2) Omicron had lower peak viral load and more variable early viral growth durations than Delta
1/13
We obtained longitudinal RT-qPCR tests (a combined anterior nares & oropharyngeal swab in each person) collected between July 5th 2021 and January 10th 2022 from the NBA occupational health program. This included 97 Omicron infections and 107 confirmed Delta infections. 2/13
Here are trajectories from individuals detected ≤1 day or ≥2 days since a previous negative or inconclusive test. The latter group reflects identification due to symptoms or concern for contact with an infected individual. Data here: github.com/gradlab/CtTraj… 3/13
We looked at the % of individuals with Ct<30 on each day post detection, noting Ct<30 is a useful but caveated proxy for LFT sensitivity and culturable virus. We saw: 1) ~50% of individuals have Ct<30 on day 5 post detection 2) All individuals have Ct≥30 by day 11
4/13
Next, building on our prior work (tinyurl.com/2szh8368, tinyurl.com/2p82efyz), we quantified the viral dynamics. Omicron infection duration was similar to Delta (both ~10 days), but Omicron infections had lower peak viral loads (higher peak Ct values). 5/13
We also saw more variability in estimated proliferation times – that is the delay from infection to reaching peak Ct. Plot shows individual-level posterior mean estimates (Omicron in red, Delta in blue). 6/13
This seems consistent with anecdata finding some Omicron infections with low viral load for a few days prior to growth, possibly reflecting suppression via immunity or different compartmentalization in the respiratory tract, while others grow rapidly.
7/13
Pre-empting a key question: “what’s the impact of immunity?”. Probably a fair bit, as most of these individuals are vaccinated and many were boosted before the rise of Omicron. We are working to update these analyses accounting for immune status. 8/13
Key limitations: 1) Ct<30 threshold is a proxy and does not necessarily predict infectiousness. 2) These infections are likely identified faster than in the general population, so “day 5 post detection” in other contexts may reflect a later point in infection than day 5 here
9/13
Take home 1: ~50% of individuals have low Cts ≥5 days post detection, even in this highly-boosted population (see also tinyurl.com/2p8858jt). A cautious approach to shortening isolation may therefore be warranted, noting that the decision balances multiple factors. 10/13
Take home 2: Peak viral load, proxied by Ct values, is lower, and peaks no sooner, than for Omicron than Delta. So Omicron’s increased infectiousness doesn’t seem to be through higher viral load (at least as measured in the nose and mouth by Ct values). 11/13
This is a preprint and hasn’t been peer reviewed. We welcome feedback! See related (and many other threads):
What is a public health test? A thread on population-level vs. individual-level optimal criteria (this is for a thread to be linked elsewhere so is probably quite incoherent)
From a clinical perspective we want a test that is optimal for helping the individual being treated. So we want to be as sure as possible that the result leads us to the right diagnosis and treatment
To achieve this, we might favour sensitivity and specificity above all else – those are the important metrics for treating our patient. We are probably happy to use a battery of expensive and time-consuming tests because we want to do the right thing for the person in front of us
Ct values can be used to estimate epidemic dynamics UPDATE! Ct values are expected to change depending on whether the epidemic is growing or declining, and we harness this to estimate the epidemic trajectory. Lots of cool new analyses and methods! 1/12
Highlights:
- Cts from symptom-based surveillance change over time, but the effect is weaker
- Methods to infer incidence using single cross-sections of Cts
- Unbiased by changing testing coverage
- Gaussian process (wiggly line) model for incidence tracking using Ct values
2/12
Ct values can be used to estimate epidemic dynamics! We show that the distribution of viral loads changes during an epidemic and develop a new method to infer growth rates from cross-sectional virological surveys without using reported case counts.
Highlights:
- The distribution of observed viral loads is determined by recent incidence trends
- The median and skew of detectable Cts in Massachusetts were correlated with R(t), as predicted
- A novel statistical method to infer the epidemic growth rates using Ct values
2/25
Cts are inversely proportional to log viral loads. The relationship depends on the instrument and sampling variation, but low Cts generally indicate high viral loads. Think about expectations & distributions rather than individual measurements
Very excited to share our updated preprint on pooled testing for SARS-CoV-2 surveillance. This has been a fantastic modeling and lab collaboration with @BrianCleary, @michaelmina_lab and Aviv Regev, and it’s all about our favorite topic: viral loads. 1/16
Highlights:
-PCR sensitivity and efficiency are linked to epidemic dynamics and viral kinetics
-Prevalence estimation without testing individual samples using a few dozen tests
-Simple (by hand) strategies optimized for resource-constrained settings
Full story below. 2/16
We (the world) still need more testing. The number of test kits is still limited in a lot of places, meaning that we are missing a lot of infections, not testing regularly, and are flying blind wrt population prevalence. Pooling has been discussed as part of the solution. 3/16
We propose 4 possible explanations for the lack of cases in children:
1) Kids haven’t been making as many contacts as normal. This may contribute, but probably isn’t the only factor: medrxiv.org/content/10.110…
2) Children are less susceptible to *infection* or adults are more susceptible. Seems unlikely now, given secondary attack rates from contact tracing, though conflicting findings: medrxiv.org/content/10.110… + previous
A recent analysis from Oxford presented a range of model scenarios consistent with observed COVID death counts. I’m going to reproduce their analysis here and then present some slight modifications to provide a conservative (if technical) perspective. (gonna be 15ish tweets)
They showed you can estimate the same number of deaths with either a high % of the population at risk of severe disease and a recent epidemic start, or a low % and an earlier start. Some media outlets have reported this as suggesting “majority of the UK has already been infected”
But that’s *not* what the authors were trying to say. The aim was: given what we do know about the virus, let’s test different assumptions for the stuff we don’t know and see which tests are consistent with observed death counts.