Impact of Intravenous Fluids and Enteral Nutrition on the Severity of Gastrointestinal Dysfunction: A Systematic Review and Meta-analysis.

INTRODUCTION: Gastrointestinal dysfunction (GDF) is one of the primary causes of morbidity and mortality in critically ill patients. Intensive care interventions, such as intravenous fluids and enteral feeding, can exacerbate GDF. There exists a paucity of high-quality literature on the interaction between these two modalities (intravenous fluids and enteral feeding) as a combined therapy on its impact on GDF. AIM: To review the impact of intravenous fluids and enteral nutrition individually on determinants of gut function and implications in clinical practice.
METHODS: Randomized controlled trials on intravenous fluids and enteral feeding on GDF were identified by a comprehensive database search of MEDLINE and EMBASE. Extraction of data was conducted for study characteristics, provision of fluids or feeding in both groups and quality of studies was assessed using the Cochrane criteria. A random-effects model was applied to estimate the impact of these interventions across the spectrum of GDF severity.
RESULTS: Restricted/ goal-directed intravenous fluid therapy is likely to reduce 'mild' GDF such as vomiting (p = 0.03) compared to a standard/ liberal intravenous fluid regime. Enterally fed patients experienced increased episodes of vomiting (p = <0.01) but were less likely to develop an anastomotic leak (p = 0.03) and peritonitis (p = 0.03) compared to parenterally fed patients. Vomiting (p = <0.01) and anastomotic leak (p = 0.04) were significantly lower in the early enteral feeding group.
CONCLUSIONS: There is less emphasis on the combined approach of intravenous fluid resuscitation and enteral feeding in critically ill patients. Conservative fluid resuscitation and aggressive enteral feeding are presumably key factors contributing to severe life-threatening GDF. Future trials should evaluate the impact of cross-interaction between conservative and aggressive modes of these two interventions on the severity of GDF.

Introduction

Gut dysfunction (GDF) is a common problem in critically ill patients. It is the leading cause of multiple organ dysfunction syndrome/failure (MODS/MOF) and a significant cause of mortality and morbidity in critically ill patients [1]. In addition to this, the treatment of acute and critical illness can exacerbate GDF. Commonly used ICU interventions such as intravenous fluid resuscitation, early aggressive enteral feeding and vasopressor therapy are key factors leading to a secondary gut injury. In critical illness, intravenous fluid is the mainstay of early management for hemo-dynamic instability. It is vital to resuscitate a patient before commencing vasopressor therapy, particularly to delay the onset of an ischemic insult commonly occurring in hemodynamically unstable patients [2]. On the flip side, over-resuscitation can lead to bowel oedema leading to an ileus, while under-resuscitation with persistent splanchnic and peripheral vasoconstriction can trigger intestinal mucosal ischemia [3]. Although, enteral nutrition is the preferred approach to meet nutritional and modest fluid requirements in these patients, the delivery of early but aggressive enteral nutrition (EN) in hemodynamically unstable patients can precipitate the development of severe GDF, potentially leading to non-occlusive mesenteric ischemia which increases the chance of sepsis, multi-organ failure and mortality [4]. Intravenous fluid and enteral nutrition are two sides of the same coin and play a crucial role in determining the outcome of GDF if used wisely. However, very few studies have evaluated the role of these two modalities, thus making it difficult to understand their relationship with relevance to the severity of GDF. The aim was to review the evidence of the impact of intravenous fluid resuscitation and enteral nutrition individually on determinants of gut function and the implications in clinical practice.

Methods

Search Criteria and Study Identification

Electronic databases (MEDLINE and EMBASE) were searched using keywords on ‘gastrointestinal dysfunction in adult intensive care unit (ICU) /surgical patients on enteral feeding and intravenous fluids. The databases screened for all publications from the earliest available until 16th October 2018 (Appendix A).

Randomised controlled trials were searched by applying the keywords. Any additional studies on the impact of ‘intravenous fluid’ and ‘enteral feeding’ were included in the screening for the systematic review and meta-analysis. The search identification, screening and selection were conducted by the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow chart (Fig1) [5]. The study selection criteria were as follows.

The inclusion criteria were:

Study design: all randomised controlled trials (intravenous fluids and enteral feeding on GDF);

Study population: Adult surgical and critically ill patients

Disease state: critical illness and postoperative conditions

Intervention: enteral feeding: route of feeding (enteral vs parenteral); timing of feeding (early vs delayed); feeding vs nil-by-mouth and intravenous fluids: restricted vs liberal regime, goal-directed vs standard/conventional, low-infusions vs high-infusion or controlled vs rapid fluid therapy; intravenous fluids type: crystalloid fluid (normal saline or plasmalyte or ringer’s lactate) or colloid fluid (hydroxyethyl starch, albumin, gelofusion).

Study outcome: the occurrence of gastrointestinal dysfunction

The studies were excluded if they were:

non-ICU or non-surgical patients

paediatric population

animal studies

published in non-English languages

conducted on healthy volunteers

non-randomized trials (intravenous fluid therapy and enteral feeding)

not relevant to either of the interventions planned to study pattern of feeding (bolus vs continuous), comparative feed compositions (standard vs immune-enriched), related routes of feeding (nasogastric vs nasojejunal or jejunal) and studies addressing medications (e.g. prokinetic therapy).

Data Extraction: Data were extracted and independently recorded by two authors using predesigned data collection forms on Microsoft Excel.

Study characteristics included baseline demographic data such as author, publication year, study setting (ICU or surgical ward), admission diagnosis, study population, the total number of patients, fluid or enteral feeding interventions applied to experimental and control groups. The effect of fluid therapy and enteral feeding on GDF was analysed by separating the severity of GDF outcomes: 1) mild to moderate and 2) moderate to severe. All studies were stratified into the Clavien-Dindo classification [6] depending on the variability of clinical aetiology and interventions applied. Any additional studies derived from other sources and reference lists of included articles were screened and included if relevant. Data were independently reviewed and cross-checked by two authors (V.A. and A.B.). Any inconsistencies or disagreements were discussed between the two authors (V.A. and A.B.), and differences of opinion were further clarified by the senior author (J.A.W.).

Methodological quality

The methodological quality of included randomised controlled trials was assessed according to the Cochrane recommendations (The Nordic Cochrane Centre, The Cochrane Collaboration, 2008) [7]. These included systematic differences between groups (selection bias and performance bias), blinding of study participants and assessors, sequence allocation and concealment of allocated groups, the validity of findings and data withdrawal, incomplete outcome data (attrition and detection bias), and differences between data reporting or unreported data. The risk of bias assessment was presented according to the Cochrane collaboration recommendations. The overall quality of the study was graded as ‘poor’, ‘fair’ and ‘good’ based on the classification in the Cochrane’s quality assessment tool.

Statistical Analysis

All data were presented as the number of episodes of GDF in patients. Data analysis and interpretation were performed using Revman 5.3 (Revman, Version 5.3 for Windows; Copenhagen, Denmark: the Nordic Cochrane Centre, The Cochrane Collaboration, 2008) [7]. The nature of the analysis was not suitable for a pooled data analysis. Within each class of interventions (intravenous fluid and enteral feeding), a meta-analysis of GDF events was performed. Quantitative data meta-analysis was performed with at least two studies reporting on GDF as the primary or secondary outcome. Studies that did not have GDF as a primary or secondary outcome were excluded from the meta-analyses (Fig 1).

Heterogeneity was assessed by using I2 and classified as < 25% - low ; 25 – 50% - moderate and > 75% as high heterogeneity (heterogeneity and subgroup analysis in Cochrane consumers and communication group reviews) [8]. Regardless of the presence or absence of heterogeneity, a random-effects model was used to provide the most conservative estimate. Pooled effects for classes of interventions were calculated as weighted mean difference (MD) with 95% confidence interval (CI). P-value < 0.05 was considered statistically significant for all analyses. Ethical approval was not necessary for a review of published trials.

Results

Study Selection and Characteristics

A total of 103 studies including intravenous fluids (n = 46) and enteral feeding (n = 57) were eligible for inclusion in the systematic review, of which 43 (n = 22 intravenous fluid; n = 21 enteral feeding) studies were included in the final meta-analyses.

In studies on intravenous fluid therapy [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54], 46 randomised controlled trials’ including 20,780 patients were systematically reviewed, of which 22 studies (n = 2696) were included in the final meta-analysis. Ten studies included mechanical ventilated critically ill patients, and the remaining 36 studies included post-operative patients. The intervention group received either restricted, goal-directed, low-infusion fluids or a controlled-expansion fluid regime given as crystalloid fluid (normal saline or plasmalyte) or colloid fluid (hydroxyethyl starch). The control group included standard, liberal, conventional, high-infusion fluids or rapid-expansion fluid regimes given as crystalloid fluids (ringers lactate, plasmalyte and saline). Five studies compared more than two groups of fluid regimes. Fifteen studies included critically ill, trauma and surgical patients with a grading of IV as per the Clavien-Dindo classification (Appendix B). The remaining studies included postoperative and acutely ill patients with Clavien-Dindo grading of II and III (Tables 1-3).

Table 1

Study Characteristics of 'good' quality studies on the impact of intravenous fluid therapy on gut dysfunction included in the systematic review

Author	Year	Study Population	Study Setting	Study type	Study patients	Admission diagnosis	Experimental	Intravenous fluid	Control	Intravenous fluid	Dindo-Clavien Classification*
Brandstrup[9]	2003	elective colorectal resection	surgery	RCT	141	postsurgical	69	restricted	72	standard
Holte[10]	2007	elective surgery	surgery	RCT	32	elective colorectal surgery	16	restricted	16	liberal
Holte[11]	2007	post-surgery	surgery	RCT	48	knee arthroplasty	24	restricted	24	liberal
Gonsalez-Fajardo[12]	2009	post-surgery	surgery	RCT	40	vascular surgery transperitoneal aorto-iliac	20	restricted	20	standard
Yates[13]	2013	elective surgery	surgery	RCT	206	elective colorectal surgery	104	starch	98	crystalloid
Ghodraty[14]	2017	post-surgery	surgery	RCT	91	abdominal surgery	46	HES	45	ringers lactate
Gómez-Izquierdo[15]	2017	post-surgery	surgery	RCT	128	colorectal surgery	4	GDFT	64	control

Abbreviations: HES- hydroxyethyl starch; GDFT-goal-directed fluid therapy; RCT-randomised controlled trial. * Appendix C

Table 3

Study Characteristics of 'poor ' quality studies on the impact of intravenous fluid therapy on gut dysfunction.

Author	Year	Study Population	Study Setting	Study type	Study patients	Admission diagnosis	Experimental	Intravenous fluid	Control	Intravenous fluid	Dindo-Clavien Classification*
Prein[27]	1990	post-surgery	surgery	RCT	18	modified Whipple's	6- ringers' lactate; 6– starch; 6-albumin				I	II	III	IV
Salim[28]	1991	elective surgery	surgery	RCT	130	Hartmann's procedure +/- cholecystectomy	71	early oral	59	conventional intravenous
Yogendran[29]	1995	elective surgery	surgical	RCT	200	surgical	100	Low-infusion	100	High infusion
Wilkes[30]	2001	elective, open surgical	surgical	RCT	47	surgical	23	Balanced	24	Saline
Lobo [31]	2002	post-surgery	surgery	RCT	20	colorectal surgery	10	restricted	10	liberal
Conway[32]	2002	major bowel surgery	surgical	RCT	57	surgical	28	GDFT	39	Standard
Venn[33]	2002	hip fracture surgery	surgical	RCT	90	surgical	29CON- VF ; CVP guided FT- 3 1 ; Doppler-guided FT- 30
SAFE[34]	2004	ICU	ICU	RCT	6997	ICU	3497	Colloid	3500	Crystalloid
Parker[35]	2004	hip fracture surgery	surgical	RCT	396	surgical	198	Colloid	198	Crystalloid
Noblett[36]	2005	elective colorectal resection	surgical	RCT	108	surgical	54	GDFT	54	Standard
Wakeling[37]	2005	large bowel surgery	surgical	RCT	128	surgical	64	GDFT	64	Standard
Mackay[38]	2006	elective colorectal surgery	surgical	RCT	80	surgical	41	Restricted	39	Standard
En-quiang[38]	2009	critically ill	S-ICU	RCT	76	severe acute pancreatitis	30	controlled fluid expansion	30	rapid fluid expansion
Senagore[40]	2009	laparoscopic colectomy	surgical	RCT	64	surgical	21 GDFT/LR; 21 GDFT/HS;2 22 standard
Futier[41]	2010	major abdominal surgery	surgery	RCT	70	postsurgical	36	Restricted-GDFT	34	Conservative GDFT
Benes[42]	2010	elective intraabdominal surgery	surgery	RCT	120	ICU surgical	60	GDFT	60	Standard
Pillai[43]	2011	post-surgery	surgery	RCT	66	radical cystectomy	34	intervention	32	control
Du[44]	2011	critically ill	ICU	RCT	41	severe acute pancreatitis	20	starch	21	ringers' lactate
James[45]	2011	Blunt and penetrating trauma	surgical	RCT	109	surgical	Penetrating trauma- HES 36 ; SAL 31 Blunt trauma- HES20 ; SAL 22
Challand[46]	2012	major elective colorectal surgery	surgical	RCT	179	surgical	90	GDFT	89	Standard
Myberg[47]	2012	ICU	ICU	RCT	7000	ICU	3500	Colloid	3500	Crystalloid
Srinivasa[48]	2012	elective colectomy	surgical	RCT	85	surgical	37	GDFT Restricted	37	Restricted
Zheng[49]	2013	post-surgery	surgery	RCT	60	gastrointestinal surgery	30	GDFT	30	control
Scheeren[50]	2013	High-risk surgery	ICU	RCT	52	ICU	26	GDFT	26	Control
Pestana[51]	2014	post-surgery	S-ICU	RCT	142	abdominal surgery	70	GDFT	72	control
Pearse[52]	2014	Major Gastrointestinal Surgery	surgery	RCT	734	surgical	368	GDFT	366	Standard
Peng[53]	2014	elective surgery	surgery	RCT	80	orthopaedic surgery	40	GDFT	40	standard
Reisinger[54]	2017	elective colorectal resection for malignancy	surgery	RCT	58	postsurgical	27	GDFT	31	Standard

Abbreviations: : HES- hydroxyethyl starch ; HS- hetastach; SAL- saline; LR- lactate ringers; GDFT-goal-directed fluid therapy; ICU - intensive care unit ; S-ICU -surgical ICU; CON-IVF- conventional intravenous fluid therapy; CVP- central venous pressure; FT-fluid therapy RCT-randomised controlled trial; * Appendix C

In studies on enteral feeding [55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111], 57 randomised controlled trials’, included nearly 50% of the cohort as critically ill patients while the remaining were admitted as acute or elective surgical patients with variable admission diagnoses. The experimental group included enteral feeding delivered based on the route of feeding (enteral vs parenteral; nasogastric vs nasojejunal or jejunostomy), the timing of feeding (early vs delayed), the pattern of feeding (bolus vs continuous), or enteral feeding vs nil-by-mouth (NBM) with/without intravenous fluid. Twenty-four studies included critically ill, multiple trauma or sepsis patients with a Clavien-Dindo grading of IV. The remaining studies included post-operative and acutely ill patients with Clavien-Dindo grading of II and III (Tables 4-6).

Table 4

Study Characteristics of studies on the impact of enteral feeding on gut dysfunction included in the systematic review

Author	Year	Study Population	Study Setting	Study type	Study patients	Admission diagnosis	Experimental	Control	Dindo-Clavien Classification#				Quality Grading*
									I	II	III	IV
Hoover[55]	1980	surgical	surgical	RCT	48	surgical	26 EF	22IVF					Poor
Adams[56]	1986	ICU surgical	ICU	RCT	46	multiple trauma	23 (EN)	23(PN)					Poor
Moore[57]	1986	major abdo trauma	surgical	RCT	59	surgical	29 (EN)	30 (PN)					Poor
Bower[58]	1986	surgical	surgery	RCT	20	GI/pancreato-biliary surgery	10 (EN-JeJ)	10 (PN)					Poor
Hamoui[59]	1989	surgical	surgical	RCT	19	major GI surgery	11EN	8PN					Poor
Von Meyenfeldt[60]	1992	surgical	surgery	RCT	101	GI/colon cancer	50 (EN)	51(PN)					Poor
Montecalvo[61]	1992	surgical	surgical	RCT	38	surgical	19 NG	19 NJ					Poor
Dunham[62]	1994	critically ill	ICU	RCT	37	trauma	12 (EN)+ 15 (PN) + 10(EN+PN)						Poor
Borzotta[63]	1994	trauma	surgical trauma	RCT	48	trauma	27 (EN)	21(PN)					Poor
Daly[64]	1995	surgical	surgical	RCT	60	surgical	18ENSD; 12SD-IP; 19 ENSD-IP-OP; 11 EN-IP						Poor
Carr[65]	1996	post-surgical	surgery	RCT	28	intestinal resection	14(EEN)	14(CEN)					Poor
Beier-Holgersen[66]	1996	post-surgical	surgery	RCT	60	major abdominal surgery	30(EEN)	30 (placebo)					Poor
Baigrie[67]	1996	post-surgical	surgery	RCT	97	oesophagectomy/gastrectomy	50 (EN)	47(PN)					Poor
VanBerge[68]	1997	post-surgical	surgery	RCT	57	pancreatoduodenectomy	30 (CON)	27(CYC)					Poor
Kalfarentzos[69]	1997	critically ill	ICU	RCT	38	Severe acute pancreatitis	18(EN)	20 (PN)
Heslin[70]	1997	surgical	surgery	RCT	195	upper GI malignancy	97 (EN)	98(IVF)					Poor
Reynolds[71]	1997	major upper GI surgery	surgical	RCT	67	surgical	33 (EN)	34(PN)					Poor
Stewart[72]	1998	elective surgical	surgery	RCT	80	colorectal resections	40 (EOF)	40 (COF)					Poor
Windsor[73]	1998	surgical	surgical	RCT	34	acute pancreatitis	16 EN	18PN					Poor
Singh[74]	1998	surgical	surgical	RCT	43	surgical	22JEJ	21IVF					Poor
Braga[75]	1998	surgical	surgical	RCT	166	surgical	55 STD-EN; 55 -STD-EN enriched; 56 TPN						Poor
Taylor[76]	1999	critically ill	ICU	RCT	82	head injury	41TRO	41 EN					Fair
Pupelis[77]	2000	critically ill	S-ICU	RCT	60	severe pancreatitis/peritonitis	30 (JEN)	30 (Control)					Poor
Minard[78]	2000	critically ill	ICU	RCT	27	head injury/trauma	12(EEN)	15(DEN)					Poor
Powell[79]	2000	critically ill	ICU	RCT	27	severe acute pancreatitis	13 (EN)	14(NBM)					Poor
Kearns[80]	2000	critically ill	ICU	RCT	44	critically ill	23 G	21 SI					Poor
Bozzetti[81]	2001	elective surgery	surgery	RCT	317	GIcancer	159(EN)	158(PN)					Poor
Braga[82]	2001	surgical	surgery	RCT	257	GIcancer	126(EEN)	131(PN)					Poor
Montejo[83]	2002	critically ill	ICU	RCT	101	critically ill	50 (JEN)	51(GEN)					Poor
Davies[84]	2002	critically ill	ICU	RCT	73	critically ill	34 (NJ)	39 (NG)					Poor
Bertolini[85]	2003	critically ill	ICU	RCT	39	Sepsis	18(EN)	17 (PN)					Poor
Kompan[86]	2004	critically ill	ICU	RCT	52	multiple trauma	27(EEN)	21(DEN)					Poor
Malhotra[87]	2004	post-surgical	surgery	RCT	164	perforated gut and peritonitis	83 (EN)	81(NBM)					Poor
Kumar[88]	2006	Surgical	surgical	RCT	31	surgical	15 NG	16 NJ					Poor
Nguyen[89]	2007	critically ill	ICU	RCT	31	critically ill	23 (NJ)	28 (NJ)					Poor
Han-Guerts[90]	2007	post-surgical	surgery	RCT	150	oesophagectomy	71 (ND)	79 (JEJ)					Poor
Descahy[91]	2008	critically ill	ICU	RCT	100	ICU	50EEN	50CEN					Poor
Tien[92]	2009	critically ill	ICU	RCT	200	ICU	98TRO	102 EN					Poor
Barlow[93]	2011	Surgical	surgery	RCT	121	upper GI malignancy	64 (EN)	57(NBM+IVF)					Poor
Altintas[94]	2011	critically ill	ICU	RCT	71	ICU	30 (EN)	41 (PN)					Poor
Rice[95]	2011	Surgical	surgical	RCT	247	surgical	EN 123	124 IVF					Poor
Davies[96]	2013	critically ill	ICU	RCT	181	ICU	91 NJ	89 NG					Poor
Zhu[97]	2013	post-surgical	surgery	RCT	68	pancreaticoduodenectomy	34(JT)	34(NJT)					Poor
Sun[98]	2013	critically ill	S-ICU	RCT	60	severe acute pancreatitis	30(EEN)	30(DEN)					Poor
Kadamani[9]9	2014	critically ill	ICU	RCT#	15	critically ill	15 (CON)	15 (BOL)					Poor
Boelens[100]	2014	elective surgical	surgery	RCT	123	rectal surgery	61(EEN)	62(EPN)					Poor
Harvey[101]	2014	critically ill	ICU	RCT	2388	critically ill	1197(EN)	1191(PN)					Poor
Ma[102]	2015	acute surgical	surgery	RCT	35	acute pancreatitis	17 (NTF)	18(NPO)					Poor
Bing Li[103]	2015	post-surgical	surgery	RCT	400	gastrectomy	200(EEN)	200 (PN)					Poor
Taylor[104]	2016	critically ill	ICU	RCT	50	critically ill	25 (NJ)	25(NG + ProK)					Poor
Ozen[105]	2016	critically ill	ICU	RCT	51	critically ill	26(no-GRV's)	25(GRV's)					Poor
Van Barneveld[106]	2016	elective surgical	surgery	RCT	123	rectal ca malignancy	61 (EEN)	62(EPN)					Good
Malik[107]	2016	critically ill	ICU	RCT	60	critically ill	30 (EF)	30 (placebo)					Poor
Fan[108]	2016	critically ill	ICU	RCT	80	Severe TBI	40 (EN)	40 (PN)					Poor
Stimac[109]	2016	acute pancreatitis	pancreatitis	RCT	214	acute pancreatitis	107 EN	107 IVF					Poor
Hongyin[110]	2017	acute surgical	surgery	RCT	161	acute pancreatitis	83 (APD)/61 EN)	78(non-APD)/68(EN)					Poor
Reigner[111]	2018	critically ill	ICU	RCT	2410	shock	1202(EN)	1208(PN)					Fair

Abbreviations : EEN— early enteral feeding; CEN— conventional enteral feeding; EN— enteral nutrition; PN parenteral nutrition; CON— continuous enteral feeding; CYC — cyclic enteral feeding; EOF — early oral feeding ; COF — conventional oral feeding ;JEN—jejunal enteral nutrition; DEN— delayed enteral nutrition ; NBM — nil by mouth; GEN — gastric enteral nutrition ; NJ — nasojejunal; NG nasogastric; ND — nasoduodenal; JEJ- jejunostomy ; JT —jejeunostomy tube ; NJT — nasojejunal tube; BOL —bolus ; EPN — early parenteral nutrition; NTF — nasogastric tube feeding; NPO —nil per oral; ProK—prokinetics; GRV— gastric residual volumes; APD — abdominalparacentesis drainage ;ICU — intensive care unit ; S-ICU — surgical ICU; RCT — randomised controlled trial, # - pseudo-RCT; GI— gastrointestinal; TBI — traumatic brain injury; IVF — intravenous fluids; TRO- trophic feeding; ENSD — enteral nutrition with supplemented diet; IP inpatient; OP — outpatient; STD — standard; # D-C classification Appendix C; * Thresholds for Converting the Cochrane Risk of Bias Tool.

Table 6

Impact of enteral feeding on variables of gut dysfunction as classified by feeding categories

Symptoms of GDF§	Intervention Enteral	Control Parenteral	Odds [95% Ratio CI]*	P Trend	I2 [%]#
A. Route of feeding
Vomiting	605/2388	350/2598	2.02 (1.74, 2.35)	<0.01	0
Diarrhoea	190/1508	421/1515	1.75 (0.39, 7.86)	0.46	92
Abdominal distension	123/1386	90/1390	1.51 (0.93, 2.45)	0.10	28
Ileus	52/347	65/347	0.97 (0.34, 2.76)	0.96	58
Anastomotic leak	28/540	54/545	0.54 (0.31, 0.95)	0.03	14
Intestinal ischaemia	33/2493	16/2495	1.87 (0.72, 4.87)	0.20	42
Peritonitis	5/265	18/268	0.31 (0.11, 0.87)	0.03	0
B. Timing of feeding	Early	Delayed
Vomiting	3/56	19/54	0.11 (0.03, 0.41)	<0.01	0
Diarrhoea	27/39	23/40	2.45 (0.26, 22.75)	0.43	69
Abdominal Distension	12/66	21/69	0.51 (0.22, 1.91)	0.12	0
C. Enteral feeding vs Nil-by-mouth (NBM)	Enteral	NBM
Vomiting	21/220	22/219	0.72 (0.18, 2.90)	0.65	0
Abdominal Distension	66/242	48/240	1.40 (0.75, 2.64)	0.29	33
GI bleed	2/133	2/133	0.99 (0.17, 5.86)	0.99	0
Anastomotic leak	12/244	24/236	0.46 (0.22, 0.95)	0.04	0

*CI - Confidence interval used; Significant P values (<0.05) are shown in bold; #I2 - heterogeneity between studies expressed as percentages; § GDF - gut dysfunction

Quality assessment

The quality of studies was graded based on the Cochrane Quality assessment tool for randomised controlled trials for intravenous fluid (Tables 1-3) and enteral feeding (Table 4) studies (Appendix C and D). All studies met the criteria for randomisation and allocation concealment, but a wide variability existed between studies for other domains (blinding of participants and personnel, blinding of outcome assessment and assessor, incomplete outcome data and selective reporting). In the intravenous fluid group, quality assessment for 7 studies [9, 10, 11, 12, 13, 14, 15] (15%) scored ‘good’ (Table 1), 11 studies [16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26] (22%) scored ‘fair’ (Table 2), and more than half (63%) of the studies [27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54] were ‘poor’ (Table 3). In the enteral feeding group, the majority (95%) of the studies [55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75,77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105,107, 108, 109, 110] scored ‘poor’; two studies scored ‘fair’ [76, 111] and 1 study [106] was of ‘good’ quality (Table 4).

Table 2

Study Characteristics of 'fair ' quality studies on the impact of intravenous fluid therapy on gut dysfunction.

Author	Year	Study Population	Study Setting	Study type	Study patients	Admission diagnosis	Experimental	Intravenous fluid	Control	Intravenous fluid	Dindo-Clavien Classification*
Gan[16]	2002	major elective general, urologic, or gynaecologic surgery	surgery	RCT	100	postsurgical	50	GDFT	50	Standard
Moretti1 7	2003	Major elective cardiac surgery	surgery	RCT	90	postsurgical	30 - HetaStarch normal saline; 30 Heta Starch Balanced salt; 30 Lactated Ringers
Nisanevich[18]	2005	elective intraabdominal surgery	surgery	RCT	157	postsurgical	77	Restrictive	75	Liberal
Kabon[19]	2005	open colonic resection	surgery	RCT	253	ICU surgical	124	Small volume	129	Large Volume
Lopes[20]	2007	High-risk surgery	surgery	RCT	33	ICU surgical	17	GDFT	16	Control
Vermuelen[21]	2009	elective major abdominal surgical procedures	surgical	RCT	62	surgical	30	Restricted	32	Standard
Mayer[22]	2010	major abdominal surgery	surgery	RCT	60	ICU surgical	30	GDFT	30	Standard
SAFE [23]	2011	ICU	ICU	RCT	1218	ICU	603	Colloid	615	Crystalloid
Guidet[24]	2012	severe sepsis	ICU	RCT	196	ICU	100	Colloid	96	Crystalloid
Perner[25]	2012	severe sepsis	ICU	RCT	798	ICU	398	Colloid	400	Crystalloid
Reddy[26]	2016	critically ill	ICU	RCT	69	critically ill	35	plasmalyte	34	saline

Abbreviations: GDFT-goal-directed fluid therapy; ICU -intensive care unit; S-ICU -surgical ICU; RCT-randomised controlled trial; * Appendix C

Quantitative data analysis

Impact of intravenous fluid therapy on GDF

Twenty-two randomised controlled trials [9,10,13, 14, 15, 16,18, 19, 20, 21, 22,26,28,31,41, 42, 43,49,51, 52, 53, 54] evaluated mild to moderate (nausea, vomiting and ileus) and moderate to severe (GI bleed, anastomotic leak, perforation and intestinal obstruction) GDF in 7368 patients, of which, 3682 (50%) were randomised to the intervention group (goal-directed/ restricted/ balanced intravenous fluids) and the remaining to the control group (liberal/ standard intravenous fluid). In the intervention group, no significant difference was observed for nausea, ileus, GI bleed, anastomotic leak, perforation or intestinal obstruction, in the intervention group in comparison to the control group. However, restricted/goal-directed fluid therapy in the form of colloids (starch/albumin) or a balanced fluid solution (plasmalyte /ringers lactate) was likely to reduce ‘mild’ GDF such as vomiting (p = 0.03) in critically ill and major surgical patients compared to a standard/liberal intravenous fluid regime (Table 5). Heterogeneity between studies ranged from 0 - 45 %.

Table 5

Impact of intravenous fluid therapy on variables of gut dysfunction

Symptoms of GDF §	Interventional	Control	Odds Ratio [95% CI]*	P Trend	I2 (%)#
Nausea	88/ 274	90/278	0.98 (0.67, 1.44)	0.92	0
Vomiting	62/462	94/447	0.51 (0.28, 0.94)	0.03	45
Ileus	66/832	80/828	0.83 (0.52, 1.32)	0.42	23
GI bleed	15/592	10/587	1.48 (0.66, 3.35)	0.34	0
Anastomotic leak	44/833	43/867	1.03 (0.54, 1.96)	0.93	31
Perforation	7/238	6/234	1.05 (0.36, 3.09)	0.92	0
Intestinal obstruction	5/451	11/445	0.53 (0.20, 1.45)	0.22	0

a: restricted, goal-directed, low-infusions or a controlled-expansion fluid therapy given as crystalloid fluid (normal saline or plasmalyte) or colloid fluid (hydroxyethyl starch)

b: standard, liberal, conventional, high-infusions or rapid-expansion fluid regimes given as crystalloid fluids (ringers lactate, plasmalyte and saline).

*CI - Confidence interval used; Significant P values (<0.05) are shown in bold; #I2 - heterogeneity between studies expressed as percentages; § GDF - gut dysfunction

Impact of enteral feeding on GDF

Twenty-one randomised controlled trials’ [60,63,67, 70,71, 72, 73,75,78,81,85, 86, 87,91,93,94,100,101,106, 108,111] enrolled 18,543 patients of which, 50% (n = 9260) patients were randomised to the enteral nutrition groups. The remaining half (n = 9283) were randomised to the parenteral nutrition group, delayed enteral feeding or nil-by-mouth group. Mild to moderate GDF (vomiting, diarrhoea, abdominal distention and ileus) and moderate to severe (GI bleed, anastomotic leak, intestinal ischaemia, peritonitis) are presented in sub-groups (route of feeding – enteral vs. parenteral; the timing of feeding – early vs. delayed and feeding vs no feeding – enteral vs. nil-by-mouth) demonstrated in Table 6.

Route of feeding (enteral vs parenteral)

In the EN group, a significant increase in vomiting episodes was observed compared to in parenteral nutrition group (p < 0.01). The EN group showed a trend in fewer events for anastomotic leaks (p = 0.03) and peritonitis (p = 0.03) compared to the parenteral nutrition group. Other variables of GDF, including diarrhoea, abdominal distension and intestinal ischemia, presented with no significant differences between the two groups (Table 6). Heterogeneity between studies ranged from 0 – 92 %.

Timing of feeding (early vs delayed)

Four randomised controlled trials’ enrolled 324 patients, of which 50% of patients were allocated to the early enteral nutrition group and the other half to the delayed/conventional enteral nutrition group. A significant decrease in the vomiting episodes was observed in the early enteral nutrition group compared to delayed/conventional enteral nutrition group (p < 0.01). No differences were observed between groups for diarrhoea and abdominal distension. Heterogeneity between studies ranged from 0 – 69 %.

Enteral feeding vs nil-by-mouth (NBM)

Six randomised controlled trials’ enrolled 1667 patients, of which 50% was randomised to the intervention group. There was a tendency of reduced anastomotic leaks in patients receiving enteral feeding (p = 0.04) compared to patients on a nil-by-mouth regimen. However, no differences were observed for events on vomiting, abdominal distension and GI bleed. (Table 6). Heterogeneity between studies ranged from 0 – 33 %.

Discussion

The results of the meta-analysis demonstrate that restricted/goal-directed fluid therapy regardless of the type of fluid reduces mild GDF (vomiting) but not other complications associated with GDF. Enteral feeding, on the other hand, significantly increased vomiting episodes compared to parenteral nutrition but ‘early’ enteral nutrition significantly reduced the incidence of vomiting compared to delayed feeding. Enteral feeding was likely to reduce severe gut complications such as anastomotic leak and peritonitis compared with parenteral nutrition or an NBM status. Other mild to moderate variables of GDF (i.e. nausea, abdominal distension, ileus or diarrhoea) and moderate to severe complications (i.e. GI bleed, perforation, intestinal obstruction or intestinal ischaemia) were not associated with significant changes in outcomes. The results suggest that although the beneficial effects of restricted/goal-directed intravenous fluids and enteral feeding are essential to reduce some form of GDF, the impact is not prevalent for other variables of GDF (e.g. ileus and intestinal ischemia) associated with poor clinical outcomes. This may reflect the paucity of high-quality literature on the interaction between intravenous fluid (resuscitation) and enteral feeding as a combined therapy on the impact of GDF. The role of these two modalities in combination should be regarded as an important aspect in identifying the impact on the severity of GDF in acute surgical and critically ill patients.

Intravenous fluid therapy is frequently the first line of treatment in acute surgical and critically ill patients but hypervolemia and hypovolemia, both, are deemed detrimental. A revival of interest emerged almost two decades ago when hypovolemia in the form of restrictive fluid therapy was associated with improved post-operative clinical outcomes [9,16,18,31]. These studies suggested that a preferred approach of ‘zero’ or ‘neutral’ fluid balance not only improves outcomes related to gut motility but also may prevent adverse long-term outcomes. The current study demonstrated that mild GDF, i.e. vomiting, was significantly lower in patients on a restrictive/ targeted intravenous fluid regime. Studies have also reported similar results when colloids have been administered postoperatively [14,17]. The benefit of this outcome may be explained by cumulative administration of smaller volumes (of colloids) compared to crystalloids. Hypervolemia from excessive or liberal fluid administration, particularly crystalloids is associated with poor outcomes in postoperative [9,18] and in critically ill patients [25,47]. It can precipitate intestinal oedema leading to an ileus, delayed gastric emptying, feeding intolerance and hence sub-optimal nutrition delivery. Another school of thought indicates that complex surgical patients with high-risk surgeries possibly require judicious amounts of fluids to avoid complications associated with circulatory failure and gut mucosal ischemia [41,54,116,119]. This may be particularly relevant when liberal intravenous fluids are necessary to resuscitate patients after massive haemorrhagic losses for haemodynamic stability. In recent decades, goal-directed fluids have been advocated to prevent tissue hypovolemia [20] but maintain euvolemia by using targeted fluid approach raising the possibility of improved clinical outcomes in high-risk patients [22, 42, 84,121]. Hence, it is expected that a modest amount of fluids might be necessary to prevent anastomotic hypoperfusion, gut mucosal ischemia and reduce postoperative complications. Although the benefit of goal-directed fluid therapy is projected at improving organ perfusion without the onset of tissue oedema [54,117,122]; a paucity of studies exists warranting more research in this area [15, 41, 52, 118].

Enteral Nutrition forms an integral part of overall fluid administration in addition to intravenous fluids. Enteral nutrition and intravenous fluids combined play a crucial role in GDF outcomes, but due to a paucity of studies, this area has not received due attention. Enteral nutrition is invariably the first choice of nutrition compared to parenteral nutrition over decades [112]. The current study demonstrated that mild GDF, i.e. vomiting significantly increased in patients receiving enteral nutrition but reduced significantly when enteral feeding was commenced earlier. This is possible because ‘early’ enteral nutrition has multiple advantages over parenteral nutrition [75,82,114,121], and these benefits are evident in high-risk surgical and critically ill patients [70,124]. The initiation of enteral feeding is known to stimulate gut motility which reduces the incidence of GDF symptoms such as nausea and vomiting postoperatively. However, a significant difference for ileus between groups was not observed, although the number of events were lower in the enterally fed group. In cases of gut failure, when enteral feeding is contraindicated, parenteral nutrition becomes the sole choice of feeding and may be commenced within 24 hrs of ICU admission or post-surgery [120]. Administering parenteral nutrition appears to be a logical clinical decision, especially if enteral feeding raises the suspicion of non-occlusive mesenteric ischemia in the critically ill, with haemodynamic compromise. Our review showed no differences for intestinal 'ischaemia' between groups, although the events were half in the control group compared to the intervention (enteral nutrition) arm. Considering that the current review included a heterogeneous mix of patients, it is evident that in a sub-set of patients, i.e. post-cardiac surgery, severe acute pancreatitis or septic shock, administration of early enteral nutrition may potentially pose more risk than benefit by increasing the risk of bowel ischemia.

Nevertheless, the use of trophic enteral feeding has been suggested in haemodynamically unstable patients to maintain gut integrity [4]. Authors have argued that enteral nutrition comes with its risks such as aspiration, pneumonia, intestinal obstruction, necrosis and pneumonitis intestinalis. However, the present study demonstrated no such differences for any of these complications. For gastrointestinal complications, a significant reduction in anastomotic leaks in the enteral nutrition group suggesting its benefits irrespective of the feeding route was observed. It is common practice in some areas, particularly intensive care, to commence patients on parenteral nutrition with anastomotic leaks before a trial of enteral nutrition. However, it should be acknowledged that a correct assessment for an enteral nutrition challenge can be countered in patients on parenteral nutrition with significant complications (e.g. anastomotic leaks), hence lowering the threshold of initiating enteral nutrition. Barlow et al. [93] found a lower incidence (2 vs 7) of anastomotic leaks in the early enteral nutrition group. They attributed a three-day shorter length of stay and reduced postoperative complications from installing early enteral nutrition. A similar effect was confirmed by a Cochrane review [115] in which enteral nutrition reduced the risk of anastomotic leaks from 27% in the standard group to 13% in early enteral group. These results affirmed with the present findings. It is hypothesised that enteral nutrition may improve perfusion at the anastomosis site, which promotes mucosal wound healing and prevents further leaks.

In comparison, Lewis et al. (2009) did not support this finding and observed mortality of 50% in the intervention group (enteral group) with anastomotic leaks [114]. However, it is likely that a smaller sample size may result in a false positive rate for mortality, thus exaggerating the magnitude of the negative result. The benefit of enteral feeding in complications such as perforation and peritonitis has been confirmed by several reports, which resonated with our findings. Early enteral feeding seems to maintain gut integrity by improving mucosal circulation and oxygen delivery that may reduce the risk of peritonitis [74, 87,113].

The present study is not without limitations:

The severity score in majority of the studies including surgical patients was low (ranging between I to III) hence the overall effect may be confounded by the clinical severity of the cohort. The majority of studies were conducted in stable postoperative patients and results may not be generalisable to a high-risk group, e.g. septic shock.

Critically ill patients are a heterogeneous group, and the effect on gut function can differ with specific sub-population. Such high-risk heterogenous patients need to be assessed in robust, well-designed, and randomised controlled trials. A possible stumbling block may be the ethical dilemma of implementing clinical trials using regimented interventions in these patients is often challenging for institutions and ethics committees.

Individualised unit protocols were variable with prescription of fluid and enteral feeding regimes possibly confounding the overall impact on GDF outcomes.

Most studies included small numbers of patients and were single-centred studies.

Postoperative morbidity manifested as GDF may be associated with the type of surgical procedure or manipulation of the bowel during surgery which may be associated with inducing a surgical stress response. However, this is expected to be low in our study, considering that the majority of the cohort included stable postoperative patients.

The majority of our studies found no differences between long-term endpoints (mortality and length of stay) but the occurrence of GDF was excluded from primary endpoints.

Most importantly, it was difficult to define or classify gut dysfunction because, until now, there is no valid, objective or a reliable scoring system to assess gut function in intensive care patients [125]. This suggests the need to develop a novel scoring tool to address this concern in future trials. Due to fewer studies on the effect of intravenous fluids and enteral nutrition on GDF, our meta-analyses may have been underpowered to see significant outcomes on GDF. Overall, studies on intravenous fluid remain mostly inconclusive, and potentially the impact of intravenous fluids may project variable outcomes when applied to a homogenous cohort instead of heterogeneous patient groups.

Further, inconclusive results from large-scale fluid and enteral feeding trials raise the suspicion that GDF may be the missing link, which perhaps may be associated with long-term outcomes. This dimension is often ignored when evaluating endpoints. To observe a difference in the key outcome, we first need to understand the combined effects of intravenous fluids and enteral nutrition in influencing clinical outcomes, including GDF. It is expected that as a result of the potential interaction between these two modalities, patients receiving liberal fluid resuscitation and early aggressive feeding are more likely to be at risk of severe GDF. More work is required to understand the implications of intravenous fluids and enteral nutrition on GDF and how this may impact overall patient outcomes. Future studies should evaluate this potential interaction and assess the combined impact of these two modalities on GDF in surgical and critically ill patients.

Conclusion

A restricted/goal-directed fluid regime and early enteral feeding compared to parenteral or a nil-by-mouth regime may reduce the risk on mild GDF in some, but not all complications of severe GDF. Because of a preventive strategy, we need to first understand the interaction between both (intravenous fluids and enteral feeding) and their impact on the gut so its implications can be translated into clinical practice eventually. Hence, it can be hypothesised that conservative fluid resuscitation and aggressive enteral feeding may potentially be the fundamental cause of developing severe life-threatening GDF (i.e. intestinal ischemia) and complications that can delay recovery and affect clinical outcomes in acute surgical and critically ill patients. Future research should evaluate and focus on an extended conceptual framework on the cross-interaction of conservative and aggressive modes across these two interventions and its impact on various levels of severity of GDF.

Appendix A

Search Strategy

# Searches Results
1 Gastrointestinal Diseases/
2 ((gastrointestinal or intestin* or digestive) adj3 (dysfunction* or failure or disorder* or injur* or disease*)).mp.
3 ((abdominal or gut or bowel or intestin*) adj3 (perforat* or infarct* or obstruct* or failure or ischemi*)).mp.
4 gastroparesis.mp. or Gastroparesis/
5 gastrointestinal motilit*.mp. or exp Gastrointestinal Motility/
6 (dysmotilit* or intestinal motilit*).mp. 5645
7 Intra-Abdominal Hypertension/
8 (abdominal compartment syndrome* or intra abdominal hypertension or intraabdominal hypertension).mp.
9 feed* intolerance.mp.
10 ileus.mp. or Ileus/
11 Intestinal Obstruction/ or Intestinal Pseudo-Obstruction/ or pseudo obstruction.mp. or ogilvie’s syndrome.mp.
12 (mesenteric or peritonitis).mp. 91360
13 or/1-12 282880
14 enteral nutrition/ or parenteral nutrition/
15 Parenteral Nutrition, Total/
16 ((enteral or parenteral) adj3 (feed* or nutrition)).mp.
17 Fluid Therapy/ or intravenous fluid*.mp.
18 (fluid* adj3 therap*).mp.
19 (resuscitation adj3 fluid*).mp.
20 vasoactive.mp.
21 Vasoconstrictor Agents/ or vasoconstrictor*.mp. or vasopressor*.mp.
22 inotrope*.mp.
23 or/14-22
24 intensive care/ or critical illness/
25 Intensive Care Units/
26 General Surgery/
27 Postoperative Complications/ or Postoperative Care/
28 (intensive care or ICU or critical care or critical* ill*).mp.
29 (surgery or surgical or postoperative).mp.
30 or/24-29
31 randomized controlled trial.pt.
32 controlled clinical trial.pt.
33 randomized.ab.
34 placebo.ab. 35 drug therapy.fs.
36 randomly.ab.
37 trial.ab. 38 groups.ab.
39 or/31-38
40 adult/ or aged/ or “aged, 80 and over”/ or frail elderly/ or middle aged/ or (adult* or middle aged or older or old or aged or
elderly or geriatric* or frail).mp.
41 13 and 23 and 30 and 39 and 40
42 exp animals/ not humans.sh
43 41 not 42

Appendix B

Dindo-Clavien Classification

Grade	Grade Definition
Grade I	Any deviation from the normal postoperative course without the need for pharmacological treatment or surgical, endoscopic, and radiological interventions Allowed therapeutic regimens are: drugs as antiemetics, antipyretics, analgesics, diuretics, electrolytes, and physiotherapy. This grade also includes wound infections opened at the bedside
Grade II	Requiring pharmacological treatment with drugs other than such allowed for grade I complications, Blood transfusions and total parenteral nutrition are also included
Grade III	Requiring surgical, endoscopic or radiological intervention Grade IIIa Intervention not under general anaesthesia Grade IIIb Intervention under general anaesthesia
Grade IV	Life-threatening complication (including CNS complications)* requiring IC/ICU management Grade IVa Single organ dysfunction (including dialysis) Grade IVb Multiorgan dysfunction)

Dindo D, Demartines N, Clavien PA. Classification of Surgical Complications: A New Proposal with Evaluation in a Cohort of 6336 Patients and Results of a Survey. Annals of Surgery. 2004; 240 (2): 205-213.

Appendix C

Quality assessment for studies on the effect of intravenous fluid therapy on gut dysfunction (Cochrane quality grading for randomised controlled trials)7

*Thresholds for Converting the Cochrane Risk of Bias Tool: Good quality: All criteria met (i.e. low for each domain); Fair quality: One criterion not met (i.e. high risk of bias for one domain) or two criteria unclear, and the assessment that this was unlikely to have biased the outcome, and there is no known important limitation that could invalidate the results Poor quality: One criterion not met (i.e. high risk of bias for one domain) or two criteria unclear, and the assessment that this was likely to have biased the outcome, and there are significant limitations that could invalidate the results OR Two or more criteria listed as high or unclear risk of bias

Appendix D

Quality assessment for studies on the effect of enteral feeding on gut dysfunction (Cochrane quality grading for randomised controlled trials)7

*Thresholds for Converting the Cochrane Risk of Bias Tool: Good quality: All criteria met (i.e. low for each domain); Fair quality: One criterion not met (i.e. high risk of bias for one domain) or two criteria unclear, and the assessment that this was unlikely to have biased the outcome, and there is no known significant limitation that could invalidate the results; Poor quality: One criterion not met (i.e. high risk of bias; for one domain) or two criteria unclear, and the assessment that this was likely to have biased the outcome, and there are significant limitations that could invalidate the results OR Two or more criteria listed as high or unclear risk of bias