Capsule Endoscopy and Double-Balloon Enteroscopy Rarely Matched in Mid-Gut Bleeding
In patients with mid-gut gastrointestinal bleeding, two of the most important tools in the diagnostic pathway did not line up as closely as clinicians would like. In a retrospective cohort from a referral hospital in Lima, Peru, video capsule endoscopy and subsequent double-balloon enteroscopy showed low overall diagnostic agreement, even though they aligned more often on lesion type and lesion location than on the final yes-or-no question of whether a lesion was present at all.
The cohort included 51 patients, 28 of them women, with a median age of 65 years. Most had overt bleeding, and the rest had occult bleeding. Hemoglobin values were deeply concerning at baseline, with a mean of 7.3 g/dL, a level that reflects substantial blood loss and helps explain why these patients were routed into advanced small-bowel evaluation rather than routine upper or lower endoscopy. The median interval between capsule endoscopy and double-balloon enteroscopy was 13 days, with a wide interquartile range of 5 to 29 days. In that gap, blood can stop, recur, or shift in intensity, and small lesions may become less visible, especially if they are intermittent vascular sources or subtle mucosal abnormalities.
The procedural mix also reflected real-world small-bowel practice. Of the enteroscopy procedures, 39 were antegrade and 12 were retrograde. That distribution likely tracks with lesion accessibility and the expected location of bleeding, but it also means that complete visualization of the suspected bleeding segment is not always straightforward in a disease process known for elusive, often transient findings.
Where the Two Tests Agreed — and Where They Did Not
Agreement between the two modalities was measured with the Kappa coefficient, and the overall figure landed at 0.37, with a 95% confidence interval from 0.10 to 0.64 and a p value of 0.002. In practical terms, that is low concordance. A positive capsule study did not reliably predict that double-balloon enteroscopy would rediscover the same lesion, and a negative capsule study did not exclude the possibility of a lesion later seen endoscopically.
The discordance cut in both directions. In 10 patients, or 19.6%, capsule endoscopy showed positive findings, yet double-balloon enteroscopy identified no lesion. In one patient, or 2%, capsule endoscopy was negative while enteroscopy was positive. Even in a carefully selected referral population, the small bowel proved difficult to interrogate with a single sequential approach.
These mismatches are not surprising to anyone who treats obscure gastrointestinal bleeding. Capsule endoscopy can capture fleeting mucosal abnormalities during physiologic transit, but it cannot biopsy, treat, or re-examine a suspicious area under direct control. Double-balloon enteroscopy, by contrast, allows deeper inspection and intervention, yet its yield depends on reaching the right segment at the right time and may miss lesions that were visible days earlier on capsule imaging. Bleeding angioectasias, small vascular lesions, and some tumors can behave intermittently or sit beyond the segment ultimately reached by the scope.
Lesion Type and Location Matched Better Than the Overall Diagnosis
The more reassuring finding was that agreement improved when the analysis shifted from the binary question of lesion detection to the more granular questions of lesion type and location. Concordance for lesion type reached 0.58, and concordance for lesion location reached 0.50. Both values fall into the moderate range, suggesting that when the two tests did identify abnormalities, they often pointed to the same broad pattern and the same general anatomic territory.
Vascular lesions and tumoral lesions each showed moderate agreement at 0.60. That matters because these categories have different bedside consequences. Vascular lesions often lend themselves to endoscopic therapy when reached, while tumoral lesions may trigger surgical, oncologic, or multidisciplinary evaluation. Moderate agreement in these subgroups suggests that capsule imaging can still function as a useful roadmap, even if it cannot be treated as a fully dependable substitute for tissue-directed enteroscopy.
The findings also fit the known pathophysiology of mid-gut bleeding. The small bowel is long, mobile, and difficult to map, and lesions may be tiny, intermittent, or hidden behind folds. A patient with severe anemia and overt bleeding may have a lesion that bleeds only sporadically, leaving behind a capsule image one day and a normal enteroscopic field days later. The reverse can happen as well when capsule transit is rapid, visualization is incomplete, or bleeding originates from a segment not captured well during the initial study.
What This Means for Referral-Center Practice
This cohort does not argue against capsule endoscopy or double-balloon enteroscopy. It argues against assuming that either one is definitive on its own in mid-gut gastrointestinal bleeding. The low overall agreement suggests that a negative finding on one test should be interpreted in the context of the clinical picture, the timing between procedures, and the likelihood of intermittent bleeding.
For referral centers, the study reinforces a practical sequencing problem. A diagnostic strategy that stretches across days may lose sensitivity as lesions evolve or disappear. In a patient with marked anemia and ongoing bleeding, longer gaps between studies may widen the chance for discordant results. That concern is especially relevant when the suspected lesion is vascular, since these abnormalities can be notoriously fickle and may not present with consistent stigmata from one exam to the next.
The study also points to a larger operational truth in small-bowel bleeding: diagnostic agreement is not the same as diagnostic completeness. Capsule endoscopy can identify a route toward the lesion, while double-balloon enteroscopy can confirm, characterize, and sometimes treat it. When the two disagree, clinicians are left to reconcile the difference with laboratory severity, bleeding pattern, and the anatomic reach of the procedure already performed.
Limits of the Dataset and the Clinical Questions Left Open
This was an observational, retrospective comparison from a single referral hospital, so the results reflect real-world practice but also inherit its limitations. The sample was small, the time between procedures varied, and the analysis does not resolve whether some discordance came from lesion evolution, incomplete visualization, or differences in procedural reach. The report also does not establish which test should come first in every patient, nor does it prove that one modality should replace the other.
What it does show is narrower and more concrete: in this Peruvian cohort, video capsule endoscopy and double-balloon enteroscopy agreed only modestly on whether a lesion was present, while their agreement improved when attention shifted to lesion type and location. For patients with severe mid-gut bleeding and low hemoglobin, that difference can determine whether the next step is repeat imaging, deeper enteroscopy, or a broader search for an intermittent source that has not yet revealed itself.