Update: Descriptive epidemiology of Parkinson disease

Abstract

We review the descriptive epidemiology of Parkinson disease (PD). PD is a prevalent neurologic disorder in high Socio-Demographic Index (SDI) nations with rising prevalence in low and middle SDI nations. PD became a prevalent disorder in high SDI nations during the 20th century. Population growth, population aging, and increased disease duration are major drivers of rising PD prevalence. Exposure to industrial toxicants may also be a contributor to rising PD prevalence. PD is an age-related disorder with incidence likely peaking in the 8th decade of life and prevalence in the 9th decade of life. PD is notable for significant sex difference in PD risk with greater risk in men. There may be ancestral differences in PD prevalence and risk. PD is associated with moderately increased mortality though this may be underestimated. Despite significant research, there is considerable uncertainty about basic features of PD epidemiology.

1. Introduction

Parkinson disease (PD) is a common and costly neurologic disorder. A recent estimate describes at least a million affected individuals in the United States with annual costs of ~$52 billion [1]. High PD prevalence is typical of high Socio-Demographic Index (SDI) nations with rising prevalence in middle and low SDI nations (see below). PD emerged as a prevalent neurologic disorder during the course of the 20th century with 19th century accounts describing PD as rare to uncommon [2]. High PD prevalence and incidence are typical of the high burden of chronic illness accompanying population growth and demographic transition secondary to modernization [3]. Recent data from the People’s Republic of China (PRC), a very populous nation completing demographic transition, suggests the existence of over 3.6 million Chinese citizens with PD [4].

It is likely that there was another significant change in Parkinsonism (Pism) epidemiology over the course of the 20th century. A significant fraction of Pism in the early-to mid-20th century was probably PostEncephalitic Parkinsonism (PEP) following the Encephalitis Lethargica (EL) outbreak at the end of World War I. EL and PEP may have been sequelae of the great H₁N₁ Influenza pandemic [5–10]. Mid-20th century data are consistent with a transient burden of PEP [11–14]. The PEP phenomenon led to Poskanzer’s and Schwab’s hypothesis that most PD was a form of PEP and that PD incidence would decline in the second half of the 20th century [15]. They were wrong, but their logic was faultless, and the PEP experience should be recalled as we follow post-COVID patients.

Assessing PD epidemiology is impeded by methodological limitations, including variation in datasets and ascertainment depth. PD epidemiology is complicated also by the clinical and pathologic heterogeneity of PD. PD is a heterogeneous aggregate with the common feature of early, prominent nigrostriatal dopaminergic neuron degeneration and consequent Pism but presently lacks sharply defined biomarkers or histopathologic signature. Genetic disorders classified as PD, Parkin (PRKN) and some Leucine Repeat Rich Kinase 2 (LRRK2) mutations, lack α-synucleinopathy. There is overlap between the intensity of nigral α-synucleinopathy in PD subjects and age-matched controls [16, 17]. PD ascertainment may be influenced by other age-related pathologies. Buchman et al. demonstrated that Pism features are common in the very elderly and often associated with non-synuclein pathologies [18,19]. PD clinical and pathologic heterogeneity is an obstacle to using epidemiologic data to uncover clues to disease etiology and pathogenesis, identify modifiable risk factors, and estimate socioeconomic disease impacts.

This narrative review focuses on summarizing descriptive epidemiology of PD, focusing on more recent studies. We used a hybrid search approach, relying older reviews to summarize older literature, and performing PubMed searches using terms such as “prevalence,” “incidence,” and “mortality.” In addition, we conducted a thorough search and screening of relevant manuscripts cited in identified articles. Through these searches, we identified greater than 400 relevant peer-reviewed articles on different aspects of PD epidemiology. Those most pertinent to descriptive epidemiology are discussed and cited.

2. Prevalence

Analyses from the Global Burden of Disease, Injuries, and Risk Factors Study (GBD) suggested that PD is the neurological disorder with the most rapid, recent increases in prevalence, disability, and deaths [20, 21]. The GBD 1990 estimate for global prevalence of PD patients was 2.5 million. The GBD 2016 estimate was 6.1 million PD patients. In a recent GBD-based update, Ou et al. estimated a 2019 global total of 8.4 million PD patients [22]. Much of this impressive increase is attributable to population aging. Identification of PD as the major neurologic disorder with the most rapid, recent prevalence increase likely also reflects declines in stroke and dementia incidence in high SDI nations.

Prevalence is a linear function of both incidence and disease duration, is influenced significantly by the age structures of studied populations, and prevalence estimates reflect ascertainment depth. In the GBD analysis, the global age-standardized prevalence rate rose by approximately 22 %, suggesting that rising PD prevalence is not due solely to population aging. The demographic transitions accompanying modernization are driven by industrialization. It is also possible that industrial toxicants are contributing significantly to rising PD prevalence. This concept has been popularized as a partially modifiable “Parkinson’s Pandemic.” [23–25] While the GBD studies are a useful point of departure for describing PD prevalence, prevalence trends, and other basic features of PD epidemiology, the GBD studies are secondary analyses of existing datasets with some reported results including modeling components. Disentangling the specific contributions of changing population age structure, changing disease duration, variable ascertainment, and potential effects of environmental exposures is challenging.

In international GBD comparisons, age-standardized prevalence increased least in high SDI nations (~9 %). Middle SDI (~60 %) and low-middle SDI (~32 %) nations exhibited the greatest rises in age-standardized prevalence with high-middle SDI and low SDI nations exhibiting intermediate age-standardized prevalence increases (~20 %) [20,21]. The greater magnitude of age-standardized prevalence increases in middle and low SDI nations could follow rising incidence secondary to increased industrial toxicant exposures, but could also be due to improved ascertainment of PD and/or increasing disease duration. Both accompany modernization and might be characteristic of nations achieving middle SDI status.

As expected with population aging as a major driver of PD prevalence, cumulative data indicates rising mean age of onset of PD. In the historic Hoehn & Yahr Columbia University clinic series, the approximate mean age of onset was 55 years [12]. Recent incidence estimates (see below) indicate highest incidence rates in the eighth decade of life and recent community-based studies of incident PD cohorts also indicate mean age of onset in eighth decade of life [26–33].

A recent GBD study on the US burden of neurologic disease suggests that the US may be a partial outlier among high SDI nations [34]. Age-standardized prevalence increased by approximately 16 % from 1990 to 2017, with a wide 95 % confidence interval (CI; 2.7%–31 %). A closer look at US data suggests caution about limitations of the GBD dataset, subsequent interpretations, and is an illustration of how varying ascertainment influences prevalence estimates. In the US GBD study, the 2017 PD total was estimated at ~550,000 (crude prevalence ~169/100,000). Marras et al. aggregated North American cohort studies to estimate prevalence in Canada and the USA [35]. Standardized to the US 2010 Census, their prevalence estimates translate to ~700,000 affected Americans in 2010 – crude prevalence ~249/100,000. Yang et al. used administrative datasets to estimate ~1,000,000 American PD patients in 2017 – crude prevalence ~307/100,000 [1]. It is possible that US PD prevalence is stable over the last generation. Strickland and Bertoni assessed PD prevalence in the US state of Nebraska during the late 1990s [36]. Nebraska established a PD registry in 1996 requiring physician and pharmacist reporting of PD patients, likely producing adequate ascertainment. Estimated Nebraska prevalence was 329/100,000. In the pioneering Copiah County (Mississippi, USA) study from the early 1980s, estimated crude prevalence was 347/100,000 [37]. In the Canadian province of Ontario, recent analysis using administrative data from 1996 to 2014 suggests rising prevalence but falling incidence and increased disease duration [38,39]. For high SDI North American nations, there is uncertainty about the relative contributions of changes in age structure, incidence, and ascertainment to estimates of PD prevalence. Uncertainty secondary to varying ascertainment is also suggested by marked transnational differences in reported prevalence among high SDI nations. A recent estimate for Israel was~500/100,000 while an estimate for the United Kingdom in 2018 was ~218/100,000 [40,41].

In a meta-analysis of mortality studies, Macleod et al. described significant increases in PD duration over the past few decades [42]. The conservative inference is that increasing PD duration reflects general improvement in later-life expectancy. Wanneveich et al. modeled PD prevalence in France from 2010 to 2030, assuming constant incidence, population aging, and increasing PD duration proportional to general improvement in later-life expectancy [43]. Projected PD age-standardized prevalence increased by 12 %. Kadastik-Eerme et al. describe a ~30 % rise in PD prevalence in regions of Estonia from the 1990s to the 2010s, but no change in incidence and rising later-life expectancy driving prevalence increases [44,45]. It’s likely that increasing PD duration is a significant contributor to high PD prevalence in high SDI nations and plausible that increasing disease duration contributes to rising PD prevalence in middle SDI nations.

Consistent with clinical experience and many individual studies, meta-analysis of prevalence studies describe increasing prevalence with age [46]. The same overall trajectory of age-related prevalence increase is seen in the recent global GBD analysis [20]. The GBD analysis extends to age 100 and prevalence peaks in the mid-80s, consistent with peaking incidence in the mid-70s (see below). Some prior epidemiologic studies describe PD as an aging-dependent disorder with continuously increasing incidence and prevalence accompanying aging [47]. The alternative is that PD is an age-dependent disorder with maximal risk focused at specific age intervals. The GBD analysis favors the age-dependent model. Myall et al. used New Zealand administrative data to address the aging-dependent versus age-dependent question. Similar to the GDB estimate, Myall et al. find a prevalence peak around age 85 [47]. Similar results were obtained by Kadastik-Eerme et al. in Estonia [44,45], and an analogous peak was seen in recent Chinese data [4]. An attractive model is that PD is an age-dependent disorder secondary to deleterious interaction of general aging effects with subpopulation-specific PD risk factors.

Sex:

The GBD prevalence analysis illustrates a common feature of PD — male predominance with a male to female ratio of approximately 1.4:1, consistent with many prior epidemiologic studies, clinical research studies, and clinical experience [20]. Moison et al. describe rising male to female prevalence ratios with advancing age [48]. In a recent meta-analysis, Zirra et al. describe a possible trend towards PD prevalence sex equality [49]. Zirra et al. reported a male to female ratio of approximately 1.2:1, with a declining ratio over that last 3 decades from approximately 2:1 to 1.15:1. Lim et al. point to Korean and Japanese epidemiologic data indicating higher PD prevalence in Korean and Japanese women than in Korean and Japanese men [50]. Female predominance was seen also in recent analyses of Chinese and Korean data [4,51]. Higher male prevalence is probably not a function of greater disease duration. Considerable data suggests that PD in women tends to be less aggressive and incidence studies indicate higher male PD incidence [52]. The basis of sex differences is unknown. An obvious candidate is gonadal steroid effects. Prior epidemiologic research produced inconsistent results [53], but recent data suggests that later menopause decreases and oophorectomy increases risk of PD [54,55]. Gonadal steroid effects are likely to be complex and could vary with PD etiology. Studies of some genetic PD variants underscore this point. Male and female LRRK2 G2019S mutation carriers appear equally likely to develop PD, but different GBA mutations may exhibit varying sex-specific risks [56].

Ancestry:

Some data suggests that ancestry affects prevalence. Pringsheim et al. suggested that standardized PD prevalence rates were lower in Asian than European-North American populations [46]. A subsequent systematic review and meta-analysis by Abbas et al. supports this inference [57]. Comparison of two high income Asian nations, Singapore and Japan, whose citizens are predominantly of Asian ancestry, with two Oceanian nations whose populations are dominated by individuals of European descent, Australia and New Zealand, suggested that Asian ancestry is associated with lower PD risk [50]. Recent analysis from ancestrally diverse New Zealand indicates lower PD prevalence among individuals of Asian ancestry [58]. Some recent studies suggest that Americans of African ancestry exhibit significantly lower PD prevalence, on the order of a 25–30 % reduction, compared with Americans of European descent [59–61]. Lower PD prevalence in Americans of African ancestry was found also in the pioneering North Manhattan study, based on careful ascertainment within a community in New York City from 1988 to 1993 [62]. In community-based studies of aging Americans, Bailey et al. compared the frequency of PD and Pism in Americans of African descent and those considered White [59]. The frequency of Pism was lower in those of African descent with a non-significant trend towards lower frequency of PD in those of African descent. In an older study of a biracial community in the US state of Mississippi, however, there was no difference in PD prevalence between individuals of European and African ancestry [37]. It is plausible that there may be ancestry-related differences in PD risk.

3. Incidence

The systematic reviews of Twelves et al. (2003) and Hirsch et al. (2016) summarize older incidence studies [63,64]. Hirsch et al. performed a meta-analysis and summarize PD incidence as 38/100,000 person years in women over the age of 40 and 61/100,000 person years in men over the age of 40. Incidence in both men and women increased approximately 35-fold from the fifth to the eighth decade of life, peaking in the eighth decade. Hirsch et al. did not provide an estimate for crude incidence. Most of the studies incorporated in their analysis were from high SDI nations and adjusting for age structure, a crude incidence rate of ~22/100,000 person years is plausible. This extrapolation is close to the estimate of Twelves et al., ~17–19/100,000 person years, close to other older estimates of crude incidence, and within the large range of incidence rates discussed by von Campenhausen et al. in their summary of European studies [65]. Willis et al. recently estimated PD incidence in the USA and Canada for the year 2012. In those 45 and over, their range of incidence rates was 47/100,000 to 77/100,00; for those 65 and over, 108/100,000 to 212/100,000 [66]. These estimates are high compared with the rates suggested by Twelves et al., but may reflect more complete ascertainments.

An unresolved issue is whether PD incidence, particularly age-standardized incidence, is changing over the last few decades. This question is most pertinent to the “Parkinson’s Pandemic” proposal. Older data from the US and Canada suggest stable incidence in recent decades [67,68]. The recent GBD study of neurologic disease trends in the USA suggests rising age-standardized incidence of PD in the USA [34]. Savica et al. estimated incident PD from 1976 to 2005 in Olmsted County, Minnesota [69]. Incidence increased in men but not in women, from ~40/100,000 person-years to ~56/100,000 person-years. Increased PD incidence was most marked in men over the age of 70. Savica et al. speculated that rising PD male incidence was linked to declining rates of smoking in men. Data from other nations returns variable results, with most studies suggesting stable to modestly declining incidence rates over the last few decades. Zheng et al. used the most recent GBD dataset to estimate PRC PD incidence over the interval from 1990 to 2019 [70]. Zheng et al. describe a relatively modest change in age-standardized incidence rates over this interval; from 13.24/100, 000 to 15.27/100,000 (and with diminishing age-standardized mortality rates). This increase may not be an actual increase in age-standardized incidence as Zheng et al. describe increased awareness of and improved diagnosis of PD over this interval. Other studies are consistent with significant declines in PD incidence. A recent systematic review and meta-analysis of Italian epidemiologic data suggests declining PD incidence [71]. Wong et al.’s analysis of Province of Ontario administrative data from 1996 to 2014 describes an incidence rate reduction of approximately 13 % [38]. Darweesh et al. compared Pism and PD incidence between the 1990 and 2000 enrollment cohorts of the community-based Rotterdam Study [72]. Follow-up at ten years revealed declines in Pism (~40 %) and PD (~60 %) incidence between these two cohorts. In a United Kingdom primary care database (The Health Improvement Network; THIN), Horsfall et al. described a 1 %/year decline in incident PD diagnoses over the period from 1999 to 2009 [73]. Goldacre et al. inferred parallel results from death certificate analysis in the United Kingdom over the period from 1979 to 2006 [74]. A recent THIN dataset analysis, covering 2002 to 2018, documents stable PD incidence [75]. In the E3N prospective study, enrolling middle-aged French women, there was stable incidence from 2002 to 2016 [76]. Isotalo et al. described a modest rise in PD incidence in Finland over the interval from 1997 to 2014 but a recent analysis of the same dataset described no changes in age-adjusted PD incidence over the past 25 years [77,78]. Evaluation of Taiwanese administrative data over recent and relatively short intervals returned conflicting results. W-M Liu et al. suggested declining age-standardized PD incidence (~18 %) over the interval from 2004 to 2011 [79], with C–C Liu et al. reporting a modest (~9 %) rise in age-standardized PD incidence over the interval from 2002 to 2009 [80]. In both Taiwanese studies, there were significantly higher relative rises in age-standardized prevalence, suggesting increased disease duration and/or better ascertainment. In combined analysis of the South Korean national insurance database and a national registry of “rare intractable diseases”, Park et al. described an 18 % increase in incidence from 2010 to 2015 [51]. In paired studies of a Japanese community, Yamawaki et al. described stable incidence from 1980 to 2004 [81]. While there may be national or regional trends in changing PD incidence, whether or not there are any recent, global trends in age-adjusted PD incidence is unclear.

4. Mortality

PD is associated with increased mortality. Important questions are the magnitude of increased mortality and whether dopamine replacement therapy (DRT) reduced mortality rates. There is little pre-DRT data. In the Columbia University clinic series, published on the threshold of the DRT era, Hoehn and Yahr estimated a standardized mortality ratio (SMR) of ~3, suggesting a marked deleterious effect of PD [12]. Roughly contemporaneous data from Olmsted County, Minnesota, however, produced a SMR of 1.6, suggesting a significant but substantially smaller effect [82].

Conventional reviews describe a range of SMRs ranging from 1.0 to 3.6 with a mean of ~1.8 [83–85]. A recent meta-analysis with strict study inclusion criteria suggested a SMR of ~2.22 [86]. The systematic review and meta-analysis of Macleod et al. describes SMRs ranging from 0.9 to 3.79 [42]. In the Macleod et al. analysis, study design impacted results. Inception cohort studies had a mean SMR of 1.52 with the mean SMR of ~2.0 in non-inception cohort studies. Inception study SMR results are consistent with data from two more recent community-based studies of incident PD cohorts. The Cambridgeshire Parkinson’s Incidence from GP to Neurologist (CamPaIGN) study reported a SMR of 1.29 and a Swedish inception cohort study reported a SMR of 1.58 [27,87]. In an Estonian inception cohort with median 5-year follow-up, the SMR was 1.12 [45]. An SMR of 1.75 (confounder adjusted) was reported in a prospective, population-based cohort study of prevalent PD within an community based study of elderly Spanish subjects (NEDICES) [88]. In CamPaIGN at 10-year follow-up, the SMR was not significantly different from a reference population, though this analysis was likely underpowered to detect a significant difference of this relatively small magnitude. Okunoye et al. recently reported a SMR of 1.14 from the United Kingdom THIN database over the interval from 2006 to 2016, a modest but significant increase compared with a reference population [89]. Duration of follow-up affects SMR estimates. Okunoye et al. noted increasing mortality with disease progression. The estimated SMR for PD subjects in a comprehensive registry of PD subjects in the Australian State of Queensland, spanning some 2 decades, was 2.75 [90]. Pinter et al. reported the final outcomes of an Austrian clinic series with 38-year follow-up and a cumulative SMR of approximately 2.0 [91]. Broadly consistent with these results is an inception cohort study based on Finnish administrative data reporting a relative case-fatality rate of 2.29. In this dataset, excess mortality was present from 1 year after diagnosis [92]. Consistent with data indicating increasing PD duration, some studies indicate declining PD mortality in recent decades [39,70,89].

The presence of common, serious co-morbid diseases in this largely elderly population likely accounts for moderate increases in SMR. In several studies, the most common proximate cause of death is cardio-vascular disease, and as expected, cancer-associated mortality is common [84,93]. It is plausible that PD as a proximate cause of death is underestimated due to death reporting practices. This conclusion is supported by the fact that pneumonia is described as a disproportionate proximate cause of death in PD patients in some studies [87].

Competing mortality effects are a function of the age at which patients experience PD onset, as younger patients with relatively fewer co-morbid disorders experience longer survival. This can be seen in a counterexample from the pioneering DATATOP trial. In 13-year follow-up of this cohort, Marras et al. reported a SMR of 1.04, a non-significant increase over the general population mortality rate [94]. With a mean age of 61 at enrollment, the DATATOP cohort was relatively young and individuals with significant co-morbid illnesses were excluded. Ishihara et al. modeled the interaction of age of onset and varying SMR levels with a Gompertz function-based model to account for increasing mortality as a function of age [85]. Individuals with younger age of PD onset exhibited the longest projected disease durations but also the largest loss of life expectancy. At a SMR of 2, for example, a patient with age of PD onset of 40 was projected to lose 7 years of life expectancy. At the same SMR, a patient with age of PD onset of 80 was projected to lose 3 years of life expectancy. This model is vindicated by recent data. In the Finnish inception cohort study of mortality, the highest case-fatality ratio was in early onset PD (onset <50 years) and the relative risk of death declined with advancing age of diagnosis [92].

Dopamine Replacement Therapy:

Whether DRT resulted in improved survival of PD patients is controversial. Macleod et al. comment that there is no convincing evidence of an effect of DRT on mortality [42]. In a retrospective analysis of data from Olmsted County, Minnesota, over the interval from 1964 to 1978 and spanning the introduction of DRT, Uitti et al. concluded there was a significant improvement in patient survival associated with DRT treatment [95]. Clarke suggested that the adoption of DRT resulted in only a transient effect on survival of PD patients [96]. If there is a DRT effect on survival, it would be most marked in the initial years of treatment (the “honeymoon period”), before the emergence of morbid DRT refractory features such as dementia. In the Queensland PD Registry, Poortvliet et al. note that PD patient mortality does not begin to diverge from the population background until approximately a decade after PD diagnosis [90]. For incident PD in the Health Professionals study, the age-adjusted relative risk of death was 1.1 at 5 years from diagnosis, 2.3 at 5–10 years from diagnosis, and 3.5 after 10 years [97]. On the other hand, recent analysis of the Norwegian ParkWest incident PD cohort indicates significantly increased mortality in the first decade after diagnosis with a hazard ratio of approximately 2.5 [98]. Interesting comparison data comes from Olmsted County, Minnesota. Savica et al. describe survival rates for several incident α-synucleinopathies — PD, PD with dementia, Dementia with Lewy bodies, and Multiple System Atrophy — over the interval from 1991 to 2010 [99]. Compared with a reference population, all exhibited increased mortality. The survival curve for PD subjects diverges from that of the reference population approximately 5 years after diagnosis, matching the “honeymoon period.” For the less treatable entities, the survival curves diverge earlier. In an analysis of American administrative data, Willis et al. reported that PD patients treated by neurologists had a lower likelihood of death [100]. One feature likely distinguishing neurologist care of PD is more aggressive DRT use [101]. The cumulative evidence suggests that DRT has a survival benefit in PD, but definitive proof is not likely to ever be available.

5. Summary (Table)

Table:

Summary of descriptive epidemiology of Parkinson disease (PD)

PD is a common neurodegenerative disorder.

PD prevalence is rising substantially across the globe secondary to population aging and increasing disease duration

Greatest future prevalence increases will occur in middle and low SDI* nations

In high SDI nations, minimum crude prevalence approximates 300/100,000

In high SDI nations, minimum crude incidence approximates 20/100,000

In high SDI nations, mean disease duration approximates 15 years

PD is an age-dependent disorder.

PD incidence rises sharply after the age of 60

PD incidence likely peaks in the mid-70s

PD prevalence likely peaks in the mid-80s

PD impact varies with sex

PD exhibits male predominance in prevalence, incidence, and disease severity with a male to female prevalence ratio of approximately 60:40

The male to female differential may be decreasing in recent decade.

In high SDI nations, PD is associated with moderate increases in relative mortality rates.

There is uncertainty as to whether age-standardized incidence of PD is changing in the past few decades. Data from high SDI nations largely suggests stable to modestly declining age-standardized incidence rates.

In high SDI nations, it is likely that disease duration increased in recent decades

There may be ancestry-based differences in PD risk.

Over the course of the 20th century, PD evolved from an uncommon disorder to a significant public health concern in high SDI nations. PD is a rapidly growing problem in middle and low SDI nations and identified recently by the World Health Organization as a world-wide concern [102]. Reported PD prevalence differs across nations, with high SDI nations exhibiting crude prevalence rates ranging between approximately 200–400/100,000. Population aging and increased disease duration are likely largely responsible for increasing prevalence. Most studies, mainly from high SDI nations, show largely unchanged incidence rates over recent years. PD incidence, however, could change markedly if there is another event such as the post-World War I outbreak of PEP. Close neurologic surveillance of COVID pandemic victims is well justified. Approximately 22/100,000 person years is a plausible minimum estimate of crude incidence rates in high SDI nations. PD is an age-related disorder with peak incidence in the 8th decade of life and peak prevalence approximately a decade later. Males exhibit higher incidence and prevalence, and PD susceptibility may vary with ancestry. PD is associated with moderate increases in mortality but its contributions to mortality may be underrated.

Despite considerable research, there are many uncertainties about PD epidemiology. There is a substantial lack of data and studies from various regions around the world including Latin America, Africa, and Southeast Asia, though a recent meta-analysis of Latin American data suggests prevalence and incidence similar to that of high SDI nations [103]. Even for high SDI nations, we don’t have consistent, reliable data about prevalence, incidence, or mortality. This is a lost opportunity as clear regional or national differences in these features might be useful in identifying significant risk factors. Good estimates of prevalence, incidence, and mortality are important for estimating the social burdens of PD and health services planning. There is uncertainty about sex and ancestry differences. Is the relative prevalence and/or incidence of PD changing between men and women? Does ancestry influence PD risk? Prior PD epidemiologic studies primarily focused on high SDI nations, but as suggested by the GBD analyses, middle and low SDI nations likely exhibit the most dynamic changes in PD epidemiology. High quality research in these nations might be particularly fruitful. A significant obstacle to characterizing basic PD epidemiology is the meta-syndromic nature of our present characterizations of PD. Use of emerging convenient biomarkers may help overcome this obstacle.