Effects of COVID-19 Vaccination Timing and Risk Prioritization on Mortality Rates, United States

During rollout of coronavirus disease vaccination, policymakers have faced critical trade-offs. Using a mathematical model of transmission, we found that timing of vaccination rollout would be expected to have a substantially greater effect on mortality rate than risk-based prioritization and uptake and that prioritizing first doses over second doses may be lifesaving.

I n December 2020, the US government issued emergency use authorization for two 2-dose severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccines, both estimated to be >94% efficacious in preventing symptomatic coronavirus disease (COV-ID-19) (1)(2)(3). The Advisory Committee on Immunization Practices immediately recommended the prioritization of frontline workers and high-risk subgroups (4). As of February 14, 2021, ≈52 million doses have been administered (5). We used a mathematical model of COVID-19 transmission to evaluate the effects of vaccine timing, risk prioritization, number of doses administered, and uptake rates on population-level mortality rates (Figure).
Focusing on Austin, Texas, USA, we projected COVID-19 deaths over 8 months for both an infection-blocking vaccine that prevents infection upon exposure (assuming 95% reduction in susceptibility in vaccinated persons) and a symptom-blocking vaccine that prevents symptoms upon infection (assuming 95% reduction in symptomatic ratio in vaccinated persons). Vaccination would begin on January 15 or February 15, with 10,000 vaccines administered weekly and allocated to cities pro rata. We compare 3 strategies: no priority groups; 1 of 3 priority groups vaccinated before the general public (adults >65 years of age, adults who have high-risk underlying During rollout of coronavirus disease vaccination, policymakers have faced critical trade-offs. Using a mathematical model of transmission, we found that timing of vaccination rollout would be expected to have a substantially greater effect on mortality rate than risk-based prioritization and uptake and that prioritizing first doses over second doses may be lifesaving. conditions, or both); and 10 phases that vaccinate age-risk groups in order of risk for severe COVID-19 outcomes. Stochastic simulations assumed that 7.6% of the overall population of the Austin-Round Rock Metropolitan Statistical Area were immunized by infection before January 15.
These projections validate the prioritizing of highrisk groups. In a pessimistic scenario in which a symptom-blocking vaccine rollout began in February 2021 with 50% uptake, prioritizing high-risk adults and adults >65 would avert ≈17,000 (95% CI 0-36,000) more deaths in the United States than a nonprioritized campaign. Given the state of the pandemic in early 2021, we expected vaccine delays to cost more lives than either imperfect prioritization or vaccine hesitancy.
The United Kingdom and Belgium have prioritized first doses over second doses (7), in an effort to provide partial immunity to more persons. The United States has publicly resisted this approach, citing the lack of clinical trial data validating the approach (8). We found that providing a single (82% efficacious) dose would be expected to save more lives than the corresponding 2-dose strategy, because partially immunizing a large number confers a greater degree of population-level protection than more fully immunizing half as many. Although a 1-dose campaign may accelerate herd immunity and require far fewer resources than a 2-dose campaign, we strongly caution that additional data and single-dose trials are needed to establish efficacy. If the single-dose efficacy is <82%, then we would expect the difference between a single-dose strategy and the corresponding 2-dose strategy to be smaller. We expect similar reductions in mortality rate from both strategies when the singledose efficacy is 52% (2) (Appendix Figure 7). We note that low-efficacy vaccines may increase the risk for vaccine-resistant variants (9) and that there may be political, commercial, and societal barriers to shifting priorities mid-campaign (10).
We assumed that vaccines provide lasting immunity and block either infection or symptoms, whereas the reality may be a hybrid of both (Appendix Table 2, Figure 2), along with riskier behavior stemming from pandemic weariness or overconfidence in the vaccination campaign. Our estimates reflect conditions in the United States in early 2021, as cases were surging toward a pandemic peak in the absence of effective mitigation. The estimated public health benefits of vaccines decrease under higher COVID-19 transmission rates that might occur with relaxed mitigation measures, lower levels of immunity before the rollout, or the emergence of more transmissible SARS-CoV-2 variants including B.1.1.7 (Appendix).
Risk prioritization is a valid approach for maximizing the impact of vaccines, but not at the expense of vaccination speed. Our projections suggest 2 immediate strategies: hybrid distributions that combine active outreach to priority groups with passive distribution to the general public; and distribution of single doses to as much of the population as possible, foregoing plans to hold second doses in reserve. A lthough airborne transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has been proven possible among humans (1), cats (2), ferrets (3), and Syrian hamsters (4), the relative roles of droplets and aerosols in the airborne transmission of SARS-CoV-2 remain controversial. A recent study showed that coronavirus disease (COVID-19) patients exhaled millions of SARS-CoV-2 particles during early infection stages (5). However, the size distribution of SARS-CoV-2 aerosol particles in exhaled breath of COVID-19 patients is not clear.
To analyze size distribution of SARS-CoV-2 aerosols shed by cynomolgus monkeys, we inoculated 3 monkeys with SARS-CoV-2 via a combination of intranasal, intratracheal, and ocular routes. Monkeys were kept in individual cages placed in an isolator (biosafety housing with HEPA filters and independent ventilation system). The exhaled breath and air in the isolator were collected by a 6-stage Andersen sampler (https://tisch-env.com) at postinfection days 2, 4, and 6, and we quantified the viral RNA copies in samples (Appendix, https://wwwnc.cdc.gov/ EID/article/27/7/20-3948-App1.pdf). We also determined size distribution of SARS-CoV-2 particles.
The virus particles monkeys exhaled peaked at postinfection day 2 and ranged from 11,578 to 28,336 RNA copies during a 40-minute period. On average, each monkey exhaled 503 virus particles/min and 209.5 virus particles/L of exhaled breath. At postinfection day 4, the number of exhaled virus particles decreased substantially, ranging from 3,369 to 5,134 RNA copies during a 40-minute period. On average, each monkey exhaled 106 virus particles/min and 44 virus particles/L of breath. At postinfection day 6, no viral RNA was detected in exhaled breath (Figure 1, panel A; Appendix Figure 1). At postinfection days 2, 4, and 6, viral RNA was detected in air within the isolator housing the monkeys; we detected 6,182-13,608 RNA copies during a 30-minute period (Figure 1, panel C).
We measured size distribution of SARS-CoV-2 aerosol particles shed by the monkeys. In exhaled breath of inoculated monkeys and in air in the isolator, viral RNA was detected in all size bins, 0.65-2.1 mm, 2.1-4.7 mm, and >4.7 mm, at postinfection days 2 and 4; most were concentrated in the 2.1-4.7-mm bin ( Figure 1, panels B, D; Appendix Tables 1, 2). For exhaled breath, virus particles in each of the 3 size bins accounted for 27.4%, 49.6%, and 23.0% of the total virus copies/40 min, respectively; for air in the isolator, virus particles in each of the 3 size bins accounted for 3.8%, 75.0%, and 21.2% of the total virus copies/30 min, respectively (Appendix Tables 1, 2; Appendix Figure 3). Most virus particles were in the smaller particle size range (0.65-4.7 mm), accounting for 77% to 79% of the total virus particles shed by the monkeys; droplets (>4.7 mm) accounted for »21%-23% (Appendix Tables 1, 2; Appendix Figure 3). We tried to isolate We analyzed size of severe acute respiratory coronavirus