Comparative effectiveness over time of the mRNA-1273 (Moderna) vaccine and the BNT162b2 (Pfizer-BioNTech) vaccine

With the emergence of variants of concern and reports of post-vaccination infections, evaluating real-world comparative effectiveness of Covid-19 vaccines is critical. Although both mRNA-1273 (Moderna) and BNT162b2 (Pfizer-BioNTech) have high efficacy against severe disease, evidence suggests mRNA-1273 elicits higher humoral antibody responses than BNT162b2 across age groups and prior infection status, and antibody levels correlate with protection. Observational reports (including during Delta variant circulation) indicate higher breakthrough infection rates after BNT162b2 compared with mRNA-1273. This study aims to compare the incidence and severity of SARS-CoV-2 infection following complete vaccination with either vaccine in a large, diverse, real-world U.S. cohort, with primary outcomes focused on infection rates at 30, 60, and 90 days post full vaccination.

Prior work shows that two doses of mRNA-1273 produce higher humoral antibody responses than two doses of BNT162b2 in both previously uninfected and infected individuals, and higher antibody titers correlate with enhanced protection against Covid-19. Observational studies during periods including Delta variant circulation reported higher infection rates among BNT162b2 recipients compared with mRNA-1273 recipients, suggesting potential differences in real-world performance of these vaccines.

Design: Retrospective cohort study following STROBE guidelines, using de-identified administrative claims for Medicare Advantage and commercially insured individuals in a research database. Study period for vaccination and follow-up: January 1 through May 31, 2021, with outcomes assessed from 14 to 151 days after the second dose. Population: Adults ≥18 years fully vaccinated (two doses) with mRNA-1273 or BNT162b2 on or before May 31, 2021. Exclusions included second doses administered outside EUA-recommended intervals (earlier than 21 days for BNT162b2 or 28 days for mRNA-1273, or >42 days after first dose). Individuals with Covid-related events within 13 days after second dose and no events after day 14 were excluded for analytic datasets. Sub-analyses stratified by age ≥65 vs ≤64 and by no prior Covid-19 diagnosis. Outcomes: Primary—SARS-CoV-2 infection defined by documented care-seeking encounters (positive PCR/diagnosis in ER, outpatient, or inpatient settings), assessed at 30, 60, and 90 days after full vaccination (second dose +14 days). Secondary—composite severe outcomes: ICU admission; ICU admission/death/transfer to hospice; hospitalization/ICU admission/death/transfer to hospice, assessed over similar windows and via time-to-event analyses. Analytic datasets: Large vaccinated cohorts were constructed for time-to-event and for binary outcomes at 30, 60, and 90 days, with counts by vaccine and restrictions based on being fully vaccinated by specific calendar cutoffs to allow adequate follow-up. Kaplan–Meier curves estimated time to infection and time to severe outcomes. Negative control outcomes were modeled to assess residual confounding. Statistical analysis: For binary outcomes at 30/60/90 days, multivariable logistic regression models estimated adjusted odds ratios (aORs) comparing mRNA-1273 vs BNT162b2. Models included: unadjusted; propensity score subclassification; models adjusted for age, timing of vaccination, residence, prior Covid-19, socioeconomic status; models further adjusting for nursing facility residence and comorbidities; with and without inverse probability of treatment weighting (IPTW). Time-to-event analyses used weighted and unweighted Cox proportional hazards models, reporting adjusted hazard ratios (aHRs). Predicted survival probabilities and marginal event rates were generated. Number needed to vaccinate (NNV) was derived from marginal risks. Covariates included demographics, vaccination timing, state of residence, urban/rural status, prior Covid-19, nursing facility transition, and comorbid conditions. Sensitivity and stratified analyses were performed for no prior Covid-19 and age strata.

- Cohort: >3.5 million fully vaccinated individuals; 8,848 documented Covid-19 infections. Among infections, 3,090 (35%) occurred in mRNA-1273 recipients and 5,758 (65%) in BNT162b2 recipients. - Infection risk: mRNA-1273 was associated with lower odds of infection than BNT162b2, with the difference emerging early and increasing over time. Examples: 30-day aOR 0.67 (95% CI 0.63–0.71); 90-day aOR 0.66 (95% CI 0.60–0.73). Time-to-event analysis similarly favored mRNA-1273 for infection: aHR 0.69 (95% CI 0.66–0.72). - Number needed to vaccinate (NNV) to prevent one infection when using mRNA-1273 instead of BNT162b2 decreased over time: approximately 1,870 at 30 days and 290 at 90 days. - Severe outcomes: No statistically significant differences between vaccines were observed for ICU admission, ICU/death/hospice, or hospitalization/ICU/death/hospice in primary interpretations. Example estimates reported include composite ICU/death/hospice aHR 0.76 (95% CI 0.50–1.16) and composite hospitalization/ICU/death/hospice with model-based estimates showing no significant difference overall; an odds ratio example reported as aOR 1.23 with a wide CI (95% CI 0.67–2.25), consistent with no clear difference. - Subgroup analyses: Results were similar when restricting to those without prior Covid-19, those ≥65 years, and those ≤64 years; models consistently indicated lower infection risk with mRNA-1273 and no clear differences for severe outcomes. - Comorbidities associated with higher risk of severe composite outcomes among vaccinated individuals included chronic heart failure (aHR 1.52, 95% CI 1.03–2.26), hypertension (aHR 2.17, 95% CI 1.30–3.62), and lymphoma (aHR 7.03, 95% CI 4.31–11.47).

The study addresses whether mRNA-1273 and BNT162b2 differ in preventing infection and severe Covid-19 outcomes in real-world use. Findings indicate that mRNA-1273 confers a modest but statistically significant advantage in preventing SARS-CoV-2 infection, with divergence apparent early after full vaccination and increasing through 90 days. Despite this difference in infection risk, both vaccines performed similarly in preventing severe outcomes such as hospitalization, ICU admission, death, or transfer to hospice in the primary interpretations, suggesting comparable protection against severe disease. Although the absolute individual-level difference in infection risk is small, it is meaningful at the population level; for example, vaccinating 1 million people with BNT162b2 instead of mRNA-1273 could correspond to thousands of additional care-seeking infections at 90 days. Results were robust across multiple modeling approaches (logistic, IPTW, Cox models) and consistent across age groups and among those without prior Covid-19.

In a large, diverse real-world cohort, mRNA-1273 demonstrated slightly greater effectiveness than BNT162b2 in preventing documented SARS-CoV-2 infection over the first 90 days after full vaccination, while both vaccines showed similar effectiveness in preventing severe outcomes (hospitalization, ICU admission, death, or transfer to hospice). These findings support the high effectiveness of both vaccines against severe Covid-19 and indicate a modest incremental benefit of mRNA-1273 for reducing infections that may have meaningful population-level implications.

- Severe adverse events were rare, limiting statistical power to detect small differences between vaccines for hospitalization, ICU admission, death, or hospice transfer. - Different analytic cohorts were used for 30-, 60-, and 90-day outcomes; between-model comparisons are not valid due to differing populations across models. - Outcomes for infection reflect a care-seeking population identified via claims (ER, outpatient, inpatient encounters), which may under-ascertain asymptomatic or non-care-seeking infections. - Follow-up was limited to 14–151 days after the second dose, so longer-term comparative effectiveness was not assessed. - As an observational claims-based study, residual confounding may remain despite adjustment, propensity methods, and negative control outcome checks.