Building better guidelines with BRIDGE-Wiz: development and evaluation of a software assistant to promote clarity, transparency, and implementability

Abstract

Objective: To demonstrate the feasibility of capturing the knowledge required to create guideline recommendations in a systematic, structured, manner using a software assistant. Practice guidelines constitute an important modality that can reduce the delivery of inappropriate care and support the introduction of new knowledge into clinical practice. However, many guideline recommendations are vague and underspecified, lack any linkage to supporting evidence or documentation of how they were developed, and prove to be difficult to transform into systems that influence the behavior of care providers.

Methods: The BRIDGE-Wiz application (Building Recommendations In a Developer's Guideline Editor) uses a wizard approach to address the questions: (1) under what circumstances? (2) who? (3) ought (with what level of obligation?) (4) to do what? (5) to whom? (6) how and why? Controlled natural language was applied to create and populate a template for recommendation statements.

Results: The application was used by five national panels to develop guidelines. In general, panelists agreed that the software helped to formalize a process for authoring guideline recommendations and deemed the application usable and useful.

Discussion: Use of BRIDGE-Wiz promotes clarity of recommendations by limiting verb choices, building active voice recommendations, incorporating decidability and executability checks, and limiting Boolean connectors. It enhances transparency by incorporating systematic appraisal of evidence quality, benefits, and harms. BRIDGE-Wiz promotes implementability by providing a pseudocode rule, suggesting deontic modals, and limiting the use of 'consider'.

Conclusion: Users found that BRIDGE-Wiz facilitates the development of clear, transparent, and implementable guideline recommendations.

Figure 1

Empirically derived classification of action types. Reviewing more than 700 randomly selected recommendations, recommended actions for the vast majority could be classified into these categories.

Figure 2

Users choose an action type (‘PRESCRIBE’) and select a transitive, active-voice verb (‘start’). Next they define the object of the verb (‘metformin as first-line treatment’).

Figure 3

After constructing individual tables that reflect benefits versus risks, harms, and costs, users are prompted to judge whether there is a balance or a preponderance.

Figure 4

Bridge-Wiz summarizes the judgments about balance or preponderance of benefits versus risks and proposes a strength of recommendation and an appropriate deontic term.

Figure 5

Panelists' responses to survey questions. Numbers on the bars represent the number of respondents choosing each level of agreement. The number of panelists responding to the item is displayed in the column entitled ‘Total responses’. Median response and 25th and 75th quartiles for each item are displayed.