Of the 34,301 persons initially screened, 184 were screened twice and counted twice. A total of 34,117 unique participants were screened for the trial; 181 persons failed screening twice and were counted twice, and 1661 unique participants failed screening, which does not include 4 persons who were screened and did not undergo randomization but are not included in the number of failed screenings. Of the total 32,451 participants who underwent randomization, 1 participant was enrolled at two separate sites under two subject identification numbers, underwent randomization at both sites, and received both doses of assigned vaccine or placebo. This participant is included once in the all-participants analysis population but is excluded from all other analysis populations. Three participants underwent randomization twice in error. The safety analysis population for each group reflects treatment actually received. The number of participants in each group who received the second dose are those included as of data cutoff. Participants could be excluded from the fully vaccinated analysis population for more than one reason, including for not receiving two doses, and may therefore be counted for exclusion twice. Group assignment was unblinded for 7635 participants (35.3%) in the AZD1222 group and 4157 participants (38.4%) in the placebo group after the second dose. Covid-19 denotes coronavirus disease 2019, RT-PCR reverse transcriptase–polymerase chain reaction, and SARS-CoV-2 severe acute respiratory syndrome coronavirus 2.
Between August 28, 2020, and January 15, 2021, a total of 34,117 unique participants were screened, 32,451 of whom met eligibility criteria and underwent randomization to receive the AZD1222 vaccine (21,635 participants) or placebo (10,816 participants) (Figure 1). The majority of participants were men (55.6%) and had at least one coexisting condition (59.2%); the mean (±SD) age was 50.2±15.9 years (Table 1). Overall, 79.0% of the participants were White, 8.3% were Black, 4.4% were Asian, 4.0% were American Indian or Alaska Native, 2.4% were of multiple races or ethnic groups, 0.3% were Native Hawaiian or other Pacific Islander, and the remainder were of unknown or unreported race or ethnic group. Across both groups, 22.3% of participants were Hispanic or Latinx. Baseline demographic and clinical characteristics were balanced between the trial groups in both the safety analysis population (Table 1) and the fully vaccinated analysis population (Table S1 in the Supplementary Appendix). A total of 347 participants (1.6%) in the AZD1222 group and 169 (1.6%) in the placebo group were living with well-controlled human immunodeficiency virus infection.
The incidence of adverse events is shown in Table S2. A total of 11,972 participants (37.0%) — 8771 (40.6%) in the AZD1222 group and 3201 (29.7%) in the placebo group — reported 23,538 adverse events. The most common adverse events, occurring in at least 5% of participants within 28 days after any dose in either group, were general pain (8.2% in the AZD1222 group and 2.3% in the placebo group), headache (6.2% and 4.6%, respectively), injection-site pain (6.8% and 2.0%), and fatigue (5.1% and 3.5%).
A similar percentage of participants in each group had a serious adverse event within 28 days after any dose: 119 serious adverse events occurred among 101 participants (0.5%) in the AZD1222 group and 59 events among 53 participants (0.5%) in the placebo group. During the entire trial period, a total of 7 adverse events leading to death occurred in 7 participants in the AZD1222 group, and 9 adverse events leading to 7 deaths occurred in the placebo group. These deaths are described in Table S2. No deaths were considered by investigators to be related to the vaccine or placebo. No deaths related to Covid-19 occurred in the AZD1222 group, and two deaths related to Covid-19 occurred in the placebo group.
Medically attended adverse events and adverse events of special interest within 28 days after a dose also occurred in similar proportions in the two groups (Table S2). The incidences of individual adverse events related to the vaccine or placebo during the entire trial period are shown in Tables S3 through S5. The incidence of potential immune-mediated conditions was similar in the two groups (1.8% in the AZD1222 group and 3.4% in the placebo group), as were the incidences of adverse events of special interest: neurologic (0.5% in the AZD1222 group and 0.4% in the placebo group), vascular (0.1% in the AZD1222 group and <0.1% in the placebo group), and hematologic (<0.1% in both groups). Specifically, the incidences of deep-vein thrombosis (<0.1% in both groups), pulmonary embolism (<0.1% in both groups), thrombocytopenia (<0.1% in the AZD1222 group and none in the placebo group), and immune thrombocytopenia (none in the AZD1222 group and <0.1% in the placebo group) were low and similar in the groups. There were no cases in either group of thrombosis with thrombocytopenia, cerebral venous sinus thrombosis, or venous thrombosis in unusual locations.
Erythema and induration were classified by size as mild (2.5 to 5 cm), moderate (5.1 to 6 cm), or moderate-to-severe (>6 cm). Fevers were graded by temperature as none (≤37.8°C), mild (37.9 to 38.4°C), moderate (38.5 to 38.9°C), severe (39.0 to 40.0°C), or life threatening (≥40.1°C). The most common solicited adverse events that occurred in at least 5% of participants within 7 days after any dose in either group were tenderness (68.4% in the AZD1222 group and 19.0% in the placebo group) and pain (58.3% and 15.7%), both local adverse events; the most common systemic adverse events were headache (50.2% in the AZD1222 group and 35.5% in the placebo group), fatigue (49.7% and 31.2%), muscle pain (41.9% and 19.5%), malaise (35.0% and 17.0%), chills (28.2% and 9.5%), nausea (15.3% and 12.1%), and temperature higher than 37.8°C (7.0% and 0.6%). The All Ages group included 1013 participants for dose 1, placebo; 968 for dose 2, placebo; 2037 for dose 1, AZD1222; and 1962 for dose 2, AZD1222. The age 18 to 64 group included 663 participants for dose 1, placebo; 629 for dose 2, placebo; 1339 for dose 1, AZD1222; and 1288 for dose 2, AZD1222. The age 65 and older group included 350 participants for dose 1, placebo; 339 for dose 2, placebo; 698 for dose 1, AZD1222; and 674 for dose 2, AZD1222.
In the substudy population, more participants in the AZD1222 group than in the placebo group had local solicited adverse events (74.1% in the AZD1222 group vs. 24.4% in the placebo group) and systemic solicited adverse events (71.6% vs. 53.0%) (Figure 2). The majority of solicited adverse events (92.6%) across both groups were mild or moderate in intensity. Events occurred less frequently after the second dose than after the first dose in both age groups, a difference that was more marked in participants 18 to 64 years of age. The majority of local and systemic solicited adverse events resolved within 1 to 2 days after onset.
Values shown for no. of events/total no. are the number of events that occurred among the participants within each group and do not account for censoring due to unblinding of group assignment or loss to follow-up. The primary efficacy end point is the first case of SARS-CoV-2 RT-PCR–positive symptomatic illness occurring 15 days or more after the second dose of AZD1222 or placebo among participants with negative serostatus at baseline. Vaccine efficacy is shown with 95% confidence intervals (CIs), except for vaccine efficacy values for the primary efficacy end point according to SARS-CoV-2 baseline serostatus, the secondary end point of severe or critical symptomatic Covid-19, and the exploratory end point of Covid-19–related intensive care unit (ICU) admissions, which are based on a one-sided 97.5% CI calculated with the exact Poisson model, owing to nonconvergence of the Poisson regression with robust variance. Race and ethnic group were reported by the participant. Other denotes participants who provided a race or ethnic group identification other than White, Black, or American Indian or Alaska Native. Key secondary end points were incidence of symptomatic illness (at 15 days or more after the second dose of AZD1222 or placebo) regardless of evidence of previous SARS-CoV-2 infection at baseline, severe or critical symptomatic Covid-19 (at 15 days or more after the second dose of AZD1222 or placebo), Covid-19–related emergency department visits, symptomatic Covid-19 as defined by Centers for Disease Control and Prevention (CDC) criteria, and first response (change from negative serostatus for SARS-CoV-2 nucleocapsid antibodies at baseline to positive serostatus after receiving AZD1222 or placebo). P values are reported for the primary and key secondary outcomes; analyses followed prespecified plan to adjust for multiple comparisons. I bars indicate confidence intervals; arrows indicate truncated values, with actual values shown in the accompanying column; the dashed vertical line represents the upper limit (i.e., 100% vaccine efficacy); and the solid vertical line represents the nominally statistically significant criterion of a lower confidence interval greater than 30% applicable to the primary end point and is shown for reference. NA denotes not available, and NE could not be estimated.
The time to the first event was relative to the time of the actual second dose administration, calculated as (date of SARS-CoV-2–positive test) – (date of second dose of AZD1222 or placebo + 14 days) + 1. For participants whose data were censored, the censoring time was from the date of the second dose of AZD1222 or placebo + 14 days to the last time observed before data cutoff (March 5, 2021). The cumulative incidence of Covid-19 was estimated with the Kaplan–Meier method. Vaccine efficacy, estimated on the basis of the supportive analysis of the time to primary efficacy end point with the use of the Cox proportional-hazards model, with the randomization and age groups at the time of informed consent as covariates, was 73.9% (95% CI, 65.3 to 80.5). Tick marks indicate censored data.
Once adjudication of all events that occurred before the data cutoff was complete, 203 symptomatic Covid-19 events met the case definition of the primary end point and were included in the updated primary analysis for the fully vaccinated analysis population (17,662 participants in the AZD1222 group and 8550 in the placebo group) (Figure 1). The efficacy analyses presented here are based on the updated primary analysis of the group whose data were censored as of the cutoff date. In the full analysis population, the median follow-up duration from the second dose to the data cutoff date, regardless of unblinding of group assignments, was 61.0 days (range, 1 to 129) in both groups (Table 1). Overall, 73 events (0.4%) occurred in the AZD1222 group and 130 (1.5%) occurred in the placebo group (Figure 3). For the primary efficacy end point, the success criterion was met in the fully vaccinated analysis population on the basis of an overall vaccine efficacy estimate of 74.0% (95% confidence interval [CI], 65.3 to 80.5; P<0.001). Results regarding the cumulative incidence of the first SARS-CoV-2 RT-PCR–positive symptomatic illness after the second dose of AZD1222 (Figure 4) showed that the effect of AZD1222 began soon after the second dose. Vaccine efficacy was consistent in the analyses in which follow-up data were not censored at unblinding of the treatment assignment or EUA vaccination (74.3%; 95% CI, 66.0 to 80.6) and also when multiple imputation was used (73.3%; 95% CI, 64.6 to 79.9).
On September 9, 2020, the trial was placed on clinical hold owing to an event of transverse myelitis reported in a different AZD1222 clinical study.2 After a review of the event and all available safety data, the Food and Drug Administration lifted the clinical hold on October 23, 2020, and the trial resumed on October 28, 2020. A total of 775 participants (2.4%) in the safety analysis population were affected by the clinical hold and received their second dose outside the planned 28-day window. Vaccine efficacy in this subgroup of participants who received their second dose at an extended dosing interval was consistent with that in the overall group (78.1%; 95% CI, 49.2 to 90.6).
Vaccine efficacy estimates according to subgroup are shown in Figure 3, although small case numbers hindered confidence in some subgroup estimates, such as those for ICU admissions and those based on data from participants in Chile and Peru. Estimated vaccine efficacy was high against symptomatic illness in participants 18 to 64 years of age (72.8%; 95% CI, 63.4 to 79.9) and those 65 years of age or older (83.5%; 95% CI, 54.2 to 94.1) and was consistent across participants of different races and ethnic groups, status with respect to coexisting conditions, baseline SARS-CoV-2 serostatus, and sex. In Chile, 4 cases of symptomatic illness were noted among 1360 participants in the AZD1222 group as compared with 2 cases among 672 participants in the placebo group. In Peru, 11 cases among 867 participants in the AZD1222 group and 9 cases among 435 participants in the placebo group were observed. Estimated vaccine efficacy against symptomatic Covid-19 regardless of evidence of previous SARS-CoV-2 infection (a secondary end point) was 73.7% (95% CI, 65.1 to 80.1; P<0.001).
The vaccine was significantly effective against all other key secondary efficacy end points (Figure 3). In the fully vaccinated analysis population, no cases of severe or critical symptomatic Covid-19 were observed among the 17,662 participants in the AZD1222 group, as compared with 8 cases (<0.1%) among the 8550 participants in the placebo group. Estimated vaccine efficacy of AZD1222 for the prevention of Covid-19 (as defined by CDC criteria) was high (69.7%; 95% CI, 60.7 to 76.6; P<0.001), as was efficacy against emergency department visits attributed to Covid-19 (94.8%; 95% CI, 59.0 to 99.3; P=0.005), with 1 (<0.1%) emergency department visit in the AZD1222 group and 9 (0.1%) in the placebo group. Estimated vaccine efficacy against Covid-19–related hospitalizations (an exploratory end point) was 94.2% (95% CI, 53.3 to 99.3) (Figure 3). One participant in the AZD1222 group who had a Covid-19–related emergency department visit had an allergic reaction to a monoclonal antibody treatment and was hospitalized. This hospitalization did not meet the criteria for severe or critical Covid-19.
The estimated vaccine efficacy for incidences of first SARS-CoV-2 RT-PCR–positive symptomatic illness occurring after the first dose of AZD1222 or placebo is described in Figure S1. AZD1222 was efficacious at preventing infection with SARS-CoV-2, as measured by nucleocapsid antibody seroconversion 15 days or more after the second dose; this included all participants who tested positive for SARS-CoV-2 nucleocapsid antibodies regardless of symptoms or severity (64.3%; 95% CI, 56.1 to 71.0; P<0.001). Additional details of the efficacy analyses are provided in the Supplementary Appendix.
Participants who received AZD1222 and were seronegative at baseline showed strong vaccine-induced serum IgG responses to the spike protein (Fig. S2). Levels of neutralizing antibodies were higher than baseline at all time points in the AZD1222 group, increasing further after a second dose, but remained low throughout the trial in the placebo group (Fig. S3).
Whole-Genome Sequencing of SARS-CoV-2 Samples
Among participants in the full analysis population (the 30,889 participants who were seronegative at baseline), whole-genome sequencing of saliva samples obtained from 176 participants in the AZD1222 group and 183 participants in the placebo group attending illness visits, regardless of qualifying symptoms, yielded four cases of variants of concern, including alpha and beta variants (one putative B.1.351 case was determined by clade). Of the variants of interest observed, epsilon was the most common (B.1.429 in 14 participants and B.1.427 in 3 participants) followed by iota (B.1.526 in 1 participant) (Table S6).