A systematic literature search was conducted from inception until August 2023 in six databases: CINAHL complete (EBSCO), Medline (Ovid), PubMed central (PMC), Physiotherapy Evidence Database (PEDro), Embase (Ovid), and the Cochrane library (Central). The search terms and strategies were developed and piloted in consultation with the research team and a specialist health science librarian.

Keywords for motivational interviewing, physiotherapy/physical therapy and rehabilitation were mapped to medical subject headings and combined with keyword search results (including truncations to account for variations in spelling). Results for ‘physical therapy’/ ‘physiotherapy’ “OR” ‘rehabilitation’ were combined with results for ‘motivational interviewing’ using the “AND” operator (see Supplementary material for full search strategies).

The inclusion criteria followed the PICOs framework (population, intervention, comparator, outcome, study design).17 Studies were included if they were peer-reviewed randomised controlled trials evaluating physical therapist-delivered MI focussed on health-related behaviour change (e.g. physical activity, dietary intake) in adults attending physical therapy / rehabilitation (including physical therapy) with or without co-intervention/usual care (full inclusion criteria provided in Supplementary material).

Trials evaluating MI for mental health conditions and/or addiction were excluded. Trials where MI was combined with other behavioural interventions, e.g. cognitive behavioural therapy, were excluded to enable this review to focus on the behavioural effect of MI. Studies that used motivational enhancement therapy, a brief therapeutic approach using MI principles, were also excluded.

After removal of duplicates, two reviewers independently screened titles and abstracts and then the remaining full-text articles for inclusion via the online platform Covidence.18 Agreement between reviewers was recorded. Disagreements between reviewers were resolved by a third reviewer, or through discussion within the team. Reference lists of included papers were screened to identify any additional papers for review. Authors of full-text papers were contacted to clarify if physical therapists provided the MI intervention if this was not clear.

Data extraction was completed independently by two reviewers (EW, CP). Data were extracted into a customised excel spreadsheet and included demographic factors (age, sex, country, year of study, health condition, setting), experimental and control interventions, primary (e.g. physical activity) and secondary outcomes (patient health outcomes) and time points, adverse events, and results. Because MI efficacy is associated with treatment fidelity,19 MI treatment fidelity and methods to confirm proficiency of the MI physical therapists were also extracted. Where the required information was not clearly reported, lead authors of included papers were contacted to request additional detail.

Internal validity of included trials were evaluated using the Physiotherapy Evidence Database (PEDro) scale; a valid, unidimensional measure of methodological quality of clinical trials.20 The PEDro scale has demonstrated ‘fair’ to ‘excellent’ inter-rater reliability for physical therapy-related clinical trials,21 and evidence of construct and convergent validity.22 Studies scoring < 4 are considered to demonstrate ‘poor’ methodological quality, 4 to 5 considered ‘fair’, 6 to 8 are considered ‘good’, and 9 to 10 are considered ‘excellent’.23 For studies evaluating complex interventions such as MI, a PEDro score of 8/10 is often the highest score obtainable as blinding of therapists is not possible and blinding of participants is challenging.20,21

Two reviewers (EW and NT or KH) independently applied the tool, and disagreements between reviewers were resolved through discussion or by a third reviewer (CP) as necessary. Agreement between reviewers was recorded.

Standardised mean differences (SMD) and 95 % confidence intervals (CI) were calculated from pooled post-intervention means and standard deviations (SD). Where results were not reported as means and SDs, they were manually converted according to recommendations.24 Clinically homogeneous data were synthesised using random effects models for outcomes using Revman software (Review Manager version 5.4).25 Strength of the SMD was reported according to Cohen26 where 0.2 represents a small effect, 0.5 a moderate effect, and 0.8 a large effect.26 Trial results were synthesised for common outcomes including physical activity, health-related QoL, self-efficacy, and endurance (e.g. 6 min walk test [MWT]). If a trial reported several different measures of an outcome (e.g. physical activity), the common measure between the trials was used. Statistical heterogeneity was assessed using the I2 statistic with significant heterogeneity defined as I2 >50 %.27

The GRADE (Grading of Recommendations, Assessment, Development, and Evaluations) approach was applied to determine the certainty of evidence for each meta-analysis.28 The GRADE approach considers four levels of certainty (very low to high) where randomised controlled trials begin with a high certainty and are downgraded if there are concerns about risk of bias, inconsistency, indirectness, imprecision, and publication bias.

Evidence was downgraded by one level: if there were concerns about the methodological quality (PEDro < 6)23 of the majority of trials (i.e. >50 %) (risk of bias); if the statistical test for heterogeneity demonstrated substantial heterogeneity (I2 >50 %)29 (inconsistency); if there were significant concerns that the participants, intervention, comparator, or outcome were inconsistent with usual clinical practice (e.g. participants who wouldn't usually see a physical therapist) or if there was evidence of indirect comparisons (e.g. participants who are seeing a physical therapist vs healthy controls)30 (indirectness); if the 95 % CI of the SMD was wide (i.e. >0.8) indicating imprecision (such that the lower band of the 95 % CI could indicate little or no effect while the upper band could indicate a large effect)26 or if the CI crossed the null effect threshold31 (imprecision); or if publication bias was strongly suspected, for example if the analysis included mainly small studies showing statistically significant results or inclusion of trials with industry influence.32 Evidence was downgraded two places if both criteria for imprecision were met. Single trials were considered inconsistent and imprecise, thereby providing low certainty evidence. This could be further downgraded to very low certainty evidence if there was high risk of bias.

Results and certainty of the evidence have been reported according to the GRADE guidelines on informative statements to communicate the findings of systematic reviews.33

After duplicates were removed, 1608 articles were screened on title and abstract, of which 118 underwent full text screening. After excluding 108 ineligible articles (Supplementary material, full-text exclusions), 10 publications from 9 independent randomised controlled trials met the final eligibility criteria for inclusion (Fig. 1). Findings from one trial were reported in two separate papers, one reporting initial findings34 and the other reporting 12-month follow up data.35 For the purpose of this review, these publications will collectively be referred to as the primary trial by Arkkukangas.34 Inter-rater agreement was ‘moderate’ at title and abstract screening (k = 0.44, 95 % CI 0.36, 0.52), and full text screening (k = 0.57, 95 % CI 0.34, 0.80). Two authors were contacted for the unadjusted post intervention data,36,37 and one provided further data.37

Fig. 1.

Flow of trials through the review. MI, motivational interviewing; RCT, randomised controlled trial.

(0.37MB).

The nine randomised controlled trials were published between 2012 and 2022 (Table 1) and included 909 participants. Five trials were completed in Europe (Sweden, Belgium, Spain, Denmark, Switzerland),34,36,38-39 two in Australia, ,40,41 one in Canada,42 and one in the USA.37

Three trials included participants with chronic obstructive pulmonary disease (COPD),36,38,39 two recruited older adults from the community,34,43 and single trials included participants with coronary artery disease,42 cancer,40 post hip fracture,41 and post total knee replacement.37 The mean age of participants ranged from 5940 to 83 years old34,41 and, overall, 61 % (554) were male (Table 1).

Physical therapist-delivered MI was added to usual care and compared to usual care alone in 8 of the 9 trials.34,36-37,42,43 In three of these trials, usual care consisted of minimal intervention (such as written advice or routine general practitioner visits)43,41-42 and in five trials, usual care consisted of comprehensive rehabilitation involving physical therapy with or without multi-disciplinary team care.34,36-37,39-40 In one of the trials where MI was delivered as an adjunct to comprehensive rehabilitation, a third group received minimal intervention.34

The ninth trial was the only one which included a co-intervention. This trial examined physical therapist-delivered MI with a structured walking program and usual care management compared with usual care management alone.38

Motivational interviewing was delivered in person in four trials,34,36-37,39 via telephone in three trials,43,40-41 and in combination in two trials with MI initially delivered in person and subsequently delivered by telephone.38,42 Physical therapists delivered between two38 and 1237 MI sessions over an intervention period ranging from seven40 to 52 weeks (Table 1).38,42

Four trials demonstrated the minimum recommended training requirements for MI proficiency,34,39-41 including between two40,41 and nine39 days of training and feedback from a MINT trainer (Table 2). One trial provided individual training sessions by an experienced health psychologist,36 and one provided minimal MI training led by the principal investigator (exercise physiology background).37 Three studies did not report who provided the MI training.38-43,42

Four trials39-41,43 confirmed MI fidelity with the validated Motivational Interviewing Treatment Integrity (MITI) Scale.44 Of these, two confirmed proficiency via role play with a MINT accredited MI trainer prior to the study commencement.40-41 Other fidelity measures included controlled coding during the study at a MI coding lab,34 unannounced observations by trained reviewers against a fidelity checklist,37 and review of audiotaped MI sessions by a health psychologist.36 Two trials reported the use of training, support, periodic supervision, scripts, and checklists to maintain intervention fidelity (Table 2).38,42

Eight trials measured the primary outcome of physical activity with an accelerometer or pedometer.36-42 Six reported daily steps,36,38-41 and one reported kilometres travelled over seven days.42 Four trials reported minutes of moderate to vigorous physical activity (MVPA) using an accelerometer37,40 or a physical activity questionnaire.43,42 A single trial measured physical activity subjectively via a 6-point self-reported activity scale.34 Two trials evaluated physical activity at long-term follow-up.35,39 No included trials reported other behaviour change outcomes such as those related to diet or medication adherence.

Commonly reported secondary outcomes included health-related QoL (n = 5), self-efficacy (n = 4), and endurance (n = 5). Health-related QoL was measured using the Assessment of Quality of Life Instrument,41 European Organization for Research and Treatment of Cancer QoL Questionaire-C30,40 euroQol-5 Domain Quality of Life questionnaire,43 chronic respiratory disease questionnaire (CRQ),39 the clinical COPD questionnaire (CCQ),38 and as a subset of the Knee Injury and Osteoarthritis Outcome Score (KOOS).37 Self-efficacy was evaluated using the ambulatory self-confidence questionnaire,41 physical activity appraisal inventory,40 falls efficacy Swedish scale,34 and the self-efficacy for exercise scale.43 Endurance was measured via the 6MWT.36–38,39–40.

The mean PEDro score of the included trials was 6.7 out of 10, ranging from 537 to 834,43,40 (Supplementary material, PEDro methodological quality assessment). Eight of the 9 trials demonstrated good methodological quality (PEDro score 6–8)34,36,38-41,42 with the remaining study demonstrating fair methodological quality.37 All trials fulfilled random allocation, baseline comparability, reported between-group differences, and provided point estimates and estimates of variability. Eight trials used blinded assessors34,36,38-42 and five trials followed an intention-to-treat analysis.34,38-43,40,42 There was almost perfect agreement between reviewers who assessed methodological quality (kappa 0.96, 95 % CI 0.90, 1.00).

This review and meta-analysis has been conducted in accordance with the PRISMA reporting guidelines for systematic reviews and meta-analysis.16 All trials included in this review were peer reviewed randomised controlled trials, and predominantly good-quality trials. The GRADE approach28 was applied to each meta-analysis to evaluate the level of certainty of each finding. A particular strength of this review is reflected through the robust MI eligibility criteria ensuring only trials which confirmed MI fidelity with a validated and reliable MI integrity tool were included.

A potential limitation of this review is the low number of trials and limited sample size in each meta-analysis, which may have contributed to imprecision in estimates of effect. In addition, few trials included long-term follow-up. Finally, it is possible that through the robust eligibility criteria regarding MI treatment fidelity, some trials may not have been included due to reporting, rather than the quality of MI within the trial.