2.1 Ethical Considerations

Ethical approval for this study was obtained from the ethics committee of Guangdong Provincial Hospital of Chinese Medicine (approval no. AF/04-07.0/10.0). Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

2.2 Study Population

This study was conducted in a 19-bed intensive care unit (ICU) of a university-affiliated tertiary-care hospital. Patients admitted to the ICU between March 2018 and September 2021 were evaluated for study inclusion. Sepsis was diagnosed according to the criteria outlined in the Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock [15], and intestinal dysfunction was diagnosed according to the criteria outlined in Gastrointestinal Function in Intensive Care Patients: Terminology, Definitions, and Management [16]. The exclusion criteria were: death within 24 h; patients who needed cardiopulmonary resuscitation; advanced malignant tumor; abnormal coagulation mechanism, such as hemophilia; pregnancy; recent abdominal trauma with the risk of digestive tract perforation; pacemaker implantation. In all, 255 patients were identified with an International Classification of Disease (ICD)-9 discharge diagnosis code of sepsis, and these patients were all diagnosed with acute bowel injury. To reduce confounding factors, 1:2 propensity score matching by sex, age, comorbidity, infection source, laboratory results, and AGI classification through multiple logistic regression analysis was used. The following patients were included after a review of their medical records. Patients (male or female) aged > 18 years and with S-AGI were included. For the EA treatment at ST36-RN4, sessions lasted 30 min and took place twice a day for 7 days. The pattern was set to a continuous wave and the current intensity was adjusted to the maximum tolerable level for the patient.

2.3 Patient Selection for EA

In the Second Affiliated Hospital of Guangzhou University of Chinese Medicine, EA is a routine diagnostic and treatment measure. Clinical doctors assess patients' eligibility for EA treatment, and patients sign a consent form for EA upon admission to the ICU. The selection of EA patients was based on historical medical records, and no active patient selection was carried out during the study.

2.4 Data Extraction

For all patients, the characteristics documented at baseline were age, sex, Acute Physiology and Chronic Health Evaluation II (APACHE II) score, comorbidities, infection source, laboratory results, and AGI classification. Data collection was completed by two independent authors.

2.5 Propensity Score Matching (PSM) and Covariate Selection

PSM was employed to enhance the robustness of our results using a 1:1 nearest neighbor-matching algorithm with a caliper width of 0.1. We used Stata module (FILLMISSING) to fill missing values in numeric or string variables. Considering the relatively low percentage of missing data (ranging between 0.3% and 5%), the missing data in vital signs and laboratory parameters (accounting for 5% of the total) were replaced with the median value. We used subgroup analysis for the sensitivity analysis in this study. The principle for selecting covariates was that they were not affected by treatment. Potential outcomes were independent of treatment assignment. Based on the research background, we chose the covariates sex, age, baseline comorbidity, infection source, laboratory results, and AGI classification.

2.6 Statistical Analyses

The patients were monitored for 28 days after enrollment or until death. Their baseline characteristics were assessed within 24 h before enrollment. The independent samples t-test was applied for continuous variables following a normal distribution (checked by the Shapiro–Wilk test with p > 0.05) and independent EA and conventional groups (e.g., age, ICU stay length); the paired t-test was used for paired data like pre- and post-treatment measurements within each patient; the Mann–Whitney U test was for non-normally distributed continuous variables (Shapiro–Wilk p < 0.05) and independent groups; the chi-square test was for comparing categorical variables (e.g., gender, comorbidities); and Fisher's exact test was used when the sample size was small and expected frequencies in contingency table cells were less than 5. Values are shown as mean ± SD. The Kaplan-Meier and log-rank tests were used to estimate differences in the cumulative incidence of mortality between the two cohorts. Cox proportional hazards regression analysis was performed to obtain the hazard ratios (HRs) with 95% confidence intervals (Cis) after adjusting for each covariate. Furthermore, subgroup analyses were undertaken to examine the consistency of our findings across different subgroups. p-values < 0.05 were considered statistically significant. Data were analyzed using SPSS software (version 25.0, USA) and STATA software (version 17.0, USA).

3.1 Participants and Baseline Characteristics

Figure 1 shows the screening and matching process for the study participants. A total of 50 S-AGI patients who received EA treatment were enrolled between 2018 and 2021, and a 1:2 propensity score was used to match a cohort of 100 participants as controls. The no-EA and EA cohorts showed no significant differences in the sex, age, baseline comorbidity, infectious source, laboratory results, and AGI classification. The demographic comparisons are presented in Table 1.

3.2 Relationship Between EA Use and S-AGI Mortality

In univariate analysis, there was an association between the use of EA and a significantly reduced 28-day mortality compared with the conventional cohort (HR = 0.52, 95% CI: 0.27–1, p = 0.05) (Table 2).

Figure 2 shows a significant intergroup difference in the 28-day survival rate: the survival rate of the EA group is significantly higher than that of the conventional group. A log-rank test was conducted to more accurately evaluate the difference between the two groups, and the results still showed a significant intergroup difference (p < 0.05).

We examined the HRs with 95% CIs in univariate and multivariate Cox regression analyses to investigate the potential association of EA with the prevalence of mortality within 28 days (Table 2). We observed that the HRs for EA use were consistently significant in all models (HR range 0.44–0.49, p < 0.05 for all). In comparison with the conventional cohort, patients who received EA treatment showed an association with a lower risk of 28-day mortality (adjusted HR = 0.46, 95% CI = 0.22–0.95, p = 0.04) after adjustment for age, sex, ICU stay days, AGI classification, diabetes mellitus, chronic kidney disease, cerebral vascular diseases, cardiovascular disease, chronic respiratory disease, tumor disease, APACHE II score, continuous renal replacement therapy, ventilation, white blood cell count, hemoglobin, platelets, arterial partial pressure of oxygen, lactate, and infection source.

3.3 Subgroup Analysis and Sensitivity Analysis

Subgroup analyses indicated the durability and dependability of the association between EA use and the outcome in the ICU. Specifically, the use of EA was found to be associated with a more substantial protective impact among individuals requiring mechanical ventilation and those classified as AGI III or IV. No other significant interactions were observed in the subgroups (P_interaction > 0.05) (Figure 3).

4.1 Main Findings

Our study presents a cohort analysis of the effect of EA use on mortality in patients with S-AGI. The findings suggest an association where patients with S-AGI who received EA had a reduced risk-adjusted 28-day mortality rate compared with those who did not receive EA. Notably, these results remained consistent even after adjusting for potential confounders such as demographic factors, laboratory results, infection source, severity-related variables, comorbidities, and treatments. We also analyzed the individual and combined effects of these covariates on the relationship between EA use and mortality, which will help to more comprehensively understand the impact of EA on the prognosis of S-AGI patients. Thus, our study provides compelling evidence supporting the potential advantages of EA use in enhancing outcomes for S-AGI patients.

4.2 Effect of Using EA on Mortality in Patients Suffering From S-AGI

This is the first retrospective trial investigating the effectiveness of a 7-day course of EA treatment in patients with S-AGI. In this study, we found that EA treatment was associated with a significant decrease in patients' mortality on day 28 (p < 0.05). It was also observed that EA use was associated with a reduced 28-day mortality rate among patients with S-AGI. Compared with previous studies [17-19], our study further adjusted for infection source, comorbidities, and treatments and still observed a significant effect of EA on reducing mortality, which provides more comprehensive evidence for the role of EA in S-AGI patients. EA at points ST36-RN4 can regulate the motility of the intestinal tract and promote passage of stool; simultaneously, it can also improve intestinal inflammation. The main therapeutic indication of EA at ST36 and RN4 is digestive system disease, and it performs the functions of dredging excretory organs and immune regulation [20]. EA at the acupoints of the Stomach Meridian at Foot Yangming can reduce the plasma levels of TNF-β and IL-6 in patients with sepsis and inhibit the progression of inflammation. EA at ST36 can shorten the recovery time of bowel sounds, ameliorate defects, and promote the recovery of gastrointestinal function in patients undergoing abdominal surgery [21, 22]. EA at ST36 can improve the gastrointestinal barrier membrane of mucus by improving neuroregulation, regulating the secretion of gastrointestinal hormones, enhancing blood supply to the gastrointestinal mucosa, and removing inflammatory mediators [19]. However, the specific anti-inflammatory mechanism of EA is still unclear. Therefore, our study shows an association where EA may be related to an improvement in the intestinal inflammation of patients with S-AGI and a reduction in the 28-day mortality of S-AGI.

The function of the gastrointestinal tract is closely related to the prognosis of sepsis [4, 23]. Imbalances in intestinal motility and destruction of the intestinal barrier are prone to cause translocation of bacteria and toxins in the intestine, aggravating intestinal infections and promoting the pathophysiological process of MODS, which leads to a vicious cycle of explosive infections and inflammatory factor storms [3, 24]. Researchers have gradually discovered that the gastrointestinal tract is not only a target organ for exacerbation of sepsis but also a triggering organ for further damage [23, 25]. Modern medical treatment mainly focuses on anti-infection, stomach protection, and intestinal microbiota, but it cannot fundamentally improve the gastrointestinal function of patients. Moreover, antibiotic resistance, the use of mechanical ventilation, and the deployment of artificial heart, lung, and kidney replacement technologies can potentially cause further damage to gastrointestinal function [26, 27].

EA reduces intestinal inflammation and improves the outcome of S-AGI. The underlying mechanism by which EA alleviates S-AGI intestinal inflammation is unclear. In an experiment involving EA treatment of S-AGI, EA at ST36 was shown to increase the expression of intestinal tight junction protein, reduce serum d-lactic acid level, and improve the permeability of the intestinal mucosal barrier [13]. EA at ST36 performs anti-inflammatory and antioxidant functions, accelerates the recovery of gastrointestinal function in patients with sepsis, and protects the intestinal mucosal barrier function [12]. Recent studies have revealed the intrinsic relationship between ferroptosis and acute intestinal injury. Ferroptosis can regulate the inflammatory cascade at multiple levels and stages, and is closely related to organ function damage in sepsis [28]. Ferroptosis is excessively activated in AGI, which is evidenced by a decline in expression of superoxide dismutase and glutathione peroxidase 4, and a decrease in the level of glutathione. Iron overload, reactive oxygen species, and malondialdehyde production contribute to this increase [29, 30]. These aspects should be verified in future basic experiments.

4.3 Strengths of Our Study

Our study had three main strengths. First, we employed a self-constructed database, ensuring the reliability and comprehensiveness of our data. Our study also used well-defined diagnostic criteria for S-AGI, enhancing the accuracy and validity of our findings. Different from previous studies that may have overlooked comorbidities and treatments, our study included infection source, diabetes mellitus, chronic kidney disease, cerebral vascular disease, cardiovascular disease, chronic respiratory disease, malignant tumor, ventilation and continuous renal replacement therapy in the models, which allows for a more comprehensive analysis of the relationship between EA use and mortality in individuals with S-AGI. Second, no prior research has specifically investigated the effects of EA use on mortality risk in S-AGI patients. Our study shows an association suggesting that the use of EA may be related to a decrease in the 28-day mortality among individuals with S-AGI. Third, we used multiple regression analysis to establish the robustness and reliability of our outcomes. This rigorous analytical approach bolsters the credibility and internal validity of our results.

4.4 Limitations of Our Study

Our study also had several limitations. A major limitation of this single-institute study is the relatively small sample size. Specifically, the treatment group consists of only 50 patients. The small sample size may not be representative of the broader population of S-AGI patients, and thus, the results may not be applicable to other clinical settings or patient cohorts. Another limitations of this study is its single-center nature. The research was carried out in a single traditional Chinese medicine hospital, which may lead to the results not being generalizable to other hospitals with different patient demographics, medical cultures, and treatment protocols. Future multi-center studies involving a more diverse range of hospitals, including both traditional Chinese medicine and Western-medicine hospitals, are needed to validate and expand our findings. In addition, this retrospective study had a potential selection bias. Therefore, more prospective studies with a larger cohort of subjects are needed to support the present findings. Another limitation was that there is no specific mechanism to explore, and further animal experiments are required to explore potential mechanisms.