Dyspnea matters: Normal oxygen levels may hide “silent suffocation”

A study led by biomedical scientist Erica Heinrich at the University of California, Riverside, highlights a critical gap in how clinicians detect and treat breathing distress, or dyspnea, particularly in patients on ventilators.

Dyspnea, the medical term for breathing discomfort or shortness of breath, is often hard to recognize. According to Heinrich, "it's often very difficult to tell when a patient is experiencing breathing discomfort," and current clinical approaches may be overlooking it.

In controlled laboratory experiments with nearly 70 healthy participants, Heinrich's team induced dyspnea by carefully adjusting oxygen and carbon dioxide levels. While both gases influence breathing, the findings point to carbon dioxide - not oxygen - as the most significant driver of the sensation of breathlessness.

"This challenges a common clinical practice," said Heinrich, an assistant professor of biomedical sciences in the UCR School of Medicine. "Clinicians often rely on oxygen saturation in the blood, or SpO₂, to assess patients. But oxygen levels are often a poor predictor of how breathless someone feels. Carbon dioxide is a much more important driver of this sensation."

The study, published in Respiratory Physiology and Neurobiology, reinforces growing evidence around "silent hypoxemia," a condition in which patients have dangerously low oxygen levels without feeling short of breath - or conversely, feel severe distress despite normal oxygen readings. The work aims to produce a predictive tool that can help clinicians predict and continuously monitor breathing comfort in patients. It will use a machine-learning algorithm to detect dyspnea from noninvasive biomarkers.

Even more concerning, outward signs of distress, or lack thereof, can be misleading, Heinrich said.

"We found that sometimes the person who looks the calmest reports the most distress," she said. "Subtle cues that appear like anxiety, such as darting eye movements, may reveal more than visible breathing patterns."

For patients on mechanical ventilation, the stakes are especially high. Ventilators often use low tidal volumes - small breaths - to protect the lungs from injury. But this can intensify the sensation of breathlessness. In some cases, patients who are sedated or unable to communicate may feel as though they are suffocating.

"This can be a truly terrifying experience," Heinrich said. "Patients describe feeling like they're dying, but they can't communicate, and no one realizes what they're going through."

Such experiences can have lasting consequences. High rates of post-traumatic stress disorder (PTSD) have been documented in patients after ICU stays involving mechanical ventilation, and untreated dyspnea may be a contributing factor.

The study also points to a broader issue in medical training. Unlike pain, dyspnea is not widely taught as a complex, subjective experience.

"There's very little education on the patient experience of dyspnea," Heinrich said. "It's underrecognized and undertreated compared to pain."

The implications extend beyond the ICU. In chronic conditions like Chronic Obstructive Pulmonary Disease (COPD), the severity of dyspnea may be a stronger predictor of mortality than traditional lung function tests. The researchers suggest that continuous monitoring of breathlessness, potentially through noninvasive tools, could improve diagnosis and care.

Looking ahead, Heinrich emphasizes the need for both cultural and clinical change.

"We need to take dyspnea as seriously as pain," she said. "That means better training, better monitoring, and rethinking how we balance lung protection with patient comfort. At its core, the message is simple but urgent: Dyspnea matters."

Heinrich was joined in the study by Karapet Mkrtchyan, who recently graduated from UCR with a doctoral degree in biomedical sciences, as well as UCR collaborators Wei Vivian Li, Shujie Ma, and Mona Eskandari.

The study was funded by a School of Medicine Dean's Collaborative Seed Grant to Heinrich and Eskandari.

The title of the paper is "A machine learning approach to predicting dyspnea with noninvasive biomarkers."

Header image credit: Mufid Majnun on Unsplash