# Construction and validation of a risk-prediction model for hypovolemic shock after cytoreductive surgery in patients with ovarian cancer: a retrospective study

## TL;DR

This study creates a model to predict hypovolemic shock risk after ovarian cancer surgery, helping doctors intervene early.

## Contribution

A novel visual prediction model for hypovolemic shock after cytoreductive surgery in ovarian cancer patients is developed and validated.

## Key findings

- The model includes operation duration, blood loss, albumin, fibrinogen, and sedative use as key predictors.
- The model showed strong predictive accuracy with an AUC of 0.800 in training and 0.821 in validation.
- Calibration and decision curve analysis confirmed the model's reliability and clinical utility.

## Abstract

Ovarian cancer patients undergoing cytoreductive surgery are prone to hypovolemic shock in the early postoperative period, resulting in tissue hypoperfusion, lactic acid accumulation, endotoxin displacement, and even multiple organ dysfunction syndrome; however, in existing studies, there is a lack of a dynamic approach to assess the risk of postoperative hypovolemic shock. This study aimed to construct and validate a visual prediction model of hypovolemic shock after cytoreductive surgery.

This is a retrospective observational study. Patients with ovarian cancer who received cytoreductive surgery at Zhejiang Cancer Hospital between January 2023 and June 2024 were retrospectively enrolled and divided into a training group and a validation group. Independent predictors of hypovolemic shock were identified using least absolute shrinkage and selection operator (LASSO) regression from the training set, and a nomogram was constructed based on these predictors. A nomogram was used to depict the weight of each variable in the logistic regression model on the event occurrence. The performance of the nomogram was assessed using receiver operating characteristic (ROC), calibration curve, and decision curve analysis (DCA).

A total of 423 patients were eligible for inclusion in this study. There were 301 cases in the training group and 122 cases in the validation group. This visual prediction model was constructed based on the duration of the operation (odds ratio (OR) = 1.273; 95% confidence interval (CI) [1.052–1.552]), the amount of blood lost during the operation (OR = 1.102; 95% CI [1.020–1.196]), the amount of albumin (OR = 0.935; 95% CI [0.879–0.993]) and fibrinogen (OR = 0.606; 95% CI [0.371–0.948]) immediately after the operation, and the postoperative use of sedative drugs (OR = 2.248; 95% CI [1.109–4.538]). The area under the ROC of the nomogram for the training and validation cohorts was 0.800 (95% CI [0.740–0.860]) and 0.821 (95% CI [0.735–0.907]), respectively. The predicted probabilities of the two groups of models were basically consistent with the actual incidence rates, with the average absolute errors being 0.016 and 0.019, respectively. The Hosmer-Lemeshow test of the training group (P = 0.722) and the validation group (P = 0.565) showed that there was no significant difference between the predicted values and the actual values, indicating that the model fitted well. The results of the DCA curve showed that the two groups had a net benefit within the probability range of risk threshold values of 0.01 to 0.84 and 0.04 to 0.80, respectively.

The model constructed in this study demonstrates improved predictive accuracy for the risk of hypovolemic shock after cytoreductive surgery in patients with ovarian cancer, and it holds the potential to provide a basis for medical staff to achieve early identification and timely intervention.