Characterization and epidemiology of influenza viruses in patients seeking treatment for influenza-like illnesses in rural BangladeshMA Fally1, M Redlberger-Fritz2, P Starzengruber1, P Swoboda1, HP Fuehrer1, EB Yunus3, WA Khan4, H Noedl1
1 Institute of Specific Prophylaxis and Tropical Medicine, Medical University of Vienna; Malaria Research Initiative in Bandarban (MARIB), Bangladesh
2 Department of Virology, Medical University of Vienna, Bangladesh
3 Department of Virology, Chittagong Medical College, Chittagong, Bangladesh
4 International Centre for Diarrhoeal Disease Research, Dhaka, Bangladesh
Correspondence Address: Source of Support: This study was supported by the Malaria Research Initiative Bandarban (MARIB), Vienna, Austria. Alere Austria (Linz, Austria) Vienna, Austria. Alere Austria (Linz, Austria), Conflict of Interest: None DOI: 10.4103/0022-3859.105441
Source of Support: This study was supported by the Malaria Research Initiative Bandarban (MARIB), Vienna, Austria. Alere Austria (Linz, Austria) Vienna, Austria. Alere Austria (Linz, Austria), Conflict of Interest: None
Context: Infections caused by influenza viruses are a major health burden, both in developed and developing countries worldwide. Nevertheless, the overwhelming majority of influenza reports originate from industrialized countries in northern and southern temperate zones. Aims: The aim of this study was to determine the epidemiology of influenza viruses in patients seeking treatment for acute febrile illnesses in rural Bangladesh. Settings and Design: As part of our research on the causes of febrile illnesses in rural Bangladesh, nasopharyngeal swabs from patients with signs and symptoms consistent with influenza were collected from 2008 onwards. Materials and Methods: Viral infection was established using two independent rapid diagnostic tests (RDTs) and later confirmed by RT-PCR. Results: A total of 314 fever cases were enrolled in a survey of febrile illnesses carried out in Bandarban District in southeastern Bangladesh, out of whom 38 (12.1%) tested positive by RDT. Molecular subtyping showed that seasonal H3 strains (N=22; 7.0%) as well as the new H1N1v pandemic influenza subtype (N=13; 4.1%) had been circulating at the time of our investigations resulting in a PCR-adjusted positivity rate of 11.1% (95% CI 8.0 - 15.3). The positive predictive values for the RDTs used were 90.9% and 94.4%, respectively. Conclusions: This study provides a first insight into influenza epidemics in one of the most remote parts of Asia. Our findings suggest that respiratory illnesses due to influenza viruses are underreported in areas with limited access to health care and show a distinct seasonality also in rural areas of tropical countries.
Keywords: Bangladesh, H1N1, H3N2, influenza, rapid diagnostic tests
Influenza viruses are a major cause of respiratory diseases and increased mortality in both the developed and the developing world.  However, information related to the prevalence and epidemiology of influenza-related diseases is limited, especially from rural areas of developing tropical countries where the situation seems to differ considerably from that in temperate zones of the world. , Several authors suggest that influenza seasonality in tropical countries may be less defined compared to temperate zones. ,,,, Neither of these reports covers rural, sparsely populated areas where access to health care is limited. ,,,
Particularly in remote settings where access to laboratory infrastructure is limited, rapid diagnostic testing (RDT) can significantly improve access to influenza diagnostics.  However, laboratory tests, such as polymerase chain reaction (PCR) remain essential for a definite diagnosis and characterization of influenza. ,
The aim of this study was to provide an insight into the influenza situation in the remote regions of south-eastern Bangladesh to identify the circulating subtypes, collect epidemiological data, and to determine the potential role of RDTs for monitoring influenza epidemics.
Collection of samples, rapid diagnostic testing and case definition
Bandarban is a remote and sparsely populated hill district in south-eastern Bangladesh, bordering Myanmar. All investigations were conducted at the local district hospital.
As part of our research assessing the causes of febrile illnesses in the Chittagong Hill Tracts, nasopharyngeal swabs from patients with signs and symptoms of influenza-like illness (ILI) were collected from 2008 onwards and analyzed using two independent rapid diagnostic tests (Clearview Exact Influenza A & B, Applied Biotech Inc., San Diego, California, USA and BinaxNOW® Influenza A & B, Binax Inc., Scarborough, Maine, USA). Both systems are FDA-cleared lateral flow immunochromatographic assays for qualitative detection of Influenza A and B nucleoprotein using monoclonal antibodies.
Nasopharyngeal sample material was obtained using standard swabs for virology specimens (Virocult® MW 950/974/975, Medical Wire & Equipment Co. Ltd., Corsham, Wilts., England). After collection, the material on the sample swabs was dissolved in 0.5 - 1.5 ml saline (for BinaxNOW® ) or buffer solution included in the test kit (for Clearview) and tested at room temperature within 10 minutes. Both rapid diagnostic tests were performed according to manufacturer's instructions. Every new batch of RDTs was checked for test reliability using positive and negative controls provided by the manufacturer.
The study protocol was approved by an independent institutional ethics committee in Bangladesh. Written informed consent was obtained from all study participants or from their legal representatives prior to enrolment.
The definition of ILI remains controversial. We therefore decided to use a classification pooling the European Centre for Disease Prevention and Control (ECDC) definition for both, acute respiratory infections and influenza-like illnesses with fever as our main enrollment criterion. The case definition for ILI used in this study was sudden onset of fever ≥37.5°C plus at least one of the following symptoms: Headache, musculoskeletal pain, cough and running nose.
As malaria remains a major cause of febrile illnesses in the Chittagong Hill Tracts, every patient was also tested for malaria.
Laboratory methods for viral detection
From patients with a positive RDT result, a second nasopharyngeal sample was obtained for confirmatory testing. Specimens were stored in liquid nitrogen or at -20°C and transferred to the Austrian National Influenza Reference Centre for virus subtyping using reverse transcriptase real-time PCR (RT-PCR) following established standard procedures. 
An anonymized MS Excel database for data entry and storage was created and double checked. Microsoft Excel and VassarStats (http://vassarstats.net/) were used for data analysis. Pearson΄s Chi-square with Yates' correction or Fisher΄s Exact test was used for categorical data. Mann-Whitney U-test was performed for comparing continuous data that did not conform to normal distribution.
Patient characteristics are shown in [Table 1]. Based on patient history, none of the participants had been vaccinated against influenza. Out of a total number of 314 febrile patients included in the febrile illnesses surveillance between January 2008 and December 2009, 78 (24.8%; 95% confidence interval (95% CI) 20.2-30.1) patients fulfilling our criteria for ILI were screened for influenza, and 38 (12.1%; 95% CI 8.8-16.4) patients were diagnosed as influenza positive based on positive RDT results. Based on molecular analysis, 35 (11.1%; 95% CI 8.0-15.3 equivalent to 44.9%; 95% CI 33.7-56.5 of all patients screened by RDT) patients were classified as true positive for influenza A. Molecular analysis shows that all 22 patients identified during the first outbreak in May 2009 carried a seasonal H3 strain. During the second outbreak in September/October 2009, another 13 patients were identified who all carried the newly identified H1N1v subtype. None of the patients carried both viruses at any time.
All patients testing positive for influenza were less than 40 years old (mean age: 11.5 years, standard deviation (SD) 10.2 for the first and 13.6 years, SD 6.9 for the second outbreak).
During the second outbreak, patients with signs and symptoms of ILI tended to seek medical care earlier than at the time of the first influenza peak (mean days of fever 2.4 vs. 3.8; P=0.038).
None of the patients presented with life-threatening illness due to influenza and all cases were self-limited. None of the patients received antivirals.
Among H3 positive patients, the most common symptoms were weakness, cough, headache, dizziness and sore throat. H1N1v positive patients showed a higher percentage of chills and gastrointestinal symptoms (diarrhea, vomiting).
No significant differences were found between the two subtypes in terms of body temperature and respiratory rate.
Each patient presenting with ILI was tested in a BinaxNOW® Influenza A & B and/or Clearview Exact Influenza A & B. Out of 36 patients positive by Clearview, 34 tested positive in PCR (22 seasonal subtype H3, 12 pandemic subtype H1N1v). The positive predictive value (PPV) of this test system was 94.4%. The PPV for BinaxNOW® was 90.9% (20 out of 22 patients true positive by PCR). The subtype distribution was 19 H3 and 1 H1N1v, respectively.
Influenza cases in context with climate
To determine whether climatic conditions (temperature, rainfall) have any influence on influenza seasonality and epidemiology, we compared our results with data obtained from the Bangladesh Meteorological Department in Dhaka.
Patient data suggest two influenza peaks in 2009 [Figure 1]. The first outbreak in May 2009 coincided with the beginning of the monsoon season and the worldwide spread of the influenza H1N1v subtype. Nevertheless, all influenza infections were due to the seasonal subtype H3. The second peak was observed at the end of the rainy season in Bangladesh in September-October 2009. All patients positive for influenza were found to carry the H1N1v pandemic influenza subtype.
This study was conducted as part of a febrile illnesses surveillance study to investigate influenza-related respiratory diseases in the Chittagong Hill Tracts. To our knowledge, this is the first report of influenza activity in this remote part of the world.
An interesting observation was that all of the patients tested positive for influenza were below 40 years of age, which is in contrast to the common u-shaped age distribution (very young, elderly) found in temperate zones of the world.  The shorter duration from onset of symptoms to the time of seeking health care during the second outbreak with pandemic influenza may be a result of higher awareness regarding influenza or due to more severe illness (e.g. higher percentage of chills, headache, weakness and gastrointestinal symptoms). A potential indication for more severe illness while carrying H1N1v was the higher percentage of patients taking medication (paracetamol and/or antibiotics).
The results of our study show that influenza is a common burden in rural areas of developing countries, especially at the beginning and at the end of the rainy season. More data will be needed to define whether the second outbreak in September - October was only due to the rapid spread of the new subtype H1N1v affecting the whole country at that time  or is an annually recurring event.
Considering the limited availability of meteorological data and the limited number of influenza-positive cases seen in this study, we decided to restrict ourselves to mere observation. However, our findings suggest that influenza-related diseases show a distinct seasonality not only in temperate climates but also in tropical countries, which is consistent with several previous reports and empirical observations in many tropical countries. ,,,,,,, As the epidemiology of influenza in tropical countries is still not fully understood, extending active influenza surveillance programs to rural settings, including more detailed data on climatic conditions, is vital for the development of effective regional prevention strategies as well as epidemic and pandemic control measures. 
The sensitivity of influenza rapid antigen tests has previously been reported as being lower than expected. This may at least in part be attributable to methodological issues, e.g. use of frozen or refrigerated samples or culture supernatant. ,, This could potentially lead to DNA degradation and therefore lower sensitivity when testing these samples.  Fresh sample material seems to result in sensitivities comparable to the manufacturer's information. , However, even if influenza RDTs show a lack of diagnostic sensitivity compared to more elaborate laboratory methods, they are an economical, easy-to-use, and fast method to determine influenza outbreaks at an early stage while showing excellent characteristics for deployment in areas with limited medical infrastructure. ,,
An important limitation of the present study is that not all patients with signs and symptoms of influenza-like illness seek medical care resulting in a relatively small number of observations. Also, RDT sensitivity could not be determined as RDT-negative patient samples could not be investigated by PCR due to limited cold chain capacities. Nevertheless, the present study provides valuable data from a rural area of a tropical developing country where access to gold-standard detection methods like PCR or viral culture is significantly limited.