ABSTRACT
OBJECTIVE: Our objective was to determine sensitivity, especificity and predictive values of transcutaneous sonography for detecting gastric wall lesions. MATERIALS AND METHODS: This prospective study was performed from March 1999 to April 2000 on 150 patients referred for transcutaneous sonography by the Endoscopic Service Unit. Sonographic examinations were performed using RT 4000 General Electric equipment with 5 Mhz transducer and replenishment of stomach with fluid. All scanning was done by the same sonographer, who was unaware of endoscopic, tomographic, or upper gastrointestinal series features. Results from sonography were compared with gastrointestinal tract endoscopy. Sensitivity, specificity, and predictive values were determined using contingency statistical procedure. Sonographic examination accuracy was calculated evaluating sensitivity and specificity confidence intervals (CI). Kappa index was calculated. Diagnostic accuracy differences observed between tumoral and non-tumoral lesions by sonography were evaluated by chi 2 probe. RESULTS: Sensitivity of 85% (95% CI, from 75.2 to 94.8%) and specificity of 90% (95% CI, from 86 to 93.9%) were obtained. Positive predictability was 78% and negative predictability was 94%. Diagnostic accuracy was 87%. Kappa index was 0.717. There were 35 no false-positive results (19 tumoral lesions and 16 non-tumoral lesions), seven false-negative results (one tumoral lesion and six non-tumoral lesions) and 10 false-positive results (two tumoral lesions and eight non-tumoral lesions). Only one of 20 tumoral lesions were diagnosed by ultrasound whereas from 22 non-tumoral lesions were not diagnosed 6 (chi 2 = 3.74, p > 0.05). CONCLUSION: Transcutaneous sonography is a rapid, low cost and non-invasive method that may be useful to establish clinic diagnosis and in the first steps of gastric wall lesions evaluation, it is valuable in assessment of diagnostic orientation for the referring clinic.
