Scholarlink

Background <p>Fracture-related infection (FRI) and subsequent persistent infections significantly affect fracture surgeries. We aimed to develop a precise, personalized risk calculator to assist orthopedic surgeons in evaluating perioperative FRI risk.</p> Methods <p>Data from 36,087 patients across two medical centers and four regional hospitals were analyzed. We assessed 29 risk factors, including patient characteristics, comorbidities, fracture location, and surgical variables using multivariable logistic regression. Each factor was weighted based on its regression coefficient. Discrimination and calibration were assessed with optimism-corrected AUC and Brier scores (1 000-fold bootstrap) and calibration plots.</p> Results <p>FRI occurred in 2396 patients (6.64%), and 453 patients (1.26%) experienced persistent infections. The top 10 risk factors included male sex, open fractures, tibiofibular fractures, ankle/foot fractures, operative time, hospital stay, peripheral vascular disease, diabetes, chronic kidney disease, and psychotic disorders. Optimism-corrected AUCs for predicting FRI and persistent infections were 0.781 [<i>p</i> &lt; 0.001, 95% confidence interval (CI) 0.772–0.791] and 0.801 (<i>p</i> &lt; 0.001, 95% CI 0.779–0.823), respectively. Optimal cutoff scores for predicting FRI and persistent infections were 213 (sensitivity 0.638, specificity 0.796) and 232 (sensitivity 0.658, specificity 0.821). Calibration plots demonstrated good predictive performance (mean absolute errors: FRI 0.006, persistent infection 0.006). Brier scores were 0.055 (FRI) and 0.012 (persistent infections), indicating good accuracy.</p> Conclusions <p>The FRI risk calculator showed good predictive abilities, with optimized cutoff points aiding perioperative planning and preventive measures. Patient engagement in understanding of infection risk can improve treatment outcomes. Limitations include participant biases and retrospective design; prospective external validation is recommended.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Development and evaluation of a risk calculator for fracture-related infection after open reduction and internal fixation: a multi-institutional study

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年10月27日

查看全文
  • Chin-Yi Liao,
  • Yu-Der Lu,
  • Yu-Jui Chang,
  • Feng-Chih Kuo,
  • Chi-Hsiang Hsu,
  • Shan-Ling Hsu

摘要

Background

Fracture-related infection (FRI) and subsequent persistent infections significantly affect fracture surgeries. We aimed to develop a precise, personalized risk calculator to assist orthopedic surgeons in evaluating perioperative FRI risk.

Methods

Data from 36,087 patients across two medical centers and four regional hospitals were analyzed. We assessed 29 risk factors, including patient characteristics, comorbidities, fracture location, and surgical variables using multivariable logistic regression. Each factor was weighted based on its regression coefficient. Discrimination and calibration were assessed with optimism-corrected AUC and Brier scores (1 000-fold bootstrap) and calibration plots.

Results

FRI occurred in 2396 patients (6.64%), and 453 patients (1.26%) experienced persistent infections. The top 10 risk factors included male sex, open fractures, tibiofibular fractures, ankle/foot fractures, operative time, hospital stay, peripheral vascular disease, diabetes, chronic kidney disease, and psychotic disorders. Optimism-corrected AUCs for predicting FRI and persistent infections were 0.781 [p < 0.001, 95% confidence interval (CI) 0.772–0.791] and 0.801 (p < 0.001, 95% CI 0.779–0.823), respectively. Optimal cutoff scores for predicting FRI and persistent infections were 213 (sensitivity 0.638, specificity 0.796) and 232 (sensitivity 0.658, specificity 0.821). Calibration plots demonstrated good predictive performance (mean absolute errors: FRI 0.006, persistent infection 0.006). Brier scores were 0.055 (FRI) and 0.012 (persistent infections), indicating good accuracy.

Conclusions

The FRI risk calculator showed good predictive abilities, with optimized cutoff points aiding perioperative planning and preventive measures. Patient engagement in understanding of infection risk can improve treatment outcomes. Limitations include participant biases and retrospective design; prospective external validation is recommended.