Diabetes group education for insulin and GLP 1 users tracks with lower HbA1c in 15 patient report

HAVANT, ENGLAND. Monthly diabetes group education for adults on insulin or GLP 1 drugs was linked to lower HbA1c among regular attendees at one Hampshire GP practice, a preprint posted on medRxiv on 9 November reports. Only 15 patients took part over 12 months, and the six who came to four or more sessions saw the largest mean fall, 12.4%. The report has no control group, has not been peer reviewed and was funded by Eli Lilly UK, Embecta UK and Novo Nordisk UK.

Who the programme reached at Homewell Practice

Homewell Practice serves around 15,000 patients in Havant, including the Leigh Park estate, which the authors describe as home to some of the most deprived wards in England. The practice cares for about 1,106 people with type 2 diabetes. According to the paper, 291 of them used injectable treatment: 187 were prescribed insulin, 136 a GLP 1 receptor agonist, and some took both. Many had poor glucose control, frailty or a high risk of hospital admission. Diabetes specialist nurse Jane Sennett and clinical pharmacist Emily Bond led the team, with Kristin Veighey of the University of Southampton as a co-author. They ran 12 monthly diabetes group education sessions in a local community hall between May 2022 and April 2023.

Invitations first went to patients who met at least two of three risk markers: moderate or severe frailty while using mixed insulin, an HbA1c of 75 mmol/mol or higher on any injectable, or a QAdmissions score of 20% or more. Only seven patients met all three markers and 50 met two, the paper reports. Uptake was low, so the team opened the sessions to anyone meeting one marker in June 2022, and to every adult on injectable therapy in August 2022. The authors say the broadest invitation brought the most engagement. Sessions had no fixed curriculum. Staff covered glucose targets, insulin adjustment, low blood sugar, food and complications, and let attendees steer the discussion.

What the HbA1c numbers show and do not show

The preprint sorts results by how often people came. Mean HbA1c fell 12.4% in the six patients who attended four or more sessions, 5.6% in the three who attended three, and 2.4% in the one patient who attended two. The five people who came once saw mean HbA1c rise 1.8%. The paper presents these as percentage changes and does not list baseline values for attendees. The authors read the pattern as a link between steady attendance and better glucose control. Every group is tiny, however, and the two session figure reflects a single person. No statistical tests are reported, which the authors put down to the small sample and the service evaluation design.

Mean HbA1c fell 12.4% in the six patients who attended four or more sessions, 5.6% in the three who attended three, and 2.4% in the one patient who attended two.

Several design limits cut against a causal reading. HbA1c was measured at routine reviews, so follow up timing differed from patient to patient. There was no comparison group of people on injectables who did not attend. The team also did not record medication changes made during the year, the paper says, so it cannot separate the effect of diabetes group education from the effect of treatment changes. Patient comments praise regular medication adjustments in the group setting. People who keep coming to a monthly class may also differ from those who drop out in motivation, health or support at home. Housebound patients could not attend at all. The authors frame the project as a test of feasibility and acceptability, not proof of effect.

Peer support and the patients who stayed away

Nine of the 15 attendees returned anonymous paper questionnaires. All nine said the sessions were useful, four calling them very useful and five somewhat useful, and all nine reported a better understanding of their diabetes. Eight said they would recommend the group to others. Eight of the nine named sharing experiences with other people living with the condition as the most valued part, ahead of topics such as understanding HbA1c, treating hypoglycaemia and injection technique, and the authors single out peer support as the main draw. Attendance averaged 6.4 people per session, from four in August 2022 to 12 at the final meeting in April 2023. Carers or spouses sometimes came along. No patient attended all 12 sessions.

The report's most practical finding concerns who did not come. The authors write that patients with the greatest need, including people with mental health conditions and several long term illnesses, were the least likely to join groups and seemed to prefer one to one appointments. They list privacy worries, low confidence in groups, competing social pressures and practical barriers as possible reasons, but say these were not studied. That gap matters because the original invitation list was built around exactly those high risk patients. The authors propose a hybrid of diabetes group education and individual reviews for higher need patients. They also suggest that several small local groups may suit people better than one large programme covering a wider area.

Industry funding and what comes next

The project was paid for by an education and development grant shared equally by Eli Lilly UK, Embecta UK and Novo Nordisk UK, according to the paper. The money covered the hall, clinician and administrator time, refreshments and stationery. The authors state that the funders had no role in design, analysis or the decision to publish, and that the clinical team controlled session content. The medRxiv listing also records past honoraria from Novo Nordisk and other companies for two authors. The full text of the medRxiv preprint carries the standard warning that it has not been certified by peer review and should not guide clinical practice. Because it was run as a service evaluation, the project did not need ethics committee approval, the authors say.

Next steps are modest. The authors call for evaluation across a Primary Care Network to test whether the model can scale, and say the learning has been shared with local practice teams and the area's diabetes specialist nurse service. They also recommend dedicated data tools so future rounds can track more outcomes in a consistent way. For readers following outcomes on peptide drugs, this report is a different kind of evidence from the real world tirzepatide results reported from England this autumn. It also sits far from the adult trials behind the weekly insulin icodec approval for children. It is a single practice audit of 15 people, and its numbers should be read that way.

Conclusion

This diabetes group education report offers a small, uncontrolled signal: among 15 adults on insulin or GLP 1 drugs at one deprived Hampshire practice, those who came most often had the largest reported HbA1c falls. Medication changes were not tracked, follow up timing varied and the paper is not peer reviewed. Its clearest lesson concerns reach, since the high risk patients it was built for were the least likely to attend. Whether the model works at scale is untested, and the authors call for a Primary Care Network evaluation next.