Improving the analysis of individual participant data
(IPD)
In our IPD meta-analyses, we collect data on all trials,
published and unpublished, on all randomised patients and on the
same outcomes for each trial. This should limit publication bias,
attrition bias and selective outcome reporting bias as far as
possible. We also update follow-up in order to reduce any
follow-up bias in long-term outcomes. As the IPD are thoroughly
checked and queried, this often improves the quality of the data
further.
These approaches should help ensure that we consistently provide
the most up-to-date, unbiased and reliable estimates of the effects
of treatments. However, we also need to ensure that we select and
develop the best methods for analysis of IPD so that we optimally
describe the effects of therapies and maximise the potential of our
data.
In many situations, we are interested in time-to-event outcomes
e.g. time to disease recurrence or time to death. Methods of
analysis of these outcomes, particularly in terms of summarising
the treatment effect, need further development and assessment.
Key
projects
- Exploring the best methods for assessing treatment by patient
subgroup interactions
- Identifying the best methods for summarising treatments effects
based on IPD
- Identifying the most suitable summary measures for
meta-analysis of time-to-event outcomes
- Using cumulative incidence estimation to allow for competing
causes of failure in meta-analysis
Selected publications
- Siannis F, Barrett JK, Farewell VT, Tierney JF. One-stage
parametric meta-analysis of time-to-event outcomes. Stat Med. 2010
Dec 20;29(29):3030-45. Epub 2010 Oct 20
- Fisher DJ, Copas AJ, Tierney JF, Parmar MK. A critical review
of methods for the assessment of patient-level interactions in
individual participant data meta-analysis of randomized trials, and
guidance for practitioners. J Clin Epidemiol. 2011
Sep;64(9):949-67. Epub 2011 Mar 16
- Bowden J, Tierney J, Simmonds M, Copas AJ, Higgins JPT.
Individual patient data meta-analysis of time-to-event outcomes:
one-stage versus two-stage approaches to estimating the hazard
ratio under a random effects model. Res Synthesis Methods, in
press
- Simmonds MC, Tierney J, Bowden J, Higgins JPT. Meta-analysis of
time-to-event data: a comparison of two-stage methods. Res
Synthesis Methods, in press
