Depression and anxiety disorders are highly prevalent conditions
in Parkinson´s disease (PD) and constitute a global burden.
Among patients in neurological clinics, concomitant anxiety
syndromes with depression were present in 75% of cases.1 Some
authors consider that significant anxiety in combination with
depression may represent a specific depressive subtype in PD.2
The average prevalence rate of depression in PD has been
calculated around of 40%,3 while the prevalence of anxiety
ranges widely from 5.3% to 40%.4 Though anxiety is a common
problem in PD, relatively little attention has been paid to the
Depression and anxiety are common in the elderly
population.6 When anxiety disorders occur later in life they
tend to be associated with medical and neurological conditions.1
In the WHO Collaborative Study, depression was found to be 9
times more likely in patients with anxiety disorders. There was
substantial overlapping between depressive and anxiety symptoms
in this study: 39% of patients with current depression also had
anxiety disorders and 44% of those with current anxiety
disorders showed comorbid depression.7 In a review of anxiety in
PD, Richard et al.8 noted that up to 60% of patients with
depressive symptoms also suffered from anxiety.
Recently, several factors have led to suggestions
that anxiety and depression are the same disease. Briefly, they
frequently co-exist, there are overlapping symptoms between the
two afflictions, the same neurotransmitters are involved in both
these mental states9,10,11 and, finally, similar agents can be
used to treat them both.
In a recent paper of the NINDS/NIMH Work Group,
the authors suggested the need for research on comorbilities
between anxiety and depression.12 The recent report of the the
Movement Disorder Society commissioned a task force to assess
the clinimetric properties of these scales in PD: “No scales met
the criteria to be ‘‘recommended,’’ and all scales were
classified as ‘‘suggested.’’ Essential clinimetric information
is missing for all scales. Because several scales exist and have
been used in PD, the task force recommends further studies of
these instruments. If these studies show that the clinimetric
properties of existing scales are inadequate, development of a
new scale to assess anxiety in PD should be considered.13”
Our study was designed to investigate the
concurrent validity of the Beck Depression Inventory (BDI) and
Beck Anxiety Inventory (BAI) evaluation scales. We evaluated
both scales against the International Classification of Diseases
version 10 (ICD-10) criteria (as the gold standard)14 for major
depressive disorders (MDD) and for generalized anxiety disorders
(GAD). A second objective of this study was to examine how
effectively the BDI and the BAI differentiate MDD and GAD, as
diagnosed using the ICD-10. Finally, another fundamental
objective of this work was to analyze the interdependency of
anxiety and depression. To this end, we did a study with a K2
factorial design to analyze correlations between the two
Patients, Materials and Methods
Design. This was an observational, analytical crosssectional,
one-point-in-time evaluation with a K2 factorial design.
Patients. One hundred and forty-seven
consecutively included patients diagnosed with PD as per United
Kingdom PD Society Brain Bank criteria,15 in stages 1 to 5 of
the Hoehn and Yahr scale (HY),16 were regularly treated and
followed-up as outpatients at the Movement Disorders Unit of the
Carlos Andrade Marin Hospital (HCAM) Neurology Service in Quito,
The exclusion criteria were severe cognitive
impairment (with a score of over 5/10 as evaluated by Pfeiffer’s
Short Portable Mental Status Questionnaire – SPMSQ17),
illiteracy, serious concomitant illness, blindness, hypoacusis,
or limb amputation. This study was approved by the HCAM
Department of Research and Teaching and all patients involved
provided prior written informed consent.
Assessment Procedure. Demographic data (age
and gender) and historical data (age at onset of PD, duration of
disease, years in treatment with levodopa and dose employed)
were recorded. Neurologist-based assessments were HY stage,16
Schwab and England scale (SES),18 and Unified Parkinson’s
Disease Rating Scale sections 1 to 3 (UPDRS).19 Alternatively
scales used as evaluation tools by a second researcher were the
BDI (Spanish version by Sanz et al.)20 and the BAI (Spanish
version by Sanz et al.)21 Measurements were applied during the
“ON” state in the case of patients with fluctuations.
Seven days after the first evaluation, a third
researcher evaluated patients with the ICD-10 criteria for MDD
and GAD. All the authors remained blind to ongoing test results
during psychometric patient evaluations.
Data Analysis. The following metric
characteristics of the BDI and BAI were explored:
Acceptability. This indicates the extent to which
score distributions adequately represent the true distribution
of health status in the sample. This was determined by comparing
observed to possible score ranges, the proximity of means to
medians, floor and ceiling effects (with <15% accepted as
satisfactory), and skewness of score distributions (accepted
limits: -1 to +1).22
Discrimination. The discriminative capacity of
each scale was statistically analyzed with ROC (receiver
operating characteristic) curves and AUC (area under the
curve).23 A cutoff point was established to determine the
Quality Index of Sensitivity (chance corrected index of
sensitivity = k (1,0)); Quality Index of Specificity (chance
corrected index of specificity = k (0,0)); Efficiency (correct
classification rate; proportion of positives and negatives
classified correctly by the test = EFF); Efficiency of a Random
Test (correct classification rate that would be expected by
chance alone = EFF_RAN);24 Prevalence & Bias Adjusted
Kappa (Kappa adjusted to take account of differences in
perceived prevalence and the relative frequency of positive and
negative observations = PABAK);25 Odds Ratio (Haldane’s
estimator; this estimator of the odds ratio and its standard
error have desirable properties, particularly when cell
frequencies are zero or small = OR’);26 Positive
Likelihood Ratio (LR+) and Negative Likelihood Ratio (LR-), as
well as their 95% confidence intervals (95% CI) to obtain post-test
probabilities.27 K2 Factorial Design.28 Several special cases of
the general factorial design are important because they are
widely used in research work. The most important of these
special cases is that of K factors, each at only two levels.
These levels may be quantitative or they may be qualitative, and
may be owing to the presence or absence of a factor. A complete
replicate of such a design requires 2 x 2 x...x 2 = 2K
observations and is called 2K factorial design. We planned
to examine the magnitude and direction of factor effects to
determine which variables were likely to be important. The
results of the experiment are easily expressed in terms of a
The 147-patient sample was composed of 106 males (72.1%). The
mean age of patients was 68.66 years, and their mean illness
duration was 6.81 years. They had been on L-dopa treatment for a
mean of 5.19 years and were receiving an average dose of 721.53
mg/day. Eight patients were in stage 1; 40 in stage 2; 92 in
stage 3; 6 in stage 4 and just 1 in stage 5, according to
the H&Y scale (see Table 1). For analytical purposes, the single
stage-5 patient was included in the stage-four group.
According to the ICD-10 criteria, 75 patients
(51%) experienced anxiety and 63 (43%) had depression (12 mild,
28 moderate, and 23 severe). The mean score on the BDI was 17.44
points and it was 16.18 on the BAI. Scores differed according to
the H&Y stage, with a tendency toward bimodality for both the
BDI and the BAI; the highest scores were seen in stages 1 and 4
(see Table 2).
We found that 44 subjects (30%) showed depression
and anxiety criteria. There were 19 (12.9%) who suffered
depression without anxiety symptoms and 31 (21%) who had anxiety
without depression. Finally, 53 patients (36%) experienced
neither anxiety nor depression (Table 2). Concerning the
acceptability of both scales, appropriate scores and statistics
were achieved, thus assuring data quality.
Using the ICD-10 criteria as the gold standard,
we did the BDI’s ROC curves and the respective AUC and found
that its ability to discriminate mild depression was 12 Revista
Ecuatoriana de Neurología / Vol. 18, N o3, 2009 low [0.476
(0.384—0.568)]. We got values of [0.740 (0.656—0.825)] for
moderate depression and [0.897 (0.835—0.959)] for severe
depression. When we grouped the entire depressed population, the
BDI showed the following values: [0.858 (0.800—0.916)] (Graph
1). The BAI’s ROC curve and AUC were [0.907 (0.860—0.955)]
To obtain the previously mentioned values, we
analyzed the various cut-off points that enable the best
discriminative qualities for both scales and got cut-off points
of 14/15 for the BDI and 13/14 for the BAI (Table 3). Important
values were obtained for the prior probability and posterior
probability (odds) for those cut-off points (Table 4).
Finally, we decided to use the BDI to assess
anxious patients (ICD-10 criteria) and the BAI to assess
depressed patients (ICD-10 criteria). The ROC curves and the AUC
were [0.739 (0.658—0.819)] and [0.771 (0.695-0.847) ,
respectively (Graphs 3 and 4). Center points were added to the
K2 factorial design, which showed a clear functional dependency
between both factors (Table 5).
The main concern in the use of two-level
factorial design is the assumption of linearity in the factor
effects. Of course, perfect linearity is unnecessary and the K2
system will work quite well even if the linearity assumption
holds only very approximately. In fact, we have noted that if
interaction terms are added to the main effects in the first-order
model, we get:
Which is a model capable of representing any
the response function. This curvature, of course,
comes from the turn in the plane induced by the interaction of
the terms βijχiχj. In our case, this first design revealed (i)
that each factor (anxiety or depression) makes the intercept
negative with values of -6.415 and -17.085, respectively; (ii)
the presence of both places this value at 8.415.
There are going to be situations where the
curvature in the response function is not adequately modeled. In
such cases, a logical model to consider is: where βij represents
pure second-order or quadratic effects.
This equation is called a second-order response
surface model. The method consists in adding center points to
the K2 design.
yF = the average of the four factorial points; nc
= the number of observations at the center (0,0), and yc = the
average of the nc center points. If the yF - yc difference is
small, the center points run near the plane that passes through
the factorial points, and there is no quadratic curve. On the
other hand, if the yF - yc difference is large, a quadratic
curve is present.
The sum of squares of the pure quadratic curve
has one degree of freedom and is rendered by:
The sum obtained may be compared with the mean
quadratic error. More specifically, when we add
points to the center of the 2K design, the curvature test
becomes a hypothesis test:
Furthermore, if the factorial points of the
design have not been replicated, the nc midpoints may be used to
construct an error estimate with nc - 1 degrees of freedom.
Adding midpoints produced significant values, so
H0 is acceptable, and HA is rejected. That is,
there is a linear relationship between both regressors (Table
Our study investigated the concurrent validity of the scales
compared to the ICD-10 and our results are similar to those
obtained by other authors.
Regarding the discriminative ability of the
scales employed, the AUC result obtained [0.858 (0.800—0.916)]
in our sample with the BDI was similar to the 0.8567 obtained by
Leentjens et al. in Parkinson´s patients, even though they used
the DSM-IV as the gold standard.29
Using the DSM-III-R in a population of
hospitalized neurological patients, Lykouras et al. found that
values for the AUC differed depending on the cutoff points,
reaching 0.925 with 20 points and 0.94 with 29 points.30
In another study,31 an AUC of 0.88 was obtained
and the cutscore with the greatest sensitivity (0.715) and
greatest specificity (0.90) was 14/15. Again, these authors used
the DSM-IV as the gold standard. We obtained an identical
cutscore in our study.
With regard to the BAI, its metric properties
have not been measured in Parkinson´s patients, as far as we
know. Hoyer et al.32 used it in an epidemiological study in
adult women, and Kabacoff et al.33 used it to
evaluate elderly psychiatric outpatients. In any case, our
results are not comparable to those from these studies. Indeed,
one of the limitations of our study is that we cannot compare
our results because we have used a different gold standard.
Our findings on the prevalence of depression
(42.8%) are similar to others reported by our group,34 and the
prevalence of anxiety (51%) falls within the range reported by
some authors,8 although we are aware that the BAI may over-assess
the presence of anxiety, as described by Higginson et al.35
It has been persistently stated that scales such
as the HDRS or BDI are not unidimensional and that they involve
too many somatization items that are confused with the selfsame
symptoms of Parkinson´s disease.36,37 The results we present
bear out this lack of dimensionality, since we found that the
BDI´s AUC-measured discriminative ability for anxiety was [0.739
(0.658—0.819)]. With regard to the BAI, which is employed to
discriminate the depressed, AUC values of [0.771 (0.695-0.847)]
We believe our results show, as do others, the
defects of both scales. As de Gruijter DNM and van der Kamp38
say, “A depression inventory, for example, may not merely tap
depression as the intended trait to measure, but also anxiety.
In this case, a reasonable decomposition of observed scores on
the depression inventory would be:
Z = τ +ED
where X is the observed score, τ is the
true score, ED is the systematic error due to the anxiety
component, and EU is the combined effect of unsystematic error”.
This, in our opinion, is due to what Bond TG and Fox CM39 have
clearly stated, “… one’s philosophy of measurement leads one to
use a statistical analysis model that will guide the development
and selection of items –the statistical model is being used as a
means of quality control of the items. This is in contrast to
the most common alternate approach where the statistical model
is augmented by parameters that are designed to accommodate the
characteristics of the item set—one could say that the
statistical model is being used to describe the items.”
Additionally, our results clearly show the
presence of both entities – depression and anxiety – as a
continuum, which is borne out by refinement of the design models.
This coexistence of depression and anxiety has been reported in
a number of studies. In general practice, for example, 39% of
depressed patients suffered anxiety and 44% of anxiety patients
suffered depression.7 Dual prevalence is also well known in the
elderly.6 In fact, one of the first reports of the coexistence
of both entities in Parkinson´s patients, Menza et al.40 found
that 92% of those suffering anxiety had comorbility with
depressive disorder and that 67% of depressed PD patients also
suffered from anxiety. Another article adds that Parkinson´s
patients with anxiety got higher scores on the HDRS.41
More recently, Nuti et al.42 reported that 19.3%
of their subject sample with Parkinson´s suffered depressive
illness and anxiety.
Despite the extensive literature on anxiety and
depression, there is still disagreement about whether these two
syndromes represent distinct clinical disorders, as evidenced in
Anxiety and Depression: Individual Entities or Two Sides of the
Same Coin?.9 This conception of their being two sides of the
same coin is based on the widely observed coexistence of both
entities in populations studied, and the fact that they both
respond to the same drugs, the SSRIs. Finally, we should
consider the monoaminergic circuits as paradigmatically
responsible for depression symptoms.43
On the other hand, it has been suggested that the
hypothesis for anxiety disorders would be the presence of a
reverberating circuit that arises in the orbitofrontal cortex
and projects to the striate, from this to the thalamus, and from
there returns to the prefrontal cortex.44
By unifying these hypotheses, we would conclude
that the monoamines serotonin, noradrenalin, and dopamine are
involved in the parallel circuitry that runs from the prefrontal
regions to the caudal regions of the basal ganglia, and from
these to the thalamus, from whence they return to the cortical
In addition, studies suggest that anxiety
disorders may be particularly difficult to distinguish from
depression46 and that different clusters of symptoms are
reported by depressed and anxious patients on clinical and self-
ating scales.47 Thus, we may also consider that these are very
deeply interpenetrating entities, just as we have shown.
One of the reasons for studying these
correlations is to see the degree of dependence or independence
of the variables. Since neither of the two – depression and
anxiety, or anxiety and depression – can be taken as an
independent regressor while the other is dependent, it is then
essential, in our opinion, to determine the degree of
interdependence between both variables. We believe we have
demonstrated that the combination of anxiety and depression is,
statistically, more significant than the presence of either one
alone. In other words, they may be elements of one and the same
1. Marsh L. Anxiety disorders in Parkinson´s disese. Int Rev
Psychiatry 2000; 12: 307-318.
2. Henderson R, Kurland R, Kersun JM, Como P. Preliminary
examination of the comorbidity of anxiety and depression in
Parkinson´s disease. J Neuropsy Clin Neurosci 1992; 4: 257-264.
3. Marsh L, Berk A. Neuropsychiatric aspects of Parkinson´s
disease: Recent advances. Cur Psychiatry Rep 2003; 5: 68-76.
4. Walsh K, Bennet G. Parkinson´s disease and
anxiety. Postgrad Med J 2001; 77: 89-93.
5. Marsh L. Neuropsychiatric aspects of Parkinson´s disease.
Psychosomatics 2000; 41: 15-23.
6. Martin LM, Fleming KC, Evans JM. Recognition nad management
of anxiety and depression in elderly patients. Mayo Clin Proc
1995; 70: 999-1006.
7. Lecrubier Y. The burden of depression and anxiety in general
medicine. J Clin Psychiatry 2001; 62 (suppl8): s4-s9.
8. Richard IH, Schiffer RB, Kurland R. Anxiety and Parkinson´s
disease. J Neuropsy Clin Neurosci 1996; 8: 383-392.
9. Nutt D. Axiety and depression: individual antities or two
sides of the same coin? Int J Psych Clin Pract 2004; 8 (suppl1):
10. Coles ME, Gibb BE, Heimberg RG. Psychometric Evaluation of
the Beck Depression Inventory in Adults with Social Anxiety
Disorder. Depression and Anxiety 2001; 14: 145–148.
11. Levine J, Cole DP, Chengappa KNR, Gershon S. Anxiety
Disorders and Major Depression, Together or Apart. Depression
and Anxiety 2001; 14: 94–104.
12. Marsh L, McDonald WM, Cummings J, Ravina B, et al.
Provisional Diagnostic Criteria for Depression in Parkinson’s
Disease: Report of an NINDS/NIMH Work Group. Mov Dis 2006; 21:
13. Leentjens AFG, Dujardin K, Marsh L, Martinez-Martin P,
Richard IH, Starkstein SE, et al. Anxiety Rating Scales in
Parkinson’s Disease: Critique and Recommendations. Mov Disord
2008; 23: 2015-2025.
14. World Health Organization. Pocket Guide to the ICD-10
Classification of Mental and Behavioural Disorders. Churchill
Livingstone, London, 1994.
15. Gibb WRG, Lees AJ. The Relevance of the Lewy
Body to the Pathogenesis of Idiopathic Parkinson’s Disease. J
Neurol Neurosurg Psychiatry 1988; 51: 745-752.
16. Hoehn MM, Yahr MD. Parkinsonism: Onset, Progression, and
Mortality. Neurology 1967; 17: 427-442.
17. Pfeiffer E. A Short Portable Mental Status
Questionnaire for the Assessment of Organic Brain Deficit in
Elderly Patients. J Am Geriatr Soc 1975; 23: 433-441.
18. Schwab RS, England AC. Projection Technique for Evaluating
Surgery in Parkinson’s Disease. In: Gillingham FJ, Donaldson IML,
editors. Third Symposium on Parkinson’s Disease. E and S
Livingstone, Edinburgh, 1969; pp. 152-157.
19. Fahn S, Elton RL, and Members of the UPDRS Development
Committee. Unified Parkinson’s Disease Rating Scale. In: Fahn S,
Marsden CD, Goldstein M, Calne DB, editors. Recent Development
in Parkinson’s Disease. McMillan, New Jersey, 1987, Vol 2: pp.
20. Sanz J, Vásquez C. Fiabilidad, validez y datos normativos
del Inventario para la Depresión de Beck. Psicothema 1998; 10:
21. Sanz J, Navarro ME. Propiedades psicométricas
de una versión española del Inventario de ansiedad de Beck (BAI)
en estudiantes universitarios [The psychometric properties of a
Spanish versiono f the Beck anxiety Inventory (BAI) in a
universitary students sample]. Ansiedad y Estrés, 2003; 9:
22. McHorney CA, Tarlov AR. Individual-patient monitoring in
clinical practice: are available health status surveys adequate?
Qual Life Res 1995; 4: 293-307.
23. Pepe MS. The statistical evaluation on medical tests for
classification and prediction. New York: Oxford University Press,
24. Byrt T, Bishop J, Carlin JB. Bias, prevalence and kappa. J
Clin Epidemiol 1993; 46: 423-439.
25. Agresti A. Categorical Data Analysis. 2nd
edition. New York, John Wiley & Sons. 2003.
26. Haynes RB, Devereaux PJ, Guyatt GH. Clinical expertise in
the era of evidence-based medicine and patient choice. Evid
Based Med 2002; 7: 36-38.
27. Deeks JJ, Altman DG. Diagnostic tests 4: likelihood ratios.
Br Med J 2004; 329: 168-169.
28. Montgomery DC, Runger GC. Applied statistics and probability
for engineers. 3rd ed. New York, John Wiley & Sons. 2003.
29. Leentjens AFG, Verhey FRJ, Luijckx G-J, Troost J. The
validity of the Beck Depression Inventory as a screening and
diagnostic instrument for depression in patients with
Parkinson’s disease. Mov Dis 2000; 15: 1221-1224.
30. Lykouras L, Oulis P, Adrachta D, Daskalopoulou E, Kalfaris
N, Triantaphyllou N, et al. Beck Depression Inventory in the
detection of depression among neurological inpatients.
Psychopatology 1998; 31: 213-219.
31. Visser M, Leentjers AFG, Marinus J, Stiggelbout AM, van
Hilten JJ. Reliability and validity of the Beck depression
inventory in patients with Parkinson’s disease. Mov Disord 2006;
32. Hoyer J, Becker ES, Neumer S, Soeder U, Margraf J. Screening
for anxiety in an epidemiological sample: predictive accuracy of
questionnaires. J Anxiety Dis 2002; 16: 113-134.
33. Kabacoff RI, Segal DL, Hersen M, Van Hasselt VB.
Psychometric properties and diagnostic utility of the Beck
Anxiety Inventory with older adult psychiatric outpatients. J
Anxiety Dis 1997; 11: 33-47.
34. Serrano-Dueñas M, Serrano M. Concurrent
validation of the 21-item and 6-item Hamilton Depression Rating
Scale versus the DSM-IV diagnostic criteria to assess depression
in patients with Parkinson’s disease: An exploratory analysis.
Park Rel Disord 2008; 14: 233–238.
35. Higginson CI, Fields JA, Koller WC, Troster AI.
Questionnaire assessment potentially ovesestimates anxiety in
parkinson’s disease. J Clin Psychol Med Set 2001; 8: 95-99.
36. Hoogendijk WJG, Sommer IEC, Tissingh G, Deeg DJH, Wolter Ech.
Depression in Parkinson´s disease. The impact of symptom overlap
on prevalence. Psychosomatics 1998; 39: 416-421.
37. Linden M, Borchelt M, Barnow S, Geiselman B, The impact of
somatic morbidity on the Hamilton Depression Rating Scale in the
very old. Acta Psychiatr Scand 1995; 92: 150-154.
38. de Gruijter Dato NM, van der Kamp Leo JTh. Statistical Test
Theory for the Behavioral Sciences. Chapman & Hall/CRC. Boca
Raton (FL). 2008, p. 11-12.
39. Bond Trevor G, Fox Christine M. Applying The Rasch Model.
Fundamental Measurement in the Human Sciences. 2nd edition.
Lawrence Erlbaum Associates. Mahwah (NJ). 2007, p.xv-xvi.
40. Menza MA, Robertson-Hoffman DE, Bonapace AS. Biol Psychiatry
1993; 34: 465-470.
41. Stein MB, Heuser IJ, Juncos JL, Uhde TW. Anxiety disorders
in patients with Parkinson’s disease. Am J Psychiatry 1990; 147:
42. Nuti A, Ceravolo R, Piccinni A, Dell’Agnello G, Bellini G,
Gambaccini G, et al. Psychiatric comorbidity in a population of
Parkinson’s disease patients. Eur J Neurol 2004; 11: 315-320.
43. Stahl SM. Symptoms and circuits, Part 1. Major depressive
disorder. J Clin Psychiatry 2003, 64: 1282-1283.
44. Stahl SM. Symptoms and circuits, Part 2. Anxiety disorders.
J Clin Psychiatry 2003; 64: 1408-1409.
45. Stein DJ, Westenberg HGM, Liebowitz MR. Social anxiety
disorder and generalized anxiety disorder: Serotoninergic and
dopaminergic neurocircuitry. J Clin Psychiatry 2002; 63 (suppl
46. Coles ME, Brandon EG, Heimberg RG. Psychometric evaluation
of the Beck depression inventory in adults with social anxiety
disorder. Depression and Anxiety 2001; 14: 145-148.
47. Riskind JH, Beck AT, Brown G, Steer RA. Taking the measure
of anxiety and depression. Validity of the reconstructed
Hamilton scales. J Nervous and Mental Disease 1987; 175: