Strategies to Improve Inpatients' Quality of Bowel Preparation for Colonoscopy: A Systematic Review and Meta-Analysis

Background and Aims Inpatients' bowel preparation before colonoscopy is frequently inadequate, and various interventions have been investigated to improve it, so far. We aimed to evaluate the efficacy of various interventions to improve inpatients' colon preparation quality. Methods We systematically reviewed the literature for publications on interventions aiming to improve the quality of inpatients' colon preparation until June, 2018. Significant heterogeneity—measured with I 2—was detected at the level of P < 0.1. Adequacy rates were measured using inverse variance, and the size effect of different interventions was calculated using random effects model and expressed as odds ratio (OR). Results Seventeen studies enrolling 2733 inpatients were included. Overall, 67% (60-75%) of the participants achieved adequate colon cleansing (I 2 = 97%; P < 0.001). In six studies assessing the impact of educational interventions to patient/physician/nurse vs. no intervention, adequate bowel preparation was achieved in 77% (62-91%) vs. 50% (32-68%) of the patients (OR (95%CI) = 3.49 (1.67-7.28), P = 0.0009; I 2 = 74%; P = 0.002). Ten studies examined variations (qualitative and/or quantitative) in bowel preparation regimens with adequate preparation detected in 71% (60-81%) of the participants, and a single study examined the administration of preparation through an esophagogastroduodenoscope, resulting in adequate prep in 71% of the patients. Conclusions Despite several interventions, only two-thirds of inpatients achieve adequate colon preparation before colonoscopy. Educational interventions significantly improve inpatients' bowel preparation quality.


Introduction
Hospitalization compared to the ambulatory setting is associated to an almost twofold higher risk of failed bowel preparation before colonoscopy, while the rate of inpatients with adequately prepared colon does not exceed 50%, as they are usually of advanced age, debilitated, and suffering from comorbidities that either prevent successful ingestion of bowel prep or affect patients' comprehension and compliance with the regimen's instructions [1]. Suboptimal bowel preparation contributes not only to increased risk of missed pathology and patient inconvenience but also to a detrimental burden for healthcare systems, due to delayed or repeated procedures and prolonged hospital stay [2,3]. In order to overcome these hardships, several studies have evaluated the efficacy of various interventions, i.e., different purgatives, alterations in timing of preparation administration, introduction of educational programs for physicians, nurses and patients. In this context, we conducted a systematic review of the current literature, to provide insights into types of interventions used to substantially improve inpatient's bowel preparation. Systematic Reviews (PROSPERO) under the registration number CRD42017078647.

Eligibility
Criteria: Study Endpoint. Eligibility criteria were a priori delineated using the PICO statement as follows; P: inpatients undergoing colonoscopy for any indication; I: any type of intervention aiming to improve the quality of inpatient bowel preparation regardless of baseline disease or comorbidities; C: patients without intervention; and O: preparation's adequacy rate. Any type of trial published as full text in English language was included, while pediatric studies; meta-analyses or systematic reviews, editorials, case reports, narrative reviews, and conference abstracts; studies that did not detail patient information; and duplicate publications were excluded.

Information Sources and Search Strategy.
A systematic computer-aided literature search of MEDLINE, Cochrane Library (Cochrane Central Register of Controlled Trials), and Google Scholar databases was performed for consistent trials. The search was initially performed on the 22 nd of July 2017 and repeated on the 9 th of June 2018-the full electronic search strategy is available in Supplementary Material A. Search was conducted independently by two investigators (PG, GT). All titles and abstracts generated from the search were screened for inclusion; further selection was conducted by obtaining full texts of identified articles to determine whether they met inclusion criteria. To fulfill a recursive search, references of all studies and reviews acquired from the electronic search were manually searched for potentially eligible studies not captured initially. Systematic reviews and meta-analyses were consulted for additional information but excluded from analysis. This systematic review was conducted according to the preferred reporting items for systematic reviews and meta-analysis (PRISMA) [4] recommendations (Supplementary Material B).

Study Selection.
All articles retrieved from the search were screened independently by two reviewers (PG, GT). In case of uncertainty, disagreement was resolved by consensus. Titles and abstracts of all results were initially reviewed; thereafter, the full text of eligible studies was obtained and independently assessed for eligibility.

Data Collection
Process. Data were extracted from eligible peer-reviewed articles by two investigators (PG, GT), independently using standardized extraction forms. Discrepancies regarding data extraction were also resolved by consensus.
2.6. Data Items. The following data were extracted from included studies: country of study origin, number of patients enrolled and their mean age, study design and setting (year, location), and number of centers. The number of patients receiving or not intervention-defined as any measure aiming to improve bowel preparation quality, including verbal or written instructions to patients, enhanced educational measures to attending healthcare professionals or other ancillary medical providers, modifications in bowel preparation regimens (qualitative and/or quantitative), and other miscellaneous measures (not classified in previous categories) before colonoscopy-was extracted. Consequently, the number of patients with adequate bowel preparation that either received or not any intervention was extracted. For the purpose of our study, bowel preparation quality was dichotomized in two groups: adequate and inadequate. Bowel preparation was considered inadequate when it scored a total of the Boston Bowel Preparation Scale (BBPS) score < 6 with at least one segment score < 2 and a total of the Ottawa Bowel Preparation Scale (OBPS) score ≥ 6, described as "poor" in the Aronchick Scale [5][6][7]. For studies using different bowel preparation scoring scales, results were adjusted based on authors' definitions and presented accordingly (Supplementary Material C, Table 1). Moreover, the number of patients accepting the preparation strategy, willing to repeat the procedure, reporting any adverse event (definitions according to each study are available in Supplemental Material C, Table 2); the amount of preparation received, the number of repeated colonoscopies due to inadequate bowel cleanliness; and the total length of stay were also extracted. In case of missing data, the corresponding author was contacted by email, and if no response was received, the study was excluded from the analysis.

Outcome Measures.
Primary aim was to investigate interventions applied to inpatients undergoing colonoscopy aiming to improve colon preparation and determine their effect on the preparation's adequacy rate. Acceptance of preparation strategies, percentage of preparation received, willingness to repeat the examination, adverse events, repeat colon examinations, and duration of hospital stay comprised the secondary aims of this review.

Statistical
Analysis. Extracted data were analyzed using the statistical software Review Manager (RevMan 5.3.5, Copenhagen, Denmark, the Nordic Cochrane Centre, the Cochrane Collaboration, 2014). Overall preparation adequacy rate and all secondary endpoints were calculated using generic inverse variance analysis, and they are presented as percentage with respective 95% confidence intervals (CI). For bowel preparation adequacy comparisons, odds ratios (ORs) and their 95% CIs were calculated. All outcomes were further compared using the random effects model (DerSimonian and Laird method). Heterogeneity among studies was measured using the I 2 with lower values representing lower levels of heterogeneity. In case of significant heterogeneity (P < 0 1), predefined sensitivity analysis was performed by repeating the analysis excluding one study at a time to assess potential excessive influence of a study in heterogeneity's significance. For the primary endpoint, an additional predefined sensitivity analysis according to the study design, pooling separately prospective and observational studies, was performed. Forest plots were created for visual display of results. Potential publication bias of included studies was assessed by simple inspection for symmetry of funnel plots (if included studies were less than 10) constructed by plotting the log ORs vs. precision of individual studies per outcome or by Egger's test [8] evaluated using StatsDirect 3 (StatsDirect Ltd., Sale, Cheshire, England) software, if included studies were more than ten. Finally, we used both overlapping confidence interval inspection and the test for subgroup differences provided by the statistical software in order to perform a per intervention-used subgroup analysis for each outcome.
2.9. Risk of Bias in Individual Studies. To assess the quality and the risk of bias of the included randomized and nonrandomized studies, we used the Cochrane collaboration tool [9] and the Newcastle-Ottawa Scale [10], respectively.

Educational Interventions.
There was one RCT [15] and 5 nonrandomized prospective studies [11][12][13][14]16] evaluating the effect of educational interventions on bowel preparation. A detailed description about each study's educational intervention is presented in Supplemental Material C Table 5. In two studies, investigators assigned participants to the intervention arms to receive either an educational booklet on colonoscopy preparation [15] or an extra brief counselling and written instructions regarding the methods and rationale of bowel preparation [12], respectively. Three studies [11,13,14] educated the personnel involved in patients' preparation. Special leaflets, lectures, and/or presentations were used to educate nurses, who guided study participants before and during bowel preparation, and consequent comparison with either the preintervention period [11,14] or wards where noneducated nurses participated [13] was made. Finally, one study [16] is aimed at educating both personnel and patients.

Various Preparation
Regimens. Ten studies including 1802 individuals examined the impact of cathartics and alterations in timing of their administration on bowel cleansing [17][18][19][20][21][22][23][24][25][26]. Heterogeneity of regimens precluded meta-analysis; thus, they are presented in a qualitative-narrative manner organized in 3 subgroups: (a) various purgatives, (b) combinations of low-volume preparations with adjunctive agents, and (c) effect of timing of preparation administration (split dose vs. single dose). Reilly and Walker [18] deemed 6-L PEG (polyethylene glycol) with an additional second preparation (e.g., laxatives, tap water enemas, and Fleet enemas) as the optimal strategy while Seinelä et al. [17] demonstrated no significant benefit of sodium phosphate over PEG regarding adequacy of cleansing (81% vs. 77%, P = not available). Müller et al. [19] randomized subjects to receive mannitol or sodium picosulfate reporting equivalent results between groups. Moreover, in a large (n = 308) German multicenter noninferiority RCT, 2 L of PEG plus ascorbic acid achieved similar colon cleanliness  compared to 4 L PEG solution (88.9% vs. 94.8%, -5.9% with a lower limit of the 1-sided 97.5% confidence interval -12.0%, within the limits for noninferiority set before the study) [20]. Similarly, in a single-center randomized pilot study evaluating the impact of a same-day, 1-liter PEG on the diagnostic rating and tolerability, no difference compared to split-dose 4 L PEG was noted (63% vs. 56%, P = 0 64). Thus, sameday, 1 L-PEG bowel preparation could be introduced for selected inpatients [26]. In another randomized Korean study, the efficacy of low-volume (2-L) PEG with ascorbic acid was comparable to that of 2 L of PEG plus bisacodyl [23]. Regarding timing of bowel preparation, split-dose PEG preparation has been reported to be superior to sameday preparation [22]. On the contrary, a randomized, single-center study reported no difference between splitdose and morning-only PEG preparations (mean total Ottawa Scale score: 7 38 ± 3 65 vs. 7 15 ± 3 58, P = 0 75) [21]. Yadlapati et al. [25] reported a higher rate of adequate bowel preparation in patients receiving 4 L PEG as split dose compared to same-day regimen (85.7% vs. 42.5%, P < 0 01). Finally, implementation of a multiday bowel preparation regimen in 53 spinal cord injury patients led to adequate bowel cleanliness in 89% of the participants [24]. The cleansing rate was not affected by the study design (73% (63-83%) vs. 65% (49-81%) for RCTs and observational studies, respectively; test for subgroup differences (chi 2 = 0 63, df = 1, P = 0 43).
Three studies [11,15,25] presented results regarding the length of stay and repeated endoscopies ( Table 2). Ergen et al. [15] reported that mean total hospital stay was 6 days in patients receiving 4 L PEG split-dose plus education vs. 5 days in those receiving the same regimen without education; Yadlapati et al. [25] reported 8 days mean hospital stay after preparation commencement for patients receiving same-day 4 L PEG compared to 6.9 days for those getting 4 L split-dose PEG. Two studies [11,15] reported on the need to repeat colonoscopies because of inadequate bowel preparation. No difference was noted between patients receiving purgatives with personnel education and those only on purgatives (OR (95% CI): 0.93 (0.47-1.81); P = 0 82, I 2 = 0%, P = 0 47).

Discussion
A number of inpatient-related factors may contribute to inadequate bowel cleansing [2]. In the absence of guidelines  or recommendations, several methods have been implemented to improve bowel preparation scores [13,15,[19][20][21]26]. Despite their application, the overall adequacy rate of bowel preparation remains low. Our systematic review and meta-analysis confirm the low preparation adequacy rate (67%) among inpatients undergoing colonoscopy.
Our analysis showed that educating either the patients or the hospital personnel or both may pose certain effect on inpatients' bowel preparation quality. Educational interventions (paper-based interventions, videos, reeducation phone calls the day before colonoscopy, or in-person education by physicians) have been established from outpatients' studies as efficient methods to optimize colon preparation outcome [32]. However, the evidence to strongly support a similar conclusion regarding inpatients is quite low deriving only from 1 RCT and 5 nonrandomized studies. Specifically designed booklets [15] and written instructions [12] have been used to assist inpatients not only to discern the importance of adequate bowel preparation but also to increase their compliance by clarifying potential queries related to the procedure (adverse events, time points of regimen administration, etc.). Training healthcare professionals by using lectures and presentations [11,13,14] about the importance of adequate preparation and how to achieve it and recording adherence to the preparation plan through electronic documentation [11,14] might also enhance the effectiveness of provided instructions and decrease the rate of inadequate preparations. Still, all the abovementioned interventions are far from being perfect, as the overall colon cleansing adequacy rate remains suboptimal.
In addition, our analysis did not find solid evidence to support that specific types of cathartics or alterations in timing of their administration could result in better mucosa visualization. Although several approaches are available, the ideal bowel preparation regimen for inpatients remains to be determined, yet. Given the fact that several predictors of inadequate preparation are to be anticipated (e.g., advanced age, deteriorated health status, multiple medications, and comorbidities), this might be a particularly difficult task [2]. PEG-based regimens could be considered as the first step in any preparation strategy as they are more likely to achieve adequate bowel cleanliness retaining at the same time excellent patient safety profile. However, even they are not the optimal choice as their efficacy may be severely hampered by poor tolerability and compliance due to inability to drink 4-L PEG formulations, unpleasant taste, lack of comprehension, and complexity of the preparation instructions. Thus, "hybrid" bowel preparations, i.e., low-dose PEG with adjunctive agents like ascorbic acid that display equal effectiveness to the standard 4 L regimen could represent a useful alternative [20,33].
Since no single intervention has been shown to be efficacious in reaching the optimal level of bowel preparation in inpatients, one could speculate that multiple, combined strategies based on a case by case decision may have the potential to influence the final outcome. Indeed, this is the key message of a recent trial, where implementation of a standardized order set with split-dosing regimen, provision of written educational material to patient, and active nursing facilitation to the process overall resulted in significant positive improvements in the rate of acceptable inpatient bowel preparation [34].
Core strengths of the meta-analysis are the comprehensive and contemporaneous search strategy, including a recursive search of the literature of selected articles. To the best of our knowledge, this is the first study systematically addressing all available interventions to improve bowel preparation in inpatients.
We acknowledge a series of limitations in our study. The principal limitation lies in the heterogeneity encountered, calling for careful interpretation of our results. The latter mainly arises from the characteristics of the meta-analyzed evidence: retrospective, single-center setting, inadequate statistical power, small samples, and combination of randomized and observational studies, arbitrary classification of the reviewed interventions, and bowel preparation scales used.
In an effort to explore the evident heterogeneity, we performed predefined sensitivity analyses; nevertheless, ecological bias cannot be excluded. Even the existing evidence supporting that educational interventions reduce the rate of inadequate colon cleansing could be of higher quality. One could argue that the presence of significant heterogeneity and questionable-in some instances-study's quality included may challenge the validity of our results; however, our review enhances existing literature by specifically highlighting the potential role of educational interventions in inpatients bowel preparation adequacy and how current studies may still offer guidance in everyday clinical practice. Moreover, information regarding the exact stationary status of inpatients was absent, while concomitant medications were not systematically analyzed. Finally, local factors (e.g., staff availability) that might affect each intervention's efficacy remain underrated.
In conclusion, this study highlights the inadequate level of bowel preparation in inpatients undergoing colonoscopy, although several interventions have been implemented to increase it. However, educational interventions provided to patients and health care personnel reduce the rate of inadequate colon cleansing.

Disclosure
Part of this study has been presented as a poster during DDW 2018, Washington DC, USA, and as an Oral e-Poster during ESGE Days 2018, Budapest, Hungary.

Supplementary Materials
Supplemental Material A: search strategy. Supplemental Material B: preferred reporting items for systematic reviews and meta-analyses (PRISMA) 2009 checklist. Supplemental Material C Table 1: bowel preparation rating scales used in the included studies. Supplemental Material C Table 2: assessment of secondary endpoints. Supplemental Material C Table 3: risk of bias assessment of included randomized studies and authors' judgement. Supplemental Material C Table 4: risk of bias assessment of included observational studies using the Newcastle-Ottawa Scale. Supplemental Material C Table 5