The GRADE working group has recently adopted guidance for assessing the risk of bias
associated with missing participant outcome data across an entire body of
evidence. This post summarizes the guidance, provides a background, relevant
references, and links to excel documents with calculators to facilitate the use
of the guidance.
The Guidance
·
Start with a primary meta-analysis using a
complete case analysis (i.e., excluding those with missing data)
·
Then conduct sensitivity meta-analyses imputing,
in each study, data for those with missing data, and then pooling across
studies.
o
For binary outcomes: we suggest use of
“plausible worst case” in which review authors assume that those with missing
data in treatment arms have proportionally higher event rates than those
followed successfully.
o
For continuous outcomes: use imputed mean values
from other studies within the systematic review, and the standard deviation
from the median standard deviations of the control arms of all studies. For meta-analyses in which investigators have
used different instruments to address the same construct, convert all scores to
standardized mean difference (SMD) or to the units of a selected reference
instrument.
·
If the results of the primary meta-analysis are
robust to the most extreme plausible assumptions, do not rate down quality of
evidence for risk of bias due to missing participant outcome data.
·
If the results of the primary meta-analysis are not
robust to the most extreme plausible assumptions, consider rating down quality
for risk of bias.
References
Links to MPD
assumptions calculators
Binary variables:
A freely downloadable Excel document to determine the numerators and
denominators to be used for each trial included in the metaanalysis according
to the selected assumption: https://www.dropbox.com/s/opstwgm45qiq57k/Assumptions%20about%20MPD%20v5.xls?dl=0
Continuous variables:
A freely downloadable Excel document to determine the means and SDs to be used
for each trial included in the meta-analysis according to the selected
assumption per strategy:
