Background Increasing inequalities in accessibility to primary care has generated medical deserts. Identifying them is key to target the geographic areas where action is needed. An extensive definition of primary care has been promoted by the World Health Organization: a first level of contact with the health system, which involves the co-presence of different categories of health professionals alongside the general practitioner for the diagnosis and treatment of patients. Previous analyses have focused mainly on a single type of provider while this study proposes an integrated approach including various ones to define medical deserts in primary care. Method Our empirical approach focuses on the first point of contact with the health system: general practitioners, proximity primary care providers (nurses, physiotherapists, pharmacies, laboratories, and radiologists), and emergency services. A multiple analysis approach was performed, to classify French municipalities using the information on the evolution and needs of health care accessibility, combining a principal component analysis and a hierarchical ascending classification. Results Two clusters of medical deserts were identified with low accessibility to all healthcare professionals, socioeconomic disadvantages, and a decrease in care supply. In other clusters, accessibility difficulties only concern a part of the health supply considered, which raises concern for the efficiency of primary care for optimal healthcare pathways. Even for clusters with better accessibility, issues were identified, such as a decrease and high needs of health care supply, revealing potential future difficulties. Conclusion This work proposes a multi-professional and multi-dimensional approach to medical deserts based mainly on an extensive definition of primary care that shows the relevance of the co-presence of various healthcare professionals. The classification also makes it possible to identify areas with future problems of accessibility and its potential consequences. This framework could be easily applied to other countries according to their available data and their health systems' specificities.