JMIR Publications

ABSTRACT

Background:

Motivational interviewing (MI) is a behavior modification technique. MI-based digital interventions offer an accessible way of initiating alteration in patient habits. Selected papers evaluated the effectiveness of digital tools and specified conversion of MI spirit into a digital platform of which thirteen studies fit the criteria. This literature review aims to examine MI-based digital interventions (MIDIs) used for behavioral transformation in participants. A general effect size of between 0.2 and 0.3 was specified by some studies (n=5). All studies (n=9) established a positive behavioral effect on participants using MIDIs of which 44% applied pure MI principles using diverse forms of digital delivery methods. Some studies explored the influence of moderators on MIDI’s efficacy through demographic or behavioral segmentation. The fascinating premise for MIDI’s extensive deployment is the cost-effectiveness and enhanced accessibility thus promoting inclusivity across diverse demographic groups.

Objective:

This paper’s objective is to conduct a literature review of digital interventions that are based on MI principles used for behavioral change in participants.

Methods:

This literature review was conducted through Feb 29, 2024, the papers included were found using PubMed using the terms “(((motivational interview) AND (mobile app)) AND ((diabetes) OR (prediabetes))); ((motivational interviewing) AND (mobile app)); (((motivational interviewing) AND (conversational agent) AND ((diabetes) OR (prediabetes))); ((motivational interviewing) AND (conversational agent)); (((motivational interviewing) AND (chatbot) AND ((diabetes) OR (prediabetes))); ((motivational interview) AND (chatbot))”. Inclusion criteria comprised papers published in the English language were randomized controlled trials (RCTs) that assessed the effectiveness of digital tools that use motivational interviewing principles. These papers also presented details on the translation of MI spirit into a digital platform either in the paper or through referenced protocol studies or feasibility tests. The papers with only face-to-face use of MI techniques and papers behind a paywall were excluded. No protocol studies or feasibility tests were included that did not have an RCT study yet. Using the terms mentioned above the total number of papers found was 126 after the inclusion and exclusion criteria were applied, 13 papers were selected.

Results:

Thirteen studies used digital interventions based on MI principles, of which 8 were RCTs, 1 was a prospective study and 4 were development or feasibility studies for the RCTs. The sample sizes for RCTs were an average of 1450 participants with a minimum of 120 participants and a maximum of 3088 participants with ages ranging from 18-70 years. While most studies (n=6) had a diverse group of participants, a couple of studies’ (n=2) target participants were narrow. 56% of studies used MI spirit in combination with self-determination theory (SDT) principles, while 44% of studies implemented pure MI principles in their digital health intervention. 78% of the studies included are from the Netherlands and 22% were set in the United States of America. All the studies focused on reducing risk factors for chronic diseases with most (n=5) focused on physical activity (PA), others looked to implement improvements in substance use(n=2) and HIV prevention (n=1). Delivery of digital interventions varies from web-based computer-tailored (n=5), text messaging (n=1), chatbot (n=1) and avatar (n=2). Most of the RCTs (n=7) had prior development or feasibility studies. All studies (n=9) demonstrated a positive behavioral impact on participants when using MIDIs and showed statistically significant results for the interventions compared to the control group. However, no significant difference was noted between MI intervention and non-MI intervention (n=2) or when MI principles were used in two different types of intervention (n=2). Still, a few studies displayed statistical significance in some aspects tested between MI intervention and non-MI intervention (n=3) and in a study (n=1) testing out MI principles in a digital format. Effect size data was provided by some studies (n=5) using Cohen’s d with a target effect size between 0.2 and 0.3 for the overall study, while a study stated the effect size of 0.2 but not the specific test used to measure that value. A couple of studies (n=2) that analyzed MI content in these digital interventions found that open-ended questions (OEQs) were associated with a statistically significant behavioral change in participants. Using MIDIs (n=3) intensely and for a long time presented a positive behavioral change in certain aspects tested. There were a few studies (n=6) that attempted to segment their data, of which one study found no effect from the moderators studied. A couple of studies (n=2) analyzed demographic modulators of the MIDIs, whereas the other studies (n=3) segmented their data based on behavioral aspects that had an impact on MIDIs outcomes. Older par­ticipants had a lower attrition rate in most studies (n=6) when compared to younger participants. In comparison, a <13% overall dropout rate was noted in some studies (n=2).

Conclusions:

In this literature review, it was discovered that MIDIs instigate a positive behavioral impact on its participants, advocating for their widespread adoption into practice. Further development in Natural Language Processing systems is vital to achieving real-time evaluation and providing timely feedback to the users within MIDIs, enhancing their effectiveness and user experience.