Development, validation, and implementation strategies for the exercise in cancer evaluation and decision support (EXCEEDS) algorithm
Date
2021
Authors
Covington, Kelley R., author
Bundy, Anita, advisor
Pergolotti, Mackenzi, advisor
Sharp, Julia, committee member
Leach, Heather, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
Background: Clinical practice guidelines recommend referral to cancer rehabilitation or exercise services (CRES) to optimize survivorship. Yet, ability to connect the right survivor with the right CRES at the right time is an ongoing challenge and barrier to utilization of these services. Objective: I aimed to develop a CRES decision support algorithm and used Delphi methodology to systematically: (1) evaluate the algorithm's acceptability and utility; and (2) establish consensus for implementation priorities including key stakeholders, platforms and strategies. Method: I performed a literature review and synthesis, then convened a multidisciplinary expert stakeholder group to participate in algorithm development. We worked iteratively and collaboratively until consensus was reached for content and format of the Exercise in Cancer Evaluation and Decision Support (EXCEEDS) algorithm and conceptual model. Then I recruited international clinical and research experts to participate in the two-part (three survey) online modified Delphi study. In Part 1, participants completed one survey including: (1) CRES recommendations for two randomized case studies in two conditions (using EXCEEDS vs. without EXCEEDS); (2) the Acceptability of Implementation Measure (AIM); and (3) open-ended feedback on the algorithm. Following this survey, I compared decision efficiency (accuracy and duration) between conditions (EXCEEDS vs. independently) for each case study using frequencies (hypothesis ≥75% accurate) and paired samples t-test (p <.05), then calculated consensus for each AIM domain ("meets approval", "is appealing", "welcome in my field/practice"; hypothesis ≥70% agreement) and overall score (hypothesized mean ≥ 4.0). These results were reported to participants in Part 2. I also performed inductive thematic analysis of open-ended feedback. In Part 2 of the Delphi study, participants completed a series of two surveys including ranking the following items using curated lists: (1) stakeholder group (1 -most likely to 7- least likely to benefit), (2) platform (1 - most likely to 6- least likely to be beneficial) and, implementation strategies (1 - most important to 15- least important for successful implementation). I performed preliminary analysis of each ranking using measures of central tendency (median and IQR), then calculated the proportion of participants who ranked each option as a high priority. Ten implementation strategies were ranked as high priority and returned to participants for the final survey where they rated each strategy in terms of effort associated with using the strategy (1 - low effort to 4 - high effort) and potential impact of the strategy on successful implementation of the EXCEEDS algorithm (1 – low impact to 4 - high impact). Following the Eisenhower Urgent-Important Matrix Method, I plotted the effort/impact scores in four quadrants representing effort and impact for each strategy to determine implementation priorities. Results: The final EXCEEDS algorithm combines biomedical and individual characteristics associated with need for supervised skilled CRES into 11 risk-stratified dichotomous (yes/no) questions, organized into two sections: (1) pre-exercise medical clearance recommendation, and (2) CRES triage recommendation. Delphi study participants (N=133) represented all CRES stakeholder groups (oncology, physical medicine and rehabilitation, exercise science, etc.). Loss to follow up between surveys ranged 28% (survey 3) to 43% (survey 2). When using the EXCEEDS algorithm, decision accuracy improved in six (of eight) conditions (75%) and duration improved in all conditions (N=4, p <.05). Consensus was achieved in three AIM domains (75%); overall AIM score was M=3.90 ± 0.473 (range = 1.0 – 5.0). Qualitative themes from participant feedback include: (1) algorithm strengths (n = 123, 40.9%), (2) implementation considerations (n=93, 30.5%), and (3) areas for revision (n=87, 28.5%). Oncology clinicians and administrators were the highest-ranked stakeholder group (Median=2.0, IQR= 1.0 – 3.75, 75.0% agreement) and the only one to achieve consensus. Open-access internet was the highest-ranked implementation platform (Median =2.0, IQR= 1.0 – 3.5, 72.4% agreement) and the only one to achieve consensus. Consensus was achieved for eight of the ten highest-ranked implementation strategies (80%, inter-rater agreement range = 93.4% - 71.1%). Two strategies were categorized as urgent/important: "develop educational materials" and "remind clinicians". Seven strategies were categorized important/not urgent. One strategy, "model and simulate change", was categorized as not important/not urgent. Conclusion: The EXCEEDS algorithm is an acceptable and efficient evidence-based solution to identify and connect the right survivor, with the right CRES, at the right time. Thus, implementation of the EXCEEDS algorithm guided by the consensus-based priorities identified in the Delphi study has the potential to improve CRES coordination and utilization. Future hybrid studies will be used to determine prospective efficacy and best practices for implementation.
Description
Rights Access
Subject
cancer
exercise
rehabilitation
clinical decision support
algorithm
health services