Development of a nomogram to predict the need for more than one Nuss bar in pectus excavatum

Dear Editor,Figure 1.: (A) Nomogram predicting the need for more than one Nuss bar. (B) Calibration curve showing good agreement between predicted and observed probabilities, indicating excellent model fit and robust predictive performance. Pectus excavatum (PE), the most common congenital deformity of the anterior chest wall, leads to abnormal thoracic morphology and varying degrees of cardiopulmonary impairment[1]. It also exerts a significant psychosocial impact, particularly among children and adolescents[2]. The Nuss procedure has become the standard treatment for PE; however, achieving optimal individualized correction remains a major surgical challenge, particularly in determining whether to place a single or multiple Nuss bars[3]. Currently, this decision relies primarily on the surgeon’s subjective assessment of preoperative imaging, lacking objective and quantitative tools for guidance. Such subjectivity may result in overcorrection or undercorrection, especially in low-volume centers or less-experienced hands, thereby affecting outcomes and increasing the risk of reoperation. This study aimed to develop and validate a nomogram model based on clinical and radiological parameters to predict the required number of Nuss bars in patients with PE, providing a quantitative and intuitive tool to support preoperative decision-making. This study adheres to the Transparency In The reporting of Artificial INtelligence (TITAN) Guidelines 2025 for transparency and integrity in the use of artificial intelligence in research and publication[4]. Patients who underwent the Nuss procedure for PE at our institution between 2020 and 2023 were retrospectively reviewed, excluding those without preoperative computed tomography (CT) at our hospital or lost to follow-up. The number of Nuss bars was jointly determined by two senior pediatric thoracic surgeons after comprehensive evaluation. Clinical variables included age, height, weight, and body mass index (BMI). Radiological parameters were derived from quantitative CT measurements: sternovertebral distance (SVD), Haller index (HI), asymmetry index (AI), correction index (CI), and flatness index (FI). All measurements were performed at the level of maximal sternal depression. SVD was the shortest sternum-to-vertebra distance; HI and CI assessed deformity severity. AI quantified asymmetry from left–right hemithorax diameter differences. FI was the transverse-to-longitudinal diameter ratio, indicating chest flatness (Supplemental Digital Content Figure 1, available at: https://links.lww.com/JS9/G813). Continuous variables were expressed as mean ± standard deviation, and intergroup comparisons were conducted using independent-samples t-tests. To construct the predictive model, least absolute shrinkage and selection operator (LASSO) regression was first applied to all candidate variables to avoid multicollinearity and identify the most informative predictors[5]. The penalty parameter λ was selected using 10-fold cross-validation based on the minimum-λ criterion. Subsequently, Firth’s penalized logistic regression was used to develop the final model, minimizing bias due to the limited sample size and outcome imbalance. Based on this model, a nomogram was constructed to visually estimate the probability of requiring dual Nuss bars. Model discrimination was evaluated using the receiver operating characteristic curve and the area under the curve (AUC). Model calibration was internally validated using 1000 bootstrap resamples to obtain the bias-corrected AUC, and calibration performance was visualized using calibration plots generated with the rms package[6]. A total of 64 patients were included in the final analysis, comprising 46 patients who received a single Nuss bar and 18 who received dual bars. No complications attributable to the number of bars occurred in any patient, and all patients reported satisfaction with both the therapeutic outcomes and overall treatment cost. Differences were observed between the two groups (Table 1), particularly in age, height, weight, and CI (P < 0.05). Table 1 - Baseline characteristics and group comparison Variable Total (n = 64) One Nuss bar (n = 48) Dual Nuss bars (n = 16) P-value Age (days) 4535.30 ± 1016.18 4266.81 ± 990.67 5340.75 ± 582.38