Prospective cohort study of influenza vaccine effectiveness among healthcare personnel in Lima, Peru: Estudio Vacuna de Influenza Peru, 2016‐2018

Abstract Background The Estudio Vacuna de Influenza Peru (VIP) cohort aims to describe the frequency of influenza virus infection, identify predictors of vaccine acceptance, examine the effects of repeated influenza vaccination on immunogenicity, and evaluate influenza vaccine effectiveness among HCP. Methods The VIP cohort prospectively followed HCP in Lima, Peru, during the 2016‐2018 influenza seasons; a fourth year is ongoing. Participants contribute blood samples before and after the influenza season and after influenza vaccination (for vaccinees). Weekly surveillance is conducted to identify acute respiratory or febrile illnesses (ARFI). When an ARFI is identified, participants self‐collect nasal swabs that are tested for influenza viruses by real‐time reverse transcriptase‐polymerase chain reaction. Influenza vaccination status and 5‐year vaccination history are ascertained. We analyzed recruitment and enrollment results for 2016‐2018 and surveillance participation for 2016‐2017. Results In the first 3 years of the cohort, VIP successfully contacted 92% of potential participants, enrolled 76% of eligible HCP, and retained >90% of participants across years. About half of participants are medical assistants (54%), and most provide “hands‐on” medical care (76%). Sixty‐nine percent and 52% of participants completed surveillance for >70% of weeks in years 1 and 2, respectively. Fewer weeks of completed surveillance was associated with older age (≥50 years), being a medical assistant, self‐rated health of fair or poor, and not receiving the influenza vaccine during the current season (P‐values < .05). Conclusions The VIP cohort provides an opportunity to address knowledge gaps about influenza virus infection, vaccination uptake, effectiveness and immunogenicity among HCP.


| INTRODUC TI ON
A multi-year, prospective cohort study of healthcare personnel (HCP) in Lima, Peru, is underway, named Estudio Vacuna de Influenza Peru (VIP). Here, we summarize the objectives and design, results of recruitment during the first 3 years of the study, and rates of participation in active surveillance during the first 2 years of the study.
A meta-analysis of studies of seasonal influenza estimated that 1/5 HCP are infected with influenza virus annually, based on serologic and clinical testing. 1 Estimates of influenza virus infection among HCP vary widely depending on the extent of active surveillance and whether studies relied on serologic 2 or molecular diagnostics. 3,4 Healthcare personnel are believed to be at increased risk because of frequent patient contact. They may also transmit influenza to their patients, though the extent of these risks is unclear. 5 Because HCP often work while ill, 3,6,7 more information is needed on the number and types of contacts HCP may have with patients while HCP are symptomatic with influenza and other viral infections. 3,[6][7][8] Recent research suggests that certain subgroups of HCP, such as those that perform aerosol-generating procedures, may be at heightened risk of infection with respiratory pathogens including influenza. 9 Our cohort study was designed to address gaps in our knowledge of influenza burden and impact among HCP. The first objective of the VIP Cohort is to describe the frequency of influenza virus infections among HCP, including acute illnesses and asymptomatic infections.
Vaccination of HCP against influenza virus infection is an important component of infection control in healthcare settings, 10 but relatively low uptake among HCP outside the United States remains a topic of international concern and debate. [11][12][13] Although numerous studies of the knowledge, attitudes, and practices (KAP) associated with influenza vaccine acceptance have been conducted among HCP in high-income countries, [14][15][16][17] less is known about barriers to vaccine acceptance among HCP in low-and middle-income countries. 13,15,18 The second objective is to identify predictors of vaccine acceptance and hesitancy in HCP.

Studies of influenza vaccine immunogenicity among HCP have
demonstrated that repeated vaccination can blunt the antibody response to hemagglutinin 19,20 and neuraminidase. 21 Further research is needed to examine how influenza vaccination across multiple seasons may affect immunogenicity 22 and how these effects are mediated by specific humoral 20 and cell-mediated immune responses. 22 The third objective is to examine how repeated influenza vaccination may modify immunogenicity.
Although recent reviews confirm that seasonal influenza vaccine is moderately effective in reducing the risk of illness among adults, 23 there are limited data regarding the value of vaccine for HCP. To date, the only randomized controlled trial of influenza vaccine efficacy among HCPs relied on serologic outcomes, 24 which are biased among vaccinees and may inflate influenza vaccine effectiveness (IVE) estimates. 25,26 Reports of reduced IVE among frequent vaccinees in some studies 20,22,27 and seasons 28 make it important to examine IVE among HCP, a population that receives frequent annual influenza vaccinations in the United States. Few data are available about the value of influenza vaccine in reducing missed work due to infection or reducing frequency of time worked while ill. 5,8 Given that influenza vaccine may only reduce the risk of influenza illness by 40%-60% during years with a good match between circulating and vaccine viruses, further research is needed on whether factors like age, patient-care responsibilities, and the use of personal protective equipment (PPE) modify the risk of vaccine failure. Limited research suggests that vaccination may also modify illness duration and severity among those who develop influenza illness despite vaccination. 8,[29][30][31] The fourth objective of the VIP Cohort is to evaluate IVE in preventing influenza illness and associated missed work and working while ill. See Appendix S1 for more detail on study objectives. of completed surveillance was associated with older age (≥50 years), being a medical assistant, self-rated health of fair or poor, and not receiving the influenza vaccine during the current season (P-values < .05).

Conclusions:
The VIP cohort provides an opportunity to address knowledge gaps about influenza virus infection, vaccination uptake, effectiveness and immunogenicity among HCP.

K E Y W O R D S
healthcare personnel, influenza, influenza vaccine 2 | ME THODS

| Setting
The VIP Cohort recruited HCP in Lima, Peru, at Dos de Mayo National Hospital, Cayetano Heredia National Hospital, and Daniel Alcides Carrión National Hospital in 2016 and expanded to include National Institute of Child Health (Del Niño) and Archbishop Loayza Hospital in 2017 (Table S1).

| Eligibility criteria
Eligible participants are HCP aged ≥18 years, working ≥30 hours/ week, with routine, direct patient contact and must have been employed by the hospital for ≥1 year. Similar to previous definitions for HCP, 32 we include a variety of HCP, including direct care providers, allied-health workers, and non-clinical personnel. Participants are ineligible if they received the current seasonal influenza vaccine prior to enrollment.

| Recruitment strategy
To minimize potential selection biases, HCP are invited to join the cohort using a stratified sampling strategy. We categorize potential participants at each hospital into 18 strata by sex, three age groups, and three occupational categories. To ensure the cohort includes participants with all combinations of sex, age, and occupation, we set a goal of ≥50 participants in each strata. We set goals for total recruitment in year 1 of 1200, year 2 of 2800, and year 3 of 2400, and set minimum enrollment goals per study hospital (Appendix S1).

| Enrollment
Participants complete an enrollment survey when they enter the cohort and complete follow-up surveys at the end of season and start of season for their remaining time in the cohort. The enrollment survey gathers information on sociodemographic characteristics, work responsibilities, health status, health behaviors, and KAP regarding influenza illness and vaccination (Appendix S1). Influenza vaccination history for five prior years is documented by self-report at enrollment and extracted from each hospital's employee vaccination registry (Appendix S1, Table S2).

| Active surveillance
Based on previous surveillance for laboratory-confirmed influenza virus infection in Lima, 33 we conduct active surveillance for ARFI during ~20 weeks each year. The start of active surveillance is informed by historical trends and early reports of During the influenza season, participants receive twice-weekly short-message-service (SMS) text messages to confirm whether they had an acute illness with one or more of the following symptoms within the past 7 days: cough, runny nose, body aches, or feverishness. Upon illness identification, staff conduct an acute illness survey and participants contribute a self-collected nasal swab. Staff conduct a follow-up survey at illness resolution. To verify surveillance completeness and mitigate information bias, the end-of-season survey asks participants whether any illness was missed during the season (Appendix S1, Figure S1).

| Influenza virus infection detection
The primary study outcome is ARFI associated with influenza virus infection confirmed by rRT-PCR. Specimens are tested by NAMRU-6 Laboratory for influenza A and B viruses, subtypes and lineages using rRT-PCR assays, with standard protocols, primers, probes, and reagents supplied by US CDC's International Reagent Resource (IRR) (Appendix S1).

| Blood specimens
All participants contribute 10 mL of whole blood at enrollment and 5 mL at the start of session and end of season; vaccinees also provide 5 mL approximately 28 days ( Appendix S1 and Figure S2 for more information on laboratory testing.

| Data management
Data collection and management were conducted using REDCap (Research Electronic Data Capture), a browser-based metadatadriven software system 34 (Appendix S1).

| Statistical analysis to date
To assess the stratified recruitment approach, we evaluated the proportion of HCP who fully enrolled out of all eligible HCP. Full enrollment is defined as providing informed consent, completing the enrollment survey and contributing the enrollment blood sample. We compared full enrollment stratified by major recruitment categories in the 18 recruitment strata (sex by occupation by age) using chi-square tests and used multivariable logistic regression to model full study enrollment as a function of these five factors.
To describe performance of surveillance activities in years 1 and 2, we examined the proportion of participants who com- 10 missing responses are denoted on the tables; data were not imputed for these analyses.

| Ethical approval and ethical considerations
The study protocol and procedures were reviewed and approved by seven institutional review boards including NAMRU-6, each study hospital and by Abt Associates (coordinating institution for US CDC). All participants completed written informed consent.
Small gifts were given to participants at study milestones. Given the research nature of the laboratory methods and time delays in batch testing, rRT-PCR results were not available to participants and did not inform decisions regarding their medical care or approval to return-to-work.

| Recruitment and retention
The recruitment flow diagram for years 1-3 is presented in Figure 1. We successfully contacted 92% (4728/5131) of potential TA B L E 2 Characteristics of fully enrolled participants, VIP cohort, 2016-2018 (N = 3050) Total N = 3050 a n (Col.%)
c Currently receiving medical care for ≥1 of asthma, cancer, lung condition, diabetes, heart condition, high blood pressure, immunosuppression/problem with immune system, kidney disease, neurologic problem, and other. d Self-reported vaccination history.
f Possible responses range from 0-30 d.
g Regularly administers ≥1 of the following: collects respiratory swab, collects sputum specimen, administers medication using nebulizer, applies nasal cannula, applies oxygen facemask, performs tracheal intubation, inserts nasogastric tube, performs manual ventilation, performs suction of fluids, performs chest physiotherapy, and performs bedside bronchoscopy.
Although most participants were healthy, 21% reports ≥1 chronic medical condition, and 20% describe their overall health as only "fair" or "poor." Most (85%) report having received the influenza vaccine at least once before enrollment.
In the end-of-season survey, a small percentage of participants reported that they had failed to report at least one possible ARFI as

| D ISCUSS I ON
The VIP Cohort is poised to address knowledge gaps regarding the burden of laboratory-confirmed influenza illness and the preven- of participants, respectively, reported that they failed to report an acute illness during the season. Gaps in surveillance data create potential for information bias; in a multivariable model, we found male sex, age ≥50, occupation as a nurse/technician or medical assistant, self-rated overall health as "fair" or "poor," and having not received the vaccination in the current season were associated with missing more weeks of surveillance. Nonetheless, the ability to quantify this missing information and address it in statistical models for IVE and influenza virus infection incidence represent a strength of the study.
This study has several other limitations. Like all studies of IVE and influenza incidence, the ability to broadly generalize results is limited by the unpredictability of circulating virus types and potential for mismatch between vaccine components and circulating strains in any year. Although conducting the study in Peru allows us to examine IVE in a middle-income and Southern-hemisphere country, where data on IVE are limited, the generalizability of findings to the United States and other countries is unknown.
Additionally, the overall intensity and impact of influenza seasons are variable, and low influenza activity in a study season could negatively affect our ability to precisely estimate IVE and incidence. There is potential for bias in recall of information collected by self-report, including vaccination history and details about illness severity and duration.
This study provides a unique opportunity to characterize and understand influenza illness among HCP and the impact of influenza illness on work in healthcare settings. In this context, we can better understand the role influenza vaccines play in protecting HCP from becoming infected, missing work, or working while sick, and the serologic response produced by influenza vaccines in a repeatedly vaccinated population.

ACK N OWLED G EM ENTS
The authors would like to thank all study staff at participating hospitals and express gratitude to all study participants.