An Introduction to Handling Missing Data in Health Economic Evaluations
Baio, G;
Leurent, B;
(2016)
An Introduction to Handling Missing Data in Health Economic Evaluations.
In: Round, J, (ed.)
Care at the End of Life: An Economic Perspective.
Springer, pp. 73-85.
ISBN 9783319282664
https://researchonline.lshtm.ac.uk/id/eprint/2535953
Permanent Identifier
Use this permanent URL when citing or linking to this resource.
Missing data is a problem commonly encountered in health intervention studies. It
is particularly prevalent in palliative care research where collection of complete data
is often hampered by the deteriorating health and sometimes death of the study
participants. Missing data is a limitation for the validity of an economic evaluation.
If missing data is not dealt with appropriately during analysis, there is a risk of bias
occurring in the estimates of the outcomes, both benefits and costs, associated with
the interventions being compared. Such biased results would in turn effectively
invalidate any inferences made to guide the decision-making process.
Given the importance of dealing properly with missing data to making correct
inferences from study results, we present here a guide paying specific attention to
problems that might arise when studying end-of-life populations in a health economic
setting. We begin by describing what it means for data to be missing and how
this affects inference. We then outline methods for dealing with different types of
missing data, progressing from simple (and often unsatisfactory) methods such as
complete-case analysis to more sophisticated and computationally expensive
approaches such as multiple imputation, which generally produce more meaningful
results. We finish by considering specific issues around end-of-life populations.