Premium
The potential value of faecal lactoferrin as a screening test in hospitalized patients with diarrhoea
Author(s) -
van Langenberg D. R.,
Gearry R. B.,
Wong HL.,
Ward M.,
Gibson P. R.
Publication year - 2010
Publication title -
internal medicine journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.596
H-Index - 70
eISSN - 1445-5994
pISSN - 1444-0903
DOI - 10.1111/j.1445-5994.2009.02102.x
Subject(s) - medicine , receiver operating characteristic , multivariate analysis , gastroenterology , confounding
Background: Nosocomial diarrhoea is common and its investigation carries a significant healthcare cost. This study aimed to determine the utility of faecal lactoferrin (FL), a readily measurable marker of intestinal inflammation, in hospitalized patients with diarrhoea. Methods: FL was quantified in consecutive faecal samples submitted to a hospital pathology laboratory. Patient data were extracted from hospital records. Receiver–operator curve (ROC) analysis was performed in a subset of patients where a decision about low or high likelihood of inflammation could be confidently made. Multivariate analyses were performed to identify associations with an elevated FL. Cost analyses were also performed. Results: A total of 511 faecal samples from 433 patients (48% male, median age 67 years) was studied. Median FL concentration was 3.4 µg/mL (range 0–288). ROC analysis indicated an optimal cut‐off value of 1.25 µg/mL (sensitivity 92%, specificity 97%, negative predictive value 97%) compared with the manufacturer's cut‐off of 7.25 µg/mL (60%, 66% and 85% respectively). Multivariate analysis at the lower cut‐off minimized potentially confounding variables. Proton pump inhibitor use independently increased (OR 2.3, 95% CI 1.5–3.8) and current smoking reduced (0.61, 0.38–0.99) the likelihood of an elevated FL. Only one out of 32 bacteriological positive samples would have been missed if FL was instituted as a screening test prior to microbiological assessment, which could have reduced laboratory‐related costs by up to 56%. Conclusion: In hospitalized patients, a normal FL effectively excludes inflammatory diarrhoea and is proposed as a screening test prior to microbiological assessment of faeces. Prospective evaluation of this approach is warranted.