Reliability of the Measurement of Glenoid Bone Defect in Anterior Shoulder Instability

0
0
0

Reliability of the Measurement of Glenoid Bone Defect in Anterior Shoulder Instability

Introduction

Anterior shoulder instability is a significant clinical concern, particularly in populations with high physical demands, where recurrent dislocations can lead to chronic joint dysfunction. The incidence of anterior shoulder dislocation in the general population is approximately 21.9 per 100,000 individuals annually, though this rate rises sharply in certain professions, reaching as high as 2,800 per 100,000. Despite non-operative management, recurrence rates remain alarmingly high, ranging from 58% to 100%. A critical factor contributing to recurrent instability is glenoid bone loss, observed in 90% of cases. Accurate quantification of this defect is essential for surgical decision-making, as the extent of bone loss influences the choice between soft tissue repair and bone augmentation procedures like the Latarjet technique. However, the reliability of existing methods for measuring glenoid defects remains contentious, necessitating further investigation.

Methods

This study evaluated 69 consecutive patients who underwent surgical treatment for recurrent anterior shoulder dislocation at a single institution. Preoperative 3-D computed tomography (CT) scans were obtained using a Toshiba Aquilion 80-slice scanner with standardized parameters (120 kVp, 125 mAs, 1 mm slice thickness). Three fellowship-trained shoulder surgeons independently performed glenoid defect measurements using Mimics (version 19.0) for 3-D reconstructions and ImageJ (version 1.52a) for quantitative analysis.

Two distinct en-face view protocols were employed:

  1. Self-Confirmed En-Face View: Each observer subjectively positioned the glenoid parallel to the screen, generating personalized en-face images.
  2. Designated En-Face View: An independent surgeon created standardized en-face views for all patients to minimize variability in glenoid orientation.

For each view, two defect metrics were calculated:

  • Areal Defect Ratio: The area of bone loss relative to the area of a best-fit circle drawn around the inferior glenoid contour.
  • Linear Defect Ratio: The chord length of the defect relative to the circle’s diameter, adjusted using the Pythagorean theorem.

Measurements were repeated after a 3-month interval to assess intra-observer reliability. Statistical analysis included intraclass correlation coefficients (ICCs) for inter- and intra-observer agreement and absolute difference calculations. ICC thresholds were categorized as very good (≥0.80), good (0.60–0.79), moderate (0.40–0.59), fair (0.20–0.39), or poor (<0.20).

Results

Key Findings

  • Linear vs. Areal Defect Magnitude: Linear defect measurements consistently exceeded areal values across all cases, reflecting inherent methodological differences.
  • Inter-Observer Reliability:
    • Self-Confirmed En-Face Views: Moderate reliability for areal defects (ICC = 0.557 and 0.513 for first and second measurements) and fair-to-moderate reliability for linear defects (ICC = 0.446 and 0.374).
    • Designated En-Face Views: Similar trends, with areal ICCs of 0.549 and 0.431 and linear ICCs of 0.402 and 0.327.
  • Intra-Observer Reliability: Higher than inter-observer reliability, with areal ICCs ranging from 0.585 to 0.783 and linear ICCs from 0.523 to 0.709.
  • Absolute Differences:
    • Inter-observer differences exceeded 5% in 24.6%–55.1% of designated views and 10.1%–26.1% of self-confirmed views. Maximum discrepancies reached 15.8% (inter-observer) and 13.2% (intra-observer).
    • Intra-observer variability above 5% occurred in 2.9%–23.2% of cases, with maxima of 13.2% and 9.8% for self-confirmed and designated views, respectively.

Discussion

Areal vs. Linear Defect Measurement

The study confirmed that areal measurements are more reliable than linear methods, aligning with prior literature indicating that linear calculations overestimate defects due to irregular glenoid contours. However, even areal measurements exhibited only moderate inter-observer agreement, underscoring the subjectivity of en-face view selection and best-fit circle placement. The variability in designated views (where glenoid orientation was standardized) further highlighted inconsistencies in defining the inferior glenoid circle, a step critical for defect quantification.

Sources of Variability

  1. En-Face View Determination: Subjective alignment of the glenoid during 3-D reconstruction introduced significant observer-dependent bias. Even minor rotational deviations could alter perceived defect dimensions.
  2. Best-Fit Circle Placement: Despite guidelines for fitting the circle to the inferior glenoid, subtle differences in contour interpretation affected both areal and linear calculations.
  3. Threshold Adjustments: Grayscale thresholding in ImageJ to isolate the glenoid area introduced another layer of subjectivity, particularly in cases with irregular bone density or partial volume effects.

Clinical Implications

A 5% discrepancy in defect size could alter surgical management (e.g., choosing Latarjet over Bankart repair). This study’s findings reveal that such discrepancies occur in up to 55.1% of cases when using designated views, emphasizing the need for standardized protocols. While intra-observer reliability was higher, the 13.2% maximum difference suggests that even experienced surgeons may inconsistently interpret the same data over time.

Methodological Recommendations

  1. Automated En-Face View Alignment: Software tools could standardize glenoid orientation using anatomical landmarks, reducing observer bias.
  2. Objective Best-Fit Circle Algorithms: Computational methods for circle fitting, rather than manual placement, may improve reproducibility.
  3. Integrated Measurement Platforms: Combining 3-D reconstruction and quantification within a single software interface could minimize steps prone to human error.

Conclusion

This study demonstrates moderate reliability for areal glenoid defect measurements and lower reliability for linear methods, reinforcing the superiority of area-based assessments. However, both approaches remain susceptible to inter-observer variability, particularly during en-face view alignment and best-fit circle determination. These findings highlight the urgent need for more objective, automated measurement tools to standardize preoperative planning and ensure optimal surgical outcomes.

doi.org/10.1097/CM9.0000000000000481

Was this helpful?

0 / 0