Article Text

Risk of incident Parkinson’s disease and parkinsonism in essential tremor: a population based study
  1. J Benito-León1,
  2. E D Louis2,3,
  3. F Bermejo-Pareja1
  1. 1
    Department of Neurology, University Hospital “12 de Octubre”, Madrid, Spain
  2. 2
    GH Sergievsky Center and Department of Neurology, College of Physicians and Surgeons, Columbia University, New York, USA
  3. 3
    Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, USA
  1. Dr J Benito-León, Avda de la Constitución 73, portal 3, 7° Izquierda, E-28821 Coslada, Madrid, Spain; jbenitol{at}


Background: A longstanding literature suggests an association between essential tremor (ET) and Parkinson’s disease (PD). However, the risk of incident PD has not been quantified in cases of ET compared with controls.

Objective: To estimate the risk of incident PD in a population based cohort study of 3813 older people (including ET cases and controls) in central Spain.

Results: After a median of 3.3 years, 12 (5.8%) of 207 ET cases developed parkinsonism compared with 56 (1.6%) of 3606 controls (adjusted relative risk (RR) 3.47, 95% CI 1.82 to 6.59; p<0.001). Six (3.0%) of 201 ET cases developed incident PD versus 24 (0.7%) of 3574 controls (adjusted RR 4.27, 95% CI 1.72 to 10.61; p = 0.002).

Conclusions: Patients with ET were four times more likely than controls to develop incident PD during prospective follow-up. These data confirm and begin to quantify the link between these two diseases.

Statistics from

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.

A link between essential tremor (ET) and Parkinson’s disease (PD) has been queried in the clinical literature for some time16 yet there are no data on the risk of incident PD in prevalent ET cases compared with age matched participants without ET. We conducted a prospective, population based study of ET711 and assessed whether baseline ET is a risk factor for incident PD.


The Neurological Disorders in Central Spain (NEDICES) study is a longitudinal, population based survey of the prevalence, incidence and determinants of PD, ET, stroke and dementia.711 The study population comprised elderly subjects ⩾65 years of age taken from the census of three communities in central Spain.711 Detailed descriptions of the study population, sampling methods and study assessments have been published.711 Signed, written informed consent was obtained from all participants on enrolment. A baseline evaluation was performed between 1994 and 1995, and a follow-up evaluation between 1997 and 1998. Each evaluation included a 500 item questionnaire (demographics, medical history, lifestyle variables, three screening questions for PD (sensitivity 86.8%) and one screening question for ET (sensitivity 68.6%)).710 As documented previously, participants who screened positive underwent an in-person neurological examination which included section III of the Unified Parkinson’s Disease Rating Scale.710 As documented elsewhere,710 for subjects who could not be examined, medical records and death certificates were reviewed. ET was diagnosed using previously published diagnostic criteria7 10 (see additional material online for details).

Parkinsonism required at least two cardinal signs (resting tremor, rigidity, bradykinesia and impaired postural reflexes) in a subject not receiving anti-parkinsonian therapy or at least one sign in a patient specifically treated.12 13 Parkinsonism was further divided into three groups:

  1. PD (ie, idiopathic parkinsonism) was diagnosed, as in a previous study,9 based on the presence of two or more cardinal signs and after exclusion of other possible causes of parkinsonism.

  2. Drug induced parkinsonism: patients whose parkinsonism was attributable to medication exposure (eg, it reversed on discontinuing the medication).14

  3. All other categories of parkinsonism (non-PD parkinsonism), including vascular parkinsonism, Parkinson-plus syndromes or unspecified (ie, the exact type of non-PD parkinsonism was not certain).15 16

Incident parkinsonism was defined as a subject without baseline parkinsonism who then developed parkinsonism by the follow-up evaluation. Person-years was the time between baseline and one of three possible outcome dates: (1) date of follow-up evaluation, (2) reported date of parkinsonism onset or (3) date of death. In the Cox proportional hazards analyses, we compared the risk of developing incident parkinsonism in participants who had baseline ET compared with those without baseline ET (“controls”). We began with an unadjusted model and then, in adjusted analyses, included baseline variables that were associated with either baseline ET or incident parkinsonism (p<0.10). Martingale methods were used to check the proportional hazards assumption.17


The 3813 participants (mean age 73.4 (6.6) years) had a median duration of follow-up of 3.3 years (range 0.03–6.6 years, interquartile range 0.8 years). At baseline, the 207 participants with ET (205 (99.0%) diagnosed by direct examination and two (1.0%) by medical record review) were, on average, 1.4 years older and less educated than the 3606 participants without ET (see table 1 online). Twelve (5.8%) ET cases were taking anti-tremor medication and 78 (37.7%) reported at least one first or second degree relative with ET. None had rest tremor.

Table 1 Baseline ET and risks of different forms of parkinsonism

Overall parkinsonism

Sixty-eight (1.8%) of 3813 participants developed incident parkinsonism (30 incident PD, 22 drug induced parkinsonism, 16 incident non-PD parkinsonism) after a mean of 1.9 years (median 1.9 years; range 0.4–3.3 years). Participants who developed incident parkinsonism were older at baseline than participants who did not develop parkinsonism, and a marginally smaller proportion were current smokers, although they were similar in other respects (see table 2 online). Twelve (5.8%) of 207 ET cases developed incident parkinsonism compared with 56 (1.6%) of 3606 controls (unadjusted relative risk (RR) 3.68, p<0.001; adjusted RR 3.47, p<0.001) (table 1). Mean latency between the onset of ET and the onset of incident parkinsonism in the 12 ET cases was 8.6 (median 6, range 2–30) years. We also conducted a series of sensitivity analyses to confirm the robustness of our findings (see online material).

Incident PD

Six (3.0%) of 201 ET cases developed incident PD compared with 24 (0.7%) of 3574 controls (unadjusted RR 4.34, p = 0.001; adjusted RR 4.27, p = 0.002) (table 1, fig 1). Mean latency between the onset of ET and onset of incident PD in the six ET cases was 8.7 (median 8.0, range 5.6–12.7) years; three had a family history of ET. We also conducted several sensitivity analyses (see online material). Additional details on the six ET cases are provided online (see table 3 online).

Figure 1

Kaplan–Meier curve demonstrating the risks of surviving without incident Parkinson’s disease (PD) among participants with essential tremor (ET) compared with those without baseline ET.

Drug induced parkinsonism

In unadjusted and adjusted analyses, the risk of incident drug induced parkinsonism was 3–4-fold higher in ET cases than in controls (table 1). Of 22 participants with drug induced parkinsonism (mean age 80.3 (6.6) years; 16 (72.7%) women), 16 were taking classic neuroleptics and six other agents (eg, cinnarizine, flunarizine, clebopride). All 22 had taken these drugs for at least 6 months prior to the onset of parkinsonism and all but one received these drugs because of behavioural disorders associated with a dementia syndrome. Mean latency between the onset of ET and the onset of drug induced parkinsonism in the four ET cases was 7.7 (median 7.5, range 6–10) years.

Non-PD parkinsonism

The risk of incident non-PD parkinsonism was similar in ET cases and controls (table 1).


Patients with ET had a 3–4-fold higher risk of developing incident parkinsonism than participants without baseline ET. This included a 4–5-fold increased risk of PD and a 3–4-fold increased risk of drug induced parkinsonism. The risk of non-PD parkinsonism was not significantly increased.

A link between ET and PD has been debated in the clinical literature for some time.16 The common coexistence of the two disorders in clinics has led investigators to question whether they are associated with one another, above and beyond chance.16 The development of PD has been described in patients with ET, leading clinicians to suspect an association between the two,1 6 and ET-like action tremor is more common in PD families than in control families.2 However, the risk of incident parkinsonism and PD among prevalent ET cases has not been formally assessed until now.

Our estimates of RR may be brought into clinical settings to educate patients with ET about their risk of developing PD. Published data suggest that the lifetime risk of developing PD is 2.0% in men and 1.3% in women.18 With an adjusted RR of 4.27, one would expect a lifetime risk of developing PD of 8.5% in men with ET and 5.6% in women with ET.

The biological basis for the association between baseline ET and incident PD is not clear. The underlying pathology of ET is not well understood, but recent postmortem studies have demonstrated that approximately 25% of patients with ET have Lewy body pathology that is anatomically restricted to the locus coeruleus.1922 These postmortem studies raise the possibility that the development of PD in some of these ET cases with Lewy bodies may represent the extension of their Lewy body pathology to the substantia nigra.

We found an association between ET and incident drug induced parkinsonism. Based on a postmortem study, it has been reported that patients with drug induced parkinsonism can harbour subclinical Lewy body pathology.23 Other reports of patients with drug induced parkinsonism who develop PD also suggest a link between drug induced parkinsonism and Lewy body PD.24 As in previous studies,25 older age and female gender seemed to predispose individuals to drug induced parkinsonism.

The ET diagnoses we used were identical to those in our previous studies.7 10 There is the possibility that some of our ET cases were misdiagnosed and that they actually had PD or other types of parkinsonism at baseline rather than ET. Imaging the dopaminergic system at baseline, which we did not perform, may have addressed this issue. However, a complete Unified Parkinson’s Disease Rating Scale motor examination was conducted at baseline and at follow-up, specifically to assess subtle motor features of parkinsonism; none of our ET cases had tremor at rest or other features of parkinsonism on these examinations and none was taking medication for PD (levodopa–carbidopa, dopamine agonists or anticholinergic medications). Furthermore, among the ET cases who eventually developed incident PD, the mean latency between the onset of their ET and the onset of incident PD was more than 6 years and in some cases several decades. While action tremor may be a prominent early manifestation of PD, it is usually accompanied by other motor features of PD (eg, reduced arm swing, reduced facial expression) and is unlikely to be present in complete isolation of any other features of parkinsonism for this number of years. Finally, even when we made the conservative assumption in our sensitivity analyses that approximately 15% of ET cases had baseline PD rather than ET (ie, excluding one of six ET cases who developed PD), the RR remained significant (RR 4.32, p = 0.004).

The limitations and strengths of this study are discussed online.


The authors thank the other members of the NEDICES Study Group: S Vega, JM Morales, R Gabriel, A Portera-Sánchez, A Berbel, A Martínez-Salio, J Díaz-Guzmán, J Olazarán, J Pardo, J Porta-Etessam, F Pérez del Molino, J Rivera-Navarro, M Alonso, C Gómez, C Saiz, G Fernández, P Rodríguez and Fernando Sánchez-Sánchez. Also, J de Pedro-Cuesta, MJ Medrano, J Almazán, and the municipal authorities, family doctors, nurses, and the populations of Getafe, Lista, and Arévalo Counties.


Supplementary materials


  • Competing interests: None.

  • Funding: The NEDICES was supported by the Spanish Health Research Agency (FIS 93/0773 and 96/1993) and the Spanish Office of Science and Technology. EDL is supported by NIH R01 NS042859 and R01 NS039422 from the National Institutes of Health, Bethesda, Maryland, USA.

  • Ethics approval: Ethics approval was obtained.

  • Contributors: EDL and JB-L, developed the concept of the article, drafted the paper and conducted the statistical analysis. All authors were involved in revising the article for intellectual content.

  • ▸ Additional material and tables are published online only at