Materials and methodology
The study was conducted in our university medical college hospital in South India. The hospital has over 9000 employees among whom are doctors, nurses, allied health professional, administrative and technical support staff.
The randomization code was developed by the department of biostatistics using SAS software. The block randomization was done with the blocks of 2 (30%), blocks of 4 (30%) and blocks of 6 (60%). The codes were given to the principal investigator in sequentially labeled sealed opaque envelopes to randomly allocate patients (1:1) to individually tailored mobile phone messaging in addition to standard care or to a control group that received standard care alone.
The list of existing healthcare workers with type 2 diabetes was obtained from the staff health service. Some participants were newly diagnosed during the course of the study and were also recruited. The study protocol was clearly explained, following which a written informed consent was taken to participate in the study trial. The mobile phone numbers of the participants was obtained and saved. The trial recruitment started in 2015 January and ended in 2015 August and follow-up continued until 2016 March. Trial was funded by hospitals’ internal research fund; they had no role in trial execution, data collection or analysis.
Participants
Patients aged 18 years or older working as staff in the hospital, diagnosed with type 2 diabetes mellitus. Those who were pregnant had other endocrine disorders associated with diabetes, steroid-induced diabetes (on steroids currently) or who refused consent were excluded.
Baseline demographic details, habits, comorbidities, physical examination including body mass index, waist circumference, diet, exercise and HbA1c were recorded. Waist circumference was measured by trained, certified staff using an anthropometric measuring tape at a horizontal plane that is 1 cm above the navel. Dietary intake was assessed by the primary investigator. Details regarding the amount in kilocalories of food intake were assessed at each visit. Calculation of energy intake for individual food items was done by the National Institute of Nutrition guidelines for India. The participant was asked to self-categorize the knowledge of their diabetic diet as good, fair, poor and not sure. Physical activity was measured using the Paffenbarger Survey (online supplementary file). The physical activity assessment was done by the primary investigator. These questions allow for calculation of self-reported leisure time, physical activity and kilocalorie expenditure per week.
Intervention
Educational messages on diet control and exercise were based on the transtheoretical model of human behavior, which has five stages—precontemplation, contemplation, preparation, action and maintenance.
Messages had educational material regarding healthy eating habits and exercise and these messages were sent twice weekly. The messages were scheduled via an automatic calendar in a way that each subject in the intervention arm received 15 educational messages per month (online supplementary file panel 1).
The messages were delivered by a commercial service provider (Unicel Technologies, India). Messages were provided only in English language. Each message contained fewer than 160 characters and 60–80 messages were created for transtheoretical model stage and were sent cyclically, such that participants did not receive the same message in a 3-month period (on the basis of them receiving 15 messages per month). The assumption being that the participants moved from a pre-action stage to an action stage. The timing (05:00–08:00 hours or 17:00–20:00 hours) of mobile phone messaging were tailored to the participant’s preferences. Participants were informed of the mechanisms for delivery of mobile phone messages. The subjects in the intervention arm received these messages for the first 3 months of the study, following which they also got only the standard care.
In addition to this intervention all patients received standard of care treatment for diabetes.
Control arm
Those in the control arm received standard of care treatment for diabetes. This included dietary advice initially by a dietician and thereafter by the treating physician at every visit. Exercise advice and medication advice are given at scheduled visits, which are usually at every 3–6 months. Usually HbA1c is tested every 3 months.
Outcome
The primary objective of the study was assessed at 6 months. This was the feasibility of this computer-generated SMS-based follow-up system for healthcare workers with diabetes, hence the primary outcome was assessed by a questionnaire which included questions regarding the SMS-based system. The secondary outcome included the glycemic control (HbA1c level), patient satisfaction and adherence to lifestyle modifications, assessed with the help of a questionnaire (online supplementary file panel 3).
Sample size calculation and statistical analysis
As per the study by Ramachandran et al, there was 8% decreased in the incidence of diabetes in a 2-year follow-up using similar SMS messages. We hypothesized a 10% reduction in HbA1c in 6 months. The SD of HbA1c was assessed as 1.5, and with alpha and beta errors of 5% and 20%, respectively; with two-sided test, we need to study nearly 140 subjects in each arm. Assuming 10% drop out, we had decided to study 160 subjects in each arm.
Data were collected in a Clinical Research Form (CRF) designed for the study; data were then entered in EpiData V.3.1 and exported to SPSS V.22 for analysis. All analyses were done using intention-to-treat principle.