A Systematic Review and Meta-Analysis of Double-Blind Placebo-Controlled Randomized Trials

Gastrointestinal (GI) dysmotility is a common complication in acute, critically ill, postoperative, and chronic patients that may lead to impaired nutrient delivery, poor clinical, and patient-reported outcomes. Several pharmacological and nonpharmacological interventions to treat GI dysmotility were investigated in dozens of clinical studies. However, they often yielded conflicting results, at least in part, because various (nonstandardized) definitions of GI dysmotility were used and methodological quality of studies was poor. While a universally accepted definition of GI dysmotility is yet to be developed, a systematic analysis of data derived from double-blind placebo-controlled randomized trials may provide robust data on absolute and relative effectiveness of various interventions as the study outcome (GI motility) was assessed in the least biased manner. To systematically review data from double-blind placebo-controlled randomized trials to determine and compare the effectiveness of interventions that affect GI motility. Three electronic databases (MEDLINE, SCOPUS, and EMBASE) were searched. A random effects model was used for meta-analysis. The summary estimates were reported as mean difference (MD) with the corresponding 95% confidence interval (CI). A total of 38 double-blind placebo-controlled randomized trials . Megill, BSc, Joh MBChB, MD, , MD, MPH, PhD dopamine receptor antagonists (MD, 8.99; 95% CI, 17.72 to 0.27; P1⁄4 0.04) and macrolides (MD, 26.04; 95% CI, 51.25 to 0.82; P1⁄4 0.04) significantly improved GI motility compared with the placebo group. The use of botulism toxin significantly impaired GI motility compared with the placebo group (MD, 5.31; 95% CI, 0.04 to 10.67; P1⁄4 0.05). Other interventions (dietary factors, probiotics, hormones) did not affect GI motility. Based on the best available data and taking into account the safety profile of each class of intervention, dopamine receptor antagonists and macrolides significantly improve GI motility and are medications of choice in treating GI dysmotility. (Medicine 95(5):e2463) Abbreviations: CI = confidence interval, GI = gastrointestinal, MD = mean difference, SD = standard deviation. INTRODUCTION G astrointestinal (GI) dysmotility is a common occurrence in acute, critically ill, and postoperative patients. It represents a significant barrier to the achievement of adequate nutritional intake. Extensive physiological evidence supports the use of early enteral nutrition, but acute and critically ill patients receive only up to half of their estimated calorie requirement, often due to feeding intolerance. Limited enteral nutrition delivery may lead to deterioration of the gut mucosa and gut wall integrity. The resulting decreased intestinal permeability, coupled with bacterial overgrowth, leads to leaky gut and elevated systemic proinflammatory mediators with an increased incidence of systemic inflammatory response, bacterial translocation, and multiple organ dysfunction. GI dysmotility may develop due to a variety of causes including systemic inflammation, postoperative state, electrolyte abnormalities, and numerous pharmacological interventions that impair motility. In clinical practice, pharmacological interventions are of particular importance in preventing GI dysmotility. Furthermore, several pharmacological and nonpharmacological interventions to treat GI dysmotility have been studied, including but not limited to D2, D3 antagonists, macrolides, m-opioid receptor antagonists, and probiotics. The outcomes of these studies are often conflicting and no consensus exists on which medications should be avoided to prevent GI dysmotility and which are the most effective to treat GI dysmotility. Part of the reason for this relates to the subjective and variable definitions for GI dysmotility. While an objective, reliable, and practical definition of GI motility is alidated, and ratified, it is possible to management of GI motility based on the -blind placebo-controlled randomized www.md-journal.com | 1 trials. This is because double blinding ensures that estimates of the treatment effects are robust, regardless of a definition of GI motility used. The other common barrier is that the conventional paradigm in clinical practice and research traditionally focuses on a particular nosology and it has largely overlooked the importance of the gut as an organ in its own right. The emerging evidence from the gut origin of sepsis hypothesis, enhanced recovery after surgery paradigm, and ‘‘gut rousing’’ concept suggests that the presence of GI dysfunction impairs clinical outcomes, regardless of nosology. Hence, the gut should be afforded the same considerations as other vital organs such as heart, lungs, and kidneys. In particular, given approaches to management of cardiovascular, respiratory, and renal dysfunctions are rather generic (inotropes, mechanical ventilation, dialysis), it is argued that a truly effective treatment of GI dysfunction would be beneficial regardless of nosology. However, to date, no comprehensive comparison of interventions that truly (as proven in double-blind placebo-controlled randomized trials) affect GI function has been published. The aim of this study was to systematically review data from double-blind placebo-controlled randomized trials to determine and compare the efficacy of various interventions that affect GI motility.


INTRODUCTION
G astrointestinal (GI) dysmotility is a common occurrence in acute, critically ill, and postoperative patients. It represents a significant barrier to the achievement of adequate nutritional intake. Extensive physiological evidence supports the use of early enteral nutrition, but acute and critically ill patients receive only up to half of their estimated calorie requirement, often due to feeding intolerance. [1][2][3][4][5][6][7][8] Limited enteral nutrition delivery may lead to deterioration of the gut mucosa and gut wall integrity. The resulting decreased intestinal permeability, coupled with bacterial overgrowth, leads to leaky gut and elevated systemic proinflammatory mediators with an increased incidence of systemic inflammatory response, bacterial translocation, and multiple organ dysfunction. [9][10][11][12] GI dysmotility may develop due to a variety of causes including systemic inflammation, postoperative state, electrolyte abnormalities, and numerous pharmacological interventions that impair motility. In clinical practice, pharmacological interventions are of particular importance in preventing GI dysmotility. Furthermore, several pharmacological and nonpharmacological interventions to treat GI dysmotility have been studied, including but not limited to D 2 , D 3 antagonists, macrolides, m-opioid receptor antagonists, and probiotics. The outcomes of these studies are often conflicting and no consensus exists on which medications should be avoided to prevent GI dysmotility and which are the most effective to treat GI dysmotility. [13][14][15][16] Part of the reason for this relates to the subjective and variable definitions for GI dysmotility. 17 While an objective, reliable, and practical definition of GI motility is yet to be developed, validated, and ratified, it is possible to make progress toward management of GI motility based on the evidence from double-blind placebo-controlled randomized trials. This is because double blinding ensures that estimates of the treatment effects are robust, regardless of a definition of GI motility used. The other common barrier is that the conventional paradigm in clinical practice and research traditionally focuses on a particular nosology and it has largely overlooked the importance of the gut as an organ in its own right. The emerging evidence from the gut origin of sepsis hypothesis, enhanced recovery after surgery paradigm, and ''gut rousing'' concept suggests that the presence of GI dysfunction impairs clinical outcomes, regardless of nosology. [18][19][20] Hence, the gut should be afforded the same considerations as other vital organs such as heart, lungs, and kidneys. In particular, given approaches to management of cardiovascular, respiratory, and renal dysfunctions are rather generic (inotropes, mechanical ventilation, dialysis), it is argued that a truly effective treatment of GI dysfunction would be beneficial regardless of nosology. However, to date, no comprehensive comparison of interventions that truly (as proven in double-blind placebo-controlled randomized trials) affect GI function has been published.
The aim of this study was to systematically review data from double-blind placebo-controlled randomized trials to determine and compare the efficacy of various interventions that affect GI motility.

Search Criteria and Study Identification
Electronic databases (MEDLINE, SCOPUS, and EMBASE) were searched for key words gastrointestinal (GI) motility, or gastric emptying, or gastrointestinal transit, or peristalsis, or ileus, or gastroparesis. The databases were screened for publications from the earliest available date until May 31, 2015. The study selection criteria were as follows: The inclusion criteria were:

Data Extraction
Data were extracted and tabulated by 3 authors (VMA, HDY, RDM) using predesigned data collection forms on Microsoft Excel. These included baseline and demographic data such as author, publication year, study setting (country), study population, total number of patients, sex, and age. As part of the data extraction process, the most significant dose was considered when several different doses of treatment were tested. Where different patient subgroups were tested (and the overall average value was not provided), the subgroup with the most significant difference was included. Any inconsistencies in data collection were discussed with the senior author (MSP).

Methodological Quality
Methodological quality of included randomized controlled trials was assessed according to the Cochrane recommendations (The Nordic Cochrane Centre, The Cochrane Collaboration, 2008). 21 These included systematic differences between groups (selection bias and performance bias), blinding of study participants and assessors, sequence allocation and concealment of allocated groups, 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. 21

Statistical Analysis
All data were presented as means AE standard deviation (SD). Data analysis and interpretation was done using Revman 5.3 (Revman, Version 5.3 for Windows; Copenhagen, Denmark; The Nordic Cochrane Centre, The Cochrane Collaboration, 2008). 21 When original data were presented as standard error (SE), they were converted to SD using the formula SD ¼ SE Â Hn (n ¼ the number of patients). Within each class of interventions, a meta-analysis was conducted, if required data from 2 or more studies had been reported. Furthermore, a sensitivity analysis constrained to a particular intervention was conducted, if appropriate. Heterogeneity was assessed using I 2 and X 2 tests, with a P < 0.05 considered to be significant for the latter. 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 to be statistically significant in all analyses.
Ethical approval was not necessary for a review of published trials.

Study Characteristics
A total of 4265 potentially relevant publications were screened, of which 39 studies 22-60 were included in the systematic review ( Figure 1). The baseline characteristics of these 39 studies are presented in Table 1. Interventions and GI motility endpoints used in these studies are presented in Table 2. The included studies investigated a total of 20 different interventions. The use of study interventions in 31 studies resulted in an improvement in GI motility while the use of study interventions in 8 studies resulted in an impaired GI motility ( Table 2). Of the 39 studies, 25 studies met the criteria for inclusion in meta-analysis. 22-26,28,29,31,33,37 -40,42,43,46-49,51 -55,59 These 25 studies recruited a total of 1339 patients which employed 6 interventions (D 2 , D 3 antagonists, macrolides, dietary factors, probiotics, hormones, and botulism toxin). Figure 2 presents the methodological quality of the 25 trials included in meta-analysis. Figures 3 and 4 present assessment of publication bias for D2, D3 antagonists and macrolides, respectively.

D 2 , D 3 Antagonists
A total of 5 studies including 198 patients employed a D 2 , D 3 antagonist as the study intervention. GI motility was significantly improved in the intervention group compared to the placebo group (MD, À9.09; 95% CI, À18.03 to À0.15; P ¼ 0.05) ( Figure 5). Three out of the 5 studies used Levosulpiride while the other 2 studies used Metoclopromide and Itopride. There was a high statistical heterogeneity between the included studies (I 2 ¼ 81%). A sensitivity analysis limited to Levosulpride showed no significant improvement with the use of this intervention (MD, À34.22; 95% CI, À76.14 to 7.70; P ¼ 0.11).

Macrolides and Its Derivatives
A total of 4 studies including 251 patients employed a macrolide or its derivative as the study intervention. GI motility was significantly improved in the intervention group compared with the placebo group (MD, À26.04; 95% CI, À51.25 to À0.82; P ¼ 0.04) ( Figure 6). Three out of the 4 studies used Erythromycin while 1 study used clarithromycin (6-O-methyl erythromycin). There was a high statistical heterogeneity between the included studies (I 2 ¼ 88%). A sensitivity analysis limited to erythromycin showed no significant improvement with the use of this intervention group (MD, À4.72; 95% CI, À20.25 to 10.81; P ¼ 0.55).

Other Interventions
A total of 7 studies including 450 patients employed a GI hormone (ghrelin, cholecystokinin, melatonin, and octreotide) as the study intervention. GI motility in the GI hormones group showed no significant improvement compared with the placebo group (MD, À7.22; 95% CI, À15.37 to 0.92; P ¼ 0.08). There was a high statistical heterogeneity between the included studies (I 2 ¼ 80%).
A total of 3 studies including 169 patients employed probiotics as the study intervention. GI motility did not show a significant improvement in the intervention group compared There was a high statistical heterogeneity between the included studies (I 2 ¼ 90%).
A total of 4 studies including 227 patients employed a dietary factor (wheat bran, soy germ, and Iberogast) as the study intervention group. GI motility showed no improvement in the intervention group compared with the placebo (MD, À15.05; 95% CI, À33.17 to 3.06; P ¼ 0.10). There was a low statistical heterogeneity between included studies (I 2 ¼ 29%).
Two studies including 44 patients employed botulism toxin as the study intervention. Gut motility was significantly impaired in the intervention group compared with the placebo group (MD, 5.31; 95% CI, À0.04 to 10.67; P ¼ 0.05). There was no statistical heterogeneity between the included studies (I 2 ¼ 0%).

DISCUSSION
This is the first systematic review of double-blind placebocontrolled randomized trials that evaluated the effect of pharmacological and nonpharmacological interventions on GI motility. Twenty interventions were included in the systematic review and 6 of them were meta-analyzed. The important finding of this study was that 2 classes of prokinetics (D2-D3 antagonists and macrolides) were effective in treatment of GI dysmotility, compared with the placebo group. Also, several  interventions impaired the GI motility, including the botulism toxin, an antibiotic (Cefazolin), a gastric secretion inhibitor (Nizatidine), an amylin analogue (Pramlintide), and a dipeptidyl peptidase-4 (DPP-4) inhibitor (Sitagliptin). These findings have important implications for routine clinical practice and future research on GI motility. Prokinetics was the most investigated class of interventions, 2 of which were suitable for meta-analysis. The first class of prokinetics found to be effective in our meta-analysis is D 2 , D 3 antagonists. Of the substantial amount of dopamine produced by the GI tract, spleen, and pancreas, nearly 46% of nonmetabolized dopamine is sourced by mucosa of the GI tract, which is partly represented by the non-neuronal cells of a dopaminergic paracrine system. 61 The blockade of dopaminergic inhibitory transmission in the gut is considered the main mechanism of its prokinetic effect. In our meta-analysis, 2 of the 4 studies included patients with diabetic gastroparesis, 47,49,54 showing an improvement in gastric motility with the administration of D 2 ,D 3 antagonists. In diabetic patients with gastroparesis, it is evident that the efficacy of treating gastric dysmotility is due to the selective antagonism of this prokinetic for dopamine antagonist receptors. 47 However, reduction in gastric emptying time may potentially improve glycemic control, though it should also be acknowledged that acute changes in glycemic control may have an irreversible effect on gastric emptying with the effect being more marked in the presence of euglycemia. Although D 2, D 3 antagonists appear to be a safe therapeutic option to improve GI motility in chronic diabetic gastroparetic patients, it will be worth investigating its effect on GI motility in diabetic patients with poor glycemic control. In the remaining 2 studies, focused on dyspeptic patients, both classes of D 2, D 3 antagonists showed improved effects on GI motility and dyspepsia or gastro-oesophageal reflux events. These studies demonstrated that most symptoms of dysmotility are manifested with dyspepsia or reflux events, and the use of D 2, D 3 antagonists proves effective in patients who may experience a combination of both. Similar to serotonin receptor agonists, D 2, D 3 antagonists have adverse effects, especially the commonly used prokinetic Metoclopramide can cause dystonic reactions with long-term use 13 and tachyphylaxis may occur after several days of administration. 62 Although the drug proved to be beneficial in a heterogeneous group such as critically ill patients, its dosing needs adjustment based on the clinical status of patients to minimize its side effects, for example, in patients with renal failure. 13 The other class of prokinetics found to be effective in this meta-analysis is macrolides. Three of the 4 studies administered erythromycin while 1 used Clarithromycin (a derivative of macrolides) as the study intervention. Two of the 4 included studies investigated Erythromycin in dyspeptic patients. These studies demonstrated that Erythromycin was effective in improving gastric emptying and interdigestive gastroduodenal motility. 24,29 However, in patients with postoperative ileus, the drug was less effective in relieving postoperative symptoms or preventing the occurrence of paralytic ileus. 28,53 Hence, it may be worth investigating the effect of various doses on GI motility   in surgical and nonsurgical patients separately. Although several studies have shown that Erythromycin improves GI motility and improves early nutritional intake in severely injured or critically-ill patients, 63-65 the duration of erythromycin use is limited by its antibacterial effect and desensitization to the therapeutic effects. 66 The present study has a number of limitations that need to be acknowledged. First, the included studies comprised several disease states and one might question the use of a meta-analysis approach. However, the main premise behind this study was that the presence of gut dysmotility worsens the outcomes of patients with all the diseases states included. [18][19][20] Therefore, timely administration of apposite gut-directed interventions proved to be effective in robust double-blind placebo-controlled trials can preserve normal gut function or curtail gut dysmotility. This is not dissimilar to cardiovascular failure, which can occur in patients with various diseases, but is virtually invariably treated with inotropes. 67,68 Second, the high statistical heterogeneity between the studies may be a possible limitation for the pooled effect. However, a random-effects model was used in all the analyses to obtain the most conservative estimate. Third, the meta-analysis did not take into account the dosage and route of administration of the studied drugs, which could have had an effect on GI motility, especially in chronic conditions such as diabetic gastroparesis. In addition, the form of administration (solution or tablet form) was not considered in this meta-analysis. It is possible that the effect of a prokinetic administered intravenously may differ from that in a tablet form. 69,70 Fourth, very few studies evaluated the effect of combined interventions, which may prove to be more beneficial in treating GI dysmotility than an individual intervention. Fifth, the sample size of some individual trials was rather small. But this systematic literature review is a necessary step toward definitive clinical studies as it provides data on which to power them. And it was encouraging that, despite the small sample size, some of the results were statistically significant, which suggests that the effect size is likely to be clinically meaningful. The high heterogeneity between studies suggests that it is challenging to obtain a homogenous population particularly in critically-ill, surgical, or acute patients who may routinely receive prokinetics to treat gut dysmotility as the first line of treatment. Last, m-opioid receptor antagonists have emerged as the new promising class of drugs that may improve GI motility, in particular opioid-induced bowel dysfunction, 71,72 but only one of the clinical studies published to date met the strict eligibility criteria for inclusion in the present review. 45 In conclusion, this is the first systematic review of best quality studies that investigated interventions affecting GI motility. Dopamine receptor antagonists and macrolides significantly improve GI motility and are safe to use in clinical practice. The dose, route, and combination therapy of these prokinetics will need to be investigated in future studies. Considering the high statistical heterogeneity, the precise effect of these interventions should be investigated in homogenous groups of patients in future studies. Interventions such as botulism toxin, gastric secretion inhibitors, cephalosporin antibiotics, amylin analogues, and DPP-4 inhibitor significantly impair GI motility and should be used with caution in high-risk patients with dysmotility.