Omicron variant a month later: what do we (not) know?
This is a follow-up to the Omicron variant article I wrote a few weeks ago.
It’s been about a month since the identification of the Omicron variant, and, while we have more questions than answers, some insights are already emerging and evolving. I’m grateful for all the thoughtful, hard-working, and likely very exhausted individuals working around the clock and around the globe to get us some answers and to care for those affected. Today, I wanted to take the time to summarize what we know so far (as of 12/21/2021), and to urge for more (and more comprehensive) data collection and sharing. As I was researching the updates to these questions, it struck me that it’s very difficult to tease out each individual question, especially when data points are missing. In particular, it was still very difficult to get US-based data (even though there’s strong belief that the Omicron variant will overtake as the main variant soon). Despite the encouraging progress, we’re still a ways away from closing our COVID-19 data gap.
As a recap, these were some of the questions we needed to answer, to determine how the Omicron variant was going to affect vaccinated and unvaccinated populations:
Below is what we know so far (think of it as a barely passable literature review).
But first, the caveats:
- Numbers are changing almost daily and should be taken as directional. Any incorrect information shared is unintentional. I will attempt to correct or edit anything as soon as I learn about it.
- Epidemiological modeling is relying on data from different global studies, and each of the analyses comes with different confounding variables at the population level.
- Given the near-real time of the data, little of what’s cited here is peer reviewed or replicated. Ideally, I would prefer to cite published scientific papers wherever I can.
- Leading indicators are used to extrapolate information such as severity of disease. As more information comes online, this will be replaced with data that will hopefully provide a more holistic, longitudinal view of disease course and outcomes.
With those caveats in mind, let’s go through each of these questions.
A. Transmissibility
A1. How infectious is Omicron?
Data suggest that Omicron is very infectious. Data from research at the Imperial College in London estimates that the risk of reinfection for vaccinated individuals is 5.4 times greater for the Omicron variant than for the Delta variant.
This is likely due to a combination of increased viral replication rates and evasion of the immune response. Laboratory tests of infections with Delta vs. Omicron pseudoviruses show an increased rate of replication in airway cells. Higher replication means higher viral loads, which could mean a more likely spread, and a higher R0 (or basic reproduction number). An article in Lancet Respiratory Medicine quoted the following R0 values: “[t]he original strain of SARS-CoV-2 has an R0 of 2·5, while the delta variant (B.1.617.2) has an R0 of just under 7. Martin Hibberd, professor of emerging infectious diseases at London School of Hygiene & Tropical Medicine (London, UK), reckons omicron’s R0 could be as high as 10.”
At these levels of transmissibility, nearly everyone in the population needs to be fully immunized against this variant in order to limit the spread of disease. We are already seeing the results of such high transmissibility: from UK data, we are seeing a doubling of Omicron cases every 2–3 days. If these numbers are consistent in other locations, we are mere months (weeks?) away from Omicron levels taking over any other SARS-CoV-2 viral variant. US data are still pending (and the past 48 hours have led to serious confusion based on a CDC Nowcast algorithm reporting bug that got propagated by the media and various thought leaders).
Sources: Nature (December 17), Imperial College of London (ICL)(Report 49), Imperial College of London (Modeling), Hong Kong University Faculty of Medicine, The Lancet Respiratory Medicine, CDC variant proportions tracker, The Prepared blog post
A2. Do the mutations make the virus more able or less able to evade the immune system? To enter epithelial cells?
We can’t conclusively answer yet, but data so far suggests the answer for both is yes, at least based on laboratory tests looking at pseudovirus infections and other in vitro studies. The immune system evasion is likely due to the mutations in the Spike protein, which render the virus less recognizable to antibodies produced either through prior infection or vaccination. Higher replication rates (at least in the airway), as discussed above, might be responsible for higher infection rates. Cellular entry needs to be further studied, although more in vitro studies are likely to be published soon.
Sources: Hong Kong University Faculty of Medicine, The Wall Street Journal
B. Severity
B1. How sick do infected patients get?
Data to answer this question is still inconclusive. Hospitalization rates are being used as a leading indicator of disease severity (with time, morbidity and mortality data will come online).
Some early insights from Johannesburg suggest ~30% fewer hospitalizations, and a flattened 7 day trailing average of hospitalizations. However, there are several confounders to consider in the data. It is not likely that data from South Africa, which has a different demographic spread, different vaccination penetration, and different levels of prior COVID infection and COVID-related mortality compared to the US, could be extrapolated to infectious models in the US or elsewhere. In fact, data coming out of Denmark and the United Kingdom suggest that hospitalization rates are on par with prior infections.
Right now, most studies have compared severity indicators to historical data from the Delta variant and prior infections. However, comparison to historical data is challenging because the population infected is changing with time (e.g., more individuals are vaccinated, more have prior infections, etc.). Layering seasonality and travel patterns (e.g., more events are indoors in wintery northern hemisphere, more people are traveling for the holidays) will make the comparison to real-world data from the Delta wave more challenging. We need to collect and connect different data types, inclusive of different patient populations, to answer this question more reliably (see question B3 for an example).
Sources: Nature (December 17), Imperial College of London (Report 49)
B2. How does this differ for vaccinated vs. unvaccinated populations?
Bottom line: everyone should get vaccinated and boosted if they can. Being unvaccinated continues to put individuals at serious risk for severe disease, hospitalization, and death. (More below on immune protection and breakthrough infections).
While we’re at it, let’s follow all other best practices to curb disease spread. Sorry, grandma, if you’re reading this: we’ll have to take a walk outside with our masks on again :).
Sources: CDC Omicron variant page, Harris County, TX press release
B3. Are certain populations more likely to get severe disease in ways we haven’t seen with prior variants?
More data points are needed. Beyond the already established risks, there’s not enough data to suggest that other populations are more at risk to get severe disease due to Omicron. To get a reliable answer here, we would need the following types of data:
- Infection status (COVID test data), AND
- Variant type status (sequencing data) AND
- Medical history and demographics (to segment different populations based on different factors we’d want to study) AND
- Outcome data (hospitalization status, recovery rate, mortality data, long-term follow-up data to determine post-acute COVID symptoms)
In particular, we need to determine how Omicron infection can affect pediatric populations, given there are no vaccine options yet for younger children.
C. Immune escape
C1. How likely is Omicron to cause breakthrough infections in vaccinated individuals?
So far, data show it’s very likely for Omicron to cause breakthrough infections (see ICL data). However, we see two insights that might provide a silver lining. First, while this mutated form of COVID-19 might evade some antibody response aimed at targeting the “wild-type” version of the spike protein, T-cell response does not seem to be affected. Secondly, early data suggests that more recent inoculations (e.g., boosting) may increase antibody production enough to counteract infection.
C2. Do any of the current vaccines offer more protection? Less protection?
Research is still ongoing here, and the answer to this question will be dependent on vaccination timing and level (e.g., boosters). With mix-and-match vaccination options, teasing this question out will be even more difficult (although not impossible).
C3. Does booster availability and timing affect breakthrough infection rates?
I think answering this question (and the highly correlated question around severity) is hugely important. Data on booster effectiveness at the population level is still inconclusive, in part because not a large enough “boosted population” has been studied. A few things need to happen before we can have a reliable answer on booster data:
- Enough individuals need to get boosters, AND
- The booster status needs to be tracked appropriately. We need to know the # of vaccine doses each individual has received, what vaccine type they have received, and when their last dose was administered. (In the US, we don’t yet have high accuracy data on the number of people who have received a first and second shot of the vaccine, and data points on boosters are lagging further behind), AND
- Individuals who show symptoms of disease need to get tested, AND
- Samples need to be sequenced to determine variant
All this information needs to be stitched together, as each piece provides key answers for how to proceed. We must be cognizant though that at every stage there could be a drop — off of information. For example, if individuals get sick but don’t get tested, that data point is missing. More likely, the fragmented nature of vaccine dose status tracking will make answering this question very difficult, at least in the US.
Sources: Nature (December 2, with updates), Schmidt et al., 2021 (preprint awaiting peer review)
Conclusion
While information is coming online much more quickly than for prior variants, there are still a series of open questions that are impacted both by data collection and data sharing challenges. This is especially true in the US, where sequencing availability is uneven and fragmented, and where infrastructural IT challenges and operational challenges of communicating across jurisdictions are making it difficult for the CDC and others to provide information in near-real time. Most of the research I reviewed to provide preliminary answers above came from ex-US data sources. The feeling that we’re still building the plane as we fly it is still very real.
I still hold hope that a milder course of the disease, coupled with redoubling of vaccination and boosting efforts, and an increased public health consciousness will lead to us overcoming this wave before the burden to the health system becomes more unmanageable than it is right now. I also hope that we take these learnings as a sign that continued robust investing in our data infrastructure and our healthcare systems is non-negotiable, both in the US and abroad.
Note 1: I am employed by Datavant, however thoughts here are my own and do not necessarily represent the views of my employer.
Note 2: If you’re excited about contributing to scientific and health economics research based on connected health data, please consider checking out the COVID-19 Research Database and submitting a research proposal: https://covid19researchdatabase.org/
Special thanks to Elenee Argentinis for providing feedback!
