Prevalence of stroke survival in rural communities living in northern Peru

Background and purpose Stroke is the leading cause of neurological impairment in the South American Andean region. However, the epidemiology of stroke in the region has been poorly characterized. Methods We conducted a staged three-phase population-based study applying a validated eight-question neurological survey in 80 rural villages in Tumbes, northern Peru, then confirmed presence or absence of stroke through a neurologist’s examination to estimate the prevalence of stroke. Results Our survey covered 90% of the population (22,278/24,854 individuals, mean age 30±21.28, 48.45% females), and prevalence of stroke was 7.05/1,000 inhabitants. After direct standardization to WHO’s world standard population, adjusted prevalence of stroke was 6.94/1,000 inhabitants. Participants aged ≥85 years had higher stroke prevalence (>50/1000 inhabitants) compared to other stratified ages, and some unusual cases of stroke were found among individuals aged 25–34 years. The lowest age reported for a first stroke event was 16.8 years. High blood pressure (aPR 4.2 [2.7–6.4], p>0.001), and sedentary lifestyle (aPR 1.6 [1.0–2.6], p = 0.045) were more prevalent in people with stroke. Conclusions The age-standardized prevalence of stroke in this rural coastal Peruvian population was slightly higher than previously reported in studies from surrounding rural South American settings, but lower than in rural African and Asian regions. The death rate from stroke was much higher than in industrialized and middle-income countries.


Introduction
Stroke is the leading cause of neurological impairment in South America's Andean region, which includes Bolivia, Colombia, Ecuador, Peru, and Venezuela [1]. Despite this significant cerebrovascular disease burden, there is little information regarding stroke rates, mortality, or associated risk factors in rural populations [2,3].
The World Health Organization Monitoring Trends and Determinants in Cardiovascular Disease (MONICA) project does not include stroke registry information for any country in the Andean region [4]. In 1988, a population-based study conducted in Cuzco, Peru, showed an age-adjusted prevalence of 5.74 per 1,000, [5] similar to Colombia (5.6/1,000), [3] but slightly higher than Bolivia (3.22/1,000) [6]. Compared with other regions worldwide, Latin America has higher proportions of patients with hemorrhagic stroke [7], small vessel disease [2,7,8] and intracranial atherosclerotic lesions [9]. Additionally, the Andean sub-region possesses unique infectious risk factors, including Chagas' disease, neurocysticercosis, malaria, leptospirosis and viral hemorrhagic fevers, and non-infectious factors such as high altitude hypoxia and snake bites [2-5, 10, 11].
Taking advantage of a prior population-based epidemiologic survey from a cysticercosis elimination program in Tumbes, Peru, we applied a validated eight-question neurological survey to estimate stroke symptom prevalence, confirm presence of stroke through a standardized neurologic examination, and evaluate stroke-associated risk factors in this region.

Study population
We cross-sectionally surveyed 24,854 individuals from 80 communities from the cysticercosis elimination program, living near sea-level along Peru's Northern coast (Fig 1). The study area, covering 4,669.2 km2, contains a mostly Mestizo population. Life Expectance in Peru from 2015-2020 was 76,26 years old. Most villages have electricity but lack sewage facilities or running water. The area has 28 basic-level health centers; each staffed by a recently-graduated general practitioner (GP) performing a one-year rural service; one nurse and one health worker. Activities were performed in three phases.

Phase I-Stroke survey
A baseline census was performed to obtain household-level information. After obtaining informed consent (IC), non-medical field workers (trained by a team of neurologists) administered an eight-question face-to-face survey to identify stroke symptoms in all individuals older than 15 years. Illiterate individuals were included through reading aloud the IC and survey to ensure understanding. This WHO stroke detection questionnaire was translated into Spanish, modified, and validated by Del Brutto et al. in the Atahualpa Project. [12] Most older individuals were living in a relative's home.

Phase II-General practitioner evaluation
Participants reporting symptoms of stroke by responding yes to question 1 or 2, or a combination of two positive responses between questions 3 to 8, were evaluated in a local health center by a GP trained by neurologists to recognize stroke and other conditions mimicking stroke. Care was taken to include the local dialect to describe stroke, such as "derrame". The evaluation included anthropometric measures (weight, height, body mass index (BMI) calculated as weight (kg) divided by squared height (m2) [13], abdominal circumference, and brachial circumference), as well as questions regarding history of high blood pressure, diabetes, dyslipidemia, cardiac disease, smoking, alcohol or illicit drug use, and amount of exercise. All participants classified as "suspected stroke" were included in Phase III (neurological evaluation) to confirm diagnosis of stroke.

Phase III-Neurological examination, blood testing and brain computerized tomography (CT) scan
A team of board-certified neurologists evaluated suspected stroke cases for case-confirmation and to rule out non-stroke events. Stroke was defined according to WHO criteria, as "rapid development of clinical signs of focal (or global) disturbance of cerebral functions, lasting >24 hours, with no apparent cause other than vascular" [6,12,[14][15][16]. Non-contrast brain CT scans were offered to all participants with a suspected diagnosis of stroke and performed using a helicoid CT scan (Siemens AG, Germany) in the Center for Global Health facility. Reproductiveaged women had a urine pregnancy test performed prior to imaging, and pregnant women did not undergo brain CT. Venous blood samples (8ml) were collected from stroke cases to assess: fasting blood glucose (FBG), glycosylated hemoglobin (HbA1c), serum lipid profiles including total cholesterol (TC), triglyceride (TG), The total cholesterol/HDL cholesterol ratio (American Heart Association [AHA] [17] target below 5 in males, and below 4.4 in females), the LDL cholesterol/HDL cholesterol ratios (AHA target below 3.5 in males, and below 3.2 in females) [17] and hematocrit.
Hypertension was defined as mean systolic blood pressure (SBP) �140mmHg or mean diastolic blood pressure (DBP) �90mmHg [13], diabetes mellitus (DM) was diagnosed as FBG �126 mg/dl or HbA1c �6.5% [18] and/or self-reported diagnosis during the GP evaluation. Heart disease history was self-reported during the GP evaluation. Participants were asked whether they regularly used tobacco (current smoking defined as �1 cigarette/day) [13], consumed alcohol (any alcoholic drink in the last week), sedentary lifestyle (office work, driving as a chauffeur, and sitting while watching television or low physical activities) [19].

Ethical considerations
The study protocol and consent forms were approved by the institutional review boards of Universidad Peruana Cayetano Heredia, the University of Washington and the Regional Directorate of the Ministry of Health (DIRESA) in Tumbes.

Statistical analysis
Prevalence of stroke was defined as number of persons with confirmed stroke divided by number of baseline survey respondents. Age-adjusted stroke prevalence was obtained by dividing the number of people with stroke in each age bracket by the number of individuals in the same age bracket as stratified in the World (WHO 2000-2025) standard population. Incidence was defined as number of persons who developed stroke in the year preceding the prevalence day (April 5, 2011), divided by total number of study participants. The world (WHO 2000-2025) standard population census was the reference population for prevalence age adjustments. Confidence intervals were estimated using exact binomial method. Prevalence ratios (PR) and adjusted PR (aPR) were estimated using Poisson family general linear models with logit link. All reported probability (p) values were two-sided with significance level set at 0.05. Variables significant at the level of p<0.25 were retained in the multivariable models from which adjusted odds ratios were estimated. All-cause mortality was defined as total number of deaths during the study period (between phase I and III; April 2011-May 2012), divided by participant population. We used Stata version 14.2 for statistical analysis (Stata Corp., College Station, TX, USA).

Results
From a population of 34,825 people across 80 rural communities, 24,854 individuals older than 15 years were invited to participate; 22,278 (89.63%) individuals provided informed consent and completed the eight-question stroke survey. The participants most frequently lived in hand-constructed mud and cane houses, had public potable water service that was functional for a few hours each day, treated water with chlorine pills, and defecated in holes/silos; although a sizable proportion (24.2%) reported open field defecation.

Comparison between positive and negative survey respondents
The proportion of positive respondents to the survey was 7

Prevalence of stroke survival
A total of 1,420 individuals were evaluated by a study physician (including two negative survey respondents self-referred by study physicians due to symptoms compatible with stroke). We classified these 318/1,420 (2.2%) as "suspected of stroke", and 1,102 had a diagnosis other than stroke (Fig 2).
Of the 318 participants with suspected stroke, 24 (7.5%) declined the neurologist evaluation, 137 had a diagnosis other than stroke, and 157 individuals were confirmed as having had a stroke event. Stroke survival prevalence was 7.05/1,000 (157/22,278) inhabitants. After direct standardization to WHO's world standard population, the adjusted prevalence of stroke was 6.94/1,000 inhabitants. Participants aged 85 years and older had much higher prevalence (>50/1,000 inhabitants) compared with other stratified ages, and some unusual cases were found in individuals aged 25-34 years (Fig 3). Crude prevalence was similar for men and women (81/11,204, 7.23 per 1,000 men vs 76/11,074, 6.86 per 1,000 women; p = 0.636) ( Table 2).

Other neurological conditions
Other relevant neurological diagnoses in non-stroke cases included epilepsy in 84 participants, Bell's palsy in 152 participants, transient ischemic attack (TIA) in 37 participants, 2 cases with sequelae of Guillain-Barre, and 2 cases of amyotrophic lateral sclerosis.

Retrospective stroke incidence
Eighteen stroke events occurred in the year before the survey, for a crude annual incidence rate of 80.80 per 100,000 person/years. The incidence rate adjusted to 2007 census age distribution [20] was 129.72 per 100,000 person/years.

Clinical sub-types of stroke and brain CT findings
From the 157 PWS, clinical history of ischemic stroke was found in 139 (88.5%; 66 females and 73 males, mean age 66, median age 68, range of time-lapse of event between 0-49 years, average 7.10 years) and hemorrhagic stroke in 21 (11.6%; 12 females and 9 males, mean age 63, median age 61, range of time-lapse of event between 1-25 years, average 9.33 years). Three (1.9%) individuals with ischemic stroke later developed hemorrhagic stroke.

Morbidity and characteristics of PWS
Differences in self-reported cardiovascular information between PWS (n = 157) and people without stroke (PWOS, n = 1,102) included non-modifiable factors such as gender and age,   (Table 3). strategy for encouraging physical activity and help decrease high blood pressure. Improving the patient-health center relationship is essential for empowering the patient on the importance of effective treatment compliance and lifestyle changes to control hypertension and avoid a stroke event.
Twenty-five of 95 deaths reported in our study were due to post-stroke complications. Although our study was not designed to evaluate mortality, our crude mortality rate (1,576/ 100,000 deaths per year) in people suspected of stroke was extremely high compared to mortality rates from stroke in China (159/100,000 deaths per year) [28] or the US (37.6/100,000 deaths per year) [39]. Low socioeconomic and educational levels and social inequality have been associated with higher mortality rates in other poor regions of Latin America and the Caribbean [2].
This study supported Saposnik and Del Brutto's hypothesis that a high incidence of stroke is associated with a high risk of mortality during the acute stroke phase [3]. Although a significant proportion of the rural population had access to free health care through the Peruvian government's SIS (Sistema Integral de Salud) program, many PWS sought initial evaluation through "sobadores" (people who massage the part of the body affected by a stroke) and only later sought traditional care at the hospital outpatient clinic. This particular behavior has been found in other neurological research in this area [40]. This and other cultural idiosyncrasies may delay seeking medical treatment, and future research should examine how improving the patient-health-care provider interaction could reduce such delay. A comprehensive primarycare training campaign to improve education for patients regarding stroke symptoms, prevention, and treatment, as well as optimal cardiovascular health measures, might encourage future stroke patients to seek their first care through hospitals.
Our study's strengths included the support of 20 GPs from local health centers to recruit and enroll patients, and 8 neurologists to confirm stroke diagnoses. We demonstrated that it is possible to apply the WHO's suggestions and accomplish stroke detection at the primary-care level, where the main health response is primary-care and prevention [41]. This strategy could be replicable by chronic disease programs inside other ministries of health.
Our study had some limitations. Neuroimaging was limited to CT, which is less sensitive than MRI for detecting stroke. Also, we did not perform EKG to assess for cardiac pathology. It is possible that stroke was under-represented as a cause of death, as "sudden deaths" caused by cardiovascular or cerebrovascular disease, are not verified by pathologic evaluation and are often reported as "cardiorespiratory arrest" on death certificates [42]. Epidemiological research with standardized methodology to identify factors associated with high mortality rates in stroke is urgently needed for a more accurate assessment of the burden and characteristics of stroke mortality in Peru.

Conclusion
Stroke prevalence at sea level in Peru was higher than prior epidemiological studies from South America, but lower than in industrialized countries where the decrease in mortality of stroke has increased survival. High mortality in PWS compared to other rural settings could be due to a lack of care-seeking at primary care levels. Although stroke is the main cause for disability, it will remain a neglected chronic disease, especially in rural settings, until health programs increase services for cardiovascular health, stroke prevention, treatment, disability, and post-stroke rehabilitation.