The practice of medicine is at heart an exercise of collecting, filtering, summarizing, managing, analyzing, and acting upon information. This information comes directly from the patient's narrative history, but also from family and caretakers, and other providers. It also is derived from diagnostic interventions, including the physical examination, laboratory tests, radiologic exams, and procedures. Combined with reference knowledge about physiology, pathology, pharmacology, and other basic science disciplines, the physician makes an expert assessment of the patient's conditions and risks, and then recommends an action plan. Information about this plan must be communicated and coordinated with a larger team and with the patient and their family, executed, and then information about how the patient responds fed back in order to make adjustments over time. If this flow of information is compromised or hampered at any point in this cycle, then the potential for quality and safety problems emerges. Given this intense information-rich environment that the clinician must navigate, especially in the inpatient setting, it is clear that the judicious application of information technology (IT) can greatly empower the hospitalist in providing high quality and safe patient care; and conversely, that injudicious application of IT can promote errors and adverse outcomes.
Information technologies that impact patient safety and quality of care can be grouped into three major categories. First, there are the interventions that impact care as it is delivered in real time—this class is generally called decision support because it involves clinicians while they are making diagnostic and therapeutic decisions. The second class of information technologies, broadly known as surveillance, monitors the immediate downstream care processes to detect anomalies and unintended consequences so that effective corrective action may be taken quickly. The last general category of IT for safety and quality is data mining, or retrospective analysis of large repositories of data, such as patient registries, electronic health records, and administrative databases in order to detect meaningful patterns and signals that may help inform ways to improve the previous two types of information systems. Data mining overlaps with classical epidemiological health services outcomes research. This classification of information technologies is analogous to the distinctions between primary, secondary, and tertiary modes of disease management, which may be more familiar to clinicians.
As defined above, decision support is any type of information system that intends to direct, guide, or alter medical decision making as it occurs in real time. This may occur via passive delivery of knowledge, such as quick access to online digital references, drug compendia, clinical calculators, or differential diagnosis tools. In this case, the user must voluntarily choose to activate the service. This type of decision support is usually well received by busy clinicians, because the clinician is motivated to get a question answered. However, passive decision support does not address latent information needs, or knowledge deficits unknown to the clinician.
Decision support may also occur via active knowledge delivery, such as alerts to avoid unsafe or undesired ...