Implementation of an algorithm for predicting exacerbations in telemonitoring: A multimethod study of patients’ and clinicians’ experiences

Abstract

Background

Prediction algorithms may improve the ability of telehealth solutions to assess the risk of future exacerbations in patients with chronic obstructive pulmonary disease. Learning from patients’ and clinicians’ evaluations and experiences about the use of such algorithms is essential to evaluate its potential and examine factors that could potentially influence the implementation and sustained use.

Objective

To investigate the patients’ and clinicians’ perceptions and satisfaction with an algorithm for predicting exacerbations in patients with chronic obstructive pulmonary disease.

Design

Multimethod study.

Setting

Three community nursing sites in Aalborg Municipality, Denmark.

Participants

One hundred and eleven adults with chronic obstructive pulmonary disease and four clinicians (three nurses and one physiotherapist) specialized in telehealth monitoring of the disease.

Methods

The study was performed from November 2021 to November 2022 alongside a clinical trial in which a prediction algorithm was integrated into an existing telehealth system. The patients’ perspectives were investigated using a self-constructed questionnaire. The clinicians’ perspective was explored using semistructured individual interviews.

Results

Most patients (84.0 %–90.8 %) were satisfied with the algorithm and the additional measurements required by the algorithm. Approximately 71.7 %–75.9 % found that the algorithm could be a useful tool for disease assessment. Patients elaborated that they could see an exacerbation prevention potential in the algorithm. Patients trusted the algorithm and found an increased sense of security. The clinicians showed a positive response toward the algorithm and its user-friendliness. However, they were concerned that the additional measurements could be too demanding for some patients and questioned the accuracy of the measurements. Some felt that the algorithm could risk being time-consuming and harm the overall assessment of the individual patient. They expressed a need for continuous information about the algorithm to understand its functions and alarms.

Conclusions

Optimal use of the algorithm would require that patients perform additional pulse and oxygen saturation measurements. Furthermore, it will require in-depth insight among clinicians regarding the algorithm's functions and alarms.

Registration

The study was performed alongside a clinical trial, which was first registered September 9, 2021, at clinicaltrials.gov (registration number NCT05218525). Date of first recruitment was September 28, 2021.