The most common method for tracking blood volume during surgery—a catheter inserted into the heart that transmits information to a monitor—is not only invasive, but not very accurate.

This leads, according to Kirk H. Shelley, M.D., Ph.D., associate professor of anesthesiology, to a delicate clinical balancing act with very high stakes. “Too little fluid can put a tremendous amount of stress on the kidneys, the cardiovascular system and the central nervous system,” Shelley said. “But if you give too much fluid for the heart to pump, it backs up, causing bloating and pulmonary edema.”

Now Shelley, who as chief of ambulatory surgery oversees about 8,000 surgeries a year at Yale-New Haven Hospital, has found a possible solution to this surgical dilemma. By combining a clinical insight from the 1870s with data provided by the modern pulse oximeter, a clothespin-like clip placed on a fingertip, ear or toe to measure the oxygen level in the blood, Shelley has discovered a noninvasive, precisely quantified method to monitor blood loss and guide difficult decisions in the operating room.

The pulse oximeter has become a common sight in hospital hallways since it was first introduced in the 1980s. The clips contain light-emitting diodes that shine both visible red and infrared light through the skin. Because deoxygenated hemoglobin allows infrared light to pass but absorbs red light, while oxygenated hemoglobin allows red light to pass and absorbs infrared, the oximeter can detect changes in the blood’s oxygen saturation by calculating the relative absorption of red and infrared light.

In those early days of the pulse oximeter, Shelley discovered that oximetry clips generated exceedingly complex waveforms that were “cleaned up” by oximeter manufacturers in favor of clear, simple signals. But Shelley’s curiosity about the wealth of information produced by early oximeters prompted him to devise software to sift through the raw oximetry signal for potentially valuable clinical information.

Shelley found that pulsus paradoxus—a drop in blood flow after a deep breath caused by the mechanical ventilation used in anesthesia—could be detected in the raw oximetry waveform.

“There’s all sorts of wild, raw data that comes off the pulse oximeter that companies have worked hard to eliminate, because it has been seen as just noise,” said L. Alan Carr, Ph.D., then a senior licensing associate in Yale’s Office of Cooperative Research who shepherded the discovery through a patent application. “What’s ironic is that the background data actually had useful information in it.”

Shelley plans to mine the pulse oximeter for even more clinical treasure, and he is adapting his method for use in nonventilated patients suffering from blood loss, such as trauma patients arriving at emergency departments.