Risk of bias and data extraction form for prognosis studies

Article type
Authors
El Dib R1, Pierre R2, Weber SAT2, Ridley G3, Williams K4
1UNESP - Univ Estadual Paulista, Brazil and; McMaster Institute of Urology, McMaster University, Canada
2UNESP - Universidade Estadual Paulista, Brazil
3Sydney Children’s Hospital Network, Australia
4University of Melbourne, Royal Children’s Hospital Melbourne, Murdoch Children’s Research Institute, Australia
Abstract
Background: Over the last few years there has been increasing interest in and production of systematic reviews of prognosis studies. One challenge is to create extraction sheets and risk of bias tables for prognosis studies, as there are many variables that could affect the internal validity of such a design and there are many relevant study designs. However few risk of bias methods are currently proposed.

Objectives: To describe a new tool of extracting data and assessing risk of bias of prognosis studies in the health field.

Methods: We have developed a data extraction form and risk of bias table based on Hayden and the MOOSE statement, and adapted by us, to be used for a systematic review entitled ‘Is the risk of cardiovascular diseases greater in children with obstructive sleep apnea than in the general population? Systematic review and meta-analysis of cohort studies’.

Results: This quality assessment method is feasible for use by all reviewers that are performing systematic reviews of prognosis studies as it focuses on factors such as study participants characteristics, sample selected, recruitment method, completeness of follow up, and blinding. In addition it presents ‘a two in one’ design form combining data extraction and risk of bias.

Conclusions: We describe a new extraction sheet and quality assessment method to evaluate prognosis studies in health care. This method is extended to be used in systematic reviews of prognosis studies, however further investigation are needed to verify if all variables were covered.