Dengue is a viral infection with both endemic and epidemic transmission cycles and has an estimated global incidence of 390 million cases per year, of which 96 million manifest symptoms of any severity1. The global burden of dengue is high, with 9,221 estimated deaths annually, resulting in 576,900 years of life lost to premature mortality in 20132. In Brazil, the incidence was 813.1 cases per 100,000 inhabitants in 20153. Workforce absenteeism caused by dengue produced a total loss of approximately 260 million dollars in 20134.
Due to its environmental characteristics and inadequate sanitary and living conditions, an increase in the number of dengue cases was observed in Brazil5–7. The management, planning, and execution of dengue control policies is one of the competencies of the national, state, and municipal spheres of the Brazilian National Health System (Sistema Único de Saúde)8.
Decreased governmental investments worldwide in public health and a lack of effective preventive actions have led to increased risks of infection in several countries9. In Brazil, overall, vector control programs have not produced satisfactory results; hence, the effectiveness of these interventions is still very limited10–12. Investments in the implementation of a rigorous and continuous national program and effective management of public services such as garbage collection and structured sanitation networks are required to reduce the prevalence of the disease in the country13–14. The identification of high-risk areas is necessary for the implementation of directed public health policies and basic services to reduce the burden of dengue15.
Amazonas, characterized by elevated temperature and precipitation and humidity rates, is located in the North Region of Brazil16. These climatic and environmental conditions favor the breeding, replication, and survival of dengue vectors, increasing the risk of transmission17,18. Previous studies have evaluated the clinical and epidemiological factors of dengue patients in Amazonas State18–24. Prevalence and risk factor data for dengue in the general population are scarce in Brazil. This study aimed to assess the self-reported prevalence of dengue and associated factors in adults living in Manaus Metropolitan Region.
This was a population-based cross-sectional study conducted between May and August 2015 with adults living in Manaus Metropolitan Region in the Amazonas State, Brazil. This study is part of a larger study that intended to evaluate the use of healthcare services and resources in this region25.
Manaus Metropolitan Region is composed of Manaus, the capital city of Amazonas State, and the following seven adjacent cities: Careiro da Várzea, Iranduba, Itacoatiara, Manacapuru, Novo Airão, Presidente Figueiredo, and Rio Preto da Eva. Over 60% of the 3,483,985 inhabitants of Amazonas live in this region26. The predominant climate is equatorial, which is characterized by elevated temperature and rainfall rate16. Regions with extremely elevated temperature and humidity rates are at increased risk for dengue transmissions between mosquitos and humans27.
Adults ≥18 years old were eligible for the study and were recruited by probabilistic sampling in three stages by cluster and stratified by age and sex25. The first stage consisted of randomly selecting 400 primary and 20 secondary tracts from the 2,647 urban census tracts of Manaus Metropolitan Region. The second phase consisted of a systematic sampling of households from each tract. The third stage consisted of the registration of all adults ≥18 years old who were present at the residence, and one participant was randomly selected for the interview according to the predefined quotas of age and sex.
The primary outcome was defined as the self-reported prevalence of dengue infections in the previous year. Individual variables included sex (male, female), age group (18-24, 25-34, 34-44, 45-59, and ≥60 years old), marital status (single, separated/divorced, widowed, married), educational level (higher education or above, high school, middle school, elementary schoolor less), ethnicity (non-White, White and Asian, where non-White included Black, Brown [Brazilian mixed race], and Indigenous), socioeconomic status (A/B, C, D/E, where A refers to the wealthiest and E to the poorest28), health status (very good or good, fair, bad or very bad), health insurance coverage (yes, no), self-reported chronic diseases (yes, no), pregnancy status (yes, no), usage of healthcare services in the last 12 months (yes, no), hospitalization in the last 12 months (yes, no), visits from a family health agent in the last 12 months (2-12 visits, 1 visit, no visit), and visits from an endemic disease control agent in the last 12 months (2-12 visits, 1 visit, no visit).
The variable at the city level was the primary care package (Piso da Atenção Básica, PAB) per capita in the Brazilian currency, and at the neighborhood level (according to the neighborhood where the household was located), the Gini index was used.
Data source and measurements
Individual variables were collected from face-to-face interviews conducted by trained interviewers who registered the participants’ responses in electronic tablets (Tab 3 SM-T110 Samsung® Galaxy). The prevalence of self-reported dengue infections was assessed through the following question: “In the last 12 months, has any doctor diagnosed you with dengue?” The following two response options could be filled: “yes” or “no”. PAB per capita was obtained from the Brazilian Ministry of Health29. The Gini index was obtained from the Brazilian Development Atlas30.
To ensure the quality of the study, a pretest was performed with 150 participants who were included in the final sample. A hired company audited 20% of these interviews through telephone contact to confirm the responses, and parts of the interviews were recorded on the electronic tablet.
The sample size was calculated based on a conservative estimate of a 50% prevalence of dengue infection in a population of 2,106,322 adults living in the region, considering a 95% confidence interval (95% CI), an absolute precision of 2%, and a design effect of 1.5. We included an additional 10% in the sample to compensate for losses, with a total of 4,000 individuals to be interviewed.
Variables were analyzed using descriptive statistics with 95% CIs. Prevalence ratios (PRs) were calculated using Poisson regression with robust variance to identify factors associated with self-reported dengue. A multilevel Poisson regression analysis with random intercept was performed to assess continuity and variables at the individual, neighborhood, and city levels. We initially used an empty model to determine the clustering of continuity by city and neighborhood levels to obtain the variance. City and neighborhood variables that reduced the variance at each level were maintained in the final model, whereas those that increased the variance were excluded. If high collinearity was observed among the variables (r>0.9), a latent variable was created by multiplying both variables.
The multilevel Poisson analyses were used to calculate the PR and 95% CIs, which were adjusted by latent variables at the city level (PAB per capita), neighborhood level (Gini index), and individual level (all variables) of self-reported dengue.
Associations were considered statistically significant if p was <0.05. Data analyses were performed using Stata V.14.2 (Stata) with consideration of the complex sampling design (svy command).
The Ethics Research Committee from the Federal University of Amazonas approved this study on March 3, 2015, with the report no. 974,428 (Certificate of Presentation for Ethical Appreciation 42203615.4.0000.5020). Before performing any procedure, all participants provided written informed consent for inclusion in the study.
We included 4,001 participants, of whom 281 self-reported dengue infections in the previous year (prevalence: 7.0%; 95% CI 6.3%-7.8%).
Half of the individuals were women (52.8%), aged from 25 to 44 years old (49.9%), single (54.3%), and had an educational level up to high school (47.5%; Table 1). Most belonged to the lower socioeconomic status (C, D/E; 84.2%) and ethnically identified themselves as Brown, Black, or Indigenous (80.7%); 63.5% of the participants had received at least one visit from a family health agent. The frequency of visits from endemic disease control agents increased as the number of family health agent visits increased (71.9%) and was higher among the wealthier population (43.8%). Regarding the participants who were hospitalized (6.8%) in the previous year, one was hospitalized due to dengue.
|Variables||Total||Frequency of||Endemic disease control agent visits*|
|dengue||None||One visit||2-12 visits|
|Age group (years)|
|Higher education or above||158||4.0||4.4||46||28.9||55||35.2||57||35.8|
|Elementary school or less||1,291||32.3||9.0||398||31.1||482||37.2||411||31.7|
|White or Yellow||774||19.4||9.0||288||37.3||265||34.3||221||28.5|
|Very good or good||2,646||66.1||6.1||924||35.1||804||30.4||918||34.5|
|Bad or very bad||247||6.2||8.9||70||28.4||104||42.2||73||29.5|
|Family health agent visits*|
The frequency of dengue infections was higher in women (8.3%) and adults aged 45 to 59 years (10.6%) and ≥60 years (9.4%), individuals who completed elementary school or less (9.0%), White and Asian individuals (9.0%), individuals belonging to the lower socioeconomic status (9.2%), individuals with bad or very bad health statuses (8.9%), individuals with chronic diseases (8.1%), women who were pregnant in the last 12 months (11.2%), and individuals without any visit from a family health agent in the last year (10.7%).
Dengue was associated with female sex (PR 1.51; 95% CI 1.06-2.13), older age (25-34 years old: PR 1.93, 95% CI 1.06-3.51; 35-44 years old: PR 2.59, 95% CI 1.41-4.78; 45-59 years old: PR 2.73, 95% CI 1.44-5.20; ≥60 years old: PR 2.54, 95% CI 1.19-5.45), White and Asian ethnicity (PR 1.57; 95% CI 1.11-2.23), and a lack of visits from endemic disease control agents in the last 12 months (PR 2.28; 95% CI 1.31-3.99; Table 2).
|Variables||PR (95% CI)||p-value||aPR (95% CI)||p-value|
|Female||1.49 (1.18-188)||1.51 (1.06-2.13)|
|Age group (years)||<0.001||0.268|
|25-34||1.14 (0.78-1.68)||1.93 (1.06-3.51)|
|35-44||1.48 (1.01-2.19)||2.59 (1.41-4.78)|
|45-59||2.22 (1.54-3.20)||2.73 (1.44-5.20)|
|≥60||1.97 (1.28-3.03)||2.54 (1.19-5.45)|
|Separated/divorce||1.72 (1.14-2.59)||1.55 (0.87-2.78)|
|Widowed||2.14 (1.35-3.38)||1.38 (0.68-2.80)|
|Married||1.50 (1.17-1.91)||1.20 (0.84-1.72)|
|Higher education or above||1.00||1.00|
|High school||1.37 (0.65-2.88)||1.07 (0.48-2.39)|
|Middle school||1.58 (0.72-3.43)||1.41 (0.61-3.27)|
|Elementary school or less||2.06 (0.98-4.33)||1.02 (0.45-2.28)|
|White or Yellow||1.37 (1.06-1.78)||1.57 (1.11-2.23)|
|C||2.00 (1.29-3.10)||1.42 (0.80-2.53)|
|D/E||2.65 (1.69-4.15)||1.81 (0.96-3.41)|
|Very good or good||1.00||1.00|
|Fair||1.44 (1.13-1.84)||1.37 (0.97-1.93)|
|Bad or very bad||1.46 (0.95-2.23)||1.02 (0.51-2.06)|
|Yes||0.85 (0.60-1.22)||0.95 (0.59-1.78)|
|Yes||1.45 (1.14-1.84)||1.26 (0.89-1.78)|
|Yes||1.40 (0.94-2.10)||1.65 (0.86-3.13)|
|Yes||1.45 (0.99-2.11)||1.08 (0.59-1.54)|
|Family health agent visits*||<0.001||0.364|
|One||1.85 (1.17-2.91)||1.33 (0.77-2.29)|
|None||2.49 (1.70-3.64)||1.45 (0.87-2.42)|
|Endemic agent visits*||<0.001||0.012|
|One||1.56 (0.94-2.58)||1.57 (0.93-2.67)|
|None||2.07 (1.21-3.55)||2.28 (1.31-3.99)|
After performing multilevel analyses, dengue infections were associated with older age (25-34 years old: PR 1.98, 95% CI 1.05-3.74; 35-44 years old: PR 2.47, 95% CI 1.28-4.75; 45-59 years old: PR 2.65, 95% CI 1.35-5.21; ≥60 years old: PR 2.52, 95% CI 1.10-5.79), White and Asian ethnicity (PR 1.57; 95% CI 1.07-2.31), and not receiving a visit from an endemic disease control agent (PR 2.07; 95% CI 1.23-3.55; Table 3). Pregnancy increased the variance and was excluded from the final multilevel analysis (data not shown).
|Variables||PR (95% CI)||p-value|
|Age group (years)|
|Higher education or above||1.00|
|High school||1.07 (0.45-2.55)|
|Middle school||1.47 (0.58-3.70)|
|Elementary school or less||1.06 (0.43-2.61)|
|White or Yellow||1.57 (1.07-2.31)|
|Very good or good||1.00|
|Bad or very bad||1.19 (0.57-2.49)|
|Family health agent visits*||0.651|
|Endemic agent visits*||0.030|
Seven out of 100 inhabitants of Manaus Metropolitan Region self-reported a diagnosis of dengue in the past year; the frequency of self-reported dengue was higher in women, elderly individuals, White and Asian individuals, and individuals whose households did not receive a visit from an endemic disease control agent. After adjusting for the funding of primary care by the municipality and inequalities in the neighborhoods, the association between dengue and female sex was not significant.
Despite the large number of participants, the study has limitations inherent in cross-sectional designs31. The self-reported outcome may not include participants who had symptoms of dengue but did not seek medical assistance or who presented with subclinical dengue. In 2008, a study conducted in Rio de Janeiro found that out of the 337 participants who had a molecular confirmation of dengue, only 23.3% presented with symptomatic infections32. Dengue may result in a variety of symptoms that can be easily confused with other pathologies, such as malaria33, and only a low proportion of affected individuals seek treatment, leading to fewer diagnoses1. Misclassification due to memory bias or confusion with another disease is also possible. Selection bias may have influenced the results because only individuals who were present at the residences at the time of the interviews were invited to participate. Due to the vector’s indoor characteristics and preference for feeding on humans during daylight hours34, transmission occurs mainly at home35, so it is possible that the interviewees were more frequently exposed to the infection.
In 2015, 34,110 probable cases of dengue infections were registered in the North Region of Brazil, of which 4,131 were reported in Amazonas State, leading to a prevalence of 0.12%, taking the state population as the denominator3,26. This official prevalence was lower than we estimated in the present analysis, probably due to the underreporting of cases to the Brazilian Ministry of Health36. A study conducted between 2009 and 2011 in a public emergency unit in Salvador, Bahia, showed that one out of 12 dengue cases was actually reported to the health authority, which suggests an underreporting of the disease36.
Dengue infections were more frequent in women, which was similar to the results of an epidemiological study conducted in Amazonas State in 2009, in which 54% of the 1,003 notifications occurred in female patients24. The same pattern was observed in a study conducted in the Southeast Region of Brazil from 1998 to 2006, in which 57% of the 1,212 notified cases were found in women37. A plausible explanation is that women tend to seek health services more than men, resulting in more diagnoses in this group38. A large study conducted in Rio Grande do Sul State from 2014 to 2016 analyzed 13,420 blood samples from patients with suspected dengue fever and reported that the infection was equally distributed between both sexes39. In a cross-sectional study conducted in rural Amazonia in 2004, after the analysis of the participants’ blood samples, male sex was a predictor of baseline dengue seropositivity40. Dengue was not significantly more common in women in the multilevel analysis than in men, which indicates that the risk of dengue based on sex may be influenced by the setting.
Older individuals had a higher prevalence of dengue, which was consistent with the results from a previous study conducted in Araraquara, Sao Paulo, which assessed 16,431 cases of dengue reported between 1991 and 2015. The authors concluded that the frequency of dengue was higher in individuals aged 20-59 years old41. A household survey performed in 2005 and 2006 in the Northeast Region of Brazil also found that older age was a risk factor among 2,833 individuals who were examined and assessed for dengue antibodies42. Dengue seropositivity tends to increase with age due to cumulative exposure over time, especially in endemic locations, leading to increased susceptibility to infections43. However, severe dengue and dengue hemorrhagic fever are most common in children44–46. In 2008, individuals below 15 years old were the most affected individuals during an epidemic in Rio de Janeiro State, accounting for 47% of dengue hospitalizations46–48.
In our study, Whites and Asians self-reported more dengue than non-White individuals. Studies performed with Cuban (2017) and Colombian (2014) individuals reported the protective effects of African ancestry genes against dengue infections after the analysis of the genotypes from 274 and 287 diagnosed patients, respectively49–50. In 2018, another epidemiological study conducted in Tanzania included 431 dengue patients with different self-reported ethnicities and found similar results, suggesting a lower risk of dengue among individuals of African ancestry than other ethnicities51. A study from Rio de Janeiro City indicated that self-declared Black individuals had higher incidence rates of severe dengue than other individuals, based on an analysis of 59,395 reported cases during an epidemic in 2008, probably due to the historical socioeconomic vulnerability of this group52.
Not receiving a visit from an endemic disease control agent in the previous year increased the risk of dengue by twofold. This association was maintained after adjustment in the multilevel analysis. Our findings suggest that visits from these agents are effective methods of reducing the incidence of dengue regardless of the status of the other determinants assessed. In 2010, endemic disease control agents were incorporated into primary care to strengthen the surveillance, prevention, and control strategies for endemic diseases53. Despite their relevance, endemic disease control agents are still experiencing difficulties in integrating with the population and family health agents in Brazil54–55. Visits from family health agents did not reduce the frequency of infections. Family health agents are also dealing with obstacles that potentially reduce their availability to the population, such as low salaries, low recognition of their function, a lack of limits on their attributions, barriers in their relationship with the community, weak professional training, and bureaucracy56–57. These results highlight the importance of endemic disease control agents in the prevention and control of diseases, particularly endemic ones.
Dengue was not associated with the presence of chronic diseases or with self-reported health status. Comorbidities such as diabetes and renal, infectious, and pulmonary diseases increase the risk of mortality by approximately 11 times in Brazil58. These patients also seem to be at higher risk of developing severe complications compared to healthier individuals59. Although our study showed no associations between these variables, the presence of concomitant chronic diseases, which leads to lower health-related quality of life60, increases mortality rates and severe complications of dengue.
Dengue was reported by 7 out of 100 inhabitants from Manaus Metropolitan Region and was predominantly observed in women, elderly individuals, Whites and Asians, and individuals living in households that did not receive a visit from an endemic disease control agent. Inequality and the funding of primary care removed the effect of sex on the prevalence of dengue, indicating an influence of the setting.