Survival meta-analysis of individual patient data and survival meta-analysis of published (aggregate) data - Is there an intermediate approach between these two opposite options?
A. Messori et al., Survival meta-analysis of individual patient data and survival meta-analysis of published (aggregate) data - Is there an intermediate approach between these two opposite options?, CLIN DRUG I, 20(5), 2000, pp. 309-316
Objective: To describe and evaluate, in comparison with traditional approac
hes, a technique of survival meta-analysis (the 'intermediate' method) that
re constructs individual patient data (IPD) from the graphs of survival cu
rves.
Patients: The examples considered in this paper were oncology patients (thr
ee data sets) and intensive-care unit patients (one data set).
Results: In two data sets from oncology patients, the intermediate method r
econstructed IPD from the graphs of the survival curves. In the comparison
between true IPD and estimated IPD (two data sets), the correlation analysi
s showed an excellent concordance between true survival times and estimated
survival times (correlation ranging from r = 0.989 to r = 1).
Conclusion: This technique is a compromise between meta-analysis of 'true'
IPD and meta-analysis of aggregate data without patient-level information.
This intermediate method of survival meta-analysis is suitable for small re
search groups that do not have the resources required to conduct a survival
meta-analysis based on true IPD.