Estimated incidence of influenza‐associated severe acute respiratory infections in Indonesia, 2013‐2016

Background Indonesia's hospital‐based Severe Acute Respiratory Infection (SARI) surveillance system, Surveilans Infeksi Saluran Pernafasan Akut Berat Indonesia (SIBI), was established in 2013. While respiratory illnesses such as SARI pose a significant problem, there are limited incidence‐based data on influenza disease burden in Indonesia. This study aimed to estimate the incidence of influenza‐associated SARI in Indonesia during 2013‐2016 at three existing SIBI surveillance sites. Methods From May 2013 to April 2016, inpatients from sentinel hospitals in three districts of Indonesia (Gunung Kidul, Balikpapan, Deli Serdang) were screened for SARI. Respiratory specimens were collected from eligible inpatients and screened for influenza viruses. Annual incidence rates were calculated using these SIBI‐enrolled influenza‐positive SARI cases as a numerator, with a denominator catchment population defined through hospital admission survey (HAS) to identify respiratory‐coded admissions by age to hospitals in the sentinel site districts. Results From May 2013 to April 2016, there were 1527 SARI cases enrolled, of whom 1392 (91%) had specimens tested and 199 (14%) were influenza‐positive. The overall estimated annual incidence of influenza‐associated SARI ranged from 13 to 19 per 100 000 population. Incidence was highest in children aged 0‐4 years (82‐114 per 100 000 population), followed by children 5‐14 years (22‐36 per 100 000 population). Conclusions Incidence rates of influenza‐associated SARI in these districts indicate a substantial burden of influenza hospitalizations in young children in Indonesia. Further studies are needed to examine the influenza burden in other potential risk groups such as pregnant women and the elderly.


| INTRODUCTION
Influenza epidemics are estimated to result in about 3-5 million cases of severe illness, and 250 000-500 000 deaths globally each year, with the heaviest burden of disease among certain high-risk groups (children, elderly, and chronically ill). 1 Increasing awareness of influenza in Indonesia and other tropical countries has highlighted the need for surveillance to provide timely and high-quality epidemiology and virology data to enable monitoring of influenza seasonality, identify and monitor high-risk groups, establish influenza activity baseline levels, facilitate vaccine strain selection, and determine influenza burden to guide policymakers. 2 While outpatient influenza surveillance for influenza-like illness (ILI) is widespread, inpatient SARI-based influenza surveillance has been increasing globally over the past decade, particularly in lower-income and tropical settings. 3 In Indonesia, a Driven by the lack of incidence-based data on influenza-associated SARI in low-and middle-income countries such as Indonesia, the World Health Organization published a manual in 2015 outlining several methodologies to estimate influenza incidence using surveillance data. The two primary methods described in the manual are community-based healthcare utilization survey (HUS) and hospital admission survey (HAS). While HUS-based incidence calculations have been performed to estimate influenza burden in India 5 and elsewhere, there are limited data available to date on influenza incidence generated through an HAS-based approach. The objective of this study was to estimate the incidence rate of influenza-associated SARI in Indonesia during 2013-2016 by age-group, using existing SIBI data and a supplemental HAS to estimate sentinel hospital catchment populations.

| Surveillance platform
SIBI is comprised of six sentinel facilities at general hospitals serving all age-groups and socioeconomic levels in six of Indonesia's 34 provinces (Figure 1). Four (Deli Serdang, Wonosari, Bitung, and Kanudjoso Djati) are primary referral district hospitals, and two (Haulussy and West Nusa Tenggara) are secondary referral provincial hospitals. Enrollment of all patients meeting the surveillance case definition is conducted every day year-round in all inpatient wards except for the obstetric and surgical wards. Surveillance staff use the WHO SARI case definition: hospitalized patients with history of fever or fever ≥38°C, cough, and onset of symptoms within 10 days. Information on demographic characteristics, medical history, clinical presentation, and risk factors is collected. 6 Respiratory specimens are collected and tested for influenza A and B viruses using real-time reverse-transcription polymerase chain reaction (rRT-PCR) based on CDC rRT-PCR protocol. 7 All data collected from the system are recorded in a secure online database. 6 Of the six SIBI sites, three (Wonosari hospital in Gunung Kidul District, Kanudjoso Djati hospital in Balikpapan District, and Deli Serdang hospital in Deli Serdang District) were selected for the current analysis because of the feasibility of accessing quality data from centralized electronic medical records.

| Hospital admission survey (HAS)
Among the three participating districts, a hospital admission survey (HAS) was conducted to estimate the catchment population for each district (Fig. S1). In accordance with WHO guidance, 8 the catchment population was calculated in a two-step process. First, the catchment area was defined as subdistricts where ≥80% of SARI cases seeking care at the SIBI hospital reside. Second, the catchment population was the portion of the total population in those subdistricts who sought care at the SIBI hospital as opposed to seeking care at other hospitals in the same areas. Using this approach, we retro- to generate an estimated catchment population for the SIBI facility, used as the denominator for incidence calculations. The number of admission by age-group for ICD.X J.00-J.99 at the three hospitals is presented in Table S1. F I G U R E 1 Location of Surveilans Infeksi Saluran Pernafasan Akut Berat Indonesia (SIBI) and Hospital Admission Survey sentinel sites

| Statistical analysis
Demographic and clinical characteristics of influenza-positive and influenza-negative SARI cases were tested for differences using Chi-squared test. Annual incidence rates and 95% confidence intervals were calculated according to the WHO manual, using catchment population denominators defined from the HAS and numerator defined from all SARI influenza-positive cases enrolled in the surveillance system, and accounting for variance around the numerator of influenza-positive SARI cases by age stratum, site, and year. Analyses were conducted in mi-

| Ethical considerations
This study was reviewed and approved by Ethical Clearance Committee  influenza-positive SARI cases (14%) were aged >14 years (Table 1).

| SARI surveillance findings
This age distribution was similar to that of all 1527 enrolled SARI cases. Asthma (7%), diabetes (3%), and chronic obstructive pulmonary disease (2%) were the most frequent comorbidities reported among influenza cases, and 4% of influenza-positive cases were current smokers. Only four influenza cases (2%) reported having influenza vaccination in the last 12 months, all were aged ≥50 years old. Oseltamivir was prescribed to less than one percent of admissions, including one influenza case and 6 non-influenza cases.
One (<1%) influenza case died during hospitalization; this was a 61-year-old patient who was admitted 1 day after illness onset, diagnosed with renal failure and given antibiotics but did not receive antiviral treatment. This fatal case had a reported influenza vaccination within the previous 12 months. During the study period, cocirculation of multiple influenza virus types and subtypes occurred, with virus detection year-round, and peaks in January or February each year (Figure 2).

| Estimated catchment population of sentinel sites
For the HAS, we collected inpatient data at 5 hospitals in Gunung Kidul, 9 hospitals in Balikpapan, and 6 hospitals in Deli Serdang including the SIBI hospitals. The overall proportion of those admitted for respiratory illness at SIBI sites out of the total respiratory admissions to all facilities in the catchment area was 52% for Wonosari hospital, 26% for

| Annual incidence rate of influenza-associated SARI
The combined incidence of influenza-associated SARI varied during 3 years, ranging between 13 and 19 per 100 000 population, with an increase in the point estimate each year (Table 3). Among the three hospitals, the influenza-associated SARI incidence was highest in children aged 0-4 years (82-114 per 100 000 population) followed by those aged 5-14 years (22-36 per 100 000 population). Incidence varied significantly across sites (P < .05, ANOVA); Kanudjoso Djati was the only hospital that reported influenza-positive cases in patients aged ≥60 years. At this site, observed incidence among those ≥60 years was higher than that among younger adults (15-59 years), but lower than that among children aged <15 years.

| DISCUSSION
During 2013-2016, our study identified influenza as an important contributor to respiratory disease hospitalizations in three geographically diverse areas of Indonesia. The greatest impact was among young children aged <5 years, who had the highest incidence across all three sites, followed by older children aged 5-14 years. This trend persisted in annual site-specific estimates, despite variability in incidence across sites and between seasons. For one site where incidence rates could be calculated for older age-groups, a bimodal distribution was observed where incidence rates were highest in young children, low in the productive age-groups, and then higher in those aged 60 years or more. This is consistent with previously observed patterns of influenza    19 (12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) young children and older adults. 9 Importantly, like other studies reporting on influenza infections and incidence in older populations in Indonesia, both the number of influenza cases detected and the associated underlying catchment population are small. 10,11 This may reflect the population age structure in Indonesia where the majority of the population is young, different health-seeking behavior for acute respiratory infections among older people or difficulties in enrolling older patients meeting the SARI case definition. Further studies are needed to better understand influenza burden in older age-groups.
Our estimates of incidence of influenza-associated SARI, which use a recently published methodology by the WHO, are lower than estimates reported by neighboring countries. Singapore had an age-  Care seeking for severe influenza likely varies by age-group in this population, with young children more likely to visit a hospital and be admitted. Differences in care seeking and public health system performance 22 may also influence the differences in our findings between the three sites, as incidence was highest in Balikpapan district, compared to two other districts. It is possible because Kanudjoso Djati hospital is the only large and modern hospital in its region, and influenza patients compared to other SARI patients may directly seek care there compared to smaller hospitals. Finally, this study was conducted only at three of the 6 SIBI sentinel sites, and captures less than 50% of the country's population.

| CONCLUSIONS
In this study, the hospitalized influenza burden was highest in young children, consistent with other countries, and in line with their inclusion as a key target group for vaccination as recommended by WHO Strategic Advisory Group of Experts on Immunization (SAGE). 23 These trends are consistent with what is found elsewhere and underscore the impact of pediatric admissions for influenza on health systems in lower-income settings. We suggest a further study on costeffectiveness of vaccines among children aged 6 months to 5 years to provide information to guide policymakers on influenza vaccination among specific target groups.
gratitude for SIBI sentinel sites team and the national SIBI team. We appreciated the scientific writing advice from Dorothy L.