The incongruities of the NPI−Q score obtained by
versus that obtained directly from the non−demented patient with
Marcos Serrano Dueñas,1,2 María
Fundamento: La sintomatología
neuropsiquiátrica acompaña frecuentemente a la enfermedad de
Parkinson. Objetivo: Este trabajo (transversal analítico)
exploró al Neuropsychiatric Inventory en su versión corta (NPI–Q)
en setenta pacientes consecutivos. Determinar si la información
del cuidador principal (CP) se correlaciona con la que
proporciona el paciente. Valorar en qué medida esta herramienta
desarrollada para evaluar sujetos dementes puede aplicarse a
enfermos con Parkinson no demenciados (PND).
Métodos: Empleamos la NPI–Q (versión validada al
español), en primer lugar con el CP a solas y, subsecuentemente,
de forma exclusiva con el enfermo PND por otro investigador que
desconocía los resultados de la entrevista previa. Obtuvimos el
estadístico Kappa ponderado (coeficiente de correlación que mide
la consistencia, fiabilidad y reproducibilidad inter-evaluadores
(una medida de la estabilidad del instrumento); valores de 0,7 o
superiores se aceptan como significativos. En vista de que los
resultados fueron insignificantes, se procedió a realizar un
análisis por hoc. Estudiamos la validez convergente (validez
relacionada con el constructo, que incluye las relaciones
lógicas que debe mostrar frente a otras mediciones) para ellos
empleamos el estadístico de Spearman, cuyos valores de: (r<0.29
son correlaciones débiles; r <0.3–0.58 son correlaciones
moderadas; y, r < 0.59 son elevadas correlaciones de validez
Resultados: La consistencia inter-evaluador
(CP–PND), no mostró ninguna concordancia significativa; las
correlaciones inter-item con mejores valores provinieron de los
pacientes; y, las correlaciones de Spearman (una medida de la
validez convergente) fueron más significativas en los PND.
Conclusiones: Nosotros podemos
asumir que los PND son la mejor fuente de información y
consideramos que el NPI–Q (reportado por el CP) no es una
herramienta adecuada para evaluar la sintomatología
neuropsiquiátrica en PND.
Background: Neuropsychiatric symptomatology
frequently accompanies Parkinson’s disease. Purpose: In this
work (a one point time analytical study), we wished to explorer
the Neuropsychiatric Inventory in its shorter format (NPI–Q) in
seventy consecutive patients. To find out in what way
information obtained from the caregiver (CG) is correlated with
that which the patients provide. If this tool, designed to
evaluate patients with dementia, can be applied to non-demented
subjects with Parkinson’s disease (NDP).
Method: The study, the NPI–Q
Spanish version was employed, first with the CG alone and
subsequently, with the individual patient alone by a researcher
with no knowledge of the previous result from the CG. The
weighted kappa correlation coefficient was measured to evaluate
the CG–NDP consistency rate (Test–retest reliability; obtained
by judging the reproductibility or stability of a instrument
over time; two or more observers; or two or more times); a value
of 0.7 or higher was accepted as significant. Because the
correlation results were found to be insignificant, a post–hoc
analysis was performed. We study the convergent validity (Validity–Construct–related,
include examining the logical relations that should exist with
other measures, know too as convergent validity) analyzed using
the Spearman rank correlation statistic values greater than (r
<0.29 are weak correlations; r <0.3– .58 are moderate
correlations and; r < 0.59 are high correlations the convergent
Results: The consistency inter-rater (CG–NDP),
were without any significant agreement; in the inter- tems
correlations the best values were for the patients; and in the
Spearman’s correlations (a measure of converging validity), the
values obtained on the NDP were more significant.
Conclusions: We can assume that the patient’s NDP
is the best source of information, and hence for these patients,
we consider that the NPI–Q (reported by the CG) is not the best
tool to evaluate the Neuropsychiatric symptomatology of NDP
Key words: Neuropsychiatric
symptomatology; Neuropsichiatric Inventory short Format (NPI–Q);
Parkinson’s disease; Consistence inter–rater; Convergent
Neuropsychiatric symptomatology (NPS),
defined as the presence of hallucinations and delusions
occurring during consciousness, frequently accompanies
Parkinson’s disease (PD).1,2 Rates between 20 and 30% have
recently been reported.3 Its appearance causes an increased
workload for the caregiver (CG)4, 5 and is often the prime
cause of institutionalization.3, 5
For this evaluation we used, among other
psychometric tools, the Neuropsychiatric Inventory (NPI)6,
which was originally designed to evaluate NPS in patients
with dementia.7 The original version was later developed
into a shorter format (NPI–Q)8, and re–designed for self–rating
by the caregiver. The most recent version has twelve
parameters, each scored on a scale of 0 to 3 points, with
zero indicating the parameter’s absence and three reflecting
its most severe form. Thus, a total score of 36 points would
indicate maximum severity in all parameters considered.
We used a Spanish version of the NPI–Q9, that
showed a correlation of the Pearson coefficient(r = 0, 89)
in the test–retest analysis and convergence in the NPI (Pearson’s
correlation r = 0,879) with the gravity scale.
In a study carried out by Aarsland et al6 in
which the NPI was used, the authors found that patients with
Parkinson’s and dementia had a more severe NPS score when
compared with non-demented or questionably demented patients
with Parkinson’s disease.
Thus, questions arise as (i) in what way can
information from the CG (defined as the person in charge of
the patient and the one having closest contact with the
medical team10, be correlated with that supplied by the
patient, and (ii) to what extent a tool that was originally
designed to evaluate the NPS in demented patients, can also
be applied to patients without dementia or even to non–dementia
subjects suffering from Parkinson’s (NDP) This study was
designed to answer these questions.
The topic is of considerable importance as it
is wellknown that not all patients have adequate supervision
and hence they are reluctant (or completely unwilling) to
discuss their symptoms with the CG11. The main objective of
the study was to determine the level of agreement between
the responses acquired by the CG with those acquired
directly from the NDP patient. To achieve this objective, an
analytical (one point) study was designed to determine the
consistency between the CG and NDP responses to NPI–Q scores.
Patients, materials and methods
Seventy patients diagnosed with PD were selected according
to the criteria of the United Kingdom Parkinson’s Disease
Society Brain Bank (UKPDSBB).12 Patients were selected
randomly from those visiting the Movement Disorders
outpatient clinic of the Neurology Service of the Carlos
Andrade Marín Hospital (HCAM) in Quito, Ecuador.
Adequate sample size was determined to be 54
patients, calculated using the following equation13: n =
z(2) [p(1-p)/e(2)] : 95% level of confidence (z = 1.96); 10%
prevalence (p = 0.10), 92% preciseness (e = 0.08) . n =
1.96(2) [0.10(1-0.10)/0.08(2)] = 54 patients.
All the patients were evaluated in the period
known as ON: the best functional situation. Patients with
any of the following conditions were not included in the
study: those with a record of psychiatric illness,
illiterate patients, those currently suffering from any
serious concomitant illness, amputees, and those with a
score of < 24 in the Mini Mental Status Examination (MMSE),14
Spanish version,15 regarded as an indicator of the onset of
Other patients’ conditions were also
evaluated: the level of sections I, II and III of United
Parkinson’s Disease Rating Scale (UPDRS)18, the state of the
illness Hoehn and Yahr staging (H&Y)19, and daily activities
(Schwab and England (S&E)20. Later in the study, the NPI–Q
Spanish version9 was employed, first with the CG alone and
subsequently, with the individual patient alone by a
researcher with no knowledge of the previous result from the
This study was approved by the Teaching and
Research Management at the HCAM; and each participating
patient provided their informed written consent. The CG had
demographic information about the patient, information about
how many other patients shared the room with the patient in
the study and, if the patient was married, how many shared
the bed. The weighted kappa correlation coefficient was
measured to evaluate the CG–NDP consistency rate (Test–retest
reliability; obtained by judging the reproductibility or
stability of a instrument over time; two or more observers;
or two or more times); a value of 0.7 or higher was accepted
as significant21. If CG–NDP correlation results were found
to be insignificant, a post-hoc analysis was performed.
Finally, because a relation between increased
neuropsychiatric disturbance and the H&Y stage was found (Aarsland
et al6 and Fenelon et al22) we study the convergent validity
(Validity–Construct–related, include examining the logical
relations that should exist with other measures, know too as
convergent validity) analyzed using the Spearman rank
correlation statistic values greater than (r <0.29 are weak
correlations; r <0.3–0.58 are moderate correlations and; r <
0.59 are high correlations the convergent validity).23 The
total of NPI–Q for the CG and the NDP were compared
according to degree of sickness: H&Y, S&E and UPDRS (I, II,
III sections and Total).
Seventy patients were sampled: 54 were male, 16 were female
with a combined average age of 70.5 years. Thirty eight
patients had attended primary school. On average, patients
had suffered this disease for a period of six years with
L–dopa therapy during 5 years and activities of daily life
activities (S&E) of 70%. Their scores in the MMSE averaged
26.7 and 61 in the first three sections of UPDRS. (Table 1)
According to information from the CG, 38
patients were husbands or wives 23 were children, and the
other nine were nephews or nieces, distant relatives or
friends. Thirty eight patients were elementary school
students, 27 were secondary school age and eleven were high
school students. Of the 70 patients, 46 shared their home
room (thirty seven couples, eight children and one employee);
thirty of the couples shared the same bed.
The score of the NPI–Q based on the CG was
between 0 and 26, with an average of 10.57; 85.7% of the
patients had a score less than 18. The patients gave
themselves an average score of 12.59 and a rank between 2
and 25; 88.5% achieved a score of less than 18. It was
obvious that in scores from both sources, CG’s and the
patients themselves, parameters that reflected mood, apathy
and irritability contributed the most to the total score
(63.5%) to the CG’s score; and (73.8%) to the patient’s own
score. It is important to point out that these parameters
have to do with the mood of the patient and are not
necessarily present in psychosis.
Table 1: Descriptive
characteristics of the sample
Mean values and standart deviation
The low kappa values obtained (Test–retest
reliability) revealed that there is no consistency between
the CG and the NDP scores (Table 2).
When the Spearman’s correlations were
examined as a measure of converging validity of the NPI–Q
(total score), based on the scales mentioned above, the
values obtained on the NDP were more significant consistency
than the ones obtained on the CG (Table 3).
Table 2: Inter – rater
< 0.7 as significant
The low kappa values obtained revealed that there is no
consistency between the CG and the NDP. Thus, we were forced
to question the relevance of the information provided to us
by the CG and its value. Also, the NPI–Q was not designed
specifically to measure psychosis in the PD (many parameters
reflected mood, apathy, irritability, sleep disturbances or
eating disorders), the only test speci- fic to the PD is the
Parkinson Psychosis Rating Scale.24
Our results support that the NDP is the best
source of information because his/her data allow us, for
example, to quantify the UPDRS, to quantify quality of life,
and to evaluate psychotic symptomatology. As well, it has
been shown that the patient may choose not to talk about all
that is happening in his/her body and mind to the CG in
order (i) not to worry this person or not to appear “crazy,”
(ii) for not considering the matter important, (iii) because
he/she doesn’t trust the CG or (iv) simply because the CG
isn’t quite observant enough.11,22,25
Correlations between sleeping disorders and
hallucinatory and delusionary elements have been previously
described.22 Incongruities were found and patients were
considered to have the best inter– tem correlation (0.44).
We conclude, therefore, that their information is the most
consistent. Similarly, the patient’s values for convergent
validity are better, probably because this information is
given directly by the NDP. As well, similarities occur due
to correlation with the motor test (third section of UPDRS)
which is not influenced by information derived from the
Here we present three hypotheses to justify
the lack of agreement in the correlation between CG and NDP
values: (i) many hallucinations, especially simpler ones, do
not cause serious impact on patients and are heavily under-reported
(only 1% are reported by patients22), (ii) many patients do
not report such events in order not to sound “crazy” or
“insane” or not to worry their relatives, or because they
don’t think they are important. It is important to be aware
of the discrepancy between spontaneous reports of
hallucinations (1% of patients) versus reports of
hallucinations following direct questioning to determine
their occurrence (20–30% of patients)22; finally, (iii) we
believe that the MMSE is not the best way to determine the
absence of dementia in Parkinson’s patients because it over–emphasizes
memory disorders that are symptoms of Alzheimer’s disease,
and that the cognitive disorders of Parkinson’s are more
circumscribes to the executive function as been previously
described.26,27 As a result, patients that suffer
Parkinson’s with dementia but who are not sensitive to MMSE
might not be distinguished under this testing protocol.
Table 3: Spearman correlation (rho)
correlation (p<0.05 = 2 tyles=)
** significant correlation (p<0.01 = 2 tyles=)
As a result of this study, we conclude that
the NPI–Q carried out by the CG is not the best tool to
value NPS in non–dementia patients with Parkinson’s disease.
It would be valuable to follow–up this evaluation with a
higher number of NDP’s using other psychometric tools to
evaluate cognitive function.
We thank Merche Boada, MD and the ACE Foundation (Barcelona,
Spain) for providing us with the Spanish version of NPI–Q;
and Margaret Stern, PhD for their editorial assistance.
1. Friedman JH, Fernández HH. The non-motor problems of
Parkinson’s disease. Neurologist 2000;6:18-27.
2. Thanvi BR, Munshi SK, Vijaykumar N, Lo TCN.
Neropsychiatric non-motor aspects of Parkinson’s disease.
Postgrad Med J 2003;79:561-565.
3. Henderson MJ, Mellers JDC. Psychosis in
Parkinson’s disease: “between a rock and a hard place”.
International Review of Psychiatry 2000;12:319-334.
4. Ellgring JH. Depression, psychosis, and
dementia. Neurology 1999;52(suppl3):S17-S20.
5. Melamed E, Friedberg G, Zoldan J.
Psychosis. Impact on the patient and family. Neurology
1999;52 suppl3): S14-S16.
6. Aarsland D, Larsen JP. Lim NG, Janvin C,
Karlsen K, Tandberg E, Cummings JL. Range of
neuropsychiatric disturbances in patients with Parkinson’s
disease J Neurol Neurosurg Psychiatry 1999;67:492-496.
7. Cummings JL, Mega M, Gray K, Rosemberg-Thompson
S, Carusi DA, Gornbein J. The Neuropsychiatric Inventory:
comprehensive assessment of psychopathology in dementia.
8. Kaufer DI, Cummings JL, Ketchel P, Smith
V, MacMillan A, Shelley T, López OL, DeKosky ST. Validation
of the NPI-Q, a brief clinical form of the Neuropsychiatric
Inventory. J Neuropsychiatry Clin Neurosci 2000;12:233-239.
9. Boada M, Cejudo JC, Tárraga L, López OL,
Kaufer D.Neuropsychiatric Inventory Questionnaire(NPI-Q)
validación española de una forma abreviada del
Neuropsychiatric Inventory(NPI). Neurología 2002;17:317-
10. Espinás J, Casabella B, Megido MJ, Rico
MM. Protocolo de demencias en atención primaria. Formación
médica continuada en atención primaria 1999;6(suppl5).
11. Aarslan D, Larsen JP. Diagnosis and
treatment of hallucinations and delusions in Parkinson’s
disease. In: Bédard M-A, Agid Y, Chouinard S, Fahn S,
Korczyn AD, Lespérance P, eds. Mental and behavioral
dysfunction in movement disorders. Totowa: Humana Press,
12. 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.
13. Tamaño muestral v. 1.1. Unidad de
Epidemiología Clínica y Bioestadística. Pontificia
Universidad Javeriana. Bogotá – Colombia, 2005.
14. Folstein MF, Folstein SE, Mchugh PR.
Mini–Mental State. A practical method for grading the
cognitive state of partients for the clinician. J Psychiatry
15. Bermejo F, Gómez TI, Morales JM. El
Mini–Mental State Examination en la evaluación del deterioro
cognitivo y la demencia. En: del Ser T, Peña–Casanova J,
editors. Evaluación neuropsicológica y funcional de la
demencia. Barcelona: Prous Science, 1994:93–107.
16. Menza MA, Robertson–Hoffman DE, Bonapace
AS. Parkinson’s disease and anxiety: Comorbidity with
depression. Biol Psychiatry 1993;34:465–470.
17. Nuti A, Ceravolo R, Piccini A,
Dell´Agnello G, Bellini G, Gambaccini G, et al. Psychiatric
comorbidity in a population of Parkinson’s disease patients.
Eu J Neurology 2004;11:315–320.
18. 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,
eds. Recent development in Parkinson’s disease. New Jersey:
McMillan, 1987, Vol 2: 153–163.
19. Hoehn MM, Yahr MD. Parkinsonism: onset,
progression, and mortality. Neurology 1967; 17: 427– 42.
20. Schwab RS, England AC. Projection
technique for evaluating surgery in Parkinson’s disease. In:
Gillingham FJ, Donaldson IML, eds. Third Symposium on
Parkinson’s disease. Edinburgh: E and S Livingstone, 1969;
21. Abraira V. El índice kappa. Semergen
22. Fénelon G, Mahieux F, Huon R, Ziégler M.
Hallucinations in Parkinson´s disease. Prevalence,
phenomenology and risk factors. Brain 2000;123:733–745.
23. Fisk JD, Brown MG, Sketris IS, Metz LM,
Murray TJ, Stadnyk KJ. A comparison of health utility
measures for the evaluation of multiple sclerosis treatments.
J Neurol Neurosurg Psychiatry 2005; 76: 58–63.
24. Friedberg G, Zoldan J, Weizman A, Melamed
E. Parkinson Psychosis Rating Scale: A practical instrument
for grading psychosis in Parkinson’s disease. Clin
25. Graham JM, Grunewald RA, Sagar HJ.
Hallucinosis in idiopathic Parkinson´s disease. J Neurol
Neurosurg Psychiatry 1997;63:434–440.
26. Hobson P, Meara J. The detection of
dementia and cognitive impairment in a community population
of elderly people with Parkinson’s disease by use of the
CAMCOG neuropsychological test. Age and Ageing
27. Marinus J, Visser M, Verwey NA, Verhey
FRJ, Middelkopp HAM, Stiggelbout AM, van Hilten JJ.
Assessment of cognition in Parkinson’s disease. Neurology