We read Sanyal and colleagues' recent article with interest [1]. In their study, transjugular intrahepatic portosystemic shunts (TIPS) were associated with a higher mortality rate than was sclerotherapy, despite similar rebleeding rates in both groups. To date, four randomized, controlled trials comparing TIPS with sclerotherapy for the prevention of recurrent variceal hemorrhage have been published in peer-reviewed journals [1-4]. In all but one trial [1], TIPS were superior to sclerotherapy for prevention of rebleeding and survival rates were similar in the two treatment groups.

We have assessed some of the potential sources of heterogeneity or reasons for differences between the trials. First, the patient populations are different. Sanyal and coworkers reported a higher percentage of patients with Child-Pugh class C cirrhosis (49%) [1] than were entered into the other trials (range, 9% to 18%) [2, 4]. Do these data show that liver cell function of patients with class C cirrhosis is better preserved after sclerotherapy than after TIPS? Were the patients with class C cirrhosis particularly vulnerable to the vicissitudes of TIPS?

Second, the interval between hemorrhage and randomization was another source of heterogeneity. The variable time of entry into the study after admission for variceal bleeding is an important confounding variable when survival data are analyzed. In these four trials, the time of entry ranged from 1 to 10 days. This may have affected the differences in mortality.

Third, in Sanyal and colleagues' study, the 32 survivors in the sclerotherapy group included 5 patients who had rescue therapy with TIPS. Failure of sclerotherapy due to uncontrolled rebleeding occurred in 8 of 39 patients. Two patients died; in the other 6 patients, TIPS stopped bleeding. At the time of Sanyal and colleagues' study, 5 of these 6 were still alive. Sanyal and coworkers noted a lower mortality rate in the sclerotherapy group. Was this due to the success of salvaging sclerotherapy failures by TIPS? The authors concluded that sclerotherapy may be superior to TIPS with respect to survival. In our opinion, the conclusion that should be drawn from Sanyal and associates' study is that sclerotherapy with TIPS rescue for uncontrolled bleeding may be superior to TIPS with respect to survival.

Finally, Sanyal and colleagues excluded 32 patients because they "could not be stabilized by endoscopic treatment and required definitive therapy within 72 hours." What was the definitive treatment? TIPS, perhaps? In this case, Sanyal and colleagues may have underestimated the potential value of TIPS as salvage therapy in such patients.