Comparative effectiveness of mRNA-1273 versus protein-based NVX-CoV2705 vaccination on COVID-19-related outcomes among US insured adults during 2024--2025: a retrospective matched cohort study

In a retrospective matched cohort study of US insured adults during the 2024–2025 season, the mRNA-based vaccine mRNA-1273 demonstrated significantly greater effectiveness against medically attended COVID-19 and hospitalization compared to the protein-based NVX-CoV2705 vaccine.

The Gist

The Big Picture: A Race Between Two Different Defenses

Imagine the SARS-CoV-2 virus is a shape-shifting thief that keeps changing its disguise (variants) to break into our homes (our bodies). To stop this thief, we have two different types of security guards available for the 2024–2025 season:

1. Guard A (mRNA-1273): This is the "high-tech" guard. It uses a blueprint (mRNA) to teach your body's own security team exactly how to build a weapon against the thief. This guard was updated to target the KP.2 version of the thief.
2. Guard B (NVX-CoV2705): This is the "classic" guard. It brings a physical model (protein) of the thief so your security team can practice recognizing it. This guard was updated to target the JN.1 version of the thief.

The Question: In the real world, which guard did a better job at keeping people out of the hospital and out of the doctor's office?

The Experiment: A Massive Match-Up

Researchers at Moderna looked at a huge database of insurance records from the US, covering nearly 1 million people who got vaccinated between August 2024 and February 2025.

To make it a fair fight, they played "matchmaker." They took every person who got Guard B (the protein vaccine) and found two people who got Guard A (the mRNA vaccine) who were almost identical in age, health history, location, and when they got their shot.

• Total Matched Group: About 103,000 people.
• The Goal: See who got sick and who ended up in the hospital over the next 6 months.

The Results: Guard A Pulled Ahead

After watching these groups for about six months (a full respiratory season), the results were clear:

1. Avoiding the Doctor's Office (Medically Attended COVID-19)

• Guard A (mRNA): Only about 1 out of every 100 people needed to see a doctor for COVID.
• Guard B (Protein): About 1.5 out of every 100 people needed to see a doctor.
• The Winner: Guard A was about 32% better at preventing people from getting sick enough to need medical help.

2. Avoiding the Hospital (Severe Outcomes)

• Guard A (mRNA): Only 0.09% (less than 1 in 1,000) ended up in the hospital.
• Guard B (Protein): 0.14% ended up in the hospital.
• The Winner: Guard A was about 41% better at preventing severe illness requiring hospitalization.

Did it matter if you were older?
Even in the group of people aged 65 and older (who are at higher risk), Guard A still performed better, preventing about 26% more doctor visits and 42% more hospitalizations compared to Guard B.

Why Did Guard A Win?

The researchers suggest two main reasons, like a "one-two punch":

1. The Blueprint vs. The Model (The Platform):
   Think of Guard A (mRNA) as a 3D printer that can instantly print a custom weapon the moment it sees a new threat. Guard B (Protein) is like a pre-made statue. The study suggests the "3D printer" approach might create a stronger, more flexible immune response that lasts longer.

2. The Target (The Variant):
   This is the tricky part. Guard A was trained on KP.2, while Guard B was trained on JN.1.

   • Imagine the thief (the virus) started wearing a JN.1 mask in the winter, but by spring, they had switched to a KP.2 mask.
   • Because Guard A was trained on KP.2, it was better at recognizing the thief as the season changed.
   • The researchers note that the mRNA technology allows scientists to update the "blueprint" faster, so the vaccine can stay closer to the actual virus circulating in the community.

The Takeaway

This study tells us that while both vaccines are helpful, the mRNA vaccine (Guard A) offered stronger protection against getting sick and going to the hospital during this specific season.

Why does this matter?
It helps doctors and public health officials decide which "security guard" to recommend. It suggests that when choosing vaccines, we shouldn't just look at what the vaccine targets, but also how it's made (the technology platform), because some technologies might adapt better to a virus that is constantly changing its disguise.

---

Note: This study was conducted by employees of Moderna, the company that makes the mRNA vaccine. While the data comes from a large, real-world database, it is important to remember that the authors have a financial connection to the winning product.

Technical Summary

1. Problem Statement and Background

• Context: SARS-CoV-2 continues to circulate and evolve, necessitating periodic updates to vaccine compositions. The 2024–2025 vaccination season in the US presented a unique scenario where concurrently available vaccines differed not only by platform (mRNA vs. recombinant protein) but also by antigen composition (KP.2-targeted vs. JN.1-targeted).
• The Gap: While there is substantial real-world evidence comparing different mRNA vaccines, there is limited head-to-head data comparing mRNA vaccines against protein-subunit vaccines (specifically NVX-CoV2705) in the US during the 2024–2025 season.
• Objective: To estimate the comparative vaccine effectiveness (cVE) of the KP.2-targeted mRNA-1273 (Moderna) versus the JN.1-targeted NVX-CoV2705 (Novavax) in preventing medically attended COVID-19 and hospitalization among insured US adults.

2. Methodology

• Study Design: Retrospective matched cohort study using administrative claims data.
• Data Source: Optum's de-identified Clinformatics® Data Mart, covering large commercial and Medicare Advantage health plans.
• Study Period: August 31, 2024, to February 28, 2025 (vaccination window); follow-up up to 180 days.
• Population:
  • Inclusion: Adults ≥18 years receiving one dose of mRNA-1273 or NVX-CoV2705 within the window.
  • Exclusion: Prior COVID-19 diagnosis/treatment within 90 days, prior vaccination within 60 days, conflicting demographic data, or death before the index date.
• Matching and Weighting:
  • Exact Matching: mRNA-1273 recipients were matched 2:1 to NVX-CoV2705 recipients without replacement based on: age group, sex, region, insurance type, high-risk condition status (CDC-defined), prior-season vaccination history, and week of vaccination.
  • Weighting: Stabilized Inverse Probability of Treatment Weights (IPTW) were applied to balance covariates including prior diagnosis timing, healthcare utilization, influenza vaccination, and Charlson comorbidity scores.
• Outcomes:
  • Primary: Medically attended COVID-19 (ICD-10 code U07.1 in any care setting).
  • Secondary: Hospitalization with COVID-19 (U07.1 in any diagnosis position for inpatient admission).
  • Timeframe: From Day 7 post-vaccination to 180 days or censoring (death, loss of enrollment, re-vaccination).
• Statistical Analysis: Cox proportional hazards models were used to estimate Hazard Ratios (HR). Comparative Vaccine Effectiveness (cVE) was calculated as $100 \times (1 - HR)$.

3. Key Contributions

• First Large-Scale US Head-to-Head Comparison: This is the first large claims-based study directly comparing the updated mRNA-1273 and NVX-CoV2705 vaccines in the US during the 2024–2025 season.
• Differentiation of Platform and Antigen Effects: The study addresses the complexity of a season where vaccines differed by both mechanism (mRNA vs. protein) and strain targeting (KP.2 vs. JN.1), providing real-world data on how these factors influence outcomes.
• Robust Statistical Rigor: The use of exact matching combined with IPTW weighting ensured excellent covariate balance (Absolute Standardized Differences < 0.02), minimizing confounding in an observational setting.
• Subgroup Analysis: Specific focus on the ≥65 age group, which bears the highest burden of severe disease.

4. Key Results

• Cohort Characteristics:
  • Final matched cohort: 69,140 mRNA-1273 recipients and 34,570 NVX-CoV2705 recipients.
  • Median follow-up: 180 days for both groups.
  • Demographics: Mean age 66 years; 69% were ≥65 years; 75% had at least one high-risk condition.
• Medically Attended COVID-19:
  • Incidence: 1.02% for mRNA-1273 vs. 1.48% for NVX-CoV2705.
  • Adjusted cVE: 31.7% (95% CI: 23.4%, 39.1%) favoring mRNA-1273.
  • Subgroup (≥65 years): cVE was 25.7% (95% CI: 15.4%, 34.8%).
• Hospitalization with COVID-19:
  • Incidence: 0.09% for mRNA-1273 vs. 0.14% for NVX-CoV2705.
  • Adjusted cVE: 40.7% (95% CI: 13.5%, 59.4%) favoring mRNA-1273.
  • Subgroup (≥65 years): cVE was 41.7% (95% CI: 14.3%, 60.4%).
• Durability: Cumulative incidence curves separated early (post-day 7) and remained distinct throughout the 180-day follow-up, indicating sustained protection without significant attenuation.

5. Significance and Discussion

• Public Health Implications: The findings suggest that in the 2024–2025 season, the mRNA-1273 vaccine provided superior protection against both medically attended illness and severe outcomes (hospitalization) compared to the protein-based NVX-CoV2705.
• Potential Drivers of Difference:
  1. Platform Mechanism: mRNA vaccines may induce more potent neutralizing antibody and CD8+ T-cell responses compared to protein subunit vaccines.
  2. Antigen Match: The mRNA-1273 vaccine targeted KP.2, while NVX-CoV2705 targeted JN.1. Given the rapid evolution of SARS-CoV-2 (with KP.3 and XEC variants emerging), preclinical data suggests KP.2-adapted vaccines may elicit broader cross-neutralizing responses against emerging JN.1-lineage variants compared to JN.1-specific formulations.
  3. Manufacturing Agility: The mRNA platform's flexibility may allow for closer alignment between vaccine antigen selection and circulating strains at the time of administration.
• Limitations:
  • Observational Nature: Residual confounding from unmeasured factors (e.g., healthcare-seeking behavior, frailty) remains possible.
  • Data Source: Based on claims data (ICD-10 codes) rather than laboratory-confirmed PCR tests, though U07.1 has high accuracy in inpatient settings.
  • Generalizability: The study population is limited to insured US adults (commercial and Medicare Advantage), excluding the uninsured and traditional fee-for-service Medicare populations.
• Conclusion: The study supports prioritizing updated vaccine compositions and highlights the potential public health benefit of mRNA platforms over protein-based platforms in the context of rapidly evolving SARS-CoV-2 variants. These results provide critical evidence for clinicians and policymakers when planning vaccination strategies for future seasons.