Clusters of long-term conditions by demographic groups and associated risk of hospitalisation

Background

The study aimed to identify clusters of multiple long-term conditions (MLTCs) across socioeconomic and ethnic groups and examine their association with unplanned hospitalisation.

Methods

We used Clinical Practice Research Datalink (CPRD) Gold, a UK primary care database. Patients aged 18+ years on 01/01/2018, with ≥2 conditions were included. Candidate MLTCs were identified based on clinical knowledge, and 15 conditions with a prevalence ≥5% were taken forward for analysis. We applied latent class analysis to divide patients into distinct and clinically meaningful clusters with similar MLTC patterns stratified by area deprivation (i.e., most and least deprived) and ethnicity (i.e. Asian, Black, and White). Cox regression was used to assess the association between cluster membership and unplanned hospitalisation risk within one year.

Results

From 3,146,578 patients, 237,156 (7.5%, of which 42.3% male) with ≥2 MLTCs were included in the analysis. Four clusters were identified in the least deprived areas, and three in the most deprived areas and across all three ethnic groups. “High-MLTC” (defined by a high prevalence of most conditions) and “mental health” clusters were observed across all subgroups. An “inflammatory” cluster was identified only in the least deprived areas and in the White ethnic group, whereas a “cardiometabolic” cluster was observed solely in the most deprived areas and among Black and Asian ethnic groups. Across all subgroups, the “high-MLTC” cluster showed the highest unplanned hospitalisation risk, which was doubled among patients in the most deprived areas and in minority ethnic groups (hazard ratio from 2.06 to 2.15).

Conclusions

This work highlighted clusters of MLTCs that have implications for healthcare policy. These clusters across socioeconomic and ethnic groups can be used to inform targeted and personalised interventions directed towards people with MLTCs to support patients and reduce disparities.

Key messages

• Distinct clusters of multiple long-term conditions exist across socioeconomic and ethnic groups.

• A high-burden of long-term conditions cluster was concentrated among socioeconomically deprived and ethnically minoritised groups and was associated with higher risk of unplanned hospitalisation.