Meta-analyses that are based on
individual patient or participant data (IPD) use the same basic
approach as any other well-conducted systematic review and
meta-analysis. However, they involve the collection of the original
data from all the relevant trials worldwide. This means trial
groups need to be persuaded to supply their trial data, and these
data need to be prepared and checked before being included in the
meta-analysis. Therefore, to succeed, an IPD meta-analysis relies
on extensive collaboration between researchers. It also takes more
time and resource than a meta-analysis based on results extracted
from published trial reports. However, the IPD approach can improve
the quality of both the data and the analyses and so the
reliability of the results. Therefore, it is considered the gold
standard.
IPD meta-analyses are less common
than other types of systematic review. They were first used to
summarise the effects of treatments for cardiovascular disease and
cancer. They are now used in a range of healthcare areas e.g. to
examine the effects of treatments for Alzheimers disease,
dyspepsia, epilepsy, malaria, HIV infection and hernia.
The CTU Meta-analysis Group is one
of only a few groups worldwide that regularly does IPD
meta-analyses. Practical guidance on IPD meta-analysis published by
the Group is listed below. Further information on IPD meta-analyses
can be found on the Cochrane Collaboration
IPD Meta-analysis Methods Group website.
For our current methodological
research relating to IPD meta-analysis see CTU Hub for Trials Methodology
Research.
Stewart LA, Tierney JF.
To IPD or Not to IPD? Evaluation & the Health Professions
2002;25(1)76-97.
abstract
Stewart LA, Clarke MJ, on behalf of the Cochrane
Working Party Group on Meta-analysis using Individual Patient Data.
Practical methodology of meta-analyses (overviews) using updated
individual patient data. Statistics in Medicine 1995;142057-79.
abstract
