Professional Documents
Culture Documents
Rating Scale-Revised-98:
Comparison With the Delirium
Rating Scale and the Cognitive
Test for Delirium
Paula T. Trzepacz, M.D.
Dinesh Mittal, M.D.
Rafael Torres, M.D.
Kim Kanary, B.S.
John Norton, M.D.
Nita Jimerson, M.S.N.
symptoms, and physical etiology) and add to specificity, chiatrically assessed. Subjects were evaluated on the
but are not easy to rate repeatedly during serial admin- hospital unit cross-sectionally, except for a few delirious
istrations within an episode of delirium. Some research- subjects who were retested after their delirium im-
ers have solved this problem by modifying the DRS into proved. Demographic data were obtained from the pa-
a 7- or 8-item scale after the initial administration. The tient, chart, staff, and/or family and friends.
DRS item for psychomotor behavior combines hypoac-
tivity and hyperactivity, thereby limiting its usefulness Procedures
in assessing motor subtypes of delirium. Various cog- Evaluations were done according to the availability of
nitive deficits are combined into one item because sepa- raters in a quasi-randomized fashion. The research team
rate bedside cognitive tests were originally suggested as contacted the relevant service and requested a list of
adjuncts to the DRS. However, not having a separate patients who could be approached for inclusion in the
item for attentional deficits makes it difficult to study study. DRS and DRS-R-98 ratings were done blind to
attention’s presumed cardinal role in delirium or what diagnosis by the study psychiatrists, who were trained
symptoms constitute “clouding of consciousness.” The to use these instruments. The research assistants
lack of items for language impairment or thought pro- screened cases for suitability. Ratings made use of in-
cess abnormalities limits study of what actually contrib- formation from all available sources, including discus-
utes to “confusion” other than cognitive deficits. Thus, sions with caregivers or visitors to obtain information,
the DRS has some limitations for use in phenomenologic as well as limited chart review under supervision of the
and longitudinal treatment research. research assistant or referring physician to maintain
Our revision of the DRS was intended to address blindness to diagnosis. Delirium scale ratings covered a
these shortcomings of the original scale. The revision 24-hour period.
includes two sections: three diagnostic items for initial Psychiatric diagnoses were made by the referring ser-
ratings and a 13-item severity scale that is used for re- vice physician using DSM-IV criteria and all available
peated measurements. Severity items were revised to em- usual clinical information to finalize the diagnosis. The
phasize gradations of symptom intensity; specific char- referring physician also completed the Clinical Global
acteristics can be noted on the score sheet. Items cover Impression (CGI) scale for overall severity of illness, un-
language, thought processes, two motoric presentations, der instructions to compare the subjects with other pa-
and five components of cognition. Neither the DRS nor tients having that same disorder. No particular training
its revision is intended to assess stupor or coma. was done for completion of the CGI. CGI ratings were
The intent of this study was to 1) validate the Delir- used to compare illness severity between groups and to
ium Rating Scale- Revised-98 (DRS-R-98), 2) establish its correlate with delirium scale ratings within the delirium
reliability, as well as its sensitivity and specificity for group.
distinguishing delirious from nondelirious psychiatric Interrater reliability for the DRS-R-98 was established
groups, and 3) assess its ability to function as a severity by using a subset of subjects representing various di-
measure of delirium. It was compared with the DRS, the agnostic groups. Research psychiatrists (P.T., D.M., R.T.)
Cognitive Test for Delirium (CTD), and a global clinical independently rated the same subjects following a sin-
impression scale. gle interview when they were blind to diagnosis, and
comparisons were made between pairs of raters.
Construct validity was established mostly by com-
METHODS paring the DRS-R-98 with the DRS and to a lesser extent
by comparing it with the CTD. Sensitivity and specific-
Subjects ity of the DRS-R-98 were determined by comparing
Adult subjects were recruited from medical, surgical, scores from delirious subjects with other diagnostic
psychiatric, rehabilitation, and nursing home care in- groups’ scores. This comparison also assisted with eval-
patient units of the University of Mississippi Medical uating criterion validity.
Center affiliated hospitals over a 5-month period in
1999. We recruited delirious, demented, schizophrenic, Scale Descriptions
depressed, and other psychiatric patients to form the The DRS is a 10-item clinician-rated scale with a maxi-
five different comparison groups. Verbal assent to be mum possible score of 32 points. Items represent symp-
interviewed was used to determine participation in this toms that are rated on a scale of 0 to 2, 3, or 4 points,
study as approved by the University of Mississippi with text descriptions for each point. The items are tem-
Medical Center institutional review board. There were poral onset of symptoms, perceptual disturbances, hal-
no exclusion criteria except an unwillingness to be psy- lucination type, delusions, psychomotor behavior, cog-
nitive status during formal testing, physical disorder, total scale scores, using Cronbach’s alpha coefficient to
sleep-wake cycle disturbance, lability of mood, and vari- assess internal consistency of the scale as a measure of
ability of symptoms. The DRS has good scale character- delirium. To assess empirical validity of the DRS-R-98
istics based on a number of studies of delirious popu- as a delirium scale, total and severity scores were com-
lations. Its interrater reliability (intraclass correlation pared among the five diagnostic groups by one-way
coefficients) ranges from 0.86 to 0.97 for psychiatric or analysis of variance (ANOVA), with post hoc pairwise
geriatric physicians and 0.59 to 0.99 for nonphysicians; comparisons to determine where the differences lie.
specificity ranges from 0.82 to 0.94, sensitivity from 0.82 Boxplots were graphed to show medians and distri-
to 0.94.5 butions of rating scale and CTD scores for each diag-
The DRS-R-98 is a 16-item scale (see Appendix A) nostic group. The Kruskal-Wallis test was performed to
with a maximum total scale score of 46 points (includes assess for between-group difference.
the three diagnostic items) and a maximum severity DRS-R-98 scores were compared with “after usual
score of 39 points. Whenever an item of the DRS-R-98 treatment” scores by use of paired t-tests in a subset of
could not be rated—which was usually dependent on delirious subjects to assess the DRS-R-98 as a severity
the degree of cooperation—it was so noted and later scale over time. Total scores were correlated with CGI
scored midway, that is, as 1.5 points; this occurred scores as another way to assess DRS-R-98 as a severity
rarely. We used three words to assess short-term mem- measure.
ory, months of the year backwards to help rate attention, Cutoff scores for the DRS-R-98 were determined by
copying intersecting pentagons and drawing a clockface using receiver-operator characteristic (ROC) analyses to
to help assess visuoconstructional ability, and parts of a determine acceptable levels of sensitivity and specificity
pen and/or watch to assess naming. when comparing the delirium group either with the de-
Each subject was administered the Cognitive Test for mented group or with all nondelirious subjects.
Delirium6 by a research assistant as a brief, broad mea- CGI scores were compared between groups by using
sure of cognitive function. The CTD was designed spe- one-way ANOVA to assess whether illness severity was
cifically for delirious patients, especially those who can- similar among diagnostic groups.
not speak. It tests orientation, attention, visual memory, Interrater reliabilities for the DRS-R-98 and DRS total
and conceptual reasoning and has been shown to cor- scores were measured by using an intraclass coefficient
relate highly with the Mini-Mental State Examination for pairs of independent raters.
(MMSE) in delirious patients. A suggested cutoff score
for delirium is ⱕ19 points.
The CGI is scored as a single overall impression of RESULTS
illness severity on a Likert-type scale ranging from 1 to
7 points.7 It was used to grossly measure severity of ill- Subjects
ness within diagnostic groups. Unlike the other scales, A total of 68 subjects were evaluated from five diagnos-
the CGI was completed without prior training by a va- tic groups: 24 delirious, 13 demented, 9 schizophrenic,
riety of treating physicians over a broad range of clinical 12 depressed, and 10 “other.” Twenty-seven were re-
sites. cruited from medical-surgical units, 17 from a medical-
psychiatric unit, 15 from general psychiatric units, 4
Data Analysis from a nursing home unit, and 5 from a rehabilitation
Data were analyzed by using SPSS-PC software. Age unit.
and rating scale data were expressed as means and stan- Table 1 describes demographic characteristics of sub-
dard deviations. Statistical significance was set at jects for each diagnostic group. There were no signifi-
Pⱕ0.05. DRS and DRS-R-98 scores from the primary cant differences among groups for race, where 56% were
rater were used for all analyses (except in calculating white, 42% black, and 3% other (1 Hispanic and 1 Choc-
interrater reliability when pairs of raters were used). taw Native American). There were 51 males and 17 fe-
Age was correlated with total rating scale scores for each males (because of the inclusion of a Veterans Affairs
diagnostic group by Pearson correlation. Chi-square Medical Center population) but no difference in gender
was used to compare race and sex among groups. or race ratios among groups. Mean age was significantly
DRS-R-98 total scores were compared with both the different between groups (F⳱7.43, df⳱4,63, P⬍0.001),
DRS and the CTD in delirious subjects, using a Pearson as might be expected from the different age distributions
correlation to assess construct validity and ability to as- of the illnesses studied. Schizophrenic subjects were sig-
sess severity over a range of impairment levels. Scores nificantly younger than delirious (P⬍0.01) or demented
for each DRS-R-98 item were correlated with DRS-R-98 subjects (P⬍0.001), and “other” subjects were signifi-
cantly younger than demented subjects (P⬍0.01). All compared with each of the other groups (P⬍0.001). The
groups had a broad range in age except the demented DRS-R-98 total score did not distinguish schizophrenia,
subjects, who were all over 60 years old. depressed, or “other” groups from one another, nor did
it distinguish dementia from the “other” group. How-
Validity ever, it did distinguish dementia from schizophrenia
Table 2 shows mean scores and standard deviations for and depressed groups (P⬍0.05).
each group for DRS, DRS-R-98 total, DRS-R-98 severity, Pairwise comparisons showed that the mean DRS-R-
CTD, and CGI. There was a highly significant difference 98 severity score was significantly higher in the delirium
among diagnostic groups for the CTD (F⳱19.3, group compared with each of the other groups
df⳱4,62, P⬍0.001), DRS-R-98 total scale (F⳱47.9, (P⬍0.001). It distinguished dementia from both depres-
df⳱4,63, P⬍0.001), DRS-R-98 severity scale (F⳱35.0, sion and “other” groups (P⬍0.05), but not from schizo-
df⳱4,63, P⬍0.001) and DRS (F⳱44.2, df⳱4,62, phrenia. Like the DRS-R-98 total scale, it did not differ
P⬍0.001). The CGI also distinguished the groups among “other,” schizophrenia, and depression groups.
(F⳱3.5, df⳱4,63, P⳱0.01). Delirium subjects had the Pairwise comparisons showed that mean CTD scores
highest mean scores on the delirium rating scales and were significantly lower in the delirium group than in
the lowest scores on the CTD compared with any di- any other group at P⬍0.001, except for dementia at
agnostic group, indicating they had more delirium P⬍0.05. The CTD did not distinguish the schizophrenic
symptoms and cognitive impairment. group from any group except delirium, but it did dis-
ANOVA pairwise comparisons showed that the mean tinguish dementia from the depressed and “other”
DRS score was significantly higher in the delirium groups (P⬍0.01).
group compared with each of the other groups Pairwise comparisons showed that mean CGI scores
(P⬍0.001). The DRS did not differ among dementia, were not significantly different among any groups ex-
schizophrenia, depression, or “other” groups. cept between delirium and “other,” where the latter was
With pairwise comparisons, the mean DRS-R-98 total less impaired than the delirium group (P⳱0.03). The
score was significantly higher in the delirium group mean CGI score was in the “moderately to markedly
Note: Groups are defined in the notes to Table 1. The total subjects compared for each scale was 68 except DRS and CTD, where it
was 67 because an ‘‘other’’ case had a missing CTD score and a ‘‘depression’’ case had a missing DRS score. DRS⳱Delirium Rating Scale;
DRS-R-98⳱Delirium Rating Scale-Revised-98; CTD⳱Cognitive Test for Delirium; CGI⳱Clinical Global Impression.
a
Includes 2 subjects whose delirium was resolving.
*One-way analysis of variance, P⬍0.001 among groups.
**One-way analysis of variance, P⳱0.012 among groups.
impaired” range for all groups except for “other” which Age did not correlate with any measure. The DRS cor-
was halfway between “mildly” and “moderately” im- related strongly with the DRS-R-98 total (r⳱0.83,
paired. This indicates that the major diagnostic groups P⬍0.001) and DRS-R-98 severity (r⳱0.80, P⬍0.001) sug-
were well matched for breadth of overall illness severity gesting the newer scale is a good measure of delirium.
levels based on a clinical global impression. As would be expected, the DRS-R-98 total score corre-
In addition to comparing group means, we graphed lates very strongly with the DRS-R-98 severity score
boxplots (Figure 1) to show the distribution of scale (r⳱0.99, P⬍0.001). Both the DRS and the DRS-R-98 cor-
scores in quartiles (middle 50% in the box) and median related somewhat less strongly with the CTD (r⳱ⳮ0.41,
scores (in solid black lines). On Kruskal-Wallis compar- P⬍0.05 for the DRS; r⳱ⳮ0.62, P⳱0.001 for the DRS-R-
isons, scores were significantly different among groups 98 total; and r⳱ⳮ0.63, P⳱0.001 for DRS-R-98 severity),
(P⬍0.001) for the DRS, DRS-R-98 total and severity, and consistent with the delirium scales measuring symp-
CTD scales. Median scores and middle quartiles for de- toms more broadly than just cognition. Correlations
lirious subjects did not overlap with those from any with the CGI were strong but less so than between the
other group on any scale, except for the dementia group two delirium rating scales with each other (r⳱0.45,
on the CTD, suggesting that the CTD is less discrimi- P⬍0.05 for DRS; r⳱0.62, P⳱0.001 for DRS-R-98 total;
nating. The DRS distribution shows very little overlap and r⳱0.61, P⳱0.001 for DRS-R-98 severity), likely re-
between delirium and any other group except between flecting the CGI’s more nonspecific, global nature.
its lower quartile and the upper quartile of the dementia
and “other” groups. One severely impaired schizo-
DRS-R-98 Pre and Post Treatment
phrenic outlier and one severely psychotically de-
Six of the delirious subjects were reassessed after treat-
pressed outlier just reached the lowest distribution of
ment when they no longer met DSM-IV criteria for de-
the DRS range, as well as one “other” patient who had
lirium. Their mean age was 55.3Ⳳ23.1 years (range 18–
a recently resolved delirium.
82). Their mean scores for CTD, DRS, DRS-R-98 severity,
The most overlap on boxplots for DRS-R-98 total or
and CGI are listed in Table 3. There were significant
severity scores occurs between the lower quartile of the
improvements on all measures after treatment. In par-
delirium group and the dementia group, with somewhat
ticular, the DRS-R-98 severity scale improved from a
more overlap for the severity distribution, which includes
mean of 21.5Ⳳ5.6 points to 5.2Ⳳ3.5 (t⳱7.13, df⳱5,
a resolving delirium group outlier. CTD scores are lowest
P⬍0.001), indicating an ability to measure change in
in the delirium group, but the boxes for delirium and
clinical status in parallel to global clinical, cognitive, and
dementia groups overlap, and 2 schizophrenic outliers
diagnostic assessments. The DRS also declined from a
(rated as severely and extremely severely impaired on the
mean score of 18.3Ⳳ3.9, clearly in the delirious range,
CGI) overlap with the delirium box.
to 3.5Ⳳ2.1 (t⳱10.6, df⳱5, P⬍0.001), clearly out of the
Two subjects in the delirium group had been delirious
delirious range. CGI improved significantly from
for a while and were improving, and one subject in the
“moderate/marked impairment” to “much/very much
“other” group had a brief nocturnal delirium that was
improved” (t⳱6.3, df⳱5, P⳱0.001). Although the delir-
mostly resolved by morning when he was rated. Be-
ium rating scale raters were aware of the delirium diag-
cause of their varying degrees of resolving status, these
nosis at the second rating, the CGI was rated indepen-
3 subjects were analyzed separately against the other 22
dently by the primary treating physician and the CTD
subjects with delirium to see if their scores reflected their
was administered and scored by a research assistant.
milder status. Compared with subjects having full-
blown delirium, those with resolving delirium had sig-
nificantly higher mean scores on the CTD (23.7Ⳳ2.5 vs. Scale Reliability
10.9Ⳳ5.7; t⳱ⳮ3.7, df⳱23, P⳱0.001), and significantly The Cronbach’s alpha coefficient in the delirium group
lower on the DRS (10.3Ⳳ2.3 vs. 19.3Ⳳ4.8; t⳱3.1, df⳱23, was 0.90 for the DRS-R-98 total scale and 0.87 for the
P⳱0.005), on the DRS-R-98 total (13.0Ⳳ2.0 vs. 28.2Ⳳ5.4; DRS-R-98 severity scale, supporting the reliability of the
t⳱4.8, df⳱23, P⬍0.001) and on the DRS-R-98 severity scale and its internal consistency. In addition, when the
(7.3Ⳳ1.5 vs. 22.6Ⳳ4.9; t⳱5.3, df⳱23, P⬍0.001). There effect of each individual item was deleted from the scale,
was no difference in age between those with resolving coefficients ranged from 0.88 to 0.90 for the DRS-R-98
and those with more full-blown delirium. total and from 0.84 to 0.87 for the DRS-R-98 severity
scale, suggesting a high degree of internal consistency
DRS-R-98 Characteristics in Delirious Subjects among items of the scale. Table 4 lists alpha coefficients
Correlations were performed between rating scales in for both the DRS-R-98 total and DRS-R-98 severity scales
the delirium group to address validity of the DRS-R-98. as each item is removed from the scale. The DRS also
showed a high alpha coefficient (0.87), ranging from 0.83 Interrater Reliability
to 0.87 when each item was removed, reflective of its Three trained psychiatrist raters were used to calculate
high internal consistency. an intraclass correlation coefficient (ICC) for the DRS
FIGURE 1. Boxplots of DRS, DRS-R-98 Total, DRS-R-98 Severity, and CTD scores for each of the five diagnostic groups. Median scores
are denoted by the solid line within the boxes. The boxes represent the middle 50% of the scores. Outliers are denoted by
open circles. DRSⴔDelirium Rating Scale; CTDⴔCognitive Test for Delirium.
50 40
40 30
30 20
20 10
10 0
0 –10
N= 24 13 9 12 10 N= 24 13 9 12 10
Delirium Schizophrenia Other Delirium Schizophrenia Other
Dementia Depression Dementia Depression
Diagnosis Diagnosis
30 40
30
20
Total CTD Score
Total DRS Score
20
10
10
0 –10
N= 24 13 9 11 10 N= 24 13 9 12 9
Delirium Schizophrenia Other Delirium Schizophrenia Other
Dementia Depression Dementia Depression
Diagnosis
Diagnosis
and DRS-R-98. When the primary rater was compared were chosen as the two best options, resulting in the
with the secondary rater for the DRS (n⳱25), ICC⳱0.99; same sensitivity (92%), but the higher cutoff had a
for the DRS-R-98 total (n⳱26), ICC⳱0.98; and for the higher specificity (95%). The best cutoff score for the
DRS-R-98 severity (n⳱26), ICC⳱0.99. When each com- DRS-R-98 severity scale was 15.25, resulting in 92% sen-
bination of pairs of raters was compared for each rating sitivity and 93% specificity.
scale, the ICCs ranged from 0.98 to 0.99 (10 or 11 cases The delirium group was then compared only with the
were used for each rater pair comparison). dementia group, and the best cutoff scores (17.75 for
total and 15.25 for severity) resulted in 92% sensitivity
Sensitivity and Specificity for both of the DRS-R-98 scales but a higher specificity
Results of receiver operating curve (ROC) analyses are for the total scale (85%) than for the severity scale (77%).
shown in Table 5. On the basis of clinical experience, diagnostic items are
ROC analyses were performed for DRS-R-98 scores expected to have less overlap between delirium and de-
comparing the delirium group versus all other diagno- mentia than other symptoms.
ses, as well as the delirium group versus only dementia A separate ROC analysis was performed that ex-
(dementia being the more likely diagnostic confound). cluded the two resolving delirium subjects because they
When delirium was compared with all other groups for had significantly lower DRS-R-98 scores (see above)
the DRS-R-98 total scale, cutoff scores of 15.25 and 17.75 than other delirium subjects and would be expected to
affect discrimination between groups. At the cutoff of
17.75, the DRS-R-98 total scale sensitivity increased to
TABLE 3. Pre/post treatment scores (meanⴣSD) in a delirium 100%, and higher cutoffs (21.5 and 22.5) each resulted
subgroup in a somewhat lower sensitivity (91%) but increased
Scale Pre Post specificities (92% and 100%, respectively). On the DRS-
DRS 18.3Ⳳ3.9 3.5Ⳳ2.1* R-98 severity scale with a cutoff score of 15.25, excluding
DRS-R-98 severity 21.5Ⳳ5.6 5.2Ⳳ3.5* these two subjects resulted in an increased sensitivity of
CTD 8.0Ⳳ5.6 24.3Ⳳ4.3** 100% with the same specificity (77%), whereas raising
CGI 4.5Ⳳ1.1 1.8Ⳳ0.75*
the cutoff score to 17 reduced the sensitivity and in-
Note: DRS⳱Delirium Rating Scale; DRS-R-98⳱Delirium Rating creased the specificity.
Scale-Revised-98; CTD⳱Cognitive Test for Delirium; CGI⳱Clinical
Global Impression.
* Paired t-tests, Pⱕ0.001.
**Paired t-test, Pⱕ0.01. DISCUSSION
changes constitute the characteristic waxing and waning Subjects with resolving delirium were included in the
of delirium during a 24-hour period. delirium group for all analyses, but these subjects had
The DRS-R-98 appears to be a valid measure of delir- less impairment on scale measurements, and their data
ium. We analyzed its characteristics for both the total affected the cutoff when delirium was compared with
scale and the severity scale. The DRS-R-98 correlated dementia. In clinical practice when such cases will be
highly with scores on the DRS and the CTD in delirious assessed, the differential from dementia will depend in
subjects. Its mean and median scores were significantly part on taking a careful history. Focusing on just the
higher in delirium subjects than in any of its comparison DRS-R-98 diagnostic items may assist in this process. A
groups: dementia, depression, schizophrenia, or different study using the DRS to compare delirious and
“other.” These patient groups were well matched for delirious-demented elderly subjects found that delirium
illness severity; such matching is especially important symptoms largely overshadowed dementia symptoms,
when comparing dementia and delirium groups. There although there were still some differences.4 Longitudi-
was little overlap in the boxplot distribution of DRS-R- nal testing of the DRS-R-98 in delirious-demented pa-
98 scores in the delirium group compared with other tients whose delirium resolves will be needed to deter-
diagnostic groups, although the DRS-R-98 total scale mine which items best distinguish these groups.
had less overlap with other diagnostic groups than did In our study the CTD was less robust in distinguish-
the severity scale, as would be expected. ing dementia from delirium, and many in our dementia
Cronbach’s alpha coefficient was high (0.90), indicat- group scored lower than the cutoff score of 19 points
ing high internal consistency among its 16 items; this recommended in Hart and colleagues’ original report.6
level of consistency was largely maintained for the se- Our dementia group had a CTD mean score of about 18
verity scale alone (0.87). Each item also individually and showed more overlap with the delirium group on
contributed strongly to the scale. Interrater reliability boxplots than did the DRS or DRS-R-98. This is likely
was excellent among pairs of three psychiatrist raters because the CTD measures only one dimension of de-
doing independent ratings during a single clinical in- lirium phenomenology, unlike these broader symptom
terview. scales.
The DRS-R-98 was compared with the CTD instead of Two delirium severity rating instruments have been
the MMSE because the CTD was designed for delirium published in recent years. The Confusional State Eval-
and can be administered to nonverbal or intubated pa- uation (CSE)12 is a three-part, 22-item scale from Sweden
tients. Also, the CTD has the advantage of assessing that was tested only in delirium patients (N ranged from
some executive and more nonverbal cognitive functions 28 to 51 patients). Thus, no comparison was made with
of the nondominant hemisphere, which complements dementia patients, nor were sensitivity and specificity
the verbal modalities we used during administration of values obtainable. In addition, patients with comorbid
the DRS-R-98. Despite this difference, the two scales cor- dementia were included in the delirium group, which
related highly. In addition, two recent studies8,9 that re- significantly confounds phenomenologic validity of the
lied on cognitive tests to assess delirium found high dis- scale as a delirium assessment tool. Dementia patients
criminating ability of the items measuring functions of have different symptom severity even for some overlap-
the nondominant hemisphere, in keeping with the neu- ping symptoms.13,14 O’Keefe’s Delirium Assessment
roanatomical hypothesis that the right-sided neural Scale (DAS)15 used operationalized DSM-III criteria to
pathways, often under-studied, are integral to delirium compare delirious, delirious-demented, demented, and
pathophysiology.10,11 not cognitively impaired groups (N⳱48). Sensitivity
ranged from 83% to 88% and specificity from 79% to peditious, it is unlikely that cognitive tests alone can
88% for delirium diagnosis, and interrater reliability adequately capture the breadth of symptoms needed to
ranged from 0.66 to 0.99. Onset of symptoms was not assess delirium severity and distinguish delirium from
specifically rated, but was arbitrarily scored as acute. dementia.
Three recent delirium studies have used only cogni- In summary, the DRS-R-98 is a valid and reliable
tive tests to assess delirium.9,16,17 Hart et al.9 used a step- symptom severity scale for delirium that has advantages
wise discriminant analysis of CTD item scores to deter- over the original DRS for flexibility and breadth of
mine that an abbreviated version using 2 of 9 items— symptom coverage. It is the only validated delirium rat-
visual attention span and recognition memory for pic- ing instrument with sufficient breadth and detail for use
tures—discriminated 19 delirium patients in the inten- in phenomenology and in longitudinal studies of delir-
sive care unit from other diagnostic groups with high ium patients, including serial measurements in treat-
reliability (alpha⳱0.79). Bettin et al.17 found correlations ment research. Moreover, unlike most other delirium in-
of about 0.50 between cognitive tests (forward digit span struments, it was validated against a dementia group
and similarities) and expert serial ratings using DSM- and other psychiatric diagnostic groups. It is currently
III-R in 22 delirious and 15 control elderly subjects. being translated into other languages. Further research
These investigators felt that backward digit span had a using the DRS-R-98 is needed to extend and replicate its
floor effect but recommended forward digit span for utility, including longitudinal comparisons of condi-
monitoring symptom severity in delirium. However, tions that can occur comorbidly with delirium.
O’Keefe and Grosney16 found that forward digit span
did not distinguish delirium from dementia patients, The authors thank Robert W. Baker, M.D., Andrea F.
suggesting that this test would be inadequate in clinical DiMartini, M.D., James Levenson, M.D., Roos van der Mast,
situations. Further, they found that a global rating of M.D., Ph.D., and Angelos Halaris, M.D., Ph.D. for their con-
attentiveness, digit span backwards, and a cancellation tributions to scale development; Joel Greenhouse, Ph.D., for
test distinguished delirium from dementia, but that the statistical consultation; and Henry Nasrallah, M.D., and
MMSE and a vigilance test did not. However, they ex- Gurdial Sandhu, M.D., for their support during this study.
cluded anyone with an MMSE score ⱕ10, which may This work was supported in part by the Mental Illness Re-
have biased the results away from more severe delirium search Education and Clinical Center, Veterans Integrated
cases. Also, 4 of 18 delirium subjects had comorbid de- Service Network 16 (MIRECC-VISN 16), Department of
mentia. Although the use of such tests is clinically ex- Veterans Affairs.
References
1. Inouye SK, van Dyke CH, Allessi CA, et al: Clarifying confu- 9. Hart RP, Best AM, Sessler CN, et al: Abbreviated Cognitive Test
sion: the Confusion Assessment Method. Ann Intern Med 1990; for Delirium. J Psychosom Res 1997; 43:417–423
113:941–948 10. Trzepacz PT: Neuropathogenesis of delirium: a need to focus
2. Albert MS, Levkoff SE, Reilly C, et al: The Delirium Symptom our research. Psychosomatics 1994; 35:375–391
Interview: an interview for the detection of delirium symptoms 11. Trzepacz PT: Is there a final common neural pathway in delir-
in hospitalized patients. J Geriatr Psychiatry Neurol 1992; 5:14– ium? Focus on acetylcholine and dopamine. Seminars in Clini-
21 cal Neuropsychiatry 2000; 5:132–148
3. Trzepacz PT, Baker RW, Greenhouse J: A symptom rating scale 12. Robertsson B, Karlsson I, Styrud E, et al: Confusional State
for delirium. Psychiatry Res 1988; 23:89–97 Evaluation (CSE): an instrument for measuring severity of de-
4. Trzepacz PT, Mulsant BH, Dew MA, et al: Is delirium different lirium in the elderly. Br J Psychiatry 1997; 170:565–570
when it occurs in dementia? A study using the Delirium Rating 13. Liptzin B, Levkoff SE, Gottlieb GL, et al: Delirium. J Neuropsy-
Scale. J Neuropsychiatry Clin Neurosci 1998; 10:199–204 chiatry Clin Neurosci 1993; 5:154–160
5. Trzepacz PT: The Delirium Rating Scale: its use in consultation- 14. Trzepacz PT, Dew MA: Further analyses of the Delirium Rating
liaison research. Psychosomatics 1999; 40:193–204 Scale. Gen Hosp Psychiatry 1995; 17:75–79
6. Hart RP, Levenson JL, Sessler CN, et al: Validation of a cogni- 15. O’Keefe ST: Rating the severity of delirium: the Delirium As-
tive test for delirium in medical ICU patients. Psychosomatics sessment Scale. Int J Geriatr Psychiatry 1994; 9:551–556
1996; 37:533–546 16. O’Keefe ST, Gosney MA: Assessing attentiveness in older hos-
7. Guy W (editor): ECDEU Assessment Manual for Psychophar- pitalized patients: global assessment versus tests of attention. J
macology. (Publ No ADM 76–338). Rockville, MD, US Depart- Am Geriatr Soc 1997; 45:470–473
ment of Health, Education and Welfare, 1976, pp 217–222 17. Bettin KM, Maletta GJ, Dysken MW, et al: Measuring delirium
8. Mach JR Jr, Kabat V, Olson D, et al: Delirium and right hemi- severity in older general hospital inpatients without dementia:
sphere dysfunction in cognitively impaired older persons. Int the Delirium Severity Scale. Am J Geriatr Psychiatry 1998;
Psychogeriatr 1996; 8:378–381 6:296–307
3. Delusions
Delusions can be of any type, but are most often persecutory. Rate if reported by patient, family or caregiver. Rate as delusional if ideas
are unlikely to be true yet are believed by the patient who cannot be dissuaded by logic. Delusional ideas cannot be explained otherwise
by the patient’s usual cultural or religious background.
0. Not present
1. Mildly suspicious, hypervigilant, or preoccupied
2. Unusual or overvalued ideation that does not reach delusional proportions or could be plausible
3. Delusional
4. Lability of affect
Rate the patient’s affect as the outward presentation of emotions and not as a description of what the patient feels.
0. Not present
1. Affect somewhat altered or incongruent to situation; changes over the course of hours; emotions are mostly under self-control
2. Affect is often inappropriate to the situation and intermittently changes over the course of minutes; emotions are not consistently
under self-control, though they respond to redirection by others
3. Severe and consistent disinhibition of emotions; affect changes rapidly, is inappropriate to context, and does not respond to redi-
rection by others
5. Language
Rate abnormalities of spoken, written or sign language that cannot be otherwise attributed to dialect or stuttering. Assess fluency, grammar,
comprehension, semantic content and naming. Test comprehension and naming nonverbally if necessary by having patient follow com-
mands or point.
0. Normal language
1. Mild impairment including word-finding difficulty or problems with naming or fluency
2. Moderate impairment including comprehension difficulties or deficits in meaningful communication (semantic content)
3. Severe impairment including nonsensical semantic content, word salad, muteness, or severely reduced comprehension
7. Motor agitation
Rate by observation, including from other sources of observation such as by visitors, family and clinical staff. Do not include dyskinesia,
tics, or chorea.
0. No restlessness or agitation
1. Mild restlessness of gross motor movements or mild fidgetiness
2. Moderate motor agitation including dramatic movements of the extremities, pacing, fidgeting, removing intravenous lines, etc.
3. Severe motor agitation, such as combativeness or a need for restraints or seclusion
8. Motor retardation
Rate movements by direct observation or from other sources of observation such as family, visitors, or clinical staff. Do not rate components
of retardation that are caused by parkinsonian symptoms. Do not rate drowsiness or sleep.
0. No slowness of voluntary movements
1. Mildly reduced frequency, spontaneity or speed of motor movements, to the degree that may interfere somewhat with the
assessment.
2. Moderately reduced frequency, spontaneity or speed of motor movements to the degree that it interferes with participation in
activities or self-care
3. Severe motor retardation with few spontaneous movements.
9. Orientation
Patients who cannot speak can be given a visual or auditory presentation of multiple choice answers. Allow patient to be wrong by up
to 7 days instead of 2 days for patients hospitalized more than 3 weeks. Disorientation to person means not recognizing familiar persons
and may be intact even if the person has naming difficulty but recognizes the person. Disorientation to person is most severe when one
doesn’t know one’s own identity and is rare. Disorientation to person usually occurs after disorientation to time and/or place.
0. Oriented to person, place and time
1. Disoriented to time (e.g., by more than 2 days or wrong month or wrong year) or to place (e.g., name of building, city, state), but
not both
2. Disoriented to time and place
3. Disoriented to person
10. Attention
Patients with sensory deficits or who are intubated or whose hand movements are constrained should be tested using an alternate modality
besides writing. Attention can be assessed during the interview (e.g., verbal perseverations, distractibility, and difficulty with set shifting)
and/or through use of specific tests, e.g., digit span.
0. Alert and attentive
1. Mildly distractible or mild difficulty sustaining attention, but able to refocus with cueing. On formal testing makes only minor
errors and is not significantly slow in responses
2. Moderate inattention with difficulty focusing and sustaining attention. On formal testing, makes numerous errors and either requires
prodding to focus or finish the task
3. Severe difficulty focusing and/or sustaining attention, with many incorrect or incomplete responses or inability to follow instruc-
tions. Distractible by other noises or events in the environment
䉷 Trzepacz 1998
These three items can be used to assist in the differentiation of delirium from other disorders for diagnostic and research purposes. They
are added to the severity score for the total scale score, but are NOT included in the severity score.
䉷 Trzepacz 1998
DRS-R-98 SCORESHEET
Name of Rater:
Orientation 0 1 2 3 Date:
Place:
Person:
Attention 0 1 2 3
Short-term memory 0 1 2 3 Record # of trials for registration of items:
Check here if category cueing helped
Long-term memory 0 1 2 3 Check here if category cueing helped
Visuospatial ability 0 1 2 3 Check here if unable to use hands
Diagnostic Item Item Score Optional Information