HOME
Return to Table of Contents
112 RETROSPECTIVE ANALYSIS OF HYDROGEN BREATH TESTS FOR SMALL BOWEL BACTERIAL OVERGROWTH TO DETERMINE CONFIDENCE INTERVALS FOR BREATH HYDROGEN BY CATEGORY OF TEST-OUTCOME S Makhija1, A Rostom1, VC Dias2, AW Lyon2 AIM: Current guidelines indicate that either a breath hydrogen of =>15 ppm at baseline or an increase of 12 ppm within the initial 120 minutes of a carbohydrate load is consistent with small bowel bacterial overgrowth. We hypothesize that the current guidelines will clearly distinguish clinical patient categories. The aim of this study is to assess confidence intervals for results of the breath hydrogen test (HBT) used to categorize patients with small bowel bacterial overgrowth and to estimate the probability of false positive test results.
1Departments of Medicine; 2Pathology and Laboratory Medicine, Calgary Laboratory Services, University of Calgary, Calgary, Alberta
METHODS: A retrospective cohort that underwent the HBT in the Calgary health region (Jan 2005 - Jul 2006, n=149) was evaluated. Patient demographics were collected. Breath hydrogen was determined using a Quintron Microlyzer DP analyzer using Quintron calibration gases. Oral glucose (75g) was administered following collection of the baseline breath specimen. Patients were categorized as: Negative (<15 ppm baseline, <=12 ppm increase, 0-120 min); Positive at baseline (=>15 ppm, time zero) or Positive trend (>12 ppm increase, 30-120 min). Confidence intervals (95% CI) were determined using Analyze-IT statistical software.
RESULTS: Eight patients were positive at baseline (mean H2: 35 ppm, 95% CI: 17.5-52.5 ppm), 23 patients were positive by trend within 60 minutes (mean H2: 35.5 ppm, 95% CI: 17-54 ppm). In approximately 30% of specimens from the 118 negative patients breath hydrogen was not detectable, <2 ppm, (mean H2: 4.1 ppm, 95% CI: 3.5-4.7 ppm). Based on the distribution of breath hydrogen negative results, the statistical probability of obtaining a false positive result at baseline time or by trend was estimated to be p <0.018.
CONCLUSIONS: Applying the current guidelines, confidence intervals for BHT results clearly distinguish negative patients from positive patients and there is a low probability of obtaining false positive results due to analytical variation.