JAGS 38:1097-1104, 1990

Assessing Treatment Decision-Making Capacity

in Elderly Nursing Home Residents
L. Jaime Fitten, MD,*Richard Lusky, PhD,f and Claus Hamann, MD$

Clinicians usually employ indirect measures of cognitive Judged against our more direct assessment of decision-
and physical function in order to assess medical decision- making capacity, primay physicians’ judgment of capac-
making capacity. We tested a reference group of well el- ity for consent was 31% to 39% sensitive in identifying
derly (Mini-Mental State Exam [MMSE] score = 29.1 k impaired decision-making and the M M S E was 53% to
0.8, mean k SD),for their understanding of three 63% sensitive. These measures were 100% and 82% to
increasingly complex, hypothetical treatment situations 83 % specific in identifying intact decision-making capac-
or “vignettes”- use of a hypnotic, need for thoracocente- ity, respectively. W e conclude that (1) more directly as-
sis, and desire for CPR From this, we have developed a sessed decision-making capacity varies noticeably among
more direct, Guttman-like assessment of decision-making elderly nursing home residents and correlates in only lim-
capacity. Of 51 Veterans Affairs nursing home residents ited fashion with frequently used cognitive screening
(MMSE score = 22.4 6.91, only 33.3% demonstrated in-
_+ methods; and (2)cognitive screening tests underestimate
tact decision-making capacity by this method, whereas the prevalence of impaired decision-making capacity in
77% were felt by their primay physicians to be capable this population. For informed consent and advance direc-
of giving consent for oral surgery; 37.3% had v e y im- tives, our study suggests that decision-making capacity
paired decision-making capacity; and 29.4% were inter- should be directly, rather than indirectly, assessed.
mediate in this ability. J Am Geriatr SOC38:1097-1104,1990

C
apacity for effective decision making is an es- ularly difficult, especially among vulnerable nursing
sential element in informed consent for medi- home residents.’p8 In the absence of generally accepted
cal treatment. The growing number of frail techniques for assessing patients’ decision-making ca-
and dependent elderly requiring decisions pacities and suitable norms for performance,9-11 clini-
about medical procedures and advanced directives cians frequently turn to indirect methods of assessing
poses a special problem for health-care professionals decision-making capacity such as brief mental status
who must consider the ability of patients to give in- examinations. Alternatively, they may rely on their own
formed consent. Such patients are both more likely to be or the staffs general impressions of patient competence.
required to make critical decisions and, as the result of It is doubtful that either one of these approaches con-
the higher incidence1and prevalence2-6of chronic brain stitutes a satisfactory means of assessing decision-mak-
disease in this population, more likely to suffer from ing capacity in older patients. It is unclear whether the
impaired decision-makingcapacity. As a result, achiev- frequently used mental status examinations adequately
ing an appropriate balance between patient autonomy assess the key mental operations required in decision
and professionalbeneficence in this population is partic- making. They may provide useful markers of impair-
ment only when cognition is already substantially com-
promised. Further, it is likely that brief mental status
screening examinations and clinical impressions ob-
From the ‘Departments of PsychiatryandMedicine/Geriatrics, Uni- scure important gradations of decision-makingcapacity
versity of California Los Angeles School of Medicine, Los Angeles,
California; *Sectionon Geriatric Psychiatry, Veterans Affairs Medical that would allow patients a level of participation in de-
Center, Sepulveda,California;the tDeparfmentsof CommunityMedi- termining treatment when complete self-determination
cine and Health Care and of $Medicine/Geriatrics, University of Con- is not possible.12 Although the ability of clinicians to
necticut School of Medicine, Hartford, Connecticut; the tGerontolo-
gical StudiesUnit,Travelers Center on Aging, Hartford, Connecticut; gauge decision-makingcapacity informally remainsun-
and the +Department of Geriatric Medicine, Saint Francis Hospital studied, brief mental status screening examinations
and Medical Center, Hartford, COMdcut. have, for example, been found wanting in both sensitiv-
Address correspondence and reprint requests to L. Jaime Fitten,
MD, Veterans Affairs Medical Center, 116A-9 16111 PlummerStreet, ity and specificity when judged against more stringent
Sepulveda, CA 91343. clinical examinations and behavioral criteria.l3,14

0 1990 by fhe American Geriatrics Society 0002-8614/90/$3.50

 1098 FITTEN ET AL IAGS-OCTOBER 1990-VOL 38, NO. 10

Until limitations in assessment methodology are those employed in selecting the study’s residential sub-
overcome, it is unlikely that the concept of marginal jects, with the additional requirement that reference
(selective)decision-making capacity,15J6widely applied group members be free of diagnosed psychiatric illness
in legal practice, will gain acceptance in the field of or neurologic disease.
medicine. In law, under this principle, patients may be
Instruments and Data Collection Tests of medical
deemed competent to execute a will but not a contract,
decision-making capacity and mental status were ad-
or incompetent to handle their financial affairs but
ministered to members of both groups by the research
competent to make decisions regarding medical treat-
team’s geriatrician-geropsychiatrist (LJF) and geriatric
ments. Similarly, if patients could be regarded as com-
medicine fellow (CH). Medical decision-making capac-
petent to make some treatment decisions but not others,
ity (decisionalcapacity, D.CAPCTY)was independently
a better balance between patient autonomy and patient
assessed using an instrument that employs ”vignettes”
protection in medical decision making might be
specifically developed for this purpose by the authors
achieved, minimizing both unwarranted intrusion on
(LJF and CH).” Mental status was measured with two
autonomy and neglect of needed protection.
commonly used screening instruments, Folstein’sls
The research described in this paper addresses these
Mini-Mental State Exam (MMSE) and Pfeiffer’s19Short
issues by (1) more directly examining the medical deci-
Portable Mental Status Questionnaire (SPMSQ).Partici-
sion-making capacity of a group of aged nursing home
pants were also given a test of short-term factual recall
residents through the use of pointed clinical “vi-
based on the clinically relevant informed consent
gnettes”; (2) exploring the relationship between resi-
(C.RECALL). Each participant was asked 14 questions
dents’ decision-making capacity and assessed mental
regarding the contents of the standard VA-approved
and physical status; and (3) evaluating the efficacy of
consent form, which they had signed before entering
brief mental status screening examinations and clinical
the study. This test was given by a graduate research
impressions in identifying those with limited decision-
assistant immediately after informed consent for partici-
making capacity.
pation in the study was obtained. In addition to these
METHODS cognitively oriented measures, nursing staff rated the
participants’ level of functional dependence using
Subjects Cross-sectional data on the decision-mak-
Katz’s Activities of Daily Living (ADL)Zoand Lawton
ing capacity, mental status, assessed ability to give in-
and Brody’s Instrumental Activities of Daily Living
formed consent for treatment, and other relevant char-
(IADL).Z1Finally, Sepulveda records provided sociode-
acterstics were collected for elderly patients residing in
mographic information and, in the case of the nursing
the Sepulveda Veterans Affairs Medical Center Nursing
home residents, the primary VA staff physician’s opin-
Home. All residents aged 60 or older and capable by
ion regarding the patient’s ability to give consent for oral
Veterans Affairs (VA)standards of consenting to survey
surgery (MD.JUDGE)- the only assessment of deci-
research, which require disclosure to and verbal consent
sional capacity regularly recorded in the charts.
only from competent patients, or the legally appointed
conservator of incompetents, were considered eligible Medical Decision-Making Vignettes In order to
for the study. Patients were generally considered com- evaluate medical decision-making capacity more di-
petent unless deemed otherwise by law. Of 117 resi- rectly, each patient was presented with a series of three
dents, 90 met the age requirement for the study. Thir- hypothetical clinical stories or ”vignettes.” Each suc-
teen of the 90 potential subjects were excluded due to ceeding vignette posed a more complex treatment situa-
patient or conservator refusal, three due to their conser- tion and more complex choices with potentially more
vators’ lack of response to our request, and 10 because serious consequences for the patient. All vignettes were
they were too demented to communicate but had no written in language understandable to persons with a
conservator and therefore could be assigned no official grade school education as determined by a readability
status in our study. Another 13residents were lost to the test.22All participants in the study had at least com-
study as a result of transfer, discharge, or death before pleted grade school. Choice of subject for each vignette
the required information could be collected, leaving a as well as estimation of potential risks in each case were
total of 51 subjects who completed the study. made by consensus of two geriatricians experienced in
For the purpose of instrument development, compara- the care of institutionalized patients. Vignettes were
ble information was also collected from a reference pilot-tested with healthy persons before starting the
group composed of 15 similarly aged, but generally study.
healthy, independent, community-dwelling persons The first vignette concerned the acceptance of a po-
whose medical decision-making capacity was assumed tentially beneficial medication (a hypnotic) and its possi-
to be fully intact. These individuals were drawn from ble side effects. The second dealt with an invasive diag-
the volunteer service and outpatient clinics of the Se- nostic procedure with remote but definite hazards
pulveda Veterans Affairs Medical Center. Eligibility cri- (thoracocentesis). The final vignette addressed the
teria for membership in this group were identical to choice of cardiopulmonary resuscitation under the con-
JAGS-OCTOBER 1990-VOL 38, NO. 10 DECISION-MAKING CAPACITY IN THE NURSING HOME 1099

dition of a chronically deteriorating but not terminal participant’s performance on each item was examined
illness. In each vignette the hypothetical nature of the in relation to his or her overall performance on the
story was explained, the underlying medical issues de- index. This analysis showed that six of the original 42
scribed, and the potential benefits and hazards of inter- questions posed problems for the “impairment-free”
vention (or inaction) clarified. reference group. Three of these questions came from
Each vignette was read to the participant twice, and vignette 2 and the remainder from vignette 3. When
any questions about the material were answered. Imme- these six questions were deleted from the vignettes,
diately after the second reading, participants were asked leaving in place the core of essential questions, 14 of the
a series of questions to test their level of understanding 15 reference group members (93.3%) achieved perfect
of the vignette. An initial series of questions requiring scores on all three vignettes. At the same time, these 14
yes/no, one-word, and short-phrase answers was used individuals (by definition), and 41/51 nursing home
to test recall and understanding of the basic facts residents (80.4’%0), displayed the desired hierarchical
presented in the vignette. Knowledge of these facts was scoring pattern.
demonstrated by participant’s identification of his or Further analysis showed that failure to achieve the
her own role and that of the examiner, nature of the desired scoring pattern in the remaining cases (by, for
medical problem, and type of intervention proposed. example, passing vignette 3 after failing one or both of
Understanding of the risks and benefits as well as of the the preceding vignettes) usually resulted from missing
patient’s ability to come to a decision were also similarly only one or two items in the simpler vignettes. Based on
tested. Answers to all these questions were objectively the near-perfect scores involved, we decided to treat
rated (e.g., as correct or incorrect, present or not these deviant scoring situations as hierarchical by as-
present). A final question requiring a brief, open-ended signing medical decision-making scores of 0, 1, 2, or 3
answer was used to determine whether or not the partic- according to the most difficultvignette passed.
ipant’s decisions were derived from the facts as they Raw scores were computed for the 30-item MMSE
understood them. Participants’ answers to this question and the 10-item SPMSQ and these tests’ recommended
were subjectively rated as ”acceptable” or “not accept- minimum passing scores of 24 and eight correct items,
able” (i.e., reasoned or not reasoned from the informa- respectively, were used to define the presence of cogni-
tion given). The rationality per se of the response, how- tive impairment. Separately, participants were given a
ever, was not considered in the rating. score of 1 if they had been judged by their primary
physicians as capable of giving informed consent for
Scoring Procedures An implicit goal in designing dental work, and a score of 0 if judged incapable of
the study’s test of medical decision-making capacity giving such consent. They were assigned ADL and IADL
was to construct the vignettes and associated questions scores based on the number of activities in which they
so that generally healthy and independent elderly, such were rated fully (1point) or partially (0.5 point)indepen-
as those in our reference group, would be able to achieve dent, yielding ADL scores of 0 to 6 (based on bathing,
perfect or near-perfect scores on all three vignettes. dressing, toileting, transferring, continence, and feed-
Such performance on the vignettes would then become ing) and IADL scores of 0 to 5 (based on shopping,
the standard of comparisonfor other groups. By framing traveling, using the telephone, taking medicines, and
increasingly difficult decision-making problems, it was money management).
hoped that impaired decision-making capacity among In the test based on the VA consent form (C.RE-
participants would be manifest in an hierarchical (Gutt- CALL), preliminary tabulationsagain showed that refer-
man-like) fashion (i.e., those passing vignette 1,but fail- ence group members had experienced difficulty with a
ing vignette 2, would also fail vignette 3, etc.). This number of specific questions even though these were
pattern would yield medical decision-making capacity asked in simple language. As a result, only 6O4b of the
scores of 0 to 3, reflecting the most difficult vignette reference group (and 696 of nursing home residents)
passed; passing a vignette was defined by a perfect or achieved perfect scores on this measure as it was origi-
near-perfect score according to the standard set by the nally constructed. Based on this analysis, four questions
reference group. were deleted from the test, a new C.RECALL score was
Table 1 summarizes the participants’ performance on calculated using the remaining ten items, and a passing
the individual vignettes. From the table it can be seen score for the test was set at nine correct items. With these
that the goal of the test was not fully met with the revisions, the test classified 14 of the 15 reference group
vignettes as they were originally constructed. Only nine members (93.3%) as free of impairment in short-term
of the reference group members (6O4b)achieved perfect memory for clinically relevant material.
scores on all three vignettes. Three members of the
group (204b) failed vignette 3, and another three (209’0) Statistical Analysis Initial analysis was directed at
deviated from the desired hierarchical pattern. This insuring that the adopted selection criteria would result
marginal performance of the index led to a reconsidera- in appropriate and useful study groups. To this end, the
tion of the questions employed in each vignette. Each two groups were compared on their cognitive impair-
1100 FITTEN ET AL JAGS-OCTOBER1990-VOL. 38, NO. 10

TABLE 1. PERFORMANCE ON DECISION-MAKING VIGNETTES BEFORE AND AFTER REVISION

Nursing Home Residents Reference Group
Original Revised Original Revised
n % n YO n % n %

Performance on vignettes V1 V2 V3
FFF 22 43.0 19 37.3 - - - -
PFF 12 23.5 8 15.7 - - - -
PPF 6 11.8 6 11.8 3 20.0 - -
PPP 3 5.9 8 15.7 9 60.0 14 93.0
FPF 1 2.0 1 2.0 1 6.7 1 6.7
FFP 1 2.0 4 7.8 - - - -
FPP - - - - - - - -
PFP 6 11.8 5 9.8 2 13.3 - -
Totals 51 100.0 51 100.0 15 100.0 15 100.0
VL vignette 1; V2, vignette 2; ~ 3vignette
, 3; P, passed; F, failed.

ment and functional dependence scores, and on key cantly greater degree (and range) of cognitive impair-
sociodemographic variables. Observed group differ- ment than members of the reference group. Their mean
ences on dichotomous variables, including pass/fail score on the MMSE was 22.4 f 6.89 (mean f SD) com-
distributions on the cognitive indicators, were tested for pared to a near-perfect 29.1 k 0.80 for the reference
statistical significance using the difference of propor- group. Their mean score on the SPMSQ was 6.74 k 3.39
tions test for two samples. Differences on other nomi- compared to perfect scores of 10 for all 15 reference
nally measured variables were tested by means of theXZ group members. Their mean score on the C.RECALL
statistic. Differences in raw score performance on the test was 6.53 f 2.56 compared to 9.60 f 1.06 in the
MMSE, SPMSQ, C.RECALL, ADL, and IADL measures reference group. These observed differences in average
were tested using the difference of means (Student’s t) performance on all three indicators of cognitive func-
test. tioning achieved statistical significance at or well
Subsequent analysis focused solely on the nursing beyond the .05 level using Student‘s t-statistic, as did
home residents. Variation in decision-making capacity differences in the proportion of group members “pass-
was expressed in terms of the frequency distribution ing” each test (58% versus 100% for the MMSE, 53%
of D.CAPCTY scores. The associations between versus 100% for the SPMSQ, and 29% versus 93% for
D.CAPCTY and other indicators of functioning were the C.RECALL test. Predictably, similar differences
measured with Kendall’s tau.23 Observed associations were found in the area of functional dependence, with
were explored in depth by establishing the mean cogni- the nursing home residents achieving mean ADL and
tive impairment scores and pass/fail distributions at IADL scores of 3.96 f 1.63 and 3.40 f 1.16, respec-
each level of decision-making capacity. Analysis of var- tively, compared to scores of 5.89 k 0.33 and 4.94 f
iance and the x2 statistic were used to determine if the 0.17 in the reference group. As with cognitive impair-
observed subgroup differences might have occurred by ment, these differences were statistically significant at
chance. or beyond the .05 level by Student’s t-test. Other char-
The final stage of analysis assessed the practical utility acteristics of the two groups were comparable, apart
of cognitive screening scores in predicting decision- from differences associated with the fact of residence in
making capacity. This was accomplished by using the a VA long-term care facility. Although the nursing
pass/fail distributions for MMSE, SPMSQ, C.RECALL, home residents were, for example, significantly more
and MD.JUDGE to calculate sensitivity, specificity, and likely than the ambulatory reference group members to
predictive value of the instruments relative to observed be male (98% versus 53%) and to be unmarried (63%
impairment in D.CAPCTY. The difference of propor- versus 23%), differencesin the age, race, and education
tions test for a single sample was then used to determine of the two groups were not statistically significant.
whether the resulting predictive values represented sig-
nificant improvement over random classification of the Decision-Making Capacity, Cognitive Impair-
subjects. ment, and Functional Dependence Although about
three out of four nursing home residents (77%) were
RESULTS
judged by their physicians to be capable of consenting to
Comparability of Study Groups Comparison of the dental treatment, only one third (33.3%) achieved a
two groups showed that the study design had produced perfect score on the revised D.CAPCTY. Another 29.4%
the desired contrasts in cognitive functioning. The 51 achieved intermediate scores of 2 (13.7%),or 1 (15.7%),
nursing home residents were characterized by a signifi- and the remaining 37.3% achieved a score of 0.
IAGS-OCTOBER 1990-VOL 38, NO. 10 DECISION-MAKING CAPACITY IN THE NURSING HOME 1101

Correlation analysis showed moderately strong and impaired (D.CAPCTY score < 2). Under this criterion,
statistically significantlevels of association between the about half (53%)of the residents were still classified as
residents’ D.CAPCTY scores and their performance on impaired.
cognitivemeasures.Kendall’s tau values for these associ- When impairment in decision-making was defined by
ations ranged from 0.418 for the SPMSQ (P = .001) to the more demanding criterion of a D.CAPCTY scoreless
.546 for the C.RECALL test (P = .001). In contrast, the than 3, sensitivity values for MD.JUDGE, the MMSE
level of association between D.CAPCTY and measures and the SPMSQ were 31%, 53%,and 58%,respectively.
of functionaldependence (ADL, IADL) was low and not Under the same condition, the proportion of unim-
statistically significant. Scores on the various measures paired residents correctly identified by these three mea-
of cognitive functioning were significantly correlated, sures was higher, with specificity values of 100%,8l%,
with coefficients ranging from .441 for C.RECALL and and 75%, respectively. The highest level of sensitivity
MD.JUDGE (P = .001) to .674 for MMSE and SPMSQ (82%), but the lowest level of specificity (534b), was
(P = .001). The correlation coefficient for the two mea- displayed by C.RECALL. With these sensitivity and
sures of functional dependence (ADL, IADL) was .458 specificity levels, the positive predictive value ranged
(P < .001). from 78Y0 (C.RECALL)to 100%(MD.JUDGE).The pre-
When residents were classified into four groups ac- dictive value of a negative test result ranged from 454b
cording to their D.CAPCTY scores, and their perform- (MMSE) to 6O4b (C.RECALL). In all instances except a
ance on cognitive measures was examined, cognitive negative MMSE test result and a positive C.RECALL test
functioning scores were found to improve steadily until result, the use of the cognitive impairment measures
vignette 2 was passed, and then to plateau. As seen in represented a statistically significantimprovement over
Table 2, mean MMSEscores advanced from 17.5 f 7.19 random classification of the subjects.
among those with D.CAPCTY scores of 0, to 22.9 f When the more liberal criterion for impairment in
6.24 among those with D.CAPCTY scores of 1, to decision-making capacity was used, the sensitivity of all
27.1 k 2.41 among those with D.CAPCTY scores of 2. cognitive indicators except for C.RECALL improved by
Residents passing vignette 3 exhibited no further im- 20% to 254b with little or no change in specificity. For
provement in their MMSE scores. Performance on the the two brief mental status examinations, the liberaliza-
SPMSQ and MD.JUDGE variables exhibited the same tion of the criterion for impairment increased the pre-
general pattern whether mean scores or pass/fail distri- dictive value of a negative test result by 50% (e.g., from
butions were used. Performanceon the C.RECALL vari- 45% to 66% for the MMSE and from 46% to 69%for the
able showed steady improvement with each increment SPMSQ). Under the more liberal criterion, use of the
in D.CAPCTY, improving from a mean score of 4.37 +_ cognitive indices represented improvement over ran-
1.8 with 5.34b achieving a passing score among those dom classificationin all instancesexcept physician deter-
failing all three vignettes, to a mean score of 8.24 f 1.86 mination that a resident was capable of consenting to
with 52.9%achieving a passing score among those pass- dental care (i.e., a negative test result on MD.JUDGE).
ing all three vignettes. The analysis of variance and xz
DISCUSSION
tests performed on the subgroup means and the pass/
fail distributions affirmed that the overall pattern of In this study, an instrument for a more direct assess-
improvement in cognition with advancing D.CAPCTY ment of medical decision-making capacity was designed
scores was unlikely to have occurred by chance. and administered to a group of elderly nursing home
patients and to a reference group of generally healthy,
Utility of Cognitive Screening Employingthe sensi- independent, ambulatory elderly whose decision-mak-
tivity/specificity methodology often used to validate ing capacity was assumed to be fully intact. After refin-
biologically based screening instruments,24 MMSE, ing the instrument and scoring procedures on the basis
SPMSQ, C.RECALL and MD.JUDGE were treated as of the reference group‘s performance, the medical deci-
”screening tests” whose results were compared to classi- sion-makingcapacity of the nursing home residents was
fication of the subjects according to their D.CAPCTY examined. Analysis centered on determining the range
scores. Table 3 presents the calculated values for the of decision-making capacity among the residents, the
sensitivity, specificity, and predictive value of each level of association between decision-making capacity
measure of cognitive functioning,under two conditions. and cognitive impairment as measured by brief mental
In the first instance, impairment in decision-making ca- status examinations and related indicators, and the
pacity is signified by a less than perfect D.CAPCTY practical utility of these measures of cognitive function-
score (i.e., <3). Under this condition, two thirds of the ing in identifying residents with impaired decision-
nursing home residents (67%)were classified as having making capacity.
a degree of impairment in decision-making capacity The approach used to measure decision-making ca-
ranging from partial to complete. In the second instance, pacity in this population meets a number of important
a more liberal criterion was adopted, allowing a greater requirements for valid measurement.%By focusing ex-
number of residents to be classified as decisionally un- plicitly on the subject’s ability to make hypothetical but
1102 FITTEN ET AL IAGS-OCTOBER 1990-VOL 38, NO. 10

TABLE 2. INDICATIONS OF COGNITIVE IMPAIRMENT IN NURSING HOME RESIDENTS BY DECISION-

MAKING CAPACITY (MOST DIFFICULT VIGNElTE PASSED)
Indicator of Cognitive Impairment
D. CAPCTY MMSE* SPMSQ* C. RECALL* MD. JUDGE*
(most difficult vignette passed)* (224)t (>7) (>8) (Yes/No)

Score: 0 (none), n = 19 (37.396)

Mean 17.5 4.22 4.37 NA
SD 7.19 3.25 1.80 NA
Range 5-28 0-10 1-9 NA
Passed 26.3% 16.7% 5.3% 55.670
n (19) (18) (19) (18)
Score: 1 (Vl), n = 8 (15.7%)
Mean 22.9 7.63 7.13 NA
SD 6.24 3.42 2.42 NA
Range 11-29 1-10 3-10 NA
Passed 62.5% 62.5% 25.0% 75.0%
n (8) (8) (8) (8)
Score: 2 (VZ), n = 7 (13.796)
Mean 27.1 8.57 7.57 NA
SD 2.41 1.40 2.07 NA
Range 22 - 29 6-10 5-10 NA
Passed 85.796 85.7% 42.9% 100.090
n (7) (7) (7) (6)
Score: 3 (V3), n = 17 (33.3%)
Mean 26.0 8.31 8.24 NA
SD 4.16 2.52 1.86 NA
Range 17-30 1-10 4 - 10 NA
Passed 81.3% 75.096 52.9% 100.096
n (1 6) (16) (17) (12)
MMSE, Mini-Mental Status Examination; SPMSQ, Short Portable Mental Status Questionnaire; C. RECALL, test of short-term factual recall;
MDJUDGE, physician's opinion regarding patients ability togive consent for oral surgery; D. CAPCTY, decisional capacity; VI, vignette I ; V2, vignette
2; V3, vignette, 3, NA, not applicable.
*Differences among D.CAPCTYgroups in their scores on cognitive impairment indices were all statistically significant at or beyond the .02 level using
analysis of variance or the x 2 statistic.
th parentheses is given the passing score.
Gee text, "Scoring Procedures. "

typical treatment choices, face validity is readily demon- performed at the decision-makinglevel of the reference
strated. By including the subject's ability to understand group, thereby demonstrating a clear ability to under-
increasingly complex facts and issues and to arrive stand the essentials of treatment issues, options, and
through deliberation (e.g., weighing of risks, benefits, consequences as portrayed in the three increasingly dif-
and alternatives) at reasoned, although not necessarily ficult vignettes. On the other hand, one third of the
reasonable, decisions, the measurement approach ad- residents was unable to achieve a passing score on any
dresses much of the domain encompassed by medical vignette, causing serious concern about these residents'
decision making, thereby achieving content validity. Fi- capacity to consent to treatment. The large group with
nally, the range in decision-making capacity found intermediate decision-making capacity (29.446) clini-
among the nursing home residents and observed pat- cally substantiates the legal principle that one can be
tern of association of this range with other study vari- competent for some decisions but not for others. In ad-
ables point to a high degree of construct validity, Resi- dition to capturing the deliberative process essential to
dents were, for example, found to be well distributed decision making, the vignette method employed in this
across the range of possible D.CAPCTY scores. Deci- study directs particular attention to intermediate deci-
sion-making capacity was found to be well, but imper- sion-making capacity in the future assessment of in-
fectly, correlated with the conceptually related indica- formed consent and advance directives in elderly popu-
tors of cognitive impairment, and poorly correlated with lations.
measures of functional dependence that reflect physical One important feature of such assessment is the use
as well as cognitive impairment. of appropriately simplified language in the explanation
In our study, 37.3% of the nursing home residents of medical treatment situations, as illustrated by the vi-
JAGS-OnOBER 1990-VOL 38, NO. 10 DECISION-MAKING CAPACITY IN THE NURSING HOME 1103

TABLE 3. UTILITY OF COGNITIVE INDICATORS IN cognitively impaired on the basis of brief mental status
PREDICTING MEDICAL DECISION-MAKING and recall tests were found to have diminished deci-
IMPAIRMENT IN NURSING HOME RESIDENTS
sion-making capacity. More important, some degree of
Medical Decision-Making impairment in decision-making capacity was found in
Capacity* 45% to 60% of the subjects passing these tests or judged
D.CAPCTY D.CAPCTY capable of consenting to treatment on the basis of clini-
Indicator < 3 (%) <2 (46) cal impressions.
MMSE (n = 50)
Overall, these findings suggest that clinicians relying
Sensitivity 52.9 63.0 exclusively on the usual tools to determine decision-
Specificity 81.3 82.6 making capacity in similar patients will often misclassify
Predictive value (+) 85.7 8l.Ot them. In such instances, failure to recognize diminished
Predictive value (-) 44.8 65.5t decision-making capacity will be the most likely type of
SPMSQ (n = 49) error due to the low sensitivity of these methods. The
Sensitivity 57.6 69.2 observed plateau in cognitive functioning scores at in-
Specificity 75.0 78.3 termediate levels of decision-makingcapacity (between
Predictive value (+) 82.6 78.3t D.CAPCTY 2 and 3-see Table 3) suggests that such
Predictive value (-) 46.2 69.2t errors will be most frequent where impairment of higher
RECALL (n = 51) cognitive functions, unlikely to be detected by simple
Sensitivity 82.4 88.9
Specificity 52.9 50.0
screening tests, diminishes the patient’s ability to handle
Predictive value (+) 77.8 66.7t complex treatment decisions. Clearly, however, the
Predictive value (-) 60.0 80.0t shortcomings of the simple screening devices are not
MD.JUDGE (n = 44) limited to such situations. Judged against passing the
Sensitivity 31.3 38.5 simplest of the decision-making vignettes employed,
Specificity 100.0 100.0 the brief screening devices still failed to achieve accept-
Predictive value (+) 1oo.ot 1oo.ot able levels of sensitivity, specificity, and predictive
Predictive value (-) 52.9t 52.9t value.
For abbreviations, see Table 2. It is often assumed that misconceptions about the el-
‘See text, ‘‘Scoring Procedures. “ derly in general and traditional patterns of professional
tsignificantly different at the .05 level from the predictive values ob- paternalism lead practitioners to underestimatethe abil-
tained through random classification of the subjects (based on two-tailed ity of nursing home patients to participate meaningfully
probabilities).
in medical decision making. The findings of this study
suggest the opposite to be the case: when clinicians rely
gnettes. However, the complex form and wording of the on brief mental status examinations, tests of recall, and
standardized VA consent form probably results in the their own bedside judgment to detect impairment in
difficulties that even the reference group members expe- decision-making capacity, they are more likely to as-
rienced in the recall of its content. Accurate but simpli- sume incorrectly that decision-making capacity is fully
fied disclosuresof the material required for consent may intact. Based on this finding, additional steps may need
be of greater benefit to patient understanding than the to be taken to avoid such a potentially serious miscon-
currently popular legally and bureaucratically inspired ception, particularly when complex treatment decisions
forms. Only recently has the difficult problem of disclo- are involved. In the absence of full psychiatric, neuro-
sure simplificationin geriatric clinical settings begun to logic, and legal evaluations for aged nursing home resi-
be addre~sed.8J7,2~--?8 More research in this area is dents, it may be insufficient to explain to the resident the
clearly indicated. nature of the medical problem, to outline treatment op-
In general, brief mental status examinations, recall tions, to offer a recommendation, and to accept his or
tests, and bedside judgment by the primary physician her decision. The findings reported here suggest that,
were found to be of limited practical value in determin- under such conditions, clinicians can and should more
ing impaired decision-makingcapacity among the nurs- systematically and directly probe the patient’s under-
ing home residents studied. Although using such fmd- standing of the issues involved and the reasoning un-
ings resulted in significantly better prediction than derlying his or her treatment decision. Such an evalua-
random classification,they failed to approach the levels tion may be carried out by utilizing an approach similar
of predictive value expected of useful screening tools. to the one demonstrated in this study.
Given the prevalence of impaired decision-making
capacity found in this study, predictive values ap- ACKNOWLEDGMENTS
proaching 100% are theoreticallypossible for both posi- We recognize the support of Laurence Z. Rubenstein, MD,
tive and negative screeningresults.24Among those stud- and are grateful for the helpful suggestionsof Robert L. Kane,
ied, however, only 774b to 86% of those identified as MD, Terrie Wetle, PhD, and Darryl Wieland, PhD.
1104 FITTEN ET AL JAGS-OCTOBER 1990-VOL 38, NO. 10

APPENDIX: VIGNETTE 1 -MEDICATION 5. Molsa PK, Martilla RJ, Rinne UK: Epidemiologyof dementia in a
Finnish population. Ada Neurol Scand 65:541, 1982
Mr./Ms. -, let’s suppose for a minute that I am 6. Rocca WA, Amaducci LA, Schoenberg BS: The epidemiologyof
your doctor and you are my patient. Let’s also imagine dementia. Ann Neurol19:415, 1986
7. Thomasma DC: Personal autonomy of the elderly in long-term
for a minute that you‘ve been having a hard time sleep- care settings. J Am Geriatr SOC33225,1985
ing. Let’s suppose that you have told the nurses about 8. Steinberg A, Fitten LJ, Kachuk N: Patient participation in treat-
this problem and that you would like some help with it. ment decision-makingin the nursing home: the issue of compe-
tence. Gerontologist 26:362, 1986
The nurses have told me about it and now I have come to 9. President’s Commission for the Study of Ethical Problems in
talk to you about this. Medicine and Biomedical and Behavioral Research: Making
Let us suppose I tell you that I want to give you a Health Care Decisions, vol. 1. Washington DC, US Government
Printing Office, 1982
sleeping pill to take at bedtime. This medicine will likely 10. Drane J F The Many Faces of Competence. The Hastings Center
help you sleep better for a while, but after two or three Report, April 1985, pp 17-21
weeks, the effect of helping you sleep may begin to wear 11. Roth LH, Meisel A, Lidz C W Tests of competency to consent to
treatment. Am J Psychiatry 134:3, 1977
off. Let us suppose I also tell you that there could be 12. Stanley B Senile dementia and informed consent. Behav Sci Law
some unpleasant side effects with taking the pills. For 1(4):57, 1983
example, during the day these pills may slow down your 13. Anthony JC, LeResche L, Niaz L, et al: Limits of the ”mini-men-
tal state” as a screening test for dementia and delirium among
thinking and learning, and if you take them for a while hospital patients. Psycho1 Med 12:397, 1982
you could become dependent on them. That means, if 14. Schwamm BA, Van Dyke C, Kieman RJ, et al: The neurobehav-
you stop taking them you may become nervous and lose ioral cognitive status examination: comparison with the cogni-
tive capacity screening examination and the mini-mental state
more sleep. You could, of course, choose not to take the examination in neurosurgical population. Ann Intern Med
sleeping pills. This would leave your present sleep 107:486, 1987
problem about the same as it is now. 15. Faden RR, Beauchamp TL The concepts of informed consent
and competence, in Faden RR, Beauchamp TL (eds): A History
Do you have any questions about this story? I will and Theory of Informed Consent. New York, Oxford University
repeat it for you. Now try to answer as many of the Press, 1986
following questions about this story as you can. 16. Freedman B Competence, marginal and otherwise. Int J Law
Psychiatry 4:53, 1981
1. Am I really your doctor? 17. Fitten LJ, Hamann C, Evans GL, et al: Relationship of cognitive
impairment to decision-making competence in nursing home
2. Are you really my patient? residents: development of a competence assessment protocol. J
3. In this story, what problem are you having? Am Geriatr Soc 32(Suppl):519, 1984
4. How do I plan to help you with your problem? 18. Folstein MF, Folstein SE, McHugh PR: ”Mini-mental state,” a
practical method for grading the cognitive state of patients for
5. Why do I want to give you a sleeping pill? the clinician. J Psychiatr Res 12:189, 1975
6. Do you think this medicine may help? 19. Pfeiffer E: A short portable mental status questionnaire for the
7. Are there any problems in taking this medicine? assessment of organic brain deficit in elderly patients. J Am Ger-
iatr SOC23:443, 1975
8. What are these problems? 20. Katz S. Studies of illness in the aged: the index of ADL, a stan-
9. What would happen if you chose not to take the dardized measure of biological and psychosocial function.
medicine at all? JAMA 185:914,1963
21. Lawton MP, Brody EM: Assessment of older people: self-main-
10. As the patient, would you take this medicine or taining and the instrumental activities of daily living. Gerontolo-
not? gist 9:179, 1969.
22. Flesch R: The Art of Readable Writing. New York, Harper and
11. Can you tell me why? Row, 1974
23. Blalock, HM: Social Statistics, 2nd ed. New York, McGraw-Hill
Scoring criteria not provided. Inc, 1972, pp 418-426
24. Mausner J: Epidemiology: An Introductory Text, 2nd ed. Phila-
REFERENCES delphia, WB Saunders, 1985, pp 217-223
25. Carmines EG, Zeller RA: Reliability and Validity Assessment.
Kokmen E, Okazaki H, Schoenberg B Epidemiologic patterns Beverly Hills, CA, Sage Publications, 1979, pp 17-26
and clinical features of dementia in a d e h e d US.population. 26. Tymchuk AJ, Ouslander JG, Rader N: Informing the elderly: a
Trans Am Neurol Assoc 105:334, 1980 comparison of four methods. J Am Geriatr Soc 34:818,1986
Akesson HO: A population study of senile and arteriosclerotic 27. Tymchuk AJ, Ouslander JG, Rahbar B, et al: Medical decision-
psychoses. Hum Hered 19:546,1969 making among elderly people in long term care. Gerontologist28
Broe GA, Akhtar AJ, Andrews GR, et al: Neurological disorders (Suppl):59,1988
in the elderly at home. J Neurol Neurosurg Psychiatry 39:362, 28. Fitten LJ, Waite MS: Impact of acute medical hospitalization on
1976 treatment decision-making capacity in the elderly. Arch Intern
Kay DWK, Beamish P, Roth M: Old age mental disorders in Med 150:1717, 1990
Newcastle upon Tyne: Part I. A study of prevalence. Br J Psychia-
try 110:146,1964