1. A few weeks ago, several Italian doctors made quite a stir in the news. They claimed that the SARS-CoV-2 coronavirus was "losing potency", and hinted at genetic changes in the virus.

2. I was skeptical of that explanation at the time, and wrote an extended thread to explain why. https://twitter.com/CT_Bergstrom/status/1267576233631469568

3. Yesterday, two papers were released that seem to present the data these claims were based upon. The first was published by Fabio Ciceri and a team including Alberto Zangrillo, who was interviewed for the previous stories.

https://www.tandfonline.com/doi/pdf/10.1080/20477724.2020.1785782?needAccess=true

4. The data presented there are very simple: hospital admissions and death rates at San Raffaele Hospital in Milano, Italy.

The result: death rate has declined over time, even correcting for patient ages.

5. The authors present a number of possible explanations. These include

i. better understanding of disease and patient care,
ii. reduced rates of influenza coinfection,
iii. social distancing and respiratory protective devices, presumably affecting innoculum size and location,

6.

iv. changes in pollution emissions
v. genetic changes in the virus

These all seem plausible to varying degrees. I don't put a lot of weight on the genetic change story compared to some of the other alternatives.

7. The authors don't mention another possibility that I wonder about: the admission criteria over time. If in May patients were more likely to be admitted with less severe symptoms than in March this could explain a portion of the effect seen in the graph above.

8. The second (and more interesting) paper was written by Nicola Clementi and colleagues. I do not know if the full text is publicly available yet, but the abstract is here. https://www.researchgate.net/publication/341901435_Lower_nasopharyngeal_viral_load_during_the_latest_phase_of_COVID-19_pandemic_in_a_Northern_Italy_University_Hospital

9. In this paper, they note the change in disease severity and test the hypothesis that it may be related to changes in viral load. They measure viral load (as cycle threshold, Ct) from 100 NP swabs taken in April, and compare that to 100 NP swabs taken in May.

10. The authors find a difference in viral load across the two periods, which seems to be driven largely by the fact that the higher loads seen in the first period (April) are rare during the second (May) period.

11. The authors mention consider a range of possible explanations, including those described in the previous paper, but are not able to distinguish among them.

But the key finding is interesting: viral load seems to have changed, and that correlates with disease severity.

12. They suggest monitoring viral load over time, which seems sensible to me. I'm partial to the explanation that PPE and other hygenic measures have reduced innoculum sizes and/or changed innoculum location (nasophraynx vs lung, for example).

13. Ultimately what these two papers give us is evidence of reduced disease severity in Lombardy as the spring progressed. But they don't give us evidence for the claims in the late May news articles. They don't provide evidence that genetic changes in the virus are responsible.