The purpose of this randomized controlled clinical trial is to study the impact of the application of MI in routine follow-up visits of patients with T1D and poor metabolic control.

The hypothesis is that an intervention with MI can increase patients’ self-care behaviors, reduce HbA1c, enhance their self-efficacy and health-related quality of life, and improve the doctor-patient relationship.

MI is defined primarily by its spirit: that is, a style that nurtures the interpersonal relationship between the therapist and the patient. It explains how the process of change is formed by different stages and how people are more likely to abandon a habit at certain stages than at others. In addition, it considers that deciding to change, to commit oneself, and to take responsibility for a therapeutic process is key to achieving behavior changes. MI uses a style of collaborative communication focused on one or several objectives and pays special attention to the language of change. It is designed to strengthen personal motivation and commitment to a specific goal, eliciting, and exploring the reasons that the person has for changing, in an atmosphere of acceptance and empathy. The approach is based on patient-centered counseling, cognitive therapy, systems theory and the social psychology of persuasion. It integrates clinical skills that promote motivation, combining managerial and non-managerial elements. In sum, it is a practical and specific contribution to daily life in which language is used to influence behavior by helping individuals express their own internal motivations, and guiding conversations in such a way that people persuade themselves to change in accordance with their own values and interests.

For maximum benefit to be derived from MI, the interviewer must have received adequate training, but must also believe in it in order to be able to influence the patient and initiate a process of change. MI involves eliciting from the patient what they already know and have, rather than giving them what they lack (for instance, knowledge, or medication) (30). This focus can improve patients’ motivation, since they choose what they want to do or change, not what others want them to do or change. By developing MI skills, practitioners can help patients identify what matters to them and then use these motivations to bring about changes in their health behavior (30).

Assessing the effectiveness of MI in the context of diabetes may appear controversial, since adherence to self-care already obliges people with diabetes to carry out multiple tasks. In fact, previous studies have focused on assessing the impact of MI on glycemic control (14, 17); to date, there are no clinical trials that study its impact on adherence to self-care behaviors in adults with T1D, or evaluate the standardized application of MI by practitioners.

This randomized, controlled, single-blind, parallel group clinical trial is carried out at the Endocrinology and Nutrition service of the Insular University Hospital, Las Palmas de Gran Canaria, a reference center in its area for the care of people with T1D.

A total of 66 people with T1D have been included in the study and will be followed over a period of 16 months. Patients who met the inclusion criteria were informed of the nature of the study and invited to participate. Subsequently, they were recruited by their endocrinologist at the outpatient clinics of the Endocrinology and Nutrition Service of the Insular University Hospital. The recruitment period lasted from March to August 2019.

Diagnosis of T1D, age over 18, at least one year of disease duration, HbA1c >= 8% and/or severe hypoglycemia in the previous 6 months.

Pregnancy either in progress or scheduled in the following 12 months; any other circumstance that, in the opinion of the investigators, might interfere with the follow-up.

Patients who meet the inclusion criteria receive written information regarding the study and are invited to participate. They are also provided with more information by phone, when they are reminded that the trial will begin at the next scheduled follow-up appointment. At this first appointment, prior to the consultation, the participants sign the informed consent document and fill in a dossier that records all the clinical, sociodemographic and psychosocial variables. These variables are listed in the Measures section below.

The participants are randomly assigned to the intervention group or control group. A computer-generated randomization list (with a 1:1 ratio) is used for each endocrinologist to prepare the labels indicating the assignment of each patient, which is kept in sealed envelopes, numbered consecutively, and stored at the unit. At the time of inclusion of a patient in the study (appointment 1), the appropriate envelope is chosen and opened to show the endocrinologist the treatment assigned. The randomization is stratified by practitioner and the treatment assigned to the first participant each day is applied to all participants scheduled on that day, in order to facilitate the application of the MI and avoid contamination between treatment groups. To limit possible bias, patients are not informed of the assigned treatment. The informed consent document explains that the study will evaluate the effects of the doctor-patient communication, though without going into detail. The researcher who generated the randomization list and prepared the envelopes for the assignment of the treatment group will not be involved in the treatment or evaluation of the participants; likewise, the researchers who will rate MI compliance and analyze the data will be blind to the treatment allocation.

The intervention will consist in the application of MI by the endocrinologist at four follow-up visits, held at 4-month intervals.

These visits proceed in the same way as standard appointments, with the addition of this new clinical approach. The visits last approximately 15–20 min, in accordance with routine hospital practices.

At each visit, the practitioner applies the four processes of MI (12):

Before the first appointment and 4 months after the fourth appointment, the participants complete the questionnaires. In each of the five visits the HbA1c value will be recorded.

A sample of 25 baseline appointments 1 was videotaped in order to assess the application of MI, and serve as ongoing training for the practitioners. A similar procedure will be followed for appointment 4. The practitioners who apply MI have previously received training.

Patients randomly assigned to the control group will receive their four follow-up visits with standard medical care. The visits will be held with the same health practitioners, at the same intervals as in the intervention group, and will last a similar length of time. At each of the visits the HbA1c will be recorded, and before the first visit and 4 months after the fourth visit controls will complete the questionnaires in the same way as the intervention group.

Participants will continue with their usual follow-up routine for diabetes care. They are not asked for anything additional.

The results of the study will be presented to the scientific community at scientific meetings and in a research article. The results will also be disseminated in the local press and on the web page of the research group. Patients will also be informed of the results at the diabetes unit, once the trial has finished.

The main variable of the study is adherence to self-care, measured through a validated questionnaire, the Diabetes Self-Care Inventory-Revised version (SCI-R). The secondary variables are: HbA1c, self-efficacy (EAG), health-related Quality of Life (ViDa1), satisfaction with the doctor-patient relationship, motivation for self-care, satisfaction with self-care, degree of fulfilment of a personal goal initially proposed by the participant, and a series of sociodemographic and clinical variables listed below. The time intervals for the measurements are shown in Table 2 .

Based on the number of patients followed in the clinic and the percentage with an HbA1c above 8%, we estimated that each of the three clinical researchers involved in the trial would be able to recruit around 20 patients in 4 months. With an expected sample size of 60 participants, we would be able to detect inter-group differences of 12.8 points on the SCI-R questionnaire and of 0.85% in HbA1c, with a power of 90% and a two-sided alpha of.05 [assuming an SD of 15 points in the SCI -R and 1% in HbA1c in line with previous studies (5)]. The equivalent detectable differences with a power of 0.8 would be 11 points and 0.74% respectively (calculated with http://hedwig.mgh.harvard.edu/sample_size/js/js_parallel_quant.html). This was an approximate estimation assuming the normality of the variable.

In order to retain most of the patients during follow-up, if someone misses an appointment, the patient is contacted and the visit is re-scheduled. Specific efforts will be made for the final evaluation visit. If a physical appointment cannot be achieved, the questionnaires will be sent and completed by e-mail or, if this is not feasible, the patient with be interviewed by phone. We expect the loss to follow-up to be below 10%.

Intention to treat analysis will be performed: i.e., all randomized participants who have attended at least one intervention (or control) session will be included in the analyses. The main analysis for primary and secondary outcomes, except HbA1c, will be a generalized linear model (GLM) including the intervention arm, the baseline score of the dependent variable and other variables imbalanced at baseline as covariates. The dependent variable will be the post-intervention score. The GLM does not need a normal distribution in the dependent variable (38), although the distribution of the variables will be analyzed using de Shapiro- Wilk test to perform an adequate descriptive analysis. The choice of the link function will depend on the distribution of the variable.

For the HbA1c, a GLM for multilevel analysis will be used including the intervention arm, baseline HbA1c and other imbalanced variables in the baseline as covariates. First-level are those corresponding to each measurement along follow-up (repeated time measurements); the second level includes patients’ variables. The model will also include an interaction term between the intervention-arm and time (follow-up), allowing for differences in the intervention effect between follow-up assessments (39). The effect that identifies the intervention arm is considered fixed, whereas the intercept is considered random. An autoregressive [AR (1)] covariance structure will be used in the model.

Other variables will be added as confounders in the model, as secondary sensitivity analyses. We will first fit separate models including each confounder, one at a time. Those variables whose inclusion in the model changes the estimates’ treatment effect by at least 10% will be considered as confounders. As suggested in the CONSORT statement, decisions about confounders will not be based on P value  (40).

In the case of missing values, these will be accommodated with multiple imputation procedure in Stata 15.0 software (Stata Corporation) (41). This procedure saves cases for the analysis and can be considered an intention-to-treat analysis. Analysis under multiple imputations is valid for randomly missed data (42). A threshold of.05 will be used to define the statistical significance of those tests.

For the MIAS/EVEM 2.0 scale, the Intraclass Correlation Coefficient (ICC) will be used.