Combination of risk factors affecting retear after arthroscopic rotator cuff repair: a decision tree analysis

• Published in: Journal of shoulder and elbow surgery 2021-01, Vol.30 (1), p.9-15
• Main Authors: Harada, Nobuya, Gotoh, Masafumi, Ishitani, Eiichi, Kakuma, Tatsuyuki, Yano, Yuka, Tatara, Daisuke, Kawakami, Junichi, Imai, Takaki, Karasuyama, Masaki, Kudoh, Yu, Shiba, Naoto
• Format: Article
• Language: English
• Subjects: combination of risk factors; decision tree analysis; prediction probability; predictive retear factors; preoperative factors; Retear after arthroscopic rotator cuff repair
• ISSN: 1058-2746; 1532-6500
• DOI: 10.1016/j.jse.2020.05.006

Several risk factors for postoperative retear after arthroscopic rotator cuff repair (ARCR) have been cited in a large number of reports; various combinations of these seem to be present in the clinical setting. Using a combination model for decision tree analysis, we aimed to investigate the combination of risk factors that affect postoperative retear the most. A total of 286 patients who underwent magnetic resonance (MR) imaging at 6 months after surgery were included in this study. Based on the structural integrity of the MR images taken 6 months after surgery, the patients were divided into a healed group (intact tendon, 254 patients) and a retear group (ruptured tendon, 32 patients). Using univariate and decision tree analyses, we selected a combination of 11 risk factors that drastically affected postoperative retear. The mean age was 64.9 ± 7.1 years, and the mean symptom duration was 9.7 ± 11.6 months. The tear was small/medium in 177 patients and large/massive in 109 patients. The technique for surgical repair was single row in 42 patients, double row in 60 patients, and suture bridging in 216 patients. On univariate analysis, both groups had significant differences in the anteroposterior (AP) tear size (P < .0001), mediolateral tear size (P < .0001), hyperlipidemia (P = .0178), global fatty degeneration index (P < .0001), supraspinatus fatty degeneration stage (P < .0001), and critical shoulder angle (CSA) (P = .0015). All of these 5 risk factors, except for mediolateral tear size, were selected as candidates for the decision tree analysis. Eight combination patterns were determined to have prediction probabilities that ranged from 4.3% to 86.1%. In particular, the combination of an AP tear size of ≥40 mm, hyperlipidemia, and a CSA of ≥37° affected retear after ARCR the most. Decision tree analysis lead to the extraction of different retear factor combinations, which were divided into 5 retear groups. The worst combination was of AP tear size ≥40 mm, hyperlipidemia, and CSA ≥37°, and the prediction probability of this combination was 86.2%. Therefore, our data may offer a new index for the prediction of retear after ARCR.