Table 1. Outcome measures, their level of measurement (patient or vertebral), definition and the statistical methods applied to analyze intra- and interobserver agreement and reliability.
Level | Outcome | Definition | Measure | |
Agreement | Reliability | |||
Patient | Fracture present | Fracture present (yes/no) | % absolute agreement | Cohen's kappa |
Worst fracture grade | Grade 0 = <20% height loss | % absolute agreement | Weighted kappa* | |
Grade 1 = 20–25% height loss | ||||
Grade 2 = 25–40% height loss | ||||
Grade 3 = >40% height loss | ||||
Cumulative fracture grade | Sum of all grades for all fractures, continuous scale | 95% Limits of Agreement | Intraclass Correlation Coefficient | |
Vertebral | Height loss | Measured height loss, expressed as percentage** | 95% Limits of Agreement | Intraclass Correlation Coefficient |
Fracture present | Fracture present (yes/no) | % absolute agreement | Cohen's kappa |
Square weighted Cohen's Kappa.
The fractured vertebra is compared to the nearest unfractured vertebra, with preference given to vertebrae cranial to (above) the fractured vertebra. The percentage of the worst height loss of each fractured vertebra (either anterior, middle or posterior part of the vertebral corpus), is given (see Figure 1).