Real-World Effectiveness of Nirmatrelvir in Protecting Long COVID for Outpatient Adult Patients – A Large-Scale Observational Cohort Study from the RECOVER Initiative

Fei Wang, Chengxi Zang, Haoyang Li, Dhru Khullar, Yongkang Zhang, Stephenson Strobel, Yong Chen, Marc A. Sala, Payal Patel, Alejandro P. Comellas, Andrew Wylam, Mark G. Weiner, Christopher R. Forrest, Thomas W. Carton, Rainu KaushalResearch Square 2024 · DOI

This large-scale observational study, part of the NIH’s RECOVER initiative, utilizes real-world data from nearly 500,000 patients to demonstrate that Paxlovid (nirmatrelvir-ritonavir) significantly reduces the risk of Long COVID, hospitalization, and death in high-risk outpatient adults. Crucially, the study finds that these protective benefits do not appear to extend to low-risk patients, providing a nuanced map for clinical decision-making.

What Problem Does This Paper Address?

While Paxlovid is a well-established tool for preventing severe acute COVID-19, its ability to prevent "Long COVID"—formally known as Post-Acute Sequelae of SARS-CoV-2 (PASC)—has remained a subject of intense debate. Previous research, such as a high-profile study from the Veterans Affairs (VA) system, suggested a protective effect, but that population was predominantly older, white, and male.

Clinicians and researchers have faced a significant gap: Does Paxlovid work for a more diverse general population? And perhaps more importantly, does it provide any long-term benefit for "low-risk" patients (younger adults without underlying conditions) who are increasingly seeking treatments to avoid the lingering exhaustion and brain fog associated with PASC?

Methods

The researchers employed a target trial emulation framework. This is a sophisticated statistical approach designed to apply the rigors of a randomized controlled trial (RCT) to observational data (in this case, Electronic Health Records or EHRs).

Key technical components included:

  • Propensity Score Weighting: To account for the fact that doctors are more likely to prescribe Paxlovid to sicker patients, the team used Stabilized Inverse Probability of Treatment Weighting (IPTW). This balances the "treated" and "untreated" groups across a massive range of covariates (age, BMI, vaccination status, and dozens of pre-existing conditions).
  • Computable Phenotyping: Long COVID is notoriously difficult to define. The authors used a "rules-based computable phenotype" leveraging ICD-10 codes across 24 distinct condition clusters to identify incident PASC cases between 30 and 180 days post-infection.
  • Competing Risk Analysis: They used the Aalen-Johansen model to estimate cumulative incidence, treating death as a "competing risk" that could preclude other outcomes.

Datasets

The study utilized the PCORnet RECOVER repository, which aggregates data from 29 different EHR repositories across the United States.

  • Scale: 497,499 SARS-CoV-2 positive patients.
  • Diversity: Unlike the VA study, this cohort included a significant percentage of female (over 60%), Black, and Hispanic patients, as well as a wide range of ages and geographic locations (urban vs. rural).
  • Timeframe: March 1, 2022, to February 1, 2023 (the Omicron era).

Key Insights and Ideas

  • The High-Risk "Shield": For patients with at least one risk factor (e.g., diabetes, obesity, or age 65+), Paxlovid acts as a multi-system shield. It didn't just reduce "general" Long COVID; it reduced risks for neurological, pulmonary, and metabolic complications.
  • The Low-Risk Limit: The most striking "aha moment" is the lack of evidence for benefit in low-risk patients. For those under 50 without documented risk factors, the Hazard Ratio for Long COVID was 1.03—essentially a coin flip. This suggests that the biological mechanism by which Paxlovid prevents PASC may be tied to its ability to dampen high-viral-load-driven inflammation, which is less of a factor in healthier individuals.
  • Demographic Generalizability: The study confirms that the benefits seen in the VA study hold true for women and demographically diverse populations, though the magnitude of benefit was notably higher in males and the elderly.
  • Rural Advantage: Interestingly, the greatest absolute risk reduction was found in patients from small towns and rural areas, potentially highlighting the drug's impact in areas with different healthcare access patterns.

Key Results

For high-risk patients, Paxlovid treatment within 5 days of infection was associated with:

  • Long COVID: A 12% relative risk reduction (Hazard Ratio 0.88) and an absolute reduction of 2.99 events per 100 persons.
  • Mortality: A staggering 47% reduction in all-cause death (HR 0.53) during the post-acute phase.
  • Hospitalization: A 30% reduction in all-cause hospitalization (HR 0.70).
  • Specific Symptoms: Significant protection against pulmonary fibrosis, cognitive problems ("brain fog"), and diabetes, but no effect on hair loss.

Limitations and Future Work

  • Residual Confounding: As an observational study, there is always a risk that "unmeasured" factors (like health-seeking behavior or wealth) influenced who got the drug.
  • Adherence: EHR data shows a prescription was written, but it cannot confirm if the patient actually finished the five-day course of Paxlovid.
  • Pre-existing Conditions: The study focused on incident (new) conditions. It did not measure whether Paxlovid prevents the worsening of conditions a patient already had.
  • Future Directions: The authors suggest more research is needed into the distinct biological subtypes of Long COVID, as Paxlovid may be more effective against some "flavors" of the disease than others.

Summary updated

Abstract

Paxlovid has been approved for use in patients who are at high risk for severe acute COVID-19 illness. Evidence regarding whether Paxlovid protects against Post-Acute Sequelae of SARS-CoV-2 infection (PASC), or Long COVID, is mixed in high-risk patients and lacking in low-risk patients. With a target trial emulation framework, we evaluated the association of Paxlovid treatment within 5 days of SARS-CoV-2 infection with incident Long COVID and hospitalization or death from any cause in the post-acute period (30-180 days after infection) using electronic health records from the Patient-Centered Clinical Research Networks (PCORnet) RECOVER repository. The study population included 497,499 SARS-CoV-2 positive patients between March 1, 2022, to February 1, 2023, and among which 165,256 were treated with Paxlovid within 5 days since infection and 307,922 were not treated with Paxlovid or other COVID-19 treatments. Compared with the non-treated group, Paxlovid treatment was associated with reduced risk of Long COVID with a Hazard Ratio (HR) of 0.88 (95% CI, 0.87 to 0.89) and absolute risk reduction of 2.99 events per 100 persons (95% CI, 2.65 to 3.32). Paxlovid treatment was associated with reduced risk of all-cause death (HR, 0.53, 95% CI 0.46 to 0.60; risk reduction 0.23 events per 100 persons, 95% CI 0.19 to 0.28) and hospitalization (HR, 0.70, 95% CI 0.68 to 0.73; risk reduction 2.37 events per 100 persons, 95% CI 2.19 to 2.56) in the post-acute phase. For those without documented risk factors, the associations (HR, 1.03, 95% CI 0.95 to 1.11; risk increase 0.80 events per 100 persons, 95% CI -0.84 to 2.45) were inconclusive. Overall, high-risk, nonhospitalized adult patients with COVID-19 who were treated with Paxlovid within 5 days of SARS-CoV-2 infection had a lower risk of Long COVID and all-cause hospitalization or death in the post-acute period. However, Long COVID risk reduction with Paxlovid was not observed in low-risk patients.
BibTeX
@inproceedings{wang2024real,
  author = {Fei Wang and Chengxi Zang and Haoyang Li and Dhru Khullar and Yongkang Zhang and Stephenson Strobel and Yong Chen and Marc A. Sala and Payal Patel and Alejandro P. Comellas and Andrew Wylam and Mark G. Weiner and Christopher R. Forrest and Thomas W. Carton and Rainu Kaushal},
  title = {{Real-World Effectiveness of Nirmatrelvir in Protecting Long COVID for Outpatient Adult Patients – A Large-Scale Observational Cohort Study from the RECOVER Initiative}},
  booktitle = {Research Square},
  year = {2024},
  doi = {10.21203/rs.3.rs-4536807/v1},
}
Figure