Multimorbidity is a major problem for patients and health services. However, we still do not know much about the common trajectories of disease accumulation that patients follow. We apply a data-driven method to an electronic health record dataset (CPRD) to analyse and condense the main trajectories to multimorbidity into simple networks. This analysis has never been done specifically for multimorbidity trajectories and using primary care based electronic health records. We start the analysis by evaluating temporal correlations between diseases to determine which pairs of disease appear significantly in sequence. Then, we use patient trajectories together with the temporal correlations to build networks of disease accumulation. These networks are able to represent the main pathways that patients follow to acquire multiple chronic conditions. The first network that we find contains the common diseases that multimorbid patients suffer from and shows how diseases like diabetes, COPD, cancer and osteoporosis are crucial in the disease trajectories. The results we present can help better characterize multimorbid patients and highlight common combinations helping to focus treatment to prevent or delay multimorbidity progression.