Cohort profile: A Prospective Household cohort study of Influenza, Respiratory syncytial virus and other respiratory pathogens community burden and Transmission dynamics in South Africa, 2016–2018

Abstract Purpose The PHIRST study (Prospective Household cohort study of Influenza, Respiratory Syncytial virus, and other respiratory pathogens community burden and Transmission dynamics in South Africa) aimed to estimate the community burden of influenza and respiratory syncytial virus (RSV) including the incidence of infection, symptomatic fraction, and to assess household transmission. Participants We enrolled 1684 individuals in 327 randomly selected households in a rural and an urban site over three consecutive influenza and two RSV seasons. A new cohort of households was enrolled each year. Participants were sampled with nasopharyngeal swabs twice‐weekly during the RSV and influenza seasons of the year of enrolment. Serology samples were collected at enrolment and before and after the influenza season annually. Findings to Date There were 122 113 potential individual follow‐up visits over the 3 years, and participants were interviewed for 105 783 (87%) of these. Out of 105 683 nasopharyngeal swabs, 1258 (1%) and 1026 (1%) tested positive on polymerase chain reaction (PCR) for influenza viruses and RSV, respectively. Over one third of individuals had PCR‐confirmed influenza each year. Overall, there was influenza transmission to 10% of household contacts of an index case. Future Plans Future planned analyses include analysis of influenza serology results and RSV burden and transmission. Households enrolled in the PHIRST study during 2016–2018 were eligible for inclusion in a study of SARS‐CoV‐2 transmission initiated in July 2020. This study uses similar testing frequency to assess the community burden of SARS‐CoV‐2 infection and the role of asymptomatic infection in virus transmission.


| INTRODUCTION
In 2015, lower respiratory tract infections caused an estimated 2.7 million deaths globally. 1 Among children aged <5 years, the highest mortality rates are in sub-Saharan Africa where the HIV-epidemic has increased morbidity of severe pneumonia. Influenza, respiratory syncytial virus (RSV), pertussis and pneumococcus are among the leading causes of pneumonia globally. [2][3][4][5] Approximately 30% of influenza and RSV transmission is estimated to occur within households. 6,7 Data on community burden and transmission of respiratory pathogens are important to guide vaccination strategies such as reduced pneumococcal conjugate vaccine dose schedules, 8 optimal timing of booster doses 9 and vaccinating community transmitters. 10,11 Illness episodes in the community may be associated with substantial community impact including absenteeism from school or work and loss of income. 12 The PHIRST study aimed to estimate the community burden of influenza and RSV (including the incidence of infection and symptomatic fraction) and to assess household transmission of influenza and RSV (Table S1). Secondary objectives included describing the community burden and transmission of Streptococcus pneumoniae and Bordetella pertussis, estimating the impact of HIV infection and age on disease burden, estimating rates of tuberculosis infection and transmission and investigating the interaction between respiratory viruses and bacteria. We also aimed to evaluate the role of asymptomatic influenza and RSV infection in household transmission.

| Study population and household eligibility criteria
A prospective cohort study of randomly selected households in South Africa was conducted in a rural and an urban site, each with established surveillance for pneumonia and influenza-like illness ( Figure 1). 13,14 The rural site in Mpumalanga Province (Agincourt subdistrict) is part of a health and socio-demographic surveillance system (HDSS), including approximately 116 000 people in 31 contiguous villages. 15,16 Approximately 30% of the population are former Mozambicans who migrated there in the 1980s. 15 The urban site, Jouberton Township in Klerksdorp, is part of the municipality of Matlosana in North West Province, with a population of approximately 180 000 people. Mining of gold and uranium, although declining, remains a primary driver of the local economy. 17 We aimed to enroll approximately 1500 individuals (approximately 500 individuals per year) over three consecutive influenza and RSV seasons to allow the estimation of 20% risk of infection and a 10% risk of illness with 95% confidence intervals (CIs) and 5% absolute precision. Assuming an average household size of five individuals and a loss to follow-up of 10%, based in previous studies, 18 we aimed to enroll approximately 55 households with >2 household members per site each year with at least 50% having at least one child aged <5 years in the house.
In rural Agincourt, each year, we purposively selected two different villages within the HDSS. Within these villages, we randomly selected households with >2 members from an enumerated list obtained from the HDSS. In urban Jouberton township, we generated a list of 450 random global positioning system (GPS) coordinates located within a polygon defining the township boundaries using Google Earth. Study staff navigated to the location represented by the coordinates and selected the nearest house. If there was no dwelling within 30 m, the coordinates were discarded. Households were approached consecutively until the desired sample size was reached.
If a household withdrew during January-April of each year, it was replaced by a new household, selected consecutively, for the remaining follow-up period.
At each household with >2 members, study staff requested permission from the head of household to inform members about the study purpose, risks and benefits. If the head of household was a minor or unavailable after three attempts, the household was excluded. Written informed consent was required to participate in the study from all household members aged ≥18 years; assent was required from children aged 7-17 years, and consent from a parent/ guardian of children aged <18 years. We included households where ≥80% of household members consented. F I G U R E 1 Location of rural (Agincourt) and urban (Jouberton) study sites in South Africa

Strengths and limitations of this study
• PHIRST was conducted in urban and rural African settings with high HIV prevalence, allowing assessment of the effect of HIV on community burden and transmission dynamics of respiratory pathogens.
• Households were selected randomly to provide a representative sample of the community. Twice-weekly sampling from each cohort of individuals for 6-10 months irrespective of symptoms allows estimation of community burden, household secondary infection risk, and serial interval including asymptomatic or paucisymptomatic episodes.
• Polymerase chain reaction testing of >100 000 nasopharyngeal swab samples for multiple pathogens (influenza, respiratory syncytial virus, pertussis and Streptoccocus pneumonia) allows detailed examination of disease burden and transmission and pathogen interactions • PHIRST was not powered to assess severe outcomes (i.e. hospitalisation and death).
• We only examined four pathogens, but other microorganisms may be important. Samples have been stored which could allow us to implement broader multipathogen testing in the future.
Household surveys were conducted once during the follow-up period to evaluate household income, housing quality, oropharyngeal carriage of meningococcus, Corynebacterium diphtheriae and Group A streptococcus and presence of S. pneumoniae DNA in blood by polymerase chain reaction (PCR). Serum samples were also collected at enrolment, before the influenza season, and at the end of the active follow-up period. In addition, sera were also collected from the 2016 and 2017 cohorts in subsequent years (Figures 2 and S1). Environmental assessments including respirable particulate matter and temperature were undertaken twice a year (summer and winter) ( Table 1).

| Baseline, symptom and health contact data
Data were collected using REDCap (Research Electronic Data Capture) 19 . Following enrolment, a baseline questionnaire was completed for each household including information on household members, relationships, sleeping arrangements and housing. For each individual, we collected baseline information on demographics, underlying illnesses, vaccinations and occupation. During the twice-weekly follow-up phase, at each visit, for each participant, a questionnaire assessing presence of symptoms, absenteeism and health system contacts was completed and nasopharyngeal (NP) swabs were collected regardless of the presence or absence of symptoms (Table 1)    (ThermoFisher, Waltham, Massachusetts, USA) 20 . RSV A and B subgroups were determined by an in-house RT-PCR. 21,22 Twice-weekly NP samples were tested for S. pneumoniae using an in-house singleplex (lytA) quantitative real-time PCR assay. 23 NP and sputum samples were tested for Bordetella spp. (including B. pertussis, B. parapertussis, B. bronchiseptica and B. holmesii) by a combination of a triplex and singleplex real-time PCR assays and results interpreted as previously described. 24 Clotted blood samples from serology surveys collected in vacutainer tubes were centrifuged, aliquoted and stored frozen before being sent in batches to NICD on dry ice. Hemagglutination inhibition (HAI) assays using turkey red blood cells were performed to determine serological reactivity titres for serum samples against influenza. 25 Virus strains for testing were selected based on the Southern Hemisphere vaccine strains and strains predominantly circulating in

| Housing quality survey
Information on housing type, construction, materials and condition, water sources, water security and water storage, fuel use and expenditure for cooking, space and water heating, waste removal services, visible dampness and smoking practices was collected annually for all households.

| Proximity and contact study
In 2018, four surveys of household contact using proximity monitors (http://www.sociopatterns.org) were conducted to capture information on intra-household contact patterns for three seasons (summer, autumn and winter). To measure contacts of participants of the study outside the home, participants were interviewed by field workers to complete a contact diary and time-use questionnaire for one day between August and October 2018.

| Costing survey
We surveyed all symptomatic household members during August-October 2018 to assess cost of medically attended and non-medically attended illness episodes.

| Limitations
This study was not powered to assess severe outcomes (i.e., hospitalisation and death). Repeated assessment of symptoms at twice-weekly visits over an extended period may lead to possible fatigue and under-reporting by participants. High rates of migration and movement in communities under study affected follow-up rates.
The study was intensive, and nasopharyngeal swabs are T A B L E 2 A Characteristics of included participants and households in PHIRST during 2016-2018 at the rural site (Agincourt) compared to those not included, using data from the 2017 Agincourt health and socio-demographic surveillance system site (HDSS) census

PATIENT AND PUBLIC INVOLVEMENT
Both study sites have community advisory boards (CAB) consisting of representatives from community-based and faith-based organisations who were involved in the planning of the PHIRST study. The CABs meet regularly and give advice on protocols, consents and recruitment plans and also provide feedback to communities on results of studies.
In addition, feedback sessions on study findings were held for participating families.

ACKNOWLEDGEMENTS
We would like to acknowledge the study participants and field and laboratory staff who worked tirelessly to make the study a success.  (9) Lost to follow-up 6 (2) 1 (1)

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1111/irv.12881.

DATA AVAILABILITY STATEMENT
The study protocol including informed consent forms is available on the NICD website (https://www.nicd.ac.za/wp-content/uploads/ 2021/02/PHIRST-SARS-CoV-2-protocol-V1-amendment-Nov2020incl-upd-consent.pdf). Analysis of the data for primary study objectives is planned to be completed by December 2023. Additional modelling and serologic studies will be concluded within one additional year, and primary de-identified data will be made publicly available not later than December 2025. The investigators welcome enquiries about possible collaborations and requests for access to the data set. Data will be shared after approval of a proposal and with a signed data access agreement. Investigators interested in more details about this study, or in accessing these resources, should contact the principle investigator, Prof Cheryl Cohen, at NICD (cherylc@nicd.ac.za).