Economic Aspects of Evaluation - Evaluation Methods in Biomedical Informatics - page 333

type of infection, the current, local resistance patterns of bacterial strains
at that hospital, and the level of illness of the individual, as well as many
other clinical factors. The program not only recommends whether or not a
particular patient needs an antibiotic, but also makes recommendations on
the particular drug to use, the dose, and the interval. Over the course of the
hospitalization, as clinical conditions and laboratory results change, the
program makes recommendations for the modification of the patient's
antibiotic regimen. Benefits of the program are measured in terms of the
number of antibiotics ordered, the duration of dose, the number of days of
excess antibiotics provided, and the length and costs of hospital and inten-
sive care unit stays.
The analysis clearly does not provide a comprehensive analysis of the
economic outcomes of the program. Many of the outcomes are clinical, and
there is no attempt to include evidence that the program improved life
expectancy, etc.; length of stay and avoidance of allergic reactions are the
only two clinical outcomes. Furthermore, with respect to whether costs and
outcomes were correctly valued, again the answer (for a comprehensive
analysis) must be no. There are no costs of developing and implementing
the program included in the analysis, yet the authors indicate that they have
been working on the program for over a decade, which must represent a
large amount of resources. As noted, clinical outcomes are not valued, and
benefits are essentially only calculated as a pre-post change in these
outcome variables; long-term effects (survival, etc.) are not included. There
is very little attention to sensitivity analysis, although ranges for the out-
comes found in the trial are provided. Finally, since the analysis takes place
in an environment identical to the expected use, the incidence of various
infections, antibiotic uses, etc., are exactly what would be expected to be
seen in clinical practice at a similar facility.
From an economic point of view, this study is clearly more a cost-
consequence than a true cost-effectiveness or cost-benefit analysis. Because
there was no attempt to include the costs of development and implemen-
tation, it is not possible to evaluate the benefits of duplicating this effort
elsewhere. However, the study provides a remarkable amount of useful
information, indicating that, if well conducted, any type of economic analy-
sis may be useful. It provides an excellent estimate of the amount of wasted
and inappropriate care that can be removed from the use of antibiotics in
seriously ill hospitalized patients through the use of computer-based treat-
ment algorithms.
References
1. Centers for Disease Control and Prevention. Total health expenditures as a
percent of gross domestic product and per capita health expenditures in dollars:
Selected countries and years 1960-2000 in Health, United States, 2003, Table
111 on the CDC Web site: http://www.cdc.gov/nchs/hus.htm.