Diabetes Drugs Often Continue Until Death in Dementia Patients, Landmark Study Finds

Millions of older people worldwide live with both dementia and type 2 diabetes, two conditions that together create one of the most difficult balancing acts in modern medicine. A new multinational study has now revealed how rarely that balance is redressed at the end of life, showing that glucose-lowering medications frequently continue to be supplied to people living with dementia right up until their final weeks. The research, published in European Geriatric Medicine, analysed the records of more than 16,500 people who died in the United Kingdom and Australia, and found that between 56.5 and 82.8 per cent of individuals were still receiving diabetes medications within the last three months of life.

The study, led by Darshna Goordeen of Monash University’s Centre for Medicine Use and Safety together with collaborators at University College London and the University of North Carolina, is the first to examine glucose-lowering medication use in the final year of life specifically among people living with dementia in these two countries. Using linked hospital and administrative data, the team identified 10,088 individuals in the United Kingdom and 6,436 in Australia who had been diagnosed with both type 2 diabetes and dementia at least twelve months before their deaths between July 2013 and June 2018. In both cohorts, women made up the majority of participants, and a large proportion were aged 85 or older, with 55.8 per cent of the UK cohort and 37.7 per cent of the Australian cohort falling into this oldest age group.

The clinical rationale for stepping back from intensive diabetes treatment in advanced dementia is well established. Dementia guidelines in both the United Kingdom and Australia recommend individualising treatment according to changing goals when life expectancy is limited, and diabetes guidelines in both countries advise less strict glycaemic control, with higher glycated haemoglobin targets, for people living with dementia and limited life expectancy. Advanced dementia commonly brings poor nutrition, irregular eating and an inability to manage complex treatment regimens. These factors raise the risk of medication-related harm through administration errors and through a reduced ability to recognise, report or treat hypoglycaemia, the dangerous drops in blood sugar that glucose-lowering drugs can provoke. Deprescribing, defined as the planned and supervised process of reducing doses or stopping medications that may be causing harm or no longer offer benefit, is therefore widely endorsed for this population.

Despite this consensus, the evidence suggests overtreatment remains common. Earlier research found that among 161 patients with dementia and diabetes, 54 per cent were potentially overtreated, defined as having an HbA1c below 7 per cent despite mild functional or cognitive impairment or multiple comorbidities. In aged care homes, rates of potential overtreatment with glucose-lowering medications have ranged from 5 to 86 per cent. Against this backdrop, the new study set out to measure the prevalence of medication supply in the final 3, 6, 9 and 12 months of life and to identify which patient factors were associated with discontinuation in the six months before death.

The technical execution of the study drew on two of the world’s most comprehensive health data systems. In the United Kingdom, the researchers used the Clinical Practice Research Datalink, a primary care database sponsored by the Medicines and Healthcare products Regulatory Agency, linked to Hospital Episodes Statistics and mortality records from the Office for National Statistics. In Australia, they analysed linked data from the Victorian Admitted Episodes Dataset, the Pharmaceutical Benefits Scheme, which records every subsidised medication dispensed, and the National Death Index. Metformin emerged as the most commonly supplied medication class in both countries during the last year of life, used by 36 per cent of the UK cohort and 41 per cent of the Australian cohort, followed by insulin at 19.5 and 33.8 per cent respectively, sulfonylureas at 21.6 and 27.9 per cent, and newer agents including DPP-4 inhibitors, SGLT-2 inhibitors, GLP-1 receptor agonists and thiazolidinediones at 11.2 and 15.1 per cent.

The headline finding concerns insulin, the medication class most likely to be discontinued in both countries. Among people supplied insulin in the last twelve months of life, use fell by 25.6 per cent in the United Kingdom and 43.5 per cent in Australia in the final three months. In the UK, 97.1 per cent of those prescribed insulin in the last year received it within nine months of death, 90.8 per cent within six months and 74.4 per cent within three months. In Australia the decline was steeper, with supply falling from 92.5 per cent at nine months to 79.6 per cent at six months and just 56.5 per cent in the final quarter. The researchers suggest this pattern reflects growing clinical awareness of the particular hazards insulin poses in dementia, including difficulty with administration, diminished recognition of hypoglycaemic symptoms and reduced oral intake. Notably, this contrasts with Danish and American studies that reported stable or even increased insulin use near the end of life in older adults generally, though those studies did not focus specifically on people with dementia.

For metformin, the picture was more encouraging and revealed a striking alignment with clinical guidelines. Chronic kidney disease, a condition that makes metformin risky because of the small chance of lactic acidosis, was associated with higher odds of metformin discontinuation in the last six months of life in both countries, with an odds ratio of 1.05 in the UK and 1.60 in Australia. This suggests that clinicians are, at least in part, tailoring decisions to individual comorbidities rather than applying blanket rules. In Australia, male sex was associated with lower odds of metformin discontinuation, while for sulfonylureas, being aged 74 or younger was associated with higher odds of discontinuation compared with those aged 85 and older in both countries. For the newer medication classes, discontinuation was minimal, possibly because of lower initial supply rates, better tolerability or the absence of clear deprescribing recommendations in guidelines, even though recent work has flagged risks of malnourishment, dehydration, urinary incontinence and hypotension with SGLT-2 inhibitors and GLP-1 receptor agonists in frail older people.

Overall, among people supplied a glucose-lowering medication within six to twelve months of death, use fell by between 9.4 and 26.9 per cent in the final six months, with discontinuation occurring more frequently in Australia than in the United Kingdom. The authors interpret this as evidence of inadequate deprescribing overall. The gap between recommendation and practice may have several roots. Qualitative studies cited by the researchers indicate that clinicians generally feel comfortable deprescribing diabetes medications when harms clearly outweigh benefits, but some express hesitancy when a medication was initiated by a specialist. Healthcare professionals also report needing confidence that they are acting in the patient’s best interests and that the health system supports personalised care. Advance care planning, promoted in both countries as an integral component of dementia care, is intended to ensure that treatment preferences, including medication management, reflect individual values, yet the persistence of medication supply suggests these conversations are not translating into changed prescriptions.

The study’s strengths lie in its scale and its deliberate cross-national design. By applying identical methods to two large, population-level datasets, the researchers produced findings that can be directly compared, something single-country studies rarely achieve. The team acknowledged limitations, including differences in how age and comorbidities were captured across the two systems, the use of prescription records in the UK versus dispensing records in Australia, and a much lower documented prevalence of chronic kidney disease in the UK cohort, at 35 per cent versus 64.5 per cent in Australia, which may partly reflect under-ascertainment of kidney disease in UK diagnosis codes. The researchers also could not assess the appropriateness of medication use at the individual level or examine clinical parameters such as HbA1c measurements, meaning that appropriateness of discontinuation was inferred from the balance of risk and benefit at the population level.

The implications reach well beyond the two countries studied. As populations age and the overlap between dementia and diabetes grows, the findings point to a pressing need for structured, patient-centred deprescribing pathways embedded in routine dementia care. The authors highlight tools such as the 4S pathway for realigning diabetes regimens in older adults, multidisciplinary medication reviews that give clinicians the support to initiate deprescribing, and electronic health record alerts that could prompt timely medication management reviews for people with dementia in aged care settings. Until such systems become routine, the study’s central message stands: for many people living with dementia, diabetes treatment continues until the very end of life, and the challenge now is to ensure that comfort, rather than glycaemic numbers, guides the final prescription.

Subject of Research: Glucose-lowering medication use and deprescribing in the last year of life among people with dementia and type 2 diabetes in the United Kingdom and Australia

Article Title: Diabetes medication use in the last year of life in people living with dementia in the United Kingdom and Australia

Article References: Goordeen, D., Hopkinson, O. K., Reeve, E., Niznik, J. D., Lau, W. C. Y., Wei, L., Ilomaki, J., & Bell, J. S. (2026). Diabetes medication use in the last year of life in people living with dementia in the United Kingdom and Australia. European Geriatric Medicine. https://doi.org/10.1007/s41999-026-01608-6

Image Credits: AI Generated

DOI: 10.1007/s41999-026-01608-6

Keywords: dementia, type 2 diabetes, deprescribing, end-of-life care, insulin, metformin, polypharmacy, geriatrics, hypoglycaemia, pharmacoepidemiology, chronic kidney disease, medication safety