How Children's of Alabama Uses Real-Time Analytics to Support ICU Liberation

USING ANALYTICS TO FORECAST RISK

Patient deterioration is a major source of mortality and morbidity in hospitals and ICUs. This is especially true for hospitalized children, who are at an increased risk of mortality. Certain types of clinical deterioration events, such as pediatric cardiac arrest, are also associated with poor long-term functional and neurological outcomes, according to research.

Being able to identify whether a patient is getting better or worse is crucial for any clinician. But to do so, care teams must have access to relevant patient data. This can be an issue if the information is siloed, not up-to-date, or spread across multiple sources. CoA uses Etiometry’s clinical decision support software for critical care to avoid the problem.

“The platform connects to our patient monitoring devices, ventilators, and near-infrared spectroscopy (NIRS) devices, which are vital in a leading cardiovascular ICU like ours,” Borasino explained. “The platform automatically captures all the data from high-acuity patients into a single consolidated, up-to-date dashboard, so our team can see a comprehensive snapshot with longitudinal patient status trends. In addition, with the click of a button, we can pull in lab results to be displayed with trends for additional content when evaluating patients.”

The platform is also used at the bedside to flag signs of deterioration and escalate care if necessary.

“When the nurse escalates care to the cardiac intensivist, we use Etiometry's contextual visualization to assess what has been happening with the patient. It helps us break down silos between care teams and unifies us with the same information so we can coordinate care specific to that patient,” Borasino stated.

In addition to tearing down data silos, the platform acts as a support tool for clinicians’ intuition and monitoring ability.

“Before implementing the platform, subtle patient condition changes could be challenging to identify," he said. "We kept a mental picture of the patient's status and used our instincts. The platform supplements our intuitions and mental algorithms with visualization of the information in our minds, giving us more of an edge to find problems, especially small ones, to improve patient outcomes.”    

The platform provides a surveillance view, allowing clinicians to see the state of all of their patients at once, from anywhere, without having to physically go to the patient's room or look at each patient's chart. This feature helps address clinicians’ inability to be in all rooms all the time.

While not a replacement for doing rounds and evaluating a patient in the room, Borasino explained that the enhanced level of patient monitoring allows him to keep an eye on multiple patients between laps and prioritize which patients to see before his next lap, which can help improve outcomes.

IMPROVING EXTUBATION READINESS

Even in the event of patient deterioration, care teams can take steps toward the goals of ICU Liberation. One of these steps is freeing patients from invasive treatments, such as mechanical ventilation, which is common in cardiovascular ICUs.

Part of CoA’s strategy to address this problem comes in the form of an Extubation Readiness Test (ERT), which is designed to evaluate whether a patient is ready to be extubated after a period of time on a supported mode of oxygen delivery, such as mechanical ventilation, Borasino stated. Accurately determining extubation readiness is crucial to prevent failures that result in reintubation within 48 hours, which can have a negative impact on patient outcomes.

As with forecasting patient deterioration, CoA’s real-time analytics strategy plays a significant role in addressing ERT.

“Initially, our ERT protocol used parameters that were assessed manually and subjectively by the care providers from different sources to determine if a patient's physiology was ready to be extubated,” said Borasino. “We wanted to simplify our protocol, automate it, and be more objective.”

As part of this process, the hospital worked with Etiometry to evaluate the parameters that would enhance the precision of predicting a patient's extubation performance using retrospective data from the platform’s Quality Improvement System.

Prior to this analysis, CoA relied on a 10-parameter protocol for ERT. But once the analysis was completed, they found that a four-parameter approach based on a combination of oxygen saturation, respiratory rate, minute ventilation upper limit, and mean blood pressure was the most precise method to evaluate extubation readiness. This approach was then integrated into the platform to streamline patient assessment.

“In the next phase of this initiative, we are doing an additional evaluation with the data from the platform to refine the thresholds of the four protocol parameters to optimize the sensitivity and specificity. Our goal is to optimize the positive predictive value when we have compliance with established parameters, and a patient can successfully be extubated,” Borasino noted.

When CoA completes this evaluation, the new thresholds will be updated within the platform to help advance the hospital’s ERT efforts.

REVIEWING CODE EVENTS

Despite these improvements in patient deterioration forecasting and extubation readiness, some code events will still occur. Code events refer to medical emergencies that require immediate attention. While not all code events are entirely preventable, learning from them is an important part of preventing them where possible and improving outcomes for those that still happen.

“As mentioned previously, the Etiometry data helps us consider all the historical data, not only what was charted hourly, but all of the data from the monitors and devices to see exactly what happened leading up to a code event,” Borasino stated. “This powerful information enables us to see where things went wrong so that we can provide education to our staff to allow them to recognize potential signs of deterioration and catch them early in the future.”

Using information gathered by the platform, CoA completes retrospective code event reviews to facilitate learning and improvement among care teams. Issues like the experience-complexity gap, which refers to a discrepancy between nursing experience and patient complexity, can often present a challenge to health systems looking to move toward ICU Liberation. But, using real-time analytics can help, according to Borasino.

“A considerable percentage of our bedside nurses have less than two years of experience, which can be a challenge in improving patient outcomes for the most acute patients,” he said. “We are addressing this experience-complexity gap by building a culture to make the more novice nurse staff feel safe and provide a supportive environment where team members can assimilate knowledge. Ultimately our goal is to provide the appropriate education and tools so the new nurses will want to stay, and we can retain our highly trained employees.”

The hospital is leveraging the real-time analytics platform as an educational tool during code event reviews. These reviews, which should ideally happen within 48 hours of the code event, involve using the platform to recreate the events that led to a code and finding potential areas for system improvement.

Overall, the platform has acted as both an extra vital sign and a support tool for ICU care teams, Borasino said.

“Essentially, the platform helps with clinical decision-making and gives peace of mind. It can give a sense of relief when a patient is getting better or provide a sense of urgency when a patient is not going in the right direction, all with data that is difficult to visualize and manipulate to see the big picture in the EHR,” he added.