Background <p>Recent studies have addressed the problem of the complex classification and discovery of multimorbidity patterns; however, to date there are still many doubts about the conceptualization of multimorbidity and its adequate measurement due to the need to work with very large samples and longitudinal designs that allow us to study the dynamics of multimorbidity patterns and the social contexts from which they emerge.</p> Methods <p>The present study makes use of a sample of 341,343 patients over 50 years of age residing in southern Spain with the aim of analysing the trajectories of multimorbidity patterns that are linked to different socioeconomic areas.</p> Results <p>Employing Hidden Markov Models for the study of changes in the latent states of multimorbidity profiles, our analysis demonstrates significant associations between multimorbidity patterns and the socioeconomic gradient of residential areas. Moreover, we provide additional evidence that enables a characterisation of transitions from initial disease patterns to multisystemic profiles of chronicity (or complex multimorbidity) that offer a worse prognosis of health and quality of life for the patient.</p> Conclusions <p>Our study adds relevant and contextualized information on the evolution of multimorbidity patterns that can serve as a basis for further research and policies aimed at the prevention and control of multimorbidity from an early age.</p>