Individual patient data meta-analyses: distribution and epidemiological characteristics of published studies

ID: 

105

Session: 

Poster session 4

Date: 

Tuesday 25 October 2016 - 15:30 to 16:00

Location: 

All authors in correct order:

Threapleton D1, Huang Y1, Tang J1
1 Hong Kong Branch of the Chinese Cochrane Centre, Hong Kong
Presenting author and contact person

Presenting author:

Jin-Ling Tang

Contact person:

Abstract text
Background: Individual patient data meta-analyses (IPDMAs) offer advantages over traditional meta-analyses and are considered the ‘gold-standard’. However, the general characteristics of existing IPDMAs are unknown and methodological features and success in obtaining IPD may affect the quality of meta-analyses.

Objectives: To identify all published IPDMAs to date, and summarise the distribution and epidemiological characteristics.

Methods: IPDMAs were sought by comprehensive searches of PubMed, Embase and the Cochrane Library on 9 August 2012. Two researchers independently screened articles and extracted data. Study characteristics were synthesised descriptively.

Results: The earliest identified IPDMA was published in 1987 and, with an annual increase of approximately 3.7 articles, 97 were published in 2011. In total, 829 IPDMAs were identified, the majority of which related to malignant neoplasms n = 267 (32.2%) and circulatory diseases n = 179 (21.6%). Each IPDMA included a median of eight studies (interquartile range (IQR) 5 to 15) and included a median of 2563 patients (IQR 927 to 8349). Over half of IPDMAs successfully identified data from all identified studies (n = 496, 59.8%) and one quarter of studies (n = 207, 25.0%) sought data from ‘grey literature’. However, a high proportion of IPDMAs (n = 229, 27.6%) did not use systematic methods to locate studies.

Conclusions: IPDMAs have grown in popularity and have focused on cancer and circulatory diseases. Methodological approaches for sourcing relevant studies differ between IPDMAs, with some not using systematic search methods or including grey literature. Results from IPDMAs are likely subject to selection bias, publication bias and poor data availability and thus, findings from IPDMAs should not be unequivocally accepted by decision makers without awareness of these limitations and an understanding of the potential impact on findings.