PREreview del Genomic surveillance reveals age-structured SARS-CoV-2 transmission across demographics and settings

Summary

This manuscript uses large-scale SARS-CoV-2 genomic surveillance from Massachusetts to examine how transmission varied by age, setting, geography, vaccination status, and viral lineage. The authors analyze more than 130,000 high-quality genomes, including more than 85,000 linked to individual-level epidemiological metadata from 666 testing settings between November 2021 and January 2023.

The main findings are that transmission signals were strongly age-structured: elevated within-facility relatedness was concentrated among older residents in skilled nursing facilities, undergraduate-aged individuals in colleges, and older adolescents in schools, with comparatively little excess signal among staff-aged groups. The authors also find that new lineages often expanded first in young adults and college settings, that viral spread followed urban-to-rural geographic patterns, that booster vaccination was associated with reduced onward transmission, and that moderate sustained sequencing can provide timely variant detection. This work advances the field by showing how linked genomic and epidemiological surveillance can reveal fine-scale transmission patterns that would be difficult to infer from case counts alone.

Major Issues

1. The central analyses depend strongly on public testing as the “community” comparator. Public testing may differ from school, college, SNF, and workplace testing in symptom status, testing frequency, access, age structure, behavior, and calendar timing. The authors should add sensitivity analyses or clearer discussion showing how robust the within-facility enrichment estimates are to these differences, ideally stratified by lineage, time period, testing mode, and age.

2. Closely related genomes are used as a proxy for transmission, but the manuscript sometimes reads as though these links identify where transmission occurred. Genomes differing by ≤2 substitutions within 10 days can also reflect shared exposure, unsampled intermediates, or background community transmission. Please consider adding sensitivity analyses using alternative SNP/time thresholds and softening causal language from “transmission” to “transmission linkage” or “within-setting relatedness” where appropriate.

3. Age is used as a proxy for role in several settings, especially staff versus residents in SNFs and staff versus students in schools/colleges. This is reasonable but imperfect. If role, residence, dormitory status, or facility-level metadata are unavailable, the authors should make this limitation more prominent and avoid overinterpreting staff-specific or resident-specific conclusions.

4. The vaccination analyses are interesting but vulnerable to confounding. Booster status is likely correlated with age, calendar time, variant, testing behavior, prior infection, health status, and risk behavior. The manuscript should more clearly frame these results as associations and, if feasible, add adjusted models or stratified analyses by age, lineage, time period, and setting.

5. The phylogeographic inference of introductions and urban-to-rural movement is compelling, but it may be sensitive to background sampling, uneven sequencing across municipalities, and phylogenetic uncertainty. The authors should report more uncertainty around inferred movements, explain how consistent results were across the two sampling strategies, and clarify the limits of directional inference from ancestral-state reconstruction.

6. The sampling-rate recommendations are useful, but their generalizability should be discussed more explicitly. The estimates assume a high-quality genomic surveillance system and may not transfer directly to settings with lower testing access, lower genome recovery, different variant growth rates, or more biased sampling.

Minor Issues

1. Please clarify the relationship between the different dataset sizes: >130,000 genomes, 134,785 high-quality genomes, 94,404 genomes used for phylogenetic inference, and 85,125 genomes with linked epidemiological metadata.

2. The “other workplaces” category is small and heterogeneous. Conclusions about workplace transmission should be stated cautiously.

3. A concise table showing metadata availability by sector, age group, time period, and vaccination status would help readers assess missingness and possible bias.

4. Some key terms would benefit from clearer early definitions, including “linked/unlinked,” “facility,” “public testing,” “SNF,” and the use of lineage labels with asterisks.

5. The figures are information-rich but dense. Consider adding clearer legends or short interpretive captions for non-specialist readers.

6. The data/code availability statement is helpful, but reproducibility would be improved by providing figure-generation scripts, repository commit hashes, and exact analysis environments.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.