Click here to access the CE quiz based on this article.

By Genevieve Diesing

In the 20 years since the publication of To Err Is Human—the landmark Institute of Medicine report that documented the extent of US medical errors—hospitals still struggle to control hospital-acquired conditions, avoidable readmissions, and healthcare-associated infections.
But some of providers’ most insidious safety problems—such as unrecognized patient deterioration, opioid-induced respiratory depression (OIRD), and sepsis—are almost entirely preventable.

Continuous clinical surveillance solutions, or tools that use real-time patient data to identify the early onset of patient deterioration, can help clinicians do just that.

These tools aggregate continuous streams of data from multiple patient monitoring devices, as well as retrospective information from electronic health records (EHRs), and combine them with advanced analytics to produce a holistic picture of a patient’s condition.

That picture could reveal subtle trends about a patient’s health early on, giving clinicians valuable lead time in cases of deleterious conditions.

Continuous surveillance is also necessary because at-risk patients, especially those who require respiratory support or monitoring, do not always visibly fit the profile of a highly sick person. These patients often exist, undetected, across the care continuum, not just within high-acuity areas.

According to a 2018 Journal of Critical Care study, 46 percent of patients on the general care floor suffer from OIRD, which causes more than half of medication-related deaths in care settings, 97 percent of which were preventable.1

How Continuous Surveillance Works

In contrast with traditional patient monitoring (or the periodic measure of a patient’s heart rate, blood pressure, respiratory rate, oxygen level, and temperature), continuous surveillance involves the continuous acquisition of patient data and uses predictive analytics to identify trends across multiple data points over time.

Specifically, continuous surveillance technology pulls information from EHRs and other data sources, sorts and analyzes that information, and, when applicable, automatically sends pertinent alerts back to clinicians.

Continuous surveillance solutions ideally enable hospitals to create tailored rules based on their needs, coordinate with their EHRs, provide support from a clinical team on the vendor’s side, and pull information from the EHR and other data sources in near-real time.

Outsmarting Alarm Fatigue

With intuitive predictive data at their disposal, clinicians can make appropriate care decisions without being inundated by alarms, says Mary Jahrsdoerfer, PhD, director of graduate healthcare informatics at Adelphi University College of Nursing and Public Health.

While an expert nurse might anticipate patient decline based on the siloes of clinical data gleaned from typical patient monitoring, “the sheer volume of data collected from patients is beyond the capacity of the human brain to analyze,” Jahrsdoerfer says.

By moving away from reactive, intermittent monitoring and toward comprehensive surveillance, clinicians can use real-time data to divert adverse events. With a complete view of a patient’s illness and past medical history, clinicians are much better prepared to diagnose and treat him or her.

While typical patient monitoring requires clinicians to individually observe patients during singular moments while using individual devices, continuous surveillance is team-based, enabling multiple caregivers to view an inclusive image of multiple patients from either a centralized location or through mobile alarm notifications. Additionally, conventional monitoring practices often result in alarm fatigue.

The flood of false alarms has desensitized caregivers, Jahrsdoerfer says, which presents its own patient safety danger.
“What will happen is, sometimes the alarm will be real,” Jahrsdoerfer says. “And [clinicians] didn’t tend to it and a patient dies. So, we have to have smarter alarms.”

Continuous surveillance presents an opportunity to screen out false alarms or artifact signals that typically contribute to alarm fatigue.

By using analytics based on multiple sources of data, providers can take a system-wide inventory of alarms, evaluate their existing algorithms and decide where real-time data can be used to improve sensitivity and specificity to reduce false positives.

Solutions might come in the form of a single risk score based on medications, nursing assessment, and medical history, says Shannon Sims, MD, physician informaticist at Vizient, a healthcare performance vendor.

As Edward Pollak, MD, medical director and patient safety officer at the Joint Commission, puts it, clinicians could choose to configure their continuous surveillance tools to cut through data overload and get “a single roll up” of the patient’s overall condition, presenting only comprehensive, need-to-know information—such as the need for potential patient interventions, department transfers, or discharges.

Although such automated patient safety monitoring systems supplement human oversight, they are not a replacement for clinical judgement, says Sims.

As such, automated safety monitoring and alerting approaches should be continuously reviewed and refined to ensure clinical validity, efficacy, and avoidance of unintentional outcomes, such as unnecessary testing or provider alert fatigue, he says.

Proactive Measures

Stephen Morgan, MD, chief medical information officer at the Carilion Clinic in Roanoke, VA, says the health system’s continuous surveillance tool has been a proactive measure in flagging patients “before they become really, really sick.”
While Carilion hasn’t yet measured the overall quantitative impact of the tool—the health system uses PeraHealth’s Rothman Index—Morgan says its ability to predictively identify at-risk patients has enabled Carilion to reduce ICU readmissions, and to potentially predict patients’ lifespans during end-of-life care.

“When we have a conversation with a family [of a dying patient], we’re able to use this as just one more data point to help people make a decision for their loved ones,” Morgan says.

Pollak notes that providers are particularly successful with continuous surveillance in areas where they’ve used medical devices the longest, such as in remote blood glucose monitoring—where wireless insulin pumps react to feedback loops via algorithms—and cardiac implants, which adjust themselves based on algorithms and can detect arrhythmias through monitoring.

“We’ve seen some pretty good outcomes,” he says.

Systemwide Support

In addition to improving patient safety, information gleaned from real-time devices and EHR data can also inform hospitals’ emerging patient safety strategies and approaches to governance.

Organizations can use EHR data and information from other devices to support retrospective analyses of safety, which enable hospitals to determine the efficacy of their patient safety policies, clinical protocols, and automated interventions such as decision support alerts, says Sims.

This might include root cause analyses or routine monitoring to support continuous improvement and regulatory compliance, he says.

Additionally, retrospective analyses help organizations to examine their compliance with clinical protocols and the success of their patient harm prevention alert strategies. For example, they might mine the EHR for use of rescue agents such as naloxone and monitor lab data for drug-related harm such as hypoglycemia.

In addition to preventing patient harm, continuous surveillance tools create efficiencies that potentially reduce a patient’s length of stay, which help hospitals comply with value-based reimbursement regulations, saving them money.

Predictive Monitoring in the Age of COVID-19

The Carilion Clinic is using real-time data to manage the availability of its beds, materials, and personal protective equipment, a process it has had to accelerate in light of COVID-19.

“We had a fairly robust process of tracking the data for bed availability, ventilator availability, etc., but we’ve really had to step that up to be able to track those more in real time to be able to manage the hospital,” Morgan says.

Carilion Clinic’s Patient Deterioration Index has also been “very, very helpful” in monitoring patients with COVID-19 remotely, Morgan says.

The health system is monitoring patients’ oxygen saturation levels at home through remote patient monitors. Although Carilion has previously used the technology to monitor patients for chronic obstructive pulmonary disease and congestive heart failure, it has become particularly useful for keeping patients infected with the virus at arms’ length while still observing them, Morgan says.

“You want to get [COVID-19 patients] out of the hospital when it’s appropriate, and not have them expose others,” he says.

“But you’re still keeping an eye on them.”

Continuous surveillance tools can help providers to improve patient outcomes, assess bed capacity, and better time the utilization of personal protective equipment, says Sims.

“By assessing risk of respiratory failure, for example, you can determine likely ICU/ventilator utilization or conduct controlled respiratory intubations, rather than emergency intubation,” he says.

Emergency intubations increase patient risk and can potentially transmit COVID-19 to the care team, so it’s important to avoid them as much as possible, he says.

Additionally, early recognition and transfer to ICU, when needed, may decrease ICU length of stay and improve mortality rates, Sims says.

Accurate data is key to meaningful analytics, and COVID-19 has shed additional light on this, Pollak says.

Because organizations can sometimes misclassify their data—just as public health officials didn’t immediately attribute COVID-19 deaths to the virus—predictive analytics predicated on those inaccuracies lose their meaning.

“The problem with all of these big data sets is that they’re vulnerable,” Pollak says. “A lot of times organizations slap really nice data analytics on really poor data sets. And you just add to confusion.”

At the very least, there will be more data to work with going forward, Morgan says, as providers across the country are sharing analytics to create better predictive models.

“I think more people will be sharing data going forward and there will be more collaboration between health systems,” he says. “It’s already really created some nice synergies to help fight COVID.”


1. Khanna, Ashish K., Frank J. Overdyk, Christine Greening, Paola Di Stefano, and Wolfgang F. Buhre. “Respiratory Depression in Low Acuity Hospital settings—Seeking Answers From the PRODIGY Trial.” Journal of Critical Care, no. 47 (October 2018): 80-87.


Genevieve Diesing ( is a freelance health writer.

Continuing Education Quiz

Review quiz questions and take the quiz based on this article, available online.

  • Quiz ID: Q2019107
  • Expiration Date: July 1, 2021
  • HIM Domain Area: Data Structure, Content and Information Governance
Leave a comment

Send a Comment

Your email address will not be published. Required fields are marked *