This study retrospectively analysed the medical records of patients with HTG-AP with the aim to screen for independent risk and protective factors closely related to the severity of HTG-AP and detect markers suggestive of disease progression and prognosis within 24 h of admission. The study aimed to improve the treatment effect of the disease, predict disease severity more accurately, and provide a reference for clinical treatment. Ultimately, this study was conducted to achieve early identification of its tendency to become severe, early intervention, and a reduction in mortality.

Overall, 287 patients with HTG-AP who were hospitalised at the Gastroenterology Department of a Grade A tertiary hospital in Xiamen between January 2019 and December 2021 were selected. Data collected included pre-hospital (emergency, outpatient) and in-patient medical records. The Ethics Committee of Zhongshan Hospital, Xiamen University, approved this study (xmzsyyky Ethics No. 2023 − 139), and the requirement for informed consent was waived.

Patients were included if they met the following study inclusion criteria: (1) met the AP diagnostic criteria in the Guidelines for Diagnosis and Treatment of Acute Pancreatitis in China (2021) formulated by the Pancreatic Surgery Group of the Chinese Medical Association Surgery Society [5]; (2) had TG ≥ 11.3 mmol/L, or TG ≥ 5.65 and chylous serum; (3) underwent relevant examinations, including CT of the abdomen, pancreas, and chest, completed within 24 h after admission, and no other important observation indicators were missing; and (4) abdominal imaging showed no biliary calculi or obstruction.

Patients were excluded if they met the following exclusion criteria: (1) received systematic treatment in other hospitals, including but not limited to fluid resuscitation, plasma exchange, and other treatments, prior to admission; (2) acute or chronic diseases of the heart, coronary heart disease, chronic obstructive pulmonary disease, liver cirrhosis, and chronic renal failure; (3) complications with haematological or psychoneurotic diseases; (4) other clear aetiological types of AP, such as biliary, trauma, drugs, and abdominal surgery; (5) presence of clear or suspected infection elsewhere; (6) acute onset of chronic pancreatitis; and (7) incomplete clinical data.

According to the above inclusion and exclusion criteria, 21 patients were excluded from the study due to chronic renal insufficiency (one case of mild disease), appendix abscess (one case of mild disease), anaemia (three cases of mild disease), previous treatment in other hospitals (five cases of mild disease, three cases of moderate-severe disease, and five cases of severe disease), and incomplete clinical data (three cases). Ultimately, 266 patients with HTG-AP were included.

According to the Atlanta classification criteria [8], patients were classified as mild (MAP; n = 180), moderately severe (MSAP; n = 44), severe (SAP; n = 42), and critical AP (n = 0) with or without persistent OF, pancreatic/systemic infection, and local or systemic complications. Subsequently, the patients were divided into the HTG-SAP (N = 42) and hypertriglyceridemia non-severe acute pancreatitis (HTG-NSAP; N = 224) groups.

The general data of the two groups collected included sex, age, body mass index (BMI), history of alcohol consumption before the onset of disease, history of ordinary alcohol consumption, history of underlying diseases (diabetes mellitus and fatty liver), co-morbidities (diabetic ketoacidosis), number of episodes of pancreatitis, onset of disease to hospitalisation, hospitalisation time, and hospitalisation costs. The data were analysed by considering the highest or lowest values of symptoms, signs, and laboratory tests within 24 h of admission, including the abdominal pain score, presence or absence of psycho-behavioural or mental abnormalities, systolic and diastolic blood pressure (SBP and DBP), pulse rate (P), respiratory rate (R), white blood cell count (WBC), haemoglobin (Hb), HCT, platelet count, SIRS, number of items that meet the SIRS diagnostic criteria, amylase (AMY), lipase (LPS), the maximum value of albumin (ALBMax), the minimum value of albumin (ALBMin), difference between maximum and minimum albumin (dALB), total bilirubin (TBIL), alanine aminotransferase (ALT), aspartate aminotransferase (AST), TG, total cholesterol (TCHOL), blood glucose (GLU), lactate dehydrogenase (LDH), BUN, serum creatinine (Cr), bicarbonate ion (HCO3-), blood calcium (Ca2+), CRP, D-dimer (D-D), PCT, pH, arterial partial pressure of carbon dioxide, arterial partial pressure of oxygen, oxygenation index, blood base residual, plasma lactic acid, and interleukin-6. In this study, the numeric rating scale (NRS) was used. The NRS is a simple scale employed to evaluate pain intensity. Patients were asked to rate their pain degree on a scale ranging from 0 to 10, with 0 indicating no pain and 10 indicating the most severe pain. This scale is routinely used in assessments for patients with pancreatitis after admission. Three days before admission, the doctor-in-charge is responsible for measuring the pain score of patients daily.

The imaging data of all patients within 24 h of admission were reviewed. CT images of the abdomen, pancreas, and chest were re-reviewed, and the conditions of abdominal effusion and pleural effusion were collected. Additionally, severity scores, including the Ranson score (0–11 points), BISAP (0–5 score), and JSS prognostic index (0–9 score) were also collected.

SPSS 26.0 and R 4.1.2 software were used for statistical analyses. Among the collected laboratory test results, the following indicators were deleted due to missing values greater than 10%: BMI, acidity, arterial carbon dioxide partial pressure, arterial oxygen partial pressure, oxygenation index, blood alkali residual, plasma lactate, and interleukin-6.

The one-way analysis of variance (ANOVA) was performed first. Normally distributed measures were expressed as means ± standard deviation, and comparisons between groups were conducted using the independent sample t-test. Non-normally distributed measures were expressed as medians (lower quartile, upper quartile) [M (QL, QU)], and comparisons between groups were performed using the Mann–Whitney rank-sum test. Counts were expressed as the numbers of cases and percentages. The chi-square test was used for comparisons between groups. P < 0.05 was considered statistically significant.

Subsequently, the candidate predictors were further selected using the least absolute shrinkage and selection operator (LASSO) regression model. They were included in the binary logistic regression equation for multifactor analysis. The resulting independent predictors for HTG-SAP were used to build the regression model.

R 4.1.2 software was used to plot the column plots of the predicted HTG-SAP models; receiver operator characteristic (ROC) curves were used to determine and compare the area under the curve (AUC), the optimal cut-off value, and the sensitivity corresponding to the optimal cut-off value for the independent predictors and the predictive models. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) corresponding to the best cut-off value were calculated. ROC curves of the model and the BISAP, MCTSI, Ranson score, and JSS were established, and the AUC of the model was compared to that of BISAP, MCTSI, Ranson score, and JSS to determine and compare the AUC of each independent predictor and predictive model. To assess the discriminative ability of the model, the Hosmer–Lemeshow test was conducted, and the decision curve analysis (DCA) of the HTG-SAP model was plotted to assess the clinical practicability of the model. Finally, the Bootstrap method was used to repeat the sampling 1000 times for internal validation.