Main

In a randomized clinical trial, the mRNA-1273 (Moderna) vaccine had a reported efficacy of 94.1% in preventing symptomatic coronavirus disease 2019 (COVID-19) due to infection with wild-type variants1. The first immunization using this vaccine in Qatar was recorded on 28 December 2020, but mass vaccination did not start until late February 2021 with the accelerated arrival of vaccine shipments. As the vaccination campaign was scaled up, the country experienced two back-to-back severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) waves that were triggered by the introduction and expansion of the B.1.1.7 (Alpha2) and B.1.351 (Beta2) variants3. This created a unique epidemiological opportunity to assess the real-world effectiveness of this vaccine against infection with these variants of concern, as well as against severe forms of COVID-19 disease.

Results

Study population

Between 28 December 2020 and 10 May 2021, 256,037 individuals in Qatar received at least one dose of the mRNA-1273 vaccine and 181,304 completed the two-dose regimen. Among those with a vaccination record, the median dates at first and second dose were 5 and 29 April 2021, respectively. The median time elapsing between the first and second doses was 28 days (interquartile range (IQR): 28–29 days), with 94.6% of individuals receiving their second dose ≤30 days after the first dose.

Flowcharts describing the population selection process for investigation and estimation of vaccine effectiveness are presented in Extended Data Figs. 13, and the demographic characteristics of the sample for each outcome of vaccine effectiveness are presented in Table 1. Median age in the study sample for estimation of vaccine effectiveness against infection with B.1.1.7 was 31 years (IQR: 19–38), against infection with B.1.351 was 32 years (IQR: 25–39) and against severe, critical or fatal disease was 43 years (IQR: 37–51). Of note is that Qatar has a young and diverse demographic, where only 9% of its resident population are >50 years of age and 89% are expatriates from >150 countries4,5.

Table 1 Demographic characteristics of (1) B.1.1.7 cases (PCR-positive) and controls (PCR-negative), (2) B.1.351 cases (PCR-positive) and controls (PCR-negative) and (3) severe, critical or fatal COVID-19 disease cases (any PCR-positive case) and controls (PCR-negative)

Weekly rounds of viral genome sequencing from 8 March to 10 May 2021 identified B.1.351 (n = 369; 64.4%), B.1.1.7 (n = 58; 10.1%), B.1.617 (n = 18; 3.1%) and wild-type/undetermined variants (n = 128; 22.3%) in 573 randomly collected polymerase chain reaction (PCR)-positive specimens. Weekly rounds of multiplex quantitative PCR with reverse transcription (RT–qPCR)-based variant screening6 from 23 March to 10 May 2021 identified B.1.351-like (n = 2,605; 66.4%), B.1.1.7-like (n = 970; 24.7%) and ‘other’ variants (n = 349; 8.9%) in 3,924 randomly collected PCR-positive specimens. Sanger sequencing of the receptor-binding domain of SARS-CoV-2 spike protein on 109 other specimens confirmed that 103 were B.1.617-like, three were B.1-like and three were undetermined.

As of 10 May 2021, 1,558 and 243 breakthrough infections were recorded among those who had received one or two doses of the mRNA-1273 vaccine, respectively. Seven COVID-19 deaths were also recorded among mRNA-1273 vaccine recipients, all after the first dose and none after the second.

Vaccine effectiveness against infection with B.1.1.7

Estimated effectiveness against infection with B.1.1.7, defined as a PCR-positive swab with the B.1.1.7 variant, regardless of the reason for PCR testing or presence of symptoms (Methods), was negligible for 2 weeks after the first dose (Table 2 and Fig. 1), but increased rapidly in the third week reaching 81.6% (95% CI: 73.1–87.8%). Effectiveness was 94.4% (95% CI: 89.1–97.5%) in the fourth week immediately before the second dose, and reached 99.2% (95% CI: 95.3–100.0%) in the second week after the second dose. Effectiveness was 88.1% (95% CI: 83.7–91.5%) 14 or more days after the first dose but before receiving the second dose, and was 100% (95% CI: 91.8–100.0%) 14 or more days after the second dose (Table 3).

Table 2 Effectiveness of the mRNA-1273 vaccine against B.1.1.7 and B.1.351 infections and against severe, critical or fatal COVID-19 disease, week by week during the first 6 weeks after the first dose
Fig. 1: Effectiveness of the mRNA-1273 vaccine.
figure 1

ac, Effectiveness of the mRNA-1273 vaccine against B.1.1.7 infections (a), B.1.351 infections (b) and severe, critical or fatal COVID-19 disease (c) after the first and second doses. Analyses were performed on independent samples of n = 25,034 PCR-positive cases and n = 25,034 PCR-negative controls examined between 1 February and 10 May 2021 for B.1.1.7, n = 52,442 PCR-positive cases and n = 52,442 PCR-negative controls examined between 8 March and 10 May 2021 for B.1.351, n = 4,497 PCR-positive cases that progressed to severe, critical or fatal disease and n = 4,497 PCR-negative controls examined between 1 February and 10 May 2021. Data are presented as effectiveness point estimates, with error bars indicating the corresponding 95% CIs.

Table 3 Effectiveness of the mRNA-1273 vaccine against B.1.1.7 and B.1.351 infections and against severe, critical or fatal COVID-19 disease ≥14 days after the first dose and ≥14 days after the second dose

Vaccine effectiveness against infection with B.1.351

Effectiveness against infection with B.1.351, defined as a PCR-positive swab with the B.1.351 variant regardless of the reason for PCR testing or presence of symptoms (Methods), was negligible for 2 weeks after the first dose (Table 2 and Fig. 1) but increased rapidly in the third week to reach 47.9% (95% CI: 39.5–55.2%). Effectiveness was 73.7% (95% CI: 67.6–78.8%) in the fourth week immediately before the second dose, and reached 96.4% (95% CI: 94.3–97.9%) in the second week after the second dose. Effectiveness was 61.3% (95% CI: 56.5–65.5%) 14 or more days after the first dose but before receiving the second dose, and was 96.4% (95% CI: 91.9–98.7%) 14 or more days after the second dose (Table 3).

Vaccine effectiveness against COVID-19 severity or fatality

Effectiveness against any severe, critical or fatal disease due to any SARS-CoV-2 infection (predominantly B.1.1.7 and B.1.351 (ref. 7); Methods) was negligible for 2 weeks after the first dose (Table 2 and Fig. 1) but increased rapidly in the third week to reach 70.3% (95% CI: 48.9–83.5%). Effectiveness was 92.1% (95% CI: 78.4–97.9%) in the fourth week immediately before the second dose, and reached 100% (95% CI: 86.9–100.0%) in the second week after the second dose. Effectiveness was 81.6% (95% CI: 71.0–88.8%) 14 or more days after the first dose but before receiving the second dose, and was 95.7% (95% CI: 73.4–99.9%) 14 or more days after the second dose (Table 3).

Additional analyses

Sensitivity analyses matching by PCR testing date in addition to age, sex, nationality and reason for PCR testing; adjusting for calendar week in logistic regression; or additionally adjusting for matching factors—that is, sex, age, nationality and reason for PCR testing, all confirmed the main analysis results (Table 4).

Table 4 Sensitivity analyses for effectiveness of the mRNA-1273 vaccine against B.1.1.7 and B.1.351 infections and against severe, critical or fatal COVID-19 disease, after (1) matching by sex, age, nationality, reason for PCR testing and PCR test date, (2) adjusting for calendar week in logistic regression analysis and (3) adjusting for sex, age, nationality, reason for PCR testing and calendar week in logistic regression analysis

An additional analysis estimated vaccine effectiveness against symptomatic infection (predominantly B.1.1.7 and B.1.351 (ref. 7)) at 98.6% (95% CI: 92.0–100.0%) 14 or more days after the second dose (Supplementary Table 1). Symptomatic infection was defined as a PCR-positive test conducted because of clinical suspicion due to the presence of symptoms compatible with a respiratory tract infection.

An additional analysis estimated vaccine effectiveness against asymptomatic infection (predominantly B.1.1.7 and B.1.3517) at 92.5% (95% CI: 84.8–96.9%) 14 or more days after the second dose (Supplementary Table 1). Asymptomatic infection was defined as a PCR-positive test conducted with no reported presence of symptoms compatible with a respiratory tract infection—that is, PCR testing was done as part of a survey, for pretravel requirements or at a port of entry following arrival in the country4,8.

An additional analysis estimated vaccine effectiveness 14 or more days after the second dose using a cohort study design that compares infection incidence in those vaccinated with that in the national cohort of persons who were antibody negative (Methods, Supplementary Table 2 and Extended Data Fig. 4). Incidence rates among those vaccinated were estimated at 0 (95% CI: 0–3.60) for B.1.1.7, 5.83 (95% CI: 2.62–12.99) for B.1.351 and 1.94 (95% CI: 0.49–7.78) for variants of unknown status per 10,000 person-weeks in a total follow-up time of 10,282.86 person-weeks. The incidence rate among those antibody negative was estimated at 20.49 (95% CI: 19.40–21.63) for B.1.1.7, 47.78 (95% CI: 46.11–49.52) for B.1.351 and 30.12 (95% CI: 28.79–31.50) for variants of unknown status per 10,000 person-weeks in a total follow-up time of 631,171.10 person-weeks. Vaccine effectiveness was thus estimated at 100.0% (95% CI: 82.5–100.0%) against B.1.1.7, 87.8% (95% CI: 73.4–95.5%) against B.1.351 and 93.5% (95% CI: 76.6–99.2%) against variants of unknown status, supporting the results from the main analysis.

Discussion

Our analysis demonstrates that the effectiveness of the mRNA-1273 vaccine is high, regardless of variant, against both symptomatic and asymptomatic infection, and against COVID-19 hospitalization and death. Development of vaccine protection against B.1.1.7 infection and COVID-19 hospitalization and death rapidly accelerated in the third and fourth weeks after the first dose, immediately before the second dose, nearly reaching the values attained after the second dose. Protection against B.1.351 infection exhibited a similar pattern, but mounted at a slower rate and did not achieve its highest value of ~95% until after the second dose. While long-term protection of only one dose could not be assessed beyond 4 weeks after the first dose, these findings might suggest that most of the protection from this vaccine is attained using only one dose, apart possibly from its protection against B.1.351. Although for optimal protection the protocol for this vaccine requires a second dose 28 days after the first, given the sizable protection achieved after only one dose, in situations involving limited vaccine supplies, these findings may support a strategy of delaying the second dose to vaccinate the largest number of people in the shortest time, as supported also by evidence for the BNT162b2 (Pfizer–BioNTech) and ChAdOx1 nCoV-19 (AZD1222; Oxford–AstraZeneca) vaccines9,10,11. Having said that, the optimal timing of the second dose is best determined formally through randomized clinical trials and could also depend on both the incidence level and circulating variants, which are factors that can vary from one setting to another.

The mRNA-1273 vaccine demonstrated similar levels and patterns of protection after the first and second doses to the BNT162b2 vaccine, as estimated recently in Qatar3,11. However, in our study, the mRNA-1273 vaccine appears to offer greater protection against both B.1.351 and B.1.1.7. It is premature to conclude that this outcome is due to a superior vaccine-induced immune response, as other confounding variables may also have contributed to observed differences. The second dose of mRNA-1273 is administered 28 days after the first, 1 week later than for BNT162b2, which might have affected the build-up of immunity and estimated effectiveness. We also note that since the mass immunization campaigns in Qatar started with BNT162b2 in December 2020, most older persons and those with comorbid conditions were vaccinated with the BNT162b2 vaccine before the mRNA-1273 vaccine became widely available in March–May 2021. Those vaccinated with the mRNA-1273 vaccine (Table 1) tend to be slightly younger than those vaccinated with the BNT162b2 vaccine3.

Effectiveness in our study was assessed based on documented infections, but other infections may have occurred and gone undocumented perhaps because of minimal and/or mild symptoms or by virtue of being subclinical infections. However, with the high rate of PCR testing in Qatar, 64.2% of infections diagnosed during the study period were identified not because of clinical symptoms but for other reasons, including random testing campaigns (surveys), contact tracing, individual requests and routine healthcare- and travel-related testing. This indicates that the vaccine was robustly effective not only against symptomatic clinical disease, but also against asymptomatic and subclinical infection. That said, there was evidence for a slight gradient in vaccine effectiveness by appearance of symptoms—that is, greater protection against more symptomatic or severe infections, as observed for the BNT162b2 vaccine3,11.

This study has certain limitations. Estimated effectiveness against severe, critical or fatal COVID-19 disease had somewhat wider 95% confidence intervals as a consequence of (1) the relatively lower severity of COVID-19 in the young population of Qatar compared to populations where the median age is higher4,12, (2) the younger cohort of those vaccinated with mRNA-1273 and (3) the majority of vaccinated persons having had their second dose after the B.1.1.7 and B.1.351 waves peaked in early March and early April 2021, respectively. The median date of receiving the second dose in our study population was 29 April 2021. For the same reasons, we could not precisely estimate effectiveness against severe, critical or fatal COVID-19 disease attributable individually to the B.1.1.7 and B.1.351 variants.

Severe, critical or fatal COVID-19 disease was assessed up to the end of the study (10 May 2021), and it is possible that cases confirmed as PCR positive immediately before the end of the study may have progressed to a severe outcome only after the end of the study and were thus not included in our sample. This, however, should not affect our estimates because the study, by design, contrasts confirmed COVID-19 disease cases with PCR-negative cases. For reasons that remain unclear, among persons 7–13 days after the first dose the risk of infection with B.1.351 was higher compared to those who remained unvaccinated, an outcome observed elsewhere for both COVID-19 vaccines13,14,15 and other vaccines16. This might reflect a higher underlying risk of infection, bias due to uncontrolled confounding such as differences in social behavior at or following vaccination, an immunological effect16 or an artifact of the estimation method, possibly because the first vaccine dose frequently coincided with the peak of the B.1.351 wave.

Imperfect assay sensitivity and specificity of PCR testing may also have affected infection ascertainment. However, all PCR testing was performed with extensively used, investigated and validated commercial platforms having essentially 100% sensitivity and specificity (Methods). The baseline analysis did not factor matching or control for calendar time, but sensitivity analyses factoring calendar time confirmed the main findings. Data on comorbid conditions were not available and hence could not be factored explicitly in our analysis. However, adjustment for age may have served as a proxy given that comorbidities are associated with old age. Furthermore, given the young population structure in Qatar4,5 we anticipate that only a small proportion of the study population may have had serious comorbid conditions. However, our findings might not be entirely generalizable to other settings where the elderly constitute a sizable proportion of the population. Effectiveness was assessed using an observational test-negative, case-control study design17,18 rather than a randomized clinical trial, but the cohort study design also yielded similar findings, supporting the validity of this approach in assessing vaccine effectiveness. In some instances there were zero events among cases and/or controls, thus precluding estimation of effectiveness and/or associated confidence intervals.

In conclusion, in the population of Qatar the mRNA-1273 vaccine is very effective against both B.1.1.7 and B.1.351 infection, whether symptomatic or asymptomatic, and against COVID-19 hospitalization and death. Most of the protection of this vaccine accrued in the third and fourth weeks after the first dose and before the second dose, although its effectiveness against B.1.351 increased substantially more after the second dose. These findings justify optimism that vaccine-induced immunity will prove robust against known variants of concern, thereby reducing the likelihood of a protracted pandemic for years to come.

Methods

Data sources and study design

This study was conducted in the resident population of Qatar. COVID-19 laboratory testing, vaccination, clinical infection data and related demographic details were extracted from the integrated nationwide digital health information platform that hosts the national, federated SARS-CoV-2 databases. These databases are complete and have captured all SARS-CoV-2-related data since epidemic onset. Nearly all individuals were vaccinated (free of charge) in Qatar and not elsewhere. In rare situations where an individual had received the mRNA-1273 COVID-19 vaccine outside Qatar, the individual’s vaccination details were still recorded in the health system at the port of entry (airport) following their return to Qatar given the national requirements and to benefit from privileges associated with vaccination, such as quarantine exemption.

Vaccine effectiveness was estimated using the test-negative, case-control study design, a standard design for assessment of vaccine effectiveness against influenza17,18. Key to this design is the control of bias arising from misclassification of infection and differences in healthcare-seeking behavior between vaccinated and unvaccinated individuals17,18. The STROBE checklist can be found in Supplementary Table 3. Cases and controls were matched one-to-one by sex, age, nationality and reason for SARS-CoV-2 PCR testing. Matching of cases and controls was performed to control for known differences in risk of exposure to the infection in Qatar4,25,26,27.

Effectiveness was estimated against documented infection (defined as a PCR-positive swab regardless of the reason for PCR testing or presence of symptoms) with the B.1.1.7 or B.1.351 variant, as well as against severe, critical or fatal disease due to any SARS-CoV-2 infection (predominantly B.1.1.7 and B.1.351 (ref. 7)). Classification of COVID-19 case severity (acute care hospitalizations)23, criticality (intensive care unit hospitalizations)23 and fatality24 followed the World Health Organization guidelines, and assessments were made by trained medical personnel using individual chart reviews.

Classification of infections by variant type was informed by weekly rounds of viral genome sequencing and multiplex RT–qPCR variant screening6 of randomly collected clinical samples7, as well as by results of deep sequencing of wastewater samples7. Based on existing evidence20,21,22, a B.1.1.7 case was defined as an S-gene target failure using TaqPath COVID-19 Combo Kits (Thermo Fisher Scientific19; >85% of PCR testing in Qatar), applying the criterion of a PCR cycle threshold (Ct) value ≤30 for both the N and ORF1ab genes and a negative outcome for the S gene22. Meanwhile, with essentially only B.1.351 and B.1.1.7 cases identified from 8 March 2021 until the end of the study (10 May 2021) in the viral genome sequencing and multiplex RT–qPCR variant screening7, a B.1.351 case was proxied as the complement of B.1.1.7 criteria—that is, any infection with a Ct value ≤30 for the N, ORF1ab and S genes3.

With these considerations in regard to ascertaining infection by B.1.1.7 and B.1.351, the study extended from 1 February to 10 May 2021 for B.1.1.7, from 8 March to 10 May 2021 for B.1.351 and from 1 February to 10 May 2021 for any severe, critical or fatal COVID-19 disease. All records of PCR testing for those vaccinated and unvaccinated during the study duration were examined. All records of vaccination with one or two doses using a vaccine other than mRNA-1273 were excluded. Every case that met the inclusion criteria (a B.1.1.7 case, a B.1.351 case or a severe or critical or fatal COVID-19 disease case) and that could be matched to a control was included in the analysis. Both PCR test outcomes and vaccination status were ascertained at the time of the PCR test. Each person with a positive PCR test result and hospital admission was subject to an infection severity assessment every 3 days until discharge or death. Individuals who progressed to COVID-19 disease between the time of a positive PCR test result and the end of the study were classified based on their worst outcome, starting with death24, followed by critical disease23 and then severe disease23.

The study was approved by the Hamad Medical Corporation and Weill Cornell Medicine–Qatar Institutional Review Boards with waiver of informed consent.

Laboratory methods

Nasopharyngeal and/or oropharyngeal swabs (Huachenyang Technology) were collected for PCR testing and placed in Universal Transport Medium (UTM). Aliquots of UTM were extracted on a QIAsymphony platform (QIAGEN) and tested with RT–qPCR using TaqPath COVID-19 Combo Kits (100% sensitivity and specificity19; Thermo Fisher Scientific) on an ABI 7500 FAST (ThermoFisher), extracted using a custom protocol28 on a Hamilton Microlab STAR (Hamilton) and tested using AccuPower SARS-CoV-2 Real-Time RT-PCR Kits (100% sensitivity and specificity29; Bioneer) on an ABI 7500 FAST; or loaded directly into a Roche cobas 6800 system and assayed with a cobas SARS-CoV-2 Test (95% sensitivity, 100% specificity30; Roche). The first assay targets the viral S, N and ORF1ab regions, the second targets the viral RdRp and E-gene regions while the third targets the ORF1ab and E-gene regions.

All tests were conducted at either the HMC Central Laboratory or Sidra Medicine Laboratory, following standardized protocols.

Statistical analysis

Descriptive statistics (frequency distributions and measures of central tendency) were used to characterize study samples. The odds ratios, comparing odds of vaccination among cases to that among controls, and its associated 95% CI, were calculated using the exact method. Confidence intervals were not adjusted for multiplicity. Interactions were not investigated. Vaccine effectiveness at different time frames and its associated 95% CI were then calculated by applying the following equation17,18:

$$\begin{array}{l}{\mathrm{Vaccine}}\,{\mathrm{effectiveness}}\\ = 1 - \frac{{{\mathrm{vaccinated}}\,{\mathrm{among}}\,{\mathrm{cases}} \times {\mathrm{unvaccinated}}\,{\mathrm{among}}\,{\mathrm{controls}}}}{{{\mathrm{vaccinated}}\,{\mathrm{among}}\,{\mathrm{controls}} \times {\mathrm{unvaccinated}}\,{\mathrm{among}}\,{\mathrm{cases}}}}\end{array}$$

Additional analyses

To ensure that vaccine effectiveness estimates were not biased by epidemic phase and the gradual rollout of vaccination during the study, two sensitivity analyses were conducted, first matching by the exact PCR testing date and second by logistic regression to adjust for calendar week17,31. To further ensure control for confounding32,33, a third sensitivity analysis was conducted adjusting additionally for matching factors in logistic regression—that is sex, age, reason for PCR testing and nationality (grouping was based on the most populous nationality groups in Qatar; Tables 1 and 4).

Additional analyses were conducted to estimate vaccine effectiveness against symptomatic infection, defined as a PCR-positive test conducted because of clinical suspicion due to the presence of symptoms compatible with a respiratory tract infection, and against asymptomatic infection, defined as a PCR-positive test conducted with no reported presence of symptoms compatible with a respiratory tract infection—that is, PCR testing was done as part of a survey, for pretravel requirement or at the port of entry into the country (Supplementary Table 1).

Vaccine effectiveness was further assessed using a cohort study design3,34,35 by comparison of infection incidence in those who completed ≥14 days after the second dose with incidence in the national cohort of individuals who were antibody negative. The incidence rate of infection was calculated for each of the variants using a Poisson log-likelihood regression model with the STATA 16.1 (ref. 36) stptime command. The incidence rate ratio was calculated using the exact method. Vaccine effectiveness (with no adjustment for multiplicity) was estimated using the equation:

$$\begin{array}{l}{\mathrm{Vaccine}}\,{\mathrm{effectiveness}}\\ = 1 - \frac{{{\mathrm{incidence}}\,{\mathrm{rate}}\,{\mathrm{of}}\,{\mathrm{infection}}\,{\mathrm{among}}\,{\mathrm{}}\,{\mathrm{vaccinated}}\,{\mathrm{individuals}}}}{{{\mathrm{incidence}}\,{\mathrm{rate}}\,{\mathrm{of}}\,{\mathrm{infection}}\,{\mathrm{among}}\,{\mathrm{}}\,{\mathrm{antibody-negative}}\,{\mathrm{individuals}}}}\end{array}$$

Further details on this type of analysis and the antibody-negative cohort can be found in our previous studies of reinfection34,35.

Analyses were conducted in STATA/SE 16.1 (ref.36).

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.