Previous Abstract | Next Abstract
Printable Version
October 15, 2013
9:00:00 AM - 10:00:00 AM
Room Room 104-Area C
The Effects of Surgical Masks on Speech Perception
Peter D. Winch, MD, MBA, Kelsi Wittum, Student, Lawrence Feth, Ph.D., Evelyn Hoglund, Ph.D.
Nationwide Children's Hospital Anesthesiology, Columbus, Ohio, United States
Surgical masks and blood shields worn by anesthesiologists and surgeons in hospital operating rooms negatively impact speech communication and put patients at needless risk. Young adult subjects listened to sentences from the Speech Perception in Noise Test (SPIN) recorded by a male and female talker. All eight SPIN lists were recorded under three different speaking conditions: 1) speaking normally without any obstruction, 2) wearing a typical surgical mask, and 3) wearing a surgical mask with an attached blood shield. Multi-talker babble was mixed with the SPIN sentences at the signal-to-noise ratio of 0dB to simulate conversation in noisy environments. Speaker gender and recording conditions were counterbalanced across listeners to control for learning and fatigue effects. SPIN test scores for each of the three types of recordings and talker genders were compared in order to determine the degradation that blood-shields and surgical masks have speech communication in the operating room. The data suggests that surgical masks, in particular the blood shields, negatively impact speech communication. Percent correct is the highest for the unmasked condition, followed by the masked condition, and poorest in the mask and attached blood shield condition.


A panel of 21 listeners was recruited for voluntary participation comprised of young female adults. Normal hearing established by standard audiometric screening tests, defined as HL< 20dB, 250 Hz - 4 kHz, both ears. The measurement involved the speech perception in noise (SPIN) test. It is made up of eight lists of 50 different sentences. Twenty-five sentences rated as “high predictability”, for example, “Stir your coffee with a spoon.” Twenty-five sentences rated as “low predictability”, for example, “Bob could have known about the spoon.”

Two professional speakers, male and female, under three different speaking conditions: unmasked, wearing a surgical mask, and wearing a surgical mask with an attached blood-shield. The order of presentation was randomized and counterbalanced with listeners seated in a sound attenuating booth. Listeners repeated the last word of each sentence and the experimenter listened to the SPIN sentences unmasked. The listener responses were monitored and they marked correct and incorrect responses on a scoring sheet.


The average percent correct across all speaking conditions: unmasked (48.5%) > masked (33.1%) > shield (20.9%). The average percent correct for talker gender: female talker (47.1%) > male talker (21.2%). The average percent correct for sentence context: high predictability (39.2%) > low predictability (28.4%). Statistical analyses using a 3-way, repeated measures ANOVA indicated that: each of the three main effects: speaking condition, talker gender, and sentence context were significant at the 0.01 level. None of the other interactions were significant at the 0.01 level.

Copyright © 2013 American Society of Anesthesiologists