Announcement

Collapse
No announcement yet.

Reliability and validation of motion capture system to measure spinal mobility

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Reliability and validation of motion capture system to measure spinal mobility

    Dear friends,

    At the last ISB2011, I presented an oral comunication titled "MEASURING SPINAL MOBILITY IN ANKYLOSING SPONDYLITIS: COMPARISON OF MOTION CAPTURE RESULTS AGAINST CONVENTIONAL METROLOGY AND RADIOLOGICAL SCORES."



    In this communication I want to evaluate the use of our motion capture system to study the reduction of mobility in ankylosing spondylitis patients in our hospital.

    First we want to analyze reliability calculating ICC between measurements, so we make three trials. Two in the same day (five minutes between tests) and another one two weeks after. The results of ICC in measurements obtained by motion capture were high (>0.90).

    In order to validate we compare these results with the obtained by measures usually used by rheumatologist (floor to finger distance, cervical rotation, ...) and correlation were also high (>0.75).

    Finally we correlated with radiological scores (mSASSS) and results were also very good (>0.84), higher than conventional metrology results.

    So we think that our system is valid and reliable to be used to assess spinal mobility in AS.

    But some collegues during discussion of the communication said that this is not the best way to demonstrate it. Unfortunatly I didn't talk with these collegue after my presentation to discuss it, because I talked with another about other topics, but I am very interested about this.

    What's your oppinion ?

    What would be the correct method to demonstrate it ?

    What journal will be the addecuate to publish these results ?

    Thanks in advance.

    Juan Luis Garrido-Castro.
    Reina Sofia Hospital Motion Laboratory
    University of Cordoba
    SPAIN

  • #2
    Re: Reliability and validation of motion capture system to measure spinal mobility

    Juan Luis Garrido-Castro

    ICC is one aspect of reliability providing a measure of relative reliability (in simple terms it can be viewed as the ratio of Between – Within subject variance to Between + Within subject variance or the proportion of Between subject variance), and with repeated measurements with one examiner indicates the ability of the measure to discriminate between subjects. However it does not give a direct measurement of either within or between subject variance, only a ratio. To measure reliability you also need a direct measure of within subject reliability such as Standard Deviation, Coefficient of Repeatability, or Standard Error of Measurement. These three are all essentially measuring the same thing. It is also worth while plotting the differences in repeated measures against the mean to see if the reliability is related to the magnitude of the mean, the Coefficient of Repeatability being 1.96 times the standard deviation of the differences can also be obtained (Bland and Altman, The Lancet, 1986).

    ICC will vary greatly depending on the homogeneity of the subjects and the nature of the measurement (ie raw score vs normalized), as with correlation. ICC’s should also be presented with a 95% confidence interval. Generally I would consider 20 subjects as the minimum required for estimations of ICC, as you have 33 subjects and high ICC values the confidence interval should be relatively small and not a major issue. For a method for calculating confidence intervals or sample size requirements for an estimated ICC see Bonett (Statistics in Medicine, 2002).

    For validity (assuming your criterion measure is indeed valid) you need to test agreement between the two measures, a simple paired T-test will do. Correlation indicates that the measures are linearly related but does not mean that the two measures are the same. Again the difference between instruments can be plotted against the respective means to see if the agreement is related to the magnitude of the mean, the bias and Limits of Agreement being 1.96 times the standard deviation of the differences can also be obtained (Bland and Altman, The Lancet, 1986).


    Allan Carman
    School of Physiotherapy
    University of Otago
    New Zealand
    allan.carman@otago.ac.nz

    Comment


    • #3
      Re: Reliability and validation of motion capture system to measure spinal mobility

      I agree with the previous post that you must establish both relative reliability (ICCs) and absolute reliability (SEM etc.). I am doing a reliability study as part of my PhD work, and so have spent a lot of time understanding the concept. I have found "Foundations of Clinical Reserach, Applications to Practice" by Portney & Watkins an invaluable resource. You might want to consult it for help.
      Good luck!
      Joanne Parsons
      Faculty of Kinesiology
      University of Manitoba
      Winnipeg, Canada
      umparso3@cc.umanitoba.ca

      Comment

      Working...
      X