Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • I2 using metaprop states "overall (I^2=0.00%, p= . )" NB: no I2 and no p-value for I2

    Hi. I am using metaprop to calculate prevalence of infections. The command I am using is

    metaprop pos1 total1 if sti=="SYPHILIS", ftt nobox sortby(Year) lcols(Study Year) nowt power(2) xlabel(0,5,10,15,20) texts(200)

    pos1=numerator (number of people positive)
    total1=denominator (number of people tested)

    why is the I2 zero and no p-value is being generated?

    help. Thanks
    Attached Files

  • #2
    The -metaprop- command is a user-contributed command as you are asked to explain. Please do read the FAQ, in particular section 12: What should I say about the commands and data I use?

    As you do not provide data, we can only make a guess. My guess is this. The metaprop command uses, by default, a fixed-effect model which means that there is no assumed residual heterogeneity. By definition then, I^2 will always be 0%, and there is no chi-squared statistic associated with the additional heterogeneity from which to compute a p-value. (Note, the p-value is associated with this chi-squared statistic, no the I^2 value. I find the common placement of a p-value adjacent to I^2 in many forest plots a poor choice to communicate this.)

    Comment


    • #3
      Thanks. The stata file in case anyone wants to look into this is here at the below link

      https://unimelbcloud-my.sharepoint.c...gB2Ew?e=hXjEb5


      The issue only happens for analysis by HIV and SYPHILIS (codes below)

      metaprop pos1 total1 if sti=="SYPHILIS", ftt nobox sortby(Year) lcols(Study Year) nowt power(2) xlabel(0,5,10,15,20) texts(200)
      metaprop pos1 total1 if sti=="HIV", ftt nobox sortby(Year) lcols(Study Year) nowt power(2) xlabel(0,5,10,15,20) texts(200)

      Thanks

      Comment

      Working...
      X