Characterizing irreparable: a retrospective machine learning analysis of patients who undergo primary labral reconstruction during hip arthroscopy

Abstract

Abstract Hip labrum reconstruction has been demonstrated to be a viable option for patients with irreparable labral tears. However, there is a lack of data analyzing patient and labral characteristics in those undergoing primary hip labral reconstruction. To use a machine learning technique to subcategorize patients who underwent labral reconstruction during primary hip arthroscopy and to determine if there may be varying pathology resulting in severe labral damage. Patients who underwent primary labral reconstruction between 2015 and 2018 were included. Patients with a prior ipsilateral hip surgery, who were unwilling to participate, or had incomplete preoperative data were excluded. Agglomerative hierarchical clustering analysis was conducted to identify the subgroups of patients. A comparison was performed for preoperative characteristics, intraoperative findings and procedures. Of the 191 patients who underwent primary labral reconstruction and were eligible, 174 were included in the clustering analysis. Two distinct groups were identified (Group 1: 112 patients, 64.4%; Group 2: 62 patients, 35.6%). Group 1 had a significantly higher proportion of females (61.6% versus 43.5%; P < 0.05), combined Seldes I and II labral tears (94.6% versus 54.8%; P < 0.05), and larger tears. Group 2 had a significantly higher rate of labral calcification (82.3% versus 3.6%; P < 0.05). The results of this study demonstrate two distinct groups of patients who underwent primary hip labral reconstruction: those with severe labral damage, and those with a calcified labrum. Approximately two-thirds were placed in the group with severe labral damage, while the other third had diminished quality secondary to calcific changes. Retrospective comparative trial; Level of Evidence, 3.