Accuracy of the water-siphon test associated to barium study in a high prevalence gastro-oesophageal reflux disease population: a novel statistical approach

作者：Fiorentino Eugenio^*; Matranga Domenica; Pantuso Gianni; Cabibi Daniela; Bonventre Sebastiano; Barbiera Filippo

来源：Journal of Evaluation in Clinical Practice, 2010, 16(3): 550-555.

DOI：10.1111/j.1365-2753.2009.01158.x

• barium study
• gastro-oesophageal reflux disease
• latent class analysis
• pH-metry
• water-siphon test

P>Rationale, aims and objectives
Gastro-oesophageal reflux disease (GORD) is 'a condition which develops when the reflux of gastric content causes troublesome symptoms or complications'. Instrumental diagnostic tests generally used for GORD are 24-hour pH-metry and upper gastrointestinal (GI) endoscopy but barium study associated with provocative manoeuvres such as the water-siphon test (WST), has also been used for GORD. The aim of this paper was to estimate the accuracy of several tests in patients with GORD in a tertiary care setting, focusing on WST, which is rapid and non-invasive, simple to perform and well-tolerated by patients.
Method
A total of 172 consecutive patients, symptomatic for reflux referred to a tertiary medical centre, were considered and data regarding the WST, 24-hour pH-metry, upper GI endoscopy with histology were analysed using latent class analysis, a multivariable statistical method for estimating the accuracy of tests when a gold standard is not available.
Results
The overall proportion of GORD in the sample was estimated at 0.664 [95% confidence interval (CI) = (0.589; 0.731)]. WST proved to be the most sensitive [Se = 0.886; 95% CI = (0.688; 1.000)] compared with pH-metry [Se = 0.620; 95% CI = (0.493; 0.745)] and endoscopy with histology [Se = 0.534; 95% CI = (0.273; 0.789)]. It was less specific [Sp = 0.537; 95% CI = (0.003; 1.000)] than pH-metry [Sp = 0.547; 95% CI = (0.281; 0.813)], and even less than endoscopy with histology [Sp = 0.862; 95% CI = (0.495; 1.00)]. Positive predictive values were estimated at 0.792 [95% CI = (0.721; 0.862)] for WST, 0.731 [95% CI = (0.643; 0.819)] for pH-metry and 0.886 [95% CI = (0.811; 0.961)] for endoscopy with histology. Negative predictive values were estimated at 0.707 [95% CI = (0.573; 0.841)] for WST, 0.422 [95% CI = (0.310; 0.534)] for pH-metry and 0.484 [95% CI = (0.387; 0.581)] for endoscopy with histology.
Conclusion
Water-siphon test might possibly be useful in patients with suspected GORD because it is highly sensitive and predictive. A positive outcome of the WST associated with a barium study can certainly justify upper GI endoscopy and support any pharmacological treatment of GORD.

• 出版日期2010-6