Transrectal ultrasonography in prostatic cancer: interexaminer variability of interpretation.

Abstract

To compare the variability of transrectal ultrasonographic (TRUS) interpretation for the decision of performing biopsies and the lesions to biopsy. We extracted at random from our videotape database 16 records of patients who had undergone biopsies, added 2 normal glands and duplicated 2 of these 18 records. Based on the records, 5 well-trained physicians had to describe the images on the tape, and to decide whether or not to biopsy the prostate. A kappa test was computed between each couple of readers, and for the whole group. The kappa test denotes the agreement between examiners. A value of kappa < 0.20 is considered poor to slight agreement, 0.2-0.40 is considered fair agreement. The agreement between the 5 readers was poor for the biopsy decision (kappa < 0.2) and the difficulty to read the records (kappa = 0.05). The results with the global kappa were similar with a highest value < 0.3. Most of the abnormalities were described in the peripheral zone. The global kappa for the seminal vesicles interpretation is poor, but better for the capsular penetration. TRUS has a poor informative value between different practitioners. This poor agreement between different practitioners must lead to more objective ultrasonographic methods.