Diagnostic Yield of Endoscopy for Weight Loss: Results of a Comparative Cohort Study vs. Non-Weight Loss Controls

Abstract

Background: Gastroenterologists are commonly asked to perform pan-endoscopy in patients with weight loss. This request is based on historical series revealing a high prevalence of GI causes of involuntary weight loss (Lankisch J Int Med 2001). However, in the absence of specific GI symptoms or signs, such as bleeding, dysphagia, diarrhea, constipation, or anemia, the yield of endoscopy may be low and its appropriateness unclear. We studied patients referred for endoscopy because of weight loss, and evaluated the yield of endoscopy vs. non-weight loss controls. Methods: We performed a retrospective analysis of patients referred to the endoscopy unit of a University-based VA medical center from 2000-2010 with a primary indication of weight loss. We excluded patients who would otherwise qualify for endoscopy regardless of weight loss, such as those with positive fecal occult blood, iron deficiency, previous imaging study with a colonic abnormality, familial GI cancer syndrome, GI bleeding, or prior colon cancer. Although we allowed for diarrhea and constipation, we performed a sensitivity analysis in which this sub-group was also excluded. We gathered data on patient characteristics, duration and amount of weight loss, body mass index (BMI), and endoscopy findings. We calculated the proportion of cases in which an explanatory luminal diagnosis was identified by endoscopy, and compared this to the yield in 2 age and sex-matched control groups: 1) group undergoing screening colonoscopy; and 2) group undergoing diagnostic colonoscopy for a non-weight loss indication. We conducted separate analyses for colonoscopy and upper endoscopy (EGD). We employed chi-squared for pairwise comparisons. Results: There were 95 colons and 61 EGDs identified (mean age=65; 93% male; BMI=23). Patients lost a mean of 22 lbs over a 9 month average period. The Figure reveals the diagnostic yield of the study groups. Two of the 95 colonoscopy patients (2.1%) had primary colon cancer and 3 had colitis as the explanation for weight loss. The screening colonoscopy control group resulted in no colon cancers, and the diagnostic colonoscopy control group revealed 4 cancers (4.2%) and 1 case of colitis (1.1%). After excluding diarrhea and constipation, there was 1 case of colitis (1.8%) and no cancer. There were no significant differences among any pair-wise comparisons. Of the EGD cases, there was 1 gastric cancer (1.6%); the non-weight loss control EGDs revealed 2 gastric cancers (3.3%) (p=NS). Conclusion: In the absence of other GI “red flag” features, patients with weight loss rarely harbor an underlying luminal disease; there is no difference in diagnostic yield of endoscopy vs. age and sex-matched controls. These data suggest that pan-endoscopy for weight loss is unlikely to be cost-effective in the absence of other concurrent indications for a GI procedure.