This prospective study was conducted in the intensive care unit (ICU) from April to August 2022.

Ethical approval was obtained from the local Institutional Review Board (Ethics Approval Committee number: 2022S424) and registered on the China Clinical Trial Registry (No. ChiCTR2200058373) and informed consent was obtained from each patient or their next-of-kin.

Patients undergoing mechanical ventilation were eligible if they had begun enteral feeding by a gastric tube more than 48 h previously, had developed feeding intolerance, and had received at least one prokinetic agent in the ICU. Patients were excluded if they had not begun enteral nutrition, had not developed feeding intolerance, had received enteral nutrition orally or by post-pyloric feeding at the beginning of treatment, did not have clear point-of-care ultrasonography (POCUS) images of the gastric antrum, or were expected to die within 48 h after admission to the ICU. Patients aged <18 years, pregnant patients, and patients who refused to participate in the study were also excluded.

Enteral feeding intolerance was defined as the presence of persistent vomiting/regurgitation or a high GRV resulting in forced interruption of enteral feeding. If any visible reflux of gastric contents occurred, vomiting/regurgitation was diagnosed. The GRV was considered high if it exceeded 200 mL in a single measurement (20, 21).

Prokinetic agents were considered effective if the energy to target energy ratio was ≥50% on day 7 of treatment (effective group). Prokinetic agents were deemed ineffective if the energy to target energy ratio was <50% on day 7 or if the patients transitioned to post-pyloric feeding after using the agents despite reaching the energy goal (ineffective group).

Enteral feeding was initiated as soon as possible (within 24–48 h) after ICU admission if there was no contraindication for enteral nutrition. Enteral nutrition was begun at 20 mL/h, and the feeding rate was gradually increased if the patients tolerated the feeding well; the feeding rate did not exceed 150 mL/h. If deemed appropriate based on daily clinical assessment, the patients were transitioned to a volume-based feeding strategy using the enhanced protein-energy provision via the enteral route feeding protocol (PEP uP protocol) during the next and subsequent day (22–24).

The daily caloric prescription was determined according to the patient’s clinical condition and aimed to achieve 25 kcal/kg per 24-h period based on the estimated dry weight of the patient at ICU admission while the prescription was adjusted by adding 25% of the difference between the estimated dry weight and the ideal body weight for obese patients (BMI ≥ 30) by using a standard formula to calculate the ideal body weight (25). The target protein intake was 1.2–1.5 g/kg of body weight per day.

We chose a semi-elemental, concentrated feeding solution (Enteral Nutrition Emulsion, TPF-T) that would be useful in both full volume and trophic fed patients (Wuxi, Jiangsu province, China). The dietitian could suggest changes after the protocol was started based on further assessment and to initiate a protein supplement.

The usage of prokinetic agents were decided by clinicians. Patients with EFI received either metoclopramide (10 mg administered as a 50-mL IV infusion over 30 min every 8 h) or domperidone (10 mg administered via tube feeding every 8 h) or mosapride (10 mg administered via tube feeding every 8 h) or the combination of two or three agents. The dose of IV metoclopramide was adjusted by the level of creatinine clearance (creatinine clearance ≤40 mL/min with 50% of normal dose and clearance ≤10 mL/min or undergoing dialysis or continuous renal replacement techniques with 25% of normal dose in patients) (26).

Point-of-care ultrasonography (POCUS) was used to measure the gastric antrum echodensity in all critically ill patients in this study. A physician with 3 years of experience in ultrasonography performed the scans before and after the administration of prokinetic agents. POCUS was performed using a curvilinear probe with specific parameters set according to a previous study (19).

The following was the procedure of examination and echodensity measurement in the study.

The following baseline characteristics of all critically ill patients were collected within the first 24 h after admission to the ICU: age, sex, height, weight, body mass index (BMI), reason for ICU admission, Acute Physiology and Chronic Health Evaluation II (APACHE II) score, and Sequential Organ Failure Assessment (SOFA) score. Laboratory examinations were also conducted; these included white blood cell and platelet counts; measurements of the concentrations of biochemical indexes (total bilirubin, hemoglobin, serum creatinine, glucose, and albumin), procalcitonin, C-reactive protein, interleukin-6, and lactic acid; and measurement of the activated partial thromboplastin time, prothrombin time, and D-dimer. All patients’ gastrointestinal function was estimated daily; this included determination of the AGI grade (I–IV), Gastrointestinal Failure (GIF) score, and Gastrointestinal Dysfunction Score (GIDS) (19, 20). Data regarding the daily energy and protein intake from the first to seventh days of using prokinetic agents and placing the nasointestinal tube were also collected. Other data collected included the use of dexmedetomidine, propofol, midazolam, fentanyl, sufentanil, remifentanil, or norepinephrine for sedation/analgesia; use of other vasoactive agents; 28-day mortality rate; hospital mortality rate; length of ICU stay; and length of hospital stay.

A sample size of 88 patients was estimated to provide 90% power at a 2-sided alpha of 5% to demonstrate a mean difference of 10 with a standard deviation for both groups of 13 based on our pilot study which demonstrated an 90% relative reduction in the mean difference of 10 (62 vs. 52) after accounting for 20% dropout (PASS v21.0.3, NCSS software, Kaysville, United States).

Continuous variables are expressed as mean ± standard deviation or median (interquartile range) and were compared using Student’s t-test or the Mann–Whitney U test. Categorical variables are expressed as frequency (percentage) and were compared with the chi-square test or Fisher’s exact test as appropriate. Receiver operator characteristic (ROC) curve analysis and the Youden index were used to determine the ability of the gastric antrum echodensity to discriminate the efficacy of prokinetic agents. The baseline characteristics were compared between high-risk and low-risk patients according to the optimal gastric antrum echodensity cutoff value of the ED₅₀ on the first day to distinguish the efficacy of the agents. The relationship between the efficacy of prokinetic agents and the patients’ clinical characteristics (including the gastric antrum echodensity) was assessed by logistic regression. Only variables with statistical significance (p < 0.05) were included in the multivariable models to identify those variables associated with the efficacy of prokinetic agents. To reach the final parsimonious model during the process of multivariable modeling, we performed forward elimination of variables that were not statistically significant. A two-tailed p-value of < 0.05 was deemed statistically significant. Subgroup analysis of the type of prokinetic agents was performed between the effective group and ineffective group. All statistical analyses were performed using Origin 2021 (OriginLab, Northampton, MA, United States), SPSS 26.0 (IBM Corp., Armonk, NY, United States), and MedCalc 20.1.0 (MedCalc Software, Ostend, Belgium).

From April to August 2022, 580 patients from the ICU were screened. Among them, 485 patients did not meet the inclusion criteria: Point-of-care ultrasonography imaging of the gastric antrum could not be performed because of abdominal surgery or abdominal bloating (n = 120), the patients were < 18 years old (n = 35), the patients stayed in the ICU for <48 h (n = 28), the patients were expected to die within 48 h after ICU admission (n = 15), the patients were readmitted to the ICU (n = 4), enteral nutrition was not initiated (n = 98), the patients did not develop feeding intolerance (n = 151), oral intake was initiated (n = 16), and enteral feeding was initiated by post-pyloric feeding (n = 18). Therefore, 95 patients who began enteral nutrition feeding by a gastric tube, developed feeding intolerance, and received prokinetic agents were followed up. With poor quality of images during the first day, 12 patients were excluded and 83 patients were performed in the analyses finally (Figure 1).

The most common cause of ICU admission were respiratory and neurologic diseases (n = 16, 19.28%) and the followed by sepsis (n = 14, 16.87%) and trauma (n = 11, 13.25%) among the patients. There were 11 patients of cardiovascular diseases and 16 patients for other diseases which were included thrombotic thrombocytopenic purpura, thermoplegia, SLE (systemic lupus erythematosus), leukemia, acute hepatic failure, CKD (chronic kidney disease), infections of laryngeal pharynges, infections with uncertain reason.

Among the 83 patients, 58 (69.88%) were men. The mean overall age was 56.14 ± 16.11 year, and the mean BMI was 23.43 ± 4.26 kg/m². The mean overall mean SOFA score and APACHE II score were 9.72 ± 3.71 and 17.29 ± 8.40, respectively. The 28-day mortality rate and overall in-hospital mortality rate was 20.48% (n = 17) and 30.12% (n = 25), respectively. Of the 83 patients, 45 patients’ energy to target energy ratio was <50% on the seventh day, while 38 patients’ energy to target energy ratio was ≥50%. Nineteen patients (42.22%) were transitioned to post-pyloric feeding.

According to the energy goal and the proportion of patients who were transitioned to post-pyloric feeding as predefined in the study protocol, the whole cohort was divided into an effective group (n = 38, 45.78%) and an ineffective group (n = 45, 54.22%) according to the efficacy of prokinetic agents. There was no significant difference between the effective group and the ineffective group in age (54.72 ± 18.30 vs. 57.41 ± 14.29 years, p = 0.451, difference 95% CI, −4.38, 9.76), BMI (23.12 ± 4.49 vs. 23.71 ± 4.08 kg/m², p = 0.280, difference 95% CI, −1.29, 2.45), APACHE II score (17.03 ± 7.41 vs. 17.52 ± 9.28, p = 0.790, difference 95% CI, −3.20, 4.20), or SOFA score (9.54 ± 3.72 vs. 9.89 ± 3.72, p = 0.672, difference 95% CI, −1.28, 1.97). The reason for admission to the ICU and the laboratory indicators between the two groups were shown in Table 1. The baseline characteristics are shown in Table 1.

The echodensity of gastric antrum was assessed using the above-described protocol in accordance with our previous study. Point-of-care ultrasonography images of the gastric antrum echodensity in the effective and ineffective groups are shown in Figure 2. The mean difference of ED50, ED85 and ED_mean were 49.88 ± 13.78 vs. 58.13 ± 14.48 (p < 0.001), 65.70 ± 16.05 vs. 74.81 ± 16.41 (p < 0.001), 51.76 ± 14.07 vs. 60.18 ± 14.31 (p < 0.001) for effective group and ineffective group during the observational period, respectively. The gastric echodensity on the first to third days was significantly different between the effective and ineffective groups (Figure 3).

The threshold of the ED₅₀ to distinguish effective from ineffective prokinetics was 58 with specificity of 48.85% [95% confidence interval (CI), 42.7–55.1%] and sensitivity of 80.83% (95% CI, 75.3–86.6%), and the ROC was 0.690 (95% CI, 64.8–73.0%; p < 0.001) (Figure 4A). The threshold of the ED₈₅ to distinguish effective from ineffective prokinetics was 69 with specificity of 58.56% (95% CI, 52.3–64.6%) and sensitivity of 73.28% (95% CI, 67.3–78.7%), and the ROC was 0.696 (95% CI, 65.4–73.6%, p < 0.001) (Figure 4B). The threshold of the ED_mean to distinguish effective from ineffective prokinetics was 60.22 with specificity of 51.71% (95% CI, 45.5–57.9%) and sensitivity of 80.58% (95% CI, 75.0–85.4%), and the ROC was 0.690 (95% CI, 64.8–73.0%, p < 0.001) (Figure 4C).

The mean differences in the ED₅₀, ED₈₅, and ED_mean before using prokinetic agents are shown in Table 2. The mean duration of using prokinetic agents was 9.97 ± 6.21 and 8.76 ± 6.38 days (p = 0.420) in the effective group and ineffective group, respectively (Table 2). When feeding intolerance symptoms were persistent, placement of a nasointestinal tube was considered even when using prokinetic agents; a tube was placed in 19 (42.22%) patients in the ineffective group (Table 2).

The differences in the AGI grade, GIDS, and GIF score were also compared between the two groups. A positive correlation was found between the gastric antrum echodensity and AGI (grade II or III) (p < 0.05) (Supplementary Figure S1). A positive correlation was also found between the gastric antrum echodensity and the GIDS (1 or 2) (p < 0.05) (Supplementary Figure S2) and the GIF score (1 or 2) (p < 0.05) (Supplementary Figure S3).

Furthermore, there were significant differences in the daily energy intake, daily protein intake, and daily volume of enteral nutrition during the observational period between the two groups (Supplementary Figure S4). The overall mean energy intake, protein intake, and volume of enteral nutrition for the 7-day period were 800.68 ± 449.78 kcal, 35.40 ± 21.66 g, and 791.87 ± 403.69 mL, respectively. The mean energy intake, protein intake, and volume of enteral nutrition in the effective vs. ineffective groups for the 7-day period were 16.21 ± 7.98 kcal/kg vs. 9.17 ± 6.43 kcal/kg (p < 0.001), 0.72 ± 0.38 g/kg vs. 0.42 ± 0.31 g/kg (p < 0.001), respectively.

The optimal ED₅₀ cutoff value for predicting the efficacy of prokinetic agents was 58. We then categorized the patients according to the ED₅₀ of the first day, before using prokinetic agents: the high-risk group (ED₅₀ > 58, n = 33) and the low-risk group (ED₅₀ ≤ 58, n = 50). Among all parameters, only the lactic acid level was significantly different between the two groups. The ED₅₀, ED₈₅, ED_mean, and the overall nutrition parameters during the observational period were significantly different between the groups, and their daily variations are shown in Figure 5. There was a significant between-group difference in the ratio of effectiveness of prokinetic agents (n = 32, 64.00% vs. n = 7, 21.21%; p < 0.001), but there was no significant difference in the incidence of placing a nasointestinal tube (n = 9, 18.00% vs. n = 10, 30.30%; p = 0.438). The baseline demographic information in the two groups is shown in Table 3.

In the univariate logistic regression analysis, the following were associated with the efficacy of prokinetic agents: ED₅₀ [odds ratio (OR), 1.051; 95% CI, 1.036–1.066], ED₈₅ (OR, 1.043; 95% CI, 1.030–1.056), ED_mean (OR, 1.050; 95% CI, 1.036–1.065), and age (OR, 1.017; 95% CI, 1.006–1.028). The univariate logistic regression analysis also showed that sepsis and cardiovascular disease as reasons for admission to the ICU, the use of sufentanil and fentanyl during the observational period, and the levels of serum creatinine, PCO₂, and sodium obtained at ICU admission were also associated with prokinetic agent efficacy. In the multivariable logistic regression analysis of prokinetic agent efficacy, a higher gastric antrum echodensity was more likely to be associated with ineffectiveness. The ORs of the ED₅₀, ED₈₅, and ED_mean in Models 1 to 3 were as follows. Model 1: ED₅₀ (OR, 1.059; 95% CI, 1.043–1.076; p < 0.001), ED₈₅ (OR, 1.041; 95% CI, 1.028–1.055; p < 0.001), and ED_mean (OR, 1.048; 95% CI, 1.033–1.064; p < 0.001). Model 2: ED₅₀ (OR, 1.049; 95% CI, 1.032–1.066; p < 0.001), ED₈₅ (OR, 1.043; 95% CI, 1.029–1.057; p < 0.001), and ED_mean (OR, 1.051; 95% CI, 1.035–1.067; p < 0.001). Model 3: ED₅₀ (OR, 1.049; 95% CI, 1.032–1.067; p < 0.001), ED₈₅ (OR, 1.042; 95% CI, 1.027–1.057; p < 0.001), and ED_mean (OR, 1.050; 95% CI, 1.032–1.067; p < 0.001). The other variables are shown in Table 4. And the logistic regression analysis of ED₈₅ and ED_mean were shown in Supplementary Tables S1, S2.

Considering that several different prokinetic agents were used in this cohort, a subgroup analysis of the type of prokinetic agents was performed. Supplementary Figures S5–S10 show the mean differences in the ED₅₀, ED₈₅, and ED_mean between the ineffective and effective groups according to subgroups of prokinetic agents. The energy intake, protein intake, and volume of enteral nutrition are also shown in Supplementary Figures S5–S10. The baseline characteristics are compared between the two groups in Supplementary Tables S3–S8.

The analysis of ROC for these risk factors such as sepsis, cardiovascular diseases were evaluated and showed in Supplementary Figure S11. There was no difference in Cardiovascular Diseases, Sufentanyl and Fentanyl in the analysis. However, some difference was existed in the level of sodium and PCO₂ and the sensitivity of ED₅₀, ED₈₅, and ED_mean were prior compared with sodium and PCO₂ with lower specificity than these parameters.