Influenza vaccine effectiveness against medically-attended influenza infection in 2023/24 season in Hangzhou, China

ABSTRACT

From 2020, influenza viruses circulation was largely affected by the global coronavirus disease (COVID-19) pandemic, notably leading to the extinction of the B/Yamagata lineage and raising questions about the relevance of the quadrivalent influenza vaccine, which includes this lineage. Evaluating vaccine effectiveness (VE) against influenza infections is important to inform future vaccine programs. A test-negative case–control study was conducted in five tertiary hospitals in Hangzhou, the capital city of Zhejiang province, China, enrolling medically-attended patients aged >6 months who presented with influenza-like illness (ILI) from October 1, 2023, to March 31, 2024. The VE was estimated using multivariate logistic regression models adjusted for sex, age, influenza detection methods, and influenza testing timing. Of the 157,291 medically-attended ILI participants enrolled 56,704 (36%) tested positive for influenza. Adjusted overall VE against any medically-attended influenza infection was 48% (95% Confidence interval [CI]: 46%–51%). The overall VE of the trivalent inactivated influenza vaccine (IIV3) was 59% (95% CI: 50%–66%), followed by the trivalent live attenuated vaccine (LAIV3) (VE = 53%, 95% CI: 42%–62%) and quadrivalent inactivated influenza vaccine (IIV4) (VE = 47%, 95% CI: 45%–50%). IIV3 provided even better protection against medically-attended influenza B infection than IIV4 (VE = 87%, 95% CI: 81%–92% for IIV3 versus VE = 53%, 95% CI: 50%–57% for IIV4). In the 2023/24 season in Hangzhou, China, the influenza vaccine offered moderate protection during a major epidemic. The results supported the World Health Organization recommendation to exclude the B/Yamagata lineage antigen in quadrivalent influenza vaccines in 2023.

Introduction

Influenza vaccination is the most effective way to prevent influenza infection;¹ however, the influenza vaccine effectiveness (VE) can vary considerably from year to year.² From 2020, the COVID-19 pandemic greatly disturbed influenza viruses circulation,^3,4 from three potential aspects: changes in health behaviors due to non-pharmaceutical interventions,³, virus–virus interactions that may alter circulation patterns,⁵ and shifts in population immunity.⁴ In the 2023/24 season, a sharp increase in influenza viruses circulation was observed in China.⁶ Historically, China, has had a relatively low influenza vaccination rate, ranging from 1.5% to 2.2% before the COVID-19 pandemic,⁷ which increased slightly after the pandemic to 3.16% in the 2020/21 season and 2.47% in the 2021/22 season.⁸ Since 2020, Zhejiang Province has provided free trivalent influenza vaccines (IIV3) to older adults aged >70 years.⁹ This program provided an opportunity to evaluate the effectiveness of influenza vaccination policies in the real world. Recently, three types of influenza vaccines have been used in China: trivalent inactivated influenza vaccine (IIV3), quadrivalent inactivated influenza vaccine (IIV4), and trivalent live attenuated vaccine (LAIV3).¹⁰ For the trivalent influenza vaccine, in the 2023–2024 season, B/Victoria was used,¹¹ and LAIV3 was used only in children aged 3–17 years.

A test-negative case–control study design has been widely used to evaluate influenza VE.¹² However, small sample sizes and large ranges of the 95% confidence intervals (CIs) of the estimated VE have limited the robustness of the results, especially in China. Isolation of the virus and testing using real-time reverse-transcription polymerase chain reaction (RT-PCR) is adopted as the gold standard reference method for the diagnosis of the influenza.¹³ Although their sensitivity is relatively low, commercially available rapid antigen detection tests have the advantage of providing results more quickly¹³ and have been more widely used in hospitals in China. Thus, the large number of influenza-like illness (ILI) cases identified by antigen detection provided the possibility of conducting a robust VE evaluation and a separate analysis of VE in subgroups in more detail.

In addition, the B/Yamagata lineage became extinct in 2020, raising concerns regarding the quadrivalent influenza vaccine use.¹⁴ In September 2023, the World Health Organization (WHO) proposed that the inclusion of a B/Yamagata lineage antigen in quadrivalent influenza vaccines is no longer warranted, and every effort should be made to exclude this component as soon as possible.¹⁵ Thus, evaluation of effectiveness of trivalent and quadrivalent influenza vaccines, especially the quadrivalent influenza vaccine, in the years without the B/Yamagata lineage has become significant.

In this study, we performed a test-negative case–control study to estimate the effectiveness of three different types of influenza vaccines against influenza infection in the 2023/24 season in Hangzhou city, China, during a period of massive influenza epidemic. We aimed to compare VE between three types of influenza vaccines (IIV3, IIV4, and LAIV3), and different age groups. Globally, several studies have evaluated early or interim influenza VE after the COVID-19 pandemic.^16–19 As the timing of influenza vaccination plays a role in estimating its effectiveness,²⁰ the influenza VE during the entire flu season may differ from the early or interim VE.

Methods

Study population

Samples were collected from five tertiary hospitals: First People’s Hospital of Hangzhou, Second People’s Hospital of Hangzhou, First People’s Hospital of Linping District, Zhejiang Xiaoshan Hospital, and First People’s Hospital of Xiaoshan District. Data were collected from October 1, 2023, to March 31, 2024, based on the seasonality of influenza epidemics in Zhejiang Province.²¹ Regarding influenza detection, physicians were required to specify on the laboratory requisition form that the patient exhibited symptoms consistent with ILI. However, there is currently no standardized case definition of ILI.¹⁶ Each month, individuals seeking medical attention for respiratory diseases with fever, accompanied by symptoms of cough and sore throat and undergoing influenza testing, were enrolled in the study across the five healthcare institutions.

Eligible participants were patients with ILI who visited these five sentinel hospitals and were older than 6 months. Patients with an influenza testing interval exceeding 14 days were eligible for repeated inclusion in the study.² Basic demographic information, including sex and age, was collected for each patient with influenza testing, along with the detailed testing date, type of test method, and corresponding test results. Influenza testing methods included influenza nucleic acid detection by RT-PCR and influenza antigen detection. These two detection methods were analyzed individually to test the effect of the detection methods on the evaluation of VE.

Vaccination status

The virus strains recommended by the WHO in the 2024 Northern Hemisphere influenza season were A/Victoria/4897/2022 (H1N1) pdm09-like virus, A/Darwin/9/2021 (H3N2)-like virus, and B/Austria/1359417/2021 (B/Victoria lineage)-like virus.¹¹ Patient vaccination information was obtained from the Zhejiang Province Electronic Medical System using the patient’s identity card (ID). This system comprehensively covers vaccination records for the majority of Hangzhou residents, as it was constructed to record COVID-19 vaccination status. Patients without vaccination records were excluded. Patient vaccination information from the Zhejiang Province Electronic Medical System included patient ID, vaccine type (IIV3, IIV4, or LAIV3), and vaccine date. This study specifically focused on the details of vaccine administration for 2022 and 2023. In accordance with Zhejiang Province’s vaccination policy, vaccination in the current season was defined as influenza vaccination from July 1, 2023, to March 31, 2024. Vaccination in the previous season was defined as influenza vaccination from July 1, 2022, to June 30, 2023. To explore the impact of previous vaccines on VE, the vaccination status of participants was categorized into four types: never vaccinated, vaccinated in both seasons, vaccinated only in the previous season, and vaccinated only in the current season. Individuals were considered immunized if they had received an influenza vaccine for at least 14 days before the specimen collection date.² Thus, influenza vaccines administered within 14 days before ILI onset were excluded from the analysis. The flu vaccine produced by manufacturers for children under 14 years of age requires two doses to be considered a complete vaccination.

Statistical analysis

We pooled 6 months of data and compared the characteristics of cases confirmed through influenza nucleic acid testing and influenza antigen testing with the characteristics of control cases that tested negative for influenza, stratified by virus type, vaccine type, and age group of patients. VE was calculated using a test-negative case–control design as (1 – OR) × 100 %, where OR is the odds ratio. Unadjusted ORs were calculated using the formula:

$$\mathrm{Unadjusted}\mathrm{OR}=\frac{\mathrm{ad}}{\mathrm{bc}}$$

where a and b were the number of vaccinated individuals who tested positive and negative for influenza, respectively; c and d were the number of unvaccinated individuals who tested positive and negative for influenza, respectively.

Adjusted ORs were determined using multivariate logistic regression models with adjustments for sex, age, influenza detection methods, and month of influenza testing. Stratified VE estimates were computed based on the age group and influenza virus type. All statistical tests were conducted using a two-sided test, and significance was defined as p < .05 or when the lower bound of the 95% CI for VE was greater than 0. Statistical analyses were performed using R statistical programming software version 4.2.1 (The R Project for Statistical Computing, Vienna, Austria).

Sensitivity analysis

We conducted sensitivity analysis to assess the robustness of the estimated influenza VE. Firstly, we estimated the influenza VEs based on two influenza detection methods individually to test the impact of influenza detection methods. Second, to test the impact of influenza testing timing, we did not adjust the influenza testing timing or adjust the influenza testing date to estimate the VE individually.

Results

Participant characteristics

Data were collected from October 1, 2023, to March 31, 2024, involving a total of 157,291 ILI patients tested for influenza (Table 1). Of whom, 151,251 were tested by antigen and 6,040 were tested by nucleic acid (Table S1). The number of ILI patients tested for influenza peaked at week 53, 2023 (Figure 1). The dominant influenza subtype in the 2 years varied. Influenza A dominated in 2023, whereas influenza B dominated in 2024 (Figure 1). The overall influenza positivity rate was 36% in the ILI patients during the study period (Table 1), and the influenza positivity rates peaked at week 52 in 2023 at 49.9% (5,798/11,629) (Figure 1). By the end of this study, March 31, 2024, influenza viruses circulation also almost ended in Hangzhou (Figure 1). The weekly influenza positivity rates of the two detection methods were highly related (Pearson r = 0.94, p < .001) (Figure S1). In the ILI patients, female was a little more than male (53.6% versus 46.4%). Additionally, people with the age between 18 and 59 years accounted for the highest proportion, comprising 50.4%. The overall influenza vaccination rate was 7.1% (Table 1), with a rate of 4.5% for positive cases and 8.6% for negative cases. The influenza vaccination rate was higher in nucleic acid-tested patients (13.7%) compared to antigen-tested ILI patients (6.8%) (Table S1). This is associated with a higher percentage of individuals aged 70+ in the nucleic acid-tested group (19.8%) compared to the antigen-tested group (1.14%) and a higher vaccination rate among the elderly (44.8% vs. 21.7%) (Table S2). For these people who were vaccinated in the 2023/24 season, the majority opted for IIV4, accounting for 86.3%, followed by IIV3 (8.8%) and LAIV3 (4.9%). Furthermore, for these who vaccinated IIV3, most of them were older people over 70 years of age, since that the IIV3 influenza vaccines were free for them (Table S3). For people who were vaccinated in 2023–2024, the period between the influenza test date and the influenza vaccination date ranged from 15 to 249 days with a median of 97 days (Figure S2).

Table 1.

Demographic and clinical characteristics of participants who tested positive or negative for influenza virus.

Characteristics	Test-positive participants			Test-negative participants	Total
	Influenza A	Influenza B	A/B coinfection
Total, n (%)	32611	24030	63	100587	157291
Sex, n (%)
Male	15386 (47.2%)	11624 (48.4%)	29 (46.0%)	45930 (45.7%)	72969 (46.4%)
Female	17225 (52.8%)	12406 (51.6%)	34 (54.0%)	54657 (54.3%)	84322 (53.6%)
Age group, n (%)
0.5–2 years	1016 (3.1%)	679 (2.8%)	1 (1.6%)	4963 (4.9%)	6659 (4.2%)
3–9 years	7138 (21.9%)	6190 (25.8%)	23 (36.5%)	22057 (21.9%)	35408 (22.5%)
10–17 years	5718 (17.5%)	4459 (18.6%)	9 (14.3%)	18532 (18.4%)	28718 (18.3%)
18–59 years	17681 (54.2%)	12228 (50.9%)	25 (39.7%)	49336 (49.0%)	79270 (50.4%)
60–69 years	680 (2.1%)	370 (1.5%)	1 (1.6%)	3352 (3.3%)	4403 (2.8%)
70+ years	378 (1.2%)	104 (0.4%)	4 (6.3%)	2347 (2.3%)	2833 (1.8%)
Vaccinated in 2023/34 season, n (%)
No	31138 (95.5%)	22965 (95.6%)	56 (88.9%)	91984 (91.4%)	146143 (92.9%)
Yes	1473 (4.5%)	1065 (4.4%)	7 (11.1%)	8603 (8.6%)	11148 (7.1%)
Types of Vaccines, n (%)
IIV3	96 (6.5%)	19 (1.8%)	1 (14.3%)	866 (10.1%)	982 (8.8%)
IIV4	1278 (86.8%)	1020 (95.8%)	5 (71.4%)	7320 (85.1%)	9623 (86.3%)
LAIV3	99 (6.7%)	26 (2.4%)	1 (14.3%)	417 (4.8%)	543 (4.9%)

Figure 1.

Weekly number of medically-attended influenza infections and influenza positivity rates in five tertiary hospitals in Hangzhou.

Overall VE

After adjustment for potential confounders, the estimated overall influenza VE against medically-attended influenza infection in the 2023/24 season was 48% (95% confidence interval (CI): 46%–51%) (Figure 2). VE estimates were consistent between subjects using two influenza virus testing methods: 48% (95% CI: 46%–51%) for antigen-tested subjects, and 55% (95% CI: 44%–63%) for nucleic acid-tested subjects (Figure S3). However, because of the larger number of patients detected by antigen testing, the estimated 95% CI of VE was narrower. In addition, without adjusting the influenza testing timing or adjusted by the influenza testing date also had a very limited impact on estimating the overall VEs (Figure S4). The overall VE of IIV3 was highest (VE = 59%, 95% CI: 50%–66%), followed by LAIV3 (VE = 53%, 95% CI: 42%–62%) and IIV4 (VE = 47%, 95% CI: 45%–50%) (Figure 2). Adjusted VE was 65% for children aged 6 months to 2 years, 52% for people aged 18–59 years, 78% for old adults aged 60–69 years, and 28% for people aged above 70 (Figure 2). The VE against medically-attended influenza B infection (VE = 57%, 95% CI: 54%–60%) was higher than that against influenza A infection (VE = 38%, 95% CI: 34%–42%). (Figure 2).

Figure 2.

Adjusted estimates of influenza vaccine effectiveness (VE), overall and stratified to age groups, virus subtypes and vaccine types. Horizontal bars indicated 95% confidence intervals (CI).

VE by influenza subtype, vaccine type, and age group

Overall, the influenza vaccine offered better protection against medically-attended influenza B than influenza A in the 2023/24 season in Hangzhou. IIV3 was more effective than IIV4 against influenza B, with an adjusted VE of 87% (95% CI: 81%–92%) for IIV3 compared to 53% (95% CI: 50%–57%) for IIV4 (Figure 3). LAIV3 showed a VE of 79% (95% CI: 69%–86%) for children aged 3–17 years (Figure 3). Adjusted by age and testing timing, IIV4 provided best protection against medically-attended influenza A infection, followed by LAIV3 and IIV3. LAIV3 was only administered to children aged 3–17 in Hangzhou. In the age group, IIV4 provided best protection against influenza A, followed by IIV3 and IIV4. While IIV3 provided the best protection against influenza B, followed by LAIV3 and IIV4 (Figure S5).

Figure 3.

Adjusted VE against medically-attended influenza A and B infection, stratified to vaccine types.

VE for different vaccination seasons

VE against medically-attended influenza infection was the highest for those vaccinated only in the current 2023/24 season (VE = 52%, 95% CI = 49%–55%) (Figure 4). Vaccination in the previous 2022/23 season could only provide low protection against medically-attended influenza infection (VE = 23%, 95% CI = 18%–27%). However, vaccination in the previous 2022/23 season could still provide moderate protection against medically-attended influenza B infection (VE = 34%, 95% CI: 28%–40%). Moreover, vaccination in both seasons provided the best protection against influenza B infection (VE = 59%, 95% CI: 55%–63%). However, for influenza A, vaccination in the current season only provided the best protection (VE = 47%, 95% CI: 52%–60%) (Figure 4).

Figure 4.

Adjusted VE against medically-attended influenza A and B infection, stratified according to receipt of vaccines for the current season (2023/24) and previous season (2022/23).

Discussion

Several studies have evaluated the effectiveness of influenza vaccines after the COVID-19 pandemic. In Denmark, VE was low at 24.8% in non-hospitalized patients aged 7–44 years in the 2021/22 season when there was a sharp increase in influenza detections.¹⁷ In China, during the 2022/23 season, the VE was 56.3% for medically-attended influenza cases.¹⁸ Several studies have estimated the effectiveness of early or interim influenza vaccines in the 2023–2024 season. For the 2023/24 season, early estimates include 61% VE against influenza A H1N1, 49% against influenza A H3N2, and 75% against influenza B in Canada;¹⁶ 51.9% VE against influenza hospitalization in the Southern Hemisphere;¹⁶ and 45% VE in California for laboratory-confirmed influenza.²² However, to the best of our knowledge, no study has estimated the VE for the entire 2023/24 season. As the timing of influenza vaccination plays a role in its effectiveness,¹³ the VE in the entire flu season may differ from the early or interim VE and would be more accurate in estimating the cost-effectiveness of the free influenza vaccination program. Our study, estimating a VE of 48% against influenza infection in Hangzhou, is comparable to the interim VEs reported in the Southern Hemisphere and the US in the 2023/24 season.^19,22 This VE was lower than Canada’s early estimate,¹⁶ likely due to differences in study timing. Similar to Canada, we observed better protection against influenza B than A in the 2023/24 season.

Evaluating influenza VE in the post-COVID-19 era is crucial due to altered transmission patterns and population immunity levels. The resurgence of influenza in the 2023/24 season underscores the need for updated VE estimates. Our study addresses this by offering season-long VE data that captures post-pandemic epidemiological shifts, particularly relevant in regions like Hangzhou with low vaccine uptake and evolving influenza dynamics. Additionally, the large sample size also strengthens the reliability of our findings.

Before the COVID-19 pandemic, several studies also had evaluated the influenza VE in China in all age group in the year from 2013 to 2017.^23–25 The estimated VE ranged from 5% (95% CI: −53%–41%) to 47% (95% CI: −20%–77%) in these studies. The observed variations could be due to the relatively small number of ILI patients enrolled in the studies, since the vaccine-effectiveness point estimates always have very wide confidence intervals in these studies. In addition, the influenza positivity rates in the ILI cases in these three studies ranged from 20% to 27%,^23,25 lower than the 36% influenza positivity rate in this study. These differences may reflect the broader epidemiological shift following COVID-19, where decreased influenza circulation and subsequent resurgence may have altered both influenza activity and VE. Additionally, influenza VE pre-pandemic in China was estimated at an overall 36% according to a meta-analysis,²⁶ whereas recent findings, including ours, suggest potentially higher effectiveness during larger influenza outbreaks post-pandemic.

In Hangzhou during the 2023/24 season, influenza vaccine coverage among ILI patients was 7.1%, compared to 2.47% nationally in the 2021/22 season.⁸ This is significantly lower than the 49.3% coverage in the United States for individuals over 6 months of age.²⁷ This discrepancy may reflect cultural differences, as seasonal influenza vaccination is often viewed as less essential in China compared to other countries with stronger vaccination advocacy and education. Additionally, structural limitations, such as fewer public health campaigns, restricted access to vaccine providers, and a preference for treatment over prevention, may further reduce vaccine uptake. The increased influenza vaccine coverage rate mainly contributed by the increased influenza vaccine coverage rate in the old people aged more than 70 years, which was 28%, higher than those in other age groups. However, in Hangzhou in the 2023/24 season, we found that the majority opted IIV4, accounting for 86.3%, only 8.8% of ILI cases vaccinated IIV3. For those who vaccinated IIV3, most of them were older people with age more 70 years because of the free influenza IIV3 program to them (Table S3).

In this study, we found that IIV3 provided better protection than IIV4 against influenza infection, especially against influenza B infection in the 2023/24 season (87% versus 52%). The main reason could be that B/Victoria used influenza vaccine components in trivalent influenza vaccine, since B/Yamagata lineage has become extinct.¹⁴ Our results supported the report from WHO about the exclusion of B/Yamagata lineage antigen in quadrivalent influenza vaccines as soon as possible.¹⁵ Given the lower cost of IIV3, shifting from IIV4 to IIV3 may improve cost-effectiveness, as well as align with recommendations to streamline vaccine composition. This adjustment would reduce production complexity and potentially lower overall program costs. Further cost-effectiveness analyses would support the transition, as shown by the U.S. CDC’s recent decision to adopt a trivalent approach for the 2024–25 season.²⁸ However, public acceptance may be a challenge due to the existing perceptions of IIV4 as more comprehensive. To address this, public health campaigns could emphasize the targeted effectiveness of IIV3, especially against dominant strains.

An observed difference in VE between individuals vaccinated only in the current season and those vaccinated in both the current and prior seasons warrants further examination. One potential explanation is the cumulative effect of repeated vaccinations. Research has suggested that annual influenza vaccinations may not provide additive immunity benefits and, in certain cases, could result in attenuated effectiveness. This phenomenon may arise from alterations in immune memory and the body's adaptive responses over successive vaccinations.²⁹ Additionally, antigenic drift between circulating strains and those in prior season vaccines may lead to partial mismatch, limiting the ability of the immune system to effectively target the strains of the current season.³⁰ Such antigenic changes could thereby reduce the efficacy of prior-season vaccination against newly circulating influenza viruses.

The findings of this study have the following limitations. First, diagnostic limitations arise from a preference for antigen tests over PCR tests due to accessibility and faster results, not individual choice. This results in a smaller sample for PCR testing.

Second, genetic data on circulating A(H1N1)pdm09 and A(H3N2) strains specific to Hangzhou were not available, limiting our ability to assess VE concerning potential strain mismatch. Because the hospitals included in this study were not part of the national influenza sentinel surveillance network, thus restricting access to strain-specific genetic information. Furthermore, we lacked detailed patient information, including chronic disease status, which could influence VE due to conditions associated with severe influenza (e.g., cardiovascular or chronic respiratory diseases).²

Misclassification of cases constitutes another significant limitation in this study, primarily due to the use of specimen collection dates as proxies for symptom onset, which may lead to misclassification and affect VE estimates. Two potential sources of misclassification are: (1) patients whose symptom onset occurred well before specimen collection, resulting in a negative test despite a positive infection at onset; and (2) patients who were vaccinated shortly before symptom onset but had a prolonged interval before specimen collection, possibly leading to the misclassification of vaccine-related illness as influenza infection. Although these misclassification risks exist, we believe that the large sample size and use of multivariate adjustments support the robustness of the main findings. Additionally, inherent selection bias in the test-negative design, which focuses on medically-attended cases and lacks standardized ILI definitions, may limit the generalizability of the results.¹²

Lastly, the absence of data on prior season influenza infections limits our understanding of how past infections might influence VE, an aspect that future studies could explore to assess the cumulative effects of repeated vaccinations.³¹ This may result in an overestimation of VE.

Conclusion

In the 2023/24 season in Hangzhou, China, the influenza vaccine offered moderate protection during a major epidemic. The large number of patients with an ILI detected by antigen testing made it possible to conduct a detailed separate analysis of influenza VE, such as by vaccine type, age group of patients, and influenza subtype. IIV3 provided better protection against medically-attended influenza infection, especially influenza B. Thus, our study results support the WHO recommendation regarding the exclusion of B/Yamagata lineage antigens in quadrivalent influenza vaccines in September 2023.

Biographies

Zhe Wang, currently serving as the Deputy Director of the Public Health Emergency Office at Hangzhou Center for Disease Control and Prevention (Hangzhou Health Supervision Institute), is a member of the National Epidemic Investigation Expert Team under the State Council’s Joint Prevention and Control Mechanism, a graduate of the China Field Epidemiology Training Program (CFETP), and a committee member of the Respiratory Disease Branch of Hangzhou Medical Association. His main research direction is the prevention and control of respiratory infectious diseases and the field epidemiology of infectious diseases. He has published more than 10 papers as the first or corresponding author, participated in 6 projects at the municipal level or above, and has a rich theoretical foundation and practical experience in field epidemiological investigations.

Zhao Gang, MD, Chief Physician, currently serves as the Vice President of Hangzhou Red Cross Hospital, a National Second-Level Psychological Counselor. He has successively engaged in various fields such as health education, infectious disease prevention and control, medical intelligence, mental health prevention and treatment, tuberculosis prevention and control, chronic disease prevention and control, and AIDS prevention and control. He possesses extensive experience in the management of disease prevention and control operations.

Funding Statement

This study was supported by the innovation group on intelligent response to infectious diseases and public health emergencies of School of Public Health, Zhejiang University.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Author contributions

HL and BN conceived and designed the study. ZS, ZW, KZ, XC, SD, and YL collected data. HL and BN wrote the drafts of the manuscript. BN cleaned and analyzed the data. GZ, ZW, and HL supervised the study. YW, SZ, and SY commented on and revised drafts of the manuscript. HL interpreted the findings. All authors read and approved the final report.

Data availability statement

Data will be made available on request.

Ethics approval statement

This study was approved by the Institutional Review Board and Human Research Ethics Committee of the School of Medicine of Zhejiang University (No. ZGL202404–1).

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/21645515.2024.2435156