Guideline appraisal with AGREE II: Systematic review of the current evidence on how users handle the 2 overall assessments

Wiebke Hoffmann-Eßer; Ulrich Siering; Edmund A M Neugebauer; Anne Catharina Brockhaus; Ulrike Lampert; Michaela Eikermann

doi:10.1371/journal.pone.0174831

Guideline appraisal with AGREE II: Systematic review of the current evidence on how users handle the 2 overall assessments

PLoS One. 2017 Mar 30;12(3):e0174831. doi: 10.1371/journal.pone.0174831. eCollection 2017.

Authors

Wiebke Hoffmann-Eßer^{1

2}, Ulrich Siering¹, Edmund A M Neugebauer³, Anne Catharina Brockhaus¹, Ulrike Lampert¹, Michaela Eikermann⁴

Affiliations

¹ Institute for Quality and Efficiency in Health Care (IQWiG), Cologne, Germany.
² Institute for Research in Operative Medicine (IFOM), University of Witten/Herdecke, Campus Cologne, Cologne, Germany.
³ Brandenburg Medical School - Theodor Fontane Neuruppin, Germany & University of Witten/Herdecke, Witten/Herdecke, Germany.
⁴ Medical Advisory Service of the German Social Health Insurance (MDS), Essen, Germany.

Abstract

Introduction: The Appraisal of Guidelines for Research & Evaluation (AGREE) II instrument is the most commonly used guideline appraisal tool. It includes 23 appraisal criteria (items) organized within 6 domains and 2 overall assessments (1. overall guideline quality; 2. recommendation for use). The aim of this systematic review was twofold. Firstly, to investigate how often AGREE II users conduct the 2 overall assessments. Secondly, to investigate the influence of the 6 domain scores on each of the 2 overall assessments.

Materials and methods: A systematic bibliographic search was conducted for publications reporting guideline appraisals with AGREE II. The impact of the 6 domain scores on the overall assessment of guideline quality was examined using a multiple linear regression model. Their impact on the recommendation for use (possible answers: "yes", "yes, with modifications", "no") was examined using a multinomial regression model.

Results: 118 relevant publications including 1453 guidelines were identified. 77.1% of the publications reported results for at least one overall assessment, but only 32.2% reported results for both overall assessments. The results of the regression analyses showed a statistically significant influence of all domains on overall guideline quality, with Domain 3 (rigour of development) having the strongest influence. For the recommendation for use, the results showed a significant influence of Domains 3 to 5 ("yes" vs. "no") and Domains 3 and 5 ("yes, with modifications" vs. "no").

Conclusions: The 2 overall assessments of AGREE II are underreported by guideline assessors. Domains 3 and 5 have the strongest influence on the results of the 2 overall assessments, while the other domains have a varying influence. Within a normative approach, our findings could be used as guidance for weighting individual domains in AGREE II to make the overall assessments more objective. Alternatively, a stronger content analysis of the individual domains could clarify their importance in terms of guideline quality. Moreover, AGREE II should require users to transparently present how they conducted the assessments.

Publication types

Review
Systematic Review

MeSH terms

Literature*
Regression Analysis
Software*

Grants and funding

Non-monetary support for this research was provided by the Institute for Quality and Efficiency in Health Care (IQWiG). No external funding was received.