Takahashi et al.

Tropical Medicine and Health (2025) 53:51 Tropical Medicine

https://doi.org/10.1186/s41182-025-00711-x
and Health

RESEARCH Open Access

Community‑based pathogen‑specific
incidence of influenza‑like illness due
to respiratory viruses in South‑central Vietnam
in 2009–2012: after a pandemic of influenza
A viruses
Kensuke Takahashi1,2*, Shinya Tsuzuki3,4,5, Minh Nhat Le6, Nguyen Hien Anh7, Dang Duc Anh7,
Koya Ariyoshi8 and Lay‑Myint Yoshida9,10

Abstract
Background Influenza-like illness (ILI) is one of the most common illnesses caused by various respiratory viruses
and directly or indirectly incurs high expenses to households. However, the pathogen-specific incidence and health-
seeking behaviour in communities have not been well described.
Methods A longitudinal cohort study using a self-recorded health calendar among 1000 households was performed
in South-central Vietnam from October 2009 to September 2012. Endemic respiratory viruses in the community were
monitored using random sampling in public health clinics (polyclinics). The monthly incidence of specific pathogens
was calculated using the Bayesian method.
Findings Among 5,016 household members, 3,687 ILI episodes were reported during the study period. The inci‑
dence rate of ILI was 21.7 (95% confidence interval 21.0–22.4) per 1,000 person-months for all ages and highest
in children under 2 years with 71.6 (64.7–81.8) followed by 2–4 years with 71.3 (65.8–78.2). Rhinovirus had the high‑
est incidence with 22.5 among the age under 2 years, followed by adenovirus and respiratory syncytial virus (RSV)
with 12.5 and 9.9, respectively. Most young children sought treatment from clinics and hospitals, whereas most
schoolchildren and adults sought treatment from drugstores. RSV outbreaks significantly increase the number
of healthcare visits among children under 2 years, but not in older age groups.
Interpretation Several surges of ILI were attributed by multiple respiratory viruses. Healthcare seeking patterns were
varied among pathogens. Highly transmissible viruses, such as rhinovirus and adenovirus, pose the potential risk
of the next pandemic.
Keywords ILI, Respiratory virus, Incidence, Healthcare-seeking behaviour, Influenza types A and B

*Correspondence:
Kensuke Takahashi
[email protected]
Full list of author information is available at the end of the article

© The Author(s) 2025. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/.
Takahashi et al. Tropical Medicine and Health (2025) 53:51 Page 2 of 11

Introduction Methods
Influenza-like illness (ILI) is defined by the World Health Health calendar study
Organization as a fever accompanied by upper respira- We conducted a prospective cohort study in Nha Trang
tory symptoms, such as cough, sore throat, or rhinor- City, Khanh Hoa Province in South-central Vietnam,
rhoea. It is one of the most common illnesses and causes between October 2009 and September 2012. This is a
substantial morbidity and mortality especially in young tropical area with a rainy season, generally during from
children and elderly worldwide [1–7]. June to September. Nha Trang consists of 22 communes
Most ILI cases are self-limiting and show low mortality (administrative divisions similar to municipalities), with
rates; however, due to the high incidence among popula- 14 in urban areas and 8 in suburban areas. Two com-
tions, the economic burden of this illness is high. Gerry munes from urban areas and two from suburban areas
et al. reported that annual direct costs for hospitalisa- were randomly selected, and households were chosen
tion and emergency departments due to influenza in the using census data and a random number table, ensuring
United States were up to $163.25 million and $278.50 roughly equal numbers of households from each com-
million, respectively. [2] Considering not only the direct mune by enrolling only those who provided consent.
costs (e.g., fees for healthcare visits and medications) but We obtained epidemiological information about the
also the indirect costs (e.g., time off from work or recrea- households and their members at the beginning of the
tion and additional childcare), the total economic burden surveillance period. The sex of the participants was iden-
of ILI is much higher than that estimated only from the tified as a reported biological factor. Health calendars
direct costs. [2, 7–9] were distributed to these households to check personal
Various respiratory viruses, including influenza, respir- daily events of illness (cough, fever, sore throat, diar-
atory syncytial virus (RSV), adenovirus, human metap- rhoea, and body pain) and healthcare-seeking behav-
neumovirus (hMPV), parainfluenza virus, bocavirus, and iour concerning whether they visited a hospital, public
coronavirus, are known causative pathogens of ILI [3, 7, health clinic (polyclinic), private clinic, or community
11–18]. However, except for specific viruses such as cor- health centre, or just took medication at home. Fever was
onavirus disease-2019 (COVID-19), influenza, and RSV, defined as an axillary temperature of 37.5 °C or higher,
pathogen-specific incidences are not well known due to as recorded in the household health calendar. House-
lack of access to diagnostic tools [7, 18–22]. Several trials hold members were instructed to measure the axillary
for estimating the incidence of respiratory viruses with temperature using a thermometer whenever fever was
meta-analysis or modelling, and most etiological sur- suspected. Changes in family composition, such as new-
veillance of respiratory viruses have been conducted in born babies, deaths, moving in and out of each house-
hospital settings and focus on severe cases. However, the hold member, and long-staying visitors, were recorded
community burden of ILI due to respiratory viruses has monthly (Supplementary Table 2, Supplementary Fig. 1).
often been neglected and a few community-based sur- Calendars were distributed to each household from
veillances of pathogen-specific incidence of ILI in tropi- October 2009 to September 2012. At the beginning of the
cal areas has been conducted [7, 23]. Ngoc et al. reported study, a trained healthcare worker visited each household
the household incidence of ILI and identification of 15 and explained how to record the calendar. Healthcare
respiratory viruses in closed cohort study in Northern workers visited households every month in the first year
Vietnam; however, due to the limitation of study design, and every 2 months in later years to collect calendars and
age-distributed incidence of ILI nor pathogen-specific check if the data were properly recorded.
incidence could not be reported. [7]
During 2009–2010, a new variant of influenza A Polyclinic data
(H1N1) virus, namely, the 2009 pandemic H1N1 virus The only two polyclinics serving the four target com-
(pH1N1), spread all over the world [24]. Ten years later, munes were selected for this study. We collected a list of
we struggled again with the COVID-19 pandemic [25, patients with ILI symptoms who visited polyclinics (nos.
26]. The pattern of viruses circulating in the community 1 and 2) from the four targeted communes between Janu-
changes over time [3]. Knowing the characteristics of ary 2010 and September 2012, including information on
respiratory viruses is important in preparing for the next age, sex, date of onset, address, and clinical diagnosis.
pandemic of respiratory viruses. Nasopharyngeal swabs were collected from the first 10
The aim of this study was to estimate the pathogen- eligible patients < 5 years with ILI symptoms who visited
specific incidences of ILI cases in the community in the polyclinics each month. This approach was chosen to
South-central Vietnam after the pH1N1 influenza A pan- monitor circulating viruses in the community in a cost-
demic and to reveal the healthcare seeking behaviours of effective manner. Based on previous studies showing
a community during the outbreak of specific viruses. higher viral detection rates in children aged < 5 years,
Takahashi et al. Tropical Medicine and Health (2025) 53:51 Page 3 of 11

we focused on this age group for virological analysis [7]. where ILI kspoly indicates the number of sampled cases of
Swabs were collected in 1.5 ml microtube containing 600
ILI with a positive PCR for pathogen k and sk indicates
μL of Skim Milk–Tryptone–Glucose–Glycerol (STGG)
sensitivity of the test for detecting each pathogen k.
media and transported in an icebox to a microbiology
Equation (2) allows us to derive the posterior distribution
laboratory at Khanh Hoa General Hospital (KHGH) on
of the probability product sk × pk (a, t), which, assuming
the same day, processed in a standard way, and stored
a uniform prior between 0 and 1, is equal to the beta
at − 80 °C. RNA was extracted using the QIAamp Viral
distribution:
RNA Mini Kit (#52,906) at the National Institute of
Higiene and Epidemiology (NIHE) and respiratory k
sk × ppoly (a, t) ∼ Beta(α(a, t), β(a, t))
viruses were tested by multiplex polymerase chain reac-
tion (PCR) assays at Nagasaki University [10]. Multiplex With
PCR was designed to detect 13 respiratory viruses, influ-
enza A and B, RSV, rhinovirus, adenovirus, coronavirus k
α(a, t) = ILIspoly (a, t) + 1
229E, coronavirus OC43, bocavirus, hMPV, and parain-
fluenzavirus serotype 1–4 [11, 27]. Outbreak seasons of k
each virus were defined as 10% or more positive results β(a, t) = ILIspoly (a, t) − ILIspoly (a, t) + 1 (3)
among tested samples. Panelised logistic multivariate
Since we needed pk (a, t) in Eq. (1), we had to assume
regression models were used to estimate the relationship
between outbreak seasons and healthcare facility visits. a strong prior on sk to avoid identifiability issues on
pk (a, t). Using a Bayesian approach, we integrated the
uncertainty of the sampling procedure and the test sensi-
Calculation of pathogen‑specific incidence
tivity (sk) in the regression. Here, sk represents the uncer-
in the community
tainty in the observed test results arising from both the
Pathogen-specific monthly incidences of viruses were
sampling procedure and the inherent test sensitivity of
calculated as observed ILI cases in the community mul-
the multiplex PCR assay. By adopting a Bayesian frame-
tiplied by the ratio of circulating viruses obtained in
work, we assumed that the test outcomes (0, 1) followed
polyclinics. Bayesian methods were used to remove the
a beta distribution. This approach allowed us to account
stochastic effects of sampling:
for the variability and uncertainty associated with sk,
k
ILIcom k
(a, t) = ILIcom (a, t) × ppoly (a, t) (1) ultimately presenting the results with confidence inter-
vals that reflected this uncertainty. The regression out-
where a is the age group, t is time in months, k indicates come was the posterior distribution of the number of ILI
each respiratory virus, ILIcom is the monthly number of patients infected with pathogen k in the community.
ILI cases in the community and ILI kcom indicates the path- The monthly incidence of ILI was described from
ogen-specific incidence that have not sought medical October 2009 to September 2012, and the pathogen-spe-
ILI kpoly (a,t) cific incidence was calculated from January 2010 to Sep-
care (mild ILI cases), and ppoly
k
(a, t) = ILI poly (a,t) is the tember 2012.
proportion of ILI cases infected with pathogen k amongst
polyclinic patients. Among individuals who visited poly- Seasonality of circulating viruses and health seeking
clinics and were diagnosed with ILI each week, we behaviour
obtained 10 nasopharyngeal swabs from randomly The outbreak season for each virus, specifically influenza
selected patients < 5 years from four target communes. A, influenza B, and RSV, was defined as when the calcu-
We assumed that the monitored households were repre- lated incidence exceeded 10 cases per 1,000 population.
sentative samples from those four communes, which For each age group, the odds of visiting any healthcare
allowed us to link the circulation of pathogens among ILI facilities during the outbreak were calculated. These odds
patients from the four communes who visited the poly- were adjusted based on the outbreak status of other cir-
clinics with the number of childhood ILI in those house- culating viruses during the same period, allowing for a
holds. To reflect the uncertainty of sampling in ppoly k
in clearer understanding of health-seeking behavior in rela-
Eq. (1) we employed Bayesian regression techniques and tion to specific viral outbreaks.
assumed the following:
  Data management

­FoxPro® 7. Unknown or unclear information were con-

k
ILIspoly k
(a, t) ∼ Binomial ILIspoly (a, t), sk × ppoly (a, t) Calendar data were entered and managed using Visual
(2)
firmed via phone calls. Daily data were converted using
Takahashi et al. Tropical Medicine and Health (2025) 53:51 Page 4 of 11

the following criteria: (a) ILI was defined as a fever and (30%), followed by manual workers (24%) and farmers/
cough or sore throat. We considered that symptoms fisheries (24%).
developed within 7 days belonged to the same episode of At the time of selection, the sample population was
illness; (b) if one person visited multiple facilities on the 4,696, which increased to 4,716 at the beginning, and
same day, the higher level of facility was applied (order 4,755 at the end of the surveillance. Monthly movements
of facility level: KHGH > other hospital > private clinic of households were monitored (Supplementary Table 2
> polyclinic > Community health centre (CHC) > drug and Supplementary Fig. 1). Eleven households consist-
store); and (c) dates when symptoms started and ended ing of 59 members dropped out during the study period;
were defined as first and final dates of symptom in one six moved, four refused to participate, and one who lived

were transferred to STATA​®14 and ­R® (ver. 4.1.2) for fur-

chain of symptoms, respectively. Spreadsheet format data alone died. There were 148 newborns and 61 deceased
members, and 116 members moved in and 93 moved out.
ther analysis. Among deceased cases, reasons for deaths were specified
in 44 cases: 13 senile cases, 10 malignant cases, 10 brain
Results strokes cases, four accident cases, three heart disease
Demographic data cases, three renal cases, and one asthma case.
The population in the four studied communes was 49,374
according to census data for 2010. Characteristics of Age‑distributed incidence of ILI in the community
selected households and their members at the beginning Several surges in ILI symptoms were observed every
of the study are shown in Supplementary Table 1. The 2–3 months (Fig. 1). The incidence rate of ILI was 21.7
median number of household members was 4.8 [inter- (95% confidence interval 21.0–22.4) per 1000 person-
quartile range (IQR) 4.0–6.0] persons/house. The median months for all ages. The highest incidence was observed
age of participants was 31 (IQR 17–46) years, which for those aged under 2 years with 71.6 [95% confidence
became 32 (IQR 18–48) years at the end of surveillance. interval (95%CI) 64.7–81.8] events/month/1,000 popula-
Half of the sample population graduated from second- tion, followed by those aged 2–4 years with 71.3 (65.8–
ary school or higher grades, and at least one person in all 78.2). The lower incidences were observed in 15–29 years
households could read and write. The majority of males and 30–45 age groups, with 14.0 (13.0–15.2), 13.0 (11.9–
aged 20–59 years worked as manual workers (49%), fol- 14.1) events/month/1,000 population, respectively, and
lowed by white-collar or office workers (23%). The major- the incidence gradually increased with age (Fig. 2).
ity of females aged 20–59 years worked as housewives

Fig. 1 Incidence of ILI per 1000 population-month in each age group. Several surges of ILI symptoms were observed every 2–3 months. Sharp
surges of ILI incidence were observed in age group under 2 years (undashed line), which were synchronized with 2–4 years (middle dashed line),
5–14 years (fine dashed line) and > 60 years (rough dashed line). No clear surges were observed in 15–59 years (dotted and dashed lines)
Takahashi et al. Tropical Medicine and Health (2025) 53:51 Page 5 of 11

Viral results
During the study period, 15,173 patients, including
12,772 (84.2%) children under 5 years of age, visited
polyclinics. Among these, 1,428 samples were collected
for further virological analysis (Supplementary Fig. 2).
Monthly numbers of positive cases are shown in Fig. 3
and the numbers of detected viruses are shown in Sup-
plementary Table 3. Rhinovirus was the most frequently
observed pathogen, accounting for 407 cases (28.5%), fol-
lowed by adenovirus with 157 cases (11.0%), RSV with
129 cases (9.0%), influenza A virus with 82 cases (5.7%),
and influenza B virus with 45 cases (3.2%). Co-infection
of respiratory viruses were seen in 172 patients. Among
Fig. 2 Incidence rate of ILI per 1000 person-months in each age them, 150 patients had co-infection of two respiratory
group. Vertical bars indicated 95% confidence intervals (CI). No
significant difference was observed between those in age group
viruses, 20 had three respiratory viruses and two had four
under 2 years and 2–4 years, 71.6 (95%CI 64.7–81.8) and 71.3(65.8– respiratory viruses. Among the patients with co-infection
78.2), respectively. The minimum incidence of ILI was 13.0 (11.9–14.1) of respiratory viruses, rhinovirus was the most frequently
in 30–44 years detected, with 127 (73.8%) samples, followed by adenovi-
rus with 102 samples (59.3%).

Healthcare‑seeking behaviour
Healthcare-seeking behaviour varied significantly by age Pathogen‑specific incidence of respiratory viruses
group. Younger children under 5 years of age most fre- and health seeking behaviour
quently visited public clinics and hospitals, with 22.8% of The monthly pathogen-specific incidence of major res-
ILI cases in those under 2 years and 18.2% in those aged piratory viruses is shown in Fig. 4. Highest incidences
2–4 years seeking care at polyclinics, while private clin- were observed in rhinovirus average around 20 episodes/
ics were also commonly used (39.8% and 38.4%, respec- months/1000 population, which raised up to 85.3 (95%CI
tively). In contrast, older children and adults increasingly 50.5–118.4) episodes/months/1000 population in maxi-
relied on drugstores, with 66.9% of cases in the 5–14 age mum amongst children under 2 years. Adenovirus, RSV,
group and over 79% in adults aged 30 years and above Influenza A virus, and Influenza B virus marked high
opting for this type of care. Polyclinic visits declined incidence in the order next to rhinovirus. No significant
sharply with age, accounting for only 8.1% and 4.2% of differences were observed between two age group.
cases in the 15–29 and 30–44 age groups, respectively Distinct seasonality was observed in influenza A
(Table 1). virus and influenza B virus and RSV. Influenza A virus
outbreaks were observed in August–September 2010,
December 2010 to January 2011, June 2012, and August
2012. Influenza B virus outbreaks were observed in

Table 1 Healthcare-seeking behaviour by each age group

Age group (years) ILI Healthcare facility, n (%)
Drugstores CHC Polyclinic Private clinic Hospital Other

<2 289 114 (39.4) 1 (0.3) 66 (22.8) 115 (39.8) 19 (6.6) 9 (3.1)
2–4 534 253 (47.4) 3 (0.6) 97 (18.2) 205 (38.4) 8 (1.5) 10 (1.9)
5–14 913 611 (66.9) 5 (0.5) 95 (10.4) 174 (19.1) 17 (1.9) 57 (6.2)
15–29 578 463 (80.1) 8 (1.4) 47 (8.1) 62 (10.7) 10 (1.7) 15 (2.6)
30–44 544 433 (79.6) 4 (0.7) 23 (4.2) 70 (12.9) 3 (0.6) 22 (4.0)
45–59 501 413 (82.4) 12 (2.4) 22 (4.4) 49 (9.8) 5 (1.0) 25 (5.0)
60 + 338 270 (79.9) 4 (1.2) 19 (5.6) 26 (7.7) 8 (2.4) 23 (6.8)
Total 3697 2,557 (69.2) 37 (1.0) 369 (10.0) 701 (19.0) 70(1.9) 161 (4.4)
CHC community health centre. Drugstores include going to a drugstore or taking drugs stocked in their households. A polyclinic is a public clinic, where the
consultation fee is basically free for children under 6 years, whereas private clinics offer paid services. Participants reported as “hospital” when they required admission
regardless of the length of stay. “Other” included nowhere to go, traditional medicine or herbal medicine
Takahashi et al. Tropical Medicine and Health (2025) 53:51 Page 6 of 11

Fig. 3 Monthly number of respiratory viruses identified in polyclinics. infa, influenza type A. infb, influenza type B. rsv, respiratory syncytial virus.
hmpv, human metapneumovirus. piv1–4, parainfluenza virus serotypes 1–4. rhino: rhinovirus. corona: coronavirus. adeno, adenovirus. boca,
bocavirus. Each bar indicates the monthly number of viruses identified in policlinics from January 2010 to December 2012

March–May 2010 and December 2010. Outbreaks of months, whereas less patients in age groups of 5–14 and
RSV were observed in April–August 2010, July–Octo- 15–29 years visited healthcare facilities. Among the age
ber 2011, and June–September 2012. In the rainy season, group of 45–59 years, the proportion of patients with
June–September, positive cases of influenza A and RSV ILI symptoms who visited healthcare facilities increased
increased but those of influenza B decreased. in influenza B virus outbreak months but decreased in
Total 15,173 patients visited polyclinics −1 and influenza A virus outbreak months, with adjusted odds
−2 from target population with symptoms of ILI. ratio of 3.0 (95%CI 1.5–5.8) and 0.4 (95%CI 0.2–0.8),
Among them, the number of visitors under 2 years was respectively (Table 2).
9344(61.5%) and 2–5 years was 3428(22.6%). Randomly
selected 1,428 samples were tested for respiratory viruses Discussion
(Supp. Table 3). Estimating the pathogen-specific incidence of ILI in the
During RSV outbreak months, patients with ILI community is challenging due to the limitations of data
symptoms in the age group under 2 years tended to sources. Researches based on health care facilities might
visit healthcare facilities [adjusted odds 2.0 (95%CI be underestimated, which do not include cases just tak-
1.2–3.5)] compared with those during non-outbreak ing medicine and staying home [28]. Incidences of some
Takahashi et al. Tropical Medicine and Health (2025) 53:51 Page 7 of 11

Fig. 4 Monthly pathogen-specific incidence per 1000 population of ILI. A incidence in < 2 years. B incidence in 2–5 years. InfA, influenza type A.
infB, influenza type B. RSV, respiratory syncytial virus. Left upper figure showed incidence of ILI in each age group from Jan 2010 to Dec 2012. The
other figures showed pathogen-specific monthly incidences of each pathogen calculated by Bayesian methods. After 1,000 times trials to obtain
incidence based on equations, median of incidences were shown in middle undashed line with 25–75% of distribution (shaded area) and 2.5–97.5%
(dashed line). Means of median incidences in each month were calculated and shown in middle of figures
Takahashi et al. Tropical Medicine and Health (2025) 53:51 Page 8 of 11

respiratory virus infections commonly diagnosed in Children under 5 years tended to visit healthcare facili-
clinical settings such as influenza RSV and coronavirus ties rather than just staying home.
disease 2019 (COVID-19) were well reported but other Rhinovirus is one of the leading causes of viral bronchi-
respiratory viruses [7]. Sylvia T, et al. reported highest olitis in infants and the most common virus associated
incidence in rhinovirus as 29.78 per 100 population/year with wheezing in children aged 1–2 years. [23] Symp-
equal to 24.82 per 1,000 population per month, which is toms of ILI caused by rhinovirus infection are relatively
in agreement to this study [18]. Our result of adenovirus frequent but otherwise a mild, self-limited syndrome.
as second most identified pathogen was not compatible Therefore, its importance as a possible causal factor of
in other studies, which indicates that endemic viruses in severe illness has often been neglected. Multiple infec-
the communities would be varied depends on the area tions including rhinovirus were observed in numbers of
and timing of the studies. [18, 22] cases in our surveillance. This study did not analyse the
During the COVID-19 pandemic, various diagnostic severity of ILI cases. However, previous research has
tools were developed for COVID-19, but not for other shown that co-infections involving respiratory syncyt-
respiratory viruses. The combination of community ial virus (RSV) and other pathogens can exacerbate res-
surveillance using calendar-based records and molecu- piratory illness severity by intensifying inflammatory
lar biological surveillance in clinics enabled us to esti- responses [11, 27]. While we did not assess disease sever-
mate the community burden of each respiratory virus. ity in our cohort, these findings suggest that co-infections

Table 2 Adjusted odds ratio of any healthcare facility visit during the outbreak of specific viruses
Outbreak season Total HC visit* (%) Crude OR (95%CI) p value aOR⁑ (95%CI) p value
of each virus

0–1 years 289 189 (65.4)

Influenza A 56 36 (64.3) 0.934 (0.510–1.709) 0.824 0.868 (0.470–1.604) 0.651
Influenza B 58 37 (63.8) 0.909 (0.501–1.649) 0.754 0.771 (0.417–1.428) 0.408
RSV 96 72 (75.0) 1.927 (1.122–3.309) 0.018 2.021 (1.162–3.512) 0.013
2–4 534 301 (56.4)
Influenza A 110 53 (48.2) 0.661 (0.435–1.005) 0.053 0.654 (0.425–1.006) 0.053
Influenza B 74 40 (54.1) 0.896 (0.549–1.462) 0.659 0.930 (0.568–1.523) 0.774
RSV 164 92 (56.1) 0.984 (0.680–1.423) 0.930 1.069 (0.731–1.564) 0.730
5–14 913 284 (31.1)
Influenza A 170 53 (31.2) 1.008 (0.704–1.443) 0.965 1.142 (0.785–1.661) 0.487
Influenza B 106 38 (35.8) 1.280 (0.839–1.953) 0.251 1.286 (0.839–1.971) 0.249
RSV 280 68 (24.3) 0.622 (0.452–0.854) 0.003 0.603 (0.434–0.837) 0.002
15–29 578 123 (21.3)
Influenza A 83 11 (13.3) 0.541 (0.280–1.043) 0.066 0.626 (0.320–1.227) 0.173
Influenza B 75 20 (26.7) 1.429 (0.824–2.479) 0.204 1.614 (0.920–2.833) 0.095
RSV 220 35 (15.9) 0.585 (0.380–0.901) 0.015 0.597 (0.382–0.934) 0.024
30–44 544 100 (18.4)
Influenza A 130 25 (19.2) 1.087 (0.660–1.791) 0.744 1.132 (0.664–1.929) 0.649
Influenza B 45 9 (20) 1.162 (0.549–2.457) 0.694 1.183 (0.559–2.504) 0.660
RSV 202 36 (17.8) 0.947 (0.604–1.483) 0.811 0.908 (0.562–1.469) 0.695
45–59 417 84 (16.8)
Influenza A 101 9 (8.9) 0.443 (0.217–0.903) 0.025 0.395 (0.191–0.816) 0.012
Influenza B 46 15 (32.6) 2.736 (1.415–5.292) 0.003 2.957 (1.508–5.797) 0.002
RSV 175 31 (17.7) 1.114 (0.687–1.808) 0.661 1.238 (0.753–2.036) 0.399
60 + yr 338 56 (16.6)
Influenza A 75 11 (14.7) 0.856 (0.423–1.731) 0.666 0.865 (0.420–1.780) 0.693
Influenza B 42 5 (11.9) 0.699 (0.272–1.797) 0.457 0.707 (0.276–1.814) 0.471
RSV 129 21 (16.3) 0.974 (0.542–1.751) 0.930 1.006 (0.552–1.836) 0.983
*
HC, healthcare facility including hospital, polyclinic, private clinic, and community health centre. Odds ratio (OR) were calculated by penalized maximum likelihood
regression. (Stata command “firthlogit”). Crude OR of each pathogen indicates odds ratio to visit HC in outbreak seasons compared with those in non-outbreak
seasons. Reference numbers were omitted. ⁑aOR, adjusted odds ratio. Since outbreak seasons of each pathogen were duplicated in several months, OR were adjusted
by being the outbreak season of influenza A and B and RSV to know the independent likelihood of each pathogen
Takahashi et al. Tropical Medicine and Health (2025) 53:51 Page 9 of 11

may contribute to more severe outcomes. It is also pos- for each pathogen with confidence intervals. Accord-
sible that factors beyond severity, such as the timing of ing to our surveillance, a higher proportion of younger
viral outbreaks or accessibility of healthcare facilities, age groups with ILI symptoms visited polyclinics, so we
influenced healthcare-seeking behaviour in this study. believe that virus detection in polyclinics reflected the
Further investigation is warranted to explore how these circulating viruses in the target communities.
factors interact with severity to shape healthcare utiliza- Second, calendar-style health recording may lead to
tion patterns. recall bias. This method was effective for daily-based
There were several surges of ILI incidence linked with recording; however, when participants forgot to record
all age groups, although the highest incidences were in on time, our research staff needed to ask them to recall
the youngest age group of under 2 years. This might indi- ILI events, which may have introduced recall bias. This
cate that most ILI symptoms started from the younger issue could have been more pronounced in elderly-
generation and spread to the older groups. This study only households due to potential memory issues. How-
revealed that those surges could not be explained by a ever, such households were rare in this study, with only
single pathogen but a complex of multiple pathogens. two identified: one consisting of a couple aged 73 and
Transmission of respiratory infection is influenced by 78 years, and another with a 79-year-old individual living
temperature and humidity [29]. In the northern hemi- alone who dropped out in the first 3 months due to the
sphere, RSV activity starts in July in tropical areas and occupant’s death. Given the rarity of such cases, the over-
later with increasing latitude. In this study, the positive all impact on our findings was minimal.
ratio of influenza A and RSV increased in the rainy sea- In the beginning of the study, participants were moti-
son, whereas rhinovirus and adenovirus were observed vated and diligent in recording events, but in the late
all year, consistent with previous reports. [25, 29, 30] phase, reporting of ILI symptoms declined, potentially
During RSV outbreak seasons, there were increased leading to an underestimation of ILI incidence dur-
numbers of visits to healthcare facilities for the age group ing that period. To address these challenges, health-
younger than 2 years, possibly indicating that severe care workers conducted monthly or bi-monthly visits to
symptoms with RSV infection leads their carers to take review and update household health calendars, which
infected children to healthcare facilities rather than just helped ensure data accuracy and mitigate recall bias.
staying home or giving medicine from drug stores. Second, calendar-style health recording may lead to
During influenza B outbreak months, healthcare- recall bias. This method was good for daily-based record-
seeking behaviour among adults was higher than dur- ing; however, when participants forgot to record on
ing influenza A outbreaks. One possible explanation time, our research staff needed to ask them to recall ILI
for this observation is the lower incidence of influenza events. In the beginning of the study, participants knew
B compared to influenza A in prior seasons, which may the importance of the study and were keen to record the
result in reduced population immunity against influenza events, but in the late phase, reporting of ILI symptoms
B among adults. This decreased immunity could lead to declined. Therefore, ILI incidence in the late phase might
more pronounced symptoms when exposed to influenza be underestimated.
B, prompting higher rates of healthcare-seeking behav- Although this study detected 13 respiratory viruses,
iour. Future studies should investigate the immunological we focused on rhinovirus, adenovirus, RSV, influenza A,
profiles and co-infection dynamics in adult populations and influenza B for pathogen-specific incidence analysis.
during influenza outbreaks. The other viruses had a low detection frequency, mak-
There were several limitations in this study. First, ing it challenging to analyse their seasonality or outbreak
viruses that cause ILI symptoms in the sample popula- patterns reliably. This limitation highlights the need for
tion and those observed in polyclinics could differ. We larger-scale studies or extended monitoring periods to
could not get samples from the persons with ILI symp- capture the epidemiological dynamics of less frequently
toms in the community in our study setting. Instead, we detected viruses.
collected the samples in polyclinics and estimated the Even with these limitations, this study is valuable in
community circulating viruses by random sampling from showing the community burden of respiratory viruses
the patients with ILI symptoms. While many common in the pre-COVID-19 pandemic era. High incidences of
pathogens were observed across age groups in a simi- rhinovirus and adenovirus recall us their transmissibility.
lar community-based study in Laos [31], differences in Continuous monitoring of respiratory viruses in commu-
pathogen distribution by age highlight the limitation of nity is required for outbreak preparedness.
generalizing pathogen-specific ILI incidence from chil-
dren under five to the entire population in this study. The
Bayesian method enabled us to estimate positive rates
Takahashi et al. Tropical Medicine and Health (2025) 53:51 Page 10 of 11

Supplementary Information References

1. Wang D, Zhang T, Wu J, Jiang Y, Ding Y, Hua J, et al. Socio-economic bur‑
The online version contains supplementary material available at https://​doi.​
den of influenza among children younger than 5 years in the outpatient
org/​10.​1186/​s41182-​025-​00711-x.
setting in Suzhou, China. PLoS ONE. 2013;8(8): e69035.
2. Fairbrother G, Cassedy A, Ortega-Sanchez IR, Szilagyi PG, Edwards KM,
Additional file 1. Molinari N-A, et al. High costs of influenza: direct medical costs of influ‑
enza disease in young children. Vaccine. 2010;28(31):4913–9.
3. Marcos MA, Camps M, Pumarola T, Martinez JA, Martinez E. The role of
Acknowledgements viruses in the aetiology of community-acquied pneumonia in adults. Int
The authors gratefully acknowledge the individuals who helped to collect and Med Press. 2006;11:351–9.
analyse the data: Luu Trung Hieu, Le Ngoc Khanh and other staff members 4. Matthew B, Jhung MA, Carrie R, Alicia MF, Lina B, Lyn F. Influenza-like
from the Khanh Hoa Health Service, Nha Trang, Vietnam, and John Edmonds, illness, the time to seek healthcare, and influenza antiviral receipt during
Stefan Flasche and Anthon Camacho from London School of Hygiene and the 2010–11 influenza season — United States. J Infect Dis. 2014;4:535.
Tropical Medicine. 5. Madhi SA, Klugman KP. World Health Organisation definition of “radiolog‑
ically-confirmed pneumonia” may under-estimate the true public health
Author contributions value of conjugate pneumococcal vaccines. Vaccine. 2007;25(13):2413–9.
KT contributed conceptualisation, data curation, formal analysis, investigation, 6. World Health Organization. Human infection with pandemic (H1N1) 2009
methodology, project administration, resources, software, validation, visualisa‑ virus updated interim WHO guidance on global surveillance. Geneva:
tion, writing – original draft. ST contributed data curation, formal analysis, WHO; 2009.
software, supervision, validation, visualisation, and writing– review & editing. 7. Nguyen DNT, Mai LQ, Bryant JE, Hang NLK, Hoa LNM, Nadjm B, et al.
LNM contributed data curation, formal analysis, investigation, methodology, Epidemiology and etiology of influenza-like-illness in households in
resources, writing– review & editing. NHA contributed data curation, formal Vietnam; it’s not all about the kids! J Clin Virol. 2016;82:126–32.
analysis, investigation, methodology, resources, writing– review & editing. 8. Principi N, Esposito S, Gasparini R, Marchisio P, Crovari P. Burden of
DDA conceptualisation, project administration, resources, supervision, and influenza in healthy children and their households. Arch Dis Child.
writing– review & editing. KA contributed conceptualisation, funding acquisi‑ 2004;89(11):1002–7.
tion, methodology, project administration, resources, supervision, validation, 9. Simmerman JM, Lertiendumrong J, Dowell SF, Uyeki T, Olsen SJ,
writing– review & editing. LMY conceptualisation, data curation, formal Chittaganpitch M, et al. The cost of influenza in Thailand. Vaccine.
analysis, funding acquisition, investigation, methodology, project administra‑ 2006;24(20):4417–26.
tion, resources, software, supervision, validation, and writing– review & editing. 10. Yin JK, Salkeld G, Lambert SB, Dierig A, Heron L, Leask J, et al. Estimates
and determinants of economic impacts from influenza-like illnesses
Funding caused by respiratory viruses in Australian children attending childcare: a
This research was supported by Japan Program for Infectious Diseases cohort study. Influenza Other Resp Viruses. 2013;7(6):1103–12.
Research and Infrastructure, Japan Agency for Medical Research and Develop‑ 11. Yoshida LM, Suzuki M, Yamamoto T, Nguyen HA, Nguyen CD, Nguyen
ment (AMED) under Grant Number JP21wm0125006. The funding source AT, et al. Viral pathogens associated with acute respiratory infections in
had not been involved in the study design nor writing and editing this central vietnamese children. Pediatr Infect Dis J. 2010;29(1):75–7.
report. This study was funded by the Japan Agency for Medical Research and 12. Papadopoulos NG, Bates PJ, Bardin PG, Papi A, Leir SH, Fraenkel DJ, et al.
Development. Rhinoviruses infect the lower airways. J infect Dis. 2000;181:1875–84.
13. Falsey AR, Walsh EE. Viral pneumonia in older adults. Clin Infect Dis.
Data availability 2006;42(4):518–24.
No datasets were generated or analysed during the current study. 14. Falsey AR, Hennessey PA, Formica MA, Cox C, Walsh EE. Respiratory
syncytial virus infection in elderly and high-risk adults. N Eng J Med.
2005;352(17):1749–59.
Declarations
15. Xiang Z, Gonzalez R, Xie Z, Xiao Y, Liu J, Chen L, et al. Human rhinovi‑
rus C infections mirror those of human rhinovirus A in children with
Ethics approval and consent to participate
community-acquired pneumonia. J Clin Virol. 2010;49(2):94–9.
This work was approved by the Ethics Committees of the Institute of Tropical
16. Foulongne V, Guyon G, Rodière M. Human metapneumovirus infection in
Medicine, Nagasaki University, Japan, and NIHE, Hanoi, Vietnam.
young children hospitalized with respiratory tract disease. Pediatr Infect
Dis J. 2008;27(5):406–12.
Competing interests
17. Takahashi K, Suzuki M, Minh LN, Anh NH, Huong LTM, Son TVV, et al. The
The authors declare no competing interests.
incidence and aetiology of hospitalised community-acquired pneumonia
among Vietnamese adults: a prospective surveillance in Central Vietnam.
Author details
1 BMC Infect Dis. 2013;13(1):296.
Acute & Critical Care Center, Nagasaki University Hospital, Nagasaki, Japan.
2 18. Taylor S, Lopez P, Weckx L, Lazcano-ponce E, Kerdpanich A, Angel M,
School of Tropical Medicine and Global Health, Nagasaki University, Nagasaki,
et al. Respiratory viruses and influenza-like illness: epidemiology and
Japan. 3 Disease Control and Prevention Center, National Center for Global
outcomes in children aged 6 months to 10 years in a multi-country
Health and Medicine, Tokyo, Japan. 4 Faculty of Medicine and Health Sciences,
population sample. J Infect. 2017;74:29–41.
University of Antwerp, Antwerp, Belgium. 5 AMR Clinical Reference Center,
19. Thai PQ, Mai LQ, Welkers MRA, Le Khanh N, et al. Pandemic H1N1 virus
National Center for Global Health and Medicine, Tokyo, Japan. 6 Antimicrobial
transmission and shedding dynamics in index case households of a
Resistance Research Centre, National Institute of Infectious Disease, Tokyo,
prospective Vietnamese cohort. J Infect. 2014;68(6):581–90.
Japan. 7 Department of Bacteriology, National Institute of Hygiene and Epide‑
20. Lambert SB, O’Grady KF, Gabriel SH, Nolan TM. Respiratory illness during
miology, Hanoi, Vietnam. 8 Department of Clinical Medicine, Institute of Tropi‑
winter: a cohort study of urban children from temperate Australia. J
cal Medicine, Nagasaki University, Nagasaki, Japan. 9 Department of Pediatric
Paediatr Child Health. 2005;41(3):125–9.
Infectious Diseases, Institute of Tropical Medicine, Nagasaki University, Naga‑
21. Leder K, Sinclair MI, Mitakakis TZ, Hellard ME, Forbes A. A community-
saki, Japan. 10 Graduate School of Biomedical Science, Nagasaki University,
based study of respiratory episodes in Melbourne, Australia. Aust N Z J
Nagasaki, Japan.
Public Health. 2003;27(4):399–404.
22. Spencer JA, Shutt DP, Moser SK, Clegg H, Wearing HJ, Mukundan H, et al.
Received: 2 October 2024 Accepted: 14 February 2025
Distinguishing viruses responsible for influenza-like illness. J Theor Biol.
2022;545: 111145.
23. Vandini S, Biagi C, Fischer M, Lanari M. Impact of rhinovirus infections in
children. Viruses. 2019;11(6):1–13.
Takahashi et al. Tropical Medicine and Health (2025) 53:51 Page 11 of 11

24. Chowell G, Bertozzi SM, Colchero MA, Lopez-Gatell H, Alpuche-Aranda

C, Hernandez M, et al. Severe respiratory disease concurrent with the
circulation of H1N1 influenza. N Engl J Med. 2009;361(7):674–9. https://​
doi.​org/​10.​1056/​NEJMo​a0904​023.
25. Audi A, AlIbrahim M, Kaddoura M, Hijazi G, Yassine HM, Zaraket H. Season‑
ality of Respiratory Viral Infections: Will COVID-19 Follow Suit? Front Public
Heal. 2020;15(8): 567184. https://​doi.​org/​10.​3389/​fpubh.​2020.​567184.
26. Pammal RS, Kreinces JB, Pohlman KL. The importance of pharmacy part‑
nerships in effective COVID-19 vaccine distribution. Disaster Med Public
Health Prep. 2022;16(6):2293–5. https://​doi.​org/​10.​1017/​dmp.​2021.​178.
27. Yoshida L-M, Suzuki M, Nguyen HA, Le MN, Dinh VuT, Yoshino H, et al.
Respiratory syncytial virus: co-infection and paediatric lower respiratory
tract infections. Eur Respir J. 2013;42(2):461–9. https://​doi.​org/​10.​1183/​
09031​936.​00101​812.
28. Nolan T, Borja-Tabora C, Lopez P, Weckx L, Ulloa-Gutierrez R, Lazcano-
Ponce E, et al. Prevalence and incidence of respiratory syncytial virus and
other respiratory viral infections in children aged 6 months to 10 years
with influenza-like illness enrolled in a randomized trial. Clin Infect Dis.
2015;60(11):e80–9.
29. Li Y, Reeves RM, Wang X, Bassat Q, Brooks WA, Cohen C, et al. Global
patterns in monthly activity of influenza virus, respiratory syncytial virus,
parainfluenza virus, and metapneumovirus: a systematic analysis. Lancet
Glob Heal. 2019;7(8):e1031–45.
30. Nokes DJ, Okiro E, Ngama M, Ochola R, White LJ, Scott PD, et al.
Respiratory syncytial virus infection and disease in infants and young
children observed from birth in Kilifi District, Kenya. Clin Infect Dis.
2008;46(1):50–7.
31. Rudge JW, Nui I, Rebecca P, et al. Epidemiology and aetiology of
influenza-like illness among households in metropolitan Vientiane,
Lao PDR: A prospective, community-based cohort study. PLoS ONE.
2019;14(4):e0214207.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in pub‑
lished maps and institutional affiliations.