Proximal Junctional Kyphosis: Inter- and Intraobserver Reliability of Radiographic Measurements in Adult Spinal Deformity.

STUDY DESIGN: Reliability study of radiographic measures of proximal junctional kyphosis (PJK) in patients with adult spinal deformity (ASD). OBJECTIVE: To assess impacts of level of proximal endpoint and vertebral fracture on reliability of measurement of junctional kyphosis. SUMMARY OF BACKGROUND DATA: Radiographic assessment is important in determining management of patients with PJK or proximal junctional failure (PJF). No study to date has evaluated the reliability of radiographic measurement of the junctional kyphotic angle after surgery for ASD. METHODS: Postoperative radiographs from 52 patients with ASD were divided into four categories based on the level of the upper instrumented vertebra (UIV) and the presence or absence of PJF: upper thoracic without failure (UT), thoracolumbar without failure (TL), upper thoracic with PJF (UTF), and thoracolumbar with PJF (TLF). Nine surgeon reviewers performed radiographic measurements of kyphosis between UIV+2 and UIV twice at least 4 weeks apart. Intraclass correlation coefficients (ICC) were calculated to determine inter- and intraobserver reliability. RESULTS: Interobserver reliability for measurements of UT, TL, UTF, and TLF were all "almost perfect" with ICC scores of 0.917, 0.965, 0.956, and 0.882, and 0.932, 0.975, 0958, and 0.989, for sessions 1 and 2, respectively. Similarly, ICCs for kyphosis measurements for the TL and TLF group had "almost perfect" agreement with means of 0.898 (range: 0.817-0.969) and 0.976 (range: 0.931-0.995), respectively. ICCs for measurements for the UT and UTF groups all had "substantial" or "almost perfect" agreement with means of 0.801 (range: 0.662-0.942) and 0.879 (range: 0.760-0.988), respectively. CONCLUSION: The present study demonstrates high inter- and intraobserver reliability of PJK measurement following instrumented fusion for ASD, independent of the presence or absence of PJF. Although slightly lower for upper thoracic than for thoracolumbar proximal endpoints, all ICCs consistently reached at least "substantial agreement" and "near perfect agreement" for most. LEVEL OF EVIDENCE: 4.