Dealing with missing data
If data are missing from reports, we will make attempts to contact
the study authors to obtain missing information. If this is not
successful, we will assume that the data are missing ’at random’ or
’not at random’.
Data are said to be ‘missing at random’ if the fact that they are
missing is unrelated to actual values of the missing data. For instance,
if some quality-of-life questionnaires were lost in the postal
system, this would be unlikely to be related to the quality of life of
the trial participants who completed the forms. Data are said to be
‘not missing at random’ if the fact that they are missing is related
to the actual missing data. For instance, in a trial if participants do
not attend the final follow-up interview because they have developed
a pressure ulcer or are in pain, they are more likely to have
missing outcome data. Such data are ‘non-ignorable’ in the sense
that an analysis of the available data alone will typically be biased.
Publication bias and selective reporting bias lead by definition to
data that are ’not missing at random’, and attrition and exclusions
of individuals within studies often do as well (Higgins 2011b).
If data turn out to be missing at random, we will analyse the available
information. If data are regarded to be notmissing at random,
we will make explicit the assumptions of any methods we use to
deal with the missing data, for example that the missing values
were assumed to indicate a poor outcome. We will undertake a
sensitivity analysis to assess how sensitive the results are to reasonable
changes in the assumptions made. We will also address the
potential impact of any missing data on the findings of the review
in the Discussion section (Higgins 2011b).