Empirical therapy versus diagnostic tests in gastroesophageal reflux disease: a medical decision analysis.

Our objective was to describe the conditions that determine the costs of empirical therapy in gastroesophageal reflux disease (GERD). Our design was a threshold analysis using a decision tree. The costs of medications were estimated from the average wholesale prices. The costs of diagnostic procedures were expressed as the sum of physician and facility costs. A decision tree was modeled to calculate the threshold probability of GERD, for which empirical therapy became the preferred management strategy. Bayes' formula was used to transform the sensitivity and specificity of various symptoms and the joint occurrence of multiple symptoms into disease probabilities. The decision in favor of empirical therapy is influenced by four factors: the probability of GERD, the duration or costs of GERD therapy, the costs of erroneous empirical therapy in patients with diagnosis other than GERD, and the costs of diagnostic procedures. In general, the expected benefit of saving the costs of a diagnostic procedure outweighs the costs of occasional erroneous empirical therapy. However, if surgical therapy is considered or antisecretory therapy is administered for a time period of 10 or more years, diagnostic confirmation of GERD should be sought. In the long run, the failure to differentiate between peptic ulcer and GERD results in the highest cost associated with erroneous empirical therapy. In patients with multiple characteristic symptoms of GERD, the diagnosis can be ascertained with sufficient confidence to warrant empirical therapy.