The Clinical Year in Review series continued Monday, May 19, at the ATS 2025 International Conference with an examination of studies with clinical practice implications and a closer look at unresolved questions and developing principles in general critical care, acute respiratory distress syndrome (ARDS), sepsis, and medical education.

General Critical Care

Patrick G. Lyons, MD, MSci, assistant professor of medicine at Oregon Health and Science University, provided a “head-to-toes” update on notable studies in critical care—TRAIN, PREOXI, DanGer Shock, and PRECISe.

The TRAIN randomized trial tested two different hemoglobin thresholds for guiding transfusion in patients with acute brain injury and anemia. A hemoglobin cutoff of 9 (restrictive approach) yielded better outcomes than that with 7 (liberal strategy).

Patients in critical care are preoxygenated whenever possible before intubation to reduce or delay hypoxemia. The PREOXI randomized study showed that noninvasive ventilation (NIV) decreased peri-intubation hypoxemia, compared to preoxygenation with an oxygen mask.

“I have started using the NIV approach, provided that intubation is not a right-this-second need,” Dr. Lyons said. “It does not take much additional work to set up NIV, as most ventilators can connect to an NIV mask with a single attachment.”

In the DanGer Shock trial, temporary mechanical circulatory support with a microaxial flow pump reduced the absolute risk of 180-day mortality by 13 percent, compared to the standard of care (SoC), albeit with a higher incidence of adverse events.

Increased enteral protein support was associated with consistently worse quality of life, over time, than standard enteral protein provision in critically ill patients, per the PRECISe study.

Dr. Lyons also reviewed two studies focused on health technologies/systems—TELESCOPE, which evaluated an ICU telemedicine approach, and SIMPLIFY, which explored an ICU clinical decision support tool.

Dr. Lyons said that TRAIN and PREOXI findings will change clinicians’ decision-making approach for some patients in the critical care setting.

Acute Respiratory Distress Syndrome

Jessica A. Palakshappa, MD, MSci, associate professor of pulmonary, critical care, allergy/immunology, and internal medicine at Wake Forest University School of Medicine, discussed baseline driving pressure’s impact on mortality, the driving pressure-limiting strategy used in STAMINA, comparison of high-flow nasal oxygen (HFNO) versus NIV in RENOVATE, the PROLABI study of lung-protective ventilation in patients with acute brain injury, and individualized treatment effects of oxygen targets in mechanically ventilated critically ill adults.

Driving pressure impacted mortality in an age-dependent manner, with very old patients being more vulnerable to high driving pressure. “A driving pressure threshold of 11 cm H₂O may be most appropriate for very old patients,” Dr. Palakshappa said.

Although driving pressure is considered a key determinant of ventilator-induced lung injury (VILI) and death in observational ARDS studies, the STAMINA randomized study showed no difference in ventilator-free days (VFDs), secondary outcomes, or safety with a driving pressure-limited strategy.

In RENOVATE, HFNO was non-inferior to NIV, regarding endotracheal intubation or death, in non-immunocompromised patients with hypoxemia, COPD exacerbation with respiratory acidosis, acute cardiogenic pulmonary edema, and hypoxic COVID-19, but not in immunocompromised hypoxemic patients.

ARDS severity based on impairment of oxygenation is not predictive of VILI risk, which suggests that the practice of recommending respiratory support and how ARDS is classified needs to be re-examined.

Sepsis

Lisa K. Torres, MD, MS, assistant professor of medicine at Weill Cornell Medicine, organized her presentation on sepsis around four key themes: tools for sepsis management, the use of electronic medical records (EMRs), sepsis in a global health context, and developing precision approaches in pediatric sepsis.

“The optimum treatment duration of antibiotics is uncertain. The decision to stop antibiotics is often based on the patient’s clinical progress,” Dr. Torres said, contextualizing the role of antibiotics in sepsis management.

However, the two measures used to guide cessation of antibiotics, procalcitonin and C-reactive protein (CRP), are supported by weak evidence or lack consensus guidance. In this context, the ADAPT-Sepsis trial showed that care guided by procalcitonin, but not CRP, shortened antibiotic duration, compared to standard care.

Dr. Torres reviewed the utility of rapid PCR-based pneumonia testing in improving sepsis outcomes and antibiotic stewardship, based on the INHALE WP3 pragmatic trial.

An electronic sepsis screening alert, implemented in the EMR and based on the quick Sequential Organ Failure Assessment score, reduced in-hospital mortality among hospitalized patients.

Sepsis research conducted in low- and middle-income countries (LMICs) is often overlooked, Dr. Torres noted. She discussed a meta-analysis evaluating how geo-economic factors affect the risk-benefit balance of fluid management strategies in sepsis, and a sepsis study in India, a country that accounts for 26.4 percent of sepsis-related deaths worldwide.

Dr. Torres also shared her thoughts on gene expression signature-based predictions of organ dysfunction in children with viral or bacterial infections. Such biomarker-guided methods may facilitate sepsis recognition and appropriate antibiotic use in children, in whom infections are mostly of viral origin.

Medical Education

Morgan Soffler, MD, assistant professor of medicine at the New York Medical College School of Medicine and associate program director for the Westchester Medical Center Pulmonary and Critical Care Fellowship, highlighted the role of artificial intelligence (AI) in medical education, the importance of procedural skills training, and the involvement of non-physicians, such as advanced practice providers (APPs), in medical education.

Dr. Soffler shared these key takeaways from a cross-sectional study that compared AI-facilitated clinical reasoning to that of attending physicians and trainees: AI can synthesize clinical data, creating and prioritizing a problem list, while performing similarly to physicians in diagnostics. However, the role of AI beyond simulated study settings remains unclear. The major takeaway, she said, is that while AI may enhance clinical reasoning in collaboration with physicians, it is unlikely to replace physicians.

Dr. Soffler also discussed how interventional pulmonology fellowships may impact the bronchoscopy skills of pulmonary critical care fellows and how APPs may be an underused resource in graduate medical education.