Intra- and inter-observer reproducibility of r-ASRM endometriosis classification system and its components, using edited videotaped laparoscopic procedures

Several classification systems have been proposed for endometriosis but the search for the ideal system is ongoing. While the most commonly used system has historically been r-ASRM, this system is not fit for purpose, particularly for deep endometriosis. In order to explore strategies to devise a new system and learn from problems with the existing ones, this study was designed to assess the reproducibility of each component in r-ASRM and its total score using videotaped laparoscopic procedures. Two surgeons independently scored 64 edited videos of laparoscopic endometriosis procedure, twice. Using the Kappa statistic, the agreement of the scores given was analyzed. r-ASRM showed a moderate inter-observer agreement (ƙ = 0.503) and good intra-observer agreement (ƙ = 0.774 and 0.682 for scorer 1 and 2 respectively) for overall disease staging. The agreement for each component of the system, however, was highly variable. The least agreement was observed for the peritoneum with ƙ = 0.157 and ƙ = 0.362 respectively for inter-observer and intra-observer. The lowest intra-observer agreement was seen for cul-de-sac for scorer 2 (ƙ = 0.382). Whilst the overall rASRM shows acceptable agreement between two scorers, this agreement seems to be the product of inconsistent scoring for each component.