High Mortality Associated with Chikungunya Epidemic in Southeast Brazil, 2023

Introduction

Until then free area, in Brazil in 2014 two lineages of the chikungunya virus (CHIKV) were detected, in the North, Asian lineage and in the Northeast, ECSA (East/Central/South Africa, CHIKV), with transmission sustained by Aedes aegypti. In the following years, only the ECSA lineage remained and caused several focal epidemics, mainly in the North and Northeast regions(1). In recent years there has been an increase in the number of cases with rapid territorial expansion, reaching densely populated areas in the South and Southeast regions. The co-circulation of dengue and chikungunya, associated with little clinical knowledge of the latter by health professionals, has made the diagnosis and treatment of the disease difficult, in addition to preventing a more reliable assessment of the impact of chikungunya in Brazil (2) and in other countries in the region.(3)

In the year 2023, Minas Gerais, a state located in the Southeast Region of Brazil, was the state most affected by the chikungunya epidemic(2). Using a methodology adopted by the World Health Organization(4), we assessed mortality associated with the chikungunya epidemic in the most affected areas

Methods

We assessed the epidemiological situation of chikungunya based on cases available in the laboratory surveillance, disease notification and mortality systems maintained by the Brazilian Ministry of Health (http://tabnet.datasus.gov.br/). The North and Northeast Health Macro-regions of Minas Gerais were evaluated together because they are contiguous areas, similar from a social, epidemiological and climatic point of view(figure 1). Together, these Macro-regions total 2.5 million inhabitants, distributed between urban clusters, such as Montes Claros (largest city in the region, 413 thousand inhabitants) and extensive rural areas with low population density. The region’s climate is tropical and during the studied period there were no extreme weather events in the region, according to data from the National Institute of Meteorology (INMET, Appendix).

• Download figure
• Open in new tab

Figure 1.

Map of Brazil highlighting the State of Minas Gerais and the North and Northeast Health Macro-regions

In table 1, we present the number of laboratory confirmed cases and calculate the coefficient of incidence and positivity of laboratory tests (positive tests/tests performed) of chikungunya, dengue and covid-19 (Severe Acute Respiratory Infection cases) for all Health Macro-regions of Minas Gerais, during the period of the epidemic. The regions selected for the present study had the highest positivity and incidences of chikungunya in the state and low positivity and incidences of dengue and covid-19.

We assessed excess mortality associated with the 2023 chikungunya epidemic. A time series of average daily all-cause mortality was constructed for each month from January 2017 to December 2023. To estimate expected mortality for the period of chikungunya epidemic, a robust Poisson regression model was constructed based on the formula below:

Where:λ_i represented the number of deaths in a particular month are time in months as a linear, quadratic and cubic to correct for secular fluctuations and are the nominal variables associated with the months, to correct seasonal fluctuations and ϵ_i is the error term.

We estimated the regression coefficients excluding the most critical period of the covid-19 pandemic (Jan/2020-Jul/2022) and the period of the chikungunya epidemic (Dec/2022-May/2023). The period of the chikungunya epidemic was defined as the months with incidence coefficients above the 95% percentile of the analyzed period. Excess deaths were considered as the difference between observed deaths and those estimated by the model for the months in which observed deaths exceeded the upper limit of the 99% CI(4)(Figure 2).

• Download figure
• Open in new tab

Figure 2.

Bottom: absolute number of monthly cases of chikungunya (blue line, right scale), number of observed monthly deaths (black line, left scale), number of monthly deaths predicted by the model (solid red line, left scale) and upper and lower limits of the 99% confidence interval (dotted and solid yellow lines respectively, left scale). In the middle: monthly incidence coefficient of chikungunya per 100,000 inhabitants (blue line, left scale), monthly number of excess deaths (red columns, right scale). Above: dengue incidence coefficient per 100,000 inhabitants (blue line, left scale). All data refer to the North and Northeast Health Macro-regions of Minas Gerais.

Results

During the chikungunya epidemic, there were 890 more deaths than the number expected for the period. This value is the mortality considered attributable to chikungunya. We also constructed a robust Poisson regression with sine and cosine seasonal components for Fourier decomposition, the results found were not statistically different (p=0.8) and we opted for the model presented due to its logical simplicity and smaller number of terms (Appendix)(5).

We found that the excess of monthly deaths has a significative Kendall’s tau-b correlation between the total number of confirmed monthly deaths, with laboratory-confirmed cases and with positivity by RT-qPCR test for chikungunya(table 2). On the other hand, there was a negative and non-significant correlation between the excess of monthly deaths and the total number of confirmed monthly deaths, confirmed cases and positive tests for covid-19.

The correlation between excess monthly mortality and the number of dengue cases and deaths was an expected phenomenon due to the seasonal coincidence with chikungunya. However, there was a negative and non-significant correlation between excess monthly mortality and the positivity of laboratory tests (RT-qPCR) for dengue.

The excess mortality rate in the period was calculated at 35.1/100,000 inhabitants. This value is a little more than half the global average excess mortality rate from Covid-19 estimated by WHO for the first year of the pandemic (60/100,000) (4), it is seven times greater than the excess rate estimated global average mortality rate for the 2009 influenza pandemic (5/100,000 inhabitants) and is close to the excess mortality rate for the 1967 influenza pandemic (40/100,000)(6). In a similar study we found that in the 2016 chikungunya epidemic in Pernambuco there was an excess death rate of 47.9 and in the 2014 epidemic on the islands of Guadalupe and Martinique the excess deaths were 81.6 per 100,000 inhabitants(7,8). Mavalankar and colleagues calculated the excess mortality rate in Ahmedabab(India) during the 2006 chikungunya epidemic at 77.5 deaths per 100,000(3,800 deaths)(9) while no deaths have been officially confirmed in the country.

The number of deaths(890 deaths) attributable to the chikungunya epidemic is almost 60 times greater than the number of deaths(15 deaths) identified by epidemiological surveillance in the two Macro-regions studied. Despite representing a small fraction of the Brazilian territory, this excess of deaths observed is almost 8 times greater than the total number of deaths from chikungunya identified by epidemiological surveillance throughout Brazil in the year 2023(106 deaths). In Pernambuco(2016, Brazil) there were 4,505 excess deaths during the chikungunya epidemic period, but surveillance confirmed only 94 deaths. In Puerto Rico in the 2014 epidemic, only 31 deaths from chikungunya were confirmed, but there were 1,310 deaths higher than expected for the period(7,10). In these examples, the underestimation of the number of deaths by the official surveillance system was 98%, the same value found in the present study.

Discussion

Covid-19 pandemic brought great interest to the topic of monitoring excess mortality as a fundamental tool for assessing the impact of emerging viruses, as it is applicable in different contexts and is less dependent on laboratory resources(4,6). Given the significant increase in the number of chikungunya cases in endemic areas and the expansion of this and other arboviruses to free areas, tools complementary to traditional epidemiological surveillance need to be added to better assess the impact in terms of morbidity and mortality.

Excess mortality is a measure that contributes to building solid evidence of the impact of epidemics. Influenza is a classic example of the use of this tool to more adequately measure the number of deaths, as a significant part of deaths occur due to bacterial complications, decompensation of pre-existing diseases, among others.(6)

Adequately quantifying the number of deaths caused by emerging viruses is a fundamental need so that risk assessment and priority-setting systems can direct resources to address epidemiological challenges.

In recent years, necropsy studies have demonstrated viral antigens and genetic material in tissues of important organs such as the CNS, heart, lung and liver(11–14). The fact that the worsening of chikungunya is caused by a severe systemic infection, immunological and hemodynamic dysfunction, associated with the dysfunction of multiple distinct organs makes it difficult to define a single clinical syndrome that characterizes the worsening, which makes clinical suspicion very difficult, especially as it is a disease emerging(11–14).

The first warning sign about the importance of severe forms of chikungunya was given in 2006 on the island of Reunion(15), the issue of underreporting of deaths was subsequently demonstrated in India (9). Since then, evidence of these warnings has been confirmed through different study designs, in different regions of the world, however, several researchers and even official documents still classify chikungunya disease as having a low risk of death.

This study did not receive funding and the authors declare that they have no conflicts of interest.