Effectiveness of Vaccines and Monoclonal Antibodies Against Respiratory Syncytial Virus: Generic Protocol for Register-Based Cohort Study

Abstract

Several immunization products are currently being developed against respiratory syncytial virus (RSV) for children, pregnant females, and older adults, and some products have already received authorization. Therefore, studies to monitor the effectiveness of these products are needed in the following years. To assist researchers to conduct postmarketing studies, we developed a generic protocol for register-based cohort studies to evaluate immunization product effectiveness against RSV-specific and nonspecific outcomes. To conduct a study on the basis of this generic protocol, the researchers can use any relevant databases or healthcare registers that are available at the study site.

respiratory syncytial virus, RSV, vaccine, monoclonal antibody, maternal immunization

Human respiratory syncytial virus (RSV) is a common respiratory pathogen that causes high morbidity among infants, young children, and the elderly [1–3]. RSV has 2 major antigenic subtypes, A and B, and its activity typically peaks between December and February in temperate regions of the northern hemisphere [3], although the coronavirus disease 2019 (COVID-19) pandemic has disrupted this pattern [4, 5]. Clinical manifestations of RSV range from mild upper respiratory tract infections to severe lower respiratory tract infections (LRTIs) [3]. Acute otitis media is also a common complication of RSV infection among children [6, 7], and RSV infection might also increase the risk of wheezing illness (including bronchiolitis and asthma) among children [8].

RSV vaccine development was started in the 1960s, but the occurrence of vaccine–associated enhanced respiratory disease in a clinical trial halted the development for decades [9]. The monoclonal antibody (mAb) palivizumab was approved for use in high-risk infants as prophylaxis for severe RSV disease in the late 1990s, but it is not widely used due to its high cost and limited efficacy [10, 11]. Currently, many new vaccines and mAbs are being tested for target groups including infants, pregnant women, and the elderly [12–14]. Nirsevimab, a long-acting mAb product for infants, and Arexvy, a protein subunit vaccine for adults aged 60 years and older, have already been authorized in the European Union [15, 16].

As several RSV immunization products are expected to be authorized in the coming years, studies to evaluate their effectiveness and impact are needed. For this purpose, we developed a generic protocol to assess the effectiveness of RSV immunization products in middle- or high-income countries using a register-based cohort study design. This generic protocol is not intended to be implemented directly and the goal is to assist researchers to draft study-specific protocols (Figure 1). This generic protocol was part of the public–private partnership Preparing for RSV Immunisation and Surveillance in Europe (PROMISE) project [17].

Figure 1.

Step-by-step guide. Abbreviations: mAb, monoclonal antibody; RSV, respiratory syncytial virus.

OBJECTIVES

The objectives are to evaluate the effectiveness of RSV immunization products in the real-world setting. These are divided into RSV-specific objectives and other objectives. The importance of the objectives in different study populations is presented in Supplementary Table 1.

RSV-specific objectives:

1. To estimate RSV immunization effectiveness against laboratory-confirmed RSV infection.

2. To estimate RSV immunization effectiveness against RSV-related primary care visits.

3. To estimate RSV immunization effectiveness against RSV-related emergency department visits.

4. To estimate RSV immunization effectiveness against hospitalization due to RSV-related LRTI.

5. To estimate RSV immunization effectiveness against intensive care unit (ICU) admission due to RSV.

6. To estimate RSV immunization effectiveness against the need for supplemental oxygen due to RSV.

7. To estimate RSV immunization effectiveness against the need for ventilation treatment (invasive or noninvasive) due to RSV.

8. To estimate RSV immunization effectiveness against mortality due to RSV.

9. To estimate RSV immunization effectiveness against all-cause mortality during the 30 days after RSV infection.

Other objectives:

1. To estimate RSV immunization effectiveness against primary care visits due to all-cause acute respiratory tract infection (ARTI).

2. To estimate RSV immunization effectiveness against hospitalizations due to all-cause LRTI.

3. To estimate RSV immunization effectiveness against hospitalization due to all-cause ARTI.

4. To evaluate if RSV immunization decreases the risk of otitis media among newborns and children.

5. To evaluate if RSV immunization decreases the risk of wheezing among newborns and children aged ≤3 years.

6. To evaluate if RSV immunization decreases antibiotic consumption.

7. To evaluate if RSV immunization decreases hospitalizations due to exacerbations of asthma, chronic obstructive pulmonary disease (COPD), or congestive heart failure (CHF) among adults.

8. To evaluate if RSV immunization decreases the risk of cardiovascular outcomes among adults.

9. To evaluate if RSV immunization decreases the risk of discharge to long-term care facilities among individuals aged 60 years or older.

10. To estimate RSV immunization effectiveness against RSV subtypes A and B.

11. To evaluate if other vaccinations administered concurrently have an effect on RSV immunization effectiveness (sensitivity analysis).

METHODS

Study Design and Study Populations

The studies can use any relevant databases or healthcare registers that are available at the study site to conduct a prospective register-based cohort study. Sample size calculations for different settings based on the method by O’Neill [18, 19] are presented in Supplementary Table 2.

Three study populations are included: (1) children; (2) linked maternal–newborn dyads with infant follow-up until 6 months of age; and (3) adults.

The inclusion for the study subjects should be based on eligibility for RSV immunization at the study site (eg, age limit). Maternal–newborn dyads in which the mother has received RSV immunization before pregnancy or within 14 days before the birth should be excluded. Follow-up of the newborns should continue until 6 months of age for an assessment of maternal immunization (Supplementary Figure 1).

Exclusion criteria should be considered if RSV immunization is restricted to specific groups (eg, chronically ill individuals). Furthermore, additional exclusion criteria are required to assess certain outcomes (see Outcomes section).

Study Period

The study period is different for RSV-specific and nonspecific outcomes (see Outcomes section). For the RSV-specific outcomes, the study period starts on the first day of the immunization campaign and ends on the day of data cutoff (Supplementary Figure 2). For the nonspecific outcomes, the follow-up period should be restricted to the RSV season. The start and end of the RSV season should be defined at the study site using surveillance data and based on an integrated, year-round, respiratory pathogen surveillance system, as recommended by Teirlinck et al [20].

Exposures

Exposure is defined as immunization against RSV and is a time-dependent variable. Individuals who have received a RSV vaccination or an mAb are considered exposed. For the assessment of maternal immunization, infants whose mother received RSV vaccination during the pregnancy are considered exposed on the day of birth (Supplementary Figure 1) and the analysis should take into account the time the gestational age at the time when the mother received the immunization. The exposure variable is categorical, and the exposed can be categorized into separate groups on the basis of immunization products (eg, vaccine 1, vaccine 2, mAb 1), number of doses, and time since immunization (Table 1).

Table 1.

Exposures, Definition of Exposed, and Categorization of Exposed

Exposure	Exposed	Categorization of Exposed
RSV vaccination	Individuals who received RSV vaccination	Specific vaccine product used Time since immunization Number of vaccine doses received
RSV mAb	Individuals who received RSV mAb	Specific mAb product used Time since immunization
Maternal RSV immunization	Newborns whose mothers received RSV vaccination during pregnancy	Specific vaccine product used Gestational age at immunization Time since the birth Number of vaccine doses received during the pregnancy

Abbreviations: mAb, monoclonal antibody; RSV, respiratory syncytial virus.

Outcomes

To conduct the analysis the study site is required to have comprehensive information considering at least one outcome. Each study site should describe the method to collect the outcome(s) in the study protocol. Register data that consider laboratory-confirmed RSV infections should be based on polymerase chain reaction or antigen tests conducted on respiratory specimens.

The outcomes are categorized as RSV-specific and nonspecific outcomes. The RSV-specific outcomes require laboratory-confirmed RSV infection and can be assessed during the RSV season and off-season. However, most RSV infections are not laboratory-confirmed, and evaluation of only RSV-specific outcomes leads to underestimation of the total impact of RSV immunization. For this reason, nonspecific outcomes are also important to evaluate.

The RSV-specific outcomes are as follows (see Supplementary Table 3):

• Laboratory-confirmed RSV infection

• RSV-related primary care visits

• RSV-related emergency department visits

• Hospitalization due to RSV-related LRTI

• ICU admission due to RSV

• Supplemental oxygen due to RSV

• Ventilation due to RSV

• Mortality due to RSV

• All-cause mortality during the 30 days after RSV infection

The nonspecific outcomes are as follows (see Supplementary Table 4):

• Primary care visits due to any ARTI

• Hospitalization due to any LRTI

• Hospitalization due to any ARTI

• Otitis media

• New wheezing*

• Antibiotic consumption

• Exacerbation of asthma or COPD*

• Exacerbation of CHF*

• New cardiovascular outcome*

• Discharge to long-term care facility

*As RSV immunization might have an impact on the outcomes during the off-season, a sensitivity analysis is recommended to analyze the outcomes from the start of the immunization campaign until the day of data cutoff (ie, similar to RSV-specific outcomes).

Some of the nonspecific outcomes require additional restrictions of the study populations. For wheezing and cardiovascular outcomes, the study populations are restricted to study subjects without previously diagnosed conditions. On the contrary, exacerbation of asthma, COPD, and CHF are restricted to study subjects who have been previously diagnosed with such conditions. Additional age restrictions are also included for outcomes of new wheezing and discharge to long-term care facility (Supplementary Table 4).

To estimate the presence of residual confounding, an analysis of negative control outcome(s) should be included, if possible. The negative control outcomes are RSV off-season ARTI, injury, or hospitalization due to urinary tract infection, which are assumed to be unaffected by the exposure (Supplementary Table 5).

Covariates

The analyses should take into account potential confounding factors and effect modifiers. Age and sex are required covariates as well as proxies for health-seeking behavior and chronic illnesses, which can be decided at the study site. The rest of the covariates should be included in the analysis if they are identified as potential confounders. For each study population, a different set of possible covariates is presented in Supplementary Table 6. Each study site should describe the method to collect these covariates in the study protocol.

In addition, the predominant subtype (A or B) of RSV during the study period can be used as a covariate, if surveillance data on the predominant subtype are available. This covariate can be categorized into 3 groups: A-subtype period (>60% of cases caused by the subtype), B-subtype period (>60% of cases caused by the subtype), and mixed-subtype period (neither subtype caused >60% of cases).

Individual Follow-up

Individual follow-up starts on the first day of the study period for study subjects or when the subject enters the study (ie, becomes eligible for RSV immunization or birth of the newborn; see Supplementary Figure 3). Each subject is followed until the first occurrence of any one of the endpoint events:

1. Outcome of interest (ie, outcome under evaluation).

2. Death.

3. Receipt of another RSV immunization product (eg, RSV-vaccinated receives mAb) or incorrect administration of an RSV immunization product (eg, too short or too long dosing interval).

4. Moved to another region or loss to follow-up.

5. Newborn reached 6 months of age (for maternal immunization).

6. End of the study period.

An outcome other than the outcome of interest is not an endpoint event for the analysis, and the individual follow-up can vary between different outcomes. For the negative control outcome analyses, the endpoints of individual follow-up are similar as above. Each study site can apply additional endpoints or right-censoring criteria if necessary.

For the RSV-specific outcomes, day 7 from the date of the laboratory-confirmed RSV infection (or day 30 for the outcome of all-cause mortality after RSV infection) is an additional endpoint. This means that if a specific outcome (eg, hospitalization due to RSV-related LRTI) occurs on day 8 or later from the date of the laboratory-confirmed RSV infection, this outcome is not included in the study. Day 7 is an additional endpoint because RSV infection induces natural immunity, which protects the individual from new RSV infections. An exception is mortality due to RSV, and for analyzing this outcome the laboratory-confirmed RSV infection is not an endpoint event (some deaths can occur after several weeks from the RSV infection). If a study is conducted over multiple RSV seasons, individuals with an RSV infection in a previous season can reenter into the study cohort for RSV-specific outcome analysis for the new season.

Exposure status of an individual can change during the study period from unexposed to exposed (Supplementary Figure 3). Similarly, exposure state can change during the follow-up period. For example, if exposure is categorized as 0–13 or ≥14 days since immunization, the exposure state can change from one category to another during follow-up.

Statistical Analysis

Descriptive Analysis

Absolute and relative frequencies of baseline characteristics will be presented, and these should be stratified by the exposure status at the end of the individual follow-up period or at the date of birth for maternal immunization. A bivariate analysis of exposure status (at the end of individual follow-up) and covariates should be performed with an appropriate statistical test to evaluate the difference in distribution of covariates between exposed and unexposed. Distribution of exposure events and outcomes by calendar time should be reported. The percentage of positive RSV samples of all sampled tests should be presented, if available.

Measure of Effectiveness

The effectiveness is analyzed separately for each immunization product and outcome. Hazard or risk of the outcome is compared between exposed and unexposed individuals, and they are estimated using either Cox regression with calendar time in the study as the underlying time scale or Poisson regression adjusted for calendar month. All analyses should be adjusted for age, sex, proxy for health-seeking behavior, proxy for comorbidities, and other covariates that are considered as confounders at the study site.

Effectiveness of the product is estimated as

$$Effectiveness=(1\text{–}RR)\times 100\text{\%}$$

where RR is relative risk or rate ratio (ratio of incidence or hazard ratio). In addition, the analyses can be stratified by

• Age group

• Presence of at least one underlying condition

• Presence of an immunocompromising condition

• Time since immunization

  • Corresponding stratification to previous studies with influenza or COVID-19 can be implemented (eg, 0–13 and ≥14 days or 0–13, 14–90, 91–180, and ≥181 days since immunization) [21–24]

• RSV periods of subtypes A and B

• Maternal immunization when analyzing RSV immunization effectiveness among children (ie, stratification on maternal immunization to establish if maternal immunization is an effect modifier for RSV immunization among children)

Sensitivity Analysis

As a sensitivity analysis, certain nonspecific outcomes (new wheezing; exacerbation of asthma, COPD, or CHF; and new cardiovascular outcomes) can be analyzed from the start of the immunization campaign until the day of data cutoff (ie, similarly to RSV-specific outcomes). As another sensitivity analysis, the effect modification of other vaccinations received simultaneously with RSV immunization could be analyzed. This analysis is conducted by stratifying the analysis by other vaccinations received within 14 days from RSV immunization.

Presence of Residual Confounding—Negative Control Outcome Analysis

The negative control outcome analysis is conducted similarly as the measure of effectiveness, but the outcome of interest is replaced with one of the negative control outcomes. The assumption is that the risk of a negative control outcome is similar between unexposed and exposed. If the exposed have a significantly lower or higher risk of a negative control outcome, the result indicates residual confounding in the analysis.

Data Collection and Management

At a minimum, the data sources should have information on an exposure, one of the outcomes, and the required covariates. If RSV-specific outcomes are analyzed, the data sources should also include dates of laboratory-confirmed RSV infections. Data sources can include population registers, health insurance, or healthcare provider databases. Each study site is responsible for data management and a plan should be developed to describe collection, validation, and procedures of data management and data checking.

Ethical Approval

Approval of the local ethics committee is required at each study site. The following issues should be considered in the ethical statement: database or register data linkage; consent of study subjects or a statement that consent from the study subjects is not required; reporting of small case numbers; and anonymization of data.

DISCUSSION

Evaluating postmarketing effectiveness for RSV immunization products is important for several reasons. First, in the clinical trials the study population does not usually include many individuals with severe chronic illnesses or the elderly. Among these groups the effectiveness of immunization might be lower compared to others. Furthermore, evaluation of effectiveness in these special groups and comparison of different immunization products is difficult on the basis of clinical trial results. Second, in the clinical trials the immunization is highly controlled, and human errors in immunization (eg, storage of immunization products) are less likely to occur compared to the real-world setting. Third, in the clinical trials rare outcomes, such as death due to RSV infection, are uncommon, and the statistical power is limited. Fourth, large immunization campaigns might have an impact on RSV evolution, and the new strains may have immune-evasive properties. Therefore, the monitoring of effectiveness after large immunization campaigns is important.

One of the main strengths of a register-based cohort study is its large population size and low cost [25]. A large population size increases statistical power of the analysis, which is beneficial especially to evaluate rare outcomes [25–27]. Furthermore, evaluation of explorative outcomes, such as immunization effectiveness against cardiovascular outcomes [28–30], is relatively simple in register-based cohort studies compared to other study designs. Furthermore, some register studies have been able to evaluate vaccine effectiveness close to real time [31–33], which enables policy changes rapidly if needed.

The main limitation is the quality of register data. The registers usually include at least some misclassifications of the exposure or outcomes. Furthermore, applicability and linkage ability of the registers varies greatly between countries, and high-quality register data might not be available for researchers. Similarly, definition of outcomes might be suboptimal, and health-seeking behavior usually causes some degree of bias in register-based studies, which are controlled better in test-negative case-control studies [25].

Another source of bias is RSV testing policies. RSV testing is not done systematically, and RSV testing can vary between individuals depending on behavior and comorbidities (high-risk patients are tested more often). However, appropriate adjusting should limit this effect. Moreover, an imbalance in undocumented RSV infections between exposure groups can cause bias. If numerous undocumented RSV cases occur among unimmunized individuals during the early phase of the RSV season, the immunity induced by the infection protects these individuals from new RSV infections in the later phase of the season. This could cause bias on estimation of vaccine effectiveness in the late period of the RSV season. Furthermore, use of lateral-flow tests for RSV (ie, home tests) might lead to bias if they are used widely as positive home tests might influence decisions to seek RSV immunization. This would cause under- or overestimation of effectiveness.

This protocol is intended to provide guidance for national and regional health authorities in the evaluation of implementation of RSV immunization products. The generic protocol of the test-negative case-control study by von Roekel et al complements this protocol [34].

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Acknowledgments. We want to thank Arto Palmu and Ritva Syrjänen for the help during the preparation of this article.

PROMISE investigators. Harish Nair and Harry Campbell (University of Edinburgh, UK); Louis Bont (University Medical Centre Utrecht, the Netherlands); Caren van Roekel, Adam Meijer, Anne C. Teirlinck, Mirjam Knol, David Gideonse, Anoek Backx, Hester de Melker, and Lance Presser (National Institute for Public Health and the Environment, Bilthoven, the Netherlands); Topi Turunen, Hanna Nohynek, Eero Poukka, and Annika Saukkoriipi (Finnish Institute for Health and Welfare); John Paget, Jojanneke van Summeren, and Michel Dückers (Netherlands Institute for Health Services Research, Utrecht, the Netherlands); Terho Heikkinen (Department of Pediatrics, University of Turku and Turku University Hospital, Finland); Berta Gumí Audenis and Maica Llavero (Teamit Research S.L., Barcelona, Spain); Leyla Kragten and Lies Kriek (Respiratory Syncytial Virus Network [ReSViNET], the Netherlands); Kristýna Faksová, Michele Giardini, Hanne-Dorthe Emborg, and Francesca Rocchi (Statens Serum Institut, Copenhagen, Denmark); Cintia Muñoz Quiles, Javier Diez-Domingo, Charlotte Vernhes, Clarisse Demont, Aurelie Robin, David Neveu, Lydie Marcelon, Mathieu Bangert, Rolf Kramer, Oliver Martyn, Corinne Bardone, Vanessa Remy, and Sandra Chaves (Sanofi Pasteur SA, Lyon, France); Daniel Molnar, Gael dos Santos, Jean-Yves Pirçon, Bishoy Rizkalla, Elisa Turriani, Se Li, Noemie Napsugar Melegh, Philip Joosten, and Victor Preckler Moreno (GlaxoSmithKline Biologicals SA, London, UK); Aigul Shambulova, Arnaud Cheret, Delphine Quelard, Jeroen Aerssens, Karin Weber, Corinne Willame, Anna Puggina, Katherine Theis-Nyland, and Natalia Nikolayeva (Johnson & Johnson Medical, New Brunswick, New Jersey); Veena Kumar, Hadi Beyhaghi, and Vivek Shinde (Novavax, Gaithersburg, Maryland); Beate Schmoele-Thoma, Elizabeth Begier, Kena Swanson, Tin Tin Htar, Jessica Atwell, Maria Maddalena Lino, and Monica-Flavia Turiga (Pfizer Limited, New York, New York); and Bahar Ahani (AstraZeneca, Cambridge, UK).

Financial support. This project has received funding from the Innovative Medicines Initiative 2 Joint Undertaking (grant agreement 101034339). This Joint Undertaking receives support from the European Union's Horizon 2020 Research and Innovation Programme and European Federation of Pharmaceutical Industries and Associations.

Supplement sponsorship. This article appears as part of the supplement “Preparing Europe for Introduction of Immunization Against RSV: Bridging the Evidence and Policy Gap.”

Author notes

Potential conflicts of interest. E. P. received a grant from the Finnish Medical Foundation. H. N. is a member of the Finnish National Immunization Technical Advisory Group and chair of Strategic Advisory Group of Experts on Immunization for the World Health Organization. E. B. is an employee and shareholder of Pfizer, Inc. R. K. is an employee and shareholder of Sanofi. T. H. has received personal fees from Janssen, Sanofi, and MSD (for activities outside the present work). All other authors have no potential conflicts of interest.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.