Nurse-led psychological interventions to improve diabetes control: Assessing competencies

How to Measure Motivational Interviewing Fidelity in Randomized Controlled Trials: Practical Recommendations

“…Some studies report the average MITI results for different interventions [25,29,31,36], others for different (groups of) practitioners [39,43], and some for individual practitioners over time [28,43]. In one study, the MITI results of two practitioners were weighed based on the number of participants each practitioner had counseled to obtain an overall MI fidelity for the study intervention [26].…”

“…Considering that practitioners will experience changes during this time, and that their level of MI skill may fluctuate [35], it is important to select sessions from different points in time throughout the entire intervention period, to get a complete and representative picture of practitioners' MI skill. In particular, if practitioners receive ongoing supervision or additional training, improvements in MI proficiency could be expected [25,36]. Moreover, the client also influences the session; therefore it is desirable to include sessions with different clients, who are at different stages of change, in the sample.…”

“…Most previously conducted RCTs report that 4%-32% of sessions were double-coded (e.g. [38] (32%; n = 11); [32] (4%; n = 6); [39] (20%; n = 19); [27] (8%; n = 115); [25] (30%; n = 54); [40] (10%; n unknown); [22] (25%; n unknown) [36] (27%: n = 15)). However, in some studies, all sessions were coded by at least two of the coders [28,31,41,42].…”

“…In the literature to date, the inter-rater reliability reported for different MITI variables has varied tremendously, ranging from 'poor' to 'excellent' (see below for thresholds) (e.g. [25,36]). …”

“…For example, it seems that coders more easily code the global Direction variable and behavior counts Giving Information and Reflections (e.g. [27,36]) reliably. The global variable Autonomy/Support and the MI Adherent and MI Non-Adherent behavior counts seem considerably more difficult for coders to reach agreement on.…”