A Novel Scoring System Can Estimate the Risk of Posttraumatic Osteonecrosis After Femoral Neck Fracture Fixation in...

Femoral neck fractures in patients under age 65 years remain challenging because up to 21% of them will develop osteonecrosis of the femoral head (ONFH) after internal fixation, which can be painful and may result in conversion to arthroplasty. This complication is particularly burdensome for patients who still are employed. However, preoperative parameters associated with ONFH remain unclear, and no reliable risk-stratification tool is currently available.

(1) Which preoperative clinical and radiographic variables were retained in a risk-stratification model? (2) What were the discrimination and calibration of the 8-point Femoral Neck Fracture Osteonecrosis (FNF-ON8) scoring system developed from these variables? (3) Does this scoring system remain effective in patients with displaced fractures, and how did this system perform compared with the Garden system?

In this retrospective, multicenter, comparative study, seven surgeons at three centers in China treated 1288 patients age 18 to 65 years with femoral neck fractures using three cannulated screws between January 2012 and January 2023. At these centers, three cannulated screws generally were used in younger patients when femoral head preservation was intended and acceptable reduction with stable fixation was considered achievable. We excluded 14% (184 of 1288) because of incomplete preoperative imaging data and 7% (92 of 1288) because of < 2 years of follow-up, leaving 79% (1012 of 1288) for analysis. Of these, 47% (476) were male, 53% (536) were female, 74% (753) were age ≥ 45 years, and 21% (208) developed ONFH. Fifteen candidate variables were assessed using least absolute shrinkage and selection operator (LASSO)-penalized logistic regression, and retained variables underwent multivariable logistic regression refitting to estimate ORs and 95% CIs and derive the FNF-ON8 score. For exploratory risk stratification, the lower cutoff of 4 was selected to achieve approximately 95% negative predictive value (NPV), with scores of 0 to 3 points classified as low risk. The upper threshold of ≥ 6 points was selected by maximizing the Youden index, with scores of 6 to 8 points classified as high risk; scores of 4 to 5 points were classified as intermediate risk. Sensitivity, specificity, positive predictive value (PPV), and NPV were calculated for both thresholds. The area under the curve (AUC) of the FNF-ON8 score was compared with that of the Garden classification.

Five variables retained by LASSO regression were assigned points based on the multivariable logistic regression to construct the FNF-ON8 scoring system: age ≥ 45 years (OR 2.1 [95% confidence interval (CI) 1.4 to 3.2]; 1 point), displaced fracture (OR 7.9 [95% CI 5.1 to 12.1]; 3 points), comminution (OR 2.4 [95% CI 1.7 to 3.5]; 1 point), a superior subcapital fracture line (OR 3.5 [95% CI 2.0 to 6.1]; 2 points), and an inferior subcapital fracture line (OR 3.0 [95% CI 2.1 to 4.4]; 1 point; all p < 0.001). At the lower threshold of ≥ 4 points, sensitivity, specificity, PPV, and NPV were 0.95, 0.34, 0.27, and 0.96; at the upper threshold of ≥ 6 points, they were 0.75, 0.74, 0.42, and 0.92, respectively. Among patients with displaced fractures, scores of ≥ 6 were associated with greater odds of ONFH than scores of 0 to 5 (OR 5.5 [95% CI 3.2 to 9.5]; p < 0.001). The FNF-ON8 showed greater discrimination for ONFH than the Garden classification (AUC 0.80 [95% CI 0.77 to 0.83] versus 0.70 [95% CI 0.67 to 0.73]).

Among 1012 patients age 18 to 65 years treated with three cannulated screws under general anesthesia at three centers in China, ONFH developed in 4% (11 of 283) of patients with FNF-ON8 scores of 0 to 3 and in 42% (155 of 367) of those with scores of 6 to 8. That being so, the score may complement existing fracture classifications in preoperative risk counseling and follow-up planning. A prospective, multicenter study should externally validate the FNF-ON8 in an independent population and using standardized clinical and imaging follow-up for at least 2 years.

Level III, therapeutic study.