In a recent editorial [1], Cross noted that several trials of antiendotoxin therapies have shown trends toward increased mortality in patients who do not have documented gram-negative bacteremia. He proceeded to speculate, as have others, that "strategies directed at modulating host defenses in sepsis may be two-edged swords" that "actually may accelerate the infectious process" in some patients.

Given these legitimate concerns, it is surprising that he also speculated that all-cause mortality may not be the appropriate end point for determining the efficacy of a potential therapeutic agent and that reversal in a relevant endotoxin-induced physiologic variable may be appropriate. The physiologic response of the sepsis syndrome is complex, and it is plausible that the "two-edged swords" aimed at interfering with this process may indeed succeed in "reversal of relevant physiologic variables" while increasing mortality by interfering with important defense mechanisms. Although mortality typically exceeds 30%, physiologic variables and associated illnesses ultimately resolve in survivors. It would not be reasonable to recommend therapies that shorten the duration of illness at the price of higher mortality. Because such effects are biologically plausible, we must test for them clinically and statistically. To exclude the possibility of increased mortality, the confidence interval around the effect of a therapy must lie completely on the side of improved mortality, thereby showing significantly improved mortality. Mortality is common and of over-riding clinical importance in sepsis. Newer agents may increase mortality, and the only way to be confident that an agent does not increase mortality is to show that it reduces it. Thus, mortality is the appropriate primary end point.

The plausibility of an antiendotoxin antibody adversely affecting persons who do not have gram-negative bacteremia is not addressed in the editorial [1] or the article [2]. It is worth noting that the possibility of antiendotoxin antibodies affecting persons without gram-negative bacteremia is real, given that such patients may have focal gram-negative infection or undocumented gram-negative bacteremia. In additional, endotoxin transported across ischemic bowel walls can play an important role in animal models of shock from various causes.