As noted by Lewis in a recent Review article (Platforms for antibiotic discovery Nature Rev. Drug. Discov. 12, 371–387 (2013))1, novel strategies for developing antibiotics are needed, as infections caused by multidrug-resistant organisms (MDROs) are a leading threat to public health2,3,4. Increasingly, industry, academia and regulators have become interested in the concept of relatively small development programmes focused on areas of high unmet need, such as single-pathogen therapies to treat highly resistant or totally resistant bacterial pathogens4,5,6,7. However, given the small number of cases affected per annum, how such agents could be valued to support development programmes has remained unclear.

To help clarify this issue, we estimated the annual incidence, cost and mortality of infections caused by one of the most concerning antibiotic-resistant pathogens: carbapenem-resistant Acinetobacter baumannii (CRAB). We then developed a cost effectiveness model to determine the cost per quality-adjusted life years (QALYs) saved at various costs with a novel, pathogen-specific agent targeting CRAB in comparison with the standard of care (Tables 1,2; see methods and references in Supplementary information S1 (table)).

Table 1 Model variables and sources
Full size table
Table 2 Cost-effectiveness sensitivity analysis
Full size table

Based on published frequencies of hospital-acquired infections, the proportion of such infections caused by A. baumannii and the proportion of carbapenem-resistant isolates, we estimated that 22,950 cases of CRAB infection occur annually in the United States and 75,000 globally (in developed nations) (Table 1). Based on the number of cases and the cost per case, carbapenem resistance costs health-care systems an annual excess of US$389 million and 4,590 excess deaths in the United States, and an annual excess of $742 million and 15,000 excess deaths globally (Table 1).

A cost of $10,000 per course of therapy was used to model QALYs. Net costs per life saved were $15,265 in the United States and $50,549 globally; net costs per life years saved were $1,908 in the United States and $6,319 globally; and costs per QALYs were $3,180 in the United States and $10,531 globally (Table 1).

In sensitivity analyses, even at pricing as high as $30,000 per treatment course, the US cost per QALY remained lower than $50,000, which is a commonly used benchmark for an acceptable cost per QALY against which to set drug pricing8 (Table 2). Costs of therapy below $8,474 (in the United States) or $4,945 (globally) per course resulted in a negative cost per QALY, indicating that the novel therapy reduced health-care costs.

Even if the costs of resistance were zero, and thus there were no health-care savings enabled by reducing those costs using the pathogen-specific therapy (and conservatively not attributing any costs to the standard-of-care therapy), the costs per QALY at a treatment cost of $10,000 per course were only $20,833 in the United States and globally. Varying global health-care costs at 10–60% of US costs had a negligible effect on the global cost per QALY for a novel pathogen-specific agent.

The variable that most affected the model was the excess mortality rate attributable to resistance (that is, with ineffective current therapy, and thus improvable with new, effective therapy). There are extensive reports in the literature to indicate that carbapenem resistance increases the mortality of A. baumannii infections (see Supplementary information S1 (table)). In the unlikely event that there is no excess mortality caused by resistance, and a new therapy cannot improve mortality compared to the currently available therapy, the costs per QALY go to infinity (as no life years are saved with the new therapy). To enable more quantitative modelling of the scenario in which there is a minimal survival advantage due to the new therapy, we used only a 1% absolute increase in mortality attributable to carbapenem resistance as the lower bound of the sensitivity analysis. In this case, the cost per QALY at a $10,000 treatment cost was $63,604 in the United States and $210,619 globally. In this scenario, the model becomes extremely sensitive to the cost per treatment course. For example slightly lowering the cost per treatment course to $9,073 (in the United States) or $5,545 (globally) achieves a cost per QALY that is lower than $25,000, even if the absolute increase in mortality that is due to resistance is 1%. Even if the mortality attributable to carbapenem resistance was as low as 2–4%, or the pathogen-specific therapy only reduced mortality by 5–10%, or gained only 0.8–1.6 life years per patient, the costs per QALY remained below the $25,000 cut-off at $10,000 per treatment course.

Thus, a novel, single-pathogen agent focused on CRAB could provide benefits to the health-care system while maintaining costs well below the typical benchmarks used to define cost-effective therapy.