Solving Colombia's chickenpox season puzzle
Researchers decipher the seasonality of Colombian chickenpox cases with the help of mathematical models
Chickenpox is found all over the world. However, the season during which most people become infected varies from region to region. Matthieu Domenech de Cellès' research group has now unraveled the seasonality of chickenpox cases in Colombia. Differences in humidity mean that the disease occurs mainly at the beginning of the year in the north of the country, while the south is hit particularly hard at the end of the year. The researchers arrived at this result with the help of a mathematical model that may be transferable to other countries in South and Central America. This model may contribute to the development of vaccination programs, for which the seasonal course of chickenpox is an important factor. The study was published in July in the Journal for Infectious Diseases.
Many infectious diseases occur seasonally. We know, for example, that chickenpox infections in Germany are most common at the end of winter. This is partly because of the strong seasonal changes in climate and partly due to the changes in contact rates of susceptible persons.
In the tropics, on the other hand, the situation is more complicated: There are no seasons as in temperate regions, but many diseases still seem to be seasonal, as cases do not occur with the same frequency throughout the year. This seasonality appears less definite and so far, it has been explained only by changes in the contact rate of susceptible persons—for example, the start of the school year, when many children gather together again in a confined space and thus infect each other more quickly. However, due to the paucity of epidemiological data and the complex climate in tropical countries, these associations have not been studied and other factors such as climate have not been considered.
Matthieu Domenech de Cellès' research group at the Max Planck Institute for Infection Biology set out to fully unravel the mechanisms behind the seasonality of chickenpox in tropical countries. Since epidemiological data for chickenpox in tropical countries are often only collected on a nationwide basis, the research group focused on one particular country: Colombia. Here, weekly chickenpox cases have been recorded at the municipality level for years and are available to researchers. Furthermore, as the country has a variety of climates, as it spans over different latitudes around the equator, it was possible to identify climate variables which could impact the transmission of chickenpox.
Chickenpox, also known as varicella, is a highly contagious viral infection that is usually spread by droplet infection. It occurs mainly in young children and is usually harmless with a mild fever, runny nose and typical red blisters on the skin. However, there are more than 4 million cases of severe complications worldwide each year and in pregnant women, infection can lead to malformations of the embryo. The best protection against chickenpox is vaccination and isolation of infected people, which is why in Germany children with chickenpox are not allowed to attend kindergarten or school.
Using descriptive models to see real patterns
Using data sets from 25 Colombian municipalities, the researchers were able to piece together chickenpox incidence from 2011 to 2014. This is where the real work began for the research group: "People are very good at recognizing patterns—so good that they sometimes think they see patterns where there aren't any," explains the study's first author, Laura Barrero-Guevara: "that's why we use descriptive models in our research group to extract patterns—if they exist—from data sets."
To compare the data in a meaningful way, the researchers created a model of chickenpox incidence. From the incidences of each municipality, they were able to model the incidence trend only as a function of latitude and week. The results were clear: chickenpox always occurred in two waves throughout Colombia, in April and October. However, the peaks of these waves varied regionally. In northern communities, such as Cúcuta, the first wave was always higher than the second, while in southern communities, such as the capital Bogotá, it was the other way around—there was the pattern.
Climate or social factors—what causes the seasonality of chickenpox?
This also surprised Barrero-Guevara: "We had expected chickenpox incidence to follow roughly the same seasonal rhythm throughout Colombia. But here we saw a clear regional trend along a north-south axis." To examine how this regional trend in seasonality emerges, the researchers looked for factors that follow the same regional pattern. Climate influences and socio-demographic factors were considered: Were northern communities possibly more densely populated, or were there different migration flows than in the south?
However, a clear correlation with chickenpox incidence was only seen in climate, specifically humidity, which is very diverse in Colombia as it spans almost 16 latitudes. Depending on the region, there are different rainy and dry seasons. These different seasons followed the same trend along a north-south axis as chickenpox incidence. Specifically, if it was less humid in a region at a certain time, there were more chickenpox cases.
A model of chickenpox transmission
To test whether a region's humidity could be the reason behind the seasonality of chickenpox incidence, the researchers created a mathematical model. They divided a model population into four groups: susceptible to chickenpox, exposed to chickenpox (but without symptoms), infected and finally recovered. The model then described how individuals move from one group to the other over time, from infection to recovery.
The key factor in determining the daily incidence of chickenpox is the transmission rate, i.e. how many people are infected with chickenpox each day or, speaking in terms of the model, how many people move from the susceptible group to the exposed group. Using their model, the researchers were able to test the influence of different factors on the transmission rate and thus verify which factors could be used to correctly model the incidence of chickenpox.
It was already known that school holidays affect transmission because children have less contact with other people. The question was whether humidity also had an effect on transmission. The researchers "fed" their model with different scenarios: First, with reduced transmission only during school holidays, then second, with reduced transmission during school holidays and when humidity increased.
Reducing contact during school holidays resulted in two waves of chickenpox cases in April and October, which was also observed in the recorded cases. Unlike the real data, however, these waves were of equal magnitude. Only when the seasonal change in humidity was added to the model did the picture resemble the observed difference in the incidence peaks, suggesting a decrease in transmission during the more humid times of the year.
Predicting chickenpox cases in Central America
To evaluate if the model also worked in a different context, the researchers tested their model with data from México. Here, too, they fed in the local school terms of all capital cities of the departments of México as well as the respective seasonal humidity. And they were successful! Their model was able to predict the course of the recorded chickenpox cases as reported before.
In Central American countries, the seasonality of chickenpox cases has not been described as it is now in Colombia and has been in México. This is where the strength of the research group's mathematical model becomes apparent. "After our model successfully passed the tests, we were able to make predictions for the seasonality of chickenpox in countries such as Guatemala, Panamá, Costa Rica and Nicaragua," explains Barrero-Guevara, “however these results are yet to be confirmed by real data”. The results could prove useful one day, as knowing the seasonal incidence of chickenpox could help make future vaccination campaigns more effective.