Influence of meteorological parameters in the seasonality of influenza viruses circulating in Northern Cameroon

Background Several studies have demonstrated the role of meteorological parameters in the seasonality of influenza viruses in tropical and subtropical regions, most importantly temperature, humidity, and rainfall. Objectives This study aimed to describe the influence of meteorological parameters in the seasonality of influenza viruses in Northern Cameroon, a region characterized by high temperatures. Methods This was a retrospective study performed in Garoua Cameroon from January 2014 to December 2016. Monthly proportions of confirmed influenza cases from six sentinel sites were considered as dependent variables, whereas monthly values of mean temperature, average relative humidity, and accumulated rainfall were considered as independent variables. A vector error correction model was used to determine the relationship between influenza activity and the meteorological variables. Results and conclusion Analysis showed that there was a statistically significant association between overall influenza activity and influenza A activity with respect to average relative humidity. A unit increase in humidity within a given month leads to more than 85% rise in the overall influenza and influenza A activity 2 months later. Meanwhile, none of the three meteorological variables could explain influenza B activity. This observation is essential in filling the gap of knowledge and could help in the prevention and control strategies to strengthen influenza surveillance program in Cameroon.


| INTRODUC TI ON
Influenza is a public health threat in the world as it is the source of important epidemics and pandemics. 1 Several studies show that influenza is thought to infect approximately 1 billion people each year resulting in 3-5 million severe cases and up to 500,000 deaths worldwide. 2 In Africa, outbreaks of A/H3N2 influenza in Madagascar and the Democratic Republic of Congo in 2002 resulted in 754 and 170 deaths, respectively. 3,4 Influenza epidemiology is better known in temperate regions than in tropical regions. 5,6 In temperate regions, several studies have shown that temperature and relative humidity (RH) play an important role in the survival of the influenza virus 7 and that influenza epidemics occur during the winter months. In tropical regions, there is a high level of influenza activity during the rainy season and epidemics occur from November to March. 8,9 However, it should be noted that influenza activity in tropical regions has a heterogeneous pattern without clear explanations. [2][3][4][5][6] In these tropical settings, transmission of the disease seems to be continuous throughout the year; meanwhile, the exact period of occurrence of epidemics is difficult to predict. 10,11 In Cameroon, a sentinel surveillance network for influenza was implemented in 2007 by the Centre Pasteur of Cameroon and was later designated as the National Reference Centre for Influenza (NIC) for Cameroon by the World Health Organization (WHO). This public health activity concerned only the southern part of the country. The first results showed the circulation of A/ H1N1/2009, A/H3N2, and B viruses 12 with a high circulation in the rainy season. 13,14 A recent study performed in Yaounde found no significant association between influenza activity and three meteorological variables (temperature, rainfall, and RH) though some synchronies were observed with rainfall. 15 The town of Garoua is located in northern Cameroon and is dominated by a Sudan tropical climate, with high temperatures that can exceed 30°C. In this city, the circulation of influenza viruses is poorly documented. Existing data are incomplete and does not provide information on the factors influencing the occurrence of the disease, with an inability to predict epidemics. Hence, this study was undertaken to explore the possible influence of meteorological variables on the circulation and transmission of influenza viruses in the North region of Cameroon. A retrospective analysis of the time series was carried out in the town of Garoua in the North region of Cameroon from January 2014 to December 2016, in order to evaluate a possible influence of the meteorological parameters in the seasonality of influenza.

| Study areas and design
Garoua is a port city and the capital of the North Region of Cameroon. It is located 9.30 latitude and 13.40 longitude and has a Sudan tropical climate with a dry season that runs from October to April and a shorter rainy season from May to September. The aver- one year to another and even from one month to another. Figure 1 shows the location of the study sites.

| Epidemiological and virological data
Six health facilities of Garoua were involved in the influenza surveillance network in Cameroon. These health facilities consult individuals of all age groups. However, the majority of those with respiratory infections are children. During the study period, all participants were enrolled based on clinical definition for influenza-like illness defined by a sudden onset of fever ≥38°C, cough, and/or sore throat with appearance in five previous days. All patients complying with this clinical definition were directly included in the study. After approving to participate in the study, socio-demographic and clinical data were collected from each patient as well as nasal swabs.
Swabs were introduced into a cryotube containing 2 mL of virus transport medium and stored at +4°C at study sites and were then transported to the Centre Pasteur of Cameroon where they were stored at −20°C prior to analysis and −80°C thereafter. In the laboratory, viral RNA was extracted from 140 μL of the sample using the

| Meteorological data
The climate variables from January 2014 to December 2016 available for this study at the Meteorological Service of the Ministry of Transport of Cameroon were monthly cumulative rainfall, average TA B L E 2 Characteristics of study variables monthly RH, and average monthly ambient temperature. These data were obtained from a ground station located at latitude 9°20′, longitude 13°23′E, and 242M altitude.

| Ethical considerations
This study was approved by the National Ethics Committee and the Ministry of Public Health of Cameroon. All participants enrolled gave their signed informed consent for those over 21 years old or the legal representative (parents, tutors, or guardians) for those under 21. Participant data were kept confidential.  As previously stated in the methods section, the three models were evaluated for the respect of conditions necessary for VECM analysis.

| RE SULTS
All variables were found to be stationary when differentiated (I(1)) based on the results of the ACF and the ADF test (P-value < 0.05). The optimal lag was determined using the VAR (vector autoregression) lag order selection criteria and was found to be lag 2 since it is the one which best minimized the different information criterion. The Johansen co-integration test identifies stable, short-run and long-run relationships between sets of variables. It revealed the existence of four cointegration relationships between the different influenza types and the three meteorological variables (Table 3). This means that four linear combinations exist between the variables that causes them to have a short-term and/or long-term relationship over time. Stability of the model was verified using the CUMSUM recursive estimates, and the models were found to be stable. All residuals were not autocorrelated, had no problem of heteroscedasticity, and were found to be white noises.
The estimation of the different models is given in It is noted that only one meteorological variable best explains the variation of overall influenza and influenza A activity, that is, RH at lag 2. A unit increase in humidity within a given month leads to more than 85% rise in the overall influenza and influenza A activity two months later. Overall influenza activity was also found to

F I G U R E 3 Monthly distribution of influenza activity
Overall influenza activity (%)

Influenza B activity (%)
be influenced by its preceding activity. Indeed, a unit rise in overall influenza activity during a given month leads to a 49.8% rise in this same variable during the following month.

| D ISCUSS I ON
In this study, we evaluated the association between influenza ac- Average RH was the only meteorological variable to show a significant and positive association to overall influenza activity and to influenza A activity. Several studies have instead reported a negative association between influenza and RH. 18 This is due to the fact that respiratory particles exposed to low ambient RH between 20% and 35% can remain airborne for up to 24 hours, thus increasing the risk of airborne transmission. 18,19 More studies are required to understand the reason for the rise in overall influenza activity and influenza A activity despite high RH in the North region.
Previous reports from tropical regions particularly in Bangladesh, Guatemala, El Salvador, and Panama also revealed a proportional association between influenza and humidity [ 2,9,20 ]. Since airborne transmission is unlikely in these circumstances, transmission of influenza probably occurs through contact and droplet modes. 21 On the other hand, none of the variables could explain influenza B Showing that transmission of influenza in Garoua probably occurs by the contact mode.
A major limit to this study was the relatively short period from which data were obtained as well as the use of monthly data instead of weekly data. Moreover, there were many months for which influenza activity was null due to low or limited samples collected. All these might have affected the precision and the power of our statistical analyses. More data are required from this region in order to confirm the trends observed in this study and to obtain more conclusive results. Also, it will be essential to consider other environmental factors not included in this study such as UV radiation, absolute hu-

ACK N OWLED G EM ENTS
We would like to acknowledge the focal points of the influenza sen-

CO M PE TI N G I NTER E S TS
The authors declare that they have no competing interests.