Accuracy and reliability of transanal ultrasound for anterior anal sphincter injury

Abstract

Introduction: Although transanal ultrasound has rapidly become the test of choice for the diagnosis of anal sphincter injury, the accuracy and reliability of this technique are unknown. This study evaluates the accuracy and reliability of transanal ultrasound for anterior (obstetric-related) anal sphincter injury.

Methods: Sixty-two women underwent transanal ultrasound with hard-copy images obtained at 0.5-cm intervals from the anal verge to 2.5 cm into the anal canal. All transanal ultrasound procedures were also recorded on videotape. Two experienced ultrasonographers blinded as to the patients' clinical history and examination independently reviewed the images and videotape recordings for the presence or absence of anal sphincter injury.

Results: The accuracy of transanal ultrasound in 22 incontinent women with known anal sphincter injury was 100 percent. The accuracy of transanal ultrasound in 20 nulliparous women with intact anal sphincters was only 35 percent but improved to 50 percent after the "real time" videotape was reviewed (P = 0.16) and further improved to 85 percent when interpretation was limited to the distal 1.5 cm of the anal canal (P = 0.004). In these nulliparous women, intact internal sphincters were more accurately predicted than intact external sphincters (95 vs. 85 percent; P = 0.24). Measurement agreement between the two ultrasonographers was 68 percent (fair; kappa, 0.26) but significantly improved to 78 percent (moderate; kappa, 0.48; P = 0.0001) when interpretation was limited to the distal 1.5 cm of the anal canal. Overall clinical agreement (final scan interpretation) was good (81 percent agreement; kappa, 0.61). Agreement was better for the internal sphincter (74 percent; fair; kappa, 0.36) than the external sphincter (61 percent; poor; kappa, 0.17; P = 0.0002).

Conclusions: Although transanal ultrasound can accurately identify anterior anal sphincter injury when present, transanal ultrasound falsely identifies sphincter injury in at least 5 to 25 percent of normal anal sphincters. Only fair agreement in the interpretation of transanal ultrasound exists between experienced ultrasonographers. Both the accuracy and reliability of transanal ultrasound are significantly improved by limiting transanal ultrasound to the distal 1.5 cm of the anal canal.