Most ordinary people assume that the intensive care unit is exclusively reserved for the most critical patients: those with a high risk of sudden deterioration or death. But hospital workers know that’s not always the case. Crowded wards with more sick patients than beds to hold them might push non-critical patients into the ICU, using it as a holding area. But when an emergency arises, what happens to a patient in need? Dr. Lena Chen of the University of Michigan argues that routinely collected EHR data can be used to evaluate the sickest patients, predicting their chance of deterioration and making better use of scarce and expensive critical care resources.
In an article
recently published in the New England Journal of Medicine
, Dr. Chen notes that despite varied efforts to respond to a shortage of intensivists, “relatively little effort has been devoted to what could be the most promising approach to the problem: the application of advances in health information technology (HIT) to triage decisions.” Only a few health systems have tapped into the wealth of structured data collected every day in EHRs: the VA and Kaiser Permanente, for example, use EHR data to generate risk of death estimates for their admitted patients.
“Yet these calculations of risk, which may combine real-time data on laboratory results, demographics, coexisting conditions, and vital signs, are not being used to inform decisions about admission to the ICU,” Chen says. “To accelerate progress in this area, we believe that more targeted incentives for meaningful use
of HIT should be considered.”
Chen and her associates examined the records of 101,912 patients admitted into VA acute care facilities for reasons other than scheduled surgery. The researchers found that severely ill patients with non-cardiac complaints were more likely to be admitted to the ICU than the less severely ill. But in cardiac cases, “severity of illness played a negligible role in ICU admitting decisions.”
“Patients with cardiac illness may have a need for critical care that isn’t captured by severity scores,” Chen explains. Doctors may simply be taking precautionary measures, since cardiac patients can deteriorate rapidly despite the appearance of good health. “However, the VA’s ICU severity score is an excellent predictor of the 30-day risk of death, with areas under the receiver-operating-characteristic curve of 88% for patients with cardiac illness and 81% for those with non-cardiac illness.”
With such useful results from the VA’s EHR-based predictions, Chen urges more research on using structured data to help predict the likelihood of negative outcomes, saving space, time, and money on expensive ICU care. “Reliable, individualized, EHR-based predictions of risk have the potential to improve our ability to triage – and hence care for – patients.”