How long will it take before we find out how bad Omicron is?
Abstract
We do not yet have enough information to determine the transmissibility, severity or immune escape of the Omicron variant, and deficiencies in early data are likely to create further delays. The time taken for us to determine each of the above parameters with a fair degree of confidence will vary somewhat depending on the extent to which they differ from those of the Delta variant. If Omicron has a very large transmission advantage over Delta (>100%), that will become apparent in South Africa within two to three weeks. Otherwise, we will have a fair estimate roughly six weeks from now. If the severity of illness is extremely different to Delta, in either direction, that will be detectable in South Africa’s messy data within two to three weeks. Otherwise, a smaller difference in severity will not be reliably detectable from South Africa’s dataset and will only become apparent in about 6 to 8 weeks based on data from countries with more representative sampling and data collection. Real-world data on immune escape depends on severity data but further complicated by low vaccine coverage in South Africa. If it is very extreme then it will be apparent within two weeks in South Africa, otherwise it will only be known in 6 to 8 weeks. Neutralization assays will provide laboratory data on immune escape in the mean-time. Estimates of the rest that are circulating sooner than the above timelines are unlikely to be accurate.
It is rare for a single question to dominate human discourse as thoroughly as this one currently does: “how bad is the Omicron variant?” There are two parts to that question, namely “how much more transmissible is Omicron than Delta?” and “how severe is the illness caused by Omicron?”
Frustratingly, it is not yet possible to answer either of those questions, even a little bit. In principle, the South African outbreak could have been informative by this point. Omicron has already become dominant in South Africa and typically, by looking at the pace at which one variant replaces another allows one to infer how much of a transmission advantage it has.
However, South Africa’s genomic surveillance, which has in the past been praised for its high quality, was in something of a quiet phase during the lead-up to Omicron’s emergence, having submitted just 44 sequences in 30 days. This low sampling rate is compounded by a systemic sampling bias present in all of SA’s genomic data. The majority of the country’s PCR tests are privately administered, at a cost that is prohibitive to the majority of South Africans. Moreover, the legal requirement to isolate in the case of a positive test imposes a severe financial disincentive for testing, in the form of loss of income.
As a consequence, there is no solid time-line of representative genomic sampling, and it is not possible to apply the calculations used to determine transmission advantage. People have tried, to frankly ludicrous results.
SA’s current dataset is unusable for determining Omicron’s transmission advantage and, because Omicron has pretty much entirely replaced Delta already, our data going forward will not be usable for that purpose either.
As a result we will only be able to determine that transmission advantage once data comes in from countries with more representative sampling. European countries are the most likely candidates for that; the UK sequences something like 400 samples a day, including a random. However, Omicron has only just been detected there in small numbers, and their pre-existing case-load is enormous, which delays the time taken for replacement by a new variant, so it will take some amount of time for solid data to emerge. The larger Omicron’s advantage, the faster that will occur so let us hope that we are kept waiting. However, if the replacement is not absurdly rapid, that would exclude the more outlying extremes of transmission advantage. Essentially, each week that passes will narrow down the possible range for that advantage. If the advantage is similar to the advantage that Delta had over prior variants then it then it will take somewhere in the region of 3 months to overtake Delta, but the growth trajectory will be somewhat apparent before then.
Likely, a fairly good estimate of Omicron’s transmission advantage will only be available about 4 to 6 weeks from now.
The severity of illness arising from Omicron will also be a bit tricky to extract from South Africa’s data. South Africa’s excess deaths throughout the pandemic have been in the region of triple our official COVID deaths, suggesting that many of our severely ill never receive formal healthcare. This in turn makes mortality and severity of illness difficult to track on a population level. This is compounded by the lack of a randomized testing pool, which imposes a selection bias on our testing because the likelihood of someone testing is somewhat proportional to the severity of their symptoms, which places severity of symptoms in both the numerator and denominator of a calculation relating hospitalizations to cases.
This creates a large margin of uncertainty when assessing severity of symptoms. We will therefore only be able to confidently pick up any difference in severity if that difference is very large, which is somewhat unlikely. If there IS a large difference either way, it will start to become apparent in South Africa’s hospitalization data within about two weeks, once a large-ish base of confirmed cases has solidified and the typical lag-time between cases and hospitalizations has elapsed. If there is a smaller difference, within the range that can be obscured by the data issues mentioned above, then we will not know much about severity until data from a more representatively-sampled country is available. This will take slightly longer than finding out about its transmission advantage, because of the lag-time between cases and hospitalizations and so, solid reliable data will take about 6 to 8 weeks.
As with transmission, people have jumped the gun on the topic of severity. One clinician’s anecdotal observation that Omicron has milder symptoms has sprinted around the world, amplified by major news outlets and social media influencers. Many outlets, including have taken pains to mention that the clinician, Angelique Coetzee, is the chair of the South African Medical Association (SAMA), but none have made the effort to point out that SAMA is a trade union and not a medical advisory body of any sort.
Media outlets presumably know this, but their audiences predominantly does not. The BBC has even gone so far as crediting Coetzee with being the first to spot the variant, a rather odd assertion which has been frequently repeated elsewhere. Cynically, one might presume that these media outlets have deliberately exploited these misunderstandings and amplified an unsubstantiated anecdote to drive traffic. If you have any doubts about how dangerous this is, see the screenshot below:
The completely unsubstantiated assertion that Omicron is more mild will doubtless have affected the behaviour of tens of thousands of people by now. Lives are likely to be lost as a result.
The other burning topic is the extent of immune escape. Experts have warned that the set of mutations held by Omicron are consistent with immune escape so this is another issue to watch closely.
The data required for evaluating real-world immune escape is, broadly speaking, the same as for severity, just further granulated by vaccination status relative to population vaccination status. This means that it will be known at roughly the same time as severity, possibly at a slight delay because of lower respective sample sizes for the vaccinated and unvaccinated cohorts. However, early indicators of immune escape can be obtained rapidly in a laboratory setting, by way of neutralization assays. Immune escape, therefore, will likely be the first aspect of Omicron that we will get data on, but real-world corroboration will take some time.
