Semi-automated review of medical records for public health investigations

During a public health emergency investigators may need to identify and contact affected residents. Medical record or log review has long been employed to identify potential cases of disease in public health investigations and is traditionally performed manually by trained staff either from the medical facility or the public health department. In 2010 we initiated development of a Rapid Data Query System which intended to, in partnership with local hospitals, design and implement methods for semi-automated queries of patient records and electronic transfer of identified records to the health department. Unlike electronic laboratory reporting, this system would be capable of identifying individuals based on clinical symptoms alone, which is especially important for illness for which no laboratory tests exist or for which a test has not been ordered. This system is also differentiated from syndromic surveillance in that it seeks to retrieve case information, including patient identifying variables, for specific but limited time periods based on a list of symptoms and diagnoses tailored for each investigation. In implementing this system we encountered a number of challenges including: (1) establishing partnerships with the medical community and institutions; (2) overcoming competing priorities at the medical facilities and within the health department; (3) working within tight budget constraints; and (4) assessing and adapting to the varying technological capabilities at the partner medical facilities. In this presentation I will discuss the obstacles we encountered and how they were overcome, and reflect upon the utility of developing a new Rapid Data Query system in the context of recent regulatory requirements for electronic medical record development.