Epidemiology and molecular characterization of influenza viruses in Burkina Faso, sub‐Saharan Africa

Background The importance of influenza viruses in respiratory infections in sub‐Saharan Africa has been historically overlooked, including in Burkina Faso. Objectives This study therefore aimed at evaluating the prevalence and seasonal occurrence of influenza viruses in children under 5 years old, at risk of influenza‐related complications, presenting with influenza‐like illness (ILI) or severe acute respiratory infection (SARI). The study also aimed at identifying the periods with increased influenza transmission for vaccination recommendations in Burkina Faso. Methods From January 2014 to December 2015, ILI and SARI (2015 only) patients were recruited in six healthcare centers in Burkina Faso. Influenza A and B molecular detection and subtyping were performed. Clade clustering of a subset of A(H1N1)pdm09 and A(H3N2) strains was deduced by performing phylogenetic analyses on hemagglutinin gene sequences. Weekly surveillance data from FluNet (2011‐2013; 2016) and this study (2014‐2015) were used to identify periods of increased influenza activity. Results Influenza A and B viruses were detected in 15.1% (112 of 743) of ILI and 6.6% (12 of 181) of SARI patients. Overall, influenza A viruses were largely predominant (81 of 124, 65.3%), with 69.1% of A(H3N2) and 30.9% of A(H1N1)pdm09 strains. Four waves of increased transmission were identified in 2014‐2015, each dominated by different influenza subtypes and clades. Between 2011 and 2016, periods of increased influenza activity varied in their frequency, duration, and timing. Conclusion Influenza A and B viruses were detected in a substantial number of ILI and SARI cases in Burkina Faso. Vaccination in September‐October would likely protect the highest number of patients.


| BACKGROUND
Despite the burden of influenza infections worldwide, 1,2 morbidity and mortality associated with influenza have been largely overlooked in sub-Saharan Africa, 3 likely due to previous lack of access to diagnostic tools. Misinterpretation of non-pathognomonic influenza symptoms with those induced by various viral and bacterial respiratory infections or by malaria or Lassa fever also complicates the diagnosis. [4][5][6][7] The implementation of the WHO Global Action Plan for Influenza Vaccines, strongly promoting seasonal influenza vaccination, 8 remains challenging in (sub-)tropical low-income countries. In contrast to temperate regions usually having a single influenza season, various patterns of circulation (one season, two seasons, or even year-round) can be observed in tropical regions. [9][10][11][12] However, proper understanding of influenza seasonality for each country, necessary for successful vaccination, is still lacking.
In Burkina Faso, an influenza sentinel surveillance for outpatients with influenza-like illness (ILI) was established at the National Influenza Reference Laboratory in 2010, in the aftermath of the A(H1N1) pandemic. In a preliminary report covering 2010-2012, 6.6% of ILI cases in the general population were influenza positive. 13 The few studies investigating the seasonal circulation of influenza in Burkina Faso were not conclusive on seasonal patterns, 10,13,14 hindering proper recommendations for a national policy for the management and prevention of influenza-related illnesses, and no seasonal vaccination is currently performed in Burkina Faso. We therefore initiated a two-year influenza surveillance in several healthcare facilities in Burkina Faso to assess the prevalence and seasonal occurrence of influenza viruses in children with ILI and severe acute respiratory infection (SARI). Prevalence data gathered in this study were analyzed together with FluNet surveillance information to identify the periods with increased influenza transmission for vaccination recommendations.

| Ethics statement
This study was approved by the Institutional Ethical Committee of Centre Muraz (Ethic Clearance number 13-2013/CE-CM). Samples were collected after written informed consent from parents or legal guardians. Laboratory results were shared with the participating healthcare centers.

| Inclusion criteria
For the purpose of this study, children up to 60 months of age presenting with ILI or SARI symptoms were enrolled. ILI cases were defined as outpatients presenting with a history of fever or measured fever (≥38°C) and cough or sore throat, with the onset of symptoms within the prior 10 days. SARI cases were defined as inpatients with a history of fever or measured fever (≥38°C), and cough or difficult breathing, with the onset within the last 10 days.

| Data and sample collection
Starting in January 2014, we implemented an active surveillance of patients presenting with ILI in four healthcare centers (Colsama, Accart-ville, Do and Leila), all situated in Bobo-Dioulasso, the second most populated city in Burkina Faso after Ouagadougou. In January 2015, the surveillance activities were extended to two other hospitals.
Outpatients with ILI, as well as hospitalized children presenting with SARI, were recruited at the hospital of Bogodogo in Ouagadougou.
Both hospitals are the referral hospitals for SARI cases in their respective regions (ie, Centre and Hauts-Bassins regions). All healthcare centers were selected based on the following criteria: geographic accessibility, high number of patients seeking medical care, presence of qualified physicians and nurses willing to participate to the study, and the ability to directly store samples at low temperatures.
The physicians and nurses were trained before the beginning of the study on inclusion criteria, sample collection, and storage. A nasopharyngeal swab was collected from each ILI patient, and a nasopharyngeal aspirate was collected from each SARI patient. All samples were resuspended in viral transport medium (Copan, Italy). Demographic, socio-economic, and clinical data were collected from all participants using a questionnaire administered by trained personnel.  Statistical analyses (Chi-square tests, Mann-Whitney Rank Sum

| Data analysis
Tests, and z-test for low proportions) were performed using SigmaPlot version 12.0 (San Jose, CA, USA). P-values lower than .05 were considered statistically significant.

| Influenza virus circulation in 2014-2015
While   Figure S1). Influenza B/Victoria strains (n = 15) were exclusively detected in the first 2014 wave together with 2 B/Yamagata strains. In 2015, B/Yamagata strains (n = 12) were detected sporadically throughout the year ( Table 2). The (sub-)types detected in SARI patients matched those found in ILI patients during the same period.
No temporal difference between Bobo-Dioulasso and Ouagadougou was observed (2015).

| Periods of increased influenza activity between 2011 and 2016
Between  (Table 3).

| DISCUSSION
Young children are a major risk group for complications due to influenza infection. [18][19][20] They often represent the majority of patients enrolled in ILI surveillance, also in Africa. 13,[21][22][23][24] In Africa, influenza accounts for 8%-20% of ILI cases in children below 5 years old [21][22][23][24][25] and around 7% of SARI cases. 26 Similarly, we found that 15.1% of ILI and 6.6% of SARI patients enrolled in our study tested positive for influenza A or B viruses over the 2 years period. For ILI cases, the prevalence of influenza increased with age, similarly to other studies. [22][23][24][25] Particularly in very young children, the broad definition of "influenzalike illness" leads to the inclusion of patients with respiratory infections caused by other pathogens, especially respiratory syncytial virus and rhinoviruses. [27][28][29] Increasing socialization with the children's age also contributes to age-dependent differences in risk of infection 30 FluNet data for Niger -not shown -and data from Mainassara et al. 22 ) and unlike Côte d'Ivoire, Ghana or Togo, 10,23,25,34 with a different climate. In the future, increased efforts toward sampling and individual strain characterization will help to better understand the temporal occurrence of influenza activity, virus spread between countries and potential persistence within West Africa. 35 Although variations across years of surveillance occur, influenza epidemics seem to happen mostly in September-October, corresponding roughly to the end of the rainy season and beginning of the Harmattan season (colder, dry, windy season), and January-March, essentially corresponding to the end of the Harmattan season (aggregated data over years, Table 3). Vaccination with the Northern hemisphere vaccine formulation as early as possible in September-October would cover this prolonged period, as suggested before. 10 Vaccination in April would have little benefit due to overall low influenza activity in April-June and waning protection with time. 36 If strains emerge during unusual periods and/or exhibit a major antigenic drift affecting vaccine-induced immunity, additional vaccination schedules may be required, 37 but would be difficult to implement. Until yearly variations are better understood, vaccination of pregnant women could be recommended at any time of the year with the most recent vaccine formulation available. 12 The vast majority of ILI patients enrolled in this study were treated with antibiotics, despite that viruses are usually the main cause of uncomplicated ILI. 29 This high proportion, which is not uncommon and not solely an issue of low-income countries, 38 highlights the need for revision of clinical best practice to identify cases of bacterial (co)infections. 39,40 Education and awareness of the healthcare personnel as well as the patients' parents toward the risks linked to antibiotic resistance will also be necessary to efficiently control over-prescriptions. 41,42

| CONCLUSION
Our findings showed that influenza viruses are implicated in a substantial number of ILI and SARI syndromes in children in Burkina Faso, and current data are supporting vaccination in September-October to decrease influenza burden. The influenza surveillance network in Burkina Faso should be sustained to further refine vaccination recommendations and to evaluate the cost-benefit of seasonal vaccination of high-risk patients.