# Can Hertel Criteria Reliably Predict Avascular Necrosis After Intracapsular Proximal Humerus Fractures in the Elderly? A Retrospective Analysis

## TL;DR

This study evaluates how well Hertel's criteria predict avascular necrosis in elderly patients with proximal humerus fractures.

## Contribution

The study confirms Hertel's criteria are effective for identifying low-risk patients but highlights limitations in predicting high-risk cases.

## Key findings

- AVN occurred in 10.8% of patients, with a 24.7% rate in high-risk Hertel fractures.
- Hertel's criteria showed high sensitivity (95.5%) and NPV (99.2%) but low PPV (24.7%).
- Medial hinge disruption and calcar extension < 8 mm were strongest predictors of AVN.

## Abstract

Background/Objectives: Avascular necrosis (AVN) of the humeral head is a severe complication after intracapsular proximal humerus fractures in the elderly. Hertel’s radiographic classification is widely used to estimate ischemic risk, yet its real-world accuracy remains debated. Methods: We retrospectively analyzed 204 patients aged ≥65 years treated between 2019 and 2022 for intracapsular proximal humerus fractures. Fractures were classified according to Hertel’s criteria and the LEGO system. The incidence of AVN and its association with radiographic predictors were assessed. Diagnostic performance metrics (sensitivity, specificity, predictive values, accuracy) were calculated for Hertel’s classification. Results: AVN developed in 22 patients (10.8%). High-risk fractures according to Hertel’s criteria showed a 24.7% AVN rate versus 0.8% in low-risk fractures (p < 0.001; OR = 38.7). Hertel’s model demonstrated high sensitivity (95.5%) and negative predictive value (99.2%) but low positive predictive value (24.7%). Medial hinge disruption and calcar extension < 8 mm were the strongest radiographic predictors (p < 0.001). Conclusions: Hertel’s classification effectively identifies elderly patients at low risk for AVN, given its high sensitivity and NPV. However, its limited positive predictive value highlights the need for integrative models combining radiographic and clinical parameters to improve ischemic risk stratification.