Systematic review found AMSTAR, but not R(evised)-AMSTAR, to have good measurement properties

Abstract

Objectives: To summarize all available evidence on measurement properties in terms of reliability, validity, and feasibility of the Assessment of Multiple Systematic Reviews (AMSTAR) tool, including R(evised)-AMSTAR.

Study design and setting: MEDLINE, EMBASE, Psycinfo, and CINAHL were searched for studies containing information on measurement properties of the tools in October 2013. We extracted data on study characteristics and measurement properties. These data were analyzed following measurement criteria.

Results: We included 13 studies, four of them were labeled as validation studies. Nine articles dealt with AMSTAR, two articles dealt with R-AMSTAR, and one article dealt with both instruments. In terms of interrater reliability, most items showed a substantial agreement (>0.6). The median intraclass correlation coefficient (ICC) for the overall score of AMSTAR was 0.83 (range 0.60-0.98), indicating a high agreement. In terms of validity, ICCs were very high with all but one ICC lower than 0.8 when the AMSTAR score was compared with scores from other tools. Scoring AMSTAR takes between 10 and 20 minutes.

Conclusion: AMSTAR seems to be reliable and valid. Further investigations for systematic reviews of other study designs than randomized controlled trials are needed. R-AMSTAR should be further investigated as evidence for its use is limited and its measurement properties have not been studied sufficiently. In general, test-retest reliability should be investigated in future studies.

Keywords: Clinimetrics; Epidemiologic methods; Measurement properties; Observer variation; Psychometrics; Reliability and validity; Systematic reviews.