Construction and validation of a risk prediction model for postoperative nausea and vomiting in craniotomy patients
TL;DR
A risk prediction model for PONV in patients who underwent craniotomy is successfully developed and validated to have good predictive efficacy, which is helpful in helping healthcare professionals identify high-risk patients with PONV after craniotomy at an early stage and lays the foundation for the precise prevention of PONV.
Abstract
Postoperative nausea and vomiting (PONV) is one of the most distressing symptoms experienced by patients. The incidence is high, and complications are dangerous. Therefore, it is important to construct a prediction model to enhance the risk assessment strategies. This study included developmental and validation cohorts. The development cohort consisted of patients who underwent craniotomy between January 1, 2021, and December 31, 2022. Data were retrospectively collected to identify the factors associated with PONV events and to construct a risk prediction model. The validation cohort comprised patients from September 1, 2023, to December 22, 2023, at the same hospital. We collected data prospectively, compared the forecasted outcomes with the actual outcomes, and assessed the external validation performance of the model using model evaluation metrics, such as the area under the receiver operating characteristic curve (AUROC), Brier score, and calibration curves. The incidence of PONV within 24 hours after surgery was 27.2% (181 patients). The independent predictors of PONV after craniotomy were sex, age, preoperative sodium level, surgical site, and postoperative intracranial air accumulation. The predictive model had an AUROC of 0.862 (95% confidence interval = 0.829–0.894, P < .001), a sensitivity of 73.5%, a specificity of 86.8%, and an accuracy of 83.2%. External validation revealed an AUROC of 0.832 (95% confidence interval = 0.744–0.920), a sensitivity of 73.2%, a specificity of 90.3%, and an accuracy of 83.5%. Its reliability has been demonstrated in various clinical settings. We successfully developed a risk prediction model for PONV in patients who underwent craniotomy. This tool has been validated to have good predictive efficacy, which is helpful in helping healthcare professionals identify high-risk patients with PONV after craniotomy at an early stage and lays the foundation for the precise prevention of PONV.