The consolidated dengue cohort was recruited between August 2014 and March 2015. A total of 3,514 children and adolescents aged two to 16 years were randomly selected from the population of Araraquara at the time of recruitment. From December 2016 to August 2018, subjects from the original cohort who presented ILI were included in the influenza cohort and prospectively followed during two consecutive influenza seasons. In 2017 and 2018, the winter started on June 21st in the southern hemisphere. Sample size was estimated based on the frequency of febrile episodes with respiratory symptoms observed in the first year of the original (dengue) cohort. During the 20-month period of the study, we hypothesized that 1,200 cases of febrile syndrome followed by respiratory symptoms would occur, of which 5% would be influenza with virological confirmation.9 Thus, at least 60 cases of proven influenza infection would be available for analysis.

ILI was defined by the presence of fever and two or more of the following signs and symptoms: cough, coryza, tonsillitis, pharyngitis, tachypnea, dyspnea, myalgia, headache, inappetence or prostration appearing within seven days. Proven influenza infection was defined by the occurrence of ILI with a laboratory confirmation of influenza A or B. Lymphopenia was defined as an absolute lymphocyte count (ALC) below 1.0×106 lymphocytes/mm3. Leukocytosis was defined as an elevated white blood cell (WBC) count greater than 11,000per mm3.

After enrollment, parents or guardians received a weekly phone call for fever surveillance. In the case of a temperature >37.5°C and respiratory symptoms, a home visit was scheduled to confirm ILI and collection of oral and nasopharynx swabs. In addition to demographic data, the following information was assessed during home visits: type of respiratory symptoms, date of initial symptoms, presence of symptomatic household contacts, presence of comorbidities, need of hospitalization and information on influenza vaccination in that year. All information was transcribed to case report forms (CRF). Children with laboratory-confirmed influenza or with more pronounced symptoms were referred to a pediatrician by the field team. Oseltamivir was introduced at physician discretion.

Respiratory virus diagnosis was performed at the Virology Laboratory, Institute of Tropical Medicine (University of São Paulo School of Medicine). Influenza A or B was diagnosed by nucleic acid test (NAT), using the Cepheid (Xpert® Xpress Flu, Sunnyvale, CA, USA) platform, or the XGen Influenza multiplex kit (Mobius Life Science, Pinhais, PR, Brazil) according to manufacturer’s instructions. Due to its rapid turnaround time, the Cepheid platform was used first, to assure early introduction of oseltamivir, at physician discretion. Aliquots of respiratory samples were frozen and stored at −80° C for further diagnosis of other pathogens using a multiplex platform (XGen Respiratory Panel 21 pathogens multiplex kit (Mobius Life Science, Pinhais, PR, Brazil), according to manufacturer’s instructions.

Sequencing was proposed to evaluate the match between circulating influenza B viruses and the lineages included in the trivalent (TIV) and quadrivalent (QIV) influenza vaccines recommended during the study. Primers for the hemagglutinin (HA) region were designed for INF B (FluBHA_F1-TTGGAACCTCAGGRTCTTGC or FluBHA_F2 – TCCTATAATGCACGAYAGAACA and FluBHA_R-TGCAGGAGGTCTATATTTGGTTC). Fragments were sequenced using Sanger method and sequences were built and analyzed in the CLC GenomicsWorkbench (Qiagen-https://www.qiagenbioinformatics.com/products/clc-genomics-workbench/). The visualization and obtaining of the information described in the report were done with the aid of the CLC. Reference sequences of Influenza vaccine strains were obtained from the GISAID - Global Initiative on Sharing All Influenza Data (https://www.gisaid.org/). These were aligned together to worldwide references samples from 2015 to 2018 available at GenBank and also to sequences generated in this study.

We first looked for amino acid differences between our samples and the vaccine strains. Then we reconstructed a phylogenetic tree from partial HA gene (47 sequences with 616 nucleotides long) to investigate the epidemiology of the circulating strains. Phylogenetic reconstructions were performed using the Maximum Likelihood method in the PhyML program within the SeaView package10 using the best nucleotide model (HKY+I) chosen by ModelTEst.11

In 2016, the components of influenza vaccine for the southern hemisphere were A/California/7/2009 (H1N1)pdm09, A/Hong Kong/4801/2014(H3N2) and B/Brisbane/60/2008 (B/Victoria lineage). The strain A/Michigan/45/2015(H1N1)pdm09 replaced the strain A/California/7/2009(H1N1)pdm09 in 2017. The quadrivalent influenza vaccine (QIV) also including the lineage B/Phuket/3073/2013-like virus (B/Yamagata lineage) and was only available in private clinics.

The study was approved by the Ethics Committee of the Faculty of Medicine of the University of São Paulo (CAAE no. 25706913.6.1001.0065). The selected households were visited by the project’s field team. The parents or guardians were asked to sign additional informed consent forms, specific for the influenza study. The child and/or the adolescent were requested to sign the Term of Assent, in accordance with the current legislation.

Patient characteristics were summarized using descriptive statistics. Differences between categorical and continuous variables were calculated using Pearson’s chi square, Student’s t-test and Mann Whitney tests, as appropriate. In all statistical analyses, a two-sided p-value≤0.05 was considered statistically significant (IBM SPSS statistics version 21 for Windows).

The study started on December 5, 2016, when the first case of ILI was included, and ended on August 30, 2018. Thus, the study cohort was followed for 20 consecutive months, including two winters. One hundred and seventy-nine patients were included, 99 males (55.3%) and 80 females (44.7%), median age 10 years, ranging from eight months to 19 years of age.

The patients had 277 episodes of ILI during the study period, with a median of one episode per patient, ranging from one to four episodes. Thirty-three patients had two episodes of ILI, eight had three episodes and two had four episodes of ILI. As expected, an increase in the number of episodes of ILI was observed in winter months, although a second wave of ILI was observed in the spring of 2017 (arrow, Fig. 1). In addition to fever, the most frequent symptoms were headache, observed in 233 of the 277 ILI episodes (84.1%), somnolence (in 208 episodes, 75.1%), cough (in 204 episodes, 73.6%) and coryza (in 193 episodes, 69.7%). Hemogram was performed in 202 of the 277 episodes of ILI (73%). Median leukocyte and lymphocyte counts were 6,200 (1,800–31,300)/mm3 and 2,419 (708–4,602)/mm3, respectively. Leukocytosis was observed in 121 (60%) and lymphopenia in four (2%) of the ILI episodes. Children were referred to the pediatrician in 87 of the 277 episodes of ILI (31.4%). No patient with ILI needed hospital admission.

Fig. 1.

Seasonality of respiratory pathogens diagnosed during influenza-like illness episodes. Dotted line represents influenza-like illness episodes.

(0.23MB).

Using the influenza platform (Cepheid Xpert® Xpress Flu) we identified 90 episodes of proven influenza infection, 73 influenza A (81.1%) and 17 by influenza B (18.9%). During laboratory work up with the multiplex platform to identify other respiratory pathogens, nine additional influenza infections were diagnosed (three influenza B and six influenza A). Thus, 90 of the 179 patients (50.3%) had 99 episodes of proven influenza infection, being 79 caused by influenza A (79.8%) and 20 by influenza B (20.2%). Nine patients had two episodes of influenza. Two patients had two episodes of INF A in the same year; four had two episodes of INF A in consecutive years; one patient had two episodes of INF B in the same year; and three patients had two separate episodes of influenza A and B in the same year (one patient), or in consecutive years (two patients).

In general, the clinical presentations of influenza A and B infections were similar and regularly mild, as shown in Table 1. Patients with influenza B were more likely to be male (85%, p=0.01) and presented more prostration than patients infected with influenza A virus (p=0.01). Among the 87 outpatient visits to the pediatrician, 61 were due to influenza (70.1%). Interestingly, the number of visits was significantly higher for INF A episodes in comparison with INF B (67.1% vs 40%; p=0.0003). Oseltamivir for five days was introduced only in one case of influenza A.

One hundred and thirty-eight of the 277 episodes of ILI (49.8%) tested negative by all techniques. In the remaining 139 episodes (50.2%), at least one respiratory pathogen was identified. Influenza A and B viruses were the most frequent agents identified (99 episodes) followed by human rhinovirus (22 episodes). Table 2 shows the respiratory pathogens diagnosed during the study. Fig. 1 shows the seasonality of ILI and of the respiratory pathogens. The second wave of ILI episodes observed in 2017 was mainly due to influenza B infections.

Information on influenza vaccination was obtained during home visits, when respiratory samples were taken. In 49 of the 277 episodes of ILI (17.7%), the information about influenza vaccine was not available. In the remaining 228 episodes of ILI, 85 had received the trivalent vaccine (37.3%), and 143 had not been vaccinated (62.7%). Excluding the cases without vaccine information, no difference was seen in the occurrence of influenza in vaccinated and non-vaccinated patients (p=0.47). According to the country recommendation, seven children were within the age range of influenza vaccine. In two of them, the vaccination status was unknown, and out of the remaining five, four (80%) had been vaccinated.

From the 20 cases of influenza B, 12 sequences could be generated and were used for analysis. The amino acid analysis showed that while 11 strains were related to the Phuket (B/Yamagata lineage), one of the circulating virus (patient BR_776_June_2017) was similar to the Brisbane virus (Victoria lineage) (data not shown). This was also noticeable through the phylogenetic reconstruction (Fig. 2) where the two main clusters comprised strains related to either Brisbane or Phuket lineages sampled at the same location and years, indicating co-circulation of both strains. The phylogeny also revealed that there was little difference among the Brazilian, European and USA circulating influenza B viruses since they all clustered together with no particular geographic pattern. A slight temporal structure can be observed in the tree, with viruses sampled in September and October clustering together, and samples from November and December 2017 clustered together to viruses sampled in 2018 (from other countries). In comparison to the vaccine lineages (Brisbane_60 and Phuket_3073), we observed that six of the eleven patients (54.5%) in the Phuket branch of the phylogenetic tree had not received influenza vaccine, one (9%) had unknown vaccination status, and four (36.3%) had received the TIV, which contained only the Brisbane lineage of influenza B virus. The only patient with influenza B in the Brisbane branch had not been vaccinated.

Fig. 2.

Maximum likelihood phylogenetic tree of influenza B HA. Black dots demonstrate the clinical samples from this study. Yamagata lineage and Victoria lineage are indicated by Y and V letters aside the respective clusters. Bootstrap values above 80% are depicted in the key nodes.

(0.48MB).