Economic Aspects of Evaluation - Evaluation Methods in Biomedical Informatics - page 332

The study by Wang et al. 15 examines the financial costs and benefits of
instituting a hypothetical electronic medical record. No attempt is made to
associate the effects with clinical outcomes; the study is only concerned with
financial results such as increased charge capture, decreased drug utilization,
and decreased costs in maintaining and pulling paper charts. The article uses
data from the published literature, the authors' own electronic medical
record system, and expert opinion to make assumptions regarding the costs
and benefits of system implementation.The setting of the study is a U.S. prac-
tice in which some of the patients are fully capitated (the practice only
receives a set amount per month to care for these individuals, and must pay
for health care expenditures out of that pool) and some of the patients have
fee-for-service insurance, in which the practice is paid according to a bill that
is generated by the practice. The practice, therefore, accrues benefit if it
reduces expensive medication use and laboratory tests in the capitated group,
and accrues benefit if it can improve and enhance billing practices in the fee-
for-service group. The analysis is carried out for 5 years, with setup and ini-
tialization costs only accruing in year 1, and maintenance costs, license fees,
and other recurring costs charged in the years that they occur. The costs and
expected changes in expenditure for various laboratory tests and the increase
in billing accuracy for patients and other benefits are listed, with ranges
placed on those estimates. The authors calculate that, on average, the net
benefit from instituting an electronic medical record was over $86,000 per
provider over a 5-year period of time, or $13,000 per provider per year.
With respect to the first question posed by Drummond et al., the answer
is mixed: the authors provide only a single alternative (electronic medical
record, EMR). However, this is the relevant comparison, and multiple pos-
sible EMRs could be evaluated through the extensive sensitivity analyses.
With respect to the second point, although extensive analysis was com-
pleted regarding the various costs and benefits, all the values represent only
the financial gains: no clinical gains are included, making the analysis under-
value the EMR. The analysis clearly incorporated uncertainties in the esti-
mates of input parameters and conducted multiple one-way and several
multiway sensitivity analysis of critical estimates to determine how robust
the analysis was to variability in assumptions. Finally, the last criterion
regarding validation asks whether the outcomes and costs are related to the
baseline risks of the population: this requirement relates much more
directly to the economic analysis of a particular disease process, and is not
particularly relevant in this study. On the whole, this study adheres to formal
methodological recommendations quite well.
The second study by Evans et al. 16 details the effects of the implementa-
tion of a computerized program to help clinicians choose the proper antibi-
otic regimen for seriously ill individuals in the intensive care unit at a large
academic hospital, using a pre-post design where the outcomes are mea-
sured for a year before implementation and then for a year after imple-
mentation. The program has multiple clinical algorithms that evaluate the