What makes a good quality indicator set? A systematic review of criteria

Abstract

Abstract Background While single indicators measure a specific aspect of quality (e.g. timely support during labour), users of these indicators, such as patients, providers and policy-makers, are typically interested in some broader construct (e.g. quality of maternity care) whose measurement requires a set of indicators. However, guidance on desirable properties of indicator sets is lacking. Objective Based on the premise that a set of valid indicators does not guarantee a valid set of indicators, the aim of this review is 2-fold: First, we introduce content validity as a desirable property of indicator sets and review the extent to which studies in the peer-reviewed health care quality literature address this criterion. Second, to obtain a complete inventory of criteria, we examine what additional criteria of quality indicator sets were used so far. Methods We searched the databases Web of Science, Medline, Cinahl and PsycInfo from inception to May 2021 and the reference lists of included studies. English- or German-language, peer-reviewed studies concerned with desirable characteristics of quality indicator sets were included. Applying qualitative content analysis, two authors independently coded the articles using a structured coding scheme and discussed conflicting codes until consensus was reached. Results Of 366 studies screened, 62 were included in the review. Eighty-five per cent (53/62) of studies addressed at least one of the component criteria of content validity (content coverage, proportional representation and contamination) and 15% (9/62) addressed all component criteria. Studies used various content domains to structure the targeted construct (e.g. quality dimensions, elements of the care pathway and policy priorities), providing a framework to assess content validity. The review revealed four additional substantive criteria for indicator sets: cost of measurement (21% [13/62] of the included studies), prioritization of ‘essential’ indicators (21% [13/62]), avoidance of redundancy (13% [8/62]) and size of the set (15% [9/62]). Additionally, four procedural criteria were identified: stakeholder involvement (69% [43/62]), using a conceptual framework (44% [27/62]), defining the purpose of measurement (26% [16/62]) and transparency of the development process (8% [5/62]). Conclusion The concept of content validity and its component criteria help assessing whether conclusions based on a set of indicators are valid conclusions about the targeted construct. To develop a valid indicator set, careful definition of the targeted construct including its (sub-)domains is paramount. Developers of quality indicators should specify the purpose of measurement and consider trade-offs with other criteria for indicator sets whose application may reduce content validity (e.g. costs of measurement) in light thereof.