Wales Data Linkage Finds Lower MS Prevalence in Asian and Black Groups

Key Takeaways

• In a Wales population-linkage analysis of more than 2.5 million records, recorded MS prevalence was lower in Asian and Black groups than in White people.
• Recorded MS prevalence in Wales was higher in females than males.
• The highest recorded MS prevalence in Wales was in people aged 51 to 60 years.
• People with MS in Wales were less represented in the most deprived WIMD quintile and were less likely to have been born outside the UK than the overall Welsh population, while variable routine ethnicity capture still left linkage gaps.

Multiple sclerosis (MS) prevalence by ethnicity and deprivation can be difficult to characterize when routine health datasets do not capture ethnicity consistently. In Wales, that matters because population-wide estimates may differ depending on whether demographic fields come from clinical records alone or from linked census data with more complete ethnicity and birthplace information. Linking census and health-service records offered a way to describe recorded MS prevalence across social and demographic groups in Wales and set up the linked dataset used to identify people with MS.

In Wales, the Welsh Demographic Service (WDS) was linked to 2021 Office for National Statistics (ONS) census data within the Secure Anonymised Information Linkage (SAIL) Databank in Multiple Sclerosis and ethnicity in Wales: a SAIL Databank study. The merged records supplied census age, sex, birthplace, ethnicity, and Welsh Index of Multiple Deprivation (WIMD) fields, and the Nicholas, 2024 algorithm was used to identify people with MS. Across 3.1 million ONS records and 3.2 million WDS records, linkage produced 2.5 million unique records, including 5,934 people with MS. That linked Wales population was the basis for the recorded prevalence estimates.

Recorded prevalence in Wales was lower in Asian and Black groups than in White people. The reported prevalence was 58.2 per 100,000 (95% CI, 41-80.3) in Asian people and 126.8 per 100,000 (95% CI, 80.3-190.2) in Black people, compared with 240 per 100,000 (95% CI, 233.9-246.3) in White people. Asian and Black comparisons were reported as significantly lower than White prevalence, and these differences reflect recorded prevalence within the linked Wales dataset.

Recorded prevalence in Wales differed across age, sex, and ethnicity; the abstract also reports that people with MS were underrepresented in the most deprived WIMD quintile compared with the Welsh population and describes a birthplace difference between those groups. Prevalence was 325.2 per 100,000 (95% CI, 315.6-335) in females versus 131.7 per 100,000 (95% CI, 125.3-138.3) in males, and it peaked at 445.9 per 100,000 (95% CI, 424.6-467.9) in people aged 51 to 60 years. The abstract reports a birthplace difference between people with MS and the Welsh population (96.4% vs 93.9%), although its wording is unclear, and 14.9% were in the most deprived WIMD quintile versus 18.8% of the Welsh population (p<0.001). Overall, recorded prevalence varied across demographic and social strata in Wales.

The authors concluded that linked population-wide datasets can provide an overview of MS prevalence by age, sex, ethnicity, and social deprivation in Wales, and that better linkage between datasets is still needed. In Wales, the linkage approach offered a descriptive overview of recorded prevalence patterns while leaving improved dataset linkage as the main unresolved issue named by the authors.

Clinician Questions

How were ethnicity and birthplace defined in the Wales MS prevalence analysis?

Ethnicity and birthplace were taken from 2021 Office for National Statistics census data after linkage with the Welsh Demographic Service, allowing the Wales analysis to use census age, sex, birthplace, ethnicity, and Welsh Index of Multiple Deprivation fields rather than relying only on routine clinical recording.

Which population does the Wales MS prevalence analysis actually describe?

The analysis describes people in Wales who were captured through linked Welsh Demographic Service and 2021 census records, with MS identified by the Nicholas, 2024 algorithm. The findings therefore apply to recorded prevalence within that linked population rather than to causal explanations for differences between groups.

What unresolved issue did the authors identify after linking census and health-service data for MS in Wales?

The authors said further work is needed to improve linkage between datasets, and they noted that routine ethnicity data collection in clinical datasets is variable. In this Wales analysis, that was the main open issue left after linkage was used to describe recorded MS prevalence patterns.