People with severe mental illness (SMI) experience significant health inequalities, but long-term trajectories of physical multimorbidity and their association with mortality remain poorly understood.

To identify distinct trajectories of physical comorbidity burden within people with SMI and evaluate their associated mortality risks.

This retrospective cohort study included 64 962 patients with a first recorded SMI diagnosis in primary care, followed for up to 10 years. Joint latent class modelling was applied to jointly model annual Charlson Comorbidity Index (CCI) measurements and all-cause mortality, identifying latent trajectory groups with distinct comorbidity burden progression and survival profiles. Subsequently, multivariable Cox regression was used to estimate adjusted mortality risks across identified trajectories, and multinomial logistic regression was used to identify baseline clinical indicators of trajectory membership.

Over a median follow-up of 4.75 years (interquartile range 2.50–8.45 years), there were 5932 deaths (9.1%). Four distinct trajectory groups were identified: ‘increase stable’ (2.6%), ‘persistent high’ (5.1%), ‘persistent low’ (89.9%) and ‘rapid increase’ (2.5%). Compared with the persistent-low reference group, adjusted hazard ratios for mortality were 4.11 (95% CI 3.75–4.50) for the increase-stable group, 2.83 (95% CI 2.62–3.07) for the persistent-high group and 11.33 (95% CI 10.54–12.18) for the rapid-increase group. Older age (≥65 years) and the use of medications for hypertension, diabetes and obstructive airway disease were strong indicators of belonging to the higher-risk trajectories, with the multinomial regression model showing good discrimination of trajectory membership (overall area under the curve 0.78).

Patients with SMI in primary care exhibit heterogeneous trajectories of CCI-defined comorbidity burden, with each trajectory associated with a different mortality risk. Routinely collected primary care indicators may support early identification of patients likely to follow higher-risk trajectories; however, prospective validation remains essential before implementing risk-stratified interventions in practice.