| |
Utility assessment
is well-suited to computer-assisted interviews. Utility elicitations
often include questions that must be repeated in sequence with
changing percentages. Visual props are helpful to explain the
task to respondents. Computer support can simplify those tasks.
Many investigators have developed their own automated elicitation
tools for specialized use, and a few have developed tools for
more general use. We describe a few of each of these here.
Software
to Automate Utility Elicitation
Developed
by Sumner and Nease (Sumner,
Nease et al., 1991), U-Titer supports the rating scale, TTO,
and SG techniques. It has been used in a number of different studies
involving more than 1,000 individuals in clinical domains such
as ischemic heart disease, psoriasis, breast cancer prevention,
coronary artery bypass graft surgery, low back pain, osteoporosis
prevention, mild hypertension, benign prostatic hyperplasia, HIV/AIDS,
and atrial fibrillation.
A demonstration
version of the program may be viewed at http://ilya.wustl.edu/~utiter/UtiterDemo/.
Several other
utility elicitation programs have been developed, including U-Maker
(Sonnenberg FA, U-Maker 1.0 Microcomputer Utility Assessment Program,
1993, New Brunswick, NJ) and Gambler (Gonzalez,
Eckman et al., 1992). iMPACT is a software development system
developed by Lenert that helps researchers build Internet-capable
multimedia utility elicitation software programs (Lenert,
Sturley et al., 2002). Lenert provides a comparison of features
of several general purpose computer software programs for utility
elicitation: U-Titer, Gambler, U-Maker, and iMPACT (Lenert,
Sturley et al., 2002) iMPACT can be accessed through this
website: http://pref1.ucsd.edu/lpi/externalsoftwaredetail.asp?id=17
Another
example of a general purpose utility elicitation program is ProSPEC
(Program for Surveying Preference Elicitations with Computers)
developed by Bayoumi. The program may be viewed at
http://individual.utoronto.ca/bayoumi/prospec
Application
of Multimedia Methods to Preference Assessment
Preference
assessments for health states with which a respondent has little
personal experience have been criticized on the grounds that people
have difficulty anticipating their future preferences, particularly
with medical outcomes that are unfamiliar and thus difficult to
envision (Redelmeier,
Rozin et al., 1993).
You can learn
more about the problems patients face when making preference assessments
in this Interactive Textbook at:
http://symptomresearch.nih.gov/chapter_4/sec5/chas5pg1.htm
Multimedia
software instruments integrate photographs, videos, animated graphics,
and sound; they can familiarize a respondent with a medical outcome
in a richer way than is possible with text alone. Such tools differ
substantially from other computer-based utility instruments such
as U-Titer (Sumner,
Nease et al., 1991), U-Maker, and Gambler. While those programs
are designed to facilitate utility assessment by a trained interviewer,
the multimedia tools conduct the entire interview, essentially
removing interviewer bias.
Early examples
of multimedia software instruments include programs assessing
utilities for:
As described
above, Lenert, one of the early innovators in use of multimedia
for health state descriptions for utility elicitation, has developed
a program for rapid construction of multimedia interviews, known
in its current version as iMPACT3 (Lenert,
Sturley et al., 2002).
Lenert and
colleagues have developed a computer architecture for providing
normative patient decision support over the World Wide Web (www)
(Scott,
Cher et al., 1997).
FLAIR Program
Multimedia
utility programs have also been developed for specialized purposes.
FLAIR (Functional Limitation And Independence Rating) is multimedia
software, with pictures and sound, developed specifically for
utility elicitation with computer-inexperienced older adults (Goldstein,
Miller et al., 2002). FLAIR describes dependence in activities
of daily living (ADLs), defined as needing assistance from another
person in order to perform the activity. FLAIR elicits SG ratings
for health states of dependence in one ADL at a time and then
in combinations of ADLs. The first version, FLAIR1, has been used
by more than 400 individuals ages 65 and older.
Figure
1: FLAIR2 Introductory Screen
|
|
|
Figure
2: Screen Showing SG Rating of the Health
State of Dependence Eating
|
|
|
Figure 1
is an introductory screen for FLAIR2, showing an older adult using
a computer. The program includes training modules for each activity
needed to proceed through the program starting with pointing and
clicking. After training and practice with SG ratings, the respondent
rates utility of health states of dependence in ADLs. Figure 2
shows one of the screens for the SG rating of the health state
of dependence in eating. Respondents have the option to answer
by various methods: either by "yes/no" responses to
offered risk levels, by using up/down arrows to move the amount
of risk to the desired number, or by directly typing in a number.
The bar on the right gives a visual indicator of the amount of
risk in red. Probabilities are shown in both a positive frame
("percent chance of cure") and negative frame ("percent
chance of death") to minimize bias due to framing effects
(McNeil,
Pauker et al., 1982; McNeil,
Pauker et al., 1988). Respondents typically start by using
the yes/no buttons then, as they become more familiar with the
task, change to using the up/down arrows or typing in the number.
|