). I'm writing up a synopsis of this for my web pages, but the
bottom line is that research in schizophrenia (like research in a lot of
areas) is poorly conducted.
It's hard to good research, especially in an area like schizophrenia. But we
need to do better. I think one of the most telling statistics is that these
2000 studies used 640 different outcome measures and that 369 of these
outcome measures were used only once. Meta-analysis is hard enough, but when
everybody invents their own way of measuring things, it is sheer chaos. You
might say that research about schizophrenia is schizophrenic.
Here's a draft of what I have written.
Thornley, Ben, and Adams, Clive "Content and quality of 2000 controlled
trials in schizophrenia over 50 years" British Medical Journal 1998; 317:
1181-1184.
Overview of research studies
- Studies published between 1948 and 1997.
- Patients with schizophrenia and other non-affective psychoses.
Variety of interventions
- Drugs (e.g., anti-psychotics and anti-depressants)
- Therapy (e.g., individual, group, and family)
- Miscellaneous (e.g., electroconvulsive treatments)
Four difficulties
1. Types of patients
- The ideal study would be community based.
- Only 14% of actual studies were community based.
2. Number of patients
- The ideal study should include at least 300 patients.
- The average number was only 65 patients.
- Only 3% of studies met the target of 300 or more patients.
3. Length of the studies
- The ideal study should last at least six months.
- More than half of the studies lasted six weeks or less.
- Only 19% of the studies met the target of six months or more duration.
4. Measurement
- The ideal studies should concentrate on a small number of standard
measures.
- These 2000 studies employed 640 different measures.
- There were 369 measures that were used once and never used again.
Conclusions
Much of the work in schizophrenia failed to meet appropriate research
standards. Too many of the studies...
- examined the wrong patients,
- studied too few patients,
- ended too soon,
- used fragmentary measurements.
Research in schizophrenia leaves much room for improvement.
Steve Simon, [EMAIL PROTECTED], Standard Disclaimer.
STATS - Steve's Attempt to Teach Statistics: http://www.cmh.edu/stats
<http://www.cmh.edu/stats>
P.S. You can find the full text of this BMJ article on the web. Look at:
http://bmj.com/cgi/content/full/317/7167/1181
<http://bmj.com/cgi/content/full/317/7167/1181>
===========================================================================
This list is open to everyone. Occasionally, less thoughtful
people send inappropriate messages. Please DO NOT COMPLAIN TO
THE POSTMASTER about these messages because the postmaster has no
way of controlling them, and excessive complaints will result in
termination of the list.
For information about this list, including information about the
problem of inappropriate messages and information about how to
unsubscribe, please see the web page at
http://jse.stat.ncsu.edu/
===========================================================================
