Background

Dengue (DENV), Zika (ZIKV) and Chikungunya (CHIKV) are arthropod-borne viruses (arboviruses), primarily transmitted by Aedes spp. mosquitoes, that increasingly threaten urban and peri-urban populations in the tropical and subtropical areas of low- and middle-income countries. A combination of rapid urban growth, sustained poverty, and warming climates can support unrestrained mosquito proliferation and create the requisite conditions for arboviral outbreaks [1–3].

With almost 85% of its population residing in high-density urbanized settings, stark social inequalities including disparities in housing and sanitation infrastructure [4], and climatic conditions and environmental features (e.g., open water reservoirs) that support year-round breeding of Aedes spp. mosquitoes in much of the country [5], Brazil is uniquely susceptible to arbovirus transmission [6,7]. In 2020 alone, Brazil reported almost 1 million notifications of suspected dengue, 82,419 of suspected chikungunya, and 7,387 of suspected Zika [8]. Nevertheless, this number is likely to be substantially underestimated. Evidence from an enhanced surveillance study investigating DENV underreporting in Salvador, Bahia, between 2009 and 2011 indicates that approximately only 1 in 20 DENV-positive persons who presented to a public emergency unit with acute febrile illness were reported to the Brazilian Information System for Notifiable Diseases (Sistema de Informação de Agravos de Notificação, SINAN) [9].

Although asymptomatic infections are common (i.e., estimated in a recent meta-analysis to be 80% for DENV, 50% for ZIKV, and 40% for CHIKV [10]), arboviral infections can lead to disease, hospitalizations, long-term complications, and death. Dengue, Zika, and chikungunya diseases have overlapping clinical presentations with common symptoms including nausea/vomiting, rash, fever, headaches, and joint and muscular pain [11]. Complications of arboviral infections (e.g., Congenital Zika Syndrome or CHIKV-associated arthritis) can have deleterious health and social impacts over the life course, and a subset of infections may trigger severe neurological sequelae (e.g., Guillain-Barré Syndrome) [12] or progress to life-threatening conditions (e.g., severe dengue) [13,14]. Whereas ZIKV and CHIKV-related mortality are rare events, DENV resulted in 554 confirmed deaths in 2020, approximately 0.05% of notified infections in this year [8]. While there is evidence that the risk of severe dengue might be related to non-communicable comorbidities (e.g., diabetes, hypertension, respiratory, cardiovascular or renal diseases) [13–15], whether DENV or other arboviral infections may exacerbate underlying conditions and lead to excess hospitalizations for noncommunicable diseases (NCDs) remains unknown.

Using records of notifications for confirmed dengue, Zika, and chikungunya cases from SINAN and records of hospitalization (Sistema de Informação Hospitalares, SIH) in the public Unified Health System (Sistema Unico de Saude, SUS) from 1 January 2014 to 31 December 2019, we investigated whether increases in arbovirus notifications rates at the municipality- and state-levels are associated with increased hospitalizations from any cause, from arboviruses, or from specific indirect causes.

Methods

Results

From 1 January 2014 to 31 December 2019, the Brazilian health surveillance system registered 7,566,330 confirmed dengue cases, 159,029 confirmed Zika cases, and 433,887 of confirmed chikungunya cases in SINAN. While dengue cases were registered in almost all Brazilian municipalities (5334/5570, 95.8%), only 1791 municipalities (32.2%) registered at least 1 case of Zika, and 2523 (45.3%) registered at least one case of chikungunya. During the study period, notifications of dengue per 100,000 were higher than for Zika and chikungunya and covered a greater geographic distribtion–Zika and chikungunya affected fewer municipalities but incurred a relatively high incidence within the subset of municipalities that registered at least one case of each of the diseases (See Fig 1A–1C and S4 Table). Whereas dengue notifications were mostly observed in municipalities from Southeast and Central-west Brazilian States, Zika notifications have been concentrated in the North, Northeast and Central-west, and chikungunya incidence has been more spread across the North and Northeast regions (See S1 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1.

Number of (A) DENV, (B) ZIKV and (C) CHIKV case notifications (bars, main Y axys) and notifications per 100,000 (red line, secondary Y axys) in Brazilian municipalities with at least one notification between 2014 and 2019. Please note that the scales differ between the three graphs. The dotted black lines in graphs B and C mark the start of ZIKV and CHIKV mandatory notification in Brazil. The dashed blue lines in graph B mark the start and the end of Zika Public Health emergency.

https://doi.org/10.1371/journal.pntd.0010602.g001

When investigating the association between arbovirus notifications and direct and indirect causes of hospitalizations in municipalities that detected at least 200 cases of each disease, we included 2366 municipalities (44.4% of the 5334 with ≥1 case) for dengue, 99 municipalities (5.5% of the 1791 with ≥1 case) for Zika, and 217 municipalities (8.6% of the 2523 with ≥1 case) for chikungunya. Dengue notifications were associated with a small increase in dengue-related hospitalizations (RR 1.0029, 95% Crl 1.0027–1.003) and similar increases in non-haemorragic (RR 1.0029, 95% Crl 1.0027–1.003) and haemorrhagic dengue hospitalizations (RR 1.0029, 95% Crl 1.0019–1.0034) (Table 1). Zika notifications were not associated with dengue (RR 1.0002, 95% Crl 0.9986–1.0019) or arthropod-borne viral fevers and viral haemorrhagic fevers hospitalizations (RR 1.0076, 95% Crl 0.9441–1.082). Chikungunya notifications, on the other hand, were associated with overall increases in dengue (RR 1.0044, 95% Crl 1.0037–1.0051), and arthropod-borne viral fevers and viral haemorrhagic fevers hospitalizations (RR 1.0067, 95%Crl 1.0034–1.0111). At the state level, increases in dengue and chikungunya notifications were also associated with increases in dengue (dengue RR 1.0053, 95% Crl 1.0047–1.0059; chikungunya RR 1.0061, 95% Crl 1.0035–1.009) and chikungunya hospitalizations (dengue RR 1.0048, 95% Crl 1.0027–1.0072; chikungunya RR 1.0206, 95% Crl 1.0115–1.0310).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Changes of monthly age-standardized hospitalization rates directly associated with DENV, ZIKV and CHIKV cases incidence at municipality and at State level.

https://doi.org/10.1371/journal.pntd.0010602.t001

At the municipality level, we found no increases in all-cause hospitalization rates associated to arboviruses notifications. Zika incidence was associated with small increases in hospitalizations for endocrine, nutritional, and metabolic diseases (RR 1.0003, 95% Crl 1.0001–1.0004), diabetes (RR 1.0004, 95% Crl 1.0002–1.0006) and inflammatory polyneuropathy (including Guillain-Barré) (RR 1.0082, 95% Crl 1.0033–1.0133) (see Table 2). Chikungunya and dengue incidence were not associated with any indirect cause of hospitalization. At the state-level, only associations between Zika and hospitalization from inflammatory polyneuropathy as an indirect cause was consistent with the analysis at the municipality-level (Table 3). Finally, in sensitivity analyses investigating potential one- or two-month delays in the association between increases in dengue, zika and chikungunya notifications and hospitalizations, we found similar point estimates for all the analyses (S5 to S10 Tables).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Changes of monthly age-standardized hospitalization rates indirectly associated with Dengue, Zika and Chikungunya incidence in municipalities with at least 200 cases of the diseases.

https://doi.org/10.1371/journal.pntd.0010602.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Changes of monthly age-standardized hospitalization rates associated with Dengue, Zika and Chikungunya incidence in the 27 Brazilian States.

https://doi.org/10.1371/journal.pntd.0010602.t003

Discussion

This nationwide ecological study found that between 2014 and 2019, dengue and chikungunya notifications in Brazil were associated with a small increase in arbovirus-related hospitalizations (i.e., arthropod-borne viral fevers and viral haemorrhagic fevers, dengue, and chikungunya). Zika was consistently associated with increases in hospitalizations by inflammatory polyneuropathy (including Guillain-Barré), but not by other indirect causes. In contrast, increases in dengue and chikungunya notifications were not consistently associated with increases in all-cause or indirectly caused hospitalizations at the municipality- or state-levels.

The association between notification rates of confirmed arbovirus cases and arbovirus-related hospitalizations is expected, but still of high public health relevance. Previous research suggests the direct and indirect cost of dengue ambulatory and hospitalization treatments in Brazil are an estimated 447,000,000 USD per year, with costs related to deaths of approximately 65,000,000 USD per year [27]. Our finding of an association between arbovirus notifications and arbovirus-related hospitalizations likely reflects local environmental and social conditions that may support the proliferation of competent vector populations and lead to high forces of infection for different arboviruses (and serotypes) within a given community over time.

The finding of associations of increased dengue and chikungunya notifications with chikungunya-related hospitalizations and associations between increased chikungunya notifications with dengue-related hospitalizations indicate temporal and spatial co-circulation of arboviruses within a given community and may also reflect locally enhanced arbovirus surveillance and diagnostic capacity. While current evidence suggests that co-infections between ZIKV, CHIKV, and DENV do not substantively increase risks of arbovirus disease severity [28], repeated flavivirus infections are a particular concern for dengue hospitalizations. Although primary DENV infections are thought to confer some degree of cross-protection against heterologous DENVs in the short-term, previous immune experience with a heterologous DENV serotype is considered, in the long-term, to be a key risk factor for developing severe dengue that requires hospitalization upon secondary infection [29,30]. Limited evidence suggests that previous ZIKV infection may similarly facilitate antibody dependendent enhancement of subsequent DENV infection, increasing risks of severe dengue [31]. Overall, our findings reinforce the value of reducing community-level arbovirus exposure over time through integrated initiatives to improve Aedes mosquito control, living conditions and sanitation, and access to personal protective measures, such as topical repellants (reviewed in [32]).

As Zika is a generally mild disease and rarely leads to hospitalizations and deaths, we could not estimate associations between increases in arboviruses notifications and ZIKV-related hospitalizations [8]. Even among children and adolescents, evidence from a recent systematic review indicates postnatal ZIKV infections cause neurological complications in less than 1% of cases [33]. Nevertheless, the fact that the notifications of Zika were restricted to a small number of municipalites and that notification rates are relatively small, small increases in hospitalization due to indirect Zika complications might not have been observed at the ecological level.

Although Zika and dengue are known to have the potential to cause neurological complications [12,34], we only found consistent association between Zika incidence and hospitalizations by Inflammatory diseases of the central nervous system or by inflammatory polyneuropathies, which include Guillain-Barré Syndrome. In endemic countries, hospitalized dengue patients have been found to present encephalopathies, encephalitis, neuromuscular complications, and neuro-ophthalmic involvement (Reviewed in [34]). ZIKV infection has been associated with severe complications, such as Guillain-Barré Syndrome in adults [12,20], but the incidence of these complications among all ZIKV infected patients is very small. In a prospective observational study carried out in Recife, Brazil, on patients presenting with neurological complications possibly related to arboviral infections, Guillain-Barré was the most prevalent complication among the ZIKV-infected patients [12]. Differently, central nervous system (CNS) diseases such as myelitis were more prevalent among CHIKV-infected patients [12].

Although Zika was associated with Endocrine, nutritional and metabolic diseases (which includes diabetes) and diabetes itself, overall there was a lack of consistent association between arboviruses increases and increases in hospitalization from all causes or other specific indirect causes, including by chronic non-communicable diseases, such as cerebrovascular diseases, hypertensive or ischaemic heart diseases. Even though we found no association between increases in dengue notifications and indirect causes of hospitalizations, previous studies provide compelling evidence that dengue disease in individuals with comorbidities, such as diabetes, stroke, hypertension, or respiratory, cardiovascular, or renal diseases, are more likely to progress to severe dengue [13–15] and death [19]. In addition, we found no increases in indirect hospitalizations associated to increases in chikungunya notifications, but previous studies have shown that 2 to 10% of chikungunya patients can present arthralgia and/or myalgia [21], and that the infection could exacerbate preexisting conditions and lead to hospitalization [22]. Acute signs of CHIKV infection (i.e., fever, arthralgia and myalgia) are often accompanied by chronic effects in the liver, muscle, joints and remote lymphoid organs that can lead to hospitalization and that may persist for weeks to years [21]. Similarly to dengue, a study conducted in the Réunion Island has shown that hypertension and underlying respiratory or cardiological conditions are independent risk factors for developing a severe case of chikungunya disease (i.e., as defined as a case that required maintenance of at least one vital function) and that could lead to death [35]. However, even though CHIKV rarely leads to severe cases and hospitalizations [36], several studies have found higher mortality during epidemics of CHIKV [37–40]. A study conducted in Brazil looking at excess deaths during the CHIKV pandemic months in 2016 compared to previous years found a 21% increase in deaths in the State of Pernambuco, 17% in Rio Grande do Norte and from 5–15% in different regions of Bahia associated to that year chikungunya epidemic [37]. Studies using the same methodology found smaller increases in mortality associated with 2014 and 2016 chikungunya epidemics were also observed in Puerto Rico, India, Guadelupe, Martinique, and the Dominican Republic, where the epidemics reached smaller incidences than in the Brazilian state of Pernambuco [37–40].

Our study analyzed nationwide data of confirmed arbovirus cases notified in Brazil and used all public hospitalizations from SUS, which comprised the majority of hospitalization registered in the country. Nevertheless, this study is subject to limitations inherent to the ecological study design—although we found no consistent association between higher arbovirus incidences at the municipality level and increased hospitalization rates by indirect causes, we cannot rule out the possibility that individuals infected by arboviral infection will have higher risk of discompensating from a previous comorbidity and being hospitalized by that disease. Similarly, we are unable to unable to discern whether a given individual experienced repeated arbovirus infections and that would increase their risk of hospitalization by both direct (arboviral-related) or indirect causes. The small proportion of arboviral cases that lead to hospitalization and the potential excess cases of hospitalizations that could occur in the peak periods of diseases transmission might be very sensitive to underreporting of cases. Dengue registries are subject to high levels of underreporting, which are likely to be higher in periods of the year when the disease is less prevalent [9], which could lead to the insufficient capacity of finding associations between an increase in notifications of arboviruses in general and admissions to hospital, especially at the municipality- or state-level. In the case of Zika, previous studies have also shown that nearly 10% of cases were reported in pregnant persons (34), indicating bias in disease reporting patterns.

Conclusions

In this study, we found evidence that increases in dengue, and chikungunya notifications were associated with small increases in age-standardized arbovirus-related hospitalizations, but no consistent association was found with all-cause or other specific indirect causes of hospitalization. We also found that increases in hospitalizations by neurological diseases have been associated with increases in Zika notifications. We showed that strong associations between arboviral infections and increases in complications that lead to hospitalization can be detected at ecological municipal or state level analysis, but further investigations at the individual-level are necessary to understand if arboviral infections have the potential to exacerbate individual risks of hospitalization from underlying conditions, such as chronic diseases. Understanding these associations could improve management of patients, lead to the investigation of arboviral infections among patients hospitalized by chronic conditions in endemic areas for arboviruses, and reinforce the need to reduce community-level arbovirus exposure.

Supporting information