Status Plus

abstract

121 - HOW REPEATABLE IS SPHINCTER IMAGING BY EXO-ANAL 4D ULTRASOUND?

121

HOW REPEATABLE IS SPHINCTER IMAGINGBY EXO-ANAL 4D ULTRASOUND?

N. SUBRAMANIAM, J. BIENKIEWICZ,T. FRIEDMAN, F. TUREL, H. DIETZ;
Sydney Med. Sch. Nepean, Sydney,Australia.

Introduction: Exo- anal ortrans-perineal ultrasound (TPUS) is rapidly gaining popularity in theassessment of anal sphincter trauma. The repeatability of TPUS in theassessment of functional pelvic floor anatomy is well documented1,but there is a lack of repeatability data for anal sphinctertrauma.
Objective: To determine the medium- to long-termrepeatability of TPUS assessment of anal sphincter trauma and comparethe performance of a novice with that of an experiencedinvestigator.
Methods: This is a retrospective studyanalysing archived ultrasound volumes datasets of 103 patients whopresented with symptoms of pelvic floor dysfunction. Each wasassessed twice between 2012 and 2016 at an average interval of 260(1-1100) days. Tomographic ultrasound imaging (TUI) was used toevaluate external anal sphincter integrity as describedpreviously2.
Offline analysis of archived volume datawas performed at a later date by the first (> 1 year’sexperience) and second (no prior experience) authors, independentlyand blinded against all clinical data and each other’s results.Assessment of data sets obtained at the second time point was blindedagainst the results obtained for the first time point (Figure 1).
Todetermine intra-observer agreement for each of the investigators,Cohen’s kappa (κ) and agreement (%) were used to assessqualitative data, while intra-class correlation coefficients (ICC)were used to test the repeatability of defect angle measurements.


Figure 1: Assessment of theanal sphincter by 3D-TPUS. The two panels show the same patient at aninterval of 216 days. The first assessment was rated borderlinenormal (3/6 slices abnormal), the second one as a ‘significantresidual defect’ (4/6 slices abnormal). When assessed by anexperienced observer, most discrepancies on re-test occurred in womenwith such borderline findings.
Results: During the studyperiod, 105 women had two TPUS assessments for anal sphincter trauma.Ultrasound volumes were missing in 2, leaving 103 data sets foranalysis. In a test retest series undertaken prior to commencement ofthis study, the novice investigator achieved good inter-observerrepeatability with a kappa of 0.71 and ICC of 0.72 for significantEAS trauma and defect angle measures, respectively. In the actualtest- retest study, the novice investigator achieved relatively highagreement (89.4%) for significant trauma and 79.3% for single slicedefects, but kappa and ICC for all parameters were low (κ 0.3-0.22;ICC 0.22-0.28). The experienced investigator, however, demonstratedgood to very good repeatability (κ 0.78-0.91; ICC 0.82-0.9) betweenthe two assessments which equates to 98.1% agreement for significanttrauma and 94.7% agreement for single slice defects.

Table1 Repeatability of EAS trauma on TPUS (n=103)

EAStrauma

NoviceInvestigator

ExperiencedInvestigator

Cohen'sKappa (95%CI)

Significanttrauma (≥4/6) (y/n)

0.30(0.04-0.56)

0.91(0.78-1.04)

Significantsingle slice defect (y/n)

0.22(0.12-0.31)

0.78(0.70-0.85)


ICC(95%CI)

Sumof significantly affected slices (0-6)

0.28(0.10-0.45)

0.90(0.85-0.93)

Defectangle on single slices (0o-180o)

0.22(0.15-0.30)

0.82(0.79-0.84)


Table2: Intraobserver comparison for percent agreement for significantEAS trauma (≥4/6)


ExperiencedInvestigator

NoviceInvestigator

SecondAssessment

FirstAssessment

Firstassessment

Yes(n)

No(n)

Total(n)

Yes(n)

No(n)

Total(n)

Yes(n)

11

1

12

5

6

11

No(n)

1

90

91

10

82

92

Total(n)

12

91

103

15

88

103


Percentagreement (98.9%), sensitivity (91.7%),specificity (98.9%)

Percentagreement (84.5%), sensitivity (33.3%), specificity (93.1%)


Conclusions: When analysisis performed by an experienced observer, the repeatability of TPUSmeasures for anal sphincter trauma seems to be high. The performanceof a novice is weaker, especially when compared to other measures offunctional pelvic floor anatomy obtained by post- processing analysisof archived volume data sets. Training and experience seem to beparticularly crucial for sphincter assessment by TPUS. Errors seem toaffect primarily the identification of the cranial EAS margin whichis crucial in order to avoid false positive results.
References:
1.Int Urogynecol J 2015: 26(11): 1667-72
2. Ultrasound ObstetGynecol 2015; 46: 363-66