Cost-effectiveness of HA-1A therapy for Gram-negative sepsis
ACP J Club. 1992 May-June;116:94. doi:10.7326/ACPJC-1992-116-3-094
Schulman KA, Glick HA, Rubin H, Eisenberg JM. Cost-effectiveness of HA-1A monoclonal antibody for gram-negative sepsis. Economic assessment of a new therapeutic agent. JAMA. 1991 Dec 25;266:3466-71.
To evaluate the cost-effectiveness of treating sepsis with HA-1A monoclonal antibody.
Cost-effectiveness analysis based on a randomized, double-blind, placebo-controlled trial.
Hospitalized patients with sepsis.
Effectiveness data were based on the results of a trial with 543 patients (mean age, 58 years) with sepsis and suspected gram-negative infection that was subsequently confirmed in 36%.
Effectiveness data were developed using the statistically significant difference in in-hospital mortality among patients with gram-negative sepsis, comparing those treated with HA-1A with those on placebo.
Main outcome measures
The incremental cost of care and years of life saved for patients with sepsis were calculated using data from the trial and using modeling techniques when clinical or economic variables were unknown. 2 different management strategies were assessed: "treat" all patients with sepsis or "test" and then treat only patients with positive test results. The "test" strategy was hypothetical and would depend on development of a rapid test for gram-negative sepsis. Sensitivity analyses were used to test several assumptions.
The incremental lifetime cost of the "treat" strategy was $5650 per patient. The average benefit to treated patients was 5.4 deaths averted per 100 treated patients. The cost-effectiveness of therapy using the "treat" strategy was
$24 100 per year of life saved. On a population basis, the annual cost of HA-1A therapy in the United States using this strategy would be $2.3 billion. Using the "test" strategy (which would reduce the number of false positives being treated), the cost-effectiveness of therapy was $14 900 per year of life saved. The corresponding annual cost for the United States population would be $1.3 billion. In the sensitivity analyses (varying the proportion of patients with gram-negative bacteremia, the life expectancy of survivors, and drug and hospitalization costs), the ratios for the treat strategy ranged from $8400 to $110 200 per year of life gained.
The cost-effectiveness ratio for treating all patients with suspected gram-negative bacteremia was $24 100 per year of life saved.
Source of funding: No external funding.
Address for article reprints: Dr. J.M. Eisenberg, Department of Medicine, Georgetown University Medical Center, 3800 Reservoir Road NW, Washington, DC 20007-2197.
This paper highlights 2 critical issues faced by hospitals and physicians as they contemplate the potential release of HA-1A, the monoclonal antibody against bacterial endotoxin that is projected to be life-saving as well as expensive. First, because the expected benefit from treatment is years of life gained, concentrating its use in patients with longer postsepsis life expectancies will be most cost-effective. At least 1 institution has considered including prognosis in their decision criteria for HA-1A (1). Second, available evidence suggests that HA-1A is only useful in treating sepsis due to gram-negative bacteremia. Hence, a critical determinant of the cost-effectiveness of HA-1A is the proportion of treated patients who have gram-negative infection. This article shows that treating populations with lower proportions of gram-negative bacteremic patients will substantially increase costs; for example, if 10% of the treated population has gram-negative infection, the cost rises to $65900 per year of life saved. This sensitivity to patient selection is stimulating pharmacy and therapeutic committees around the country to create policies that will restrict HA-1A use to well-defined patient groups. This sensitivity also led Schulman and colleagues to model the effect of a hypothetical rapid diagnostic test for gram-negative bacteremia. Including the cost of the test, the authors estimate that costs per year of life saved would decrease 38% and save $1 billion annually.
This lucid economic analysis shows how combining available data with sensitivity analysis can lead to important insights. Nevertheless, these insights, like most of the discussions of HA-1A, derive from a single clinical trial, and further trials are needed to confirm the drug's effectiveness.
James P. Patton, MD
University of Pennsylvania Philadelphia, Pennsylvania