Adverse Events of Video Capsule Endoscopy Over the Past Two Decades: A Systematic Review and Meta-Analysis

Background: A full spectrum of video capsule endoscopy (VCE) adverse events over the past two decades has not been evaluated. We aimed to determine pooled rates, predictors and temporal-trend of VCE adverse events over the past two decades. Methods: Systematic search of PubMed and EMBASE for English-language publications reporting VCE adverse events (January 1, 2000 to March 31, 2019). Data were extracted independently by two investigators. Pooled VCE adverse event rates were calculated using random-effects model. Predictors and temporal-trend of each adverse event were performed by meta-regression analyses. Results: In total, 402 studies were identied, including 108,079 VCE procedures. Rate of retention, swallow disorder, aspiration, technical failure, and procedural adverse events were 0.73% (95% condence interval [CI], 0.59%-0.89%), 0.75% (95% CI, 0.43%-1.13%), 0.00% (95% CI, 0.00%-0.00%), 0.94% (95% CI, 0.65%-1.28%), 0.67% (95% CI, 0.32%-1.10%), respectively; incomplete examination rate of esophagus, stomach, small bowel, and colon were 9.05%, 7.69%, 12.08%, 19.19%, respectively. Patency capsule reduced retention rate by 5.04%, whereas known inammatory bowel disease increased retention rate by 4.29%. Elder was the risk and protective factor for small bowel incomplete examination (0.30%) and swallow disorder (-0.72%), respectively. Rates of retention and small bowel incomplete examination signicantly declined over time (P = .0006 and P < .0001). Conclusions: VCE adverse event rates were generally low, and retention and small bowel incomplete examination rates declined over the past two decades. Patients with known inammatory bowel disease or elder should be alerted to high risk of retention or small bowel incomplete examination (PROSPERO: CRD42019139595). meta-analysis to report pooled rates of each VCE adverse event, assess the predictors and provide time-trend analysis over the past two decades. The results demonstrated that VCE is a safe procedure worldwide with low rate of adverse events. The risk for retention should be assumed in patients with known IBD, which is a contraindication unless intestinal patency is proven, best by the passage of an intact PCE. Elder age is the risk or protect factor for small bowel incomplete examination or swallow disorder. In addition, the rate of retention and small bowel incomplete examination were declined over the years.


Introduction
Since its introduction by Iddan et al [1] in 2000, video capsule endoscopy (VCE) has established itself as a noninvasive diagnostic tool for gastrointestinal diseases over the past two decades. It has become the rst-line investigation procedure in small bowel disorder evaluation [2]. Recently, the invention of esophagus capsule endoscopy (ECE) [3,4], magnetically controlled capsule endoscopy (MCE) [5][6][7], and colon capsule endoscopy (CCE) [8,9] widened the range of applications and made VCE available for entire gastrointestinal tract examination.
Although VCE has been widely used, potential VCE adverse events could happen and deserved consideration. Retention, the most noticed adverse event that may lead to acute small bowel obstruction and usually required surgical intervention. It has been reported of approximately 1.4% in most recent review [10] and varied from 0%-13% [11][12][13][14][15][16][17]. Related systematic reviews reported pooled retention rate of different indications (1.2-2.6% and 2.1-8.2%, respectively) [18,19]. However, no systematic review or meta-analysis estimated the rate of a full spectrum of VCE adverse events, and predictors of each adverse event have never been evaluated. Additionally, with the advance of technology, VCE adverse event rates tend to decline and an update is warranted. Moreover, previous SRMAs are limited to single VCE type, while several novel diagnostic VCEs have been invented and widely used in clinical practice, adverse event rates of other VCE types are needed.
Herein, we aimed to perform a comprehensive systematic review of the contemporary literature to quantify the rates of all VCE adverse events, assessing the potential predictors of each adverse event and demonstrating whether rates changed over the past 20 years.

Data Sources and Searches
This systematic review and meta-analysis was conducted and reported in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines (Supplementary Method 1) [20]. The identi er of systematic review registration was PROSPERO (CRD42019139595). We searched PubMed and EMBASE databases for English-language publications on VCE from January 1, 2000 through March 31, 2019 using the keywords related to "capsule endoscopy", which were based on Medical Subject Headings. Additional studies were identi ed by manually searching the reference lists of the included studies. Detailed search strategy is available in Supplementary Method 2.

Study Selection
Studies reporting adverse events of VCE were included. Exclusion criteria included (1) Case reports or studies with fewer than 50 patients; (2) letters, editorials, correspondences, perspectives, reviews, guidelines, conference abstracts or presentation without formal publication; (3) Animal and in vitro studies, trainee participation; (4) Duplicated publications from the same trial (only the most recent and most extensive data was included); (5) Studies focused on non-VCE (i.e. motility capsule endoscopy, patency capsule endoscopy [PCE] only, BRAVO pH capsule, tethered capsule endoscopy, balloon capsule endoscopy, et al).
Studies that performed an initial PCE before VCE to exclude potential small bowel obstruction were included. Three independent reviewers (Y.-C.W., J.P., and Y.-W.L.) selected the abstracts and determine their inclusion. Full texts of the potentially eligible studies were further evaluated whether it contained relevant information.

De nitions
We de ned VCE adverse events as retention, swallow disorder, aspiration, technique failure, procedural adverse events, and incomplete examination of con rmed with abdominal radiograph, or if a directed medical, endoscopic or surgical intervention has to be implemented to remove or add its passage [18,21]. Swallow disorder was de ned as patients unable to swallow the VCE, or require endoscopic delivery system assistance [22,23]. The de nition of aspiration was bronchial aspired VCE [24]. Technical failure was de ned as malfunction of the equipment, including gaps in recording, short duration of batteries of VCE or recorder, failure to activate VCE, failure to download or upload [25]. Procedural adverse events meant discomfort during VCE examination [26]. Esophagus incomplete examination was de ned as no image of Z line was obtained by VCE [27]. Stomach incomplete examination was de ned as incomplete visualization of all six landmarks (i.e. Cardia, fundus, body, angulus, antrum, and pylorus) [28]. Small bowel incomplete examination, meaning that VCE failure to reach the caecum during the recording time [18,29]. Colon incomplete examination was regarded as VCE was not excreted or did not reach the rectum during the recording time [30].

Data Extraction And Outcomes Assessment
Data were extracted independently by two investigators (Y.-C.W., F.-Y. S.). The characteristics of study (i.e. rst author, publication year, study period, study design, area), patient (i.e. Simple size, mean age, male percentage, indications, history), procedure (i.e. total number of VCE, VCE types), and each adverse event (i.e. Type, events number, reasons, and interventions) were independently collected. Patient groups were classi ed by indications according to the clinical practice guidelines [10], as for the case-controlled studies and randomized controlled trials (RCT), the data of each group was extracted separately.
Our primary outcome was to estimate pooled rate of each VCE adverse event. The secondary outcomes were factors associated with the rate of retention, small bowel incomplete examination, swallow disorder, and procedural adverse events. Time-trend of all VCE adverse events were analyzed to determine whether rates changed over the past two decades.

Data Synthesis And Analysis
All statistical analyses were performed using the meta package in R version 3.5.1, and statistical signi cance was reported when the P < .05 unless speci ed otherwise. The VCE adverse events rates were pooled using metaprop command. We applied Freeman-Tukey Double Arcsine transformations since low rates were expected [31]. Heterogeneity was assessed with the I 2 statistic, which values of 0%, < 25%, 25-75%, and ≥ 75% denoted no, low, moderate, and high heterogeneity, respectively [32]. Owing to the degree of heterogeneity found, random effects model described by Dersimonian Laird was used [33]. Publication bias was assessed mathematically using Egger's test [34]. Sensitivity analysis were performed by systematically removing each study in turn to explore its effect on each VCE adverse event rate.
Subgroup analysis was done according to different VCE type (ECE, Gastric VCE [GCE], Small Bowel CE [SBCE], CCE, and After PCE). The metareg command was used in univariable and multivariable meta-regressions to test the in uence of study-level moderators on the rate of retention, small bowel incomplete examination, swallow disorder, and procedural adverse events [35]. Seven moderators were tested including study midpoint period, study design, study region, patient groups, male percentage, mean age, and VCE type. Covariates meeting our signi cance criterion (P ≤ .1) were entered into a multivariable metaregression model. The study period midpoint and each adverse event rate were meta-regressed to explore which adverse event's rate have changed over time.

Results
The literature search resulted in 13,168 citations, 811 potentially relevant studies met the eligibility criteria were reviewed in full. After excluding ineligible reports, 402 studies were selected for systematic review and meta-analysis ( Fig. 1).

Study Characteristics
The baseline characteristics of the included studies were summarized in Table 1   The Egger's test did not indicate the existence of obvious publication bias for retention rate (P = .6063), incomplete examination rate of esophagus (P = .7632), small bowel (P = .1315), and colon (P = .1393), while for the rate of stomach incomplete examination (P = .0017), swallow disorder (P < .0001), aspiration (P < .0001), technical failure (P < .0001), and procedural adverse events (P < .0001) showed signi cant asymmetry ( Table 2). The effect estimated from the sensitivity analysis showed little change ( Supplementary Fig. 7).

Reasons and interventions of retention and Time-trends of each VCE adverse event rate
The de nite reasons for retention were reported in 610 VCEs according to 119 studies ( Supplementary Fig. 8). Crohn's disease was the most common retention reason (n = 216, 35.41%). Among the 766 retained capsules, surgery was the most frequently used intervention (n = 352, 45.95%), followed by endoscopically management (n = 199, 25.98%), no intervention (n = 176, 22.98%) and medical therapy (n = 39, 5.09%). Although there was no signi cant change in time-trend analysis of retention interventions, surgery had a downward trend and other interventions had upward trends (Supplementary Fig. 9).

Discussion
To our knowledge, this is the rst meta-analysis to report pooled rates of each VCE adverse event, assess the predictors and provide time-trend analysis over the past two decades. The results demonstrated that VCE is a safe procedure worldwide with low rate of adverse events. The risk for retention should be assumed in patients with known IBD, which is a contraindication unless intestinal patency is proven, best by the passage of an intact PCE. Elder age is the risk or protect factor for small bowel incomplete examination or swallow disorder. In addition, the rate of retention and small bowel incomplete examination were declined over the years.
Retention is the most focused adverse event since retained capsules may cause partial or complete gastrointestinal obstruction, which limits wider utilization of VCE. Compared with previous study [18], we detected a lower pooled retention rate of 0.73%. This could be attributed to the usage of PCE, which predicts small bowel strictures in high-retention risk patients [36,37]. As showed in the results, retention rate of after PCE group was 0.09% in subgroup analysis, and PCE signi cantly decreased retention rate by 5.04% in multivariable meta-regression. These ndings con rm that performing an initial PCE before VCE is useful to avoid retention. The known IBD is the predictor increase retention rate by 4.29%, which may due to the presence of associated underlying in ammatory strictures [38]. Surgery is frequently performed for retained capsules in the early years [39]; however, for asymptomatic patients or with slight abdominal pain, later studies reported more favorable clinical outcomes using endoscopic methods or medical treatment [40,41]. Our time-trend metaregressions also showed decreasing trend for surgery and increasing trends for non-surgical management.
As VCE is usually swallowed in standing position, the esophageal transit time is very short due to gravity function, resulting in few images taken and causing esophagus incomplete examination. Right supine position [27], acquired image from both ends of VCE [42], increased frame rate [43], and stringed VCE [44,45] were studied to overcome gravity effect and improved Z-line visualization. The invention of MCE provided a more viable approach for gastric examinations [6]. Since MCE cannot perform ushing and suctioning, visualization may be impaired by the presence of bubbles and mucus. The investigators have used detergents in gastric preparation, while the fundus still poor visualized [46,47]. Because battery life is limited, incomplete examination of small bowel and colon could easily occur (12.08% and 19.19%, respectively). In this meta-analysis, age was an independent predictor affected small bowel incomplete examination rate, which is similar with the previous study of Girelli et al [48]. Small bowel incomplete examination rate declined over the past two decades, which may contributed by various investigations, such as prolonged battery life span in newer capsule generation [49,50] or reduced gastric transit time (GTT). CCE was introduced with prolonged battery life, and we con rmed a signi cant relationship between CCE and lower small bowel incomplete examination rate. The methods included real-time viewer [51], administered prokinetic agents (such as metoclopramide [52,53], mosapride [54], and erythromycin [55]), endoscopically placement [56,57], and magnetic steering [58] can be used to improve the likelihood of a complete small bowel examination in routine clinical practice.
Since the VCE was introduced, it has been proven useful for many indications across a wide age range, the youngest child used VCE was only 8 months of age [59]. However, young age was an independent predictor signi cantly associated with higher swallow disorder rate. In one series, 63 of 83 children < 8 years old required AdvanCE™ placement device to deliver the VCE into the duodenum [60]. In this study, 0.75% patients were unable to swallow the capsule. It's notable that, the capsule aspiration is an adverse event relating to swallowing disorder. Although very rare (5/23,449), case reports described it may cause life threatening acute respiration distress, and over half of patients required bronchoscopy intervention after capsule aspiration [24,61,62]. Aging, neurological or swallowing disorder and patients with a weak or absent cough are high risk for VCE aspiration [63]. Fortunately, the patients included in this meta-analysis had no respiratory distress, and the problem can resolved quickly by spontaneously coughing [23,25,48,64].
The overall technical failures rate is signi cantly lower than previous study by Rondonotti et al (0.67% vs 8.59%) [25], this re ects the immature of early capsule and software prototypes. During VCE examination, male and ECE were signi cantly increased procedural adverse events rate. In esophagus examination, the use of string attachment could be able to prevent VCE rapid and unpredictable transmission, allowing controllable movement and real-time visualization. However, the retrieval of the capsule caused discomfort, usually lead to nausea and vomiting [3]. The detachable technique in string VCE avoid this problem, and comfort assessment was better than previous reports [45].
To date, this work is the largest overview including over 100,000 VCE procedures from 402 literatures. Our study has several strengths. First, compared with previous studies, this is the rst systematic review and meta-analysis comprehensively summarized the full spectrum of VCE adverse events, ranging from retention, swallow disorder, aspiration, technique failure, procedural adverse events, and incomplete examination, demonstrating low adverse event rates and safe clinical application over its two decades of use. Second, we performed meta-regression to determine the predictors of each adverse event, and identi ed new risk or protective factor (age) for small bowel incomplete examination or swallow disorder. Third, this is the rst study providing temporal changes of VCE adverse event rates. The retention and small bowel incomplete examination rates in this study were lower than previous systematic review [18] (0.73% vs 1.4%, and 12.08% vs 16.5%, respectively), and our time-trend analysis indicated rates of those adverse events declined over the past two decades, encouraging continued efforts to achieve and maintain safety targets in VCE practice.
There are several limitations. First, there were obvious heterogeneity in most VCE adverse event rates, and Egger test indicated potential publication bias for stomach incomplete examination rate, swallow disorder rate, aspiration rate, technical failure rate, and procedural adverse events rate, which may have compromised the precision of our study. Second, exclusion of studies with fewer than 50 patients may introduced selection bias to this analysis, the rate of rare VCE adverse events such as capsule aspiration may underestimated. Last, meta-regression analysis was conducted on the level of the studies, and the characteristics of studies, individual patients, and VCE could not be retrieved to identify other risk factors.
In conclusion, this systematic review and meta-analysis comprehensively summarized the full spectrum of VCE adverse events, demonstrating low adverse event rates and safe clinical application. Retention and small bowel incomplete examination rates declined over the past two decades. Patients with known in ammatory bowel disease or elder should be alerted to high risk of retention or small bowel incomplete examination. Future clinical practice and research will bene t from this knowledge and potential adverse events would be prevented. Supplementary Files