TL;DR
While early reports confirmed that the SARS-CoV-2 lineage B.1.1.7 (the Alpha variant) was significantly more transmissible, its impact on mortality remained unclear. This study analyzes a massive dataset of 2.2 million community tests in England, finding that B.1.1.7 is associated with a 55% to 61% higher hazard of death compared to earlier variants, even after adjusting for age, sex, and socioeconomic factors.
What Problem Does This Paper Address?
By late 2020, the emergence of B.1.1.7 in the UK had fundamentally changed the trajectory of the pandemic due to its increased transmissibility. However, a critical question remained: Does "more contagious" also mean "more severe"?
Measuring disease severity is notoriously difficult during an active outbreak. Crude death rates can be misleading because they are heavily influenced by who is getting tested, the age of the infected population, and whether hospitals are currently overwhelmed. The authors sought to isolate the specific "variant effect" on mortality by linking individual-level community testing data with comprehensive death records, allowing for a granular adjustment of confounding variables.
Methods
The researchers employed several sophisticated statistical techniques to handle the complexities of real-world epidemiological data:
- SGTF as a Proxy: B.1.1.7 carries a specific deletion in the spike gene. In certain PCR assays (Thermo Fisher TaqPath), this causes the "S-gene" target to fail while other targets succeed. This S-gene target failure (SGTF) served as a reliable proxy for identifying B.1.1.7 cases without needing full genomic sequencing for every sample.
- Cox Proportional Hazards Models: This is the standard for survival analysis. The authors used it to estimate the "hazard ratio"—the relative increase in the risk of death for B.1.1.7 vs. non-B.1.1.7 cases over a 28-day period.
- Inverse Probability Weighting (IPW): Since SGTF status was only available for about half of the tests (depending on which lab processed the sample), the authors used IPW to correct for "missingness." This ensures that the results aren't biased by the fact that certain regions or age groups might have had better access to SGTF-capable labs.
- Misclassification Analysis: They accounted for the fact that, early in the study period, a small number of other variants also caused SGTF, using a logistic growth model to estimate the probability that a given SGTF sample was truly B.1.1.7.
Datasets
The study utilized three primary datasets provided by Public Health England:
- Pillar 2 Community Testing: A line list of 2,245,263 individuals who tested positive in the community between September 2020 and February 2021.
- Lighthouse Lab Ct Values: Cycle threshold (Ct) values for the ORF1ab, N, and S genes, used to determine SGTF status.
- COVID-19 Death Line List: A comprehensive, deduplicated list of all COVID-19 deaths in England, linked via a common numeric identifier.
Key Insights and Ideas
- The Divergence Over Time: Interestingly, the hazard of death for B.1.1.7 wasn't just higher; it appeared to increase relative to other variants the longer the patient survived post-test. The hazard ratio was 1.14 on day 1 but climbed to 2.24 by day 28.
- Consistency Across Strata: The increased risk was remarkably consistent. Whether looking at different age groups, sexes, or regions, B.1.1.7 consistently showed higher mortality rates in 94% of the 104 different subgroups (strata) analyzed.
- The Viral Load Connection: The authors noted that B.1.1.7 infections are associated with higher viral concentrations (lower Ct values) on nasopharyngeal swabs. This higher viral load is a likely candidate for the biological mechanism driving increased severity.
- Test-Seeking Behavior: By comparing their data to random community sampling (ONS data), the authors showed that B.1.1.7 cases weren't just "more likely to seek tests because they felt sicker." The variant's prevalence in the testing data matched its prevalence in the general population.
Key Results
- Hazard Ratio: The adjusted hazard of death associated with SGTF was 1.55 (95% CI 1.39–1.72). When correcting for misclassification and missing data, this estimate rose to 1.61 (95% CI 1.42–1.82).
- Absolute Risk Increase: For a male aged 55–69, the absolute risk of death within 28 days of a positive community test increased from 0.6% to 0.9%.
- Elderly Impact: In females aged 85 or older, the risk jumped from 13% to 19%; for males in the same age bracket, it rose from 17% to 25%.
Limitations and Future Work
- Community Focus: The study only included people tested in the community. It excludes those who went directly to the hospital without a prior community test, which might miss the most severe immediate presentations.
- Vaccination Status: The analysis was conducted during the very early stages of the UK's vaccination rollout and could not account for individual vaccination status.
- Mechanism: While the study establishes a strong correlation between B.1.1.7 and mortality, it does not definitively prove the biological mechanism (e.g., whether it's purely due to higher viral load or different tissue tropism).
- Future Variants: This framework provides a "gold standard" for how to evaluate the severity of future variants of concern as they emerge.
Summary updated