# Prediction model for unplanned extubation of thoracoabdominal drainage tube in postoperative inpatients: a retrospective study

## TL;DR

This study develops a model to predict unplanned removal of thoracoabdominal drainage tubes in postoperative patients, helping clinicians identify high-risk cases early.

## Contribution

The study introduces a validated predictive model for unplanned extubation using clinical factors and statistical validation methods.

## Key findings

- The model achieved an AUC of 0.897 in internal validation with high sensitivity and specificity.
- External validation showed an AUC of 0.839, confirming the model's generalizability.
- Key risk factors included drainage method, fixation method, and patient self-care ability.

## Abstract

It is crucial to identify the risk factors for unplanned extubation (UEX) in thoracoabdominal drainage tubes as early as possible and establish applicable risk prediction model to reduce the incidence of UEX.

A retrospective survey of patients who underwent Thoracoabdominal drainage tubes placement at a tertiary hospital was conducted in Zhejiang Province, China, between January 2020 and January 2023. A training set was established to build the predictive model and conduct internal validation, which was assessed for discrimination using ROC curves and for Calibration using the Hosmer–Lemeshow test and Calibration curves. A nomogram was constructed to visually present the results of the logistic regression analysis. An external validation dataset was created for assessing the external validation of the model.

a total of 2220 patients were enrolled. Multiple logistic regression analysis showed that negative pressure ball drainage, adhesive fixation method, self-care ability (self-care vs. complete dependence), self-care ability (partial dependence vs. complete dependence), and Thoracoabdominal drainage tubes were statistically significant factors associated with UEX (P < 0.05).The predictive model equation was as follows: a = 0.95–1.66 × drainage method + 2.45 × fixation method −4.17 × self-care ability (self-care vs. complete dependence) −2.79 × self- care ability (partial dependence vs. complete dependence).In the internal validation, the AUC was 0.897 (95% CI = 0.87–0.92; P < 0.001), with a sensitivity of 0.75 and specificity of 0.93, indicating a high level of discrimination for the model. The Hosmer–Lemeshow test yielded a chi-square (χ2) value of 2.823 with 8 degrees of freedom and a P-value of 0.945, indicating high accuracy of the model. In the external validation, the AUC was 0.839 (95% CI = 0.75–0.93; P < 0.001), with a sensitivity of 0.73 and specificity of 0.96. The Hosmer–Lemeshow test yielded a χ2 value of 12.85 with 8 degrees of freedom and a P-value of 0.117. The DCA plot shows that the DCA curve is consistently higher than the two extreme curves, indicating a good fit of the model.

The predictive model for the risk of unplanned extubation of thoracoabdominal drainage tubes in postoperative patients demonstrates good discrimination and Calibration. It can provide reference for clinical nursing staff in predicting the risk and early development of personalized preventive strategies for drainage tube UEX.