For patients with suspected small-bowel bleeding, a three-variable scoring system predicted the diagnostic outcomes of video capsule endoscopy, according to the findings of a multicenter study.
“Admission to the hospital with overt bleeding and having a hemoglobin [level] of less than 6.4 g/dL were two positive predictors of a diagnosis. Being younger than 54 years old at the time of the video capsule endoscopy exam was a significant negative predictor of a diagnosis,” Neil B. Marya, MD, of the University of California, Los Angeles, and associates wrote in Techniques and Innovations in Gastrointestinal Endoscopy.
To develop the scoring system, they analyzed retrospective data from 162 adults with suspected small-bowel bleeding who received video capsule endoscopies at the University of Massachusetts or the University of California, Los Angeles, in 2016 or 2017. Most of these individuals were outpatients with occult gastrointestinal bleeding, but nearly one-third were inpatients with overt bleeding.
In all, 70 (43%) patients had a relevant finding on video capsule endoscopy, most frequently arteriovenous malformation (26%), blood (10.5%), or ulceration (10.5%). On multivariable analysis, the odds of positive video capsule endoscopy were significantly higher when patients had been admitted to the hospital with overt bleeding (adjusted odds ratio, 2.38; 95% confidence interval, 1.05-5.39) or had a baseline hemoglobin level less than or equal to 6.4 g/dL (aOR, 2.68; 95% CI, 1.11-6.48). In contrast, patients who were younger than 54 years were significantly less likely to have diagnostic lesions (aOR, 0.25; 95% CI, 0.11-0.58). After designating these variables as A, B, and C, respectively, the researchers derived the following equation to produce the score: (0.87 x A) + (0.99 x B) – (1.38 x C). Each variable was scored as 1 (present) or 0 (absent). Based on this equation, the highest score possible score was 1.86, and the lowest possible score was –1.38.
The researchers validated this scoring system by analyzing data from 152 adults with suspected small-bowel bleeding who were examined prospectively at the two centers. The development and validation cohorts resembled each other except that the validation cohort had lower mean hemoglobin levels (8.2 g/dL versus 9.0 g/dL in the development cohort; P < .01) and higher mean blood urea nitrogen levels (25.3 mg/dL vs. 19.9 mg per dL; P =.03). Receiver operating curves were similar between the two groups (respective C-statistics, 0.70 and 0.69; P = .91).
However, the scoring system’s maximum specificity was only 30.6%, yielding a positive predictive value of only 48.6%. The associated cutoff score was greater than or equal to 0. At this cutoff, sensitivity was “at least 90%,” and negative predictive value was 83.6%. Thus, the scoring system is best suited for identifying patients who are unlikely to have a diagnostic lesion found on video capsule endoscopy, the researchers said.
“Patients with scores of less than 0 could conceivably have capsule examinations deferred,” they concluded. “For example, consider a clinician taking care of a hospitalized patient who meets criteria for undergoing video capsule endoscopy for the indication of suspected small intestinal bleeding, but finds that the patient has a suspected small-bowel bleeding capsule diagnosis score of less than 0. The clinician could decide to have their patient undergo the video capsule endoscopy as an outpatient, since the likelihood of detecting an actionable lesion is low. This decision could have a [beneficial] financial impact.”
No external funding sources were reported. Dr. Marya disclosed a recent consulting relationship with AnX Robotica. Two coinvestigators disclosed a consulting relationship with Medtronic and research support from Olympus and Medtronic.
SOURCE: Marya NB et al. Tech Innov Gastrointest Endosc. 2020 Jun 19. doi: 10.1016/j.tige.2020.06.001.
