Incident Diabetes Linked to More Hospitalisations and Death in Multiple Sclerosis

People with multiple sclerosis who developed diabetes after their MS diagnosis had more hospital admissions, longer hospital stays and higher subsequent mortality than people with MS who did not develop diabetes, according to a population-based cohort study using linked primary and secondary care records from England. The analysis included 47,247 people with MS and 458,951 matched controls without MS, with a mean follow-up of 15.2 years. It found that MS was associated with a modestly higher incidence of diabetes than in controls, while the relationship between incident diabetes and mortality was broadly similar in the two groups.

The researchers analysed anonymised records from the Clinical Practice Research Datalink Gold and Aurum datasets, which cover approximately 60 million patients, or about 20% of the UK population, through more than 2,400 general practices. These records were linked to Hospital Episode Statistics for admissions and outpatient and emergency care, as well as Office for National Statistics death registrations. Adults with MS were matched to up to 10 controls without MS on general practice, sex and year of birth, and controls were assigned the same index date as their matched MS case.

MS was identified through two or more diagnostic or management codes, or one MS code accompanied by an appropriate prescription. Diabetes was defined as suspected or confirmed type 1, type 2 or unspecified diabetes, with or without antidiabetic medication; people whose only record was gestational diabetes were excluded. A six-month wash-out period after the MS diagnosis date was used to reduce the chance that hospitalisations were directly related to the diagnostic process. The models adjusted for age, sex, ethnicity, area-level deprivation, rural or urban residence, smoking status and practice-level random effects, with additional sensitivity analyses restricting the MS group to diagnoses made from 2002 onward and considering confirmed diabetes only.

Over the study period, 2,119 people with MS, or 4.48%, developed suspected or confirmed diabetes, compared with 21,720 controls, or 4.73%. The crude incidence rate was 3.14 per 1,000 person-years for people with MS and 3.00 per 1,000 person-years for controls. After adjustment, the incidence rate ratio was 1.16, meaning the observed rate among people with MS was about 16% higher than among matched controls, with a 95% confidence interval of 1.11 to 1.21. The association was similar when diabetes was analysed with a cause-specific Cox model and when baseline hypertension was added to the adjustment set. Diabetes type was available for only 35.79% of people with incident diabetes, but the increased incidence was observed in both recorded type 1 and type 2 groups.

Diabetes was associated with substantially greater use of hospital services among people with MS. In unadjusted models, those with incident diabetes had 2.83 times as many hospitalisation episodes as MS patients without incident diabetes. The association was smaller after adjustment, but remained clear: the adjusted incidence rate ratio was 1.52, with a 95% confidence interval of 1.42 to 1.64. The result was unchanged when the analysis was restricted to MS diagnosed from 2002 onward, when only confirmed diabetes was counted, or when hypertension was included as an additional covariate. Hospitalisation duration was also longer among MS patients with incident diabetes. The median cumulative duration of hospital admissions was 12 days, with an interquartile range of 0 to 83 days, for those with diabetes, compared with 2 days, with an interquartile range of 0 to 22, for those without diabetes. The adjusted regression coefficient was 0.89, and the association persisted in the sensitivity analyses.

Controls showed an even stronger relationship between incident diabetes and hospitalisation. Among controls, the adjusted incidence rate ratio for the number of admissions was 2.36, and the adjusted association with hospitalisation duration was also significant. When both groups were examined together, the diabetes-related excess in hospitalisation episodes was stronger among controls than among people with MS, with an interaction estimate of 0.43 and a 95% confidence interval of 0.40 to 0.47. The difference for hospitalisation duration was likewise stronger in controls. The researchers suggested that admissions related to MS itself may dilute the relative effect of diabetes among MS patients. Another possible explanation is closer monitoring of people with MS, although that possibility was not directly tested, and the data did not show a corresponding reduction in mortality.

Diabetes was also associated with higher all-cause mortality among people with MS. In the primary model, incident diabetes was associated with an adjusted hazard ratio of 1.08 for death, with a 95% confidence interval of 1.01 to 1.15. When diabetes was modelled as a time-varying exposure, the hazard ratio was higher, at 1.41, with a 95% confidence interval of 1.31 to 1.49. This model was intended to reduce immortal time bias by accounting for the time before diabetes developed. The association remained significant in analyses limited to MS diagnosed from 2002 onward, when diabetes was restricted to confirmed cases, and after further adjustment for hypertension. In controls, incident diabetes was associated with mortality as well, with an adjusted hazard ratio of 1.25 in the primary analysis and 1.69 when diabetes was modelled as time-varying.

The mortality association did not differ significantly between people with MS and controls. The interaction between MS status and incident diabetes was not statistically significant in the main mortality model, nor when diabetes was treated as a time-varying exposure. Thus, the study does not show that diabetes has a uniquely stronger or weaker effect on death in people with MS than in comparable people without MS. It does show that, in both populations, the emergence of diabetes was associated with a higher subsequent risk of dying. The findings should therefore be interpreted as observational associations rather than proof that diabetes caused the excess hospital use or mortality.

Interaction analyses found limited evidence that ethnicity, deprivation or rural–urban residence modified the overall relationships in the MS cohort. Within people with MS, however, the association between diabetes and the number of hospitalisation episodes was stronger among people living in rural areas. Among controls, the relationship was stronger in White than non-White groups for some hospitalisation measures, and diabetes-related mortality was stronger among people in the least deprived groups and in rural areas. These subgroup findings came from smaller numbers and require caution. The study authors noted that regular healthcare contact among people with MS could increase detection of asymptomatic diabetes and partly contribute to the higher observed incidence in the MS group. Baseline comorbidity, physical activity, obesity, immune dysregulation and shared autoimmune susceptibility for type 1 diabetes were discussed as possible explanations, but the mechanisms were not established.

Several limitations constrain the interpretation and generalisability of the results. The database relies on routinely collected clinical information, so some conditions may be under-recorded, ethnicity was missing for some participants, and people who are less engaged with healthcare services may be underrepresented. Diabetes type was unrecorded for 64% of people with incident diabetes, and haemoglobin A1c was not reliably available for analysis. People who develop diabetes may differ from those who do not in ways that were not fully measured, so residual confounding cannot be excluded. The MS index date was the first recorded code rather than the date of clinical onset, and disease duration before that point was therefore unknown. Finally, the cohort was predominantly White and based in England, while the epidemiology of MS and diabetes differs by geography and ethnicity, limiting extrapolation to other populations. The authors concluded that further studies are needed and that equity-informed diabetes screening, prevention and management may benefit people with MS.

Subject of Research: Associations between incident diabetes, hospitalisation and mortality in people with multiple sclerosis

Article Title: Incident Comorbid Diabetes and Outcomes in Multiple Sclerosis: An English Population‐Based Cohort Study

Article References: Lau, Y., Zabihi, S., Hartmann, M., Mathlin, G., Banerjee, S., Marouf, E., Hadley, C., Cooper, C., & Dobson, R. (2026). Incident Comorbid Diabetes and Outcomes in Multiple Sclerosis: An English Population‐Based Cohort Study. Annals of Clinical and Translational Neurology, Article acn3.70548. https://doi.org/10.1002/acn3.70548

Image Credits: AI Generated

DOI: 10.1002/acn3.70548

Keywords: multiple sclerosis, incident diabetes, hospitalisation, mortality, population-based cohort, English health records, health inequalities, observational study, type 2 diabetes, Incident, Comorbid, Diabetes