Behaviour change interventions based on automated and tailored communication systems to improve health outcomes in adult chronic patients: a systematic review

Abstract

Background: Behaviour change digital-assisted interventions can be a low-cost solution to enduring healthy behaviours. Understanding how to automate and tailor such interventions to patients’ needs can improve health outcomes. This study aims at reviewing the evidence of the use of such interventions in a clinical setting in adult chronic patients. Methods: This study followed the PRISMA guidelines for comprehensive search, appraisal, and synthesis of research evidence. A literature search was performed to find studies published between 1 January 2017 and 26 March 2022 in peer-reviewed journals and written in English. Behaviour change interventions in adult chronic patients with automated and tailored communication systems were considered. Results: Seven quantitative studies (6 RCT and a pilot RCT) with sample sizes from 54 to 506 participants and timeframes varying from 4 to 30 weeks met the eligibility criteria. Populations comprehended patients with cardiovascular disease, chronic kidney disease, diabetes, or osteoarthritis. Communication channels included SMS, email, and voice records. Rule-based models were used to automate content in all studies and bidirectional communication was used in four of them. Content and service delivery were tailored given the patient’s goals, needs, functional ability, activity, personal characteristics, and communication channel preferred. Most data management platforms used were developed in prior studies. Interventions were designed based on 9 different behaviour change models. There was no evidence of improvements in clinical outcomes after the intervention. Some studies showed improvements in self-efficacy (n=1), the use of dietary data info (n=1), physical activity (n=2), social support (n=1), and quality of life (n=1). The risk of bias analysis revealed that one study had moderate risk and the remaining had low risk. Conclusion: This study supports researchers that are designing digital behaviour change interventions by putting in evidence the intervention’s features needed to be considered during this step. It revealed that the most efficient solutions were tailored to patients’ needs and disease stages and that further research on the mechanisms of change and content automation needs to be done.