An Online Delphi Process to Identify ASD Diagnosis Guidelines for Best Practice Evaluation and Implementation

Background:  

Active mobilization of research evidence into clinical practice requires knowledge transfer processes which systematically implement activities which support better outcomes based on clear scientific evidence. Numerous guidelines, based on considerable evidence and expertise, have been developed to describe best practice in the diagnosis of ASD. A large tertiary children’s hospital, with approximately 500 ASD query referrals annually, chose an implementation science framework to enable implementation of best practice guidelines. This abstract describes consensus development processes used to identify initial best practice implementation targets.

Objectives:

Report a case study using implementation science principles to adapt and operationalize multiple best practice guidelines to establish an ASD diagnostic model for a tertiary clinic setting.   

Methods:

Four rounds of voting facilitated consensus development and decision-making over a 3-month period using a modified Delphi process which combined online and in-person group discussion. The process was designed to be iterative, using data and discussion to modify methodology. Decisions regarding changes were made prior to beginning each round with input from all stakeholders. The Delphi system for choosing best practice guidelines had the following characteristics:

• Representative stakeholders including different disciplines, levels of leadership, researchers and community representatives (n=9).
• Provision of evidence-based background information including ASD guidelines from 1) National Institute for Health and Clinical Excellence, 2) Miriam Foundation’s Canadian Best Practice Guidelines, and 3) British Columbia’s Standards and Guidelines for the assessment and diagnosis of ASD.
• Delphi rules/principles included anonymous on-line voting, criteria to guide Likert scale voting and between-round decision-making, between-round data synthesis and presentation, and adaptation of  the process based on data synthesis and stakeholder discussion.

 Results:

All guidelines were reviewed to determine which had complete consensus. These included a) which professionals could be included in the diagnostic team, b) team competencies, c) need for collaborative communication amongst team members and families, d) need for written documentation, e) need for formal observation and parental interview without reliance on a single diagnostic tool, and f) consideration for co-morbid conditions. Forty-seven guidelines were then used in the next round of online voting.  Participants voted using 6 criteria and a 5-point Likert scale. Data from this round resulted in changes to criteria for the next round, changes to the Likert scale, and 20 guidelines for consideration in the next round. A final voting round voting required participants to rank remaining guidelines with the goal of systematically implementing these practices for the diagnosis of all children with a query of ASD. An evaluation plan and implementation process, using the National Implementation Research Network framework, was developed to address implementation of guidelines identified as priorities for ASD diagnosis.

 Conclusions:

 This strategy was effective in achieving consensus across multiple stakeholders. We propose that implementation science provides an inclusive approach to determining service delivery guidelines and enacting change in a large institution. The implementation science framework will be used to establish evaluation tools for outcomes and implementation process.