Surprising interobserver and intra-observer variability in pediatric testicular ultrasound volumes

Abstract

Testicular volume (TV) can be obtained by either scrotal ultrasound (SU) or orchidometer. Scrotal ultrasound allows for a more objective measurement; however, the interobserver and intra-observer variability of TV measurements has not been rigorously studied. The authors measured intra-observer and interobserver variability of SU TV measurements in pediatric patients to assess the reliability and reproducibility of SU. Special attention was paid to how often a 20% discrepancy in TV was noted as this has previously been utilized as an indication for varicocelectomy. Patients with an indication for SU or undergoing an ultrasound for another reason were prospectively recruited. Two different urologic specific ultrasound technicians (A and B) performed SU to assess interobserver variability. A second measurement was taken by technician A within 90 days to assess intra-observer variability (A vs A1). The technicians were blinded to other ultrasound results. Fourteen patients (28 testes, 56 volume measurements) were included in the intra-observer group and 17 patients (34 testes, 68 volume measurements) in the interobserver group. The mean time to repeat intra-observer ultrasound measurements (range) was 46 days (23-84). Mean age (range) in the intra-observer group was 14.3 years (11-19) and 14.1 years (11-19) in the interobserver group. Indication for ultrasound was varicocele (n=6), scrotal pain (4), hydronephrosis (3), hydrocele (2), epididymal cyst (2), posterior urethral valves (1), and testis asymmetry (1). Utilizing Bland-Altman analysis and plots, variability was seen in both intra-observer and interobserver measurements. The mean values for testicular sizes for technician A and technician B were 13.0±9.7cm3 vs 13.8±9.9cm3, respectively. The mean values for TV measurement for technician A's first and second measurements (A, A1) were 14.3±9.7cm3 and 14.8±8.9cm3, respectively. An errant 20% difference in TV measurements for the same testis was seen in 25% (7 of 28) of intra-observer measurements and 35% (12 of 34) of interobserver measurements. These 20% differences were more common with a lower body mass index (odds ratio, OR=0.74, p=0.01) in the interobserver group, and lower TV was a predictor in the intra-observer group (OR: 0.82, p=0.009). Variability exists in both interobserver and intra-observer measurements of TV by dedicated urologic ultrasonographers, and greater than 20% of differences in measured TV in same testicles occurred in over 25% of cases. Caution should be exercised in basing operative decisions and scientific studies on limited measurements of TV.