Evaluation of Health and Lifestyle Tool

Considerable evidence suggest that lifestyle changes can prevent or delay the onset of type 2 diabetes, and self-care behaviors largely determine HbA1c. Modifiable lifestyle factors have been established as key drivers of disease onset, progression, and prognosis, motivating the use of "lifestyle as medicine". Despite this, a substantial number of patients with type 2 diabetes (T2D) have difficulties attaining adequate glycemic control.

Digital health tools are increasingly incorporated into diabetes care, and have the potential to improve both behavioral and clinical outcomes on a broad basis. However, low levels of uptake, reduced user engagement over time, and low acceptance among patients, raise concerns about their effectiveness. Knowledge on how to design interventions aimed at inducing change by promoting motivation and personal engagement, as well as a better understanding of patient interaction with digital health interventions, could help overcome these challenges and inform the development of a novel and effective health support.

The objective or this study is to evaluate a new web-based tool, developed at the University Gothenburg, Sweden, that aims to support patient autonomy and motivation to make sustainable lifestyle changes.

The investigators will test the hypothesis that individuals with type 2 diabetes in routine care who have access to the tool get improved HbA1c relative to baseline as compared with individuals with type 2 diabetes on usual care not exposed to the tool.

The study is an investigator-initiated single-center study conducted at Scania University Hospital, Sweden, and will follow participants over three years.

Participants with type 2 diabetes will be recruited by sending letters with study information to patients with type 2 diabetes in the region. If HbA1c is 52 mmol/mol or above participants are included and attend study visits at Scania University Hospital every three months during the first year and every sixth month during the following two years for blood sampling and physical examination.

During the first 3 months, only half of the participants get access to the tool immediately and the other half remain on usual care (randomized who get access or not). After 3 months, all participants get access to the tool and used it during an open-label observation period of up to three years.

Study visits: Every physical study visit lasts appr. 20 minutes and includes measurements of length and weight, blood pressure, and estimations of fat and muscle mass by bioimpedance. Fasting blood samples are taken for analysis of HbA1c and other cardiometabolic proteins. Participants will not receive any counselling or lifestyle advice at the study visits. Technical problems are referred to the study coordinator, who may also respond to requests to clarify content in a general manner without providing personal advice. Patients are followed by their ordinary physician throughout the study, i.e. the tool is provided on top of ordinary anti-diabetic treatment and healthcare contacts.

The tool: The tool is web-based and used via a computer or mobile phone. It is used at each individual's preferred pace but participants are recommended to login at least every other week. Every round the participants choose a themes (out of appr. 80 possible covering e.g. food, exercise, stress, self-reflection aspects), which takes appr. 15-30 minutes to complete. Participants then reflect on the content and how it could be implemented in daily life. When returning for next round participants are asked to reflect on any changes done since last time. There is no interaction between individual participants.

Statistics: The study has two primary endpoints. One primary endpoint is change of HbA1c between participants with and without access to the tool during the first 3-month period. Patients lost to follow-up will not be included in the analysis.

The standard deviation of change of HbA1c is 6 mmol/mol over 3 months. With 80% power at alpha=0.05, 142 participants are needed to each randomization arm to detect a significant difference between the groups, assuming that the true treatment effect of the intervention is 2 mmol/mol over 3 months.

Enrollment will continue until the required number of participants are reached, accounting for those lost to follow-up during the randomization period. The participants use the tool at their own pace but are within the tool recommended to complete a theme at least every other week to enable implementation of changes and maintain an awareness of the tool and its content.

As another primary endpoint, investigators will during the open-label observation period compare change of HbA1c at one year relative to baseline between patients using the tool as recommended (at least every other week) and matched controls on usual care (1:2 ratio between exposed and controls). The controls are matched on gender, age, BMI and HbA1c from existing clinical registries in the region.

The standard deviation of DeltaHbA1c is 7 mmol/mol over one year. With 80% power at alpha=0.05, 24 participants using the tool as recommended and 48 matched controls are needed to detect a significant difference between the groups, assuming that the true treatment effect of the intervention is 5 mmol/mol.

Subanalysis of MOD patients A data-driven cluster analysis of 9,000 diabetes patients has been performed in the ANDIS registry in Sweden based on six variables measured at diagnosis: GAD antibodies, age, BMI, HbA1c (reflecting long-term blood glucose), HOMA2-B (reflecting insulin secretion) and HOMA2-IR (reflecting insulin resistance). Four clusters of T2D patients were highlighted, each with different characteristics and risk of complications. One of these cluster, MOD (Mild Obesity-related Diabetes), is characterized by high BMI and insulin resistance but relatively well-preserved insulin secretion. It has been suggested that patients with MOD characteristics would benefit particularly from lifestyle changes. The investigators therefore hypothesize that patients with MOD characteristics would benefit particularly well from the intervention. As a subanalysis the investigators will analyse the interaction between MOD/non-MOD subgroup and exposure to the intervention using a linear model with a term for the cluster variable, a term for the exposure to the intervention and an interaction term.