Challenges in conducting an overview of reviews evaluating diagnostic accuracy and predictive ability of frailty screening tools: a practical example

Tags: Poster
Apóstolo J1, Bobrowicz-Campos E2, Holland C3, Cooke R3, Santana S4, Marcucci M5, Vollenbroek M6, Germini F7, Cano A8
1Health Sciences Research Unit: Nursing, Nursing School of Coimbra, Portugal Centre for Evidence-Based Practice: a Collaborating Centre of the Joanna Briggs Institute, Portugal, 2Health Sciences Research Unit: Nursing, Nursing School of Coimbra, Portugal Centre for Evidence Based Practice: a Collaborating Centre of the Joanna Briggs Institute, Portugal, 3Aston Research Centre for Healthy Ageing (ARCHA), Aston University, Birmingham, UK, 4Economics, Management and Industrial Engineering Department, University of Aveiro, Portugal, 5Geriatric Unit, Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico and Department of Clinical Science and Community Health, University of Milan, Milan, Italy, 6Roessingh Research and Development, The Netherlands, 7Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Italy, 8Department of Paediatrics, Obstetrics and Gynaecology, Universitat de Valência, Spain

Background: Within the context of the project 664367/FOCUS funded under the European Union’s Health Programme (2014-2020), we have conducted an overview of systematic reviews (SRs) to examine diagnostic accuracy and predictive ability of available screening tools for frailty. This review process was based on Joanna Briggs Institute (JBI) procedures. Of 420 records identified through searching in databases for published and unpublished studies, 20 full-texts were assessed for inclusion criteria and then 10 for risk of bias (RoB). We encountered various limitations when we started to appraise the methodological quality of the SRs eligible for inclusion.

Objectives: To describe the potential bias of the SRs eligible for inclusion in one overview of SRs related to diagnostic accuracy and predictive validity of screening tools for frailty.

Methods: Detailed analysis of the RoB by applying the 11 items of the JBI critical appraisal checklist for systematic reviews and research syntheses to the SRs eligible for inclusion and by data extraction based on the JBI data extraction form for review of systematic reviews and research syntheses.

Results: One of the 10 analyzed SRs was a Cochrane SR, and nine were non-Cochrane and non-JBI SRs. In the Cochrane Review only the likelihood of publication bias was not controlled. With regard to the remaining nine SRs: in two the inappropriate definition of inclusion criteria was identified; in five the reference standard using for comparison of the index tests was not considered; in two the critical appraisal of the included studies was missing, and in one an inappropriate tool for this purpose was used. None of these nine SRs evaluated likelihood of publication bias. Related to data extraction, we identified cases of bias in the selection of the reported results, lack of uniformity of provided statistics, and inconsistency in conferring significance to the obtained results.

Conclusions: RoB were mainly identified in the SRs that did not follow standardized international collaboration procedures. There is a need for wider use in future SRs of standardized procedures in order to improve the quality of evidence synthesis.