Delays in Transfer to the ICU
A Preventable Adverse Event?
Article first published online: 28 JAN 2003
Journal of General Internal Medicine
Volume 18, Issue 2, pages 155–156, February 2003
How to Cite
Kaboli, P. J. and Rosenthal, G. E. (2003), Delays in Transfer to the ICU. Journal of General Internal Medicine, 18: 155–156. doi: 10.1046/j.1525-1497.2003.21217.x
- Issue published online: 28 JAN 2003
- Article first published online: 28 JAN 2003
Intensive care units (ICUs) evolved nearly 50 years ago as an organizational structure to improve quality of care for critically ill patients by concentrating specialized interventions and personnel in a centralized location.1 By the late 1960s, nearly all U.S. hospitals had at least 1 ICU.2 ICUs provide life-sustaining treatments and monitoring that are often unavailable on typical hospital wards. However, outcomes of ICU care are dependent on patients’ severity of illness at the time of ICU admission.3,4 Thus, from a theoretical perspective, patients transferred to the ICU earlier in the course of a downward clinical spiral should fare better than patients transferred later.
Among ICU admissions, patients who are transferred from hospital wards may be at particularly high risk of adverse outcomes.5 In the current issue of JGIM, Young et al. report the findings of a well-designed observational study that examined the impact of delays in ICU transfer in this high-risk group.6 The authors found that patients who were transferred to the ICU from the wards of a community hospital more than 4 hours after a marker of clinical instability was first noted (i.e., slow transfers) had a nearly 5-fold higher adjusted risk of death than patients transferred earlier (i.e., rapid transfers). This increased risk of death is striking and is similar in magnitude to the difference found in a prior study comparing mortality in a 30-year-old and a 90-year-old patient admitted to the ICU.7
A further significant finding of the study was that physicians responsible for slow-transfer patients were less likely to be notified within 2 hours after a marker of clinical instability was first noted, and only 23% of slow-transfer patients received a physician bedside evaluation within 3 hours after a marker was noted, compared to 86% of rapid-transfer patients. The authors also found that slow-transfer patients suffered greater declines in physiological function during the period prior to transfer and were sicker at the time of ICU transfer. While causation is often difficult to establish from observational studies, these findings provide plausible evidence for a causal relationship between delays in transfer and increased mortality.
Taken together, the findings have implications for the organization and delivery of hospital care and raise 3 important questions: 1) Is it possible to identify unstable patients who might benefit from earlier ICU care or other aggressive interventions? 2) Do delays in ICU transfer represent preventable adverse events? 3) Is this an important patient safety issue?
The decision to transfer a clinically unstable patient to the ICU is a complicated process that considers the acuity of the patient's condition and deterioration as well as the patient's underlying diagnoses, prognoses, and treatment preferences. The ICU transfer process is also intertwined with such considerations as nurse staffing, physician availability, ICU bed availability, time of day, and hospital policies regarding the use of specific diagnostic and therapeutic modalities outside of the ICU. Delays in transferring unstable patients are clearly undesirable. However, some delays may be unavoidable because of the aforementioned constraints. Specifically, delays in transfer may be markers for deterioration that occurred during off-hours or on the weekends. During these periods, nurse staffing is typically lower, physicians are less likely to be available, and patients may be at increased risk for adverse outcomes.8 The link found between delays in notifying physicians and delays in transfer is also noteworthy, given recent associations at a hospital level between nurse staffing and rates of hospital complications and mortality.9 While increasing nurse staffing levels may be difficult for hospitals because of fiscal constraints or labor shortages, the roughly 60% higher costs associated with slow-transfer patients in the current study raises the possibility that increasing nurse staffing on hospital floors may be cost effective.
Clinical deterioration that ultimately leads to cardiopulmonary arrest or death often follows a common physiological pathway.3 While physiologic deterioration in some patients can be unpredictable and sudden, as with massive pulmonary embolism or ventricular fibrillation, deterioration is often more gradual and insidious, as typically seen in respiratory insufficiency or sepsis. As suggested by the findings of Young et al., earlier identification of patients whose condition is deteriorating may enable providers to effectively intervene and forestall further deterioration and possibly eliminate the need for ICU transfer.
In the current study, 11 physiological and laboratory markers of clinical instability were very sensitive (88%) with respect to identifying patients who were transferred to the ICU. Many of these markers are routinely recorded in electronic medical records, and could be used to generate computerized alerts to physicians and nurses. Unfortunately, the markers lacked specificity (13%) and had a very low positive predictive value (8%). Thus, relying on the markers alone to identify patients for increased vigilance by nurses and physicians, early aggressive intervention, or early ICU transfer would expend considerable resources and would likely have an inordinately high cost–benefit ratio. Moreover, the relative scarcity of ICU beds in many institutions would likely make such early intervention and transfer strategies impractical from an organizational perspective.
Nonetheless, it is possible that much of the predictive power of the markers used in the current study is driven by a few select factors. For example, a prior study found the occurrence of a respiratory rate of 28 or higher during a 72-hour period had a sensitivity of 54% and specificity of 83% in predicting subsequent cardiopulmonary arrest.10 Clearly, empirical research is needed to identify more specific markers of clinical deterioration and to determine if such markers will be clinically useful and practical.
Finally, we believe the current study raises important issues in patient safety. A delay in ICU transfer that leads to a poor clinical outcome should be considered a preventable adverse event. Knowledge of such events, even if after the fact, may be helpful in designing better systems of care. While the importance of eliminating delays in transporting patients to the hospital for certain clinical conditions (e.g., acute myocardial infarction and stroke) has received a great deal of public attention, it is perhaps ironic that little attention has focused on delays in transferring patients already in the hospital to an ICU that may be just down the hall. Although there may be substantial organizational challenges to eliminating delays in ICU transfer, the findings of Young et al. suggest that these challenges must be addressed.—Peter J. Kaboli, MD, MS, Gary E. Rosenthal, MD,Division of General Internal Medicine, Department of Internal Medicine, University of Iowa College of Medicine and Program for Interdisciplinary Research in Health Care, Iowa City Veterans Affairs Medical Center, Iowa City, Iowa.
- 1Quality assessment and assurance in the intensive care unit. In: SivakED, HigginsTL, SeiverA. eds. The High Risk Patient: Management of the Critically Ill. Philadelphia: Williams & Wilkins; 1995: 1576–86., .
- 2Gatekeeping in the Intensive Care Unit Chicago: Health Administration Press; 1997., .