AIM: To compare the diagnostic yield of capsule endoscopy (CE) with that of double-balloon enteroscopy (DBE).

METHODS: Pubmed, Embase, Elsevier ScienceDirect, the China Academic Journals Full-text Database, and Cochrane Controlled Trials Register were searched for the trials comparing the yield of CE with that of DBE. Outcome measure was odds ratio (OR) of the yield. Fixed or random model method was used for data analysis.

RESULTS: Eight studies (n = 277) which prospectively compared the yield of CE and DBE were collected. The results of meta-analysis indicated that there was no difference between the yield of CE and DBE [170/277 vs 156/277, OR 1.21 (95% CI: 0.64-2.29)]. Based on sub analysis, the yield of CE was significantly higher than that of double-balloon enteroscopy without combination of oral and anal insertion approaches [137/219 vs 110/219, OR 1.67 (95% CI: 1.14-2.44), P < 0.01), but not superior to the yield of DBE with combination of the two insertion approaches [26/48 vs 37/48, OR 0.33 (95% CI: 0.05-2.21), P > 0.05)]. A focused meta-analysis of the fully published articles concerning obscure GI bleeding was also performed and showed similar results wherein the yield of CE was significantly higher than that of DBE without combination of oral and anal insertion approaches [118/191 vs 96/191, fixed model: OR 1.61 (95% CI: 1.07-2.43), P < 0.05)] and the yield of CE was significantly lower than that of DBE by oral and anal combinatory approaches [11/24 vs 21/24, fixed model: OR 0.12 (95% CI: 0.03-0.52), P < 0.01)].

CONCLUSION: With combination of oral and anal approaches, the yield of DBE might be at least as high as that of CE. Decisions made regarding the initial approach should depend on patient’s physical status, technology availability, patient’s preferences, and potential for therapeutic endoscopy.

Small intestine is the middle part of digestive tract, due to its anatomic location, structural characteristics, and physiological functions, it is difficult to inspect via gastroscopy and colonoscopy. Push endoscopy is limited in that it only allows inspection of the proximal small intestine for variable distances. Intraoperative endoscopy has not met with widespread acceptance for the requirement of open laparotomy with surgically assisted passage of the endoscope through the intestine, although it has been considered the most reliable procedure for several years[1]. The diagnostic yield and accuracy of other modalities such as barium study, DSA, radionuclide and computerized tomography, magnetic resonance imaging is poor. All these make the management of small bowel disease exceptionally difficult.

The techniques of capsule endoscopy (CE) and double-balloon enteroscopy (DBE) are regarded as excellent tools for the diagnosis and treatment of small-bowel disease, as well as for complementary procedures. Considerable research has demonstrated that they were more effective than other diagnostic modalities[2,3]. It’s confusing, however, when a choice should be made between the two advanced tools. Gay et al[4] concluded that indication and route for DBE could be determined according to the outcome of CE. The protocol has been widely accepted in developed countries and it seems effective, but it means that patients should undergo and pay for both of these two expensive items. It’s not clear whether the above protocol was a good choice judging from the cost benefit. The aim of the present study was to assess the diagnostic efficacy of CE and DBE and to determine the optimal diagnostic procedure for patients with small bowel disease.

One hundred and sixty-three pieces of literature were initially identified using the search strategy described. One hundred and fifty-three articles were excluded after preliminary review for not having comparative studies, leaving 10 for detailed evaluation by two independent reviewers. Among these 10 potentially appropriate studies, an article retrospectively studied the yield of CE and DBE[5] and a review[6] were excluded. Finally, eight pieces of literature with 277 subjects met inclusion criteria, to include 5 fully published articles and 3 abstracts (Figure 1).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Trial search flow for meta-analysis.

Description of the studies

Most patients suffered obscure GI bleeding. Besides, two patients with diarrhea and one patient with unclear weight loss were included in Damian’s study[7]. Three patients in Wi’s study[8] complained of chronic abdominal pain, one of them was excluded from statistics because his final diagnosis was functional GI disease. Nine patients with polyposis in Matsumots’s study[9] were excluded from data analysis because their diagnosis had been made in advance and multi-segments of GI tract were involved.

In Nakamura’s study[10], two patients did not undergo DBE for severe cardiopulmonary impairment, one patient with anal bleeding was identified as having colon cancer before DBE, one patient suddenly declined DBE just before the examination, so the data of the 4 patients were deleted from statistical analysis. All patients of the other 7 studies underwent both CE and DBE.

Double blind method was used in five studies: physicians evaluating CE and DBE should be blinded to the results of the other method. Matsumoto et al[9] did not mention this aspect in their paper. In Mehdizadeh’s[11,12] study, CE was performed prior to DBE. In Hadithi’s trial[13], the endoscopists were aware of the identity and clinical presentations of the patients, and of the results of the VCE at the time of performing DBE, but the outcome of his study indicated that the detection rate of CE was superior to DBE. Lesions causing GI bleeding were classified and counted in six studies. Capsule retention occurred in three patients. No DBE related adverse events were reported (Table 1).

Researcher (yr)	Case numberenrolled in study	Mean age	Male patientsnumber	Case numberenrolled in analysis	Positive findings	Complication	Double blindmethod
Kameda N[14] (2006)1	24	62	9	24	15/16	1/0	Yes
Hadithi M (2006)	35	63	22	35	28/21	0	No
Nakamura M (2006)	32	59	21	28	17/12	0	Yes
Matsumoto T (2005)	13	48	7	13	10/6	0	Yes
Wi JH (2006)1	11	44	5	10	7/9	0	Yes
Damian U (2006)1	28	60	Not reported	28	19/14	0	Not reported
Mehdizadeh S (2006)	188	58	85	115	63/57	0	No
Zhong J[15] (2004)	24	39	16	24	11/21	2/0	Yes

¹Abstracts from national meetings.

Among 277 subjects of the 8 studies, one hundred and seventy patients produced findings via CE and 156 by DBE. Homogeneity test of each OR displayed that heterogeneity did exist, the random model was used for calculations. The summary OR was 1.21 (95% CI: 0.64-2.29). There was no significant difference between the yield of CE and DBE (Figure 2).

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Comparison of the yield of CE and DBE (All studies). The existence of heterogeneity was confirmed (chi² test, P = 0.02), random model was used. The summary OR was 1.21 (95% CI = 0.64-2.29). There was no significant difference between the yield of CE and DBE (P = 0.56).

Sub analysis

Homogeneity test displayed that there was heterogeneity in the meta-analysis. After reviewing each study intensively, we discovered that there were two insertion approaches applied in DBE procedures: oral and anal approaches. In some studies, two insertion approaches were combined: if no lesions were identified by one approach, the other one was performed, or all patients underwent both of the two approaches; in the other studies, most patients underwent just single DBE procedure, regardless of whether lesions were found or not. Single insertion approach could hardly cover the whole GI tract, so the detection rate it yielded was inevitably lower compared to the combination of both approaches. Therefore, we performed sub analysis according to the DBE insertion approaches used in each study. The method that DBE should be performed in each patient with combination of two insertion approaches was not used in 5 studies including 219 subjects, while the combination method was used in 2 studies including 48 subjects. Wi et al[8] did not mention the DBE insertion approaches used in his abstract, therefore the data was given up in further sub analyses. The result of sub analysis indicated that the yield of CE was significantly higher than that of DBE without the combination of the two insertion approaches [137/219 vs 110/219, fixed model: OR 1.67 (95% CI: 1.14, 2.44), P < 0.01)] (Figure 3A); while the detection rate of DBE with combinatory oral and anal routes was higher, though not significantly, than that of CE, [26/48 vs 37/48, random model: OR 0.33 (95% CI: 0.05, 2.21), P > 0.05)] (Figure 3B). Unfortunately, the sample size in this study was not large enough to lead to an absolute conclusion, we expect that this difference will tell us more if the we enlarge the sample size.

Open in New Tab   Full Size Figure   Download Figure

Figure 3 A: Comparison of the yields of CE and DBE without combination of the two insertion approaches. The yield of CE was significantly higher than that of DBE when the combination of two approaches was not used (fixed model, P = 0.009); B: Comparison of the yields of CE and DBE with the two insertion approaches combined. The yield of DBE with combinatory oral and anal routes was moderately higher than that of CE (random model, P = 0.25).

Meta-analysis focused on fully published literature on obscure GI bleeding

Since meta-analyses are rigorous exercises that need to be subject to peer review, it seems that the data of abstracts from national meetings may not be reliable; abdominal pain and diarrhea will have incredibly low yields on DBE and CE and most will be functional, therefore, we focused the analysis on the fully published papers concerning obscure bleeding. Focused analysis also revealed the significantly higher yield of CE compared to DBE when the combination of two approaches was not used [118/191 vs 96/191, fixed model: OR 1.61 (95% CI: 1.07, 2.43), P < 0.05)] (Figure 4A); and the yield of CE was significantly lower than that of DBE when combinatory insertion approaches were used [11/24 vs 21/24, fixed model: OR 0.12 (95% CI: 0.03-0.52), P < 0.01)], just one study was included however (Figure 4B).

Open in New Tab   Full Size Figure   Download Figure

Figure 4 A: Comparison of the yield of CE and DBE without combination of the two insertion approaches (focused on the fully published papers on obscure GI bleeding). The yield of CE was significantly higher than that of DBE when the combination of two approaches was not used (fixed model, P = 0.02); B: Comparison of the yield of CE and DBE with 2 insertion approaches combined (focused on the fully published papers on obscure GI bleeding). The yield of DBE with combinatory oral and anal routes was significantly higher than that of CE (fixed model, P = 0.004), just one study was included however.

CE and DBE are both advanced inspection items. They have common indications and quite different features. CE can cover the whole GI tract; the procedure requires no sedation[6] and is better tolerated. Its major limitations are the inability to obtain a biopsy, precisely localize a lesion, or perform therapeutic endoscopy. Additionally, it may provide false-positive and false-negative findings due to its incontrollable movement and low-resolution pictures it takes. It was recently reported that CE missed an advanced small-intestinal cancer that was later diagnosed with push enteroscopy[16], suggesting that CE is not an exclusive procedure but should instead be complementary to other diagnostic tools in the assessment of small intestinal pathology. On the contrary, DBE has more advantages in that it is much better adjusted because the movement of the scope can be handled according to the viewing angle and observation time needed; it can provide high-quality pictures, biopsy availability and therapeutic endoscopy[17,18]. DBE can be considered the gold standard if the whole small intestine were inspected. It appears to be equally as effective for the management of small-bowel lesions compared with intraoperative enteroscopy, and is associated with fewer complications[19,20]. Alternatively, it is not able to visualize the whole GI tract unless a combination of oral and anal insertion approaches is performed. The procedure is invasive and not as well tolerated as CE, requiring additional staff, typically two physicians or an additional nursing assistant.

The meta-analysis indicated that the yield of CE was higher compared to DBE with a single insertion approach, but might be lower than that of DBE with a combination of oral and anal approaches. Currently, most patients will have DBE that is capsule-directed, and therefore physicians should be able to find the majority of lesions that are causing bleeding, taking into account that the course is less invasive and more likely to be accepted. Though CE is a highly sensitive modality for the diagnosis of small bowel disease, it cannot always provide a right instruction for DBE due to the potential false-positive and false-negative findings produced. In Mehdizadeh’s series, CE found a potential bleeding source in 63 patients, but 22 (34.9%) of them received negative results in the following DBE procedure. In Hadithi’s series, eight (28.6%) patients had positive findings on CE, whereas these lesions could not be verified by DBE. The inclusion of all lesions detected by CE, even trivial, may partly explain the large number of false positives. The fact that in the majority of cases DBE actually visualized only the jejunum, thereby decreasing its detection rate may provide an alternate explanation. It is clear that a single insertion procedure of DBE could not provide a full-scale evaluation for the multi-segment involving or multi-segment originating diseases like Crohn’s disease and polyposis. In fact, we cannot determine whether an identified lesion is localized until the whole GI tract was inspected, therefore a second DBE should usually be considered. The procedure of DBE in combination with the two insertion approaches may prove to be more sensitive and the outcomes would be more reliable. At the same time, there will be no need for the course and charge of CE if the diagnosis was made by a single DBE procedure or when a second DBE could not be avoided. This would be more economical in some countries like China (Table 3).

Funnel plot (Figure 5) for the analysis of publication bias was performed on the six studies in which the DBE was carried out with single insertion approach, visual inspection of the plot revealed no evidence of publication bias. The number of the fully published papers was too small to analyze the publication bias and more studies on comparison of the yield of DBE with combinatory insertion approaches to CE are needed.

Open in New Tab   Full Size Figure   Download Figure

Figure 5 Funnel plot for the analysis of publication bias on the studies in which the DBE was performed without combination of the two approaches.

In summary, our study displayed that CE and DBE are both effective modalities for diagnosis of small bowel diseases. A capsule-directed DBE procedure might be better tolerated; the procedure of DBE with combination of oral and anal approaches might be more sensitive and the outcome might be more reliable (Figure 6). Choice for initial test should depend on patient’s physical status, available technology, patient’s preferences, and potential for therapeutic endoscopy. On the other hand, more studies are needed to evaluate the yield and accuracy of DBE with combinatory oral and anal insertion approaches compared to CE.

Open in New Tab   Full Size Figure   Download Figure

Figure 6 Diagnosing flow according to DBE performed based on a capsule-directed strategy or a combinatory oral and anal routes strategy.