A new retrospective analysis of the first 250,000 hospitalized individuals with SARS-COV-2 infection (>20 years) across Brazil may provide some possible explanation for the surge in Manaus. https://www.thelancet.com/journals/lanres/article/PIIS2213-2600(20)30560-9/fulltext

The North (Manaus is located in Northern state of Amazonas) and northeastern states of Brazil are younger than other parts of the country. Moreover, the N and NE states have greater concentration of healthcare capacity in capital cities.

The study looked at general ward inpatients, ICU inpatients and ventilated individuals. However, across all categories higher mortality rate in the 20-39, 40-49, 50-59 and 60-69 age groups is seen in both the north and northeast as compared to other parts of the country.

Clinical evidence suggest patients in the N and NE arrive hospital in worse shape than others from other parts of the country. This likely impacts survival rates among this cohort.

In a similar vein, General hospital mortality nationwide was 38% but this varied significantly by region

- 50% in the North
- 48% in the Northeast
- 35% in the Midwest
- 34% in the Southeast
- 31% in the South

Hospitalization per 100k suggests younger patients in the N and NE were hospitalized at lower rates than other areas (apart from the South) but HC resource use is high both in the North and the older & densely populated SE. Signs of younger cohorts delaying going to hospital?

However, most of the 1st wave in the N and NE came earlier in the pandemic so some patient outcome may have been affected by fewer amounts of treatment options at the time.

Overall, as espoused in an earlier thread, a higher CFR in areas with patient overload/fewer healthcare resources could result in significant overestimation in back calculating seroprevalence from mortality data. https://twitter.com/CovidSerology/status/1349639190955270144

This suggests caution is warranted in direct interpretations of possible community prevalence from age-adjusted calculations from mortality data in Manaus (and N & NE Brazil in general).

Depending on how high CFR are in these settings. Demographic advantages conferred by a high populations may be cancelled out resulting in indistinguishable IFR between diff. populations. E.g Manaus could have similar IFR to much older Geneva/Sao Paulo https://twitter.com/CovidSerology/status/1349639188736425986

Due acknowledgment to @otavio_ranzani and his team for this very important study and hopefully this study helps shed more light on the evolving nature of the epidemic in Brazil (especially second waves in parts of the North and Northeast).