Relative effectiveness of a heterologous booster dose with adenovirus type 5 vectored COVID-19 vaccine versus three doses of inactivated COVID-19 vaccine in adults during a nationwide outbreak of omicron predominance, in China: a retrospective, individually matched cohort-control study

ABSTRACT

Effectiveness of heterologous booster regimes with ad5 vectored COVID-19 vaccine in a large, diverse population during the national-scale outbreak of SARS-CoV-2 omicron predominance in China has not been reported, yet. We conducted a large-scale cohort-control study in six provinces in China, and did a retrospective survey on the COVID-19 attack risk during this outbreak. Participant aged ≥18 years in five previous trials who were primed with 1 to 3 doses of ICV received heterologous booster with either intramuscular or orally inhaled ad5 vectored COVID-19 vaccine were included in the heterologous-trial cohort. We performed propensity score-matching at a ratio of 1:4 to match participants in the heterologous-trial cohort individually with the community individuals who received three-dose of ICV as a control (ICV-community cohort). From February 4 to April 10, 2023, 41504 (74.5%) of 55710 individuals completed the survey. The median time since the most recent vaccination to the onset of the symptoms of COVID-19 was 303.0 days (IQR 293.0-322.0). The attack rate of COVID-19 in the heterologous-trial cohort was 55.8%, while that in the ICV-community cohort was 64.6%, resulting in a relative effectiveness of 13.7% (95% CI 11.9 to 15.3). In addition, a higher relative effectiveness against COVID-19 associated outpatient visits, and admission to hospital was demonstrated, which was 25.1% (95% CI 18.9 to 30.9), and 48.9% (95% CI 27.0 to 64.2), respectively. The heterologous booster with ad5 vectored COVID-19 vaccine still offered some additional protection in preventing COVID-19 breakthrough infection versus homologous three-dose regimen with ICV, 10 months after vaccination.

KEYWORDS:

• Effectiveness
• COVID-19 vaccine
• heterologous immunization
• homologous immunization
• hazard ratio

Introduction

Strategic Advisory Group of Experts on Immunization (SAGE) of the World Health Organization (WHO) recommended that homologous or heterologous COVID-19 booster immunization should be performed 4–6 months after the prime immunization, particularly for those primed with inactivated COVID-19 vaccine [1]. In China, over 90% coverage of primary immunization of COVID-19 vaccines in the whole population has been achieved, with the majority receiving inactivated COVID-19 vaccines [2]. Thus, an effective boosting strategy with heterologous or homologous COVID-19 booster immunization following the priming of inactivated vaccines is important for the establishment of COVID-19 herd protection in China. Previous studies with different COVID-19 vaccines suggested that heterologous prime-boost COVID-19 immunization could induce significantly superior and broader immune responses compared with that of homologous booster immunization dose [3,4]. However, evidence on the superior protection of the heterologous prime-boost regimen over the homologous booster immunization against SARS-CoV-2 breakthrough infection was limited, particularly regarding the protection against omicron-associated outpatient visits, and admission to hospital in real-world situation.

To protect the population from the pandemic of SASRS-CoV-2, China adopted the “dynamic zero-case policy” for almost three years and only sporadic or small-scale regional outbreaks of SARS-CoV-2 were observed till the end of the year 2022, leaving almost 1.4 billion Chinese people had no natural exposure to any SARS-CoV-2 variants such as alpha, beta, gamma, and delta before. Thus, after removing the control policy, a nationwide outbreak of omicron predominance was triggered, causing a nationwide COVID-19 wave affecting the majority of the population in China from the end of 2022 to the early period of 2023 [5,6].

In 2021 and 2022, we conducted a series of trials of homologous prime-boost regimens with intramuscular or orally inhaled adenovirus type 5 vectored COVID-19 vaccine in adults in China, who were primed with different doses of inactivated COVID-19 vaccines before. Participants receiving heterologous booster immunization with adenovirus type 5 vectored COVID-19 vaccine after the primary immunization of inactivated COVID-19 vaccines demonstrated significantly higher immune response than the homologous immunization of the inactivated COVID-19 vaccine [7-12]. However, quick antibody waning after the vaccination was noted in both heterologous and homologous groups at a similar pace, resulting in an uncertainty of enhanced protection associated with heterologous prime-boost regimens, particularly for a long time. Furthermore, more comparative research studies are needed between different booster regimens over extended periods to evaluate mucosal vaccines’ immunogenicity and efficacy [13].

Here, we reported the relative effectiveness of heterologous booster regimes with adenovirus type 5 vectored COVID-19 vaccine following primary immunization of inactivated COVID-19 vaccines in a large, diverse population of people aged 18 years or older, during the nationwide outbreak of omicron predominance at the end of 2022 in China.

Methods

Study design and participants

This is a retrospective, cohort-control study conducted in Jiangsu, Chongqing, Shandong, Hunan, Anhui, and Yunnan provinces in China (Figure S1). We invited all the healthy participants aged 18 years or above from five previous heterologous prime-boost immunization clinical trials with COVID-19 vaccines (ClinicalTrials.gov, NCT04892459; NCT04952727; NCT05043259; NCT05303584; and NCT05204589) to participate a survey. Detailed information on these five rime-boost immunization trials is shown in additional file 1: Table S1.

All participants in these trials had received 2 doses or 3 doses of ICV as primary immunization, and those who received a heterologous booster were included in the heterologous-trial cohort, involving five different heterologous priming-boost regimes (2 doses ICV + 1 dose Ad5-IH, 3 doses ICV + 1 dose Ad5-IH, 1 dose ICV + 1 dose Ad5-IM, 2 doses ICV + 1 dose Ad5-IM, and 3 doses ICV + 1 dose Ad5-IM). While participants, who received only ICV, were included in the ICV-trial cohort (2 doses ICV, 3 doses ICV, and 4 doses ICV). We linked the database of the five trials and the National Vaccination Integrated Service Management Information System, which includes information on the demographic characteristics of all individuals who received at least one dose of the COVID-19 vaccine in China, as well as the date and type of vaccine for each dose. Then, we individually matched each participant in the heterologous-trial cohort at a ratio of 1:4 with community individuals who have received three-dose ICV, using propensity score weighting to adjust for age (±5 years), sex (same gender), region (same county or district), and time of the last dose COVID-19 vaccination (±6 months). By doing this, we have built the ICV-community cohort from the general population matched to the heterologous-trial cohort.

This study was approved by the Ethics Review Committee of the Chinese Center for Disease Control and Prevention, and conducted by Jiangsu, Chongqing, Shandong, Hunan, Anhui, and Yunnan Provincial Center for Disease Control and Prevention in collaboration. Oral informed consent was obtained from each participant before starting the survey.

Procedures

Individuals in either the heterologous-trial cohort, ICV-trial cohort, or the ICV-community cohort were interviewed by trained and qualified investigators, using a computer-assisted telephone interview system. The system guided interviewers through the questionnaire, and the investigators entered the participants’ answers directly into the system. All information obtained was self-reported by the respondents and was checked and judged by the investigators according to the uniform criteria specified. All investigators signed a confidentiality agreement to keep the information collected confidential. The contents of the survey mainly include basic demographic information, lifestyle, medical history, and the confirmed or suspected SARS-CoV-2 infection during the outbreak, including symptoms, severity, treatment-seeking, and duration.

Outcomes

The primary outcome was the breakthrough COVID-19 disease, COVID-19-associated outpatient visits, and hospitalization during the nationwide epidemic of omicron predominance from 1 November 2022 to 31 January, 2023. COVID-19 cases in our study were defined as participants with positive antigen rapid test or nucleic acid test results (aetiologically confirmed cases), or those having acute onset of at least two typical symptoms or signs of COVID-19 with a history of exposure to probable or confirmed COVID-19 cases, without aetiological test results (clinical cases), during the outbreak period. Typical symptoms or signs of COVID-19 include fever, cough, fatigue, headache, muscle pain, sore throat, rhinitis, dyspnoea, nausea, vomiting, diarrhoea, anorexia, eye discomfort, bone pain, hyposmia, hypogeusia, chills, concentration disorders, and hair loss [14,15]. The severity of the breakthrough COVID-19 cases was categorized according to the WHO guideline, grading as mild, moderate, severe, or critical [16]. All eligible COVID-19 cases, including both aetiologically confirmed cases and clinical COVID-19-like cases were included in the primary analysis.

Statistical analysis

A total of 11,030 participants, who received heterologous boosters, were involved in the heterologous-trial cohort, which was matched at a ratio of 1:4 with individuals receiving thee-dose ICV in the community. Considering an approximate response rate of 60% in the invited population, we estimate that about 6618 and 26,472 individuals in the heterologous-trial cohort and ICV-community cohort would participate in the survey, respectively, which provided at least 98% power to identify a difference of 3% in incidence between the groups when about 50% incidence of infection in the community assumed.

We described the characteristics of the omicron-predominate outbreak of COVID-19 in China at the end of year 2022, according to the data collected from all the surveyed population, and by age, sex, and provinces. We estimated a time-varying reproduction number (R_t) with our outbreak data, based on incidence time series and serial interval distribution using the R package EpiEstim (version 2.2–4). We assumed the serial interval to be Gamma-distributed with a mean of 3.5 days, and a standard deviation of 2.4 days for Omicron, which was reported in a previous study [17]. The time window for the moving average smoothing was 7 days as default.

We compare the overall relative hazard of COVID-19 and vaccine effectiveness in heterologous-trial cohort versus ICV-community cohort or ICV-trial cohort. Stratified relative hazard and relative effectiveness in subcohorts by the type of the booster vaccine or different prime-boost regimens were also performed. Relative vaccine effectiveness was calculated as (1-hazard ratio) × 100%. Besides, hazard ratio and relative effectiveness in the heterologous-trial cohort versus ICV-community cohort were stratified by case type (aetiologically confirmed and clinical cases), age, severity, and treatment-seeking. We further assessed the impact of vaccination on COVID-19 severity by scoring the reported symptoms of COVID-19-associated case. Briefly, symptoms were scored on a scale of 0–5 according to each symptom grade, and symptom duration was also scored on a scale of 1–4 based on the median and quartile of each symptom among all the surveyed participants (Additional file 1: Table S2).

Logistic regression was used for comparing the hazard between the heterologous cohort and ICV cohort, adjusted for age, sex, BMI, lifestyle, comorbidities, tetanus and influenza vaccination, and the time since vaccination between the receipt of the most recent COVID-19 vaccine dose and the start date of the outbreak. Furthermore, bootstrapping (100,000 times) was used in the analyses to estimate the hazard ratio and relative effectiveness of COVID-19 in different heterologous subcohorts versus their matched controls from the ICV-community cohort and 95% confidence intervals.

Missing data were not inputted. For all analyses, two-sided p < 0.05 was considered statistically significant. The Chi-square test or Fisher exact test was used to compare categorical data. Statistical analyses were conducted using SAS 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

A total of 11,030 participants receiving heterologous booster immunization from five clinical trials were individually matched with 44,120 participants receiving 3 doses of ICV, and combined with the addition of 560 participants receiving ICV from clinical trials, a total of 55,710 potential participants were determined (Figure 1). From 4 February to 25 March 2023, we telephoned all potential participants to invite them to participate in the survey. Of them, 41,504 (74.5%) participants responded and completed the survey, with 9550 (86.6%) in the heterologous-trial cohort, 522 (93.2%) in the ICV-trial cohort, and 31,432 (71.2%) in the ICV-community cohort. The baseline characteristics of participants between the heterologous-trial cohort and ICV-community cohort were generally comparable (Table 1). A total of 242 participants, who had a SARS-CoV-2 infection history before the start of the omicron-predominate outbreak, were removed from the analysis. Thus, a total of 41,262 eligible participants were included in the final analysis, of which 9480 (23.0%), 31,260 (75.8%), and 522 (1.3%) participants from heterologous-trial, ICV-community and ICV-trial cohorts, respectively.

Figure 1. Study profile. ICV = inactivated COVID-19 vaccine. Ad5-IH = adenovirus type 5 vectored COVID-19 vaccine through oral inhalation. Ad5-IM = adenovirus type 5 vectored COVID-19 vaccine through intramuscular injection.

Table 1. Baseline characteristics of the participants in the surveyed population.

	ICV-trial cohort (n = 522)	Heterologous-trial cohort (n = 9480)	ICV-community cohort (n = 31,260)	p-value*	Total (n = 41,262)
Age, years
18–59	366 (70.11)	5873 (62.0)	19,477 (62.3)	0.5323	25,716 (62.3)
≥60	156 (29.89)	3607 (38.1)	11,783 (37.7)		15,546 (37.7)
Mean age, years	51.8 (15.1)	52.6 (15.6)	52.7 (15.6)	0.8752	52.6 (15.6)
Sex
Male	276 (52.9)	5228 (55.2)	17,644 (56.4)	0.0260	23,148 (56.1)
Female	246 (47.1)	4252 (44.9)	13,616 (43.6)		18,114 (43.9)
BMI, kg/m²	24.3 (3.1)	24.2 (3.7)	23.8 (3.5)	<0.0001	23.9 (3.6)
Time from last dose vaccination to November 1, 2022 (days)	353.0 (250.0, 444.0)	251.0 (248.0, 257.0)	252.0 (247.0, 258.0)	0.1097	252.0 (247.0, 258.0)

Data are n (%), mean (SD), or median (IQR). *p-value was calculated for the comparison between the heterologous-trial cohort and the ICV-community cohort. ICV = inactivated COVID-19 vaccine. BMI = body mass index.

A total of 25,880 (62.7%) breakthrough COVID-19 cases were identified from November 1, 2022 to January 31, 2023, of which 8035 (31.0%) were aetiologically confirmed cases and 17,845 (69.0%) were clinical cases (additional file 1: Table S3). Among aetiologically confirmed cases, 4010 (49.9%) participants were confirmed by the nucleic acid test, and 4025 (50.1%) participants were only confirmed by the antigen rapid test. The median time from the most recent vaccination to the onset of the symptoms of COVID-19 was 303.0 days (IQR 293.0–322.0). The overall epidemic wave was concentrated between 1 December 2022 and mid-January 2023, with a peak in mid-to-late December 2022 (Figure 2). The cumulative infection rate in adults aged between 18 and 59 years was higher than that in the elderly population aged ≥60 years (67.1% vs. 55.4%, p < 0.0001). Furthermore, females demonstrated a higher cumulative infection rate than males did (70.2% vs. 56.9%, p < 0.0001). Although the overall trends of the outbreak across the six provinces were similar, the particular patterns in different provinces varied a bit. The increase of the epidemic was first noted in Chongqing, followed by Anhui, Yunnan, Jiangsu, Shandong, and Hunan. During the epidemic outbreak period, the highest attack rate of COVID-19 in the population was reported by participants in Jiangsu (69.0%), while the lowest was in Shandong (51.7%) (Additional file 1: Figure S2). Among all COVID-19 cases, the proportion of mild, moderate and severe COVID-19 cases was 97.8%, 2.2% and 0.02%, respectively, no critical cases were reported. The COVID-19-associated outpatient visits and hospitalization accounted for 15.5% and 1.2% of the total reported cases, respectively. However, 77.6% of cases reported self-medication or on the doctor's prescription during their infection. The median duration of the symptomatic period of COVID-19 was 7.0 days (IQR 4.0–15.0) among the breakthrough cases. Despite the lower overall infection in the elderly population, participants aged 60 years and above reported higher proportions of COVID-19-associated outpatient visits (19.7% vs. 13.4%) and hospitalization (2.5% vs. 0.5%) than those aged 18–59 years did.

Figure 2. Number of COVID-19 cases and cumulative infection rate in the surveyed cohort during the omicron outbreak from 1 November 2022 to 31 January 2023. A: COVID-19 cases in all participants. B: Outpatients and hospitalized COVID-19 patients in all participants. The abscissa indicated the date of COVID-19 onset among outpatients and hospitalized COVID-19 patients. C: COVID-19 cases in participants aged 18–59 years. D: COVID-19 cases in participants aged ≥60 years. The bars indicate the number of cases, the lines indicate the cumulative infection rate.

During the outbreak, the estimated R_t of the total population increased to a maximum of 2.33 (95% CI: 2.24–2.42) on December 10, 2022, and subsequently declined, dropping below 1 since December 25, 2022 (Additional file 1: Figure S3 A). In the results of estimated R_t stratified by provinces, we observed similar trends in six different provinces but the peaking and the declining of Rt varied a bit (Additional file 1: Figure S3 C, F-G).

The attack rate of COVID-19 in the heterologous-trial cohort was 55.8%, while that in the ICV-community cohort was 64.6%, resulting in a hazard ratio of 0.86 (95% CI 0.85–0.88) and relative effectiveness of 13.7% (95% CI 11.9–15.3) (Table 2). In addition, the incidence of COVID-19-associated outpatient visits and hospitalization were much lower in the heterologous-trial cohort than those in the matched ICV-community cohort, with a hazard ratio of 0.75 (95% CI 0.69–0.81) for outpatient visit and 0.51 (95% CI 0.36–0.73) for hospitalization, demonstrating a more robust relative effectiveness of 25.1% (95% CI 18.9–30.9) in preventing outpatient visit, and 48.9% (95% CI 27.0–64.2) in preventing hospitalization. However, the hazard ratio and relative effectiveness against COVID-19 of the heterologous-trial cohort versus the ICV-community cohort were similar between participants aged 18–59 years and those aged 60 years and above.

Table 2. Relative effectiveness against COVID-19 of the heterologous-trial cohort versus the ICV-community cohort.

	Heterologous-trial cohort (n = 9480)	ICV-community cohort (n = 31,260)	p-value	Hazard ratio (95% CI)	Relative effectiveness % (95% CI)
Number of cases (%)	5293 (55.8)	20201 (64.6)	<0.0001	0.86 (0.85, 0.88)	13.7 (11.9, 15.3)
Case type
Antigen rapid test or nucleic acid test-confirmed cases	1989 (37.6)	5850 (29.0)	<0.0001	1.12 (1.07, 1.17)	−12.1 (−17.3, −7.2)
Clinical cases	3304 (62.4)	14351 (71.0)		0.78 (0.76, 0.80)	22.0 (19.8, 24.1)
Age group, years
18–59	3525 (66.6)	13447 (66.6)	0.9656	0.87 (0.85, 0.89)	13.1 (11.1, 15.0)
≥60	1768 (33.4)	6754 (33.4)		0.85 (0.82, 0.89)	14.6 (11.4, 17.7)
Severity*
Mild	5151 (97.9)	19484 (97.8)	0.9672	0.86 (0.85, 0.88)	13.6 (11.8, 15.3)
Moderate	114 (2.1)	433 (2.2)		0.86 (0.70, 1.05)	14.2 (−5.3, 30.0)
Severe	1 (0.0)	5 (0.0)		0.66 (0.08, 5.61)	34.5 (−460.9, 92.3)
Outpatient visit†	623 (12.2)	3241 (16.3)	<0.0001	0.75 (0.69, 0.81)	25.1 (18.9, 30.9)
Hospitalization‡	34 (0.7)	259 (1.3)	0.0002	0.51 (0.36, 0.73)	48.9 (27.0, 64.2)
Score based on symptom grade	4.8 (3.2)	4.9 (2.7)	<0.0001	NA	NA
Score based on symptom grade and duration	15.2 (11.8)	15.6 (10.2)	<0.0001	NA	NA

Data are n (%) or mean (SD). *Data were missing for 308 participants. †Data were missing for 454 participants. ‡Data were missing for 452 participants. NA = not applicable. ICV = inactivated COVID-19 vaccine. CI = confidence interval.

The results for subcohorts grouped by the type of the booster vaccine showed the hazard ratio and relative effectiveness against COVID-19 of the ICV + Ad5-IH cohort versus the ICV-community cohort was 0.86 (95% CI 0.85–0.88) and 13.7% (95% CI 11.9–15.4), respectively (Table 3). While, the hazard ratio and relative effectiveness against COVID-19 for ICV + Ad5-IM cohort versus the ICV-community cohort were 0.88 (95% CI 0.82–0.95) and 11.7% (95% CI 5.3–17.6), respectively.

Table 3. Relative effectiveness against COVID-19 of the heterologous-trial cohort versus the ICV-community cohort stratified by the type of the booster vaccine.

Heterologous schedule type	Heterologous-trial cohort	ICV-community cohort	p-value	Hazard ratio (95% CI)	Relative effectiveness % (95% CI)
ICV + Ad5-IH	n = 9113	n = 29935
Number of cases (%)	5031 (55.2)	19130 (63.9)	<0.0001	0.86 (0.85, 0.88)	13.7 (11.9, 15.4)
Case type
Antigen rapid test or nucleic acid test-confirmed cases	1853 (36.8)	5481 (28.7)	<0.0001	1.11 (1.06, 1.16)	−11.1 (−16.4, −5.9)
Clinical cases	3178 (63.2)	13649 (71.4)		0.77 (0.75, 0.79)	22.7 (24.9, 20.5)
Age group, years
18–59	3337 (66.3)	12810 (67.0)	0.3953	0.86 (0.84, 0.88)	13.8 (11.7, 15.8)
≥60	1694 (33.7)	6320 (33.0)		0.87 (0.84, 0.90)	13.2 (9.90, 16.4)
Severity*
Mild	4905 (97.9)	18459 (97.8)	0.7714	0.86 (0.85, 0.88)	13.5 (11.7, 15.3)
Moderate	107 (2.1)	402 (2.1)		0.86 (0.70, 1.07)	13.7 (−6.6, 30.1)
Severe	0 (0.0)	5 (0.0)		NA	NA
Outpatient visit†	571 (11.8)	3015 (16.0)	<0.0001	0.74 (0.68, 0.80)	26.5 (20.1, 32.4)
Hospitalization‡	32 (0.7)	247 (1.3)	0.0002	0.50 (0.35, 0.73)	49.7 (27.4, 65.1)
Score based on symptom grade	4.76 (3.2)	4.90 (2.7)	<0.0001	NA	NA
Score based on symptom grade and duration	14.89 (11.5)	15.46 (10.0)	<0.0001	NA	NA
ICV + Ad5-IM	n = 367	n = 1325
Number of cases (%)	262 (71.4)	1071 (80.8)	0.0001	0.88 (0.82, 0.95)	11.7 (5.3, 17.6)
Case type
Antigen rapid test or nucleic acid test-confirmed cases	136 (51.9)	369 (34.5)	<0.0001	1.33 (1.14, 1.56)	−33.1 (−56, −13.5)
Clinical cases	126 (48.1)	702 (65.6)		0.74 (0.66, 0.84)	25.7 (15.9, 34.4)
Age group, years
18–59	188 (71.8)	637 (59.5)	0.0002	1.02 (0.96, 1.09)	−2.3 (−9.4, 4.3)
≥60	74 (28.2)	434 (40.5)		0.65 (0.56, 0.77)	34.6 (23.0, 44.4)
Severity**
Mild	246 (96.9)	1025 (97.1)	0.2686	0.87 (0.81, 0.94)	12.6 (5.9, 18.8)
Moderate	7 (2.8)	31 (2.9)		0.83 (0.37, 1.86)	17.4 (−86.1, 63.3)
Severe	1 (0.4)	0 (0.0)		NA	NA
Outpatient visit††	52 (20.5)	226 (21.4)	0.7398	0.96 (0.73, 1.25)	4.4 (−25.0, 26.9)
Hospitalization‡‡	2 (0.8)	12 (1.1)	0.6263	0.69 (0.16, 3.07)	30.8 (−207.4, 84.4)
Score based on symptom grade	6.1 (4.0)	5.6 (3.3)	0.1297	NA	NA
Score based on symptom grade and duration	20.4 (15.1)	18.8 (12.7)	0.3005	NA	NA

Data are n (%) or mean (SD). *Data were missing for 285 participants. †Data were missing for 430 participants. ‡Data were missing for 428 participants. §Data were missing for 426 participants. **Data were missing for 23 participants. ††Data were missing for 24 participants. ‡‡Data were missing for 24 participants. §§Data were missing for 24 participants. NA = not applicable. ICV = inactivated COVID-19 vaccine. Ad5-IH = adenovirus type 5 vectored COVID-19 vaccine through oral inhalation. Ad5-IM = adenovirus type 5 vectored COVID-19 vaccine through intramuscular injection. CI = confidence interval.

Relative effectiveness of difference across the subgroups defined by different heterologous prime-boost regimens, with the highest of 36.6% (95% CI 23.0–47.9) in the 1 dose ICV + 1 dose Ad5-IM cohort, followed by 14.1% (95% CI 12.3–15.9) in the 2 doses ICV + 1 dose Ad5-IH cohort, 13.4% (95% CI 4.8–21.1) in the 2 doses ICV + 1 dose Ad5-IM cohort, and −23.9% (95% CI −40.7 to −9.1) in the 3 doses ICV + 1 dose Ad5-IM cohort, with the lowest of −26.1% (95% CI −45.4 to −9.3) for the 3 doses ICV + 1 dose Ad5-IH cohort (Additional file 1: Table S4). In addition, the relative effectiveness of the heterologous-trial cohort versus the ICV-community cohort after bootstrapping from subcohorts grouped by the type of the booster vaccine or different prime-boost regimens was similar to the above results (Additional file 1: Table S5-S6).

The lowest hazard ratio against COVID-19 of the heterologous-trial cohort versus the ICV-community cohort was 0.72 (95% CI 0.67–0.77) in Yunnan, followed by 0.79 (95% CI 0.74–0.85) in Shandong, 0.83 (95% CI 0.79–0.87) in Hunan, 0.90 (95% CI 0.85–0.95) in Chongqing, 0.91 (95% CI 0.88–0.94) in Jiangsu, and 0.93 (95% CI 0.89–0.98) in Anhui (additional file 1: Table S7). The ranking from highest to lowest for relative effectiveness against COVID-19 of heterologous-trial cohort versus the ICV-community cohort was Yunnan (28.3%, 95% CI 23.0–33.1), Shandong (20.6%, 95% CI 15.2%–25.6), Hunan (16.9%, 95% CI 12.8–20.9), Chongqing (9.8%, 95% CI 4.5–14.7), Jiangsu (9.3%, 95% CI 6.2to 12.3), and Anhui (6.7%, 95% CI 2.5–10.7). Furthermore, the average scored symptoms of COVID-19 cases as well as the severity in the heterologous-trial cohort was 4.8 and that in the ICV-community cohort was 4.9, which were numerically similar, but statistically significant (p < 0.0001) (Table 2). Meanwhile, symptom scores along with durations were also numerically similar, but statistically significant (15.2 vs 15.6, p < 0.0001).

Multivariate logistic regression analysis showed the hazard of breakthrough infection risk was negatively associated with Ad5-IH compared to ICV (OR 0.63, 95% CI 0.60–0.66), but positively associated with the number of priming doses of ICV (2 vs. 1: OR 2.98, 95% CI 1.70–5.21; 3 vs. 1: OR 4.79, 95% CI 2.66–8.66) (Table 4). In addition, ageing, male, and smoking were associated with a lower hazard of breakthrough infection risk, but overweight, alcohol, tea, or coffee consumption, comorbidities, allergy history, and the time since the most recent COVID-19 vaccination were associated with higher hazard. Furthermore, multivariate logistic regression involving specific prime-boost regimens showed that 2 doses ICV + 1 dose A5d IH (OR 0.63, 95% CI 0.60–0.66), and 1 dose ICV + 1 dose Ad5 IM (OR 0.27, 95% CI 0.17–0.44) were negatively associated with the hazard of infection risk compared with 3 doses ICV (additional file 1: Table S8).

Table 4. Multivariate logistic regression analysis of COVID-19 incidence based on the type of the booster vaccine and the prime doses of inactivated COVID-19 vaccine.

Factors	OR	95% CI	p-value
Type of the booster vaccine
ICV	1 (reference)
Ad5-IM	0.81	0.58, 1.12	0.1973
Ad5-IH	0.63	0.60, 0.66	<0.0001
Prime doses of ICV
1	1 (reference)
2	2.97	1.70, 5.21	0.0001
3	4.79	2.66, 8.66	<0.0001
Age, years			0.0001
18–29	1 (reference)
30–39	0.92	0.84, 1.01	0.0924
40–49	0.69	0.62, 0.75	<0.0001
50–59	0.50	0.45, 0.55	<0.0001
60–69	0.41	0.38, 0.45	<0.0001
70–79	0.36	0.33, 0.40	<0.0001
≥80	0.31	0.26, 0.36	<0.0001
Sex
Female	1 (reference)
Male	0.68	0.64, 0.72	<0.0001
BMI
<18.5	0.96	0.86, 1.07	0.4511
≥18.5 and <24	1 (reference)
≥24	1.06	1.01, 1.11	0.0097
Smoking
No	1 (reference)
Yes	0.59	0.56, 0.63	<0.0001
Alcohol drinking
No	1 (reference)
Yes	1.06	1.00, 1.12	0.0336
Tea drinking
No	1 (reference)
Yes	1.11	1.06,1.16	<0.0001
Coffee drinking
No	1 (reference)
Yes	1.20	1.11, 1.31	<0.0001
Comorbidities
No	1 (reference)
Yes	1.41	1.34, 1.48	<0.0001
Allergic history
No	1 (reference)
Yes	1.23	1.11, 1.35	<0.0001
Tetanus vaccination history
No	1 (reference)
Yes	1.04	0.93, 1.16	0.4759
Influenza vaccination history
No	1 (reference)
Yes	1.00	0.90, 1.11	0.9969
The interval from last dose vaccination to 1 November 2022 (months)	1.16	1.13, 1.18	<0.0001

ICV = inactivated COVID-19 vaccine. Ad5-IM = adenovirus type 5 vectored COVID-19 vaccine through intramuscular injection. Ad5-IH = adenovirus type 5 vectored COVID-19 vaccine through oral inhalation. BMI = body mass index. OR = odds ratio. CI = confidence interval.

Besides, we also calculated the hazard ratio and relative effectiveness against COVID-19 of the heterologous-trial cohort versus the ICV-trial cohort, which showed a slightly lower overall hazard ratio of 0.76 (95% CI 0.72–0.80) and higher relative effectiveness of 24.5% (95% CI 20.3–28.5), respectively (additional file 1: Table S9).

Discussion

A nationwide outbreak of COVID-19 associated with omicron-predominate subvariants BA.5.2, BA.2.76, and BF.7 occurred in China at the end of the year 2022, causing a wave of large-scale infections among the population, despite a high ICV vaccination coverage [18]. In this study, we did a retrospective survey of a large, diverse population of people aged 18 years or older, from six provinces in China, and identified an overall breakthrough infection proportion of 62.7% among the responders. The reasons for the high attack rate during the study period are complicated, but the main reason should be it has been a long time (about 8–12 months) since the last dose of vaccination. The durability of the protection of COVID-19 vaccines is not good and the protection waned quickly after the vaccination, regardless of the types of COVID-19 vaccines, which has been wildly observed [19]. In addition, another probable reason is that the population in China had taken a zero-COVID policy and experienced a long period of lock-down with no natural exposure to any SARS-CoV-2 virus, which hurdled to achieve herd immunity against COVID-19, while the SARS-CoV-2 were spreading around in other areas. Another sentinel community-based surveillance in rural areas reported SARS-CoV-2 infection peaked between December 20 and –22[5], which is consistent with our survey.

During the omicron-predominate outbreak, the relative effectiveness of heterologous prime-boost regimen versus three-dose vaccination of ICV in preventing breakthrough COVID-19 cases was lower (13.7% [95% CI 11.9–15.3]), while that preventing the COVID-19 associated with outpatient visit and hospitalization was higher (25.1% [18.9–30.9] and 48.9% [27.0–64.2]), respectively. These data indicated that the advantages of the heterologous booster with Ad5-IM or Ad5-IH are not only reflected in the enhancement of the immune responses induced by ICV priming in the short term but also in increasing the vaccine protection against COVID-19-associated hospital admission, even about 10 months after the vaccination, which was of big significance in reducing the burden of medical resources.

We found that the elderly reported significantly lower COVID-19 attack rates compared with young adults, showing a reduced hazard ratio associated with age increasing. This result seems to be contrary to the fact that SARS-CoV-2 infection risk increases with age, but it may reflect that the elderly population is more likely to take behaviours to reduce exposure risk like mask using, avoiding close contact with persons with COVID-19, or reducing outdoor activities during the outbreak period. Besides, the results of another survey conducted in China from December 2022 to February 2023 also showed that the COVID-19 attack rate of the elderly was about 10% lower than young adults [18], which was consistent with our findings. Although the elderly reported a lower overall attack rate of COVID-19 breakthrough cases, they accounted for a higher proportion of outpatient visits and hospitalization. Taken together, these findings suggest that the benefit of a heterologous booster with Ad5-IM or Ad5-IH following the ICV priming could be substantially higher in the elderly.

The overall infection proportion found in our study was relatively lower compared with a previous report on a survey conducted across 31 provinces in China [18], which reported a proportion of 82.4% of individuals with SARS-CoV-2 infection between December 2022 and February 2023. The differences may be explained by the following three reasons. First, the surveyed populations were different in geographical area and age range in the two studies. Participants in the previous study were much younger than those in our study, the proportion of participants aged ≥ 60 years was 4.9%, while that of our study was 37.7%. Second, the previous survey study was conducted through a self-reported anonymous questionnaire accessed through social media. Thus, the infected individuals may be more inclined to respond to the survey, while those who were not infected were less likely to respond. Third, the determination of clinal COVID-19 cases in our study required at least two typical symptoms combined with a COVID-19 exposure, while the previous survey reported any suspected symptoms during that outbreak.

Although the overall patterns of epidemic outbreaks are similar across provinces, variations in infection escalation and cumulative incidences were noted. The relative effectiveness against COVID-19 may be affected by the different epidemic intensities in different provinces, and the lower epidemic intensity is associated with higher vaccine protection.

Results of multivariate logistic regression analysis suggested that the increase in numbers of ICV priming doses reduced additional protection offered by the heterologous booster. It may be related to the immunological imprinting effect that repeated vaccination of ICV mainly recalled memory B cells elicited by ICV-based wild-type strain, but rarely produced variant-specific B cells, thereby impairing the protection against omicron [20-22]. Similarly, immune imprinting has indeed also been observed in mRNA COVID-19 vaccines [23]. After controlling prime doses of ICV, a booster dose of Ad5-IH showed a significantly lower hazard ratio (OR 0.63, 95% CI 0.60–0.66) compared with three doses of ICV, but not Ad5-IM (OR 0.81, 95% CI 0.58–0.12). Although previous studies have reported that influenza and tetanus vaccination might have a non-specific protective effect on SARS-CoV-2 infection and/or COVID-19 severity [24-27], this was not observed in this study. We found drinking tea or coffee was positively associated with the hazard of breakthrough infection, reflecting that people who work long hours and experience fatigue or high levels of stress are more vulnerable to breakthrough infection. In addition, overweight participants suffered a higher risk of breakthrough infection compared with those with normal weight, and smoking was associated with a lower risk for non-smoking, which is consistent with previously reported studies [28,29].

We have reported the results of the survey based on a clinical trial of heterologous boost with an aerosolised Ad5-nCoV after two-dose ICV in adults (NCT05204589), showing a comparative vaccine protection of 35.1% for Ad5-IH versus ICV against COVID-19 [12]. However, due to the small number of participants receiving ICV in the trial (<200) in that trial, it is impossible to access the increase of vaccine protection against more severe COVID-19 diseases.

In this study, we individually matched community individuals who received three-dose ICV based on the characteristics, we expected to be important for the COVID-19 risk, like age, sex and region, and vaccination time for the last dose to emulate a randomized trial. However, the two different sources of participants may still differ in their self-awareness and reporting of other health conditions, as well as from unmeasured behaviours which may potentially lead to bias. We did note a higher breakthrough infection reported in the ICV-trial cohort than did the ICV-community cohort (73.9% vs. 64.6%). Besides, the response proportions of the heterologous-trial cohort, ICV-trial cohort, and ICV-community cohort varied from 71.2% to –93.2%, which may also cause some bias. However, there were 3 peaks of infection on December 10, 15, and 20 in our study, respectively, which was much higher than the days before or after. One of the reasons may be some participants may not be able to accurately recall the specific date of infection, they are more likely to estimate by the first, middle, and last ten days of a month (i.e. December 10, 15, and 20).

We ensure the reliability of the survey data from three aspects. First, all the investigators had received unified training before conducting the surveys. Individuals in either the heterologous-trial cohort, ICV-trial cohort, or the ICV-community cohort were interviewed by the same group of trained and qualified investigators. Second, we used a computer-assisted telephone interview system to conduct the online survey and generate raw data in a real-time. So, the designated quality control personnel could check and supervise the data promptly to ensure the authenticity of the data. Third, we checked the consistency of the survey by carrying out a proportional sampling (≥5%) of the completed questionnaires by each investigator for repeated surveys.

There are some other limitations of our study. First, the sample sizes of participants receiving heterologous boosters with Ad-IH or Ad5-IM for subgroup analysis by prime-boost regimens. Therefore, we employed bootstrap sampling to simulate an expanded sample size. Second, only a small part of the reported COVID-19 cases was aetiologically confirmed, due to the limited accessibility of antigen rapid tests or nucleic acid tests during the outbreak. Clinical cases of COVID-19 confirmed by symptoms and exposure history may be misidentified. Third, since it is a retrospective study, there would inevitably be some recall bias in the acquisition of breakthrough infection. In addition, our results could be affected by survivor bias since some life-threatening or fatal cases were unable to respond to our survey. At last, our survey only covered the populations in six provinces, the findings might not be generalizable to the whole China population.

Conclusion

Our findings showed that the heterologous boost regimen offered low additional protection in preventing breakthrough COVID-19, but higher in preventing outpatient visits and hospitalization compared with the homologous ICV regimen, during the omicron-predominate outbreak at the end of year 2022, in China, about 10 months after the vaccination.

Availability of data and materials

The study protocol is available in the Supplementary Information file. Researchers, who provide a scientifically sound proposal, are allowed to access to the de-identified individual participant data. Individual participant data can be obtained with a request to [email protected], [email protected], or [email protected].

Acknowledgements

We acknowledge all investigators of Jiangsu, Chongqing, Shandong, Hunan, Anhui, and Yunnan.

Disclosure statement

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

Additional information

Funding

National Natural Science Foundation of China (grant numbers 82341031, 82173584, and 82222062), Major Research Plan of the National Natural Science Foundation of China (grant number 92269205), Science Fund for Distinguished Young Scholars of Jiangsu Province (grant number BK20220064), and Jiangsu Provincial Key Project of Science and Technology Plan (grant numbers BE2021738 and BE2023601).