HR indicates hazard ratio.
Likelihood ratios (LRs) for incident parkinsonism by decreasing follow-up duration.
eMethods. Statistical Analysis
eTable 1. Overview of Clinical Parkinsonism Diagnoses
eTable 2. The Distribution of Cognitive Dysfunction and Subtle Motor Features at Baseline, Stratified by Incident Parkinsonism Status
eTable 3. Mini-Mental State Exam and the Risk of Parkinsonism
Customize your JAMA Network experience by selecting one or more topics from the list below.
Darweesh SKL, Wolters FJ, Postuma RB, et al. Association Between Poor Cognitive Functioning and Risk of Incident Parkinsonism: The Rotterdam Study. JAMA Neurol. 2017;74(12):1431–1438. doi:10.1001/jamaneurol.2017.2248
Is poor cognitive functioning associated with a risk of parkinsonism and Parkinson disease in community-dwelling individuals?
In this population-based cohort study including 7386 participants of the Rotterdam Study with median 8.3 years of follow-up, poor baseline cognitive functioning indicated the probable onset of parkinsonism and probable Parkinson disease. The association remained robust beyond the first 8 years and after removing individuals with onset of parkinsonism after dementia.
The findings from this study suggest that cognitive dysfunction can be considered a sign of prodromal Parkinson disease.
Cognitive dysfunction is a common feature among patients with parkinsonism, including Parkinson disease (PD). However, there is a scarcity of data on cognitive functioning before parkinsonism diagnosis, a stage at which patients may still respond to putative disease-modifying interventions.
To assess whether poor cognitive functioning is associated with an increased risk of parkinsonism.
Design, Setting, and Participants
Between January 8, 2002, and December 14, 2008, baseline cognitive function was assessed in 7386 participants of the Rotterdam Study who were free of parkinsonism and dementia. Four tests were administered (Stroop color word test, letter-digit substitution, verbal fluency, and word learning) and a global cognition score was derived from principal component analysis. Subsequently, participants were followed up until January 1, 2015, for the onset of parkinsonism through serial in-person examinations and complete access to medical records. Parkinsonism was defined as the (1) presence of hypokinesia or bradykinesia plus at least 1 other cardinal sign and/or (2) clinical diagnosis by a neurologist or geriatrician. Patients with dementia diagnosis before parkinsonism diagnosis were considered to have probable PD.
Main Outcomes and Measures
Hazard ratios (HRs) for incident parkinsonism per SD decrease in global cognition, adjusted for age, sex, and study subcohort.
A total of 7386 patients were included in the analysis; of these, 4236 (57.4%) were women and mean (SD) age was 65.3 (10.2) years. During follow-up (median, 8.3 years; range, 0-15 years), 79 (1.1%) individuals received a diagnosis of incident parkinsonism; of these, 57 (72.2%) received a diagnosis of probable PD. Among patients with incident parkinsonism, 24 (30.4%) also developed dementia (10 before and 14 after parkinsonism onset). Poor global cognition at baseline was associated with a higher hazard of incident parkinsonism (hazard ratio [HR], 1.79; 95% CI, 1.37-2.33). The association remained robust beyond the first 8 years (HR, 1.59; 95% CI, 1.01-2.59) and after removing individuals with dementia onset before parkinsonism (HR, 1.72; 95% CI, 1.28-2.27). Poor global cognition at baseline was also associated with incident probable PD (HR, 1.52; 95% CI, 1.11-2.08). Letter-digit substitution (HR, 1.59; 95% CI, 1.22-2.04), verbal fluency (HR, 1.61; 95% CI, 1.23-2.08), and inverted interference task Stroop color word test (HR, 1.56; 95% CI, 1.25-1.96) scores were each associated with incident parkinsonism, whereas the association with word learning delayed-task scores was weaker (HR, 1.18; 95% CI, 0.92-1.52).
Conclusions and Relevance
Poor cognitive functioning is associated with an increased risk of incident parkinsonism, including probable PD. Cognition indicates the probability of parkinsonism over long intervals and extends beyond patients with onset of parkinsonism after dementia. The findings suggest that cognitive dysfunction can be considered a sign of prodromal PD.
Poor cognitive functioning is common among patients with parkinsonism, and a substantial subgroup of patients has cognitive dysfunction at the time of parkinsonism diagnosis. Between 15% and 43% of patients with newly diagnosed Parkinson disease (PD) are cognitively impaired1,2 and cognitive impairment at diagnosis is a risk factor for subsequent dementia in patients with PD.3 In patients with dementia with Lewy bodies (DLB), the dementia diagnosis typically precedes parkinsonism onset. In addition, poor cognitive functioning is common among patients with multiple system atrophy, progressive supranuclear palsy, and vascular parkinsonism.4-6
Cognitive function has also been reported to be worse in individuals who are free of parkinsonism but have impaired olfaction and reduced dopamine transporter binding, both of which are strong proxies of PD.7 Recently, a case-control study showed that subtle decline in executive cognitive functions may precede diagnosis in patients with PD, in parallel with subtle motor signs.8 In addition, poor manual dexterity is associated with an increased risk of incident parkinsonism and PD, even when measured years before diagnosis.9
However, it is unclear whether a single global assessment of cognitive functioning is associated with an increased risk of parkinsonism over longer intervals, to what extent the different cognitive domains predict parkinsonism, and how poor cognitive functioning combines with subtle motor signs in patients with prediagnostic parkinsonism. Such insight is critical because it may further unravel the prodromal phase of common parkinsonism diseases, in particular prodromal PD, a phase in which patients may still respond to putative disease-modifying interventions.
We hypothesized that poor cognitive functioning is associated with an increased risk of parkinsonism and tested this hypothesis in a prospective, population-based study. Furthermore, we hypothesized that such an association would be especially driven by performance on tests of executive functions and that a combination of poor cognitive functioning and subtle motor signs would be a strong indicator of the probability for incident parkinsonism.
The present study was embedded in the Rotterdam Study, a large, prospective, population-based study in the Netherlands.10,11 In 1990, inhabitants of the well-defined Ommoord district in the city of Rotterdam who were aged 55 years or older were invited to participate and 7983 individuals agreed (first subcohort). In 2000, all inhabitants who had become 55 years or older or had moved into the study district since the start of the study were invited to be included in the Rotterdam Study and 3011 agreed (second subcohort). The cohort was further extended in 2006 (third subcohort, age ≥45 years) to a total of 14 926 participants (overall response, 72.0%). By 2015, the first subcohort had a total of 6 visits (mean interval between visits, 8 years), the second subcohort had 3 visits, and the third subcohort had 2 visits.
The Rotterdam Study has been approved by the medical ethics committee according to the Population Study Act Rotterdam Study, executed by the Ministry of Health, Welfare, and Sports of the Netherlands. All participants provided written informed consent to participate in the study; participants did not receive financial compensation.
Although cognitive screening tests had been used since 1990, a more comprehensive cognitive test battery was introduced during center visit rounds between January 8, 2002, and December 14, 2008: 2002-2004 (first subcohort), 2004-2006 (second subcohort), and 2006-2008 (third subcohort). Of the original 14 926 participants, 8414 (56.4%) were alive at the time of and participated in this broader assessment, which is considered the baseline for the present study. Participants were extensively screened for parkinsonism and dementia12,13 and we excluded individuals who had parkinsonism or dementia at baseline as well as individuals who had missing data on more than 1 cognitive functioning task (n = 1028). We followed up the remaining 7386 participants until the onset of parkinsonism; onset of dementia; January 1, 2015; or death. Study follow-up for incident parkinsonism was virtually complete (98%).14
A detailed description of the assessment methods of objective cognitive function as well as age- and sex-specific normative values in the Rotterdam Study have been published.15 In short, we used the Stroop color word test (comprising 3 tasks),16 letter-digit substitution test,17 verbal fluency test,18 and 15-word list learning test.19Table 1 lists the domain that is primarily represented by each test. Of these domains, executive abilities, information processing speed, attention, and semantic fluency are commonly impaired in patients with clinical PD, including in those with PD and mild cognitive impairment.20-22 Furthermore, impairment in semantic fluency is associated with a subsequent risk of dementia in patients with PD.23
We used the Purdue Pegboard Test to assess manual dexterity.24 In this test, participants are asked to place as many cylindrical metal pegs into 1 of 25 holes in a pegboard as possible in 30 seconds. The test is performed 3 times, 1 each using the left hand, right hand, and then both hands simultaneously. The mean Purdue Pegboard Test score is the sum of each trial divided by 3.
A detailed description of parkinsonism assessment methods in the Rotterdam Study has been published.25 In the present study, we used 4 overlapping modalities to screen for potential parkinsonism: in-person examinations, in-person interviews, use of antiparkinson medication, and clinical monitoring alerts. In-person examinations comprised 2 phases.
In the first phase, participants underwent standardized assessments of the following parkinsonian signs: tremor (resting, positional, and intentional), hypokinesia and bradykinesia (including arm swing, gait, finger tapping, and general impression), cogwheel rigidity, and postural reflex. These screening assessments were conducted by research nurses who were repeatedly trained by one of us, an experienced neurologist (P.J.K.) during the study period. Individuals who had positive screening results were subsequently invited for a structured physical examination by a research physician who specialized in neurologic diseases.
Of all individuals who had positive screening results in any of these methods, complete medical records (including letters from medical records of specialists and general practitioners) were studied and case reports were prepared covering all potentially relevant information to establish the presence and cause of parkinsonism. These case reports were evaluated by a panel led by the experienced neurologist (P.J.K.).
Parkinsonism was defined as (1) the presence of hypokinesia or bradykinesia in combination with at least 1 other cardinal sign (resting tremor, rigidity, or postural imbalance) as observed by any physician and/or (2) a clinical diagnosis of parkinsonism by a neurologist or geriatrician (if motor examination details were not available).
A detailed description of PD assessment methods in the Rotterdam Study has previously been published.25 Parkinson disease was diagnosed after exclusion of parkinsonism associated with preexistent dementia, use of antidopaminergic drugs, cerebrovascular disease, multiple system atrophy, progressive supranuclear palsy, and evidence for other rare causes (eg, corticobasal degeneration). To diagnose PD, there also had to be (1) a clinical PD diagnosis by a neurologist or geriatrician and/or (2) a positive response to dopaminergic treatment.
During the study period (2002-2015), dementia at parkinsonism diagnosis was used as an exclusion criterion for PD by the neurologists who examined patients and adjudicated case reports in the Rotterdam Study protocol. As a consequence, there were 9 patients with parkinsonism with a clinical presentation consistent with PD who were not originally classified as having PD because their parkinsonism was preceded by a dementia diagnosis. At the time of diagnosis these patients did not have typical clinical features of DLB, such as visual hallucinations or pronounced variations in attention and alertness, or a clear secondary cause of parkinsonism. In the recent Movement Disorder Society criteria of PD,26 prior dementia is no longer an exclusion criterion for PD, suggesting that these patients would now likely receive a clinical PD diagnosis. Therefore, we classified these 9 patients together with those who had already received a clinical PD diagnosis (n = 48) as having probable PD.
A detailed description of assessment methods has previously been published.27 In short, participants were screened for dementia at baseline and follow-up examinations using a 3-step protocol.13 Individuals with a positive screening result on either the Mini-Mental State Examination (MMSE)28 or the Geriatric Mental State Schedule organic level29 underwent the Cambridge Examination for Mental Disorders of the Elderly.30 Additional information was obtained from routinely performed, in-person neuropsychological examination, and the total cohort was continuously monitored for dementia through a computerized linkage of medical records from general practitioners and the regional institute for outpatient mental health care with the study database. Available neuroimaging data were used when required for establishing a diagnosis. Basic activities of daily living and instrumental activities of daily living were routinely assessed at every center visit using the Stanford Health Assessment Questionnaire31 and Lawton Instrumental,32 respectively. Additional interview data on activities of daily living impairment during between-visit intervals was obtained from medical records. For all suspected cases of dementia, a consensus panel led by a consultant neurologist (P.J.K.) decided on the final diagnosis in accordance with standard criteria for dementia (Diagnostic and Statistical Manual of Mental Disorders, Third Edition Revised).
Because this was a population-based study, we used internal age- and sex-specific norms in the full cohort to determine the presence of cognitive dysfunction. Age- and sex-specific norms of cognitive test scores in the Rotterdam Study have previously been published.15 Cognitive dysfunction was defined as age- and sex-adjusted global cognition z scores of less than −1. Subtle motor features were defined as any parkinsonian sign during screening or Purdue Pegboard Test z score less than −1 compared with age- and sex-specific internal norms.
Statistical methods are presented in full in the eMethods in the Supplement. In short, we performed a principal component analysis of 4 tests (Stroop color word test, letter-digit substitution, verbal fluency, and word learning) to derive a global cognition score. We studied the association of cognitive function test scores with incident all-cause parkinsonism and incident probable PD using Cox proportional hazards models with adjustment for age, sex, and study subcohort.
To enable translation of our findings to clinical practice, we present likelihood ratios (LRs) for the baseline presence of isolated or combined cognitive dysfunction and subtle motor features for incident parkinsonism during follow-up. Furthermore, we consecutively subtracted 1 year of follow-up in a stepwise fashion to explore whether the LRs of combined cognitive dysfunction and subtle motor features at baseline for incident parkinsonism during follow-up varied by duration of follow-up. To determine statistical significance, we used 2-tailed tests with the threshold set to α = .05. Data analysis was conducted with SPSS, version 126.96.36.199 (IBM Corp) and R, version 3.2.4 (R Foundation).
Baseline characteristics of the study population are reported in Table 2. Individuals with at least 1 parkinsonian sign had lower Purdue Pegboard Test scores compared with other individuals (age- and sex-adjusted difference in Purdue Pegboard Test z score, 0.25 [95% CI, 0.20-0.31]; P < .001). In individuals without any parkinsonian sign, cognitive dysfunction was associated with lower scores on the Purdue Pegboard Test (age- and sex-adjusted difference in Purdue Pegboard Test z score, 0.48 [95% CI, 0.42-0.54]; P < .001).
During follow-up, 7386 participants amassed 61 660 person-years (median, 8.3 years; range, 0-15 years) and 79 (1.1%) participants received a diagnosis of incident parkinsonism. Of those with incident parkinsonism, 24 (30.4%) also received a diagnosis of incident dementia (10 before and 14 after onset of parkinsonism), and 446 (6.1%) individuals who remained free of incident parkinsonism received a diagnosis of incident dementia. The median interval between baseline cognitive assessment and detection of incident parkinsonism was 5.6 years. Fifty-four (68.4%) patients received a diagnosis of incident parkinsonism without a structured physical examination by a research physician, whereas 13 (16.5%) patients received a diagnosis without a diagnostic examination by a neurologist or geriatrician. Overall, 8 (10.1%) patients had neither type of examination; these patients received the diagnosis from a general practitioner or nursing home physician. Of patients with incident parkinsonism, 57 (72.2%) received a diagnosis of probable PD. A complete overview of clinical parkinsonism diagnoses is presented in eTable 1 in the Supplement.
Poor global cognition at baseline was associated with a higher risk of incident parkinsonism (hazard ratio [HR] per SD decrease, 1.79; 95% CI, 1.37-2.33). Interaction terms of global cognition with age (coefficient [SE], 0.015 [0.014]; P = .27), sex (coefficient [SE], 0.14 [0.23]; P = .54), and study subcohort (subcohort 2, coefficient [SE], 0.14 [0.31], P = .65; subcohort 3, coefficient [SE], 0.43 [0.37], P = .25) were not statistically significant. The association between poor cognitive functioning and incident parkinsonism remained robust beyond the first 8 years after censoring individuals with incident dementia and restricting the analysis to patients with incident parkinsonism who were examined by a neurologist or geriatrician (Figure 1). Additional adjustment for educational attainment did not affect the association between poor cognitive functioning and incident parkinsonism (HR 1.81; 95% CI, 1.37-2.38). The association slightly weakened after additional adjustment for subtle motor signs (HR, 1.69; 95% CI, 1.38-2.34) (Figure 1). The association attenuated somewhat and was no longer statistically significant after exclusion of individuals with subtle motor signs at baseline (HR, 1.33; 95% CI, 0.92-1.92). Poor cognitive functioning was also associated with probable PD (HR, 1.52; 95% CI, 1.10-2.08) and with a joint end point of probable PD or DLB (HR, 1.59; 95% CI, 1.17-2.17).
With respect to the separate cognitive function tests, letter-digit substitution test, Stroop color word test tasks 2 and 3, verbal fluency, and immediate 15-word list learning test scores were significantly associated with all-cause parkinsonism, each with an HR exceeding 1.3 per SD decrease (Table 1). For probable PD, effect estimates were generally direction consistent but somewhat attenuated and not statistically significant, with the exception of verbal fluency (Table 1).
Almost half (49.4%) of participants who received a diagnosis of incident parkinsonism during follow-up already had subtle motor features at baseline, cognitive dysfunction at baseline, or both. As reported in Table 3, individuals with cognitive dysfunction at baseline had an almost doubled risk of incident parkinsonism compared with individuals with normal cognitive functioning. Those who had both cognitive dysfunction and subtle motor signs had a 3.3-fold increased risk of incident parkinsonism at baseline. This finding translates to a moderately increased posttest probability of developing incident parkinsonism for individuals who had both features at baseline (positive LR, 2.66; 95% CI, 1.64-4.32), and the positive LR increased further after follow-up was restricted to shorter intervals (Figure 2). When a stricter threshold for cognitive dysfunction and low manual dexterity was used (−2 or less z score), the positive LR of combined cognitive dysfunction and subtle motor features at baseline also increased (Table 3).
Of 1181 individuals with cognitive dysfunction at baseline, 486 (41.2%) also had subtle motor signs at baseline, including most (14 of 22 [63.6%]) of those who received a diagnosis of both incident parkinsonism and incident dementia (eTable 2 in the Supplement). In individuals who received a diagnosis of both incident dementia and incident parkinsonism, baseline cognitive dysfunction was not associated with incident dementia (HR, 1.10; 95% CI, 0.43-2.80).
The MMSE was not included in the main analyses because it is less sensitive to subtle cognitive deficits than the other cognitive function tests. Still, lower total MMSE scores were associated with a higher risk of incident all-cause parkinsonism and nonsignificantly with incident probable PD (eTable 3 in the Supplement). The pentagon copy item of the MMSE was also associated with an increased risk of incident all-cause parkinsonism and nonsignificantly with incident probable PD. Inclusion of the pentagon copy in global cognition did not substantially affect its association with incident parkinsonism (eTable 3 in the Supplement).
In this prospective, population-based cohort, poor cognitive functioning was associated with an increased risk of parkinsonism. This association was present even after restricting the analysis to patients with incident parkinsonism without dementia. The association was also present for diverse domains of cognition, including executive, attention, cognitive speed, and memory domains. Our findings suggest that cognitive dysfunction can be considered a sign of prodromal PD.
We observed that poor cognitive functioning is prospectively associated with parkinsonism when analyses were restricted to patients with probable PD. Our data provide important insight into the association of cognitive functioning with incident parkinsonism in the general population, especially given the fact that, to our knowledge, only 2 community-based studies have previously published data on this topic.33,34 In the Honolulu-Asia Aging Study of Japanese American men, an increased risk of PD was seen across quartiles of worse executive function as well as an increased risk with slow reaction time; however, the full study results have not yet been published.33 In the Religious Orders Study and Memory and Aging Project, associations between several cognitive function tests and incident parkinsonism (without diagnosis of PD vs atypical parkinsonism) were reported in an elderly population with 5-year follow-up.34 Separately, in a cross-sectional study of parkinsonism-free individuals, participants with strong PD proxies (hyposmia and dopamine transporter binding reduction) had worse global cognition than did other participants and particularly performed worse on tasks relating to fluency, task switching, attention, and working memory.7 The observations in these 3 studies are consistent with our findings: we noted strong associations between incident parkinsonism and test scores relating to fluency and moderate associations with processing speed, reading speed, interference of automated processing and attention, and recognition from verbal memory. These results are also in line with the observation that nonamnestic impairment is more common than amnestic impairment in patients with clinical PD with mild cognitive impairment.
Associations of separate cognitive function tests with probable PD were generally direction consistent. However, with the exception of verbal fluency and the word learning recognition task, HRs were lower for probable PD than for all-cause parkinsonism. Also, most associations of separate cognitive function tests with probable PD were statistically nonsignificant; this finding may be due to the relatively small number of incident probable PD cases and the fact that the tests used were not designed specifically to detect cognitive dysfunction in PD.35
We consider 3 explanations for an observed link between poor cognitive performance and incident parkinsonism. The first would be that parkinsonism only precedes cognitive dysfunction. This is an unlikely explanation for our results. We actively screened for parkinsonism and dementia at baseline and excluded individuals with baseline parkinsonism and dementia. Furthermore, the association remained robust beyond the first 8 years, making it unlikely that our findings were driven by a delay in parkinsonism diagnosis. The second explanation would be that individuals who probably will develop parkinsonism in mid- or late life never attain a high level of cognitive functioning in early life. However, adjustment for educational level did not affect the association between cognitive functioning and incident parkinsonism in our population. Still, because educational attainment is only a proxy of early life cognitive functioning, we cannot exclude this explanation. Third, low baseline cognitive scores may indicate ongoing cognitive decline in prediagnostic patients who probably will develop parkinsonism, most of whom have prediagnostic PD. Consistent with this explanation is the high proportion (30.4%) of patients with incident parkinsonism who also received a diagnosis of incident dementia during follow-up, either before or after parkinsonism diagnosis. Almost half of the patients who had a diagnosis of incident parkinsonism during follow-up already had some cognitive dysfunction, subtle motor features at baseline, or both. These findings suggest that both motor and cognitive decline are common in patients with prodromal PD and that their sequence of occurrence is variable. Future observational studies will unravel the underlying abnormalities that drive the variable sequence of motor and cognitive decline in patients with prodromal PD.
In the Movement Disorder Society criteria for prodromal PD, cognitive loss was not included as a variable because of the absence of prospective evidence of predictive value.36 Our results, as well as those of other published studies, suggest that cognitive dysfunction now warrants inclusion as a prodromal marker. Although subtle motor abnormalities and cognitive dysfunction often occur in the same individuals, their predictive utility for incident parkinsonism only slightly overlaps. In the present study, having a global cognitive score 1 SD below the mean would be associated with a positive LR of 1.97 for parkinsonism, whereas normal cognition would translate to a negative LR of 0.86, and a combination of cognitive dysfunction and subtle motor signs was an even stronger predictor for parkinsonism. The predictive value further increased for more stringent thresholds of cognitive and subtle motor dysfunction and also for shorter prediction intervals. The correlation between subtle motor features and cognitive dysfunction will likely also be taken into account in risk estimates for the next Movement Disorder Society criteria for prodromal PD.
Before further interpreting these findings, several methodologic considerations should be noted. First, it is possible that some misclassification of the diagnosis of parkinsonism occurred. However, after restricting our analyses to patients who were examined by neurologists or geriatricians, we observed a similar overall association. Furthermore, we classified patients who had a clinical presentation consistent with PD as having probable PD irrespective of whether and when they also developed dementia. This categorization was done to be consistent with the 2015 Movement Disorder Society clinical diagnostic criteria for PD26 in which PD and DLB are also no longer considered mutually exclusive diagnoses. It is likely that many of these patients would also have met criteria for DLB once additional features developed. However, the removal of early dementia as an exclusion criterion for PD diagnosis remains a highly debated issue with opposing viewpoints.37,38
Strengths of our study include its population-based design, which ensures a representative sample of unselected patients with incident parkinsonism, active screening for parkinsonism and dementia at baseline, complete access to medical records of study participants (including specialists’ letters), and the extensive (median, 8 years) follow-up for incident parkinsonism. This study extends findings from the Rotterdam Study’s recently published nested case-control study,8 in which we showed that poor executive functioning and subtle motor signs are increasingly more common among patients with prodromal PD. Although our previous publication documented the temporal association, it could not be used to estimate risk with a single measure, which is needed to further refine diagnostic criteria for prodromal PD.36 In the present project, we used a prospective design based on a single global cognitive function measure, which allowed us to assess how common poor cognitive functioning is in patients with prodromal parkinsonism, both in isolation and in combination with subtle motor features.
Poor cognitive functioning is associated with an increased risk of developing all-cause parkinsonism and probable PD, even at intervals longer than 8 years. The association is present for diverse domains of cognition, including executive, attention, cognitive speed, and memory domains. Our findings suggest that poor cognitive functioning can be considered a prodromal sign of PD.
Accepted for Publication: May 11, 2017.
Corresponding Author: M. Arfan Ikram, MD, PhD, Department of Epidemiology, Erasmus MC University Medical Center, Wytemaweg 80, 3015 CN, Rotterdam, the Netherlands (firstname.lastname@example.org).
Published Online: September 25, 2017. doi:10.1001/jamaneurol.2017.2248
Author Contributions: Dr Darweesh had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Darweesh, M. A. Ikram.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Darweesh.
Critical revision of the manuscript for important intellectual content: Wolters, Postuma, Stricker, Hofman, Koudstaal, M. K. Ikram, M. A. Ikram.
Statistical analysis: Darweesh, Stricker.
Obtained funding: M. A. Ikram.
Administrative, technical, or material support: Stricker, M. A. Ikram.
Study supervision: Stricker, Koudstaal, M. K. Ikram, M. A. Ikram.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study received support from Stichting ParkinsonFonds. The Rotterdam Study is further supported by the Erasmus MC University Medical Center and Erasmus University Rotterdam; the Netherlands Organization for Scientific Research; the Netherlands Organization for Health Research and Development; the Research Institute for Diseases in the Elderly; the Ministry of Education, Culture, and Science; the Ministry of Health, Welfare, and Sport; the European Commission; the Netherlands Genomics Initiative; and the Municipality of Rotterdam.
Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: The contribution of the inhabitants, general practitioners, and pharmacists of the Ommoord district to the Rotterdam Study are acknowledged.
Create a personal account or sign in to: