A comparison of nine scales to detect

depression in Parkinson disease

Which scale to use?

J.R. Williams, PhD ABSTRACT

E.S. Hirsch, MS Objective: The Methods of Optimal Depression Detection in Parkinson’s Disease (MOOD-PD)
K. Anderson, MD study compared the psychometric properties of 9 depression scales to provide guidance on scale
A.L. Bush, PhD selection in Parkinson disease (PD).
S.R. Goldstein, MD
Methods: Patients with PD (n ⫽ 229) from community-based neurology practices completed 6
S. Grill, MD
self-report scales (Beck Depression Inventory [BDI]–II, Center for Epidemiologic Studies Depres-
S. Lehmann, MD
sion Rating Scale–Revised [CESD-R], 30-item Geriatric Depression Scale [GDS-30], Inventory of
J.T. Little, MD
Depressive Symptoms–Patient [IDS-SR], Patient Health Questionnaire-9 [PHQ-9], and Unified
R.L. Margolis, MD
Parkinson’s Disease Rating Scale [UPDRS]–Part I) and were administered 3 clinician-rated scales
J. Palanci, BS
(17-item Hamilton Depression Rating Scale [HAM-D-17], Inventory of Depressive Symptoms–
G. Pontone, MD
Clinician [IDS-C], and Montgomery-Åsberg Depression Rating Scale [MADRS] and a psychiatric
H. Weiss, MD
interview. DSM-IV-TR diagnoses were established by an expert panel blinded to the self-reported
P. Rabins, MD
rating scale data. Receiver operating characteristic curves were used to estimate the area under
L. Marsh, MD
the curve (AUC) of each scale.
Results: All scales performed better than chance (AUC 0.75–0.85). Sensitivity ranged from 0.66
Correspondence & reprint to 0.85 and specificity ranged from 0.60 to 0.88. The UPDRS Depression item had a smaller
requests to Dr. Marsh: AUC than the BDI-II, HAM-D-17, IDS-C, and MADRS. The CESD-R also had a smaller AUC than
[email protected]
the MADRS. The remaining AUCs were statistically similar.
Conclusions: The GDS-30 may be the most efficient depression screening scale to use in PD
because of its brevity, favorable psychometric properties, and lack of copyright protection. How-
ever, all scales studied, except for the UPDRS Depression, are valid screening tools when
PD-specific cutoff scores are used. Neurology® 2012;78:998–1006

GLOSSARY
AUC ⫽ area under the curve; BDI ⫽ Beck Depression Inventory; CESD-R ⫽ Center for Epidemiologic Studies Depression Rating
Scale–Revised; GDS ⫽ Geriatric Depression Scale; H&Y ⫽ Hoehn and Yahr; HAM-D-17 ⫽ 17-item Hamilton Depression Rating
Scale; IDS-C ⫽ Inventory of Depressive Symptoms–Clinician; IDS-SR ⫽ Inventory of Depressive Symptoms–Patient; MADRS ⫽
Montgomery-Åsberg Depression Rating Scale; MMSE ⫽ Mini-Mental State Examination; MOOD-PD ⫽ Methods of Optimal De-
pression Detection in Parkinson’s Disease; NPV ⫽ negative predictive value; PD ⫽ Parkinson disease; PHQ-9 ⫽ Patient Health
Questionnaire-9; PPV ⫽ positive predictive value; ROC ⫽ receiver operating characteristic; SCID ⫽ Structured Clinical Interview
for DSM Disorders; UPDRS ⫽ Unified Parkinson’s Disease Rating Scale.

Depressive syndromes affect an estimated 40% of patients with Parkinson disease (PD).1 How-
ever, depressive syndromes in PD are often unrecognized or inadequately treated.2 Routine use
of depressive symptom rating scales may improve detection of depression in PD. In 2009, the
US Preventive Services Task Force recommended use of depression scales in primary care but
did not recommend a specific scale.3 Similarly, no one scale is recommended in PD.4
Depression screening tools should be both sensitive and specific to the broad differential
Supplemental data at
diagnosis of depressed mood in PD.5,6 Although not a substitute for a diagnostic evaluation, scales
www.neurology.org
From the Food and Drug Administration (J.R.W., S.R.G.), Silver Spring, MD; Departments of Psychiatry and Behavioral Sciences (J.R.W., E.S.H.,
S.L., J.T.L., R.L.M., J.P., G.P., P.R., L.M.) and Neurology (S.R.G., S.G., H.W., L.M.), Johns Hopkins University School of Medicine, Baltimore,
MD; Department of Psychiatry (K.A.), University of Maryland Medical Center, Baltimore; Mental Health Care Line (A.L.B., L.M.), Michael E.
Supplemental Data DeBakey Veterans Affairs Medical Center, Houston, TX; Menninger Department of Psychiatry and Behavioral Sciences and Department of
Neurology (A.L.B., L.M.), Baylor College of Medicine, Houston, TX; Parkinson’s and Movement Disorder Center of Maryland (S.R.G., S.G.),
Baltimore, MD; Department of Psychiatry (J.T.L.), Georgetown University Medical Center, Washington, DC; and Mental Health Service (J.T.L.),
Veterans Affairs Medical Center, Washington, DC.
Study funding: Funding information is provided at the end of the article.
Disclosure: Author disclosures are provided at the end of the article.

998 Copyright © 2012 by AAN Enterprises, Inc.

should distinguish normal emotional variability Screening visit. During an initial screening visit, subjects
completed 6 self-reported depression scales plus other demographic
from symptoms that reflect major depression
and clinical measures. All scales were completed before the diagnos-
or a disabling nonmajor depressive syndrome.5,7 tic visit and were presented in random order. The time to complete
Multiple clinician-rated depression scales (the each self-reported scale was recorded. The MMSE provided a mea-
24-item and 17-item Hamilton Depression Rat- sure of global cognitive status.28 The UPDRS Motor Subscale and
Hoehn and Yahr (H&Y) staging criteria assessed motor defi-
ing Scale [HAM-D], the Montgomery-Åsberg cits and status of disease progression, respectively, with the
Depression Rating Scale [MADRS], and subject in an optimally medicated (on) state.17,29
the Unified Parkinson’s Disease Rating Diagnostic visit. A geriatric psychiatrist conducted the diag-
Scale [UPDRS] Depression item),8 –18 and nostic examination using the Structured Clinical Interview for
self-reported scales (Beck Depression Inventory DSM Disorders (SCID) for DSM-IV-TR, supplemented with
questions to capture psychiatric disturbances not included in
[BDI]–Version I) and the 30-item and 15-item the SCID (e.g., cognitive dysfunction, apathy, and impulse
Geriatric Depression Scale [GDS]) have been control disorders).30 Three clinician-rated scales were rated in
shown to be valid in PD.14,15,18 –24 The GDS-15 the context of the psychiatric interview. If an informant was
available, a trained research coordinator conducted the Com-
and Patient Health Questionnaire-9 (PHQ-9)
posite International Diagnostic Interview–Short Form and
have also been investigated as diagnostic instru- the Informant Interview for the Diagnosis of Dementia and
ments in PD.25,26 Depression in Older Adults to obtain collateral psychiatric
The comparative performance of these scales and cognitive information.31,32 Informant information was
not used to rate clinician-rated scales.
is unknown. Previous studies evaluated 1 or 2
scales at a time in tertiary care samples and thus Rating scales. Subjects were administered 6 self-reported and
3 clinician-rated depression scales. The self-report scales (BDI-
provide limited guidance. Differences in diag- II, GDS-30, Center for Epidemiologic Studies Depression
nostic approaches and subject characteristics Rating Scale–Revised [CESD-R], Inventory of Depressive
across studies also prevent direct comparisons. Symptoms-Self-Report [IDS-SR], PHQ-9, and the UPDRS
Mentation, Behavior, and Mood Subscale) and the clinician-rated
In addition, the influence of patient characteris- scales (HAM-D-17, Inventory of Depressive Symptomatology–
tics on psychometric properties of depression Clinician-Rated [IDS-C], and MADRS) were selected because
scales in PD needs more study. of previous validation in PD or because their content mapped
onto DSM-IV-TR criteria for major depression.9 –18,20,22,26,33–36
This study concurrently evaluated 9 de-
This analysis modified the UPDRS Mentation, Behavior, and
pression scales relative to a comprehensive Mood Subscale by using the depression item as a self-report scale
clinical assessment for depressive disorders in (UPDRS Depression).17 An inclusive symptom attribution ap-
a community-based sample. Differences in proach was used to rate all clinician-rated depression scales.5 This
approach rates all symptoms as related to depression, regardless of
performance were tested across scales and symptom overlap with PD or other medical conditions.5
across patient subgroups. Accordingly, the
Psychiatric diagnoses. Final consensus psychiatric diagnoses,
present report provides additional guidance based on DSM-IV-TR criteria, were established using a modifica-
on depression scale selection in PD. tion of best-estimate diagnostic procedures.33,37 A panel of 6 psy-
chiatrists with expertise in geriatric psychiatry or movement
METHODS Participants. Subjects were recruited from 3 disorders reviewed each subject’s history and all data collected
community-based movement disorder practices. Patients were from diagnostic interviews, informants, and medical records.
mailed letters by their neurologist inviting participation in a The panel was completely blinded to information from self-
study evaluating the psychometric performance of depression report depression scales. An inclusive symptom attribution ap-
scales in PD (the Methods of Optimal Depression Detection in proach was used to diagnose depression.5 Subjects were classified
Parkinson’s Disease [MOOD-PD] study). Subjects were in- as actively depressed, if diagnosed with a current major depres-
cluded in the study if they met United Kingdom Brain Bank sive episode, major depressive episode in partial remission, minor
criteria for idiopathic PD, had a Mini-Mental State Examination depression, dysthymia, adjustment disorder with depressed
(MMSE) score ⱖ24, and spoke English fluently.27,28 mood, or depressive disorder not otherwise specified. Subjects
A total of 269 patients met the screening criteria and completed with a depressive disorder in full remission (asymptomatic) or
questionnaires, which identified by self-report or informant those emotionalism (pathologic crying) were classified as not actively
depressed. No instances of mood disorders due to a general
who 1) endorsed any degree of depression, apathy, anxiety, or irrita-
medical condition or substance-induced mood disorders were
bility, 2) had been prescribed psychiatric medications, or 3) reported
observed.
a history of or a current depressive disorder diagnosis. Initially, sub-
jects endorsing criterion 1, 2, or 3 and every fourth subject not Statistical methods. Subjects were omitted from the analysis
meeting criterion 1, 2, or 3 were asked to take part in diagnostic if any depression scale items were missing. Receiver operating
psychiatric interviews (figure 1). After only 10 of the first 143 sub- characteristic (ROC) curves measured the psychometric prop-
jects failed to qualify for diagnostic interviews using these criteria, all erties of the depression scales, with the consensus diagnosis
subsequent participants meeting the initial screening criteria were serving as the gold standard.6 Psychometric indices included
considered eligible for diagnostic interviews. the area under the curve (AUC), sensitivity, specificity, posi-

Neurology 78 March 27, 2012 999

 Figure 1 Flow of participants through the Methods of Optimal Depression Detection in Parkinson’s Disease
(MOOD-PD) study

BDI ⫽ Beck Depression Inventory; CESD-R ⫽ Center for Epidemiologic Studies Depression Rating Scale–Revised; GDS-
30 ⫽ 30-item Geriatric Depression Scale; IDS-SR ⫽ Inventory of Depressive Symptoms–Patient; MMSE ⫽ Mini-Mental
State Examination; PD ⫽ Parkinson disease; PHQ-9 ⫽ Patient Health Questionnaire-9.

tive predictive value (PPV), and negative predictive value and ␹2 tests as appropriate. ␹2 tests, as implemented by the ROC-
(NPV).6 Indices were evaluated at cutoffs defined as the max- COMP command, were used to evaluate between- and within-scale
imum sum of sensitivity and specificity for each scale. The AUC differences. Within-scale analyses stratified the ROC curves
maximum sum of specificity and sensitivity was chosen be- by gender, H&Y stage (⬍2.5 vs ⱖ2.5), MMSE score (30 –28 vs
cause these measures are not affected by prevalence rates for 27–26 vs 25–24), and tertiles of age, education (years), PD symp-
depression, unlike the PPV and NPV.6 Internal reliability was tom duration (years), and UPDRS Motor score. Repeated-measures
measured by the Cronbach ␣.6 analysis of variance compared self-reported scale completion times.
Analyses were conducted using STATA statistical software (ver- For post hoc pairwise AUC comparisons, ␣ ⫽ 0.001 was chosen a
sion 9.0). Between-group differences were evaluated using t tests priori. For all other analyses, ␣ ⫽ 0.05 was chosen a priori.

1000 Neurology 78 March 27, 2012

 Standard protocol approvals, registrations, and patient Motor scores. Depressed subjects scored higher on all
consents. The Johns Hopkins Institutional Review Board ap- depression scales. There were no statistically significant
proved the study. Subjects and their informants provided written
differences for any variable in table 1 between the 229
informed consent.
subjects included in the analyses (figure 1) and the 21
RESULTS Table 1 shows the prevalence of depres- subjects who were excluded because of missing depres-
sion in addition to the clinical characteristics and de- sion scale data (data not shown).
pression scale scores of depressed and nondepressed Mean scale scores by depressive subgroups are
subjects. On average, depressed subjects had fewer years shown in figure 2. Mean scores were higher for subjects
of education, lower MMSE scores, and higher UPDRS with major depression than for subjects with no active
depression. Mean scores were also higher for subjects
with minor depression compared with subjects with no
active depression on all scales except the CESD-R,
Table 1 Sample characteristics and scale scores
PHQ-9, and UPDRS Depression. Subjects with major
Overall depressive depression had higher scores than subjects with minor
disorder prevalence
depression on all scales except the IDS-SR.
Depression, symptomatic, % 93 (40.6, 95% CI: 34.2–47.3)
The mean time to complete the self-report scales
Major depression, % 78 (34.1, 95% CI: 27.9–40.6) was as follows: BDI-II, 296.16 seconds (SD ⫽
Nonmajor depression, % 15 (6.6, 95% CI: 2.7–9.0) 138.14; n ⫽ 216); CESD-R, 145.36 seconds (85.99;
Depression, full remission, % 12 (5.2, 95% CI: 2.7–9.0) n ⫽ 214); GDS-30, 189.45 seconds (89.27; n ⫽
No active Active 217); IDS-SR, 594.48 seconds (297.71; n ⫽ 215);
depressive
disorder (n ⴝ 136)
depressive
disorder (n ⴝ 93) p Value
PHQ-9, 111.52 seconds (80.55; n ⫽ 214); and
UPDRS–Part I, 66.98 seconds (51.46; n ⫽ 215).
Age, y, mean (SD) 66.1 (10.0) 66.0 (10.8) ⬍0.951
Completion times differed among scales in the
Male, n (%) 93 (68) 60 (65) ⬍0.542
groupwise comparison (F5,217 ⫽ 610.69, p ⬍ 0.001)
White race, n (%) 122 (89.7) 86 (92.5) ⬍0.476
and in all pairwise comparisons ( p ⬍ 0.05).
Education, y, mean (SD) 16.5 (3.1) 15.6 (2.6) ⬍0.025
The AUCs for the discrimination of actively from
PD symptom duration, y,
mean (SD)
8.4 (6.7) 8.7 (6.2) ⬍0.721
nonactively depressed subjects and Cronbach ␣ for each
scale are provided in table 2, in addition to the sensitiv-
H&Y stage, n ⬍0.137
ity, specificity, PPV, and NPV associated with the cut-
I 21 10
off point that maximized sensitivity and specificity. For
I1⁄2 7 0
comparison, results from other validity studies in PD
II 61 43
are provided. The ROC curves and the sensitivity, spec-
II1⁄2 27 18
ificity, PPV, and NPV for additional cutoff points are
III 13 18 provided in figure e-1 on the Neurology® Web site at
IV 5 3 www.neurology.org and table e-1, respectively.
V 2 1 The AUCs were not equivalent in comparisons of
Depression scale score all scales (␹2[8] ⫽ 25.03, p ⫽ 0.0015) or in compar-
UPDRS Motor, mean (SD) 15.9 (9.9) (n ⫽ 133) 21.6 (12.0) (n ⫽ 89) ⬍0.001 isons of all self-reported scales (␹2[5] ⫽ 14.09, p ⫽
MMSE, mean (SD) 28.7 (1.3) 27.9 (1.8) ⬍0.001
0.015). The AUCs for clinician-rated scales were
BDI-II, mean (SD) 6.5 (5.2) 14.7 (7.4) ⬍0.001
equivalent (␹2[2] ⫽ 1.9, p ⫽ 0.3853). Pairwise post
hoc comparisons are shown in table 3. At ␣ ⫽ 0.001,
CESD-R, mean (SD) 9.3 (10.1) 22.1 (15.1) ⬍0.001
the UPDRS Depression had a smaller AUC than the
GDS-30, mean (SD) 5.8 (5.2) 13.7 (6.8) ⬍0.001
BDI-II, HAM-D-17, IDS-C, and MADRS. The
IDS-SR, mean (SD) 13.3 (8.0) 24.8 (10.1) ⬍0.001
CESD-R also had a smaller AUC than the MADRS.
PHQ-9, mean (SD) 3.8 (3.8) 8.9 (5.2) ⬍0.001
The AUC for each depression scale was not signifi-
UPDRS Depression, 0.2 (0.5) 1.0 (0.9) ⬍0.001 cantly different when stratified by age, gender, educa-
mean (SD)
tion (years), PD symptom duration (years), H&Y stage,
HAM-D-17, mean (SD) 4.5 (3.2) 11.1 (5.2) ⬍0.001
UPDRS Motor score, or MMSE score (table e-2).
IDS-C, mean (SD) 7.4 (5.6) 19.7 (9.3) ⬍0.001
No meaningful changes in AUC values or other psy-
MADRS, mean (SD) 3.8 (4.0) 13.5 (7.8) ⬍0.001
chometric indices were observed in sensitivity analyses that
Abbreviations: BDI-II ⫽ Beck Depression Inventory-II; CESD-R ⫽ Center for Epidemiologic excluded subjects with minor depression (data not shown).
Studies Depression Rating Scale–Revised; GDS-30 ⫽ 30-item Geriatric Depression Scale;
H&Y ⫽ Hoehn and Yahr; HAM-D-17 ⫽ 17-item Hamilton Depression Rating Scale; IDS-C ⫽
DISCUSSION This study compared the perfor-
Inventory of Depressive Symptoms–Clinician; IDS-SR ⫽ Inventory of Depressive Symptoms–
Patient; PD ⫽ Parkinson disease; PHQ-9 ⫽ Patient Health Questionnaire-9; UPDRS ⫽ Unified mance of 9 self-reported and clinician-rated depres-
Parkinson’s Disease Rating Scale. sion scales in community-based patients with PD

Neurology 78 March 27, 2012 1001

 Figure 2 Mean depressive symptom rating scale scores by depressive disorder subgroup

Major depression ⬎No active depression (all scales: p ⬍ 0 䡠 001). Nonmajor depression ⬎No active depression (17-item
Hamilton Depression Rating Scale [HAM-D-17], Inventory of Depressive Symptoms–Clinician [IDS-C]: p ⬍ 0 䡠 001; Beck
Depression Inventory-II [BDI-II], Inventory of Depressive Symptoms–Patient [IDS-SR]: p ⬍ 0 䡠 002; Montgomery-Åsberg
Depression Rating Scale [MADRS]: p ⬍ 0 䡠 005; 30-item Geriatric Depression Scale [GDS-30]: p ⬎ 0 䡠 011; Center for
Epidemiologic Studies Depression Rating Scale–Revised [CESD-R]: p ⬎ 0 䡠 128; Patient Health Questionnaire-9 [PHQ-9]:
p ⬎ 0 䡠 291; Unified Parkinson’s Disease Rating Scale Depression [UPDRS-Dep]: p ⬎ 0 䡠 382). Major depression ⬎ Nonmajor
depression (MADRS, PHQ-9, UPDRS-Dep: p ⬍ 0 䡠 001; HAM-D-17, IDS-C: p ⬍ 0 䡠 002; GDS-30: p ⬍ 0 䡠 006; CESD-R: p ⬍
0 䡠 009; BDI-II: p ⬍ 0 䡠 046; IDS-SR: p ⬍ 0 䡠 060). CI ⫽ confidence interval.

against a consensus panel depression diagnosis. All between “sustained depression” and “sustained de-
scales performed better than chance (AUC ⬎ 0.50) pression with vegetative symptoms.” In addition, a
in detecting depressive disorders and were not influ- single item may not adequately assess all relevant
enced by patient characteristics. The GDS-30, with phenomena. For example, the UPDRS Depression
its favorable psychometric properties, short adminis- assumes that vegetative symptoms cannot occur
tration time, and lack of copyright protection, without dysphoria and does not query for anhedonia.
should have broad appeal as a depression screening An explanation for the AUC difference between the
tool in PD; however, the BDI-II and clinician-rated CESD-R and MADRS is not readily apparent, be-
scales tested also had strong psychometric properties cause the CESD-R has questions similar to those of
and may be useful for depression screening. other scales studied.9,10,17,22,26,34 –36 However, the ob-
Although all scales performed better than chance, served AUC difference is unlikely to be a type I error,
all did not discriminate depressed patients equally because there was also a trend for differences between
well. Groupwise tests of the AUC revealed an in- CESD-R and BDI-II, HAM-D-17, and IDS-C ( p ⬍
equality across all scales and all self-reported scales, 0.01).
but not across clinician-rated scales. In subsequent Most scales were sensitive to differences in symp-
pairwise comparisons, all scales had similar AUCs ex- tom severity between major and nonmajor depres-
cept the CESD-R and UPDRS Depression. sion; however, the IDS-SR did not differentiate
Examining the scale questions provided a poten- nonmajor from major depression and the CESD-R,
tial explanation for the AUC of the UPDRS Depres- PHQ-9, and UPDRS-Depression did not differenti-
sion but not that of the CESD-R.17,35 The UPDRS ate nonmajor depression from no active depression.
Depression was not developed as a single item self- Lack of sensitivity to depression severity suggests that
report scale but rather as a clinician-rated scale.17 As further study is needed before the IDS-SR, CESD-R,
such, its responses are not written in lay terms. For PHQ-9, or UPDRS Depression is used to assess mi-
example, a patient may not understand differences nor depression. These results also question whether

1002 Neurology 78 March 27, 2012

 Table 2 Psychometric properties of depressive symptom rating scales (n ⴝ 229): MOOD-PD and comparison
screening studies

Measure AUC ␣ Cutoff scorea Sensitivity Specificity PPV NPV

BDI-II

Current study 0.85 0.90 ⱖ7 0.95 0.60 0.62 0.94

CESD-R

Current study 0.79 0.92 ⱖ12 0.72 0.70 0.62 0.79

GDS-30

Current study 0.83 0.92 ⱖ10 0.72 0.82 0.73 0.81

Reference14 0.86 ⱖ10 0.81 0.84 0.58 0.94

Reference24 0.89 ⱖ14 0.78 0.85 0.84 0.79

IDS-SR

Current study 0.83 0.88 ⱖ14 0.90 0.60 0.61 0.90

PHQ-9

Current study 0.81 0.85 ⱖ6 0.66 0.80 0.69 0.77

UPDRS Depression

Current study 0.75 N/A ⱖ1 0.70 0.77 0.68 0.79

Reference16 0.79 N/A ⱖ2 0.66 0.81 0.81 0.66

HAM-D-17

Current study 0.86 0.77 ⱖ7 0.77 0.76 0.69 0.83

Reference11b 0.95 ⱖ14 0.88 0.89 0.74 0.96

Reference14 N/A ⱖ13 0.81 0.82 0.58 0.93

IDS-C

Current study 0.88 0.86 ⱖ12 0.81 0.79 0.73 0.86

MADRS

Current study 0.88 0.83 ⱖ8 0.74 0.88 0.81 0.83

Reference11b 0.90 ⱖ15 0.88 0.89 0.74 0.96

Reference15 0.84 ⱖ8 0.72 0.82 0.72 0.82

Abbreviations: AUC ⫽ area under the curve; BDI-II ⫽ Beck Depression Inventory-II; CESD-R ⫽ Center for Epidemiologic
Studies Depression Rating Scale–Revised; GDS-30 ⫽ 30-item Geriatric Depression Scale; HAM-D-17 ⫽ 17-item Hamilton
Depression Rating Scale; IDS-C ⫽ Inventory of Depressive Symptoms–Clinician; IDS-SR ⫽ Inventory of Depressive Symp-
toms–Patient; MADRS ⫽ Montgomery-Åsberg Depression Rating Scale; MOOD-PD ⫽ Methods of Optimal Depression De-
tection in Parkinson’s Disease; NPV ⫽ negative predictive value; PHQ-9 ⫽ Patient Health Questionnaire-9; PPV ⫽ positive
predictive value; UPDRS ⫽ Unified Parkinson’s Disease Rating Scale.
a
The cutoff point that maximized the sum of sensitivity and specificity are presented for comparison with other studies,
not as a recommendation for a cutoff score to be used in clinical practice.
b
Results from Reijnders et al.40 and Naarding et al.12 were not included in the table because they reported on an expanded
sample first reported by Leentjens et al.11

the aforementioned scales are appropriate to monitor whereas the other scales rate the presence or fre-
treatment response; studies assessing sensitivity to quency of symptoms.9,10,17,22,26,34 –36 In addition,
change are needed to address this issue. patients may have difficulty comprehending the IDS-SR,
This study provided novel information on the because study staff noted that subjects often asked clarify-
time needed to complete self-report depression scales ing questions when completing this scale.
in PD. The CESD-R, GDS-30, PHQ-9, and The AUC and other psychometric indices for the
UPDRS Depression took most subjects less than 3 GDS-30, HAM-D-17, MADRS, and UPDRS De-
minutes to complete, whereas the BDI-II took ap- pression were similar to those previously reported in
proximately 5 minutes and the IDS-SR took approx- PD, but cutoff scores based on the maximum sum
imately 10 minutes. Scale length does not explain of sensitivity and specificity differed across stud-
these differences because all the self-report scales ex- ies.11,12,14 –16,24 In addition, cutoff scores were gener-
cept for the UPDRS Depression are composed of 20 ally lower than suggested cutoffs scores for primary
to 30 questions.9,10,17,22,26,34 –36 Differing response op- care patients with major depression.34,35,38,39 This
tions are a more likely explanation. The BDI-II and study is the first to evaluate the BDI-II, CESD-R,
IDS-SR are rated in terms of symptom severity, IDS-C, and IDS-SR in PD. One study evaluated the

Neurology 78 March 27, 2012 1003

 Table 3 Between-scale AUC comparisons (n ⴝ 229)a

UPDRS-
BDI-II CESD-R GDS-30 IDS-SR PHQ-9 Depression HAM-D-17 IDS-C

CESD-R 0.01

GDS-30 0.41 0.13

IDS-SR 0.36 0.06 0.98

PHQ-9 0.06 0.34 0.34 0.22

UPDRS Depression ⬍0.001 0.27 0.01 0.01 0.07

HAM-D-17 0.48 0.01 0.19 0.17 0.03 ⬍0.001

IDS-C 0.11 0.002 0.03 0.03 0.004 ⬍0.001 0.17

MADRS 0.14 ⬍0.001 0.04 0.05 0.004 ⬍0.001 0.32 0.85

Abbreviations: AUC ⫽ area under the curve; BDI-II ⫽ Beck Depression Inventory-II; CESD-R ⫽ Center for Epidemiologic
Studies Depression Rating Scale–Revised; GDS-30 ⫽ 30-item Geriatric Depression Scale; HAM-D-17 ⫽ 17-item Hamilton
Depression Rating Scale; IDS-C ⫽ Inventory of Depressive Symptoms–Clinician; IDS-SR ⫽ Inventory of Depressive Symp-
toms–Patient; PHQ-9 ⫽ Patient Health Questionnaire-9; UPDRS-Depression ⫽ Unified Parkinson’s Disease Rating Scale.
a
p Values for pairwise ␹2(1) tests for AUC values are listed in table 2. A priori statistical significance was set at ␣ ⫽ 0.001.

PHQ-9 as a diagnostic instrument in PD but not for ses. Limitations include the nonrandom sample
depression screening, as used in this study.25 potentially enriched for psychiatric morbidity. Patients
Sampling and methodologic differences may ex- were sequentially mailed recruitment letters inviting
plain the variability in reported cutoff scores. One them to participate in a study to assess psychiatric scales.
possibility is that this community-based sample had In addition, two-thirds of the subject pool elected not to
less psychopathology compared with the previously participate. Although the prevalence of depression is
studied tertiary care samples. Thus, lower cutoff within previously reported ranges, sampling biases are
scores may be a function of a scale score distribution possible.1 Furthermore, exclusion of patients with
whose mean is shifted to the left. Different diagnostic MMSE scores ⱕ23 limits generalizations to patients
methods are another potential source of variability. with significant cognitive impairment. In addition, gen-
In the absence of an objective diagnostic test for eralization of these results may be limited by the under-
depression, the gold standard is the clinician’s representation of racial minorities and the high
diagnosis; therefore, this and previous studies used psychia- educational attainment in the sample. In addition,
trists’ diagnosis of depression according to DSM- completion of clinician-rated scales during the psychiat-
IV-TR criteria as the standard. However, application ric interview might have introduced a bias. To limit
of DSM-IV-TR criteria in PD has limits that can af- potential biases, each case was presented to the consensus
fect diagnostic reproducibility.5 In particular, the de- panel by a psychiatrist other than the interviewing psychia-
cision to attribute symptoms to PD or to a mood trist absent of any self-reported depression scale data or any
disorder and the thresholds used to decide when a clinician-rated scale scores. Furthermore, the time to ad-
sign or symptom is clinically relevant might vary minister clinician-rated scales could not be estimated be-
among examiners. For this reason, the decision to cause they were rated in the context of the clinical
exclude or include somatic items in the evaluation of interview. Finally, self-report scale administration times do
patients with PD is controversial. A NIH workgroup not account for the time to score or interpret them.
recommended an inclusive approach, as used in this The ability to differentiate depressed from nonde-
study, for symptom assessment and diagnosis to en- pressed patients should be the primary consideration
hance sensitivity and reliability of diagnostic criteria.5 when a scale for screening in clinical practice or research
This approach is supported by evidence of similar is chosen, but ease of administration is also an impor-
psychometric properties for the BDI-I in PD when tant consideration.6 This study supports the use of select
somatic items were compared with the affective and self-report scales as alternatives to clinician-rated scales.
cognitive items.23 A final source of variability may be Self-report scales can be administered in waiting rooms
the explicit inclusion of minor depression in our case and discussed with the patient during the clinical exam-
definition. However, this is not likely because exclu- ination, a more practical approach for clinicians in com-
sion of minor depression in sensitivity analyses did parison to the 15- to 20-minute interview required for
not change the psychometric indices appreciably. clinician-rated scales.
Strengths of this study include its use of a Of the self-reported scales tested, the GDS-30
community-based sample, a standardized diagnostic in- had a strong overall combination of psychometric
strument (SCID), and expert consensus panel diagno- and administration characteristics. Although the

1004 Neurology 78 March 27, 2012

BDI-II and GDS-30 had psychometric properties DISCLOSURE
similar to those of the clinician-rated scales, the Dr. Williams is currently employed by and owns stock in Biogen Idec.
E.S. Hirsch reports no disclosures. Dr. Anderson has served on scientific
GDS-30 took less time to administer and has no
advisory boards for Lundbeck, Inc. and CHDI Foundation; has received
copyright restrictions. Other self-reported scales ei- funding for travel from CHDI Foundation; serves as a Section Editor for
ther had a smaller AUC or did not distinguish be- Current Treatment Options in Neurology; serves as a consultant for Guide-
tween major and nonmajor depression. In particular, point Global; serves on the speakers’ bureau for Lundbeck, Inc. and the
FDA; and is supported by a VA Merit Award. Dr. Bush and Dr. Goldstein
the UPDRS Depression performed more poorly than report no disclosures. Dr. Grill has received speaker honoraria from
clinician-rated scales and the IDS-SR took consider- Medtronic, Inc. Dr. Lehmann receives research support from the NIH.
ably more time to complete. Dr. Little has received research support from Neuro Hi Tech and the
This study does not provide explicit guidance on NIH/NIMH. Dr. Margolis has received research support from Janssen,
Forest Laboratories, Inc., Amarin Corporation, Medivation, Inc., the
selection of PD cutoff points. Although some studies NIH (NINDS/NIMH), and NARSAD; has received payment from As-
have used specific formulas to identify screening and traZeneca for expert testimony, royalties from Lippincott Williams &
diagnostic cutoff points, selection of a cutoff point Wilkins for a textbook chapter and licensing fees and payments from
patents related to stem cells; and payments for the development of educa-
should correspond to a scale’s intended use and the
tional presentations from Johns Hopkins University. J. Palanci reports no
resources available for follow-up.6,18,21 In general, disclosures. Dr. Pontone has received research support from EMD Se-
cutoff scores with high sensitivity are desirable to rono, ACADIA Pharmaceuticals, the NIH, Parkinson’s Disease Founda-
identify patients who merit further evaluation for de- tion/Parkinson Study Group Mentored Clinical Research Award, and
The Donna Jeanne Gault Baumann Fund. Dr. Weiss reports no disclo-
pression; cutoff scores with high specificity are desir-
sures. Dr. Rabins serves on a data safety monitoring board for the VA
able when a scale is used as a substitute for a clinical Health System; serves on the editorial boards of International Journal of
interview, such as when a scale score is used as a Geriatric Psychiatry and International Psychogeriatrics; receives publishing
criterion for entry into a clinical trial. royalties for 36-Hour Day (Johns Hopkins Press 2011); receives research
support from the NIH (NIA, NIMH); and has provided legal testimony
Effective screening is a critical first step toward re-
on behalf of Janssen. Dr. Marsh serves on scientific advisory boards for the
ducing the morbidity associated with depression in PD. National Parkinson Foundation, American Parkinson’s Disease Associa-
The GDS-30 may be the most efficient scale to use in tion, and the Parkinson Study Group; receives publishing royalties for
neurology clinics because of its brevity and favorable Psychiatric Issues in Parkinson’s Disease: A Practical Guide (Taylor & Fran-
cis, Informa, 2005); served as a consultant for Merck Serono, Boehringer
psychometric properties. However, with the exception Ingelheim, ACADIA Pharmaceuticals, Lundbeck, Inc., and Ovation
of the UPDRS Depression, all scales studied are valid Pharmaceuticals; received research support from Forest Laboratories, Inc.,
depression screening tools when PD-specific cutoff Eli Lilly and Company, Boehringer Ingelheim, the Michael J. Fox Foun-
dation, and the NIH; and currently receives research support from the
scores are used. Furthermore, all 9 scales were unaffected
NIH, American Psychiatric Association, and the Dystonia Foundation.
by patient characteristics, suggesting that each can be ad-
ministered to a wide range of patients with PD. Received June 14, 2011. Accepted in final form November 14, 2011.

AUTHOR CONTRIBUTIONS
REFERENCES
Dr. Williams: study design, statistical analysis, interpretation of data, and
1. Slaughter JR, Slaughter KA, Nichols D, Holmes SE, Mar-
drafting manuscript. E.S. Hirsch: acquisition of data, study coordination, in-
terpretation of data, and revising manuscript. Dr. Anderson: expert panel
tens MP. Prevalence, clinical manifestations, etiology, and
member, interpretation of data, and revising manuscript. Dr. Bush: statistical treatment of depression in Parkinson’s disease. J Neuro-
analysis and revising manuscript. Dr. Goldstein: neurologic examination, in- psychiatry Clin Neurosci 2001;13:187–196.
terpretation of data, and revising manuscript. Dr. Grill: neurologic examina- 2. Shulman LM, Taback RL, Rabinstein AA, Weiner WJ.
tion, interpretation of data, and revising manuscript. Dr. Lehmann: expert Non-recognition of depression and other non-motor
panel member, interpretation of data, and revising manuscript. Dr. Little: symptoms in Parkinson’s disease. Parkinsonism Relat Dis-
interviewing psychiatrist, expert panel member, interpretation of data, and ord 2002;8:193–197.
revising manuscript. Dr. Margolis: expert panel member, interpretation of
3. O’Connor EA, Whitlock EP, Beil TL, Gaynes BN.
data, and revising manuscript. J. Palanci: acquisition of data, study coordina-
Screening for depression in adult patients in primary care
tion, interpretation of data, and revising manuscript. Dr. Pontone: interview-
settings: a systematic evidence review. Ann Intern Med
ing psychiatrist, expert panel member, interpretation of data, and revising
manuscript. Dr. Weiss: neurologic examination, interpretation of data, and 2009;151:793– 803.
revising manuscript. Dr. Rabins: expert panel member, interpretation of data, 4. Schrag A, Barone P, Brown RG, et al. Depression rating
and revising manuscript. Dr. Marsh: study design, interviewing psychiatrist, scales in Parkinson’s disease: critique and recommenda-
expert panel member, study supervision, obtaining funding, interpretation of tions. Mov Disord 2007;22:1077–1092.
data, and revising manuscript. 5. Marsh L, McDonald WM, Cummings J, Ravina B. Provi-
sional diagnostic criteria for depression in Parkinson’s dis-
ease: report of an NINDS/NIMH Work Group. Mov
STUDY FUNDING
Disord 2006;21:148 –158.
The MOOD-PD study was funded by NIMH grant R01-MH069666.
6. Maruish ME. The Use of Psychological Testing for Treat-
Coauthors were also supported by NINDS grant P50-NS58377 (the Morris
ment Planning and Outcomes Assessment. Mahwah, NJ:
K. Udall Parkinson’s Disease Research Center of Excellence at Johns Hop-
kins), NIA grant T32-AG027668, the Department of Veterans Affairs, the Lawrence Erlbaum Associates; 2004.
Donna Jeanne Gault Baumann Fund, and the Weldon Hall Trust. This arti- 7. Ravina B, Camicioli R, Como PG, et al. The impact of
cle reflects the views of the authors and should not be construed to represent depressive symptoms in early Parkinson disease. Neurol-
the Food and Drug Administration’s views or policies. ogy 2007;69:342–347.

Neurology 78 March 27, 2012 1005

 8. Alexopoulos GS, Abrams RC, Young RC, Shamoian CA. Parkinson’s disease. J Neurol Neurosurg Psychiatry 2005;
Cornell Scale for Depression in Dementia. Biol Psychiatry 76:1445–1447.
1988;23:271–284. 25. Thompson AW, Liu H, Hays RD, et al. Diagnostic accu-
9. Hamilton M. A rating scale for depression. J Neurol Neu- racy and agreement across three depression assessment
rosurg Psychiatry 1960;23:56 – 62. measures for Parkinson’s disease. Parkinsonism Relat Dis-
10. Montgomery SA, Åsberg M. A new depression scale de- ord 2011;17:40 – 45.
signed to be sensitive to change. Br J Psychiatry 1979;134: 26. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity
382–389. of a brief depression severity measure. J Gen Intern Med
11. Leentjens AFG, Verhey FR, Lousberg R, Spitsbergen H, 2001;16:606 – 613.
Wilmink FW. The validity of the Hamilton and 27. Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy of
Montgomery-Åsberg depression rating scales as screening clinical diagnosis of idiopathic Parkinson’s disease: a
and diagnostic tools for depression in Parkinson’s disease. clinico-pathological study of 100 cases. J Neurol Neuro-
Int J Geriatr Psychiatry 2000;15:644 – 649. surg Psychiatry 1992;55:181–184.
12. Naarding P, Leentjens AFG, van Kooten F, Verhey FRJ. 28. Folstein MF, Folstein SE, McHugh PR. “Mini-mental
Disease-specific properties of the Hamilton Rating Scale state”: a practical method for grading the cognitive state of
for depression in patients with stroke, Alzheimer’s disease, patients for the clinician J Psychiatr Res 1975;12:189 –
and Parkinson’s disease. J Neuropsychiatry Clin Neurosci 198.
2002;14:329 –334. 29. Hoehn MM, Yahr MD. Parkinsonism: onset, progression
13. Guy W. ECDEU Assessment Manual for Psychopharmacol- and mortality. Neurology 1967;17:427– 442.
ogy. Rockville, MD: US Department of Health, Education, 30. First MB, Spitzer RL, Gibbon M, Williams JBW. Struc-
and Welfare/National Institute of Mental Health; 1976. tured Clinical Interview for DSM-IV-TR Axis I Disorders,
14. McDonald WM, Holtzheimer PE, Haber M, Vitek JL, Research Version, Patient Edition (SCID-I/P). New York:
McWhorter K, Delong M. Validity of the 30-item geriat- Biometrics Research, New York State Psychiatric Institute;
ric depression scale in patients with Parkinson’s disease.
2002.
Mov Disord 2006;21:1618 –1622.
31. Kessler RC, Andrews G, Mroczek D, Ustun B, Wittchen
15. Silberman CD, Laks J, Capitao CF, Rodrigues CS,
HU. The World Health Organization Composite Interna-
Moreira I, Engelhardt E. Recognizing depression in pa-
tional Diagnostic Interview Short-Form (CIDI-SF). Int J
tients with Parkinson’s disease: accuracy and specificity of
Methods Psychiatr Res 1998;7:171–185.
two depression rating scale. Arq Neuropsiquiatr 2006;64:
32. Lewis S, Hinchcliffe K, Katona C, Livingston G. An infor-
407– 411.
mant interview for the diagnosis of dementia and depres-
16. Starkstein SE, Merello M. The Unified Parkinson’s Dis-
sion in older adults (IDD-GMS). Int J Geriatr Psychiatry
ease Rating Scale: validation study of the mentation,
1998;13:298 –309.
behavior, and mood section. Mov Disord 2007;22:
33. American Psychiatric Association. Diagnostic and Statisti-
2156 –2161.
cal Manual of Mental Disorders: DSM-IV-TR. Washing-
17. Fahn S, Elton RL. Unified Parkinson’s Disease Rating
ton, DC: American Psychiatric Association; 2000.
Scale. In: Fahn S, Marsden CD, Calne DB, Goldstein M,
34. Beck AT, Steer RA, Brown GK. Manual for Beck Depres-
eds. Recent Developments in Parkinson’s Disease II. Flo-
sion Inventory–II. San Antonio, TX: Psychological Corpo-
rham Park, NJ: Macmillan Health Care Information;
ration; 1996.
1987:153–164.
35. Eaton WW, Muntaner C, Smith C, Tien A. Revision of
18. Weintraub D, Oehlberg KA, Katz IR, Stern MB. Test
characteristics of the 15-item geriatric depression scale and the Center for Epidemiologic Studies Depression (CESD)
Hamilton depression rating scale in Parkinson disease. scale. In: Maruish ME, ed. The Use of Psychological Test-
Am J Geriatr Psychiatry 2006;14:169 –175. ing for Treatment Planning and Outcomes Assessment.
19. Beck AT, Steer RA. Beck Depression Inventory Manual. Mahwah, NJ: Lawrence Erlbaum Associates; 2004:363–
San Antonio, TX: Psychological Corporation; 1987. 377.
20. Sheikh JI, Yesavage JA. Geriatric Depression Scale (GDS): 36. Rush AJ, Giles DE, Schlesser MA, Fulton CL, Weissen-
recent evidence and development of a shorter version. In: burger J, Burns C. The Inventory for Depressive Symp-
Brink TL, ed. Clinical Gerontology: A Guide to Assess- tomatology (IDS): preliminary findings. Psychiatry Res
ment and Intervention. New York: Haworth Press; 1986: 1986;18:65– 87.
165–173. 37. Leckman JF, Sholomskas D, Thompson WD, Belanger A,
21. Leentjens AFG, Verhey FRJ, Luijckx GJ, Troost J. The Weissman MM. Best estimate of lifetime psychiatric diag-
validity of the Beck Depression Inventory as a screening nosis: a methodological study. Arch Gen Psychiatry 1982;
and diagnostic instrument for depression in patients 39:879 – 883.
with Parkinson’s disease. Mov Disord 2000;15:1221– 38. Kroenke K, Spitzer RL. The PHQ-9: A new depression
1224. diagnostic and severity measure. Psychiatr Ann 2002;32:
22. Yesavage JA, Brink TL, Rose TL, et al. Development 509 –515.
and validation of a geriatric depression screening scale: a 39. Brink TL, Yesavage JA, Lum O, Heersema PH, Adey M,
preliminary report. J Psychiatr Res 1983;17:37– 49. Rose TL. Screening tests for geriatric depression. Clin
23. Visser M, Leentjens AF, Marinus J, Stiggelbout AM, van Gerontol 1982;1:37– 43.
Hilten JJ. Reliability and validity of the Beck depression 40. Reijnders J, Lousberg R, Leentjens AFG. Assessment of
inventory in patients with Parkinson’s disease. Mov Disord depression in Parkinson’s disease: the contribution of so-
2006;21:668 – 672. matic symptoms to the clinimetric performance of the
24. Ertan FS, Ertan T, Kiziltan G, Uygucgil H. Reliability and Hamilton and Montgomery-Aasberg rating scales. J Psy-
validity of the Geriatric Depression Scale in depression in chosom Res 2010;68:561–565.

1006 Neurology 78 March 27, 2012

A comparison of nine scales to detect depression in Parkinson disease: Which scale to
use?
J.R. Williams, E.S. Hirsch, K. Anderson, et al.
Neurology 2012;78;998-1006 Published Online before print March 14, 2012
DOI 10.1212/WNL.0b013e31824d587f

This information is current as of March 14, 2012

Updated Information & including high resolution figures, can be found at:
Services http://www.neurology.org/content/78/13/998.full.html

Supplementary Material Supplementary material can be found at:

http://www.neurology.org/content/suppl/2012/03/14/WNL.0b013e3182
4d587f.DC1.html
References This article cites 31 articles, 6 of which you can access for free at:
http://www.neurology.org/content/78/13/998.full.html##ref-list-1
Citations This article has been cited by 1 HighWire-hosted articles:
http://www.neurology.org/content/78/13/998.full.html##otherarticles
Subspecialty Collections This article, along with others on similar topics, appears in the
following collection(s):
Depression
http://www.neurology.org//cgi/collection/depression
Parkinson's disease/Parkinsonism
http://www.neurology.org//cgi/collection/parkinsons_disease_parkinso
nism
Screening in epidemiology
http://www.neurology.org//cgi/collection/screening_in_epidemiology
Permissions & Licensing Information about reproducing this article in parts (figures,tables) or in
its entirety can be found online at:
http://www.neurology.org/misc/about.xhtml#permissions
Reprints Information about ordering reprints can be found online:
http://www.neurology.org/misc/addir.xhtml#reprintsus

Neurology ® is the official journal of the American Academy of Neurology. Published continuously since
1951, it is now a weekly with 48 issues per year. Copyright Copyright © 2012 by AAN Enterprises, Inc.. All
rights reserved. Print ISSN: 0028-3878. Online ISSN: 1526-632X.