Intraseasonal waning of influenza vaccine effectiveness and implications for the optimal timing of seasonal vaccination programs

Abstract

Background: Seasonal influenza vaccination programs aim to provide protection before the virus circulation begins. However, vaccine effectiveness (VE) may decline within a single influenza season due to waning immunity, antigenic drift, and season-specific factors, raising questions about optimal vaccination timing. The purpose of this review was to synthesize evidence on intraseasonal waning of seasonal influenza VE and examine the implications and key considerations for optimizing the timing of seasonal vaccination in Canada.

Methods: A scoping review was conducted, with searches of MEDLINE, Embase, Scopus, Cochrane Library, and ProQuest Public Health Database that identified studies published from 2010 to June 2024, with an update in July 2025. Eligible studies included clinical trials, observational studies, systematic reviews, and modelling studies reporting at least two VE estimates by time since vaccination within a season or assessing vaccination timing and its impact on influenza outcomes.

Results: Forty-nine studies met inclusion criteria, including 37 assessing intraseasonal waning and 12 examining vaccination timing. Overall, VE was generally highest within one to three months post-vaccination and declined over the season (e.g., three to six months post-vaccination). Waning was more consistently observed for influenza A, particularly A(H3N2), and among adults aged 60 years and older. Modelling studies suggested that delaying vaccination could reduce influenza burden under select late-peaking or fast-waning scenarios; however, estimated benefits were generally small and highly sensitive to assumptions.

Conclusion: Intraseasonal waning of influenza VE is consistently observed, particularly for influenza A and in older adults, but represents a gradual decline that does not typically undermine season-long effectiveness. Evidence does not support substantial or consistent population-level benefits from delaying vaccination.

Introduction

Seasonal influenza in Canada follows a predictable annual pattern, typically beginning in December and lasting 12–16 weeks, although seasons may start as early as October or as late as February and can extend up to 20 weeks ^{Footnote 1}. Influenza activity generally progresses from west to east across the country ^{Footnote 1}. Annual vaccination is the primary strategy for preventing influenza and its complications, and in Canada and other northern hemisphere countries, immunization programs typically begin in October to ensure protection before virus circulation increases ^{Footnote 1Footnote 2Footnote 3}.

While fall vaccination is well established as the recommended timing for seasonal influenza immunization, research continues to examine how vaccination timing may influence effectiveness across the influenza season. Emerging evidence indicates that vaccine effectiveness (VE) may decline within a season due to waning immunity, prior immunity, underlying health conditions, and changes in circulating strains ^{Footnote 4Footnote 5Footnote 6}. Observational studies have reported increasing odds of laboratory-confirmed influenza (LCI) with longer time since vaccination, particularly for A(H3N2), raising concerns that early vaccination could result in reduced protection later in the season, especially among populations at higher risk of severe outcomes ^{Footnote 7}.

Beyond waning immunity, VE and vaccination impact are influenced by season-to-season variability in influenza circulation, antigenic drift, vaccine-strain match, and programmatic factors, such as supply and uptake ^{Footnote 8}.

While prior reviews have examined intraseasonal waning, an integrated assessment that jointly considers waning and evidence informing vaccination timing has been lacking. This scoping review addresses this gap by mapping the available evidence and synthesizing insights to support evidence-based public health decision-making and ongoing national work to optimize influenza vaccination strategies in Canada. The objectives were to: i) summarize recent evidence on intraseasonal waning of protection conferred by seasonal influenza vaccines; and ii) identify key considerations related to the optimal timing of seasonal influenza vaccination.

Methods

This scoping review followed the Joanna Briggs Institute methodological framework and is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for scoping reviews (PRISMA-ScR) ^{Footnote 9Footnote 10}. The protocol was registered on the Open Science Framework (OSF ID: Rkzc9) ^{Footnote 11}.

Search strategy

A comprehensive search strategy was developed with a research librarian and applied to MEDLINE, Embase, Scopus, Cochrane Library, and the ProQuest Public Health Database (Appendix, Supplemental material, Tables S1–S8). Searches included English and French language publications from January 1, 2010 (post-2009 H1N1 strain introduction) to June 17, 2024, and were updated on July 15, 2025.

Eligibility criteria

Studies included individuals six months of age and older in temperate regions of the northern or southern hemisphere, as well as Australia and New Zealand, due to their research quality and mostly having a temperate climate. Eligible study designs included primary studies (i.e., clinical trials and observational studies), systematic reviews, and meta-analyses.

For evidence on waning protection conferred by seasonal influenza vaccination, studies were required to report two VE estimates by time since vaccination within a single influenza season, using LCI or influenza-related hospitalization outcomes. For evidence related to optimal timing of influenza vaccination, eligible studies assessed vaccination timing through comparisons of earlier versus later vaccine administration in relation to influenza outcomes or modelled the impact of vaccination timing on influenza-related outcomes.

Studies of pandemic monovalent or investigational vaccines, immunogenicity-only outcomes, relative VE without absolute estimates, non-human studies, and studies focused solely on vaccination coverage or uptake were excluded.

Study selection and data extraction

Citations were imported into Zotero for deduplication and uploaded to DistillerSR (Evidence Partners Inc, Ottawa, Canada) for screening. Titles and abstracts were screened by three independent reviewers. A record was included if at least one reviewer deemed it potentially eligible; exclusion required agreement by two reviewers. The DistillerSR artificial intelligence (AI) review tool was used to support exclusion at title and abstract level using a conservative relevance threshold of 0.2 (scores near 0 indicating high-confidence exclusion). All AI-flagged exclusions were manually reviewed. A record was excluded only when at least one reviewer agreed with the AI’s recommendation. Disagreements were resolved through discussion. Records not flagged by AI underwent standard dual human screening. Full-text screening was conducted by two independent reviewers, with discrepancies resolved through discussion.

Data extraction was completed by one reviewer and validated by a second, capturing study characteristics (e.g., population, setting, design), interventions, outcomes, and findings relevant to each objective.

Risk of bias assessment

Risk of bias was assessed to provide context for policy interpretation. Critical appraisal was conducted by one reviewer and validated by a second reviewer. Tools included the revised Cochrane Risk of Bias Tool for randomized trials ^{Footnote 12}, the Risk of Bias in Non-randomized Studies of Interventions ^{Footnote 13Footnote 14} and A MeaSurement Tool to Assess Systematic Reviews (AMSTAR 2) ^{Footnote 15}. Modelling studies were not assessed for risk of bias.

Evidence synthesis

Findings were summarized narratively and VE estimates stratified by time since vaccination were presented graphically in a forest plot. Findings related to optimal timing of influenza vaccination were synthesized separately, highlighting key assumptions and factors influencing timing decisions.

Results

Overall, 49 studies met inclusion criteria (Figure 1), covering influenza seasons from 2005–2006 to 2023–2024 across multiple countries (Table 1 and Table 2). Most studies used a test-negative design (TND) and assessed intraseasonal waning of VE against LCI.

Figure 1 - Text description

Study selection flow diagram based on PRISMA framework for identification, screening, eligibility, and inclusion of studies in the review (n=49).

The diagram illustrates the process of study identification, screening, and selection following a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) approach. A total of 9,333 records were initially identified through database searches, including MEDLINE (n=2,013), Embase (n=3,264), Scopus (n=2,595), Cochrane Central (n=631), and ProQuest (n=830). Prior to screening, 4,626 duplicate records were removed, resulting in 4,707 unique records for title and abstract screening.

During the screening stage, 4,376 records were excluded based on predefined eligibility criteria. A total of 331 reports were then sought for full-text retrieval and assessed for eligibility. Of these, 282 reports were excluded for the following reasons: inappropriate outcomes (n=221), inappropriate intervention or comparator (n=5), inappropriate study design (n=6), lack of original data or insufficient/relevant data (n=30), studies conducted in tropical climates outside the scope of the review (n=10), and other reasons (n=10).

Following full-text assessment, 49 studies met the inclusion criteria and were included in the final review.

Intraseasonal waning of seasonal influenza vaccine protection

Thirty-seven studies assessed intraseasonal waning, including 34 TND studies, one cohort study, one randomized controlled trial (RCT), and one systematic review and meta-analysis (Table 1).

The RCT was assessed as low risk of bias (Table 1). Most non-randomized studies were rated low (n=27) to moderate (n=7) risk of bias, with one at serious risk. The systematic review received a moderate confidence rating. Common limitations included residual confounding (e.g., health status) and missing data.

Overall population-level estimates

Across 28 studies reporting population-level estimates without age stratification, VE was highest shortly after vaccination and generally declined over the course of the influenza season (Figure 2) ^{Footnote 4Footnote 7Footnote 16Footnote 17Footnote 18Footnote 19Footnote 20Footnote 21Footnote 22Footnote 23Footnote 24Footnote 25Footnote 26Footnote 27Footnote 28Footnote 29Footnote 30Footnote 31Footnote 32Footnote 33Footnote 34Footnote 35Footnote 36Footnote 37Footnote 38Footnote 39Footnote 40Footnote 41}. Waning was more consistently observed for influenza A than for influenza B, particularly A(H3N2) (Supplemental material, Table S9). For LCI, most studies reported waning within three to six months post-vaccination, although magnitude varied; one study reported no evidence of decline (Figure 2) ^{Footnote 30}.

Figure 2 - Text description

Author year	Age group	VE (95% CI)
Abou Chakra 2025	0 to 3 mo	49.8 (46.1–53.2)
Abou Chakra 2025	3 to 6 mo	42.8 (37.2–56.1)
Castilla 2013	<3 mo	61.0 (5.0–84.0)
Castilla 2013	4 mo	42.0 (−39.0–75.0)
Castilla 2013	>4 mo	−35.0 (−211.0–41.0)
Hu 2022	<2 mo	50.0 (41.0–58.0)
Hu 2022	<2 to 4 mo	39.0 (31.0–47.0)
Hu 2022	4 to 6 mo	17.0 (0.0–32.0)
Jimenez-Jorge 2013	1 to 3 mo	84.0 (−138.0–99.0)
Jimenez-Jorge 2013	3 to 4 mo	19.0 (−149.0–73.0)
Jimenez-Jorge 2013	4 to 5 mo	58.0 (−19.0–86.0)
Pebody 2016	<3 mo	51.4 (29.9–66.3)
Pebody 2016	>3 mo	52.7 (39.2–63.2)
Pebody 2019	<3 mo	18.7 (−7.7–38.6)
Pebody 2019	>3 mo	10.4 (−19.9–33.0)
Radin 2016	<1 mo	65.0 (6.0–87.0)
Radin 2016	<1 to 2 mo	53.0 (14.0–75.0)
Radin 2016	2 to 3 mo	67.0 (38.0–82.0)
Radin 2016	3 to 6 mo	60.0 (41.0–73.0)
Radin 2016	6 to 12 mo	−16.0 (−117.0–38.0)
Regan 2019	<3 mo	60.0 (26.0–78.0)
Regan 2019	3 mo	42.0 (1.0–66.0)
Regan 2019	4 mo	25.0 (−30.0–57.0)
Regan 2019	>4 mo	19.0 (−73.0–62.0)
Sullivan 2014	<3 mo	37.0 (−29.0–69.0)
Sullivan 2014	>3 mo	18.0 (−83.0–63.0)
Zhang 2024	<3 mo	70.1 (−145.–96.4)
Zhang 2024	3 to 5 mo	18.7 (−42.4–53.6)
Zhang 2024	>5 mo	21.2 (−18.8–47.7)
Zhu 2025a	<1 mo	59.0 (56.0–61.0)
Zhu 2025a	1 to 2 mo	52.0 (50.0–53.0)
Zhu 2025a	2 to 3 mo	39.0 (38.0–41.0)
Zhu 2025a	3 to 4 mo	34.0 (32.0–36.0)
Zhu 2025a	4 to 5 mo	35.0 (32.0–37.0)
Zhu 2025a	5 to 6 mo	34.0 (30.0–37.0)
Zhu 2025a	>6 mo	23.0 (20.0–27.0)
Zhu 2025b	<3 mo	39.0 (32.0–46.0)
Zhu 2025b	3 to 6 mo	30.0 (19.0–40.0)
Zhu 2025b	>6 mo	28.0 (5.0–46.0)

Three studies evaluated VE against influenza-related hospitalization ^{Footnote 25Footnote 26Footnote 36}. Two reported declining VE with increasing time since vaccination, with reductions of approximately 20%–35% beyond 120 days ^{Footnote 26Footnote 36}, while one observed relatively stable VE (~39%) with wider confidence intervals (CIs) later in the season ^{Footnote 25}. Strain-specific estimates from Lewis et al. showed greater stability for influenza B than influenza A, with VE declining from 72% to 62% for influenza B and from 42% to 16% for influenza A across comparable time intervals ^{Footnote 26}.

A meta-analysis by Young et al., which included studies also captured in this review, reported significant pooled declines in VE between early (15–90 days) and later (91–180 days) intervals for influenza A(H3N2) (∆VE=−33%, 95% CI: −57%–−12%, n=11 studies) and influenza B (∆VE=−19%, 95% CI: −33%–−6%, n=6 studies), but not for A(H1N1) (∆VE=−8%, 95% CI: −27%–21%, n=5 studies) ^{Footnote 7}. These results aligned with patterns observed in individual studies.

Analyses modelling time since vaccination as a continuous variable further supported waning. Ray et al. reported a 16% increase in the odds of influenza infection per 28 days post-vaccination (odds ratio [OR]: 1.16, 95% CI: 1.13%–1.20%) ^{Footnote 33}. Ferdinands et al. estimated average VE declines per 30 days of 7.5% (95% CI: 0.3%–16.3%) for A(H3N2), 8.5% (95% CI: 3.0%–17.0%) for A(H1N1), and 8.0% (95% CI: 1.4%–21.9%) for influenza B/Yamagata among hospitalized adults, with similar findings for infection outcomes reported by Ferdinands et al. ^{Footnote 40Footnote 41}.

Overall, population-level evidence indicated progressive intraseasonal waning, with more consistent evidence available for LCI than for hospitalization outcomes, and more pronounced for influenza A than for influenza B.

Children

Eight studies reported paediatric-specific VE estimates ^{Footnote 4Footnote 16Footnote 42Footnote 43Footnote 44Footnote 45Footnote 46Footnote 47}. Overall, VE was highest shortly after vaccination and generally remained stable or declined modestly over the season. For LCI, early–late differences were small, with changes in VE of approximately 5%–8% across the season ^{Footnote 4Footnote 16Footnote 42Footnote 43} (Supplemental material, Table S10).

Two studies assessed influenza-related hospitalization in children and reported VE declines of approximately 20% and 30% within six or nine months post-vaccination ^{Footnote 44Footnote 45}. Age-stratified analyses showed broadly similar patterns across paediatric age groups (e.g., <2 years, 3–5 years, and 6–17 years), with modest seasonal variation for LCI and larger declines observed for hospitalization outcomes ^{Footnote 16Footnote 44Footnote 45}.

Two studies reported increasing odds of influenza positivity with time since vaccination ^{Footnote 46Footnote 47}, although significant waning was observed for A(H3N2) in only one study ^{Footnote 46}. Overall, VE in children remained protective throughout the season, with limited evidence of substantial intraseasonal waning.

Adults 18 to 64 years of age

Three studies specifically assessed intraseasonal waning of influenza in adults aged 18–64 years ^{Footnote 4Footnote 16Footnote 46}. In general, VE was highest shortly after vaccination, with variable changes over time. One study reported a modest decline (~5%) at three to six months, another reported larger declines (~30%) beyond five months, and a third found no significant association between time since vaccination and influenza positivity (adjusted OR: 1.00; 95% CI: 0.95%–1.06%) ^{Footnote 4Footnote 16Footnote 46}.

Overall, evidence in this age group was limited and heterogeneous, precluding firm conclusions regarding the extend of intraseasonal waning.

Older adults

Ten studies evaluated intraseasonal waning in adults 60 or 65 years and older ^{Footnote 4Footnote 16Footnote 19Footnote 22Footnote 24Footnote 40Footnote 46Footnote 48Footnote 49Footnote 50}. Most reported declining VE over time, with patterns varying by outcome and strain. For LCI, VE was generally highest within one to three months post-vaccination and declined thereafter (Table 3), with steeper waning for A(H3N2) than influenza B in one study ^{Footnote 24}.

Hospitalization outcomes showed similar temporal patterns, often with less pronounced declines and widening CIs later in the season. Mira-Iglesias et al. reported VE decreasing from 54% at 15–82 days to 50% at 123–177 days post-vaccination ^{Footnote 48}. In Seppälä et al, VE among adults 65–79 years declined from 34% at 7–89 days to −3% beyond 180 days, whereas VE remained around 40% in adults aged 80 years and older, with CI widening over time ^{Footnote 49}.

Continuous-time analyses generally supported waning ^{Footnote 40Footnote 46Footnote 50}. Domnich et al. reported a significant 6%–7% increase in odds of influenza per week since vaccination for any influenza and A(H3N2), starting 14 days post-vaccination ^{Footnote 46}. Ferdinands et al. estimated average declines in VE of approximately 10% per 30 days post-vaccination for hospitalization ^{Footnote 40}. However, another study observed seasonal declines in VE, but sensitivity analyses accounting for calendar week and time since vaccination did not demonstrate a clear gradual decline, with overlapping CIs across weekly VE estimates ^{Footnote 50}.

Overall, VE in older adults was highest shortly after vaccination and declined over the course of the season, with considerable heterogeneity across studies.

Optimal timing of influenza vaccine administration

Twelve studies assessed optimal vaccination timing, including ten modelling studies and two observational studies (Table 1 and Table 2) ^{Footnote 51Footnote 52Footnote 53Footnote 54Footnote 55Footnote 56Footnote 57Footnote 58Footnote 59Footnote 60Footnote 61Footnote 62}. Modelling studies evaluated vaccination timing under varying assumptions related to VE, intraseasonal waning, season timing, and vaccination coverage (Table 2), while observational studies compared influenza outcomes following earlier versus later vaccination in specific population (Table 1).

Modelling results suggested that under some late-peaking or fast-waning scenarios, delaying vaccination by one to two months could reduce influenza burden when coverage was held constant ^{Footnote 51Footnote 52Footnote 53Footnote 57Footnote 58Footnote 60}. However, the estimated gains were generally small, typically corresponding to less than a 5% difference in prevented cases or hospitalizations between timing scenarios ^{Footnote 51Footnote 57Footnote 58}. Several models further showed that even modest reductions in vaccination coverage associated with delayed uptake could offset or reverse these potential benefits ^{Footnote 51Footnote 52Footnote 58}. Economic modelling favoured earlier fall vaccination, particularly for children and pregnant individuals ^{Footnote 54Footnote 55Footnote 56}. Models of compressed vaccination periods (e.g., October–May vs August–May) or two-dose schedules in older adults showed benefits under select late-peaking or fast-waning scenarios, which diminished when coverage or adherence declined ^{Footnote 58Footnote 59}.

Two observational studies reported mixed findings. Glinka et al. observed higher influenza-related outcomes among HIV-infected individuals vaccinated earlier (before mid-November) compared with later (after mid-November) vaccination ^{Footnote 61}. In contrast, a population-based cohort study in young children found that those born in October, who were disproportionately vaccinated in October, were least likely to receive an influenza diagnosis, particularly compared with children born in August, who tended to be vaccinated earlier, and those born in December, who tended to be vaccinated later ^{Footnote 62}. Both studies were subject to potential residual confounding.

Overall, modelling evidence suggests that the impact of modifying vaccination timing is highly sensitive to assumptions about waning, season timing, and vaccination coverage, with potential gains from delayed strategies generally small and easily negated by reductions in uptake, while observational evidence remains limited and inconsistent.

Discussion

This scoping review synthesized evidence on intraseasonal waning of influenza VE and its implications for seasonal vaccination timing, addressing gaps identified in recent national evidence syntheses. Across age groups and study designs, VE was generally highest within one to three months after vaccination and declined progressively over the influenza season (e.g., 3–6 months post-vaccination). Waning was most consistently observed for influenza A, particularly A(H3N2), and among older adults. Evidence for hospitalization outcomes was more limited than for LCI, but available studies suggested similar temporal patterns with greater uncertainty.

These findings align with earlier evidence syntheses, including the 2017 meta-analysis by Young et al. and a prior systematic review and meta-analysis of immunogenicity and antibody persistence, which demonstrated biologically plausible declines in vaccine-induced immunity by six months post-vaccination ^{Footnote 6Footnote 63}. However, the magnitude and consistency of waning varied across studies, likely reflecting differences in circulating strains, season timing, population characteristics, and analytic approaches. Importantly, observed late season declines may reflect not only waning immunity but also antigenic drift and reduced vaccine-strain match, particularly during A(H3N2)-dominant seasons ^{Footnote 64}.

Evidence on vaccination timing suggested that potential benefits of delaying vaccination are context-dependent and generally modest. Modelling studies indicated that in specific scenarios (e.g., late-peaking seasons or rapid waning), later vaccination could reduce influenza burden when coverage was held constant ^{Footnote 51Footnote 52Footnote 53Footnote 57Footnote 58Footnote 60}. However, projected gains were small and highly sensitive to assumptions about waning, season timing, and uptake, with several models showing that modest coverage reductions could negate benefits ^{Footnote 51Footnote 52Footnote 58}. Observational studies examining vaccination timing reported mixed findings and were subject to residual confounding ^{Footnote 61Footnote 62}.

Overall, maintaining high vaccination coverage appears to be a dominant determinant of population-level impact. While waning is a consistent feature of influenza VE, the implications for modifying vaccination timing remain uncertain and vary by population, outcome, and season. Early-season vaccination provides protection across most seasons, whereas delayed strategies may leave individuals unprotected during early circulation, given variability in season onset and peak timing.

Limitations

Interpretation should consider methodological limitations. Most studies relied on observational designs and were susceptible to residual confounding, time-varying biases and misclassification. Heterogeneity in definition of time since vaccination and adjustment for calendar time and circulating strains further complicates comparisons. Fewer studies assessed hospitalization outcomes, often with widening CIs later in the season. Modelling studies additionally depend on assumptions that may not reflect real-world conditions.

This review provides a comprehensive synthesis across multiple seasons, regions, age groups, and outcomes, integrating evidence on waning with vaccination timing. However, heterogeneity across studies, and limited data for certain populations, including pregnant individuals and immunocompromised persons, remain important gaps.

Conclusion

Intraseasonal waning of influenza VE is consistently observed, particularly for influenza A and among older adults, but represents a gradual decline that does not typically undermine season-long effectiveness. Although timing adjustments may theoretically influence protection under specific scenarios, current evidence does not demonstrate substantial or consistent population-level benefits from modifying vaccination timing. These findings highlight the importance of considering waning alongside seasonal variability, strain dynamics, and programmatic factors, and underscore the need for further research on how these elements interact to influence influenza outcomes.

Authors' statement

Both, Pamela Doyon-Plourde and Fazia Tadount contributed equally to the conceptualization, analysis and writing of the manuscript.

PD-P — Conceptualization, data curation, formal analysis, methodology, validation, visualization, writing–original draft, writing–review & editing
FT — Conceptualization, investigation, formal analysis, methodology, validation, writing–original draft, writing–review & editing
AG — Conceptualization, investigation, writing–review & editing
CL — Investigation, validation, writing–review & editing
M-MU — Investigation, validation, writing–review & editing
NS — Conceptualization, writing–review & editing
WS — Conceptualization, writing–review & editing
AS — Conceptualization, supervision, writing–review & editing

Competing interests

None.

ORCID numbers

Pamela Doyon-Plourde — 0000-0002-8738-0055
Fazia Tadount — 0009-0001-4867-5942
Anabel Gil — 0009-0001-6179-0371
Calin Lazarescu — 0000-0001-5778-9362
Marie-Michelle Ursu — 0009-0004-4541-2567
Winnie Siu — 0009-0001-5772-1509
Angela Sinilaite — 0009-0001-4488-7454

Acknowledgements

This review was developed and conducted at the request and under the guidance of the National Advisory Committee on Immunization (NACI) Influenza Working Group (IWG). The authors gratefully acknowledge the contributions of NACI IWG content experts in refining the protocol to support the feasibility and relevance of the review for key knowledge users. The authors also thank Katarina Gusic for her assistance with screening and study selection.

Funding

This work was supported by the Public Health Agency of Canada.

Appendix

Supplemental material is available upon request to the author: naci-ccni@phac-aspc.gc.ca

• Previous Page
• Table of contents
• Next page