Extraction of estimates of prognostic association for meta-analysis: simulation methods as good alternatives to trend and direct method estimation

Session: 

Oral session: Diagnostic test accuracy review and prognostic methods (2)

Date: 

Tuesday 18 September 2018 - 14:30 to 14:40

Location: 

Ochil 1

All authors in correct order:

Pérez T1, McLellan J2, Perera R2
1 Complutense University of Madrid, Spain
2 University of Oxford, UK
Presenting author and contact person

Presenting author:

Rafael Perera

Contact person:

Rafael Perera
Abstract text
Background:
Systematic reviews and meta-analysis are the standard methods to assess the association between prognostic markers and major events/conditions. However, the summary measures reported are not always explicitly presented and, therefore, different indirect methods of extracting estimates have been proposed.

Objectives:
To present two new, alternative methods for obtaining summary statistics to be included in a meta-analysis of prognostic studies based on simulating individual patient data, and to compare them with the already known, generalized least squares for trend estimation method and direct method.

Methods:
We have checked the performance of these methods using a between-study comparison, including 122 studies, and a within-study comparison, based on individual patient data from one of the studies.

Results:
The results obtained in this study show that generalized least squares for trend estimation method appears to overestimate the effect size when reported information is incomplete. For the within-study comparison, the closest approximation to the direct estimates was obtained using the approach based on simulating individual participant data.

Conclusions:
The proposed simulation methods are a good alternative when other well-known indirect methods cannot be used.

Patient or healthcare consumer involvement:
None

Relevance to patients and consumers: 

The identification of published studies and inclusion in summary analyses is critical for the evaluation of association between risk factors and relevant conditions. The work presented evaluates two alternative methods which should provide reviewers with further tools to incorporate data into pooled meta-analyses. This should improve our estimates of association potentially leading to better evidence and therefore better health care decision.