A total of 236 adult patients undergoing emergency surgery were prospectively and continuously included from April 2022 to February 2023 at the Second Affiliated Hospital of Fujian Medical University. We included patients meeting the following criteria: (1) non-pregnant adults undergoing emergency surgery; (2) American Society of Anesthesiologists grade I–II; (3) patients without complications such as hypertension, coronary heart disease and diabetes before surgery; and (4) patients without other severe systemic disease. We excluded patients undergoing chemoradiotherapy before surgery, patients with preoperative pyloric obstruction, patients with hypoproteinaemia and anaemia before surgery, patients undergoing total gastrectomy or exploratory laparotomy and patients admitted to the intensive care unit after surgery. This study was approved by the Ethics Committee of the Second Affiliated Hospital of Fujian Medical University and performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments.

A colour Doppler ultrasound diagnostic apparatus (Mindray M6, Shenzhen, China) with a convex array probe (frequency 2–5 MHz) was used to select the abdominal system imaging mode to detect the gastric antrum of the patient. The patient was asked to lay in the right decubitus position. Point-of-care gastric ultrasound is more effective in detecting gastric contents at the right decubitus position because the fluid and solid fluid mixture flow with gravity to the antrum, while the gas collects upward at the bottom of the stomach [17,18,19]. At this point, images of the gastric antrum could be continuously observed through the sagittal plane of the upper abdomen, and the probe was then placed in the subxiphoid region of the patient. The gastric antrum could be explored through the sagittal section, and the standard section was positioned behind the left liver and in front of the abdominal aorta. After the standard section was determined, the anteroposterior diameter (AP) and craniocaudal diameter (CC) of the antrum were measured, and the images were retained (Fig. 1). The formula for estimating the cross-sectional area (CSA) was as follows: [20].

A Schematic diagram of ultrasonic probe placement. B Ultrasound examination of the gastric antrum. GA, gastric antrum; L, liver; P, pancreas; SMA, superior mesenteric artery; CT, coeliac trunk; AO, aorta. C The CSA measurement is based on the anteroposterior diameter and craniocaudal diameter. CSA, cross-sectional area

$$(\mathrm}^)=(\mathrm\times \mathrm\times\uppi )/4$$

CSA was measured three times for each patient and averaged. The ultrasound examination was completed by a highly trained sonographer, and the obtained ultrasonogram was submitted to a sonographer with the title of associate senior or above for review.

The outcome index of this study was whether the patient had PONV; this was determined via follow-up with patients in the ward on the second day after surgery. The diagnostic criterion for PONV was the occurrence of postoperative nausea and/or vomiting within 24 h after surgery. The diagnosis of postoperative vomiting was mainly obtained through follow-up with the patient, the patient’s family, and the assigned nurse.

The diagnosis of postoperative nausea was obtained using a visual analogue score [21]; the scale plate was approximately 10-cm long and marked with a zero at one end and 10 at the other. Zero was classified as no nausea, and 10 was classified as intolerable nausea. Patients were asked to score the degree of nausea within 24 h after surgery, and postoperative nausea was defined as a score > 2.

The patient’s clinical and surgical data were recorded by accessing the electronic medical records system and anaesthesia system. The recorded items included patient sex, age, smoking history, alcohol history, PONV history, motion sickness history, migraine history, body mass index (BMI), duration of surgery, surgical position, mode of anaesthesia, type of inhaled anaesthetics, postoperative patient-controlled analgesia, intraoperative sufentanil dosage, duration of anaesthesia, and intraoperative use of neostigmine and glucocorticoids.

SPSS version 27.0.1.0 (SPSS Inc., Chicago, IL) and R-language 4.2.2 (R Foundation for Statistical Computing, Vienna,Austria) were used to analyse the data. The intraclass correlation coefficient (ICC) was used to assess the consistency of CSA between the same physician and other similarly qualified physicians. All patients were divided into the training cohort (n = 177) and the verification cohort (n = 59) in a ratio of 3:1, according to a random number table. Measurement data conforming to a normal distribution were expressed as the mean ± standard deviation (x̄ ± s), and quantitative data between the two groups were compared using an independent sample t-test. Non-normally distributed data were expressed as the median (interquartile range), and the Mann–Whitney U-test was used for comparisons between the two groups. Enumeration data were expressed as constituent ratios, and the chi-squared test was used to compare differences between the two groups.

The significance of each variable for PONV in the training cohort was evaluated by univariate logistic regression analysis. Variables with statistically significant differences in univariate logistic regression analysis were included in multivariate logistic regression analysis to identify independent risk factors related to the occurrence of PONV. The rms package of R version 4.2.2 was used to build a nomogram to predict PONV occurrence. The predictive performance of the nomogram was measured by the concordance index, and 1,000 bootstrap samples were drawn to decrease the overfit bias. For the application of the model, the probability of PONV in each patient was calculated based on the nomogram. The receiver operating characteristic (ROC) curve was used to calculate the optimal threshold, which was determined by the maximum Youden index (i.e., sensitivity + specificity – 1), and the accuracy of the optimal threshold was evaluated with the sensitivity, specificity, predicted value and likelihood ratio. The calibration curve, decision curve analysis (DCA) and clinical impact curve (CIC) were used to further evaluate the predictive efficacy, accuracy and clinical practicability of the model.

This study included interviews with patients undergoing emergency surgery at our hospital.