Predictors of shoulder level after spinal fusion in adolescent idiopathic scoliosis

Jan Henrik Terheyden, Mark Wetterkamp, Georg Gosheger, Viola Bullmann, Ulf Liljenqvist, Tobias Lange, Albert Schulze Bövingloh, Tobias L. Schulte

For patients with adolescent idiopathic scoliosis, shoulder balance influences their treatment satisfaction and psychological well-being. Several parameters are known to affect postoperative shoulder balance, but few prognostic models are as yet available. This study aimed to identify independent predictive factors that can be used to assess preoperatively which patients are at risk of postoperative shoulder elevation, and to build a linear prediction model. N = 102 patients with all Lenke types were reviewed radiographically before surgery and 1 year afterward. The outcome measures were coracoid height difference (CHD), clavicular angle (CA), and clavicle-first rib intersection difference (CiRID). Predictive factors commonly used in the literature were investigated using correlation analysis and statistical testing. Significant contributing factors were included in three multiple linear regression models (for CHD, CA, and CiRID). The mean shoulder level (CHD) significantly changed from a lower left shoulder value of -8.5 mm before surgery to 3.3 mm at the follow-up examination. A high preoperative left shoulder level by CiRID, a large amount of Cobb angle correction of the distal thoracic curve, a low preoperative Cobb angle in the lumbar curve, and a structural proximal thoracic curve proved to be determinants and thus risk factors for left-sided shoulder elevation after surgery. The three models predicting CHD, CA, and CiRID at the follow-up examination included these four risk factors and were significant. Preoperative variables have the strongest influence on shoulder level after spinal instrumentation. Additionally, extensive correction of the distal thoracic curve can cause elevation of the left shoulder.