Parkinsonian Signs in Older People in a Community-Based Study: Risk of Incident Dementia

Background  Mild parkinsonian signs occur in 30% to 40% of community-dwelling older people. In a cross-sectional study, the severity of these signs was greater in people with dementia than in people without dementia.

Objective  To determine whether baseline mild parkinsonian signs are a predictor of incident dementia.

Design  A prospective, longitudinal study of community-dwelling older people who did not have dementia or Parkinson disease at baseline.

Methods  A neurological examination was performed on 1028 residents aged 65 years or older in the Washington Height–Inwood community in northern Manhattan, NY. Parkinsonian signs were rated with an abbreviated Unified Parkinson’s Disease Rating Scale, resulting in a parkinsonian sign score (range, 0-40). The risk of incident dementia was assessed using Cox proportional hazards models. In some analyses, data from the modified Unified Parkinson’s Disease Rating Scale were divided into 3 domains: rigidity, axial function, and tremor at rest.

Results  The mean duration of follow-up was 5.6 years, and 224 participants (21.8%) developed dementia. In a Cox model, the risk of incident dementia was 57% higher in participants with a baseline parkinsonian sign score of 2 vs 0 (relative risk, 1.56; 95% confidence interval, 1.04-2.33; P = .03). In a second Cox model, the baseline parkinsonian sign score was associated with incident dementia (relative risk, 1.08; 95% confidence interval, 1.01-1.16; P = .02) independent of associations with baseline age, education, ethnicity, diabetes mellitus, and stroke.

Conclusions  Baseline mild parkinsonian signs are a predictor of incident dementia. Although these signs are mild, they are not prognostically benign. Brain imaging and postmortem examinations might further our insight into the anatomical and pathological basis for mild parkinsonian signs.

Mild parkinsonian signs, including bradykinesia, rigidity, gait disturbance, and resting tremor, have been reported in 30% to 40% of community-dwelling older people, and their prevalence increases with age.^1-4 These signs are associated with a 2-fold increased risk of mortality.³ Their basis is not well understood, and it is not clear whether they reflect a decline in nigrostriatal dopaminergic activity.

In a cross-sectional study of community-dwelling older people,⁴ the severity of these signs was greater in people with dementia than in people without dementia.⁴ In patients with Alzheimer disease, the severity of parkinsonian signs was correlated with the burden of nigral tangle pathologic features.⁵ However, the prognostic significance of these signs is not clear, and more specifically, there are few data that address whether these signs are a marker for the subsequent development of dementia.² Using data from a prospective longitudinal study of community-dwelling older people, we tested the hypothesis that baseline mild parkinsonian signs are associated with incident dementia.

Participants in the Washington Heights–Inwood Columbia Aging Project cohort were drawn by random sampling of healthy Medicare beneficiaries aged 65 years or older residing within a geographically defined area of northern Manhattan, New York, NY. The participants were recruited between 1992 and 1996, and 2126 were enrolled. Their mean age was 78.2 years, and the mean period of education was 8.2 years; 69.2% were women, and 44.9% were Hispanic. Each participant underwent a structured medical interview at study entry and, during the ensuing year, was referred for a standardized neurological examination, which included an abbreviated (10-item) version of the motor portion of the Unified Parkinson’s Disease Rating Scale (UPDRS)⁶ and a standardized neuropsychological battery of tests.⁷ This evaluation was repeated on a yearly basis.

We excluded data from 352 individuals who did not have a neuropyschological battery of tests and 339 who had dementia at baseline. Participants were considered to have dementia if they met established criteria⁸ and if functional difficulties could be attributed to cognitive rather than physical disability.⁹ The diagnostic criteria for dementia required a disturbance of intellectual function that interfered with work or social activities, demonstrable impairment in memory based on neuropsychological testing, impairment in at least 2 other cognitive domains, and an absence of delirium.

We assigned a diagnosis of Parkinson disease (PD) or Parkinson plus syndrome based on research criteria,¹⁰ and participants were considered to have PD or Parkinson plus syndrome if (1) they had previously received 1 of these diagnoses or (2) they had on the standardized neurological examination 2 or more cardinal signs of parkinsonism. Cardinal signs were bradykinesia, rigidity, postural instability, and rest tremor. A cardinal sign was considered to be present when 1 UPDRS rating was 2 or higher. Twelve (0.84%) of the remaining 1435 participants had a diagnosis of PD or Parkinson plus syndrome, consistent with a prevalence of 0.7% to 1.2% for PD that has been reported for persons aged 65 years or older in northern Manhattan.¹¹ We further excluded 29 participants who were taking a neuroleptic medication because parkinsonian signs can result from the use of these medications; we excluded 160 participants because of incomplete neurological examinations and 206 participants who were not observed beyond baseline. The final sample of 1028 participants had a mean age of 76 years and a mean period of education of 8.5 years; 69.8% were women, and 47.1% were Hispanic. The study was approved by our institution’s internal review board. Written consent was obtained from all participants.

Each year, a standardized neurological examination was conducted by 1 of 3 neurologists; the examination included a 10-item version of the motor portion of the UPDRS.⁶ The 10 items were speech, facial expression, tremor at rest, rigidity (neck, right arm, left arm, right leg, and left leg), posture, and body (axial) bradykinesia. Each item was rated from 0 to 4. A rating of 1 indicated a mild abnormality; ratings of 2 or higher indicated an abnormality of moderate or greater severity. The reliability of the assessment of parkinsonian signs by the 3 neurologists has been reported and is excellent.⁴ A parkinsonian sign score (range, 0 [no parkinsonian signs] to 40 [maximum]) was calculated for each participant.

All participants underwent a standardized neuropsychological battery of tests.⁷ If the participant had dementia, information on demographics was obtained from the most knowledgeable informant or an informant in conjunction with the patient. Demographic data included age, sex, and ethnic group based on self-report using the format of the 1990 US Census. A detailed medical history was obtained, and participants were asked whether they had hypertension, diabetes mellitus, heart disease (including myocardial infarction, congestive heart failure, and valvular heart disease), or stroke, and whether they had ever smoked cigarettes or consumed alcohol at least once per week during their lifetime.

All statistical analyses were performed using SPSS version 11.0 (SPSS Inc, Chicago, Ill). Pearson correlation coefficient (r) was used to assess the association between 2 continuous variables. To model the relative risk (RR) for incident dementia, we used Cox proportional hazards models. The onset date for incident dementia was the date of diagnosis of dementia in our study, and the time variable was the disease duration. In the Cox model, we adjusted for other covariates, including age (continuous variable), education, ethnicity, baseline diabetes mellitus, and baseline stroke. In the initial Cox model, the parkinsonian sign score was stratified into groups (score = 0, score = 1, score = 2, score ≥ 3). In additional analyses, the parkinsonian sign score was treated as a continuous variable. In addition, for some Cox models, data from the modified UPDRS were stratified into 3 domains, based on the results of a factor analysis.¹² These domains were rigidity (neck, right arm, left arm, right leg, and left leg), axial function (speech, facial expression, posture, and body [axial] bradykinesia), and tremor at rest.

The 1028 participants had a mean duration of follow-up of 5.6 years (median, 5.7 years; range, 1-13 years). The mean baseline parkinsonian sign score was 0.89 (range, 0-12). The score was 0 in 673 participants (65.5%), 1 in 126 (12.3%), between 2 and 4 in 178 (17.3%), and 5 or higher in 51 (5%).

At baseline, the parkinsonian sign score was correlated with age (r = 0.21; P<.001) but not with education (r = −0.04; P = .24). The baseline parkinsonian sign score was higher in participants who had hypertension or stroke (Table 1).

Of the 1028 participants, 224 (21.8%) developed dementia. At baseline, they were older, more likely to have diabetes mellitus or stroke, had fewer years of education, were less likely to be non-Hispanic white (Table 2), and had a higher baseline parkinsonian sign score (1.31 ± 2.06 vs 0.78 ± 1.61; t = 4.08; P<.001) than the 804 participants who did not develop dementia.

The baseline parkinsonian sign score was stratified into 4 groups (score = 0 [reference group], score = 1, score = 2, score ≥ 3), and in a Cox model that adjusted for baseline age, education, ethnicity, diabetes mellitus, stroke, sex, hypertension, heart disease, drinker (ever vs never), and smoker (ever vs never), the risk of incident dementia was higher in participants with a baseline parkinsonian sign score of 3 or higher (RR, 1.57; 95% confidence interval [CI], 1.07-2.32; P = .02) or of 2 (RR, 1.56; 95% CI, 1.04-2.33; P = .03) but not a score of 1 (RR, 1.22; 95% CI, 0.82-1.83; P = .33). When the baseline parkinsonian sign score was further stratified, the risk of incident dementia was increased in participants with scores of 4 or higher, compared with participants with a score of 0 in adjusted Cox models. In a Cox model, the baseline parkinsonian sign score was associated with incident dementia (RR, 1.08; 95% CI, 1.01-1.16; P = .02) after adjusting for baseline age, education, ethnicity, diabetes mellitus, stroke, sex, hypertension, heart disease, drinker (ever vs never), and smoker (ever vs never), meaning that with each 1-point increase in the parkinsonian sign score, the risk of incident dementia increased by 8% (Table 3).

We divided the modified UPDRS into 3 domains: rigidity, axial function, and tremor at rest. In an adjusted Cox model, only the axial function domain was associated with incident dementia (RR, 1.21; 95% CI, 1.09-1.35; P<.001).

In this prospective longitudinal study, we assessed whether baseline mild parkinsonian signs were a predictor of incident dementia in a cohort of community-dwelling older people. If so, this would indicate that these signs have some prognostic significance. This study had important strengths. We observed a large multiethnic cohort of individuals for up to 13 years (mean, 5.6 years), and parkinsonian signs were assessed using a standardized rating scale. Also, we collected data on vascular risk factors and events, which are potentially important confounders in the association between mild parkinsonian signs and dementia.

Mild parkinsonian signs were a predictor of incident dementia. Participants with baseline parkinsonian sign scores of 2 and 3 or higher were more than 1.5 times (56%-57%) more likely to develop incident dementia than were the participants without these signs. These data, along with the report that mild parkinsonian signs are associated with an increased risk of mortality,³ indicate that these signs, although mild, are not prognostically benign.

The pathological basis for the reported association between mild parkinsonian signs and incident dementia is not clear. One possibility is that mild parkinsonian signs occur when there are neurofibrillary tangles in the basal ganglia and those tangles later spread to other brain structures, resulting in dementia. There are other possibilities. We observed that participants with vascular risk factors had higher baseline parkinsonian sign scores than those without these conditions, raising the possibilities that vascular risk factors and vascular events might result in pathological changes in the basal ganglia and in mild parkinsonian signs and that these pathologic changes subsequently become more widespread, resulting in dementia.

We stratified data from the modified UPDRS into 3 domains, but only changes in the axial function domain were associated with incident dementia. The basis for this finding is unclear, although there is some evidence that the control of midline motor activity may involve brainstem nuclei such as the pedunculopontine nucleus, which is distinct from the classical striato-pallido-thalamic loops.¹³

The risk of incident dementia was associated independently with a number of baseline factors in addition to parkinsonian sign score. These were higher age, lower education, nonwhite ethnicity, diabetes mellitus, and stroke. Several of these associations (eg, age and education) have been reported previously. The independent association of vascular risk factors with incident dementia suggests that these factors might play a role in the onset of dementia.

We can compare our results with those of a study of 824 older Catholic clergy members who were observed for an average of 4.6 years.² The risk of dementia was increased in participants who had moderate and rapid progression in their parkinsonian signs. That cohort differed from ours in several respects. The earlier study’s subjects were recruited from a religious order, whereas the present study was community-based. In addition that cohort was predominantly non-Hispanic white (89.8%) and highly educated (mean, 18.2 years). In a preliminary study of our own cohort,¹ we found an association between parkinsonian signs at baseline and incident dementia, but the small sample size (228) and small number of follow-up assessments (1, for most participants) precluded the use of statistical procedures such as Cox proportional hazards models.

Two participants developed incident PD, but neither developed dementia. While it is possible that some of our participants could develop incident PD during additional follow-up, given the low reported incidence of PD in this age group (0.5-2.5 per 1000 persons per year), the number is expected to be small.

Baseline mild parkinsonian signs are a predictor of incident dementia. This furthers the notion that these signs, although mild, may not be prognostically benign. The basis for these signs is unclear, and future studies should include brain imaging as well as postmortem examinations, when possible, to gain further insight into these signs’ anatomical and pathological basis.

Correspondence: Elan D. Louis, MD, MS, Unit 198, Neurological Institute, 710 W 168th St, New York, NY 10032 (EDL2@columbia.edu).

Accepted for Publication: January 28, 2004.

Author Contributions:Study concept and design: Louis and Mayeux. Acquisition of data: Mayeux. Analysis and interpretation of data: Louis, Tang, and Mayeux. Drafting of the manuscript: Louis. Critical revision of the manuscript for important intellectual content: Louis, Tang, and Mayeux. Statistical expertise: Louis, Tang, and Mayeux. Obtained funding: Mayeux.

Funding/Support: This study was supported by grants NIH AG07232, R01 NS39422, and R01 NS42859, National Institutes of Health, Bethesda, Md.