Severity of acute gastrointestinal injury grade is a predictor of all-cause mortality in critically ill patients: a multicenter, prospective, observational study

This prospective, observational, multicenter study was conducted in 14 general ICUs of Zhejiang Province from 1 March to 31 August 2014. The patients were recruited from 1 March to 30 April 2014. The last follow-up was completed on 30 June 2014; the remaining 2 months of the study were for data management and analysis. To be eligible, ICUs had to have a minimum of 12 beds and have a dedicated senior physician with adequate knowledge of clinical nutrition as well as a dietitian responsible for data collection. The AGI grade was assessed daily according to the recommendation of the ESICM grading system during the first week of the subject’s ICU stay. This system is based mainly on GI symptoms and intra-abdominal pressure (IAP) on days 1–3 of ICU admission, and it is concomitantly combined with FI and organ dysfunction on the remaining 4 days. At ICU admission, the nutritional target was set for all patients at 20 kcal/kg body weight/day within the first week of ICU admission.

EN was initiated according to current clinical practice guidelines for nutritional support in critically ill patients [13‐15]. The 2006 European Society for Clinical Nutrition and Metabolism guidelines on EN in intensive care on page 215 recommend that “during the acute and initial phase of critical illness, an exogenous energy supply in excess of 20–25 kcal/kg body weight/day may be associated with a less favorable outcome, while during the recovery phase, the aim should be to provide 25–30 total kcal/kg body weight/day” [13]. The 2012 Surviving Sepsis Campaign Guidelines suggest avoiding mandatory full caloric feeding in the first week, but they suggest low-dose feeding, advancing only as tolerated (grade 2B) [15]. Therefore, we routinely used a caloric goal of 80–100% of the caloric requirement (25–30 kcal/kg body weight/day after the first week of ICU admission), and only as tolerated the caloric target within the first week of ICU admission (20 kcal/kg body weight/day). The Harris-Benedict equation was also used to determine the caloric target for EN. In addition, the calorie goal for nonobese patients was calculated using the actual body weight, whereas the calorie goal for obese patients (body mass index >30 kg/m²) was calculated using the ideal body weight [14]. Supplemental parenteral nutrition (SPN) was added as described in the “Nutrition protocol” section below [16].

EN was administered continuously by the primary care team according to routine protocols that include semirecumbent positioning, preferred use of nasogastric tubes, and the use of prokinetic agents if necessary (gastric residual volume [GRV] ≥200 ml). EN products consisted of polymeric, fiber-enriched formulas containing 1.05–1.62 kcal/ml of energy (18% proteins, 29% lipids [8% medium-chain triglycerides], 53% carbohydrates).

Patients were screened for eligibility within 24 h of ICU admission. The inclusion criteria were (1) >18 years of age, (2) Acute Physiology and Chronic Health Evaluation II (APACHE II) score >8, and (3) expected to stay for at least 24 h in the ICU. The exclusion criteria were (1) AGI could not be evaluated for any reason, (2) advanced cancer, or (3) any terminal stage disease. In addition, the patients were excluded for delayed initiation of EN (>48 h) in the absence of contraindication to EN, evidence for intolerance of EN, or hemodynamic instability. When EN was not feasible, patients at low nutritional risk (Nutritional Risk Screening [NRS-2002] ≤3) who received early SPN (<4 days) or patients at high nutritional risk (NRS-2002 ≥ 5) or who were severely malnourished and received late SPN (>4 days) at admission to the ICU were also excluded from the analyses.

The study protocol was approved by the local ethics committee of each hospital. All patients or their legal representatives provided informed written consent according to the local ethics rules.

After ICU admission, if the patient had stable hemodynamics or had no EN contraindications, the patient was recommended to receive EN starting 24–48 h after ICU admission. The patients were required to remain in a semirecumbent position. EN was preferentially carried out using a nasogastric tube. The GRV was monitored every 4 h under the condition of indwelling nasogastric tube for a total of six measurements per day. Mean daily GRV was recorded. The EN infusion rate depended upon the total daily infusion of EN. The initial infusion rate was 25 ml/h, and the maximum target infusion rate was determined according to the patient’s feeding tolerance condition and the total daily infusion amount. If the patient reached feeding tolerance of EN (without severe abdominal distention, diarrhea, vomiting, or 4-h GRV <200 ml), then the rate was incrementally increased by 25 ml/h until the EN infusion rate could reach the target (with maximum infusion rate of 100 ml/h). If the GRV of the patients was >200 ml, then intestinal motility drugs were given to improve the feeding tolerance to EN and to strive to reach the calorie target within 48–72 h. After 2-h EN implementation, if GRV was >250 ml, then the initial EN infusion rate was maintained; 2 h later, the GRV was evaluated again, and if the GRV was <250 ml, then the rate was incrementally increased by 25 ml/h until the EN infusion rate could reach the target (with maximum infusion rate of 100 ml/h). If the patient had protein calorie malnutrition at ICU admission and EN could not be implemented, or if EN could not reach 60% of the nutrition target (20 kcal/kg body weight/day within the first week of ICU admission), then the patient was required to receive SPN from the fourth day of ICU admission [16].

A specific case report was used for data collection. Data regarding baseline demographic and clinical characteristics (clinical profile, admission category, presence/absence of sepsis, APACHE II score, Sequential Organ Failure Assessment [SOFA] score, and blood measurements) and nutritional status were collected within the first 24 h of ICU admission. The GI symptoms (vomiting/regurgitation, high GRV, abnormal bowel sounds, diarrhea, bowel distention, and GI bleeding), IAP (minimum, maximum, and mean daily values), and feeding details were documented each day. In addition, the type and amount of nutrition were recorded. Patients who survived were followed by telephone. An unfavorable outcome was defined as 28- and 60-day all-cause mortality after admission to the ICU.

GI symptoms and FI were predefined according to the recommendations of the ESICM [5]. The global AGI grade was determined on the basis of worst AGI grade within the first week of ICU admission and included GI symptoms, IAP, and FI. The IAP was measured via the bladder with patients in the supine position and using the closed-loop system repeated measurements technique. The IAP was measured at least twice daily in the presence of normal values and at least four times daily if IAP was found to be >12 mmHg. FI was considered if at least 20 kcal/kg body weight/day via the enteral route could not be reached within 72 h of a feeding attempt or if EN had to be stopped for any clinical reason (vomiting, high GRV, diarrhea, GI bleeding, or presence of enterocutaneous fistulas) [5]. If EN had to be interrupted for computed tomographic examination, endoscopy, tracheotomy, or any other intervention, then FI was not considered. EN was resumed as soon as possible after the intervention was completed. In the present study, GI bleeding was defined as any bleeding into the GI tract lumen confirmed by macroscopic presence of blood in vomited fluids, gastric aspirate, or stool, according to the recommendations of the ESICM Working Group on Abdominal Problems. GRV was assessed every 4 h after ICU admission for a total of six times per day. GRV monitoring was part of EN feeding. The maximum value within 1 day was assessed, as well as the mean daily value (representing the mean of all the single GRV measurements for 1 day).

SAS 9.13 software (SAS Institute, Cary, NC, USA) was used for database management and statistical analyses. Comparisons of means and proportions relied on the standard normal z-test and Fisher’s exact test, respectively. Continuous variables with a skewed distribution were normalized by logarithmic transformation and presented as geometric mean and 95% CI. Because of the relatively small sample size, AGI was handled as a binary variable as I/II vs. III/IV for the multivariate analyses. The prognostic value of the variables was assessed using a univariate Cox proportional hazards regression model. The variables with P values <0.10 were entered in a multivariate Cox proportional hazards regression to determine the independent predictors of all-cause mortality. Covariables in the baseline multivariate regression model included traditional clinical risk factors (age, source of ICU admission, sepsis, diabetes mellitus, coronary artery disease, use of vasoactive drugs, serum creatinine and lactate levels, mechanical ventilation, and APACHE II score). To further evaluate the additional prognostic value of different clinical variables, significant improvements in three different models with the AGI grade on the first day of ICU stay and FI within the first week of ICU stay, respectively, were assessed by the likelihood ratio test, based on the comparison of the χ² value of each model. Kaplan-Meier survival analysis was performed to estimate the cumulative survival. Survival rates of different subgroups of patients stratified on the basis of different AGI grades were compared using the log-rank test. P < 0.05 was considered to indicate statistical significance.