Prevalence of Parkinson’s Disease in a Town of the Central Highlands of Peru: A Population-Based Study

Abstract

Background:

The global prevalence of Parkinson’s disease (PD) is estimated to be 315 per 100,000 inhabitants, with variations across regions and depending on the epidemiological methods used. Door-to-door prevalence studies in Andean communities are notably underrepresented in the literature.

Objective:

The aim of this study was to estimate the PD prevalence in Jauja.

Methods:

A cross-sectional community-based door-to-door prevalence study was conducted from April 2018 to September 2019. We used a Spanish version of the 9-item screening questionnaire (9-item SQ). Trained interviewers conducted face-to-face screening surveys. Participants presenting “suspected parkinsonism” underwent further neurological assessment. PD cases responded to the LARGE-PD questionnaire and “pesticide exposure questionnaire.” We calculated the crude prevalence, gender- and age-specific rates, and age-adjusted rates.

Results:

Out of 2527 eligible individuals, 1960 completed the screening survey. Of these, 563 were selected for neurological assessment, and 337 underwent neurological examination. PD was confirmed in 21 individuals, 16 of whom were identified using the 9-item SQ with a crude prevalence of 816/100,000. We observed higher rates of PD in older age groups and in males compared to females (1.6 vs. 0.3%, P < 0.05). The age-standardized prevalence was 566/100,000 (95% confidence interval [CI]: 496.6–635.2). Among PD cases, the most common environmental exposures were living near farming fields and pesticide exposure.

Conclusions:

The observed prevalence of PD in this town in the central highlands of Peru aligns with previous reports, with notably higher rates among individuals over the age of 80. Further prospective studies are warranted to investigate the potential influence of environmental factors on PD prevalence in this region.

Parkinson’s disease (PD) is the second most prevalent neurodegenerative disorder worldwide, and its global burden has increased substantially in the past few decades.¹ The global prevalence of PD is estimated at 315 per 100,000 inhabitants, with higher prevalence rates in Europe and the Americas.² These differences in prevalence rates arise from the combination of sociodemographic factors, diverse environmental exposures, genetic background, and variations in methodological approaches.² The most frequently used approaches for estimating the prevalence of PD are screening of clinical records, drug tracing, and community-based door-to-door surveys.²

The Latin American region is made up of emerging economies with significant disparities in healthcare services, and approximately 660 million inhabitants have a variable admixture of Amerindian, European, and African continental ancestries.³ PD prevalence studies are limited in the Latin American region, with prevalence rates having been reported in 13 out of 33 countries from Latin America and the Caribbean.⁴ Door-to-door prevalence studies have been conducted in the Andean countries, due to geographic, socioeconomic, and cultural factors.^5-7

Peru is a Latin American country with about 33 million inhabitants, who reside in a complex territory with an extensive Pacific Ocean coastline, the Andes highlands, and the Amazonian jungle.⁸ The communities living in the Peruvian highlands are made up of a rural population with farming-related activities as the main occupation and a high exposure to pesticides,⁹ an environmental risk factor consistently reported for PD.¹⁰ Clinical observations from local physicians raised the question regarding PD prevalence in central highlands, supported by a previous door-to-door study conducted in 2008 in the central Andes of Peru, which reported high rates of PD.¹¹ Additionally, limited access to healthcare services has restricted their participation in epidemiological studies, resulting in their historical underrepresentation. For these reasons, we conducted a two-phase door-to-door prevalence study in Jauja, a town located in the central Andes highlands of Peru.

Methods

Setting

The town of Jauja is situated at 3340 meters above sea level and is one of the 34 districts in the province of the same name, located on the slope of the Mantaro Valley in the central highlands of Peru (Fig. 1). According to the last 2017 population census developed by the Instituto Nacional de Estadistica e Informatica (INEI, 2017),¹² the town of Jauja has a total population of 17,908 inhabitants, with a predominance of women (52%). Livestock and agriculture are the main production activities in the region, with potato, corn, bananas, barley, and cassava being the main transitory crops, whereas coffee, orange, cocoa, and avocado are among the permanent crops. Domestic and street commerce have proliferated in recent years, although there are still no data available on these informal activities.¹³ Jauja is the only town in the central highlands of Peru with proper transportation access, including a local airport. The healthcare system in the town is fragmented, with two main hospitals and a limited-resource network of primary care centers. Neurological diseases are managed by internal medicine and psychiatry specialists, as there are no neurologists working in the region.

Fig. 1.

Map of Jauja.

Sample Size and Sampling

The study sample size was estimated using the following parameters: (a) population of 12,434 inhabitants over 18 years of age, (b) 0.335% margin of error, and (c) 95% confidence intervals (CI). The reference value of the prevalence rate of 0.671% was obtained from a prevalence study conducted in 2003 in 5 districts of the central highlands of Peru.¹¹ We obtained a sample size of 1928 individuals for the study. The sampling strategy combined both random walk and cluster approaches, using 14 different clusters consistent with the same number of known neighborhoods of the town of Jauja. Sampling was designed to select approximately 15% of the eligible participants for each cluster.

Study Design and Participants

A cross-sectional community-based door-to-door prevalence study was conducted from April 2018 to September 2019. The study was approved by the Research Ethics Committee of Instituto Nacional de Ciencias Neurologicas and the University of Washington IRB committee.

Residents were eligible if they were aged 18 or older and had been living in Jauja for at least 9 months. Individuals who had not been living permanently for at least 9 months during that year were excluded. The study was conducted in three phases as follows:

Phase 0: Preparation

After the IRB approval was obtained, we organized meetings with the local community, including the main public hospital, the mayor’s office, and representatives from the 14 neighborhoods of the district. During these meetings, we introduced the main objectives of the study and requested support with public announcements through local radio and newspapers. We utilized the 9-item screening questionnaire (9-item SQ) developed by Tanner et al.¹⁴ This questionnaire was translated and validated in the Spanish population, demonstrating almost 100% sensitivity and specificity.¹⁵ We piloted the 9-item SQ with 12 healthy individuals to assess comprehension of the wording, adapting three questions to use more commonly understood words within the local context. This process resulted in a revised Spanish version of the 9-item SQ (Fig. S1). Subsequently, we conducted two training sessions lasting 3 hours each for both local researchers and interviewers, who were selected from the community beforehand. The map delineating the districts was obtained upon request from the local government office.

Phase 1: Screening

During phase 1, 10 trained non-physician interviewers conducted face-to-face surveys of Jauja residents over a 9-month period. During each door-to-door visit, residents were interviewed using two strategies: (1) an open-ended question: “Do you know anyone with Parkinson’s disease?” and (2) the adapted version of the 9-item SQ designed for screening parkinsonism. If individuals were unable to respond on their own, family members answered the questions. A positive response to at least 1 of the 9 questions classified the participant as having “suspected parkinsonism” status (Fig. 1). The 9-item SQ was administered door-to-door to participants and commenced 73 weeks before the prevalence assessment date. Up to 2 visits were made at different times of the day or on different days to ensure the participation of eligible individuals. Participants identified as having “suspected parkinsonism” were invited to participate in phase 2. Additionally, we included the open-ended question, “Do you know anyone else with Parkinson’s disease?,” in the survey as part of the strategy to increase the chance of identifying PD cases. However, cases identified through this question were not included in the prevalence estimation.

Phase 2: Defining Diagnosis

Participants who responded positively to at least 1 question of the 9-item SQ during phase 1 were further examined by trained neurologists with expertise in movement disorders (MC-O, ES-C, AR-V). The diagnosis of PD was made according to the Queen Square Brain Bank criteria.¹⁶

Sociodemographic, Clinical, and Genetic Assessments

Each participant identified as a PD case was asked to complete the LARGE-PD questionnaire consisting of 97 questions, including demographic data, clinical symptoms, clinical scores to assess the PD severity, including modified Hoehn and Yahr (H&Y),¹⁷ Movement Disorder Society-Sponsored Revision of the Unified Parkinson’s Disease Rating Scale Part III (MDS-UPDRS III)¹⁸ and the Montreal Cognitive Assessment (MoCA),¹⁹ family history, and environmental exposure,²⁰ and those reporting a current or previous farming occupation also completed the pesticide exposure questionnaire, consisting of 91 questions.²¹ All identified PD participants underwent genetic testing as part of a larger genetics association study across Latin America, performed through a Neurobooster microarray²² that screened at least for 7 PD risk genes including LRRK2, GBA, PRKN, PARK7, SNCA, VPS35, and PINK1.

Statistical Methods

We defined prevalence as the number of participants affected with PD divided by the number of total participants in the population sample recruited on phase 1, including responders and nonresponders. We reported crude prevalence and 95% CIs and calculated gender- and age-specific rates. Age-specific rates were stratified into 6 different groups: 18–39, 40–49, 50–59, 60–69, 70–79, and ≥ 80 years old. We used descriptive statistics to summarize the demographic and clinical characteristics of individuals diagnosed with PD. Categorical variables were reported as counts and percentages and continuous variables as medians (interquartile range [IQR]). Additionally, we calculated age-adjusted rates and 95% CI based on the 2017 Peruvian census.¹² The statistical analyses, figures, and tables were conducted using R software version 4.4.1.

Results

Workflow and Population Characteristics

Out of the 2527 eligible individuals, 1960 completed the screening survey (77.6% of participation rate in phase 1). The remaining 567 were excluded from the study due to refusal to participate, death, or incomplete forms. Of the 1960 individuals who completed the screening survey, 563 (28.7%) responded positively to at least 1 question. Of the 563 individuals identified in phase 1, 337 were neurologically examined (respondents), and 226 were unavailable for neurological examination (nonrespondents). The response rate in phase 2 was 60%. We compared the age and gender distribution between the respondents and the nonrespondents, with minor differences in age, with respondents being slightly older compared to nonrespondents (Table 1). Gender did not differ between the 2 groups. The complete study workflow and participant selection are summarized in Figure 2.

TABLE 1.

Comparison between responds and nonrespondents

	Nonrespondents (N = 226)	Respondents (N = 337)	Total (N = 563)	P-value
Age				< 0.001
Median [IQR]	51 [40–67]	60 [44–73]	55 [42–71]
Mean (SD)	52.55 (18.17)	57.79 (18.32)	55.69 (18.42)
Gender, no. (%)				0.594
Female	139 (61.5%)	215 (63.8%)	354 (62.9%)
Male	87 (38.5%)	122 (36.2%)	209 (37.1%)

Fig. 2.

Flowchart of study workflow and participants’ selection.

Prevalence Estimates

The diagnosis of PD was confirmed in 21 individuals. Of these, 16 were identified through the 9-item SQ, resulting in a crude prevalence of 816 (95% CI: 483.7–1353.4) per 100,000. Additionally, we identified 5 PD cases, including 1 early-onset case, with the open-ended question: “Do you know anyone with Parkinson’s disease?”. We further estimated prevalence rates by age and gender segments, revealing higher rates in older age groups and a higher frequency in males compared to females (1.6% vs. 0.3%, P = 0.001). The age- and gender-specific crude prevalence rates of PD are shown in Table 2. Additionally, we found an age-standardized prevalence of 566 per 100,000 (95% CI: 196.2–1878.7), standardized to the 2017 Peruvian census.

TABLE 2.

Age- and gender-specific prevalence of Parkinson’s disease

	Population, no.	Cases, no.	Prevalence/100,000 [95% CI]	P-value
Age (yr)
18–39	691	0	NA	<0.001a
40–49	400	1	250 [13.1–1607.4]
50–59	340	0	NA
60–69	254	1	393.7 [20.6–2516.9]
70–79	194	7	3608.2 [1590.2 – 7593.5]
>80	81	7	8642 [3840.0 – 17,541.7]
Gender
Female	1165	3	257.5 [66.5–817.5]	0.001b
Male	795	13	1635.2 [911.6–2856.0]
Crude prevalence	1960	16	816.3 [483.7–1353.4]

^aFisher’s exact test for count data (adjusted for multiple comparisons).

^bFisher’s exact test for count data.

PD Cases

Among the 21 PD cases identified, 10 (47.6%) were newly diagnosed during the study (7 of whom were identified with the door-to-door strategy). The median age was 78 years [IQR, 73–85], with a median age at onset of 75.0 [IQR, 68–82 and a median disease duration of 2 [IQR, 2–8] years. About 71% of PD cases reported tremor as the most frequent initial symptom, with a median MDS-UPDRS III score of 40 [IQR, 27.5–50.8] and H&Y scoring ranging from 1 to 4.5. Only 15 out of the 21 cases completed the MoCA cognitive screening scoring 12.5 (9.3–16.8). Among the previously diagnosed patients (52.4%), all were receiving symptom-relieving drugs, including levodopa/ carbidopa alone in 7 cases and levodopa/carbidopa with biperiden in 4 cases. Seven cases (35%) reported a positive family history of PD. Notably, in 2 out of the 10 de novo patients, PD onset was before 50 years of age. Although almost all the PD cases identified (20 cases) lived near farming fields, only 7 declared active or previous exposure to pesticides, with 4 reporting farming as their major occupation. The main pesticides or fertilizers used were organophosphorus (parathion), urea, and nitrate. Agricultural products included tubers (potato, olluco, and mashua), corn, beans, peas, and barley. The clinical and sociodemographic characteristics of the cases identified are shown in Table 3. Additionally, 26 other neurodegenerative disorders (half of which presented isolated postural predominant tremor) were also identified (Fig. 2). Additionally, microarray genotyping was performed in 16 out of 21 cases to screen for major PD-risk genes. No pathogenic variants were identified in any of the tested individuals.

TABLE 3.

Characteristics of individuals diagnosed with Parkinson’s disease

	Total (N = 21a)
Demographics and past medical history
Age at prevalence estimation, years, median [IQR]	78.0 [73.0–85.0]
Female gender, no. (%)	6 (28.6%)
Years of education, median [IQR]b	10.0 [6.0–11.3]
Previous diagnosis of PD, no. (%)	11 (52.4%)
Previous medications, no. (%)c
Levodopa	7 (63.6%)
Levodopa plus biperiden	4 (36.4%)
Family history of PD, no. (%)d	7 (35%)
Current disease
Time since diagnosis, years, median [IQR]	2.0 [2.0–8.0]
Age at onset, years, median [IQR]	75.0 [68.0–82.0]
Initial symptom, no. (%)
Bradykinesia	2 (9.5%)
Rigidity	4 (19%)
Tremor	15 (71.4%)
Initial symptom location, no. (%)
Right lower extremity	1 (4.8%)
Left lower extremity	1 (4.8%)
Right upper extremity	12 (57.1%)
Left upper extremity	7 (33.3%)
Hoehn and Yahr score, no. (%)
Stage 1	1 (4.8%)
Stage 1.5	3 (14.3%)
Stage 2	3 (14.3%)
Stage 2.5	5 (23.8%)
Stage 3	6 (28.6%)
Stage 4	2 (9.5%)
Stage 4.5	1 (4.8%)
MDS-UPDRS III, median [IQR]	40.0 [27.5–50.8]
MoCA score, median [IQR]e	12.5 [9.3–16.8]
Risk factors
Tobacco, no. (%)	7 (33.3%)
NSAIDs, no. (%)	4 (19.0%)
Caffeine, no. (%)	19 (90.5%)
Proximity to crops, no. (%)	20 (95.2%)
Water source, no. (%)
City	4 (19.0%)
Springs	1 (4.8%)
Springs and city water	7 (33.3%)
Springs, well, and city water	1 (4.8%)
Well and city water	6 (28.6%)
River and city water	2 (9.5%)
Alcohol consumption, no. (%)	13 (61.9%)
Previous cranioencephalic trauma, no. (%)	4 (19%)
Occupation, no. (%)
Farm worker	4 (19%)
Football referee	1 (4.8%)
Housewife	2 (9.5%)
Driver	2 (9.5%)
Cook	1 (4.8%)
Salesperson	3 (14.3%)
Construction	1 (4.8%)
Teacher	2 (9.5%)
Cattle rancher	1 (4.8%)
Miner	1 (4.8%)
Baker	1 (4.8%)
Painter	2 (9.5%)
Exposure to heavy metals, no. (%)	5 (23.8%)
Exposure to pesticides, no. (%)	7 (33.3%)

Abbreviations: IQR, interquartile range; PD, Parkinson’s disease; MDS-UPDRS III, Movement Disorder Society-Sponsored Revision of the Unified Parkinson’s Disease Rating Scale Part III; MoCA, Montreal Cognitive Assessment; NSAIDs, nonsteroidal anti-inflammatory drugs.

^aSixteen PD cases were identified through door-to-door strategy, and we identified 5 PD cases more with an open-ended question. Missing data.

^bn = 1.

^cn = 10.

^dn = 1.

^en = 7.

Discussion

We conducted a community-based study to estimate the prevalence of PD in a town in the central highlands of Peru. With the use of a 2-phase door-to-door approach we calculated the crude prevalence of PD to be 816 per 100,000 for individuals aged 18 years or above, similar to rates reported in 2 previous meta-analyses, ranging from 427 to 1794 per 100,000.^4,23 When adjusting the crude prevalence for individuals aged 65 or above, these estimates go up to 3.7% (2.1–6.2), which is slightly higher than the overall rate reported in a recent multinational Latin American study (2.0%, 1.7–2.3), including Peru, also obtained from door-to-door surveys.²⁴ These contrasting differences may be explained by the fact that the estimates in this multinational study were obtained from cities on the Peruvian coast (Lima and Cañete), and PD diagnosis was not established by movement disorder specialist, which may have resulted in inaccurate reports.²⁴

Standardized rates are used to compare the PD prevalence of 2 or more populations. Overall, our age-standardized prevalence rate of 566 (196.2–1878.7) per 100,000 is comparable to previous global prevalence rates of up to 440.3 per 100,000 reported for Western and Eastern countries²⁵; however, it is much higher than the rates reported by the Global Burden of Disease (GBD) in 2019, ranging from 86.7 to 115.7 per 100,000 in Andean Latin America.²⁶ These findings could be due to methodological and sociocultural factors, and the interplay between aging, genetic, and environmental factors may also play an important role. The estimated prevalence in our study may be related to the concurrence of many social determinants of health and known environmental exposures, including the common use of pesticides in crops fields,²⁷ the recurrent evidence of high concentrations of heavy metals in the river due to mining contamination, and the limited access to potable water.^28,29 However, in our study, environmental exposures are presented in a purely descriptive and exploratory manner. We intentionally avoid drawing causal inferences or making regional comparisons with prior PD prevalence studies due to the small sample size and absence of corresponding environmental data from healthy control subjects. Furthermore, we must keep in mind that PD is a complex disorder, and environmental factors alone have a weak association with PD.³⁰

We found that the prevalence of PD in this region was higher among participants aged 80 years and above. Although older publications reported decreased prevalence rates among people aged 80 years or above,^31,32 recent large epidemiological studies have identified aging as the greatest risk factor for PD, showing ascendant curves across lifetime, particularly among those aged 80 and above.³³ The rate of elderly population over 80 years in the Jauja region is a bit higher than in the overall Peruvian population (2.1% vs. 1.9%).¹² It is then possible that the high PD prevalence rate among participants over 80 years old may be explained, at least in part, due to the increased life expectancy in this region.

The high rate of newly diagnosed PD cases (~50% of the PD participants) is related to a combination of health system–related issues and sociocultural beliefs and behaviors. The higher sensitivity for identifying newly diagnosed cases is a well-known advantage of the door-to-door approach methodology, as evidenced by a study conducted in Brazil, where 72% of the identified PD patients were de novo cases.³⁴ Many family members of PD cases attributed symptoms such as tremor, rigidity, or bradykinesia to the natural aging process rather than a neurological disease. Local health behaviors favored the use of traditional medicine, including the use of medicinal plants, special beverage preparations, and other unconventional treatments.³⁵ Primary care centers are primarily oriented to manage acute diseases and face tremendous infrastructure and equipment constraints.⁸ In addition, given the absence of local neurologists, neurodegenerative disorders, such as PD, are usually managed by primary care physicians and internal medicine specialists. The nearest neurology specialist is located in the closest city, approximately 1-hour away by bus.

The clinical features of the PD cases identified were mostly concordant with a typical PD diagnosis, albeit with some relevant differences. Most PD cases reported tremor as the most frequent initial symptom, with MDS-UPDRS III scores and H&Y stages mostly consistent with moderate PD. The mean MoCA scores suggest significant cognitive impairment, potentially indicative of Parkinson’s disease dementia (PDD).³⁶ However, as this finding is based solely on a screening tool, commonly employed for the initial assessment of cognitive function in population-based studies, it should be interpreted with caution. Further comprehensive neuropsychological evaluation is warranted, considering the complex interplay between dysexecutive and visuospatial deficits—hallmarks of cognitive dysfunction in PD—and the well-documented impact of education level and the floor effect on MoCA performance, as evidenced even in cognitively healthy individuals within a Peruvian cohort.^19,37 Despite having a relatively low male to female proportion (0.7:1) of suspected cases during phase 1, we obtained an extremely high male to female prevalence ratio (6.4:1) compared to previous reports.³⁸ The higher frequency of men with farming-related occupations, gender differences in access to care, gender differences for tobacco and caffeine consumption, among other sociocultural factors, might explain this difference.³⁸

Local intercultural challenges increased the difficulty in carrying out this community-based study. We faced difficulties in contacting participants due to mostly unpredictable work schedules, with some available in the morning and others in the afternoon and always modified depending on local festivals. Many houses were difficult to locate due to chaotic and incomplete numbering systems. Furthermore, street dogs are extremely common in Jauja, posing difficult access to some places due to safety concerns. We also observed some global mistrust toward foreigners, especially when discussing signing papers such as informed consent. This mistrust stemmed from previous community-based political and social activities that exploited the community with the sole intention of obtaining signatures. Despite efforts during the preparation phase, including communication with local authorities and neighborhood leaders, extensive radio broadcasts, and even the utilization of a garbage truck loudspeaker system, the high level of mistrust toward foreigners and reluctance to sign documents among residents negatively affected the response rate.

Our study has limitations, mostly related to the door-to-door approach. Limitation on inference analysis for environmental exposures, selection bias, low rate of respondents, and increased logistic costs are some factors that might negatively affect results when using this strategy.² The major estimation bias of our study may be the underestimation of the true prevalence, given the possibility to have missed PD cases among nonresponders on phase 2. Despite having a robust sample size to address the main outcome of our study—PD prevalence estimation—the limited number of identified PD cases did not allow inference analysis on environmental risk factors. A lower response rate was observed in phase 2 (60%) despite using mitigating strategies, such as meetings with local community representatives to share the study plan and to get them actively involved in the two-visit study, access to local communication resources such as radio and advertisements through large-range speakers, extensive training sessions with the interviewers emphasizing strategies to secure participants engagement for both phase 1 and phase 2 visits, and conducting phase 2 evaluations through interviews at medical centers as well as home visits. Across the entire sample, no significant gender differences were observed between respondents and nonrespondents. We found mild age differences, with nonrespondents being slightly younger than respondents. We acknowledge that the higher nonresponse rates among younger individuals, as consistently observed in survey-based studies, may lead to biased prevalence estimates.³⁹ We did not conduct a comparative analysis of nonresponders across clusters due to logistical limitations during community-based fieldwork. Furthermore, a cluster sampling approach may lead to limited internal homogeneity within the clusters themselves.

In conclusion, using a door-to-door, two-phase community-based approach, we identified a PD prevalence rate in the town of Jauja that is consistent with previous reports from the Andean region of Latin America, particularly among individuals over the age of 80. The clinical aspects were mostly consistent with previous studies performed in other communities, with a strong male predominance among the PD patients identified. Environmental exposures, such as rural work habits, were the most frequently reported; however, this finding should be cautiously interpreted given the small number of identified PD cases and the absence of non-PD reference group. Further prospective studies are required to explore the association between environmental exposure and the prevalence of PD in this region. Future studies performed in rural or Andean communities should include a comprehensive socioeconomic and cultural analysis of the regions to overcome mistrust in these communities to ensure a better participation rate. Greater awareness of PD by the primary care system in the highland regions of Peru is needed.

Supplementary Material

survey

Figure S1. Nine-item screening questionnaire.

Supporting information may be found in the online version of this article.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.