Critical Care Informatics
- 1 Introduction
- 2 Perceived Need
- 3 Specialties
- 4 Informatics in Clinical Practice
- 5 Research
- 6 Education
- 7 References
Intensive care medicine involves care of the sickest patients across the age spectrum, from neonates through pediatrics and into adulthood . Patients typically require invasive monitoring or advanced organ support, including invasive ventilation, dialysis, blood pressure support and extracorporeal membranous oxygenation (ECMO). The role of clinical informatics is expanding in the critical care environment, with opportunities to advance clinical care, research and training.
The practice of intensive care has a great need for the application of informatics principles. In a fast-paced clinical environment, errors are unfortunately frequent and often preventable. One study estimated the rate per 1000 patient-days of adverse events at 80.5, and preventable adverse events at 36.2 . Care in the ICU is also expensive, accounting for approximately 1/7th of hospital care and 4.1% of national health expenditures . Monitors and frequent observations generate a vast amount of data, but this data is often not cataloged, stored and mined appropriately to be considered valuable “information” . A major task of informaticists in the ICU will be to aggregate this data into information that can be used by clinicians at the bedside , however there are many hurdles to this task including costs, technologic limitations and physician buy-in .
The practice of critical care occurs by specially trained providers in different locations, depending on the patient. Much is shared among these specially-trained physicians, however much is different. Below we briefly review the most common subspecialties of intensive care practitioners as well as challenges specific to their specialty.
Medical & Surgical Critical Care
Adult critical care providers are often divided into medical and surgical practitioners. Adult medical critical care providers are routinely trained as pulmonary and critical care physicians, working both in an ICU and with adult pulmonology patients. Surgical practitioners are also specially trained in critical care. An additional subspecialization in surgery includes trauma intensive care medicine, the care of patients with traumatic injuries. These providers routinely work in Surgical ICUs or Trauma ICUs.
Pediatric Critical Care
The pediatric critical care unit is an intensive care specifically for patients between birth and approximately 18 years of age, and practitioners of this specialty are called "Pediatric Intensivists" . Pediatric intensivists are subspecialty-trained pediatricians who have undergone special training in the care of critically ill children. Despite the commonly-held belief that "children are just little adults", studies have shown repeatidly that critically ill children will demonstrate decreased mortality when cared for by pediatric intensivists as compared to care in an Adult ICU .
Neuro Critical Care
Neuro critical care involves intensive care with a focus on injuries of the nervous system, including stroke, traumatic brain injury and status epilepticus. Typically these units are staffed with providers trained in both neurology and critical care medicine, bringing together the right specialists for the care of these patients.
Not necessarily unique to neuro critical care is the practice of multi-modal monitoring, or the integration of several streams of data into a single monitoring system. For example, important to the care of a seizing patient with a non-invasive airway is both the respiratory monitoring (from a ventilator) and the electro-encephalographic (EEG) signals. Integration of these signals for prediction of patient course is an area of active research.
Neonatalogy is the specialty of medicine tasked with the care of the neonate, often premature or with other complicating medical conditions requiring intensive care . Common conditions include congenital disorders requiring surgery such as congenital diaphragmatic hernia, neonatal seizures, neonatal respiratory distress syndrome and extreme prematurity. Neonatal intensive care units (NICUs) are often designated as "closed" units, meaning they accept only patients delivered at that hospital, or "open" units which accept patients from other hospitals or admitted through the emergency department. In the US, NICUs are classified by the American Academy of Pediatrics as level I, II, III or IV depending on the level of service and specialization they provide 
Maternal & Fetal medicine
One unique informatics challenge with NICUs is the care of linked patients - namely, the mother and baby. Much of the history of the fetus comes from the narrative of the delivery, as well as the maternal medical history. Therefore, careful linking of these charts is helpful for the care of the fetus-turned-patient at the time of delivery. Patient "registration" must occur quickly such that orders may be safely placed on the new patient.
The combination of critical care nursing and informatics creates a specization of the field of Nursing Informatics. Nursing informaticists are vital as links from the front lines of nursing, who perform most of the day-to-day tasks in the ICU, to information systems analysts who design, maintain and analyze health systems .
Informatics in Clinical Practice
The clinical practice of intensive care medicine requires many of the same informatics princples that are required of inpatient hospital medicine, including:
- Computerized physician order entry (CPOE)
- Clinical data capture and documentation
- Clinical decision support (CDS)
- Interoperability & Ancillary System Integration
- Secondary use & analytics
However, many of these issues are magnified when we consider the speed of decisions, complexity of patients, number of medications, variety of ancillary devices and magnitude of data. Below we discuss some of these challenges and some solutions to overcome these challenges.
The intensive care unit is a technology-rich environment. Patients are often exposed to numerous monitoring modalities, including for example cardio-respiratory, neurologic, and skin - surface monitoring. This data is often streamed to the electronic health record (see Device Integration below), however the question becomes, what happens to the data next? New applications are being developed to pull this information from the medical record and help clinicians make decisions about patient care, however there have been few randomized clinical trials of these technologies . In addition to data re-use and analytics, biomedical devices such as ventilators require specialized knowledge for operation. Within intensive care units are often specialists whose role is focused on the operation of just one type of device (e.g. ECMO perfusionist).
In the technology-rich ICU environment, the technical aspects of device integration are vital to data extraction and subsequent analysis. There are numerous bedside devices which export data elements, including physiologic monitors, ventilators & infusion pumps as examples. Beyond the physical (hardware) connections needed to bring these devices together, they must speak and understand a common language. Typically HL7 is the communication framework used, however EHR venders also provide specific connection options. Data storage capabilities are also important to consider - for example, continuous vital sign data sampled at 50 Hz from a bedside monitor will generate 180,000 data elements per patient per hour for a single vital sign. When multiplied by several vital signs (heart rate, pulse oximetry, continuous blood pressure, capnography) this can easily exceed 1 million data elements per hour. Storage and analysis of this information will require high performance computing environments even before predictive algorithms can be developed .
Several companies have developed integration devices for bringing together data from multiple devices into a single location; these are just examples and not an exhaustive list:
Tele-Medicine & Remote Monitoring
Tele-medicine in a broad sense enables the practice of medicine in geographically-disparate locations . The ICU environment is an appealing one for telemedicine because of the vast array of monitoring systems limit the need for physically examining the patient - although certainly do not completely remove this need. Hospital systems are using telemedicine systems to evaluate and care for critically ill patients, and research in this field is ongoing 
Despite the promises made by the lay press, not all of intensive care medicine can or should be replaced by a camera and a remote television screen . Hands-on care by nurses, physicians and other skilled professionals will be essential, even as we move towards more tele-medicine solutions.
The ICU is an alarm-rich environment with numerous physiologic monitors causing excess noise and alarm fatigue on practitioners . Every bedside monitor, ventilator, IV pump, dialysis machine and ECMO pump has the ability to produce audible alarms of different frequencies and durations, which must be responded or acknowledged by a practitioner. Many of these are vital to the care of the patient (e.g. ECMO pump failures, ventricular fibrillation detected) but separating the signal from the noise (pun intended) is challenging (e.g. 'Tachycardia' when the heart rate falls above the alarm parameters programmed into the monitor). Researchers continue to search for more effective alarm management strategies to limit alarm burdon while maintaining safety .
Critical care informatics research is an exciting field, bridging specialties of engineering, computer science, medicine and sociology to define the future of critical care. Research is focusing on bringing novel technologies, data analytics and predictive models from the computer to the bedside. Several notable research groups that have published extensively in this area are highlighted below, however this is by no means a comprehensive list:
- Clinical Informatics in Intensive Care (Mayo Clinic, Rochester, MN) - Directed by Vitaly Herasevich and Brian Pickering Link
- Integrative Neuromonitoring and Critical Care Informatics Group (MIT, Cambridge, MA) - Directed by Thomas Heldt Link
- Laboratory for Computational Physiology (MIT, Cambridge, MA) - Directed by Roger Mark Link
- This group is responsible for maintenance of PhysioNet, a collection of physiologic data available for analysis Link
- Virtual PICU (UCLA, Los Angeles, CA) - Directed by Randall Wetzel Link
- T3 Tool (Sick Kids Hospital, Toronto, CA) - Directed by Peter Laussen Link
Databases for Research
In addition to promoting research, some of these groups have established datasets for public and hospital research use. Among these is perhaps the most popular critical care dataset, the Medical Information Mart for Intensive Care or MIMIC dataset which can be found here . This freely-available public dataset consists of de-identified health data associated with over 58,000 admissions of 40,000 critical care patients admitted to Beth Isreal Deaconess Medical Center from 2001 throught 2012. In addition to the data provided in the core dataset, the newest version of the MIMIC dataset includes a subset of physiologic waveform data on approximately 3,000 patients. It has been a rich source of data for testing and refining machine learning algorithms on critical care patients.
From a quality improvement standpoint, the VPS group (linked here) founded by Randall Wetzel (of the Virtual PICU, a distinct research entity) collects, analyzes and distributes safety and quality data to member institutions. This data is also available for research questions, with VPS acting as a central repository for questions.
Fellowship training is available in critical care and clinical informatics. The Accredication Council for Graduate Medical Education (ACGME)  certifies fellowship training programs that meet criteria for academic rigor as well as institutional support and monitoring. A list of applicable fellowships can be found below:
- Pulmonary & Critical Care Fellowships
- Neonatal - Perinatal Fellowships
- Pediatric Critical Care Fellowships
- Clinical Informatics Fellowships
Societies, Conferences & Meetings
Academic societies relevant to the practice of critial care informatics are listed below in alphabetical order, with appropriate links:
- American Medical Informatics Association (AMIA): Academic society of informaticists Link
- AMIA Intensive Care Informatics Working Group Link
- American Pediatric Society: Research-focused pediatric soceity that hosts PAS meeting Link
- Society of Critical Care in Medicine: Academic society of adult and pediatric intensivists Link
- Wikipedia: Intensive Care Medicine (Accessed 2017-10-18) Link
- Rothschild JM et al. The Critical Care Safety Study: The incidence and nature of adverse events and serious medical errors in intensive care. Crit Care Med. 2005 Aug;33(8):1694-700 PMID: 16096443
- Halpern NA & Pastores SM. Critical care medicine in the United States 2000-2005: an analysis of bed numbers, occupancy rates, payer mix, and costs. Crit Care Med. 2010 Jan;38(1):65-71 PMID 19730257
- Schmidt, J. Michael. Lecture: Biomedical Engineering and Informatics Applications in Critical Care. Columbia University Medical Center (Accessed: 2017-10-18) Slides
- Pinsky MR & Dubrawski A. Gleaning knowledge from data in the intensive care unit. Am J Respir Crit Care Med. 2014 Sep 15;190(6):606-10 PMID 25068389
- Martich GD, Waldmann CS & Imhoff M. Clinical informatics in critical care. J Intensive Care Med. 2004 May-Jun;19(3):154-63 PMID: 15154996
- Wikipedia: Pediatric Intensive Care Link
- Czaja, Angela S. Children Are Not Just Little Adults ... Pediatric Critical Care Medicine: February 2016 - Volume 17 - Issue 2 - p 178–180 Link
- Wikipedia: Neonatal Intensive Care Unit Link
- American Academy of Pediatrics, Committee on Fetus and Newborn. Levels of neonatal care. Pediatrics. 2012;130(3):587–597. doi:10.1542/peds.2012-1999 PMID 22926177
- Jastremski CA. Nursing informatics. Issues for critical care medicine. Crit Care Clin. 1999 Jul;15(3):563-76. PMID: 10442263
- Adhikari N & Lapinsky SE. Medical informatics in the intensive care unit: overview of technology assessment. J Crit Care. 2003 Mar;18(1):41-7. PMID: 12640613
- Jacono FJ, De Georgia MA, Wilson CG, Dick TE, Loparo KA. Data Acquisition and Complex Systems Analysis in Critical Care: Developing the Intensive Care Unit of the Future. Healthcare Engineering. 2010.
- Telemedicine | Medicaid (Accessed 2017-10-18) Link
- Slamon N & Greenspan JS. Considerations When Implementing a Pediatric Telemedicine Program. J Pediatr. 2016 Dec;179:5-6 PMID: 27634628
- The Economist. Prescription for the future: How hospitals could be rebuilt, better than before. April 8, 2017 Link
- Paine CW et al. Systematic Review of Physiologic Monitor Alarm Characteristics and Pragmatic Interventions to Reduce Alarm Frequency. J Hosp Med. 2016 Feb;11(2):136-44 PMID: 26663904
- Pickering BW et al. Novel Representation of Clinical Information in the ICU: Developing User Interfaces which Reduce Information Overload. Appl Clin Inform. 2010 Apr 28;1(2):116-31 PMID: 23616831
- Johnson AE et al. MIMIC-III, a freely accessible critical care database. Sci Data. 2016 May 24;3:160035 PMID: 27219127
- Accreditation Council for Graduate Medical Education: Website (Accessed 2017-10-18) Site
Submitted by Adam Dziorny