important in finding this baseline. The goal is to
identify a benchmark, documenting thestatus quo.
This baseline is critical to being able to show the
impact (improvement) that the new drug will have.
Table 23.2 listssomeoftheimportantquestionsto
consider when documenting the baseline burden of
illness. Answers will not be available for every ques-
tion, neither will perfect data be always available for
those answers that can be provided. The risk–benefit
assessment of taking the time to answer each ques-
tion thoroughly versus applying some ‘quick and
dirty’ estimates to the questions should be consid-
ered. Not every program requires a large-scale major
prospective study to answer each question, for many
of the reasons discussed above. However, in the long
run, it is usually less costly in terms of time and
money to research the unknown issues before com-
mitting to the pharmacoeconomic development plan.
Thepost hocpiecemeal approach almost always
fails.
Case study: data sources
A study to document the outcomes of epilepsy
treatment, conducted by Hirsch and Van Den
Eeden (1997), illustrates some of the challenges
associated with collecting burden of illness data.
The traditional clinical measure of seizure fre-
quency is no longer considered appropriate as the
sole measure of outcome of treatment or surgical
intervention. The additional variables to document
the burden of illness that were found illustrate the
gap between the type of data desired and what is
available. Hitherto, QOL had been assessed in
epilepsy patients using no fewer than 12 different
instruments (both disease-specific and general).
The economic impact of epilepsy had previously
been assessed at a national level and in a few small
studies.
These authors wanted to describe the overall
disease impact for patients with chronic epilepsy,
using a retrospective cross-sectional design in a
managed care organization. Multiple data sources
were required, as no single data base served as a
repository for the various types of data required,
and included administrative databases, medical
charts, pharmacy databases, outpatient databases,
hospitals, laboratories, outside services, member-
ships and so on. They found that all the identified
sociodemographic variables were available in at
least one automated database, as were two of the
clinical variables, and 26 of the economic vari-
ables. None of the humanistic variables were avail-
able in any database.
In this case, about half of the data desired were
available electronically, most of which were
related to health as heavily weighted toward eco-
nomic information. To gather the remaining
desired data the investigators needed to collect
prospectively humanistic as well as some addi-
tional clinical variables (Hirsch and Van Den
Eeden, 1997). It is quite typical that clinical data
available electronically are often not complete and
therefore not very useful, and that humanistic data
are missing completely from the databases held by
Health Maintenance Organizations.
When setting out to document the burden of ill-
ness, it is critical to ensure that the patients in the
databases really are patients with the disease. In
some cases, the ICD-9 codes are known to be inac-
curate regarding patient capture, and means otherTable 23.2 Considerations when documenting baseline burden of illness
Who has the condition (men, women, children, elderly, Blacks, Asians, Caucasians, etc.)?
How long does the condition last?
What is the impact of the condition and current treatment on the patient’s functional status or QOL?
How satisfied are patients with current treatment options?
Does the disease impact productivity?
What healthcare and other resources are currently used to manage the condition?
What percentage of patients with the disease are seeking treatment?
What is the economic and humanistic impact if treatment is not received?
What treatments or interventions are currently used to manage the condition?296 CH23 PHARMACOECONOMICS: ECONOMIC AND HUMANISTIC OUTCOMES