This study was an unmasked, cluster RCT. The randomized, parallel group trial was performed at 50 community pharmacies in Japan. Diabetic subjects were randomly assigned to either the lifestyle intervention group (IG) or the standard care group (CG). The subjects were followed for 6 months. The randomization was stratified using matched cluster size. The cluster was a community pharmacy that included patients with diabetes every day because of the need for prescription drugs. Community pharmacies were listed by three chain pharmacies and we allocated 50 pharmacies to the IG and CG. These community pharmacies were located in fifteen prefectures in Japan. Pharmacists in the attending pharmacy recruited 1 - 7 patients with diabetes by using a leaflet and poster when they visited the pharmacy.

The recruitment started in March 2011 and August 2011. The intervention for participants started in April 2011, and the last case was completed in December 2012. Patients were recruited by pharmacists by using standardized brochures and posters. Each pharmacy was asked to recruit 6 patients. They provided consent forms to conduct this study among patients. Eligibility was verified by requesting the most recent clinical data measured in a hospital or clinic within the previous 3 months for HbA1c by using a diabetic note.

The pharmacist who attended this study had been trained for 6 hours on a brief version of motivational interviewing (MI) skills in community pharmacies. This training program was a modified version of the MI (Table 1). This program of mostly than 3 minutes was developed for community pharmacies. This program helped the pharmacists’ skill and confidence that was useful for advice-giving. Generally speaking, readiness, importance, and confidence are the key factors to change behavior [12] . Our program focused on the improvement of patients’ confidence because we thought that all patients in this study had already been getting ready mentally and felt deeply the importance of changing their lifestyle. To improve patients’ confidence, pharmacists were trained on how to communicate with patients to give approval for the measures they have already taken in their daily life. This short version of MI skills based on empowerment includes three steps: 1) Using an open question; 2) Setting goals with patients 3) Closing with encouraging. This training program included three cases on snacks,

exercise, and drinking. These cases were discussed in small groups with role-playing, and were completed within 3 minutes. We prepared fourteen types of brochures that were passed to the patients in the IG at every visit.

Patients in the CG were given basic explanation on medication and their HbA1c levels were checked at every visit. The CG was sent newsletters with general information, except for that on healthy lifestyle, and just included a simple greeting in order to prevent drop-out from the study.

Patients in the IG were given pedometers when they attended this study. In addition, their HbA1c levels were checked and were advised about a healthy lifestyle by a pharmacist when they visited the pharmacy for their prescription drugs. Pharmacists who attended this study were given fourteen types of brochures regarding 1) healthy diets and snacks, 2) exercise tips for improving your HbA1c levels, 3) the goal of blood tests, and 4) reducing alcohol consumption. Patients were given one of these and pharmacists provided short explanations on how to use these leaflets. During every patents visit to the pharmacy, the pharmacists motivated their patients to improve their confidence for making change to their lifestyle within a three minute consultation.

OMRON HJ-205IT was the pedometer used. The IG was sent newsletters about healthy lifestyles including diet and physical activity every month during this study period.

In the first visit, a pharmacist in the IG and a patient discussed goal setting and the way of life style changes. Pharmacists provided information using a brochure and the patient decided the method. At the second visit and after every month, pharmacists checked patient’s HbA1c levels, whether patients had achieved their goals, and patients’ barrier to life style changes. In addition, they discussed new goal setting at every visit.

According to previous intervention studies in community pharmacies that used self-monitoring of blood glucose (SMBG), there was 0.5% reduction in HbA1c levels in T2D patients [13] [14] . Therefore, it was assumed that there was a 0.5% reduction in HbA1c levels in the intervention group because of the non-use of SMBG in the study. The present study was designed to detect a HbA1c 0.5% reduction in participants due to the intervention. Thus, the sample size required was a cluster of 90 with an average of 2 patients with a type 1 error of 5%, and a 90% power (beta = 10%) at the two-tailed 5% significance level, that allowed for a withdrawal rate of 10%.

The main outcome was change in HbA1c levels. The other outcomes were medication change, medication adherence, life style change, and change in the attitude and knowledge about diabetes. To analyze the effects of the intervention on changes as well as body mass index (BMI), IPAQ [15] , Medication adherence: Modified Morisky Scale 5-point response options [16] and DTSQ [17] , multilevel mixed effects regression models using Stata and mixed and meqrlogit command were used, after adjusting for covariates.

Medication data was collected from the prescription data. HbA1c data were checked on diabetic notes are usually used in Japanese hospitals and clinics for diabetics. These diabetic notes have not only physical and blood test data including HbA1c but also the stage of nephropathy and retinopathy. Diabetics, physicians and pharmacists can determine patients’ medical data from these notes. Height and weight were also checked it. BMI was calculated as weight (kg)/height² (m²).

In chain-pharmacies, an explanatory meeting was undertaken for this study after pharmacists agreed to participate. In addition, we made a list of over 90 prospective pharmacies. After allocation, the pharmacists of the IG were trained by the pharmacy MI program for 6 hours in a single day. After this program, the pharmacists in both groups recruited participants and collected informed consent data. When patients agreed, they send the coded data on the application form by fax. Each time the patients visited, the pharmacists sent data the same way for a total of 6 months. Patients filled out the guest questionnaire at the beginning and after 6 months.

All data are presented as mean ± SD. Primary analyses used the last observation carried forward (LOCF) for missing data. Comparisons between the groups were made with a two-tailed unpaired t-test. A two-tailed paired t-test was used to analyze the difference within groups between the baseline and the 6-month result. A p value less than 0.05 was considered statistically significant. Descriptive statistics (means, standard deviations, and percentages) were included as crude values.

Internal consistency reliability scores were assessed for diabetic knowledge using Cronbach’s alpha, where >0.7 was considered as minimum and >0.9 was desirable. All data analyses were conducted with STATA V.13. We took into account the clustering effect.

Simple randomization was performed through the Center of Randomization at the Division of Prevention Medicine of Clinical Research Institute, National Hospital Organization Kyoto Medical Center in Japan.

The study could not be blinded for participants and community pharmacists and health providers because of the nature of the intervention. The study team was not involved in the assessment of eligibility. The collected data was blindly analyzed by an independent statistician.

The IG had 26 pharmacies and the CG included 24 pharmacies. Pharmacies in the IG recruited 106 patients, of which 90 patients completed the study. Pharmacies in the CG recruited 57 patients, of which 42 patients completed the study. Figure 1 shows a flow diagram for this study. Table 2 shows Baseline characteristics of individual and cluster level.

Changes in HbA1c levels in the IG were significantly greater than changes in the CG. In addition, the score of diabetic knowledge and the number of diabetes medications was significant by the intervention. There were no differences in other healthy lifestyle indexes between the two groups (Table 3). After adjusted for the clustering

Missing data was not included in the analysis. ^aPharmacies was classified by the number of patients a day. Small: 0 - 50, Middle: 51 - 100, Large: 101-. ^bRetinopathy was checked through patients’ self-report. ^cWe counted how many kinds of diabetes medication were being taken by each person.

^aHbA1c, glycated hemoglobin. ^bBMI, Body mass index. ^cIPAQ, the International Physical Activity Questionnaire. ^dMedication adherence, a modified, 6-item. Morisky Medication Adherence Scale (MMAS). ^eDTSQ, Diabetes Treatment Satisfaction Questionnaire. ^fMedication type of drugs, Oral hypoglycemic agent(OHA) was classified into 6 groups. Sulfonylurea (SU), Biguanide (BG), Dipeptidyl Peptidase-4 Inhibitor (DPP-4I), α-glucosidase Inhibitor (a-GI), Thiazolidinediones (TZ), and Glinide. ^gKnowledge, measured by 10 true or false questions. ^hHealthy lifestyle, Lester Breslow: The “Seven Healthy Habits” 1) Get a good night’s sleep of seven or eight hours. 2) Exercise 30 minutes at a time, several times a week. 3) Eat moderately to maintain weight in relation to height. 4) Eat breakfast every day. 5) Eat regularly, whether that’s two meals a day, three or five. 6) Don’t drink at all or drink moderately. 7) Don’t smoke. ^*Adjusted B, Adjusted for the clustering effect.

effect the IG had significantly improved HbA1c (IG: −0.6% ± 0.9% vs. CG: −0.2% ± 0.9%; p = 0.021).

No adverse events occurred in either group.