Surgeons lack of agreement on determining preoperative radiographic and clinical shoulder balance in adolescent and adult idiopathic scoliosis patients.

Purpose: Preoperative shoulder balance is an important factor in determining the upper instrumented vertebrae (UIV). In adolescent and adult idiopathic scoliosis (AIS/AdIS) patients, we studied the intraobserver and interobserver reliability of spinal surgeons' assessment of preoperative shoulder balance using X-rays (XR) and anterior/posterior photographs.

Methods: An observational review of a prospective multicenter database (AIS Lenke Type 1/5/6) and prospective single-institution database (AdIS) was conducted. Ten spine surgeons reviewed AIS cases; 12 spine surgeons reviewed AdIS cases. Surgeons rated the higher shoulder: left/right/same/unsure. Reliability was calculated using Fleiss' kappa coefficient.

Results: Among 145 Type 1 AIS cases, intraobserver reliability was moderate-to-substantial: XR (κ = 0.59), anterior photographs (κ = 0.68), posterior photographs (k = 0.65). Interobserver reliability was fair to moderate for XR (κ = 0.31), anterior photographs (κ = 0.20), and posterior photographs (κ = 0.30). Among 52 Type 5/6 AIS cases, intraobserver reliability was substantial: XR (κ = 0.70), anterior photographs (κ = 0.76), posterior photographs (κ = 0.71). Interobserver reliability was fair to moderate for XR (κ = 0.49), anterior photographs (κ = 0.47), and posterior photographs (κ = 0.36). Among 66 AdIS cases, intraobserver reliability was substantial: XR (κ = 0.68), anterior photographs (κ = 0.67), posterior photographs (κ = 0.69). Interobserver reliability was moderate for XR (κ = 0.45), anterior photographs (κ = 0.43), posterior photographs (κ = 0.49). Within Type 1 AIS patients, attendings had better intraobserver reliabilities compared to fellows using X-rays (κ = 0.61 vs. 0.53), yet no effect of surgeon experience was seen with clinical photographs.

Conclusion: Though surgeons' ability to agree with themselves was moderate to substantial, surgeons' ability to agree with each other was fair to moderate. Combined measures to assess preoperative shoulder balance are needed for UIV selection.