The burden of influenza‐associated respiratory hospitalizations in Bhutan, 2015‐2016

Background Influenza burden estimates help provide evidence to support influenza prevention and control programs. In this study, we estimated influenza‐associated respiratory hospitalization rates in Bhutan, a country considering influenza vaccine introduction. Methods Using real‐time reverse transcription‐polymerase chain reaction laboratory results from severe acute respiratory infection (SARI) surveillance, we estimated the proportion of respiratory hospitalizations attributable to influenza each month among patients aged <5, 5‐49, and ≥50 years in six Bhutanese districts for 2015 and 2016. We divided the sum of the monthly influenza‐attributed hospitalizations by the total of the six district populations to generate age‐specific rates for each year. Results In 2015, 10% of SARI patients tested positive for influenza (64/659) and 18% tested positive (129/736) in 2016. The incidence of influenza‐associated hospitalizations among all age groups was 50/100 000 persons (95% confidence interval [CI]: 45‐55) in 2015 and 118/100 000 persons (95% CI: 110‐127) in 2016. The highest rates were among children <5 years: 182/100 000 (95% CI: 153‐210) in 2015 and 532/100 000 (95% CI: 473‐591) in 2016. The second highest influenza‐associated hospitalization rates were among adults ≥50 years: 110/100 000 (95% CI: 91‐130) in 2015 and 193/100 000 (95% CI: 165‐221) in 2016. Conclusions Influenza viruses were associated with a substantial burden of severe illness requiring hospitalization especially among children and older adults. These findings can be used to understand the potential impact of seasonal influenza vaccination in these age groups.


| INTRODUC TI ON
Each year, influenza virus infections are a major contributor to hospitalizations worldwide. 1 Many high-and upper-middle-income countries have used influenza-associated burden estimates to identify target groups for influenza prevention and control and to explore the cost-effectiveness of these interventions. [2][3][4][5][6] Despite increasing global influenza surveillance capacity and an improved understanding of influenza virus seasonality and epidemiology, estimates of influenza-associated burden are limited in low-and middleincome countries. 1,7 Country-specific influenza-associated burden estimates generated through the use of viral surveillance data can provide useful information to ministries of health (MOH) and national immunization technical advisory groups (NITAG) as they seek to understand the potential value of introducing seasonal influenza vaccines in their country. 3,8 The Kingdom of Bhutan is a lower-middle-income country located in Eastern Himalaya between China to the north and India to the east, west, and south, with a population of roughly 760 000 The majority of Bhutan's population lives in rural settings with rugged terrain making cross country travel challenging and access to healthcare services difficult. 11 The government of Bhutan provides all healthcare services, including medications and routine immunizations, free of charge to its citizens. While seasonal influenza is not part of Bhutan's routine immunization schedule, the government of Bhutan is exploring the utility of its introduction among WHO recommended target groups. 12 In addition, Bhutanese referral hospitals also have a stockpile of antivirals from WHO but have not yet identified priority groups to receive this medication during seasonal epidemics. In this analysis, we used SARI viral surveillance data, nationally reported respiratory coded hospital discharge data, and census data to estimate age group-specific influenza-associated respiratory hospitalizations. These findings are intended to help the Bhutanese MOH and NITAG assess the value of influenza vaccination among children and older adults (two of the WHO recommended target groups), the targeting of empiric antiviral use among hospitalized patients during the influenza season, and other influenza prevention and control measures. 12 2 | ME THODS

| Design
We used influenza viral laboratory results from nationally representative SARI surveillance to estimate the proportion of respiratory hospital discharges in six districts attributable to influenza each month for 2015 and 2016; the approach used methods adapted from the WHO Manual for Estimating Disease Burden Associated with Seasonal Influenza. 14 We then summed estimates of influenzaassociated hospitalization across each of the years and district by age group and divided the sum by the age-specific census population to generate influenza-associated hospitalization rates among persons aged <5, 5-49, and ≥50 years. We first focused on seven of the 11 SARI sentinel sites that were district-level hospitals with a defined catchment population, which was the known population of persons residing around a hospital that seek care at that particular hospital ( Figure 1). After evaluating the SARI surveillance and respiratory discharge data, we excluded one district hospital because of the poor concurrence of discharge diagnosis data in hospital records verses what the hospital reported to the national office. The four remaining sentinel hospitals were referral hospitals, without defined catchment populations; although we incorporated their virological surveillance data into this analysis, we did not generate influenzaassociated hospitalization rates specifically for these sites. Laboratory staff at RCDC tested all specimens by real-time reverse transcription-polymerase chain reaction for the presence of influenza A and B viruses using methods adapted from CDC protocols. 15 Specimens positive for influenza A were further tested for influenza A virus subtypes. In addition to testing for seasonal virus subtypes, patients with recent exposure to avian influenza outbreaks among poultry or with recent international travel history were also tested for influenza A(H5) or A(H7) viruses, respectively.

| Evaluation of SARI case-patient identification
In order to determine whether the number of SARI case-patients reported to RCDC accurately reflected the totality of patients meeting the SARI case definition, we conducted a retrospective medical chart review at four of the seven district-level SARI sentinel site hospitals, which we selected for their accessibility. At each hospital, we reviewed approximately 100 hospitalized patient medical charts from three to eight randomly selected weeks during 2015 and collected demographic, symptom, and discharge information. As annual hospital admissions varied widely between these four hospitals, the number of randomly selected weeks varied by hospital in order to review a similar number of charts at each hospital. We classified each patient according to the Bhutanese SARI case definition based on recorded symptom data. We then compared the number of SARI case-patients identified during the chart review to the number of SARI case-patients reported to RCDC during the same time period.

| Evaluation of ICD-10 coded hospitalizations
In all seven SARI sentinel sites that are district-level hospitals, we additionally conducted a retrospective hospital-based logbook review to validate the number of respiratory ICD-10 coded hospitalizations reported to the MOH each month. We conducted this evaluation to validate data in the national MOH database before it was included in the pooled analysis. At each hospital, we reviewed discharge diagnoses (both free text and ICD-10 coded) in hospital logbooks in randomly selected months during 2015-2016. We recorded demographic information (age and district of residence) and primary discharge diagnosis information (free text or ICD-10 coded) for every hospitalized patient with a reported respiratory discharge diagnosis. We summarized the number of respiratory hospitalizations by month, age group, and hospital and compared it to the number of respiratory hospitalizations reported to the MOH national hospitalization database for the same month.

| Calculation of influenza-associated respiratory hospitalizations
After excluding one SARI sentinel site following the data evaluations, we estimated annual influenza-associated respiratory hospitalizations for the six remaining districts in Bhutan where the SARI surveillance sentinel site was the only admitting hospital within that district. These six districts represented 31% of the national population. We excluded Hospital A because the number of respiratory diagnoses reported to the MOH and the number recorded in the hospital logbooks were so divergent. We calculated the monthly proportion of specimens testing positive for influenza virus across all age groups and sentinel sites for 2015-2016. We were not able to stratify these data by age group or region because of the low number of influenza virus-positive specimens identified each year. We then estimated influenza-associated respiratory hospitalizations in each district using the following formula: where:

| Ethics statement
This project was part of an influenza control program evaluation; only anonymized surveillance data were used for this analysis.

| Evaluation of SARI case-patient identification
In hospitals where the study team evaluated SARI case-patient identification, the combined number of SARI case-patients identified and reported by hospital staff through routine surveillance during the review period (N = 14) represented 36% (95% CI: 21-51) of those identified through our medical chart review (N = 39) (Table S1). Each week, however, the number of SARI case-patients identified in medical charts was very small; in 10 (56%) of the 18 weeks reviewed, there was ≤1 SARI case-patient identified in the medical charts or reported through the surveillance system.  (Table S2). On average, these differences were small (0.1%-12.8% fewer discharges recorded in national datasets than recorded in the hospital logbooks). In Hospital A, however, there was a 7-fold difference in what was recorded in the hospital logbook vs the national database.  (Table 3).

| D ISCUSS I ON
We observed a high burden of seasonal influenza virus infections among respiratory hospitalizations in 2015-2016 across Bhutan, especially among children <5 years and older adults. As seasonal influenza vaccine is not currently available in Bhutan, the Bhutanese MOH could use these burden estimates to assess the potential value of seasonal vaccine introduction and other influenza control measures. For example, the Bhutanese MOH could use these hospitalization rates to explore seasonal influenza vaccine program cost-effectiveness and cost benefit or to determine priority groups to receive stockpiled antivirals during seasonal epidemics.
The Bhutanese MOH is also updating its pandemic plan, and these influenza-associated hospitalization rates might help them better plan for the number of vaccine and antiviral doses needed during an influenza pandemic. Our all age rate estimates were similar to previously published estimates from neighboring countries where seasonal influenza vaccine use was low, such as rural India, central China, and Thailand. 17,21,22 The greatest influenza-associated hospitalization burden we observed was among children <5 years of age, which is consistent with other published studies from low-and middleincome countries such as Bangladesh, 23 Egypt, 24

| CON CLUS ION
Our findings suggest that each year hundreds of influenzaassociated respiratory hospitalizations occur throughout Bhutan, especially among young children and older adults. These findings demonstrate the magnitude of seasonal influenza burden among different age groups and could be used to explore the potential value of introducing influenza prevention and control measures.
For example, these estimates could be used to estimate the number of influenza cases that could be averted through an influenza vaccine program, information that would be valuable for the Bhutanese MOH and NITAG when considering seasonal influenza vaccine introduction.

ACK N OWLED G EM ENTS
We would like to thank Jit Bahadur Dhanal, Tsheten, Sangay

D I SCL A I M ER
The findings and conclusions in this report are those of the author(s) and do not necessarily represent the official position of the US

Centers for Disease Control and Prevention or the World Health
Organization.

CO N FLI C T O F I NTE R E S T
Co authors have no conflict of interests to declare.