Different definitions of feeding intolerance and their associations with outcomes of critically ill adults receiving enteral nutrition: a systematic review and meta-analysis

Abstract

Background: A unified clinical definition of feeding intolerance (FI) is urged for better management of enteral nutrition (EN) in critically ill patients. We aimed to identify optimum clinical FI definitions based on reported evidence.

Methods: We searched clinical studies comparing FI with non-FI with a clear definition, summarized the evidence by random-effect meta-analyses, and rated the certainty of evidence by the Grading of Recommendations Assessment, Development and Evaluation frameworks.

Results: Five thousand five hundred twenty-five records were identified, of which 26 eligible studies enrolled 25,189 adult patients. Most patient-centered outcomes were associated with FI overall. Low to very low certainty evidence established FI defined as large gastric residual volume (GRV) ≥ 250 ± 50 mL combined with any other gastrointestinal symptoms (GIS) had a significant association with high mortalities in particular all-cause hospital mortality (odds ratio [OR] 1.90, 95% confidence interval [CI] 1.40-2.57), the incidence of pneumonia (OR 1.54, 95% CI 1.13-2.09) and prolonged length of hospital stay (mean difference 4.20, 95% CI 2.08-6.32), with a moderate hospital prevalence (41.49%, 95% CI 31.61-51.38%). 3-day enteral feeding (EF) delivered percentage < 80% had a moderate hospital prevalence (38.23%, 95% CI 24.88-51.58) but a marginally significant association with all-cause hospital mortality (OR 1.90, 95% CI 1.03-3.50).

Conclusions: In critically ill adult patients receiving EN, the large-GRV-centered GIS to define FI seemed to be superior to 3-day EF-insufficiency in terms of both close associations with all-cause hospital mortality and acceptable hospital prevalence (Registered PROSPERO: CRD42022326273).

Trial registration: The protocol for this review and meta-analysis was registered with PROSPERO: CRD42022326273. Registered 10 May 2022.

Keywords: Critically ill adults; Definitions; Enteral nutrition; Feeding intolerance.

Fig. 1

PRISMA flow diagram. New studies were identified via databases and registers together with other methods for reporting of grey literature searches and results. The databases of CBM, CNKI, and WanFang are the main Chinese medical databases. FI feeding intolerance

Fig. 2

General summary plots. The number of pooled studies together with involved patients in two arms as well as ORs for all-cause mortality overall and by different FI definitions is summarized. Each node (solid circle) together with the number in it stands for the number of patients with FI (filled with orange) or non-FI (filled with blue) under a certain definition or overall situation with definition indiscriminately. The size of nodes is logarithmically proportional to the number of patients (i.e., the sample size in the FI or non-FI arm) involving the specific classification situation. The color of the arrows stands for different mean NOS score levels and the number right to the arrows as well as arrow thickness stands for the number of independent cohorts involving data pooling. The number string left to the arrows stands for pooled OR and its 95% CI when FI versus non-FI. *The number of cohorts exceeds the total number of included studies because some studies provided one more different FI definitions-based independent cohorts of eligible data for pooling. FI feeding intolerance, EF enteral feeding, GRV gastric residual volume, GIS gastrointestinal symptoms, NOS Newcastle–Ottawa Scale, No. number, OR odds ratio, CI confidence interval

Fig. 3

Data summary of relative effects overall and by different FI definitions. The relative effects overall and across 18 different kinds of FI definitions were measured as ORs for all-cause hospital mortality, long-term mortality, and incidence of pneumonia, as well as MDs for the length of hospital stay, and mechanical ventilation, along with 95% CIs. The color of each cell indicates the certainty of evidence according to the Grading of Recommendations Assessment, Development, and Evaluation. FI feeding intolerance, GIS gastrointestinal symptoms, GRV gastric residual volume, EF enteral feeding, No. number, OR odds ratio, MD mean difference, CI confidence interval

Fig. 4

Subgroup analyses according to FI definitions-related key elements. The certainty of the evidence was rated by the Grading of Recommendations Assessment, Development, and Evaluation criteria. GIS here no matter in candidate GIS or selected GIS to determine FI at least included large GRV. GIS is referred to as a large GRV alone or a large GRV combined with another one or any combination of symptoms including vomiting, absent bowel sounds, abdominal distension, and diarrhea. Data were expressed as ORs and MDs along with their 95% CIs. FI feeding intolerance, GIS gastrointestinal symptoms, GRV gastric residual volume, No. number, OR odds ratio, MD mean difference, CI confidence interval

Fig. 5

Data summary of absolute effects on all-cause hospital mortality, long-term mortality, and incidence of pneumonia overall and by FI definitions. Effects of FI and non-FI changed together in pairs with the different definitions of FI. Absolute effects of the FI in excess of non-FI were estimated overall or across 18 different FI definitions via a random-effect meta-analysis of rates. This pooled effect represents how many more patients with poor outcomes can expect to occur due to FI. *Axis starting point is -200 per 1000 events. FI feeding intolerance, GIS gastrointestinal symptoms, GRV gastric residual volume, EF enteral feeding, No. number, CI confidence interval

Fig. 6

Two-dimensional graphs of FI prevalence versus all-cause hospital mortality, long-term mortality, and the incidence of pneumonia overall and by different FI definitions. A All-cause hospital mortality versus FI prevalence. B All-cause long-term mortality versus FI prevalence. C The incidence of pneumonia versus FI prevalence. Effect sizes for FI by different definitions are represented by colored nodes, with bars representing the corresponding 95% CIs. FI prevalence by EF percentage < 20% or < 50% was not pooled due to their cohorts being completely covered by the same cohort with EF percentage < 80% in our study, if not, significant errors of selectivity were expected. FI feeding intolerance, EF enteral feeding, GRV gastric residual volume, GIS gastrointestinal symptoms, No. number, OR odds ratio, CI confidence interval