Impact of a delayed second dose of mRNA vaccine (BNT162b2) and inactivated SARS-CoV-2 vaccine (CoronaVac) on risks of all-cause mortality, emergency department visit, and unscheduled hospitalization

Mass vaccination programs have been recognized as a key strategy to achieve high levels of population immunity to mitigate the impact of the coronavirus disease 2019 (COVID-19) pandemic. Hence, various SARS-CoV-2 vaccine candidates have been developed at an unprecedented speed using different platform technologies [1]. While most types of vaccine require a two-dose regimen for adequate protection, some governments have been exploring the option of delaying the second dose to beyond the recommended interval of 3–4 weeks between doses, as a contingency measure to offer at least partial immunity to a larger population in the midst of limited vaccine supply [2, 3]. In fact, the ChAdOx1 nCoV-19 vaccine (AZD1222) has been observed to demonstrate higher efficacy and antibody titers with longer prime-boost intervals, such as receiving the second dose ≥ 12 weeks compared to < 6 weeks after the first dose [4]. A recent study on this viral vectored vaccine has identified even higher antibody titers with extended dosing intervals of up to 44–45 weeks, along with further boosts of both humoral and cellular immune responses following the administration of a third dose 28–38 weeks after the second [5]. Similarly, preliminary evidence suggested higher neutralizing antibody levels with the third dose of CoronaVac (an inactivated SARS-CoV-2 vaccine), administered 6 months compared to 4 weeks after the second dose, regardless of a 2-week or 4-week interval between the first and second doses [6]. The theoretical benefit of delaying the second dose was obvious, achieving higher vaccine uptake in a short time and decreasing COVID-19 infection rate, hospitalization, and death under circumstances of limited vaccine supply.

Delaying the second dose of SARS-CoV-2 mRNA vaccines has been supported by mathematical models to reduce infection, hospitalization, and mortality due to COVID-19, conditioned on factors including the vaccination rate, the efficacy of the first dose, and the level of pre-existing immunity [7, 8]. At the individual level, people may postpone or even refuse the second dose vaccination for various reasons, for example, feeling unwell at the scheduled time or having experienced adverse reactions following the first dose [9]. Such fear or hesitancy is valid in light of some real-life observations suggesting that those with adverse reactions to the first dose are at an increased risk of reporting them again after the second dose, especially of the same side effects [9, 10]. In general, the safety profile of a vaccine includes “all adverse events that could potentially be caused, triggered, or worsened at any time after vaccination,” which may consist of acute or chronic health conditions, such as allergic or anaphylactic reactions, and diseases diagnosed following vaccination [11, 12].

In addition to potentially discouraging timely and complete vaccination for adequate protection, adverse events following vaccination may prompt emergency department (ED) visits or even hospitalization if they are severe enough, for instance, cardiac and neurologic symptoms have been identified as primary causes of these medical encounters after the administration of SARS-CoV-2 vaccines [13‐15]. Nevertheless, studies on the utilization of healthcare services following vaccination remain scarce [12, 16‐18], and to the best of our knowledge, no clinical trials have specifically examined such outcomes with varying intervals between doses. Therefore, this retrospective study aims to address the research gaps of describing the use of health care services after SARS-CoV-2 vaccination in a real-life setting, when the second dose is administered on a delayed schedule compared to that within the recommended interval.

Baseline characteristics of fully vaccinated individuals in the two groups by brands of vaccine are presented in Table 1. Among 771,780 individuals administered with two doses of COVID-19 vaccines from February 23 to July 3, 2021, eight receiving BNT162b2 as the first dose followed by CoronaVac were excluded. A total of 654,900 individuals received the second dose within the recommended interval of respective vaccine (401,473 BNT162b2 recipients and 253,427 CoronaVac recipients), while 16,024 (3.8%) and 100,856 (28.5%) individuals received the second dose of BNT162b2 and CoronaVac beyond the recommended intervals, respectively (Fig. 1). There were < 5 daily new COVID-19 infections in the community (500 local cases in 159 days) during the study period, while one new polymerase chain reaction-positive case following the second dose of BNT162b2 vaccination and one COVID-19 case following the second dose of CoronaVac vaccination were reported in our cohort. Among those who received the second dose late, the mean (SD) dosing intervals were 34.4 (8.2) days and 31.8 (6.7) days in the BNT162b2 and CoronaVac groups, respectively (Table 1). Additional file 1: Figure S1 shows the distributions of the second dose interval of BNT162b2 and CoronaVac recipients. The median follow-up was 28 days (interquartile range [IQR] 28–28) for the overall cohort. Additional file 1: Figure S2 shows the distribution of propensity scores by vaccine types. After weighting, all SMDs of baseline characteristics were less than 0.1 (Additional file 1: Table S3).

There were 25,421 individuals (3.3%) who visited EDs after the second dose during the follow-up period (14,453 [3.5%] BNT162b2 recipients, of which 635 [4.0%] had a delayed second dose; 10,968 [3.1%] CoronaVac recipients, of which 3112 [3.1%] had a delayed second dose). The median time from receiving the second dose to ED visits was 10 days (IQR 3–19) for all BNT162b2 recipients, while the median duration was 12 days (IQR for the recommended interval group 5–20; IQR for the delayed group 6–20) in CoronaVac recipients (Table 2). Delaying the second dose was not associated with increased or decreased risk of ED visit compared to individuals receiving it within the recommended interval (BNT162b2: 14.49 vs 12.55 per 10,000 person-days, HR = 1.037, 95% CI 0.951–1.130, P = 0.411; CoronaVac: 11.21 vs 11.26 per 10,000 person-days, HR = 0.966, 95% CI 0.926–1.008, P = 0.113).

Regarding other outcomes of interest, there were 40 deaths and 5902 unscheduled hospital admissions within 28 days risk period after vaccination. The median time from the second dose administration to unscheduled hospitalization was 11 and 13 days for all BNT162b2 and CoronaVac recipients, respectively. No statistically significant differences in all-cause mortality and risk of unscheduled hospitalization (BNT162b2: 2.78 vs 2.62 per 10,000 person-days, HR = 1.054, 95% CI 0.867–1.281, P = 0.597; CoronaVac: 2.78 vs 2.93 per 10,000 person-days, HR = 0.956, 95% CI 0.878–1.040, P = 0.294) were identified for people who delayed the second dose of either vaccine. Primary causes of unscheduled hospitalization of people receiving the second dose in recommended or delayed dosing interval by the type of vaccine are shown in Additional file 1: Table S4. There were no marked differences in the top rank of disease diagnosis between the recommended and delayed groups. Unscheduled hospitalization due to “symptoms, signs, and ill-defined conditions” accounted for the most in the recommended and delayed groups for both BNT162b2 and CoronaVac recipients. Four young male (age range 19–35 years) myocarditis cases occurred at 5–20 days after the second dose of BNT162b2 in the recommended group, and no myocarditis cases occurred after the second dose of CoronaVac and delayed second dose of BNT162b2.

Subgroup analyses (Additional file 1: Table S5) showed delaying the second dose had no significant impact on the outcomes for elderly people, consistent with the main analysis. Slightly delaying or more delaying the second dose was not associated with increased or decreased risk of ED visit or unscheduled hospitalization (Additional file 1: Table S6).

The results of sensitivity analyses that excluded COVID-19-related outcomes or physical injuries (Additional file 1: Table S7), included infinite delayers to the delayed group (Additional file 1: Table S8), and used the doubly robust method (Additional file 1: Table S9) were consistent with the result of the main analysis.

This current study examined the safety following the second dose of two different vaccine platforms (BNT162b2, an mRNA vaccine, and CoronaVac, an inactivated virus vaccine), from the aspect of mortality and emergency service utilization, in relation to all adverse events occurring after vaccination that would have been severe or persistent enough to prompt emergency care services. Accordingly, delaying the second dose of BNT162b2 or CoronaVac was not associated with any change in all-cause mortality, risks of ED visit, and unscheduled hospitalization.

While a limited number of studies have explored the alternative dosing regimens with extended prime-boost intervals on the efficacy and immunogenicity outcomes of SARS-CoV-2 vaccines [4, 41‐44], comparing delayed to recommended dosing intervals on vaccine safety and utilization of healthcare services has not been addressed in the current literature. In fact, the efficacy and immune responses induced by SARS-CoV-2 vaccines can vary substantially depending on the different platform technologies [1], which would have contributed to the discrepancies in their safety profile. For instance, a systematic review and meta-analysis of clinical trials have found that SARS-CoV-2 mRNA vaccines were associated with relatively higher incidences and risks of adverse reactions versus control, in comparison with vectored or inactivated vaccines, and in particular, for systemic adverse reactions, an increased risk was only observed with mRNA vaccines but not inactivated ones [45]. Therefore, our results should be considered based on the respective vaccine strategies.

As vaccine reactogenicity is likely immune-mediated, it seems logical to expect more side effects of potentially higher severity among individuals with more robust immune responses and inflammation to vaccination [11]. For example, a number of studies have generally concluded that essentially more adverse events, especially systemic reactions, were identified after the second dose of BNT162b2 compared to the first dose, which could be related to the greater immunogenicity following the priming or sensitization to viral antigen with the first dose, and similar associations could be detected with younger age, female sex, individuals with allergies, and COVID-19 survivors [10, 17, 34, 46‐53]. Studies had quantified that the amount and lifespan of neutralizing antibodies fell remarkably before the day of the recommended second dose. In contrast, the antibody levels were significantly higher following the second dose of BNT162b2 when the dosing interval was extended to 6–14 weeks instead of the recommended 3–4 weeks, whereas T cell responses were moderately sustained [41, 43, 44, 54]. Therefore, it is proposed that the second encounter of the viral antigens would enhance the immune response when the antibody levels were dropped too low. Yet, a further investigation into the immunogenicity outcomes of delaying the second dose is needed to clarify its impact on reactogenicity, if there is any, and the possible mechanisms involved.

Regarding the inactivated vaccine CoronaVac, data on its safety and efficacy with varying prime-boost intervals remain scarce. One study has recognized that while the seropositive rate was comparable among individuals who had administered their second dose 21 days (recommended) or 1–7 months (extended) after the first dose, the latter group was associated with a significantly lower median antibody level post-vaccination, despite no significant differences were detected with varying extended intervals [42]. Besides, with reference to the platform technology of CoronaVac, antigen immunogenicity of inactivated vaccines could be hampered by the inactivation process, and the antibody titers tend to decrease over time, hence necessitating the administration of several booster doses [1]. While CoronaVac has also been demonstrated to elicit T cell responses in vaccinated individuals [55], the activation and durability of such cellular immunity with varying dosing intervals remain to be verified. Overall, it could be speculated that the immunogenicity of CoronaVac with delayed second dose would unlikely trigger substantial reactogenicity, as evidenced by the comparable risks of medical encounters to those associated with the recommended dosing interval in our study, and in line with studies reporting lower or similar incidences of adverse events following the second dose versus the first [9, 56].

While the relationship between immunogenicity and reactogenicity of SARS-CoV-2 vaccines remains to be elucidated, several studies have observed that a minority of vaccinated individuals did experience adverse reactions that prompted them to seek medical care, at either outpatient clinics or ED, but eventually required hospitalization for further management [10, 12, 16‐18, 50]. While some have explored such utilization of healthcare services among vaccinated individuals stratified by the number of doses and any previous SARS-CoV-2 infection, none has described them in relation to varying prime-boost intervals. Further research is needed to determine if the magnitude of inflammation induced by adaptive immune responses of SARS-CoV-2 vaccination parallels the incidence and severity of reactogenicity symptoms [11], with reference to the potential factors aforementioned. In combination with data on the utilization of healthcare services by vaccinated individuals, health professionals can be informed about any trends or specific populations at risk of requiring medical attention and further contribute to monitoring the safety of SARS-CoV-2 vaccination.

Considering the potential waning of immunity from the standard two-dose regimens and the emerging variants of concern, many governments are now initiating the administration of a third dose (or booster) as an attempt to maintain immune protection in the population, especially for the elderly and immunocompromised individuals who are more at risk of severe COVID-19 [57, 58]. Both randomized and observational studies have recently demonstrated that the third dose could significantly enhance the neutralizing antibody responses that have been waning over time among individuals fully vaccinated with two doses, which in turn protect them against SARS-CoV-2 infection, severe COVID-19, and related death, and possibly reducing transmission in the community; these have been observed with both BNT162b2 [59‐63] and CoronaVac [6, 58, 64, 65], notably in extended dosing intervals and heterologous vaccination. However, the massive rollout of booster doses remains an ethical controversy given the unequal access and distribution of vaccines worldwide, as well as the lack of long-term data on their safety and efficacy, especially in view of the increasing cases of breakthrough infection with the Omicron variant despite vaccination with booster doses [66]. Moreover, as protection against severe COVID-19 and related death relies mainly on the standard two vaccine doses, further studies are needed to evaluate the cost-effectiveness and optimal timing of booster doses, in addition to any unexpected increases in the incidence of adverse reactions or utilization of healthcare services following vaccination [6, 60, 66].

In this real-life setting, mortality, ED visits, and unscheduled hospitalizations of vaccinated individuals were captured, which were likely contributed by adverse reactions following vaccination that were severe or persistent enough for the people to seek urgent medical care. In contrast to surveillance studies that relied on self-reported adverse reactions, our study would have only included those severe enough to require emergency medical care, which could be an essential piece of information in the evaluation of healthcare services and allocation of resources in the midst of the current pandemic.

Our study is unique in presenting the safety of two different SARS-CoV-2 vaccines via their associated incidences of emergency services and all-cause mortality, based on the comparison of delayed versus recommended dosing intervals. Nevertheless, several limitations of this observational study should also be addressed. First, reasons for delaying the second dose were not available for further analysis, such as mild adverse events not captured by hospital records after the first dose. Hence, the impact of extended dosing interval could not be evaluated in association with the safety of second dose vaccination for individuals developing particular adverse events. Second, our results could be confounded by several residual or unmeasured factors, including but not limited to the lack of data on the utilization of private healthcare services and on individuals who had adverse reactions to the first dose and therefore avoided the second dose totally. Nevertheless, the effect of infinite delayers on our findings was minimal as demonstrated by sensitivity analysis (Additional file 1: Table S8). Furthermore, event occurrence (i.e., ED visits and unscheduled hospitalizations) prior to the second dose were well balanced after the weighting (Additional file 1: Table S3), so confounding effects by the event outcome prior to the second dose were negligible. Third, the lack of immunogenicity data for our study has prevented us from making further inferences about vaccine reactogenicity. The impact of delaying the second dose on COVID-19 infection or vaccine effectiveness was not quantified because of low COVID-19 infection rates during the study period. Fourth, given the relatively small number of more serious events such as deaths and unscheduled hospitalizations, our study would not have the power to distinguish very small differences in event rates, although we are able to confirm that any differences in rates were no greater than 27% for deaths for the BNT162b2 vaccine. Lastly, our results may not be generalizable to other settings, such as those with different population demographics, inadequate vaccine supply, or experiencing COVID-19 outbreaks, as they could confound the utilization of healthcare services and implementation of vaccination strategies.

In conclusion, this study has provided empirical data to describe the trends and populations at risk of mortality and utilizing emergency healthcare services following SARS-CoV-2 vaccination. Delaying the second dose of BNT162b2 or CoronaVac was not associated with increases in all-cause mortality, risk of ED visit, and unscheduled hospitalization compared to vaccination within the recommended interval. Regardless of the recommended or delayed schedule for SARS-CoV-2 vaccination, the second dose of both vaccines should be administered to obtain better protection against infection and serious disease. Second dose should be administered within the recommended interval following the manufacturer’s product information, until further studies support the benefits of delaying vaccination outweighing the risks. Future research should explore the relationship between immunogenicity and reactogenicity of SARS-CoV-2 vaccines in various population subgroups and dosing intervals.