A prospective, comparative cohort analysis of influenza antibody waning in Michigan and Hong Kong during periods of low influenza circulation

Abstract

Background:

Reduced influenza transmission during the COVID-19 pandemic prompted concern about waning of population immunity that could lead to subsequent surges in circulation. We evaluated this by comparing longitudinal influenza antibody titers in Michigan and Hong Kong, two regions with reduced influenza transmission during the COVID-19 pandemic.

Methods:

In two prospective cohort studies (HIVE, Michigan; EPI-HK, Hong Kong), we analyzed longitudinal serum samples collected from 2020 through 2023 from participants without documented influenza virus infection or vaccination. Sera were tested using hemagglutination inhibition assays (HAI) against relevant vaccine strains. Geometric mean titers (GMTs) and fold changes were estimated by region and time. Linear mixed-effects models were used to assess temporal trends.

Results:

We analyzed 173 sera from 57 HIVE participants and 259 sera from 60 EPI-HK participants. Initial GMTs in 2020–21 ranged from 12.3–123.4 in HIVE and 6.3–40.9 in EPI-HK (B/Yamagata–H1N1). Fold changes in GMTs ranged from 1.2–2.6 in HIVE and 0.7–1.0 in EPI-HK. In HIVE models, no significant change in HAI titers over time was detected. In EPI-HK, small but statistically significant monthly declines were observed for select H1N1 (A/Michigan) and H3N2 (A/Hong Kong) strains (e.g., A/Hong Kong: −0.98%, 95% CI: −1.82% to −0.11%).

Conclusion:

Minimal HAI titer waning was observed in both regions. In some cases, antibody levels increased in Michigan, possibly indicating cryptic circulation of strains prior to the 2022/23 influenza season. These findings do not support an “immunity debt” during pandemic restrictions and could help explain the lack of a substantial surge in influenza impact after the COVID-19 pandemic.

Background

The COVID-19 pandemic and the implementation of mitigation strategies changed the global circulation of respiratory viruses, including seasonal influenza. Interventions such as international travel restrictions, school closures, masking, and social distancing led to low influenza activity between 2020 and 2022, with some regions reporting the absence of influenza transmission^1–3. While these interventions effectively reduce transmission, concerns remained regarding the longer-term impact on population immunity to respiratory viruses. As a consequence of influenza’s seasonality, waning of antibody levels has long been a topic of interest, particularly in informing vaccine timing and the potential need for a second dose. However, it has been challenging to isolate the role of natural infection in antibody waning, given frequent infections even among vaccinated individuals. The COVID-19 pandemic provided a unique opportunity to study antibody waning in the near absence of both influenza virus exposure and vaccination. Specifically, the interruption of routine exposure to influenza viruses through infection or vaccination had led to waning humoral immunity, leaving populations potentially more susceptible when influenza circulation resumed^4–7.

In the US, influenza activity demonstrated a clear seasonal pattern prior to the COVID-19 pandemic, typically beginning in November and continuing through March each year⁸. In response to rising COVID-19 cases, Michigan implemented a series of early and aggressive mitigation measures in March 2020, including statewide K–12 school closures beginning March 12, followed by the “Stay Home, Stay Safe” executive order issued on March 23⁹. These interventions, which remained in place until June 1, 2020, significantly reduced transmission of SARS-CoV-2 and effectively disrupted the 2019–2020, and 2020–2021 influenza seasons. Following the implementation of these measures, including school closures, widespread masking, social distancing, and heightened public awareness, no influenza cases were detected in the cohort from March 2020 through October 2021. Influenza transmission re-emerged during the 2021–2022 season, dominated by H3N2 strains. In Hong Kong, universal masking was mandated from mid 2020 until early 2023. Social distancing measures were introduced in early 2020, including restrictions on public gatherings, closures of high-risk venues, and dining limitations. Schools were closed in late January 2020 and shifted to remote learning, with intermittent reopenings followed by reclosures during subsequent waves. Quarantine for inbound travelers began in February 2020 and remained in effect until September 2022¹⁰. As a result of these layered and prolonged interventions, no community-detected influenza cases were reported in the community surveillance between March 2020 and January 2023. Influenza began to spread again in Hong Kong in early 2023 after the relaxation of mitigation measures, initially predominated by H1N1 viruses, followed by H3N2 strains ¹¹.

Understanding the dynamics of influenza antibody at a population level is important for future vaccination strategies and disease burden estimation. In particular, the rate by which antibody wanes during periods of minimal antigenic drift or shift has been difficult to estimate. Although short-term kinetics of antibody waning after vaccination or infection have been characterized in pre-pandemic studies^12–15, there remains limited evidence on the long-term antibody responses in the absence of repeated exposure. Moreover, regional differences in vaccination coverage, previous exposure histories, seasonality, circulating strains, and demographic factors may influence patterns of antibody waning, which could therefore differ across populations^12,16–19.

To address these questions, we evaluated longitudinal antibody responses in two prospective cohort studies: the Household Influenza Vaccine Evaluation (HIVE) cohort in Michigan, United States, and the Evaluating Population Immunity in Hong Kong (EPI-HK) cohort. Both studies collected serial serum samples from participants who had neither laboratory-confirmed influenza virus infections nor influenza vaccinations during the pandemic period (2020–2023). These cohorts represent two distinct epidemiological and public health contexts: Michigan experienced moderate influenza vaccination uptake and a relatively early relaxation of COVID-19 mitigation measures⁹, while Hong Kong maintained stringent control policies till 2022, including prolonged mask mandates, quarantine protocols, and limited social mixing²⁰. By comparing cohorts in two distinct epidemiological and cultural settings, we aimed to gain unique insights into the durability of influenza immunity.

Methods

2.1. Participant Enrollment and Data Collection

HIVE Cohort

The Household Influenza Vaccine Evaluation (HIVE) study is a prospective cohort of households with children conducted in Ann Arbor, Michigan, and the surrounding areas. Recruitment into the cohort occurred annually, with households committing to one year of participation at a time while being encouraged to participate for multiple seasons⁸. For this analysis, we included individuals who participated in the study between 2020 and 2023, incorporating available exposure history data from prior years.

Each year, participants completed an enrollment survey and were prospectively followed for acute respiratory illness (ARI) symptoms. Household members who reported at least two ARI symptoms were asked to provide combined nasal and throat swab specimens, which were tested for influenza using reverse transcription polymerase chain reaction (RT-PCR)⁸. Annual influenza vaccination status, including historical vaccination records, was obtained from electronic medical records and the state of Michigan’s vaccination registry.

Since 2011, blood samples have been collected twice annually from participants. Pre-season serum specimens were collected in the fall and post-season serum specimens were collected in the summer. We also collected blood samples before and after receipt of a respiratory virus vaccine, such as the influenza or SARS-CoV-2 vaccine. Since fall 2021, pre- and post-vaccine bloods were collected year-round with the timing of collection occurring within 14 days prior to vaccination and 7 and 28 days after vaccination.

For this analysis, eligible participants met the following criteria:

1. Participated in the study from fall 2020 to fall 2022.

2. No self-reported or influenza vaccination records from fall 2020 to fall 2022.

3. Availability of longitudinal serological samples (≥1 blood sample/year) across four collection periods: pre-season 2020–21, post-season 2020–21, pre-season 2021–22, and post-season 2021–22.

All adult household members provided informed consent for themselves and their children; children age 7 years and older provided assent to participation. The study was approved by the University of Michigan Medical School Institutional Review Board (HUM00118900; HUM00198212).

EPI-HK Cohort

The Evaluating Population Immunity in Hong Kong (EPI-HK) study is a longitudinal community-based cohort study involving intensive follow-up of around 2000 individuals of all ages enrolled from the general community in Hong Kong²¹. Participants are followed up with blood draws every 6 months, at which information on health status, vaccination history, and hospitalization are also collected. Participants are also monitored for ARI events but will not be the focus of this study.

For this analysis, eligible participants met the following criteria:

1. Participated in the study from fall 2020 to fall 2022.

2. No self-reported or influenza vaccination records from fall 2020 to fall 2022.

3. Provide longitudinal blood samples collected approximately every six months for four periods from fall 2020 (July–December 2020) to fall 2022 (July–December 2022).

4. Be randomly selected from six age groups (0–17, 18–29, 30–44, 45–54, 55–64, 65–74 years), with 10 participants per age group.

An additional 30 participants per age group were randomly selected for inclusion in sensitivity analyses.

Participants aged 18 years or above provided written informed. For participants aged 11–17 years, proxy written consent were obtained from parents or legal guardians in addition to their own written assent. For participants aged 0–10 years, proxy written consent from parents or legal guardians was obtained. The study was approved by the University of Hong Kong Hospital Authority Hong Kong West Cluster Institutional Review Board (UW19–720).

2.2. Laboratory Methods

All serum specimens were tested for antibody responses against each influenza vaccine strain using hemagglutination inhibition (HAI) assays in each cohort. Antigens representing seasonally relevant vaccine strains were used for testing across both cohorts (Table 1). The additional EPI-HK participants were tested for a subset of antigens (H1N1: A/Brisbane/2/2018, H3N2: A/Hong Kong/4801/2014, B/Phuket/3073, B/Washington/02/2019). Serologic testing was performed independently in the investigators’ laboratory at the University of Michigan School of Public Health and at the University of Hong Kong School of Public Health using standard procedures²² (detailed procedures described in the supplementary methods).

Table 1:

HAI antigen tested in each cohort

Influenza Virus Strain	HIVE HAI antigen (year included in seasonal vaccination1)	EPI-HK HAI antigen (year included in seasonal vaccination1)
A(H1N1)	A/Michigan/45/2015 (2017–2019)2, A/Brisbane/2/2018 (2019–2020), A/Idaho/07/2018 (2019–2020)2	A/Michigan/45/2015 (2017–2019), A/Brisbane/2/2018 (2019–2020), A/Hawaii/70/2019 (2020–2022)2, A/Victoria/2570/2019 (2021–2022),
A(H3N2)	A/Singapore/INFIMH-16-0019/2016 (2018–2019)2, A/Kansas/14/2017 (2019–2020)2	A/Hong Kong/4801/2014 (2016–2018), A/Singapore/INFIMH-16-0019/2016 (2018–2019), A/Kansas/14/2017 (2019–2020), A/Cambodia/e0826360/2020 (2021–2022)
B/Yamagata	B/Phuket/3072/2013 (2015–2023)	B/Phuket/3073/2013 (2015–2023)
B/Victoria	B/Colorado/06/2017 (2018–2020)2	B/Washington/02/2019 (2020–2022)

¹Influenza strain included in the seasonal influenza vaccine formulation for the indicated year.

²Cell-culture derived strains

2.3. Statistical Methods

Baseline participant and household characteristics were summarized and compared between the two cohorts using chi-squared (χ²) tests or Fisher’s exact tests for categorical variables. HAI titers were log2-transformed, and geometric mean titers (GMT) with 95% confidence intervals (CI) were calculated for each sample collection period in both cohorts. Fold changes in GMT were calculated by comparing titers from the last and first sample collection periods within the study period.

To evaluate antibody waning, we modeled the association between log2-transformed HAI antibody titers and time since baseline (in months), stratified by antigen strain and adjusting for participant’s age. Both univariate and multivariate linear mixed-effects models were fitted separately for each strain. Models included a random intercept for each participant to account for individual baseline differences.

The multivariate model was specified as:

$${\log }_2({\mathrm{titer}}_{\mathrm{ij}})={\mathrm{\beta }}_0+{\mathrm{\beta }}_1{(\mathrm{months}\mathrm{since}\mathrm{baseline})}_{\mathrm{ij}}+{\mathrm{\beta }}_2{(\mathrm{age})}_{\mathrm{ij}}+{\mathrm{b}}_{0\mathrm{i}}+{\mathrm{\epsilon }}_{\mathrm{ij}}$$

Model selection was conducted using the Akaike Information Criterion (AIC), where the model with the lowest AIC value was chosen as the best-fitting model, appropriately capturing both main effects and relevant random effects. All analyses were conducted using R (version 4.3.1).

2.4. Sensitivity Analyses

To assess the robustness of our findings, we included an additional 30 EPI-HK participants per age group in the descriptive and statistical analyses to examine whether the inclusion of these participants affected the overall results for the sensitivity analysis.

Results

3.1. Characteristics of Households and Participants

A total of 57 participants (173 serum samples) from the HIVE cohort and 60 participants (259 serum samples) from the EPI-HK cohort were included in the analysis (Figure 1, table 2, table S1). There was no significant difference in sex distribution between the two cohorts (table 2). Participants in the HIVE cohort were enrolled at different times, with the majority enrolling before 2019; however, two participants enrolled in 2020. This later enrollment limited our ability to accurately capture their past exposure to influenza. Participants in the EPI-HK cohort were enrolled from July 2020 to December 2020.

Figure 1:

Influenza surveillance trends and serum sample collection density in Michigan and Hong Kong, 2016–2024 The influenza epidemiology curve for Hong Kong reflects surveillance data from the Hong Kong government laboratory. Percent positivity represents the proportion of respiratory specimens that tested positive for influenza among all specimens submitted to the Public Health Laboratory at the Centre for Health Protection and the Hospital Authority. These specimens were collected from patients presenting with influenza-like illness (ILI) in public/private outpatient clinics or during hospitalization. The serum sample collection density is derived from cohort-based sampling in Michigan and Hong Kong.

Table 2:

Characteristics of participants

Characteristic	EPI-HK, N = 601	HIVE, N = 571	p-value2
Age	43.1 (10.0, 74.9)	29.5 (2.5, 55.2)	<0.001
Age categories			<0.001
0–17	10 (17%)	21 (37%)
18–29	10 (17%)	6 (11%)
30–44	10 (17%)	18 (32%)
45–54	10 (17%)	11 (19%)
55–64	11 (18%)	1 (1.8%)
65+	9 (15%)	0 (0%)
Sex			0.40
F	28 (47%)	32 (56%)
M	32 (53%)	25 (44%)
Any high-risk condition3	8 (16%)	18 (32%)	0.10
(Missing)	10	0
Any influenza infection 2015–204	-	19 (59%)
(Missing)	60	25
Any influenza vaccination 2015–20	4 (6.7%)	45 (82%)	<0.001
(Missing)	0	2
Total children in the households	0.4 (0.0, 3.0)	3.0 (1.0, 6.0)	<0.001
(Missing)	4	0
Total adults in the households	2.5 (1.0, 6.0)	2.3 (1.0, 6.0)	<0.001
(Missing)	4	0

¹Mean (Range); n (%)

²Welch Two Sample t-test; Pearson’s Chi-squared test

³High risk conditions: any of the following conditions: EPI-HK: Asthma or any other chronic lung problems; Heart problems; Hypertension; Strock; Diabetes mellitus; Hypercholesterolemia; Kidney disease, such as chronic kidney disease (excluding kidney stones, bladder infection or incontinence); Liver disease, such as chronic hepatitis; Immune system-related diseases; Neurologic disorders, such as epilepsy or Parkinson’s disease; Cancer; Dermatological disease, such as eczema; Cataract; Osteoarthritis; Self-reported other comorbidities. HIVE: High-risk respiratory disease, High-risk cardiac illness, Diabetes mellitus, Chronic renal disease, Malignancy or cancer, Immunosuppressive disorders, Liver disease, Neurological or neuromuscular disorder, Hemoglobinopathies diagnosis, Diseases of arteries, arterioles, and capillaries, Cerebrovascular disease, Metabolic disease, Endocrine disorder, Diseases of veins, lymphatic vessels and/or lymph nodes, and Chronic medication use, Self-reported other comorbidities

⁴EPI-HK did not collect data on infection history prior to 2020 for all participants

Regarding historical influenza vaccination from 2015 to 2019, 74% of HIVE participants and 7% of EPI-HK participants had received the vaccine. 32% of HIVE participants had lab-confirmed symptomatic influenza infections during this period (table 2). Infection history of EPI-HK participants was unknown. During the 2022–2023 influenza season, 2 (0.2%) HIVE participants were infected and developed symptoms, whereas no infections were reported in the EPI-HK cohort. In terms of household composition, HIVE participants had a larger household than EPI-HK participants, the mean numbers of children and adults in HIVE households were 3 (range: 1–6) and 2 (range: 1–6), respectively. For the EPI-HK cohort, the averages were 0 (range: 0–3) for children and 3 (range 1–6) for adults.

3.2. Hemagglutination inhibition (HAI) specific antibody responses

H1N1

At baseline (2020–21 season), GMTs against H1N1 strains in HIVE ranged 77.1–106.4 (lowest- highest strains; A/Brisbane–A/Idaho;95%CI: 52.9–112.4;67.5–167.5); In EPI-HK, baseline GMTs ranged 7.2–15.3 (A/Hawaii–A/Michigan;95%CI: 5.9–8.7;10.8–21.7)(Figure 2, Figure S1, Table S2). The fold change in GMT for H1N1 strains ranged from 1.55– 1.69 (A/Michigan -A/Brisbane; 95% CI :0.53–4.53, 0.48–5.98) to in HIVE and 0.88–0.94 (A/Brisbane–A/Hawaii; 95% CI:0.57–1.36, 0.73–1.23) in EPI-HK. The proportion of participants with ≥2-fold increases ranged from 21.4% (A/Brisbane) to 28.6% (A/Idaho) in HIVE and 6.7% (A/Brisbane & A/Hawaii) to 16.7% (A/Victoria) in EPI-HK (Table S3).

Figure 2:

GMT over time by region

H3N2

Baseline GMTs for H3N2 strains in HIVE were 53.8 and 123.4 (A/Singapore and A/Kansas; 95%CI: 35.97–80.59; 84.05–181.1, respectively). In EPI-HK, GMTs ranged 6.3–17.8 (A/Kansas – A/Hong Kong; 95%CI: 5.4–7.3; 12.3–25.9) (Figure 2, Figure S1, Table S2). Fold changes for H3N2 strains were 2.59 for A/Singapore and A/Kansas (95% CI:0.81–8.25, 1.10–6.11) in HIVE, likely reflecting H3N2 transmission in Michigan during the 2022–23 season (Figure 1). In EPI-HK, fold changes ranged 0.81–0.99 (A/Hong Kong–A/Kansas; 95% CI:0.49–1.34, 0.82–1.21). The proportion with ≥2-fold increases ranged from 14.3% (A/Kansas) and 23.2% (A/Singapore) in HIVE and 3.3% (A/Singapore) to 10% (A/Kansas) in EPI-HK (Table S3).

Influenza B

Baseline GMTs for influenza B strains in HIVE were 37.1– 56.6 (B/Colorado and B/Phuket; 95%CI: 29.5–46.8; 41.1–77.8, respectively). In EPI-HK, they were 8.0– 11.9 (B/Washington and B/Phuket; 95%CI: 6.1–10.5; 27.8–60.21, respectively) (Figure 2, Figure S1, Table S2). The fold change in GMT was lowest for B/Yamagata: 1.23 (95% CI: 0.45–3.33) in HIVE and 0.73 (95% CI: 0.42–1.26) in EPI-HK. The proportion of participants with ≥2-fold increases were 16.1% (B/Colorado) and 21.4% (B/Phuket) in HIVE and 15% (B/Washington) to 23.3% (B/Phuket) in EPI-HK (Table S3).

When stratified by age groups (<18 and ≥18 years), no significant differences in serum GMTs or fold changes were observed for any antigens in either cohort. In HIVE, the lack of difference may be due to the limited number of participants in the younger age group. In EPI-HK, GMTs and fold changes were similarly consistent across age categories (Figure 2).

3.3. Statistical model

The univariable and multivariable linear mixed-effects models were used to evaluate factors associated with antibody titers in the HIVE and EPI-HK cohorts, evaluate the differences in temporal and age-related patterns between the two populations.

H1N1

In the HIVE cohort, no significant monthly change in HAI titers was observed for H1N1 strains, with an estimated fold change of 0.32% – 0.78% per month (A/Michigan – A/Brisbane; 95% CI:−0.78%– 1.44%; −0.35% –1.91%). However, age was significantly associated with lower titers: a 2.39% decrease per year (95% CI: −4.66% – −0.06%) for A/Michigan, but not significant for A/Idaho and A/Brisbane (figure 3, tableS4). In the EPI-HK cohort, A/Michigan titer showed a small decline but significant over time, with an estimated fold change of −0.56% per month (95% CI: −1.07% to −0.04%). Age was significantly associated with lower titers for H1N1 strains except A/Victoria, with an estimated decrease of 0.99–2.21% per year (A/Hawaii–A/Brisbane; 95% CI: −1.78%– −0.19%, −3.57%– −0.84%) (figure 3, table S5). These associations remained consistent in the multivariable model for both cohorts (figure 3, table S4; table S5).

Figure 3:

Multivariable Mixed-Effects Model of HAI Antibody Titers

H3N2

In the HIVE cohort, no significant fold changes in HAI titers per month were estimated for H3N2 strains. The largest (but non-significant) decline was observed for A/Singapore (95% CI: −0.01% –2.29%). Age was significantly associated with lower titers for A/Singapore, with a 3.40% decrease per year (95% CI: −5.59% – −1.16%). Multivariable models confirmed this age effect, while no time-related declines reached statistical significance (figure 3, tableS4). In the EPI-HK cohort, a significant decline in HAI titers over time was observed for A/Hong Kong, with a fold change of −0.97% per month (95% CI: −1.82% – −0.11%). Age was associated with lower titers for all H3N2 strains, with decreases ranging from 1.07% to 3.10% per year (A/Kansas – A/Hong Kong: 95% CI: −1.61%– −0.53%, −4.52% – −1.66%) (figure 3, table S5). In the multivariable model, both age and time remained significant predictors for A/Hong Kong (figure 3, table S4; table S5).

Influenza B

In the HIVE cohort, no significant associations were found between time and antibody titers for influenza B strains, including B/Phuket and B/Colorado. Age was significantly associated with higher titers for B/Colorado, with a 1.64% increase per year (95% CI: 0.17% – 3.14%) (figure 3, tableS4). In the EPI-HK cohort, age-associated declines were observed for B/Phuket, with a 1.92% per year (95% CI: −3.52%– −0.28%), whereas no significant association was found for B/Washington. No time-related changes were detected for any B strains (Figure 3, Table S5). Multivariable models confirmed the age effects and absence of time effects for both strains (Figure 3, Table S4; Table S5).

Across both cohorts, temporal and age-related trends in HAI titers varied by strain type. No significant time-related declines were observed in the HIVE cohort for any strain, whereas the EPI-HK cohort showed modest but statistically significant waning for select H1N1 (A/Michigan) and H3N2 (A/Hong Kong) strains. Age was more consistently associated with lower titers in the EPI-HK cohort across H1N1 and H3N2 strains, while in HIVE, significant age effects were limited to a subset of strains. Influenza B strains showed no time-related decline in either cohort; age associations were inconsistent, with an increase in B/Colorado titers by age in HIVE and a decline for B/Phuket in EPI-HK.

3.4. Sensitivity analysis

A sensitivity analysis was conducted by including additional participants from the EPI-HK cohort. Overall, no significant differences in time- or age-associated changes in antibody titers were observed for most selected strains. An exception was noted for A/Cambodia, where a significant increase in titers over time was detected, with an estimated monthly increase of 0.66% (95% CI: 0.18% to 1.15%). No significant differences were observed in age-associated titer changes in this expanded analysis (Table S6).

Discussion

In this study, we evaluated longitudinal antibody changes in two geographically and epidemiologically distinct populations in Michigan (HIVE cohort) and Hong Kong (EPI-HK cohort). We observed minimal antibody waning over the study period during the COVID-19 pandemic when influenza transmission was supressed. In the HIVE cohort, no significant time-related decreases in titers were observed for any strain group (H1N1, H3N2, or influenza B), and in some cases, modest increases were detected—likely reflecting renewed local influenza circulation, particularly of H3N2 viruses, during the 2022/23 season. In contrast, the EPI-HK cohort exhibited small but statistically significant declines in HAI titers for A/Hong Kong and A/Michigan. Across both cohorts, age was consistently associated with lower baseline titers, although the strain-specific patterns varied: only A/Michigan (H1N1) and A/Singapore (H3N2) showed significant age-related differences at baseline in HIVE, while age effects were observed for most strains in EPI-HK, except for A/Victoria and B/Washington. While site-specific variation in HAI results could have explained the overall, absolute differences in titer between regions, we would not expect this to impact our analysis of the relative magnitude of change over time^23,24.

HIVE and EPI-HK represent two regions with both historical and contemporary differences in both influenza virus circulation, public health measures and vaccine distribution. To the degree that these patterns have resulted in different levels of population immunity prior to the pandemic, the impact of the lapse in influenza circulation would be expected to differ between the cohorts. It has been established that local factors such as humidity, population density, and behavioral policies influence seasonal influenza dynamics, contributing to regional variability in transmission intensity and timing. Different vaccine policies and levels of uptake, as well as distinct pre-pandemic circulation patterns, likely shaped the baseline immunity in these populations and influenced their responses during the pandemic^11,17,18,25 These differences continued after the re-emergence of influenza. H3N2 strains (clade: 3C.2a1b.2a.2.) circulated extensively in Michigan during the 2022–2023 season²⁶, but were largely absent in Hong Kong, where masking and other mitigation strategies remained in place for a longer duration^27–29. Among the two study locations, HIVE participants generally had higher pre-pandemic vaccine coverage and documented prior infections, while EPI-HK participants included a larger proportion of individuals with no recent influenza vaccination history. Differences in vaccine strain formulations and vaccination schedule between the U.S. and Hong Kong could also have contributed to variability in strain-specific responses³⁰.

These findings provide important insight into the immunological landscape of influenza following the COVID-19 pandemic. A major concern raised early in the pandemic was that the near-complete absence of influenza virus circulation, driven by mitigation interventions, would result in widespread population-level immunity gaps and lead to more severe epidemics upon the resurgence of virus^2,4,31. Indeed, influenza virus infection and transmission resurged in many regions after mitigation policies were relaxed^4,6,27. However, our findings do not support the hypothesis that such resurgence was driven by rapid individual-level waning of antibody titers as measured by HAI. Despite the 2–3 year lapse in influenza exposure, most participants retained stable HAI antibody levels, indicating more durable antibody immunity than anticipated. Given that influenza virus transmission slowed during the COVID-19 pandemic^1,32, and hence, relatively little antigenic drift occurred, the relatively stable level of antibodies likely resulted in a similar level of population immunity against resurging influenza strains in 2022/23 and 2023/24 compared to the pre-pandemic levels in 2019/20. If influenza viruses had continued to evolve at the pre-pandemic pace, population immunity to influenza would have been lost during the pandemic not because of loss of homologous protection which our findings indicate was stable, but because of viral antigenic drift. While much of the literature has emphasized antibody waning among vaccinated individuals after the peak levels have occurred at 3–4 weeks after vaccination ^12,14,15, and waning over 6–12 months after that peak is evident from many past studies, our findings suggest that in the absence of new viral variants, antibody titers can remain stable in the longer term.

While we detected no substantial waning in functional anti-HA antibodies levels based on HAI assays, we cannot exclude the possibility that other immune mechanisms, such as reductions in anti-neuraminidase antibodies, mucosal immunity, or T cell memory, which could contribute to the observed epidemiological patterns in both regions. Additionally, a proportion of participants experienced rising antibody titers over the study period, suggesting undocumented natural exposures. This is likely due to more relaxed and shorter-duration mitigation measures in Michigan compared to Hong Kong, especially during the Omicron period. In HIVE, 14.3%–28.6% of participants showed increases in titer (A/Kansas and A/Idaho, respectively), compared to 3.3%–23.3% in EPI-HK (A/Singapore and B/Phuket, respectively), likely reflecting additional infections (asymptomatic or too mild to meet the case definition) not captured in active surveillance (Table S4).

The strengths of this study include the use of prospective longitudinal cohorts with detailed individual-level data on vaccination and timing of serum collection, and the ability to compare findings across distinct geographical regions. However, several limitations must be acknowledged. We were unable to comprehensively capture infection history in the EPI-HK cohort, limiting our ability to fully account for prior natural exposures, and our HAI-based analysis may not fully represent the breadth of immune changes occurring during this period that could have impacted susceptibility.

In conclusion, we found no evidence of substantial waning of influenza antibody titers at the individual level during the pandemic period in either the U.S. or Hong Kong. These findings suggest that individual HAI antibodies remained at steady state despite prolonged interruption in virus circulation. These results do not correspond to recent vaccine effectiveness studies that would suggest significant antibody decline within a year following infection or vaccination. Further work is warranted to understand the impact of baseline antibody levels on these patterns and how exposures from natural infection versus immunization differ in subsequent waning. Regional variation in strain circulation, public health responses, and vaccine uptake has provided a unique and valuable opportunity to understand the impact of these combined factors on population susceptibility to re-emerging influenza strains.

Data sharing:

The data used in this study can be made available upon request. Due to Institutional Review Board (IRB) regulations, data access is controlled. Per the guidelines of the Centers for Excellence in Influenza Research and Response (CEIRR) Network, individuals seeking access must complete a data and specimen collaboration form. Requests received will be reviewed by the study investigators.