The influence of clinical study design on cost-effectiveness projections for the treatment of gram-negative sepsis with human anti-endotoxin antibody.

PURPOSE This study was performed to compare the effect of entry criteria, patient population, and study design on outcome and projected cost-effectiveness of human anti-endotoxin antibody (HA-1A). MATERIALS AND METHODS Patients with suspected or documented gram-negative bacteremia (GNB) with sepsis syndrome or shock received HA-1A during an open-label protocol. The patient characteristics and outcome measures of this series were compared with those of a placebo-controlled randomized clinical trial (RCT) of HA-1A. Both data sets were subjected to three published cost-effectiveness models of anti-endotoxin therapy, which were derived from RCT data. RESULTS One hundred thirty-one patients (43 with gram-negative bacteremia) received HA-1A during a 19-month open-label protocol. Comparison with the RCT results demonstrated greater severity of illness and higher 28-day mortality in the open-label protocol. When projected for open-label recipients, HA-1A was considerably less cost-effective than in the original projections based on RCT-derived data. This reduction in cost-effectiveness was consistent across all three models and their respective sensitivity analyses. CONCLUSIONS Extrapolating cost-effectiveness from RCT-derived analyses to open-label usage may yield widely inaccurate projections because of only small differences in patient population and the drug administration protocol.