Professional Documents
Culture Documents
Six-Month Neuropsychological Outcome of Medical Intensive Care Unit Patients
Six-Month Neuropsychological Outcome of Medical Intensive Care Unit Patients
Objective: To examine neuropsychological function, depres- the study population was markedly higher than population norms
sion, and quality of life 6 months after discharge in patients who for mild dementia. Scores on the Geriatric Depression Scale–Short
received mechanical ventilation in the intensive care unit. Form were significantly more abnormal in the neuropsychologi-
Design: Prospective cohort study. cally impaired group than in the nonimpaired group at hospital
Setting: Tertiary care, medical and coronary intensive care unit discharge (p ⴝ .04) and at 6-month follow-up (p ⴝ .02), and
of a university-based medical center. clinically significant depression was found in 27% of impaired
Study Population: A total of 275 consecutive, mechanically subjects at hospital discharge and in 36% at 6-month follow-up.
ventilated patients from a medical intensive care unit were pro- No differences were observed between groups in quality of life as
spectively followed. At 6 months, 157 were alive, of whom 41 measured with the Short Form Health Survey-12 at discharge or
(26%) returned for extensive follow-up testing. 6-month follow-up.
Measurement and Main Results: Neuropsychological testing Conclusions: Prolonged neuropsychological impairment is
and assessment of depression and quality of life were performed common among survivors of the medical intensive care unit and
at 6-month follow-up. Seven of 41 patients were excluded from occurs with greater than anticipated frequency when compared
further analysis due to preexisting cognitive impairment deter- with relevant normative data. Future investigations are warranted
mined via surrogate interviews using the Modified Blessed De- to elucidate the nature of the association between critical illness,
mentia Rating Scale and a review of medical records. On the basis neuropsychological impairment, depression, and decreased qual-
of strict criteria derived from normative data, we found that 11 of ity of life. (Crit Care Med 2003; 31:1226 –1234)
34 patients (32%) were neuropsychologically impaired. Impair- KEY WORDS: cognitive impairment; critical illness; delirium; de-
ment was generally diffuse but occurred primarily in areas of pression; encephalopathy; mechanical ventilation; neuropsycho-
psychomotor speed, visual and working memory, verbal fluency, logical assessment; psychoactive medications; quality of life;
and visuo-construction. The rate of neuropsychological deficits in respiratory disease
R apid technological and medi- lationship between critical illness and 80% of survivors in one cohort were found
cal advances have combined cognitive outcomes, and a small but im- to have impaired memory, attention, con-
to facilitate the treatment of pressive body of evidence is emerging centration, or decreased processing speed a
critically ill patients. Indeed, that documents pervasive neuropsycho- year after hospital discharge (18), and in
many patients recover from critical ill- logical impairment among patients after another report, nearly 25% had mild cog-
ness they may not have survived a decade critical illness (14). Among patients with nitive impairment even 6 yrs after their
ago (1–5). Those who survive often fail to sepsis, encephalopathy has been reported intensive care unit (ICU) stay (19). How-
return to baseline levels of health and to occur acutely in as many as 70% of ever, there are no prospective reports de-
report diminished quality of life (6 – 8). In cases (15, 16), and diffuse neuropsycho- scribing neuropsychological impairment in
addition, the psychiatric consequences of logical deficits have been documented in the general medical ICU population. In ad-
critical illness are being studied with individuals with toxic shock syndrome (17), dition, no information is currently available
heightened scrutiny (9 –13). yet few data exist regarding the long-term on the prognostic significance of delirium
In the last decade, researchers have neuropsychological consequences of sepsis. during an ICU stay in regard to long-term
become increasingly interested in the re- In the acute respiratory distress syndrome, neuropsychological outcome. The few data
that exist for general medical patients in
studies that take into account preexisting
From the Department of Internal Medicine, Divi- Virginia Commonwealth University Health Systems, cognitive impairment suggest that long-
sions of General Internal Medicine and Center for Richmond, VA (RPH).
Health Services Research and the Geriatric Research Dr. Ely was supported, in part, by the AFAR Phar- term mental status is worse in patients
Education and Clinical Center of the Veterans Admin- macology in Aging Grant, the Paul Beeson Faculty with a history of delirium (20 –22).
istration Tennessee Valley Healthcare System, Nash- Scholar Award from the Alliance for Aging Research, We therefore undertook the current
ville, TN (JCJ, SMG, AS, BT, LM, EWE); the Division of and a K23 from the National Institutes of Health investigation to study the prevalence and
Allergy/Pulmonary/Critical Care Medicine, Vanderbilt (AG01023-01A1).
University School of Medicine, Nashville, TN (EWE); the
types of neuropsychological impairment
Copyright © 2003 by Lippincott Williams & Wilkins among medical ICU patients who had de-
Department of Psychiatry, Vanderbilt University School
of Medicine (SMG); and the Department of Psychiatry, DOI: 10.1097/01.CCM.0000059996.30263.94 veloped respiratory failure necessitating
APACHE II Knaus et al., 1985 (24) A severity of illness scoring system Global measure of severity of illness
Brussels Bernard 1997 (25) A scoring system that grades organ dysfunction Measure of organ dysfunction in six different
systems
CAM-ICU Ely et al., 2001 (28, 29) A 2-min assessment tool to detect delirium in A measure of delirium
mechanically ventilated patients
GCS Teasdale and Jennett, 1974 (30) A 15-point scale measuring pupillary, verbal, and Global measure of cognitive status
motor responses
GDS-SF Sheikh & Yesavage, 1986 (33) A 15-point screening tool designed to measure Global measure of depression
depression among geriatric populations
mBDRS Kay, 1977 (23) A 17-point surrogate instrument to screen for Gross measure of dementia and cognitive
dementia impairment
MMSE Folstein et al., 1975 (35) A brief 30-point tool assessing mental status Global measure of a restricted set of cognitive
functions
RASS Sessler et al., 2002 (31) A 10-point scale (⫹5 to ⫺5) used to rate Measure of the presence and/or degree of
agitation and sedation agitation/sedation
SOFA Vincent et al., 1996 (26) A scoring system (0–4) that provides a severity of Describes a sequence of complications in the
illness score critically ill
SF-12 Ware et al., 1996 (34) A 12-point generic quality of life measure Global measure of physical and mental quality
of life
APACHE II, Acute Physiology and Chronic Health Evaluation II; Brussels, Brussels Organ Failure Assessment; CAM-ICU, Confusion Assessment Method
for the Intensive Care Unit; GCS, Glasgow Coma Scale; GDS-SF, Geriatric Depression Scale–Short Form; mBDRS, modified Blessed Dementia Rating Scale;
MMSE, Mini Mental State Exam; RASS, Richmond Agitation and Sedation Scale; SOFA, Sequential-related Organ Failure Assessment; SF-12 ⫽ Short Form
Health Survey-12.
Digit Symbol Coding Wechsler, 1997 (36) 133 133 Psychomotor speed
Thurstone Word Fluency Thurstone and Thurstone, 1962 (37) Unlimited 1 point per word Verbal Fluency
Letter Number Sequencing Wechsler, 1997 (36) 21 21 Working Memory
Sequencing
Verbal Paired Associates Wechsler, 1997 (36) Verbal Memory
Immediate 32 32
Delayed 8 8
Digit Symbol Paired Recall Wechsler, 1997 (36) 18 18 Visual Memory
Recall
Faces Wechsler, 1997 (38) Visual Memory
Immediate 24 24
Delayed 48 48
Rey Osterrieth Complex Figure Rey, 1958 (39)
Figure
Copy 36 Visuo-construction
30 Minute Recall 36 Visual Memory
Digit Symbol Coding, Letter Number Sequencing, Verbal Paired Associates, and Digit Symbol Paired Recall are from the WAIS-III (Wechsler Adult
Intelligence Scale Battery). Thurstone Word Fluency requires a subject to list as many words as possible beginning with the letters S and C during two
timed intervals (5 mins and 4 mins). Verbal Paired Associates Immediate consists of four 8-point subscales. The REY Osterrieth Complex Figure requires
a subject to copy a complex geometric figure and to reproduce it from memory after a 30-min delay.
perceptual organization and visual memory ency Test, T-scores adjusted for age, educa- vs. 24.9, p ⫽ .92). Significant differences
(39). For a more comprehensive description of tion, and sex were obtained from the Heaton et existed between groups on measures of
these tests, which together comprise the neu- al. (43) manual. For the MMSE, Z-scores were depression and mental status at the time
ropsychological battery, refer to Lezak (40). derived from the age- and education-adjusted of hospital discharge. The mean GDS-SF
As previously detailed, this battery was de- norms of Crum et al (44). All scores were later
score was significantly lower for the pa-
signed to assess seven cognitive domains: converted to T-scores for consistency of pre-
mental status, psychomotor speed, verbal flu- sentation. Comparisons of baseline character-
tients not receiving follow-up testing
ency, working memory, verbal memory, visual istics between the impaired and nonimpaired than for the follow-up group (2.5 vs. 4.5,
memory, and visuo-construction. Each patient patients and comparisons of neuropsycholog- p ⫽ .0004). Similarly, the mean MMSE
was classified as impaired or nonimpaired on ical test scores were performed using Wilcox- score was significantly lower for nonfol-
the basis of normative population data. Pa- on’s rank-sum tests, as were comparisons of low-up than for follow-up patients (20.4
tients were considered impaired if they had follow-up and nonfollow-up patients. Compar- vs. 24.4, p ⫽ .01).
either two neuropsychological test scores that isons of GDS-SF, Short Form Health Survey, For purposes of comparison, the final
were at least 2 SD below the norm-referenced and MMSE scores at discharge and 6-month population was divided into two groups—
mean or three scores that were at least 1.5 SD follow-up and the degree of change occurring impaired (n ⫽ 11) and nonimpaired (n ⫽
below the norm-referenced mean. These cri- between hospital discharge and 6-month fol-
23)— using criteria defined a priori
teria for impairment were determined using low-up between impaired and nonimpaired pa-
published methodology according to the num- tients were also made using Wilcoxon’s rank- (“METHODS”). The baseline demograph-
ber of tests in the battery, which minimized sum tests. SAS 8.12 (Cary, NC) was used for all ics of the patients are shown in Table 4.
the probability that an individual could exceed statistical analysis. An alpha of .05 was used to The mean age of the impaired group was
the cutoff for impairment by chance at the indicate significance for all analyses. 60.3 ⫾ 16.2 yrs and, for the nonimpaired
␣-level of significance of .05 (41). For the group, was 49.2 ⫾ 13.5 yrs (p ⫽ .07),
purposes of this report, patients who had ab- with four patients 65 yrs or older in each
normal neuropsychological test results but RESULTS
respective group. The mean years of ed-
who did not meet the above stringent criteria Patient Baseline Characteristics. The ucation were significantly lower among
were classified as nonimpaired.
present study focused on a population the impaired than among the nonim-
Data Management and Statistical Analy-
sis. Descriptive analyses on these data were composed of 41 patients. Of these, seven paired group (11.3 vs. 14.1 yrs, p ⫽ .03).
performed, and mean scores and standard de- were determined to have preexisting cog- This cohort was found to have a high
viations were obtained. On Letter-Number Se- nitive impairment and were therefore ex- severity of illness and a broad range of
quencing, Digit Symbol Coding, Verbal Paired cluded from our final analysis and im- causes of respiratory failure, with no sig-
Associates, and Faces, age-corrected scaled pairment characterization, leaving a final nificant differences between the two
scores were obtained from the Wechsler Adult sample size of 34. To ensure that our final groups (p ⬎ .20 for all) in any of the three
Intelligence Scale–Third Edition (36) or sample was representative, we compared measures of severity presented in Table 4.
Wechsler Memory Scale–Third Edition (38), baseline demographic and other data ob- Hospital Outcomes (Length of Stay
and for Digit Symbol Pair Recall, age- tained at the time of hospital discharge and Disposition). The duration of me-
corrected percentiles were converted to T-
between our final sample and the 146 chanical ventilation was similar between
scores. For Verbal Paired Associates and Faces,
the averages of age-corrected scaled scores for patients not receiving follow-up testing the impaired and nonimpaired patients
immediate and delayed trials were used. For (Table 3). No significant differences were (median, 6 days [interquartile ranges,
the Rey Osterrieth Complex Figure, age norms found between groups with regard to age 4 –12 for impaired] vs. 7 days [4 to 11], p
from Spreen and Strauss (42) were used to (54.2 vs. 53.2 yrs, p ⫽ .58) or severity of ⫽ .94). The length of stay in the ICU for
obtain Z-scores. For the Thurstone Word Flu- illness as measured by APACHE II (24.9 impaired vs. nonimpaired patients was 10
DISCUSSION
This study demonstrates that in a gen-
eral medical ICU cohort, neuropsycho-
logical impairment is common among
mechanically ventilated patients after
hospital discharge. Using conservative
criteria to define impairment, we found
that nearly one third of patients were
impaired on neuropsychological testing
at 6-month follow-up. Furthermore, we
found that the mean MMSE score at hos-
Figure 1. Histogram plotting the mean T-scores of the impaired and nonimpaired groups according to pital discharge among patients not par-
the primary domains of the neuropsychological functioning tested. T-scores represent a mean of 50 ticipating in follow-up was both below
and a SD of 10. The visual memory score is the average of Faces I and II, Digit Symbol Pair Recall, and the commonly recognized impairment
Rey Osterrieth Complex Figure Recall. Mean composite T-scores are 35.4 (impaired) and 49.3 cutoff of 24 and significantly lower than
(nonimpaired). the mean score of our final sample. This
suggests that cognitive impairment is
likely more pervasive among survivors of
covariates was .0005 (i.e., a highly signif- ⫺2.2, p ⫽ .44), as were mean scores on critical illness than our estimate reflects.
icant difference between groups as cate- the Glasgow Coma Scale (9.7 vs. 10.5, p In addition, as shown by other investiga-
gorized). In other words, our categoriza- ⫽ .48). tions (8, 13, 45), this study also con-
tion scheme for impairment was Depression. At hospital discharge, the firmed that diminished quality of life and
maintained. mean score of the impaired group on the depression were experienced by survivors
Delirium and Sedation Scale Mea- GDS-SF was significantly higher (i.e., of mechanical ventilation after their ICU
surements. During the ICU stay, duration more abnormal) than the mean score of stay.
of delirium (as measured by the Confu- the nonimpaired group (6.2 vs. 3.7, p ⫽ Abnormalities found in our impaired
sion Assessment Method for the ICU) was .04) (Table 6), and 27% and 8% of pa- sample were commensurate with clinical
slightly greater for the impaired than for tients met criteria for clinically signifi- dementia of at least mild severity and
the nonimpaired group (4.5 vs. 4.2 days, p cant depression (ⱖ6) in the impaired and were most striking in specific neuropsy-
⫽ .24). Over the duration of the ICU stay, nonimpaired groups, respectively. At chological domains. For example, impair-
sedation scale scores (i.e., Richmond Ag- 6-month follow-up, group means were ment in visuo-construction was particu-
itation Sedation Scale) were lower (al- again significantly different (6.4 vs. 3.0, p larly severe, as was impairment in other
though not significantly so) for impaired ⫽ .02), with four impaired patients (36%) domains such as visual memory, psy-
than nonimpaired subjects (⫺2.6 vs. and four nonimpaired patients (17%) chomotor speed, and verbal fluency. Us-
GDS-SF 6.2 (3.1) 3.7 (3.1) .04 6.4 (4.3) 3.0 (3.4) .02
MCS-12 43.6 (12.1) 44.1 (9.0) .88 45.0 (11.6) 51.5 (11.0) .15
PCS-12 32.8 (8.9) 34.0 (8.8) .81 32.7 (11.9) 39.5 (12.6) .20
GDS-SF, Geriatric Depression Scale–Short Form; MCS, Mental Component Summary; PCS, Physical Component Summary.
All data presented as mean (⫾SD); higher scores on the GDS-SF reflect greater levels of depression; higher scores on both the MCS-2 and PCS-12 reflect
more favorable ratings of quality of life (i.e., less impairment); the range of possible scores on the GDS-SF is 0 –15; the range of possible scores on both
the MCS-12 and the PCS-12 is 0 –100.
ing behavioral rating scales for dementia logical domain involving planning, per- times higher than the expected number
as a reference (46, 47), the deficits of the ceptual organization, and visual of patients with mild dementia at a sim-
impaired group were of the type and se- memory—would likely lead to difficulties ilar or greater age (49).
verity to impair social and occupational in such tasks as driving, map reading, and The precise etiology of the neuropsy-
functioning. For example, the impaired assembling common household items chological impairment in our patients is
patients likely have recent memory prob- such as children’s toys or a set of book- unclear, although there is little doubt
lems sufficient to interfere with everyday shelves. that it is multifactorial. Chief among pos-
activities, decreased ability to perform The frequency of impairment in our sible explanations is that the toxic effects
complex tasks or handle complicated patients was much higher than the prev- of syndromes such as acute lung injury
problems, and decreased ability to inde- alence of mild dementias in older popu- and sepsis may cause brain damage, pos-
pendently perform business, financial, or lation-based samples (48). Our finding of sibly due to the systemic nature of in-
community affairs. Impairment in visuo- serious neuropsychological dysfunction flammatory and coagulopathic derange-
construction—a complex neuropsycho- in 11 of 34 patients is more than ten ments found in these processes (16, 18).
P
ings that demonstrate an association be- neuropsychological outcomes associated
rolonged neuro- tween cognitive deficits and diminished with critical illness. Neuropsychological
quality of life (63, 64). Although physical impairment is problematic for numerous
psychological im-
illness has long been associated with poor reasons and generally contributes to
pairment is com- quality of life (65, 66), it is increasingly strain on family and social support net-
apparent that even mild cognitive diffi- works, reduced occupational capacity, di-
mon among survivors of the culties can have an adverse impact as well minished life satisfaction, and the in-
(67) and that cognitive impairment is as- creased utilization of resources (73–78).
medical intensive care unit sociated with functional decline and a These results provide a mandate for both
and occurs with greater than decrease in subjective well-being (68). critical care physicians and health ser-
The prevalence of clinically significant vices researchers to invest their consid-
anticipated frequency when depression in our sample of both im- erable energies toward recognizing, un-
paired and nonimpaired patients is within derstanding, and intervening in ways that
compared with relevant nor- expected ranges based on relevant litera- will minimize the development and con-
ture that estimates that up to 30% of sequences of prolonged cognitive impair-
mative data. elderly patients experience depression ment in ICU survivors.
while in the hospital (69) and that as
many as 40% of medically ill patients are
REFERENCES
Another likely culprit is hypoxemia, depressed up to a year after hospital dis-
which has already been implicated in nu- charge (70). Although the precise causes 1. Hudson LD: Survival data in patients with
merous investigations regarding the de- of depression among the patients in our acute and chronic lung disease requiring me-
velopment of cognitive dysfunction (18, sample are unclear, they are likely related chanical ventilation. Am Rev Respir Dis
50, 51). Although we did not demonstrate to the variables of medical illness, neuro- 1989; 140:S19 –S24
a significant relationship between delir- psychological dysfunction, diminished 2. Bennett RL, Gilman SC, George L, et al:
Improved outcome in intensive care units for
ium and neuropsychological decline, this quality of life, and in some cases, ad-
AIDS-related Pneumocystis carinii pneumo-
investigation was not powered to detect vanced age—all of which have been re-
nia: 1987–1991. J Acquir Immune Defic
this association, and its absence may re- ported to be risk factors for depression Syndr 1993; 6:1319 –1321
flect a type II error. We believe that future (71, 72). 3. Kuhn C, Muller-Werdan U, Schmitt DV, et al:
studies of larger cohorts should examine In addition to the small sample size of Improved outcome of APACHE II score-
the association of severe or prolonged our study, an additional important limi- defined escalating systematic inflammatory
delirium and negative long-term cogni- tation of the present study was that pre- response syndrome in patients post cardiac
tive outcomes (20, 28, 29, 52–55). Psy- existing cognitive impairment was iden- surgery in 1996 compared to 1988 –1990:
choactive medications, used extensively tified by chart review and surrogate The ESSICS-study pilot project. Eur J Car-
and in significant doses among critically interview rather than by previous psycho- diothorac Surg 2000; 17:30 –37
4. McCarthy JT: Prognosis of patients with
ill patients, may play a causal role in the metric testing. It is important to ac-
acute renal failure in the intensive care unit:
persistence of cognitive deficits even knowledge that this will be an inevitable
A tale of two eras. Mayo Clin Proc 1996;
months after their discontinuation (56 – reality of all similar studies in which the 71:117–126
58, 59). The lower education levels found ICU population is admitted “nonelec- 5. Jardin F, Fellahi JL, Beauchet A, et al: Im-
among impaired patients may be signifi- tively” due to critical illness. We used a proved prognosis of acute respiratory distress
cant as well, particularly in light of data sensitive cutoff score for the mBDRS to syndrome 15 years on. Intensive Care Med
showing education to be protective eliminate those patients with any detect- 1999; 25:887– 889
against the development of neuropsycho- able baseline deficits to avoid falsely at- 6. Davidson TA, Caldwell ES, Curtis JR, et al:
logical deficits (60, 61). In addition, al- tributing their decline to factors related Reduced quality of life in survivors of acute
though it appears unlikely that age was to mechanical ventilation, critical illness, respiratory distress syndrome compared with
critically ill control patients. JAMA 1999;
the sole reason for the impairment ob- or ICU hospitalization. Furthermore, the
281:354 –360
served among our patients, it may also be low test scores in our 11 impaired pa-
7. Daffurn K, Bishop GF, Hillman KM, et al:
that their greater age made them partic- tients indicate deficits of a severity that Problems following discharge after intensive
ularly vulnerable to neuropsychological presumably would have been identified by care. Intensive Crit Care Nurs 1994; 10:
insults (62). Whereas our multivariable a close informant, if not documented in 244 –251
analysis confirmed highly significant dif- patient medical records, had they been 8. Hurel D, Loirat P, Saulnier F, et al: Qualify of
ferences in the neuropsychological pro- preexisting. Another limitation of this life 6 months after intensive care: Results of
files of our impaired vs. nonimpaired pa- study was that we did not formally assess a prospective multicenter study using a ge-
tients even after adjusting for age, the work status or activities of daily living neric health status scale and a satisfaction
educational level, and baseline dementia, of the cohort at long-term follow-up. In- scale. Intensive Care Med 1997; 23:331–337
9. Stoll C, Schelling G, Goetz AE, et al: Health-
future studies enrolling larger numbers vestigations of these issues would be
related quality of life and post-traumatic
of patients would clearly be more capable helpful in determining the functional im-
stress disorder in patients after cardiac sur-
of determining the interactions between pact of neuropsychological impairment. gery and intensive care treatment. J Thorac
these covariates and long-term neuropsy- The current investigation provides ev- Cardiovasc Surg 2000; 120:505–512
chological sequelae. idence of the prevalence of cognitive im- 10. Schelling G, Stoll C, Haller M, et al: Health-
Our documentation of depressed qual- pairment in patients after ICU hospital- related quality of life and posttraumatic
ity-of-life scores in the population as a ization and contributes to a growing body stress disorder in survivors of the acute re-