A parallel guideline development and formalization strategy to improve the quality of clinical practice guidelines

Abstract

Purpose

Clinical practice guidelines often contain ambiguities, inconsistencies, and logical errors that hamper implementation of these guidelines in practice. As guideline formalization is useful to verify the logical structure, consistency, and completeness of guidelines, several authors have argued that the formalization of guidelines concurrent with their development may improve their quality. However, experiences with such a parallel guideline development and formalization approach have not yet been reported. The goal of this study was to develop such a strategy and evaluate its application in practice.

Methods

Existing methodologies for guideline development and guideline formalization were analyzed and used as a basis to develop a strategy in which guideline formalization is performed concurrently with guideline development. The developed strategy was applied in the development of a clinical practice guideline for cardiac rehabilitation.

Results

A parallel guideline development and formalization strategy was developed that intertwines the processes of guideline development and guideline formalization. Central assets are early involvement of guideline formalization specialists and formalization tools, cooperation between guideline authors and guideline formalization specialists in the development of clinical algorithms, access to domain knowledge when formalization identifies inconsistencies or omissions, and formal verification of the guideline model prior to guideline dissemination. This strategy was applied in the development of a guideline for cardiac rehabilitation and helped to identify several vague and inconsistent recommendations and impracticabilities in the narrative guidelines that could be resolved before publication. In addition, the strategy ensured consistency between the narrative and formalized guideline.

Conclusions

Based on our experience, formalizing a guideline concurrent with its development is feasible in practice and we recommend applying such a strategy as it can be beneficial to the quality of and consistency between the guideline's narrative and formalized version.

Introduction

In contemporary healthcare, clinical practice guidelines are considered essential instruments to improve the quality of care [1] as they are found to improve patient outcomes, reduce practice variation, and reduce costs [2], [3], [4]. However, the adherence to guidelines by care professionals in clinical practice is often low [5]. One important reason for this is that, despite the great efforts put into their development, guidelines often contain ambiguous and vague concepts and recommendations, omissions, inconsistencies, and other errors [6], [7], [8], [9], [10]. Although no systematic studies are conducted on the number of such errors in guidelines in general, findings from different studies suggest that most guidelines contain from several up to dozens of such errors [10], [11], [12], [13]. As these issues hamper the use of guidelines in practice they often do not have the desired quality improvement effect on healthcare [8], [10], [14].

A powerful method to verify the logical structure, consistency, and completeness of practice guidelines is formalization, i.e. translating them into a formal, symbolic representation that allows for automated interpretation and reasoning by computers [8], [10], [11], [12], [15], [16]. Traditionally, guidelines are formalized to either provide computerized decision support or to facilitate the sharing of guidelines between institutions [17], [18]. Guideline formalization requires structuring and summarizing the knowledge and recommendations in the guidelines, and this process is increasingly recognized as helpful in identifying different types of errors that hamper the application of guidelines in practice. In addition, some guideline formalization tools include functionalities to help to identify errors by automatic verification of the logical consistency of the formalized guideline [10], [18].

In published accounts, guideline formalization was done after publication of the guideline concerned [7], [8], [10], [11], [12], [13], [16], [19]. In such an approach the errors in narrative guidelines make proper guideline formalization difficult and time-consuming without close collaboration with guideline authors. In guideline formalization, it is critical that the formalized guideline closely corresponds to the narrative guideline and reflects the intentions of the guideline authors. Also, to facilitate and reduce the variability in the guideline's interpretation and execution, vague concepts and recommendations in the guideline (e.g., ‘low’ or ‘poor’) need to be clearly delineated. Also, the usefulness of identifying problems by formalization is limited when done after guideline publication as the narrative guideline is already published and cannot be changed anymore [8], [10]. Also, resolution of these problems is difficult as it would require the involvement of guideline authors, and if they are resolved this will often compromise the correspondence between the narrative and formalized guideline.

Several authors have suggested to formalize a guideline concurrent with its development [10], [12], [15], [16]. First, this can be beneficial to the quality of the narrative guideline as this helps to identify problems that can still be resolved prior to publication. Second, the cooperation with guideline authors has the advantage that they can be consulted in case of implementation issues. Finally, the concurrent guideline formalization improves the consistency between the narrative and formalized guideline. Biondich et al. [15] have recently reported on two pilot projects where experience was built up with a collaboration between medical informatics and guideline authoring communities. In this paper, we extend this approach by developing a structured, parallel guideline development and formalization strategy that combines the principles of existing methodologies for guideline development, knowledge engineering, and guideline formalization. In addition, we report on our experience with the application of this strategy in the development of a guideline for multidisciplinary cardiac rehabilitation.