J. Tsevat et al., HEALTH VALUES OF HOSPITALIZED-PATIENTS 80 YEARS OR OLDER, JAMA, the journal of the American Medical Association, 279(5), 1998, pp. 371-375
Context.-Health values (utilities or preferences for health states) ar
e often incorporated into clinical decisions and health care policy wh
en issues of quality vs length of life arise, but little is known abou
t health values of the very old. Objective.-To assess health values of
older hospitalized patients, compare their values with those of their
surrogate decision makers, investigate possible determinants of healt
h values, and determine whether health values change overtime. Design.
-A prospective, longitudinal, multicenter cohort study. Setting.-Four
academic medical centers. Participants.-Four hundred fourteen hospital
ized patients aged 80 years or older and their surrogate decision make
rs who were interviewed and understood the task. Main Outcome Measures
.-Time-trade-off utilities, reflecting preferences for current health
relative to a shorter but healthy life. Results.-On average, patients
equated living 1 year in their current state of health with living 9.7
months in excellent health (mean [SD] utility, 0.81 [0.28]). Although
only 126 patients (30.7%) rated their current quality of life as exce
llent or very good, 284 (68.6%) were willing to give up at most 1 mont
h of 12 in exchange for excellent health (utility greater than or equa
l to 0.92). At the other extreme, 25 (6.0%) were willing to live 2 wee
ks or less in excellent health rather than 1 year in their current sta
te of health (utility less than or equal to 0.04). Patients were willi
ng to trade significantly less time for a healthy life than their surr
ogates assumed they would (mean difference, 0.05; P=.007); 61 surrogat
es (20.3%) underestimated the patient's time-trade-off score by 0.25 (
3 months of 12) or more. Patients willing to trade less time for bette
r health were more likely to want resuscitation and other measures to
extend life. Time-trade-off score correlated only modestly with qualit
y-of-life rating (r=0.28) and inversely with depression score (r=-0.27
), but there were few other clinical or demographic predictors of heal
th values, When patients who survived were asked the time-trade-off qu
estion again at 1 year, they were willing to trade less time for bette
r health than at baseline (mean difference, 0.04; P=.04). Conclusion.-
Very old hospitalized patients who could be interviewed were able, in
most cases, to have their health values assessed using the time-trade-
off technique. Most patients were unwilling to trade much time for exc
ellent health, but preferences varied greatly. Because proxies and mul
tivariable analyses cannot gauge health values of elderly hospitalized
patients accurately, health values of the very old should be elicited
directly from the patient.