The sudden emergence and detection of Influenza A H1N1 (pdm09) in KUMACA in the Ashanti region of Ghana in 2017 was totally unexpected. This tested the preparedness of local health response systems and exposed the weaknesses in these systems as it took close to two weeks and four deaths  to confirm aetiology. This necessitated the need for strong local capacities for preparedness and response against influenza and emphasised the importance of multisectoral collaboration in epidemic situations to establish aetiology in a timely manner to inform patient management strategies. The importance of a strong local capacity was manifested during the second and third outbreaks in May and December 2019 respectively where aetiology was confirmed within a day.
The different influenza strains detected during the three outbreaks may be attributable to the frequent genetic shift of the surface antigens of the influenza virus and the circulation of these viral subtypes in Ghana. This calls for a concerted effort by all stakeholders to make provisions and enhance preparedness for epidemics by the different strains of the virus at any point in time.
It is apparent from our results that influenza A virus activity is highest during the months of November and December compared to May. These months (November and December) are the transition months from the minor rainy season to the dry season in Ghana with moderate intensity of warmth and low humidity. Low temperatures and humidity favour influenza virus infectivity as has been demonstrated by a number of studies [13–16]. The virus keeps maximal infectivity at relatively low humidity while the viral inactivation occurs at relatively high humidity and temperature. Consequently, the virus is rendered non-viable after being carried in respiratory droplets for relatively longer periods in high humid air. This moderate warmth and low humidity could explain the high influenza activity during the months of November and December. The high humidity of the rainy season around the month of May potentially limits the extent of spread and viability of the virus resulting in the low influenza A activity around that period.
The overall high prevalence of influenza A observed under this study and at the different time points could be due to the relative activity of the influenza virus all year round in tropical regions of Africa as reported by Radin et al after 5 years surveillance of influenza activity in some African countries . Prevalence rates for the first (December 2017) H1N1 (pdm09) and the second (May 2019) H3N2 outbreaks conform to the findings of a three year surveillance report on influenza viruses in Ghana  and could represent the prevalence rate of these subtypes in Ghana. The high prevalence rate (71.2%) recorded during the December 2019 H3N2 period is characteristic of a temperate pattern of influenza activity between November and January in North African countries and other temperate regions [17–19], and suggestive of the influence of climatic conditions on the transmission of influenza viruses . This high prevalence also points to the endemicity of the virus subtype in Ghana and other West African states .
In our analyses, we found that influenza A was more frequently detected in subjects < 20 years which is consistent with reports that suggest that influenza A, especially H1N1 has a higher attack and relative mortality rates in people under 20 years [21, 22]. The predominance among this group could be explained by the fact that there is a high level of physical contact and interaction among this group due to overcrowding of students as witnessed in many Ghanaian second cycle schools . Consequently, when one is infected, there is a high rate of spread of the viral infection among the students . This makes them efficient transmitters of the virus and could represent an at risk group. This study could not establish contact with an infective person as a risk factor (OR = 0.37) of influenza A infection as evidenced by the large proportion of influenza positive cases (70.2%) who had not had any prior physical contact with an infective person. This is because the virus could be transmitted through infectious droplets and aerosols and as such physical contact with an infective person might not be necessary for the viral transmission especially in an epidemic situation . This underlies the effect of overcrowding on the transmission of the virus through infectious droplets and aerosols.
During outbreaks of zoonotic origins, information on contact with animals or wildlife and/or consumption of their meat becomes important [25, 26]. Our results indicated no history of animal contact. Even though our limited data may not warrant such a conclusion, an observed high prevalence of Influenza A H1N1 (pdm09) in pigs in Kumasi  provides hints of the possibility of human-to-swine or vice versa transmission. This underscores the need for systematic swine surveillance in Ghana considering the close contact between humans and livestock especially pigs which are now seen in many developing countries including those of sub-Saharan Africa [28–30].
Although many influenza patients may not present with specific signs and symptoms of illness, the acute onset of respiratory illness is typical for influenza. This includes coughing, fever and headache . Although the triad of headache, coughing, and pharyngitis has been found to constitute the major predictors of influenza disease , our study found headache as the most important predictor of influenza infection during the outbreaks, as those who presented with headache were five times more susceptible to influenza A infection. This contrasts the findings of Monto et al, (2000) who found fever and coughing to be the most important predictors of influenza disease .
Our data comes with some limitations. Key limitation is the lack of data for some of the cases who were recruited. This affects the sample size and statistical power of the study and consequently does not allow for generalisation of findings. Also, information provided by this study does not depict possible seasonal variability as data was only taken during outbreaks and not a year-round surveillance.
In conclusion, our study provides epidemiological information to better understand the incidence and burden of influenza to allow for prioritisation of prevention and control strategies. Information on the current circulating strains would aid in controlling and preparing for the next influenza epidemic and this requires information on the seasonality of influenza, which is sorely lacking. The recent influenza outbreak experience is necessary to develop the next generation of strategic, operational, and tactical plans for any anticipated or unplanned outbreaks.