Maximizing limited resources is a continuing concern for ICUs in hospitals of all sizes. Six abstracts presented at CHEST 2022 during Optimizing Resources in the ICU on Sunday, October 16, explored the effects of resource levels on patients as well as new technologies and approaches that could help improve care.
Predictive modeling and artificial intelligence have been integrated into a growing number of health care areas, including sepsis. But it can be difficult to determine how well those predictive models perform in the real world, as many are based on proprietary algorithms and have not been validated externally.
“One of the big problems with sepsis is a lack of current tools for early and accurate diagnosis,” said Dan Burgin, MD, a third-year resident at Louisiana State University Internal Medicine in Baton Rouge and part of a team that investigated the performance of a sepsis predictive model, the Epic Sepsis Model (ESM), in the ED at Our Lady of the Lake Health.
The ESM calculates a sepsis score every 15 minutes based on a variety of clinical factors, blood pressure, age, medical history, and other elements, Dr. Burgin said. The model triggers a Best Practice Advisory that includes clinical recommendations and extended order sets.
Antibiotic order is directly associated with antibiotic delivery, with earlier initiation of orders resulting in earlier delivery to patients, he explained. The question was whether the predictive ESM model can improve antibiotic order times for patients with sepsis compared with triage clinicians’ clinical judgments—and according to Dr. Burgin, currently, the answer is no. His team saw no consistent pattern between the sepsis alert time and the antibiotic order time from triage.
“Given the positive predictive value and how poor it is in this, and in conjunction with other findings, we question the utility of this tool in its current state,” he said.
Sepsis survivors also experience multiple problems after discharge from their initial stay in the hospital, and more than one-third are discharged to post-acute care because they are not stable enough to go home.
“They experience higher readmission rates as well as mortality rates than those who are discharged home,” said Nicholas Colucciello, MD, a resident at University of Toledo Medical Center. “It’s a very vulnerable group of patients.”
Dr. Colucciello presented a secondary analysis of patients from the randomized Improving Morbidity During Post-Acute Care Transitions for Sepsis (IMPACTS) trial, comparing usual care for patients discharged to a post-acute care facility vs the effects of a Sepsis Transition and Recovery (STAR) program. STAR used nurse navigators to deliver best practice proactive and sepsis-specific care during and after hospitalization for sepsis.
Additional studies covered during the session covered a range of issues and technology important to improving care for critically ill patients.
Palak Rath, MD, a fourth-year internal medicine resident at Cleveland Clinic Akron General Hospital, covered the results of a study exploring outcomes for critical care patients boarded in the ED. Per Dr. Rath, among other results, the study indicated that “the average patient, with an average boarding time of 205 minutes, experienced a 5.33% increase in ICU length of stay and a 4.72% increase in ventilator days because of boarding.”
Amy J. Montgomery, MD, MPH, a third-year internal medicine resident at Mayo Clinic, presented results from a study exploring the use of “digital twins,” virtual counterparts to real-world individuals and populations that are used for testing and training. Mayo researchers convened a multinational Delphi survey panel of 30 critical care and respiratory physiology experts across six countries to assess distinct statements that would be used to code a respiratory system for a digital twin.
“This can be used to improve ICU medical education, research, as well as support for clinical decision-making to look at outcomes of interventions without having to expose real patients,” Dr. Montgomery said.
Faculty also explored increases to the number of ventilator-assisted pneumonia (VAP) hospitalizations in recent years, as well as data suggesting that patients treated in the ED and other hospital units over weekends may be at increased risk of mortality compared with similar patients cared for during the week.
According to Namratha Meda, MBBS, a second-year resident at MedStar Health-Georgetown/Washington Hospital Center who presented the results of a study exploring the recent increase in VAP-related hospitalizations, “Cost associated with VAP hospitalizations for just 2019 was an impressive $2.8 billion.”
Dr. Meda noted that the results also indicated significant differences by race and gender. For example, White patients tended to be older, compared with Black and Hispanic patients, and were likelier to die in the hospital, and male patients tended to have a longer length of stay than female patients.
Nastaran Haghani Rad, MD, an internal medicine resident at John H. Stroger Hospital of Cook County, presented the results of a retrospective analysis exploring whether weekend staffing and diagnostic resource availability had an impact on hypertensive emergencies.
The analysis found no significant differences in inpatient mortality, length of stay, or total hospital charges for weekend vs weekday admissions, but researchers noted that patients admitted over the weekend had slightly higher rates of acute coronary syndrome, acute respiratory failure, and acute decompensated heart failure compared with weekday admissions.
“This calls for especially careful monitoring and proper management when it comes to a weekend admission,” Dr. Haghani Rad concluded.