Time lag between infection and hospitalization.
In addition to Great Britain and South Africa, Norway and Denmark have a high proportion of Omicron COVID-19 cases and a rapidly increase new-case count.
In the prior post, I looked at the data from Great Britain and South Africa and concluded that Omicron results in cases with far less average severity that Delta did. For example, in Great Britain, the current case hospitalization rate for Omicron patients seems to be about 0.3%, versus the 1.7% observed for the prior Delta wave.
In this post, I’m going to look at the case hospitalization rates for Norway and Denmark. But first, I need to clear up some misleading verbiage.
As with Great Britain and South Africa, all the official sources in Norway and Denmark appear to downplay the apparent lesser severity of Omicron relative to Delta. At present, everyone is urging extreme caution. And, I suppose, as public health agencies, that’s their job.
I continue to see this phrase repeated: “There’s a lag between infection and hospitalization, so we can’t make much of the low hospitalization rates yet”. This is the universal excuse for not paying attention to the observed case hospitalization rate for Omicron to date.
This statement is both true and not true. But first, let’s assess the lag between the count of diagnosed cases, and hospitalizations, using U.S. data.
Source: CDC COVID data tracker, accessed 12/18/2021. Notes and marks are mine.
I think it is reasonably clear, by eye, that in the U.S., there is essentially no lag between new case counts and new hospitalizations. The peaks and troughs in one curve are mirrored in the other, almost to the day.
I can now show you a counterexample — deaths — where there’s an average of about a two week lag, owing to the median time from diagnosis to death for COVID-19 decedents. Not that in the graph below, the peaks and troughs don’t coincide, but deaths generally lag new cases. You can simply look at it, and see that it looks like the orange line is shifted to the right, relative to the maroon line.
Source: Same as above.
So, yes, there is a lag between infection and hospitalization. Absolutely. I would guess that it is, on average, roughly the amount of time it takes for symptoms to emerge, plus a day or to to see that those symptoms are serious. So, perhaps six days on average?
But the point is, that lag between infection and (reporting of) hospitalization is more-or-less the same length as the lag between infection and reporting of that infection. With the net result being that the new cases, as reported, and the new hospitalizations, as reported, are more-or-less perfectly in sync.
And so, if my goal is to calculate the case hospitalization rate — the fraction of new cases that get hospitalized — it’s perfectly legitimate to take these early numbers and do that. There’s no excess lag in hospital reporting that would result in my under-stating the case hospitalization rate. (At least, not in the U.S. data.)
More to the point, the presence of some lag between infection and hospitalization is not a legitimate excuse to ignore the early hospitalization numbers. At least, not in any country where data reporting is at least as good as it is in the U.S.A.
There is, in fact, a lag between infection and hospitalization. We haven’t yet seen all the hospitalizations that will arise from cases infected today. But, equally, we haven’t see the positive COVID-19 tests reported, from individuals who were infected today. And, by and large, we’ll see the hospitalizations just about the same time we see those positive tests.
So there is a lag, but that’s not a legitimate excuse for ignoring the early estimates of case hospitalization rate.
Finally, it is completely true that (e.g.) the use of ventilators will lag the hospital admission rate. As a case progresses from admission to (say) death, it takes some time for them to get bad enough to get onto the ventilator. Empirically, because that ventilator step lies somewhere between admission and death, the corresponding times series data will have a lag somewhere between (roughly) zero (admission) and two weeks (death). So I would not make use of preliminary data on ventilator dependence in the early hospitalization data to argue that hospitalized cases are particularly mild.
Early estimates of Omicron case hospitalization rates for Norway and Denmark.
First, there are the anecdotes.
Norway already had one Omicron superspreader event, and it made the news that despite 150 cases, there were no hospitalizations. Officials of course immediately dismissed this as due to age, good heath, vaccination rate, and so on. The lack of hospitalizations was also noted in a scholarly analysis of the incident.
In terms of systematic data, I’m having a hard time finding any one system that reports both cases and hospitalizations for known Omicron-infected individuals.
For Denmark, as of Wednesday, this news reporting puts total known Omicron cases as 6047, and total known hospitalizations of Omicron cases at 10, based on a statement from the Danish government agency (SSI) that tracks this. That would be an Omicron case hospitalization rate of 0.2%. That’s not hugely different from the 0.3% that seems to be holding in Great Britain (see just prior post).
For Norway, all I can find is a government projection. In an assessment dated 12/13/2021, the Norwegian Institute of Public Health projected cases and hospitalizations. The Government of Norway published a summary of the results here. “According to a preliminary scenario from the Norwegian Institute of Public Health, the forecast is that the number of cases could rise to 90 000–300 000 per day 3 weeks from now, and 50–200 hospital admissions per day if the measures do not adequately slow down developments in the pandemic.” Those numbers imply an expected case hospitalization rate of well under 0.1%.
What I don’t have for either of these is the case hospitalization rate for Delta, which still accounts for the majority of cases in both countries. (That is, maybe somehow they just don’t hospitalize COVID cases).
Unfortunately, a lot of European countries track the number of people actively in the hospital, but not the total ever hospitalized.
That said, even a cursory glance at the Denmark hospitalization data show that they hospitalize quite heavily. At the peak of the 2020 winter wave, they had almost 1000 persons in the hospital with COVID-19 (via this source, which I believe is the Government of Denmark). At the peak of that winter wave, they were only getting about 3500 new infections per day.
Maybe I just keep finding the information I expect to find. Maybe I’m just missing all the reporting showing the hospitalization rates closer to what would have been expected under Delta.
But so far, every place I look, it’s the same story for Omicron infections.
- Lots of cases, growing rapidly
- Very few hospitalizations
- Vanishingly few deaths.
- Complete dismissal of the notion that Omicron is materially less severe, by every public health agency.
What I find most annoying is that with the exception of South Africa, in all these cases, we are looking at the same country, same population, and almost the same time period, for Delta a few weeks ago, compared to Omicron today. (In South Africa, the time gap is closer to six months for their comparisons).
And even though it’s the same country, same population, and almost the same time, when faced with the dramatically lower observed hospitalization rate of Omicron, public health officials uniformly point to factors like vaccination rates and immunity from prior infection as the explanation.
Well, no. To a close approximation, all of those factors were the same under Delta two weeks ago, as they are under Omicron today. To a close approximation, the only thing that has changed is the virus. If there is something like an order-of-magnitude reduction in the hospitalization rate, that’s far more likely to be due to the virus than to those other, nearly-constant factors.
And so, on this issue, I find myself in a position contrary to mainstream thinking, and aligned with the nutsos. It still looks to me as if Omicron is vastly less virulent than Delta.
So much so that my prediction is that Great Britain’s hospitalization rates should now start falling, even as total COVID new case counts are rising. That’s the next thing I’ll be on the lookout for.