Influence of comorbid diabetes mellitus on outcomes in multiple sclerosis: an English population-based matched cohort study

Technical Summary: Influence of Comorbid Diabetes Mellitus on Outcomes in Multiple Sclerosis

Problem Statement
Multiple sclerosis (MS) is a chronic neurological condition where increasing life expectancy, driven by disease-modifying therapies, has shifted clinical focus toward the management of comorbidities. Diabetes mellitus (DM) is a major global contributor to morbidity and mortality, yet existing evidence regarding its comorbidity with MS is limited and inconsistent. Previous studies have reported mixed findings on mortality associations, and a 2024 Danish cohort found no difference in DM incidence between MS patients and controls. Furthermore, no prior studies have examined the impact of incident (newly diagnosed) DM on hospitalization metrics in MS, nor have they explored how social determinants of health (ethnicity, deprivation, urbanity) modify these outcomes. This study addresses these gaps to determine if incident DM is more common in MS populations and how it influences hospitalization and mortality.

Methodology
The study utilized a population-based matched cohort design using linked English primary and secondary care data from 2000 to 2023.

• Data Sources: Data were drawn from the Clinical Practice Research Datalink (CPRD) Gold and Aurum datasets, linked to Hospital Episode Statistics (HES) for admissions and the Office for National Statistics (ONS) for mortality records.
• Cohort Definition: The study included 9,010 adults diagnosed with MS and 78,121 matched controls without MS. Matching was performed on age, sex, and general practice, with controls assigned the same index date as their matched MS case.
• Exclusion Criteria: Individuals with pre-existing DM (prevalent cases) were excluded to focus on incident disease. Exclusions also applied to those with missing birth years, invalid registration dates, or insufficient follow-up data (<2 years post-MS diagnosis).
• Exposure and Outcomes:
  • Exposure: Incident DM (suspected or confirmed) defined by diagnostic/management codes or prescriptions.
  • Outcomes: All-cause hospitalization (number of episodes and duration) and all-cause mortality.
• Statistical Analysis:
  • Incidence: Poisson regression models adjusted for demographics and socioeconomic factors to calculate incidence rate ratios (IRR).
  • Hospitalization: Negative binomial regression for episode counts and linear regression for duration, adjusted for covariates and practice-level random effects.
  • Mortality: Cox proportional hazards models with DM modeled as a time-varying exposure to account for immortal time bias.
  • Inequalities: Interaction terms were tested to determine if ethnicity, Index of Multiple Deprivation (IMD), or rural/urban classification modified the associations.
  • Sensitivity: Analyses were restricted to MS diagnoses post-2002 (McDonald criteria implementation) and to confirmed DM cases only.

Key Contributions and Results
The study provides robust, large-scale evidence on the burden of incident DM in MS populations over a mean follow-up of 13.2 years.

1. Increased Incidence of DM in MS: Individuals with MS were significantly more likely to develop incident DM compared to matched controls. The adjusted incidence rate ratio (aIRR) was 2.26 (95% CI: 1.96–2.61), indicating that people with MS are more than twice as likely to be newly diagnosed with DM.
2. Impact on Hospitalization: Among people with MS, the development of incident DM was associated with:
   • Higher Hospitalization Rates: An adjusted IRR of 1.82 (95% CI: 1.47–2.28) for the number of episodes.
   • Longer Duration: A median hospital stay of 18 days for those with incident DM versus 4 days for those without (adjusted β = 0.53, p < 0.001).
   • Note: While DM increased hospitalization in both groups, the relative increase in episode frequency and duration was statistically stronger in the control group than in the MS group.
3. Impact on Mortality: When modeled as a time-varying exposure, incident DM was associated with a 46% higher hazard of all-cause mortality in people with MS (HR = 1.46, 95% CI: 1.17–1.82). A similar 49% increase was observed in controls. In the main MS model without time-varying exposure, the association was not statistically significant (HR = 1.21, p = 0.10), but sensitivity analyses restricting to post-2002 diagnoses and confirmed DM showed significant mortality risks.
4. Health Inequalities: Interaction analyses revealed that the relationship between DM and outcomes varied by context:
   • Among people with MS, the association between DM and increased hospitalization episodes was stronger in rural areas.
   • Among controls, the association between DM and hospitalization was stronger in White ethnic groups.
   • Among controls, the association between DM and mortality was stronger in the least deprived groups.

Significance and Claims
The paper claims that individuals with MS face a significantly elevated risk of developing diabetes compared to the general population, which subsequently drives higher rates of hospitalization and mortality. The study highlights that while DM negatively impacts both populations, the specific manifestation of these risks (e.g., hospitalization frequency) differs between MS patients and controls, potentially due to higher baseline healthcare utilization in MS.

The authors emphasize that the findings underscore the necessity for equitable screening, prevention, and management of diabetes in people living with MS. They suggest that the observed variations by geography and ethnicity indicate that health inequalities may exacerbate the burden of comorbid DM in specific subgroups. The study concludes that effective DM management has the potential to reduce hospitalizations and extend lives in the MS population, particularly when tailored to address geographical and demographic disparities. The authors maintain a modest tone regarding subgroup analyses, noting that small sample sizes in stratified groups (e.g., by DM type) require cautious interpretation.