To the Editor.
—In the March 1988 issue of the Archives, Shah and Variyam1 presented an analysis on the association of ascites and pericardial effusion. I would like to amend and to expand their conclusions.Shah and Variyam1 examined the clinical, echocardiographic, and laboratory data of 27 patients with ascites secondary to alcoholic cirrhosis. Twenty-eight patients served as a control group. The investigators found pericardial effusions (PE), by echocardiography, in 63% of patients with cirrhosis and in 11% of control patients, but they were unable to find clinical, electrocardiographic or laboratory variables that were correlated with PE status. The high proportion of PE in both patients with cirrhosis as well as in the normal group suggests that the criterion used by Shah and Variyam1 for diagnosing PE was excessively sensitive. To test the reproducibility of Shah's and Variyam's1 observations, I retrieved the records of all patients with