Article Text

Download PDFPDF

The spectrum of neuropsychiatric symptoms in patients with early untreated Parkinson’s disease
  1. D Aarsland1,2,3,
  2. K Brønnick1,
  3. G Alves1,4,
  4. O B Tysnes3,5,
  5. K F Pedersen1,4,
  6. U Ehrt1,2,
  7. J P Larsen1,3,4
  1. 1
    The Norwegian Centre for Movement Disorders, Stavanger University Hospital, Stavanger, Norway
  2. 2
    Department of Psychiatry, Stavanger University Hospital, Stavanger, Norway
  3. 3
    Institute of Clinical Medicine, School of Medicine, University of Bergen, Bergen, Norway
  4. 4
    Department of Neurology, Stavanger University Hospital, Stavanger, Norway
  5. 5
    Department of Neurology, Haukeland University Hospital, Bergen, Norway
  1. Professor D Aarsland, Stavanger University Hospital, The Norwegian Centre for Movement Disorders, PO Box 8100, 4068 Stavanger, Norway; daarsland{at}


Background: Neuropsychiatric symptoms are common in Parkinson’s disease (PD) and have important clinical consequences for patients, caregivers and society. Few studies of neuropsychiatric symptoms in early untreated PD exist.

Objective: To explore the range, clustering and correlates of neuropsychiatric symptoms in an incidence cohort of untreated subjects with PD.

Methods: All cases with incident PD identified during a 22 month period in four counties of Western and Southern Norway were included. Standardised criteria were used to diagnose PD. The Neuropsychiatric Inventory (NPI) was administered to 175 PD and 166 healthy control subjects with similar age and sex distributions. Cluster analysis was used to investigate the interrelationship of NPI items.

Results: The proportion with any NPI symptoms was higher in PD (56%) than in controls (22%) (p<0.001). Depression (37%), apathy (27%), sleep disturbance (18%) and anxiety (17%) were the most common symptoms. Clinically significant symptoms occurred in 27% of the PD group compared with only 3% in the control group (p<.001). Subjects with clinically significant neuropsychiatric symptoms had more severe parkinsonism than those without. Two neuropsychiatric clusters were identified, one characterised by mood symptoms and one by apathy.

Conclusions: Although the majority of patients with early untreated PD do not have clinical significant neuropsychiatric symptoms, these symptoms are more common in patients than in people without PD. Both psychological stress and brain changes associated with PD are likely to contribute to the higher frequencies.

View Full Text

Statistics from

A wide range of neuropsychiatric symptoms have been described in Parkinson’s disease (PD) with important clinical consequences for patients, caregivers and society.1

However, no population based studies of the spectrum of neuropsychiatric symptoms exist in patients with early unmedicated PD. As antiparkinsonian drugs have the potential to influence a range of psychiatric symptoms, including affective symptoms, cognition, impulsivity and psychotic symptoms,2 studying drug naïve patients is required to explore behavioural changes that are related to the disease itself.

Therefore, we administered the Neuropsychiatric Inventory (NPI)3 to a large unselected sample of patients with incident, non-demented, untreated PD and matched controls, derived from the Norwegian ParkWest project,4 a multicentre population based longitudinal cohort study.


Patients and controls

Case ascertainment, patient selection and diagnostic procedures are described in an accompanying paper.4 From 206 subjects who were spouses or friends of the patients, or unrelated persons from the study area, the 175 subjects who provided the best possible match regarding age and sex distribution and years of education to the eligible patients with PD were included as a control group.

Neuropsychiatric assessment

The NPI was used to assess the frequency and intensity of 12 neuropsychiatric symptoms. The maximum possible score is 12, and a score of ⩾4 is usually considered to represent clinically significant severity. In two subjects with missing data on one or two items, a score of 0 was recorded on the missing item whereas subjects with missing data on more than two items were excluded. Motor symptoms were rated using the Unified Parkinson’s Disease Rating scale (UPDRS), subscale III.5 Neuropsychological assessment included the Mini-Mental State Examination (MMSE),6 a battery of neuropsychological tests and standardised history of cognitive impairment, and subjects with possible dementia associated with PD at examination were excluded. See Aarsland and colleagues7 for a detailed description of the procedures and criteria used to diagnose dementia.


Because of the non-linearity and non-normality of NPI subscores, non-parametric analyses (Spearman correlation and Mann–Whitney test) were used to estimate associations between NPI and demographic and clinical features. To explore how the different psychiatric symptoms clustered in individual patients, we performed cluster analyses. The NPI subscores of each patient were standardised to z score values based on the mean and SD of each NPI variable. This was performed in order to ensure equal weighting of the different symptoms in the clustering procedure. A similarity matrix was calculated using a Euclidean distance measure.3 A Ward method cluster analysis was used on the similarity matrix. For validation purposes, we also used an average linkage cluster analysis. The material was considered too small to perform a split sample validation procedure. To identify the appropriate number of clusters, we plotted the overall within group sum of squares over all the included variables and investigated the plot for an “elbow”. A p value <0.05 was considered statistically significant.


Among the 201 eligible PD patients and 175 normal controls, 19 PD and nine control subjects were not administered the NPI or had missing data on more than two items. A further seven patients were excluded because of possible dementia, leaving 175 PD patients and 166 healthy control subjects for inclusion. The patients had a lower mean MMSE score (27.8 (SD 2.3)) than the control subjects (28.6 (1.5)) (p = 0.001), and slightly less years of formal education (10.9 (2.3) vs 11.6 (3.5) years; p = 0.04) but did not differ significantly regarding age (67.8 (9.0) vs 67.3 (9.1) years; NS) or sex distribution (58% vs 60% male; NS).

The NPI scores and proportion with symptoms are shown in table 1. The total NPI score (median and interquartile range 1 (6) vs 0 (0)), number of symptoms (1 (2) vs 0 (0)) and number of symptoms with a clinically significant severity (0 (1) vs 0 (0)) differed significantly between the groups (all p values <0.001). Ninety-eight (56%) of the PD group had a positive score on at least one item compared with 36 (21.7%) in the control group (χ2 = 42.0, p<0.001). In the PD group, depression (34.3%), apathy (27.4%), sleep disturbance (17.8%) and anxiety (16.4%) were the most common symptoms. The number of positive NPI items was higher in PD patients than in control subjects (z = −7.2, p<0.001), with 61 (34.9%) patients having more than one symptom compared with only 10 (6.0%) in the control group.

Table 1 Composite scores and proportions with positive and clinically significant item scores on the NPI in early PD and control subjects

Forty-seven (26.9%) PD patients had at least one symptom of clinical significance, the most common being apathy (11.4%), depression (10.3%), appetite (8%) and sleep (7.5%) disturbances, compared with only five (3.0%) in the control group having a symptom of clinically significant severity (p<0.001). Patients with PD had more symptoms of clinically significant severity than control subjects (z = −6.1, p<0.001), and 22 (12.6%) PD patients had more than one clinically significant symptom, compared with only two (1.2%) in the control group.

Patients with at least one clinically significant score had a higher UPDRS motor subscale score (29.5 (13.7)) than those without clinically significant neuropsychiatric symptoms (21.5 (9.7), t = 4.3, p<0.001) whereas the MMSE score or sex distribution did not differ significantly.

Antidepressant drugs were used by 23 (13.1%) patients, one (0.6%) used an antipsychotic, and a benzodiazepine was used by 21 (12%) patients (regularly by 19, intermittently by two). The proportions taking antidepressants (35.6% vs 5.8%; χ2 = 26.1, p<0.001) and benzodiazepines (26.7% vs 5.5%; χ2 = 15.3, p<0.001) were higher among those with at least one significant neuropsychiatric symptom compared with those without significant neuropsychiatric symptoms.

Cluster analysis was performed to explore whether groups with specific neuropsychiatric syndromes could be identified. Five clusters were identified, based on the finding of a very clear “elbow”, indicating that there was little added benefit of increasing the cluster count from 5 to 6. When divided into two samples, 166 of 175 patients (94.8%) were classified in the same clusters, supporting the validity of the cluster solution. Two of the clusters contained only one and two patients and were discarded, with three clusters remaining. The majority of patients (n = 158, 90.3%) fell into a cluster with no or very mild symptoms (cluster 1). Cluster 2 consisted of patients with mood symptoms (n = 10; 5.7%) (ie, high scores on the depression, anxiety and apathy items) and a final small cluster 3 (n = 4, 2.3%) consisted of patients with a predominantly high apathy score. The three clusters differed significantly in cognition and motor impairment, with patients in the apathy cluster being more impaired than the other groups (table 2).

Table 2 Demographic and clinical features of early PD patients in the two neuropsychiatric clusters and in those with no or mild neuropsychiatric symptoms


This study demonstrates for the first time that a characteristic neuropsychiatric symptom pattern can be found in early drug naïve patients with PD. More than half of the patients exhibited at least one symptom, and more than 25% had at least one symptom of clinically significant severity. Several patients had two or more symptoms, and 13% had two or more clinically significant symptoms. Depression, anxiety, sleep disturbances and apathy were the most common symptoms whereas psychotic symptoms were very rare.

The cluster analysis identified two neuropsychiatric syndromes, one with mood symptoms and one with predominantly apathy. This observation is consistent with previous studies. Studies of medium staged PD (ie, Hoehn and Yahr stages 3–4) have reported neuropsychiatric clusters, including depression, anxiety and apathy clusters, and a psychosis cluster.8 In a cohort consisting of patients with advanced PD and dementia, a group with agitation and a variety of symptoms was added to the mood, apathy and psychosis groups.9 Thus the spectrum of neuropsychiatric symptoms tends to widen with more advanced disease. In early PD, mood and apathy are the predominating neuropsychiatric symptoms. Subsequently, with advancing disease, a group with hallucinations occurs, and in later stages with dementia, a group of patients exhibit multiple and severe neuropsychiatric symptoms including agitation.

The aetiology of neuropsychiatric symptoms in PD is most likely multifactorial, with a possible contribution from both psychological stress and brain changes associated with PD. At this early disease stage, the motor symptoms are usually mild with little functional impairment, but the knowledge of having been diagnosed with a chronic progressive and debilitating brain disease with a potentially negative effect on their job situation, economic security and quality of life10 may induce mood symptoms such as sadness and anxiety. On the other hand, both depression and anxiety can develop before the onset of motor symptoms and the diagnosis of PD,11 strongly suggesting that brain changes occurring early in the disease may render the patients more likely to develop psychiatric symptoms.

Neuropsychiatric symptoms were associated with cognitive impairment. Care was taken to exclude any patient with dementia, and thus this finding adds to previous studies in demonstrating a relationship between psychiatric and cognitive disturbance even in early non-demented patients. This provides further support for distinguishing PD patients with mild cognitive impairment from those with normal cognition, as a harbinger of dementia, as has been shown in imaging12 and longitudinal13 studies.

We found a relationship between motor severity and psychiatric symptoms in PD, consistent with previous studies.8 14 Demonstration of such a relationship in drug naïve patients is a novel finding, and suggests that to some degree, brain changes leading to motor symptoms and psychiatric morbidity develop in parallel, or have similar underlying brain substrates (ie, dopaminergic, nigrostriatal disturbance).

The possibility of studying patients with early drug naive PD allows us to rule out the potential confounding effect of dopaminergic treatment on neuropsychiatric symptoms, which may affect a wide range of behavioural, motivational and emotional functions.2 Antidepressants and anxiolytics were used by some patients, which might have influenced the findings by reducing the severity and frequency of depression and anxiety. Other limitations of the current data set include the cross sectional design, precluding conclusions as to aetiology and course of neuropsychiatric symptoms.

In addition, important psychiatric symptoms such as obsessive–compulsive or impulse control disorders were not assessed. Although the NPI inquires about psychotic phenomena, it does not provide a systematic way of capturing the presence or character of the “minor” forms of psychosis in PD, such as illusions, passage and sense of presence hallucinations. Thus such symptoms may have been present but not detected. Finally, the NPI assesses only severity of depression and anxiety but provides no formal diagnosis, which would require a structured psychiatric interview.

Although the majority of patients with early untreated PD did not have clinically significant neuropsychiatric symptoms, this study shows that such symptoms are clearly more prevalent than in controls. We recommend that screening of neuropsychiatric symptoms should be performed at the time of diagnosis of PD. Longitudinal studies are required to explore the predictive significance of neuropsychiatric symptoms, and how the profile of these symptoms evolve with time.


View Abstract


  • See Editorial Commentary, p 830

  • Funding: This study was funded by the Western Norway Regional Health Authority (grant No 911218) and the Research Council of Norway (grant No 177966).

  • Competing interests: None.

  • Ethics approval: The study was approved by the Regional Committee for Medical and Health Research Ethics, Western Norway, University of Bergen, Bergen, Norway.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles