Implementing the PREP2 algorithm to predict upper limb recovery potential after stroke in clinical practice: a qualitative study

Abstract

Background
Predicting motor recovery after stroke is a key factor when planning and providing rehabilitation for individual patients. The PREP2 algorithm has been developed to help clinicians predict upper limb functional outcome. Translating evidence-based interventions into clinical practice can be challenging and slow. However, shortly after its external local validation, PREP2 was successfully implemented into clinical practice at the same site in New Zealand. In parallel to further model validation, useful lessons can be learned from this experience to aid future implementation.

Objective
To explore how PREP2 was implemented in clinical practice within the Auckland District Health Board (ADHB) in New Zealand.

Design
A case study design using semi-structured interviews.

Methods
Nineteen interviews were conducted with clinicians involved in stroke care at ADHB. To explore factors influencing implementation, interview content was coded and analysed using the Consolidated Framework for Implementation Research. Strategies identified by the Expert Recommendations for Implementing Change (ERIC) project were used to describe how implementation was undertaken.

Results
Implementation of PREP2 was initiated and driven by therapists. Key factors driving implementation were the support given to staff from the implementation team; the knowledge, beliefs and self-efficacy of staff, and the perceived benefits of having PREP2 prediction information. Twenty-six ERIC strategies were identified relating to three areas: the implementation team, the clinical/academic partnerships and the training.

Limitations
Limitations included potential self-selection bias, reliance on clinicians’ ability to recall events, and potential social desirability bias affecting interview content.

Conclusions
The PREP2 prediction tool was successfully implemented in clinical practice at ADHB. Barriers and facilitators to implementation success have been identified, and implementation strategies described. Lessons learned can aid future development and implementation of prediction models in clinical practice.