It may be feasible to use cry acoustics to identify neonatal pain. Although research is still very early, investigators from Vanderbilt University Medical Center, in Nashville, Tenn., hope their study evolves to the point at which health care personnel can use a hand-held device to identify the nature of neonatal pain.
“We all know that identifying pain is a tremendous challenge in dealing with neonates,” said Carrie Menser, MD, assistant professor of anesthesiology at Vanderbilt. “And although we have a lot of subjective tools that we use, none have really ever been validated against any gold standard of neonatal pain.
“With that in mind, we wanted to take an actual gold standard—which in this case was neonatal somatosensory processing in the cortex—and compare it with neonatal cries and the acoustical features associated with those cries, to see if we can learn anything about an infant’s cry and whether it is an indicator of pain.”
Study Used Three Stimuli
The investigators hypothesized that cry acoustics can reliably distinguish crying associated with painful and nonpainful stimulation.
Fifty-four healthy, full-term newborns were enrolled in the prospective, observational study. The researchers digitally recorded and analyzed cries elicited in response to three standardized stimuli: brief room temperature air puffs, cold air puffs and a heel lance clinically required for blood sampling. Acoustical recordings were edited with software to remove any identifying information and background noises; cry signals were then analyzed.
“How did you know the electrodes were positioned correctly?” asked session moderator J. Lance Lichtor, MD, professor of anesthesiology and pediatrics at Yale School of Medicine, in New Haven, Conn. “For adults it’s easy; not so with neonates.”
“The caps are based on the circumference of the head; there are five different cap sizes,” Dr. Menser replied. “It takes a couple minutes to get a baseline reading, and then we started the study.”
A total of 79 acoustical features were extracted from the cry samples for analysis. Cortical somatosensory processing was measured through event-related potential (ERP) recordings simultaneously obtained during each stimulus. Previous research indicates that ERP responses to pain from heel lances occur in the late time windows (500-700 msec) in frontal and centroparietal locations, while responses to light touch occur in earlier time windows (50-200 msec).
“This has been very well studied in adults, and replicated in children prior to this study,” Dr. Menser said. “So that was our gold standard of whether or not these babies experienced pain with a heel lance. And the data showed that they did.”
Indeed, as Dr. Menser reported at the 2016 annual meeting of the International Anesthesia Research Society (abstract S-245), the researchers analyzed the difference in ERP amplitude between the heel lance and light touch stimuli within the 535- to 635-msec window in frontal and centroparietal locations.
Interesting Pilot Data
This derived difference score was then computed to represent pain-specific ERP activations. Based on scatterplots and correlations between the derived ERP pain index and various acoustical cry features, seven acoustical features were identified as potential indicators of pain.
“This is very much pilot data at this point, but the thought is that the human ear is not very good at distinguishing between these cries,” Dr. Menser said.
Should the results pan out, Dr. Menser sees a time when analyzing acoustical features may help practitioners treat neonatal pain better. “The way the technology stands now, it’s not very useful because you need a lot of technology and people to analyze it,” she said.
“But if you’re on the floor with an app on your phone and have a postoperative 3-month-old that’s crying, it would be great to say that this cry has features that have a high or low likelihood of pain. I don’t know that we’ll ever be able to say definitively one way or the other, but being able to generate a likelihood based on gestures would be a huge help.”
“One other thing that has historically been valuable is the eye muscles, to look for grimacing,” noted Myron Yaster, MD, the Richard J. Traystman Distinguished Professor of Anesthesiology and Critical Care Medicine at the Johns Hopkins University School of Medicine, in Baltimore. “Have you thought about incorporating this into your study?”
“We have video recordings to do facial coding,” Dr. Menser replied. “But that part has not been completed at this point.”
Despite the potential for the technology, Dr. Menser noted that crying is not the only way to determine whether a neonate is experiencing pain or not. “Even babies that don’t cry generate a cortical somatosensory process of pain,” she said. “So crying is not the only answer. Because even a child that you may score very low on the FLACC [Face, Legs, Activity, Cry, Consolability] scale could be processing pain at a cortical level.”