Helo Robin,

ICC values can easily be inflated if the variability in the subjects is
great, yielding a false idea of "good" correlation.
But what is reliability anyway? If it is just repeatability then ICC will
tell you a bunch of informaiton. But reliability has to do with how free of
random or non systematic error a measurement can be. IN That sense ICC gives
no informaiton on the subject. Other statistics must be used to assess and
evaluate reliability through that angle. If you email me privately I will be
happy to quote a few studies that make this very plain.

I am sorry I am not familiar with the study you mentioned. WHich data bases
are you using ? Have you found the study full text? is it a recent study?

Will
> Hi will,
> I am currently studying the reliability of a two dimensional video analysis
> tool, and have searched the literature to get an idea of the relative
> reliability of the current measurement tools. I was wondering if you
> were aware of the paper by Rome K and Cowieson F that determined the
> sd of these during ankle ROM testing. I am also intrigues by your
> statement that -"ICC ( intraclass correlation coefficients are not
> enough to fully describe the more exhaustive picture of reliability
> of a process/instrument." Have you any references for this? Thanks Robin
>
> >-----Original Message-----
> >From: - for physiotherapists in education and practice
> >[mailto:[log in to unmask]] On Behalf Of Will Remigio LLU
> >Sent: 07 October 2003 18:58
> >To: [log in to unmask]
> >Subject: Re: Is our goniometry very imprecise?
> >
> >
> >HIi Patrick,
> >
> >Thank you for your insightfull comments.
> >Reliability studys measure intra rater reliability and ours do
> >to. However, I
> >am interested in the application of these in the clinical setting. For
> >instance. Reliability studies have been mostly done with
> >normals. But we
> >apply it to non normal population ( poor inferences!)
> >Intra Rater is good to determine reliability itself but in
> >clinical setting
> >we don treat patients intra rater. The patient has one
> >therapist that takes
> >care and measures him/her.
> >In reliability studies you want to measure things in a short
> >time i.e. in one
> >session, to avoid the effects of improvement with time. But in
> >real life you
> >measure it along a period of treatment.
> >How do we make this bridge? Our study is exploring this. We
> >were not able to
> >cluster patients per pathology or diagnosis so others studies
> >might be able
> >to refine on this. Yes other studies could measure the same
> >joint at the same
> >time, with two instruements and compare. It would be
> >interesting to see how
> >the fluid goniometer compares it done simultaneously.
> >Yes, if the standardization process is not rigorously followed
> >even a gravity
> >based reading can set you off and coufound.
> >ICC ( intraclass correlation coefficients are not enough to
> >fully describe
> >the more exhaustive picture of reliability of a process/instrument.
> >
> >We still want realiability studies. But these studies need to
> >come closer to
> >the clinical reality. They are good and necessary inasmuch as
> >they can also
> >inform clinical decision making and bridge the gulf between
> >statistical and
> >the clinical vocabularies.
> >
> >Basically, I am suggesting that PT's don't trust taking just
> >one measurement
> >and reporting that as though this was the ROM on kneed flexion
> >for a given
> >patient but improve on this. I am recommending that an average of 3
> >measurements be taken.  I would like to know what Pt's out
> >there would view
> >these recommendation.
> >
> >Thanks for the inputs,
> >
> >Will
> >
> >
> >> Wilton;
> >>
> >> Did I read it correctly......  You're looking for change across time,
> >>  not within the same session?  Wouldn't you want to check for intra-
> >> rater reliability first?
> >>
> >> In effect looking for change across time is really looking at
> >> effectiveness of treatment if in case rom was a goal.
> >>
> >> I would also suggest trying to narrow the subjects to atleast the
> >> same type of surgery, for example, TKA, ACL recon, or menisectomy.
> >>
> >> You may want to even test using goniometer vs. fluid gonio.
> >> Although it is more accurate I think there are some inherent
> >> problems if the piece of equipment doesn't measure change or if it
> >> is "zeroed" at a position the joint is in and then that joint's
> >> position relatively changes.
> >>
> >> YOur thoughts?
> >> Patrick Zerr
> >> www.apluspt.com
> >> The easiest way to prepare for the National PT Exam!
> >> www.summitpt.com
> >> Summit Physical Therapy; Tempe, Arizona
> >>   ----- Original Message -----
> >>   From: Wilton
> >>   To: [log in to unmask]
> >>   Sent: Tuesday, October 07, 2003 9:07 AM
> >>   Subject: Is our goniometry very imprecise?
> >>
> >>   Helo,
> >>
> >>   I would like to get your feedback on the reliability of
> >> measurements PTs take   for knee range of motion. I have conducted a
> >> study measuring the reliability of knee ROM u
> >>
> >>   Using the fluid goniometer. Regardless of which type of goniometer
> >> I found that
> >>
> >>   none of the  studies done in the 80's or 90-'s
> >>   report measurement error and most were done in normals. When I did
> >> my   study on patients and looked at the variability you get it
> >> struck me that   taking just one measurement for knee flexion in a
> >> clinical setting cannot give you a     value that could be very far
> >> from the true value of that patient
> >>
> >>   because of the amount of error present in the measurement process
> >>
> >>   and that subsequent different measurements you get, may not mean
> >> there was improvement   but may be part of the normal fluctuation
> >> you get from measuring  sick joints. Therefore
> >>
> >>   I am recommending that PT's take 3
> >>   measurements take a Mean and try to get a Std Dev of their
> >> patients so as to come closer to the
> >>
> >>   True value that joint has. .
> >>
> >>   For instance. Let's say  I measure a knee joint of a patient just
> >> once (first visit) and get, say   100degres of flexion, treat the
> >> patient  and then measure it again a couple of   weeks later, and
> >> get 115degrees. Did change occur? According to our study it
> >>  didn't! ( you may be still be measuring just the regular
> >> fluctuation or typical error   variability of the patient.)
> >>
> >>   I used the fluid goniometer for being a more precise instrument
> >> than the UG. Because it has
> >>
> >>   Less to vary on since it is gravity based.
> >>   I found a huge variability and Measurement error when studying
> >> sick knee   joints. If we want our measurements to be able to base
> >> true change we might   need to revisit the way we measure knee ROM
> >> with goniometers.
> >>
> >>   May I have your input or comment, disagreements, or impressions?
> >>
> >>   Thanks
> >>
> >>   WIlton Remigio
> >>   PT from Brazil doing post Grad work in the US
> >