Introduction

The rapid development of vaccines against SARS-CoV-2 played a central role in reducing the global burden of the COVID-19 pandemic [1,2], with substantial early reductions in COVID-19 symptomatic and severe outcome rates following vaccination [3]. High vaccination rates and naturally acquired immunity have helped reduce cases and deaths from COVID-19, culminating in the end of the Public Health Emergency of International Concern in 2023 [4,5].

The CoronaVac inactivated whole-virion SARS-CoV-2 vaccine (Sinovac) was one of the first COVID-19 vaccines extensively used globally, particularly in low-and-middle income countries [4]. Clinical trials and real-world studies have demonstrated that the vaccine effectively reduced the severity of the disease and mortality, with minimal adverse events reported [4–7]. Furthermore, this vaccine has shown efficacy against various SARS-CoV-2 variants, including the Gamma variant first identified in samples from Manaus, Brazil [8–10].

CoronaVac was among the first vaccines approved and distributed on a large scale in Brazil in 2021. In Manaus, where COVID-19 cases and death rates exceeded those of the rest of the country during the second wave caused by the Gamma [10], the limited vaccine availability necessitated targeted campaigns to ensure protection of the most vulnerable populations [11,12]. Brazilian national guidelines began vaccinating the older population and healthcare professionals first. This targeted approach aimed to alleviate the burden on healthcare systems and protect those at higher risk of severe outcomes from the virus [12,13]. Due to the limited availability of more vaccines, younger groups, even those living with comorbidities, were not initially prioritized for vaccination at the beginning of 2021.

In 1996, Finkelstein et al [14] proposed allocating experimental interventions according to the individual participant’s risk to avoid the ethical challenge of potentially denying the best opportunity to those in the highest need. A control group, deliberately selected to have a lower risk, would also be concurrent; therefore, it would differ only in terms of risk factors and the intervention. The idea behind it is to demonstrate that the intervention can decrease the risk of an adverse outcome in those more likely to have one to the level of those with a better prognosis. Therefore, a clear identification of risk factors and an understanding of ethical constraints are required to consider the proposed risk-based allocation trial design. Nonetheless, no clinical study has been published to date using this design.

Given the resource-limited setting at the start of the COVID-19 vaccination campaign, we proposed a risk-based allocation clinical trial anticipating the use of CoronaVac in patients aged 18–49 with comorbidities, a population known to have increased COVID-19 mortality, but not initially targeted according to the Brazilian Ministry of Health Guidelines. We hypothesized that COVID-19-associated mortality in this group would be similar to that of a same age group without comorbidities and not vaccinated.

Methods

Results

Between March 18 and April 8, 2021, a total of 16,517 participants expressed their intention to participate in the study via online screening. After eligibility screening, 6,226 participants were included: 1,139 (18%) in the low-risk control group, and 5,087 (72%) in the High-risk vaccinated group (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Flowchart of participant inclusion.

Recruitment and screening occurred simultaneously for both groups.

https://doi.org/10.1371/journal.pone.0351566.g001

Table 1 summarizes the baseline data of the participants. In the high-risk vaccinated group, the most common comorbidities were obesity (81.1%), systemic arterial hypertension (20.1%), and diabetes (10.5%). Of 5,087 vaccinated participants, sixteen (0.3%) did not receive the second dose.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics of study participants.

https://doi.org/10.1371/journal.pone.0351566.t001

At baseline (D0), 61.9% of participants had reactive serology (greater than 0.8 U/mL): 60.3% in the unvaccinated comparison group and 62.2% in the vaccinated group (Table 1). The latter also exhibited highly reactive serology (≥250 U/mL) compared to the low-risk control group at baseline and D90, with a decline in highly reactive titers observed at D180 (Table 2). There was a significant reduction in COVID-19 infections among vaccinated participants during the first 28 and 90 days of follow-up, with a continued, albeit smaller and not statistically significant, reduction at 120 and 180 days (Table 2). Vaccinated high-risk individuals had similar severe outcomes (hospitalization and death) to unvaccinated low-risk individuals at all points analyzed (Table 2).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. COVID-19 endpoints and serodetection among participants during follow-up.

https://doi.org/10.1371/journal.pone.0351566.t002

Hospitalization rates were low in both groups, with 0.14–0.24% observed among high-risk vaccinated individuals across all follow-up periods. In comparison, low-risk unvaccinated individuals had hospitalization rates ranging from 0.0% to 0.29%, though the number of participants in this group decreased substantially over time. The absolute risk reduction between groups was modest (up to 0.15%), yielding an estimated number needed to treat (NNT) of approximately 667 to prevent one additional COVID-19-related hospitalization. This NNT reflects a pragmatic estimate of vaccine effectiveness in reducing hospitalizations among high-risk individuals in a real-world setting. However, given the small number of hospitalization events and the lack of statistical significance in group comparisons, this estimate should be interpreted cautiously. Furthermore, the initially calculated sample size (n = 10,156) was not reached, and the control group declined substantially during follow-up (from 1,139 to fewer than 150 participants by day 180), limiting the power of comparative analyses. The high baseline seroprevalence (61.9%) in both groups may also have attenuated the measurable effect of vaccination by conferring partial immunity at study entry.

Discussion

Several authors have discussed theoretical and mathematical models for allocating interventions based on risk [19]. Still, this study is the first reported in the literature to be prospectively designed as a risk-based allocation clinical trial. In the original proposal, Finkelstein et al [14] acknowledged that evidence generation to prove the efficacy of an intervention can eventually utilize other controls, such as historical controls, rather than the classical randomized control groups. Therefore, considering low-risk concurrent controls might also be acceptable under certain circumstances, particularly when clinical equipoise is far from a coin flip [20]. In the current situation, where uncontrolled real-world evidence is gaining widespread acceptance, it may also be worthwhile to use risk-based allocation in clinical trials more extensively.

In Brazil, individuals with comorbidities accounted for the majority of COVID-19 hospitalizations [21], and the burden in this group also increased across successive COVID-19 waves [22]. Studies have shown that vaccination provided substantial protection against severe COVID-19 outcomes, especially after a booster dose [23]. This study utilized a risk-based allocation of vaccination priority to individuals with comorbidities, resulting in two distinct groups: those who received the vaccine (high-risk of severe COVID-19 outcomes if infected) and those who did not (low-risk comparator). This approach enabled meaningful comparisons across key outcomes despite the lack of randomization and practical, ethical, social, and logistical restrictions [24,25]. The data demonstrate that vaccinated individuals with risk factors had similar rates of hospitalization and death as unvaccinated individuals without risk factors, suggesting that the vaccine effectively offset the elevated baseline risk associated with comorbidities. These findings align with broader real-world studies of CoronaVac in Chile and Brazil, which reported substantial reductions in severe COVID-19 outcomes during periods of different variant circulation [7,13,26,27].

At baseline (D0), both groups exhibited comparable rates of seropositivity, indicating a high prevalence of prior exposure to SARS-CoV-2 in the study population, which is consistent with findings from surveys conducted in Manaus [28,29]. Despite this initial immunity, the significant increase in anti-Spike IgG levels following vaccination, particularly by day 28 and day 90, demonstrates that CoronaVac successfully boosted humoral responses, even among individuals who were previously exposed. However, a decline in antibody titers was observed by D180, consistent with reports of waning immunity following inactivated vaccines [30,31]. Protective antibody titers decrease about a year after vaccination, and the vaccine does not perform as effectively as the BNT162b2 in providing a booster response against the SARS-CoV-2 Delta and Omicron variants [30,31]. Randomized controlled trials initially showed that the vaccine had 50% and 84% efficacy against symptomatic COVID-19 and hospitalizations due to COVID-19, respectively, among healthcare workers before the emergence of the Gamma variant [25]. Evidence from a test-negative study in Manaus during Gamma variant transmission was mixed, with some protection observed after at least one CoronaVac dose against symptomatic SARS-CoV-2 infection, but low estimated effectiveness after the two-dose schedule [26].

Severe outcomes (hospitalization and death) were similarly rare among vaccinated high-risk and unvaccinated low-risk individuals across all time points. This finding suggests that CoronaVac may effectively mitigate the elevated baseline risk conferred by comorbidities, particularly obesity, and reinforces prior evidence that the vaccine is more effective in preventing severe COVID-19 outcomes than in preventing infection itself [27]. Indeed, the most likely approach to understanding the value of COVID-19 vaccines is to analyze their effectiveness against progression to severe disease [32]. Notably, most individuals in the early vaccination group were classified as obese, a condition known to be strongly associated with severe COVID-19, ICU admission, and mortality [33,34]. Further studies are needed to clarify how varying degrees of obesity influence serodetection and immune responses following COVID-19 vaccination.

This study has some limitations. The low denominator in the low-risk control group, which limited our power to further analysis, is due to the progressive rollout of vaccines to the general population and losses to follow-up. The non-randomized design introduces the potential for confounding, although baseline characteristics between groups were generally balanced. Furthermore, the analysis employed a frequentist approach rather than a Bayesian statistic, as initially proposed for this type of trial [35]. Prior asymptomatic SARS-CoV-2 infections may have influenced outcomes, despite baseline serological screening. The study was also limited to a single urban population of working-age adults. Additionally, viral sequencing was not performed, although local epidemiology strongly suggests that Gamma was the predominant circulating variant during the study period [36,37].

In conclusion, this risk-based allocation clinical trial with CoronaVac demonstrated an apparent reduction in the risk of hospitalization and death among individuals with comorbidities, aligning their risk with that of lower-risk, unvaccinated individuals. Beyond the clinical findings, this design serves as a model for rapidly generating actionable evidence under real-world constraints. It should be considered not only in future pandemic response plans but also as a valid routine strategy to generate evidence of the clinical benefit of an intervention, whether for regulatory or public health purposes, thereby providing the best opportunity to those at higher risk.

Supporting information