Use of medical data, also known as electronic health records, in research\nhelps develop and advance medical science. However, protecting patient\nconfidentiality and identity while using medical data for analysis is crucial.\nMedical data can be in the form of tabular structures (i.e. tables), free-form\nnarratives, and images. This study focuses on medical data in the free form\nlongitudinal text. De-identification of electronic health records provides the\nopportunity to use such data for research without it affecting patient privacy,\nand avoids the need for individual patient consent. In recent years there is\nincreasing interest in developing an accurate, robust and adaptable automatic\nde-identification system for electronic health records. This is mainly due to\nthe dilemma between the availability of an abundance of health data, and the\ninability to use such data in research due to legal and ethical restrictions.\nDe-identification tracks in competitions such as the 2014 i2b2 UTHealth and the\n2016 CEGS N-GRID shared tasks have provided a great platform to advance this\narea. The primary reasons for this include the open source nature of the\ndataset and the fact that raw psychiatric data were used for 2016 competitions.\nThis study focuses on noticeable trend changes in the techniques used in the\ndevelopment of automatic de-identification for longitudinal clinical\nnarratives. More specifically, the shift from using conditional random fields\n(CRF) based systems only or rules (regular expressions, dictionary or\ncombinations) based systems only, to hybrid models (combining CRF and rules),\nand more recently to deep learning based systems. We review the literature and\nresults that arose from the 2014 and the 2016 competitions and discuss the\noutcomes of these systems. We also provide a list of research questions that\nemerged from this survey.\n