What I meant to say is that MEDLINE/PubMED allows you to combine disease of interest (P) with intervention of interest (I) but not with C-omparator or O-utcome or T-ime period of interest...
-----Original Message-----
From: Djulbegovic, Benjamin
Sent: Tuesday, January 19, 2010 4:00 PM
To: 'Kevork Hopayian'; [log in to unmask]
Subject: RE: Survey: framing questions on diagnosis
Kev, although I am aware of some advances in the field, for the most part the search is at mercy of coding in Medline/PubMed and other relevant databases. This leads to discrepancy between what clinicians/users want (PICO or PECOT) vs. how actually entries were coded ("tagged") in databases of interest. For this reason, I rarely found possible to use PICO/PECOT. A typical strategy consists of combining key words for disease (MeSH) (which broadly can relate to P, I guess) + filter for the study design (e.g. RCT).
There are number of information scientists/librarian on the list, and I would really appreciate some additional thoughts related to development of search filters using PICO/PECOT format. I am aware of some filters, but it would be good to learn how good they are. More specifically, can any of these (PICO) filters replace Clinical Queries etc.
Thanks
ben
-----Original Message-----
From: Evidence based health (EBH) [mailto:[log in to unmask]] On Behalf Of Kevork Hopayian
Sent: Tuesday, January 19, 2010 3:11 PM
To: [log in to unmask]
Subject: Re: Survey: framing questions on diagnosis
Hello,
Last year I asked you to take part in a survey on what structure would you use for a
focused question on diagnosis, using PIO PICO or PECOT. The aim was to find out people
use for the P, I/E C and O. There were 16 respondents (thank you all). The discussion was
continued at the 5th International EBHCTD conference in Sicily in Oct/Nov 2009. I said
that I would feed back the results.
There was almost virtual agreement about P = population with the symptom, but there
was a tremendous variation in the other parts of the question..
The most favoured combination was this:
I/E = the test
C = the reference standard
O = presence of disease
Also popular was a slight variation on this:
I/E = test result positive
C = the reference standard
O = presence of disease
There were powerful, even though minority, arguments against these variants. Many
people commented that this is a difficult area (and some that PICO is better forgotten for
diagnostic questions altoghether!). I can make these arguments available (eventually) to
those who are interested. Please email me if you want more information as it becomes
available.
FYI, the scenario presented for discussion was:
Sciatica is a musculoskeletal symptom in which pain is felt along the length of the leg. A
test, known as the straight leg raise (SLR), puts the sciatic nerve under tension. A
positive SLR is taken to suggest one cause of sciatica: intervertebral disc herniation or
disc rupture.
This would translate in the popular variants as:
P = people with sciatica
I/E = SLR OR SLR positive
C = MRI showing disc herniation
O = disc herniation
Best wishes,
Kev Hopayian
GP Suffolk and GP Tutor, School of Medicine, Health Policy and Practice, University of
East Anglia, Norwich, England
