From my email from Ratufa:
I wanted to point out an issue with some of the metrics that are being used to assess the severity of Omicron.
The growth rate of an outbreak impacts the observed ratios of outcomes. Early in an outbreak those ratios will be biased towards lesser severity for faster spreading strains because more severe outcomes take longer to develop.
For example, it takes on average two days to be admitted to an ICU after hospital admission. The SA Omicron outbreak looks to have a growth rate of .21/day and the original Delta outbreak one of .1/day. Based on that we would expect the proportion of ICU admissions to hospital admissions to be ~20% lower [1-e^(-2*.11)] than Delta early in the pandemic. And incidental admissions have the potential to confound that number even more.
The impact on hospitalizations and deaths is more dramatic. Positive tests tend to lead hospitalizations by about 5 days and deaths by 2 to 3 weeks. So we would expect the ratio of hospitalizations to positive tests to be ~40% lower and deaths/positive tests to be ~80% lower than in the delta outbreak holding severity constant. Though both those estimates are quite sensitive to the lag and estimate of r.
The growth effect probably doesn’t explain the majority of the difference in outcomes that have been observed. But it is potentially material. And makes me more skeptical of claims of lesser severity I’ve seen so far.