Skip to main content
. 2022 May 10;28(7):1476–1485. doi: 10.1038/s41591-022-01807-1

Table 2.

Temporal fluctuations in COVID-19-attributable in-hospital fatality rate and avoidable COVID-19-attributable deaths in hospitals

Location Observation period COVID-19 attributable hospital admissions in unvaccinated residents Age-standardized weekly COVID-19 in-hospital fatality rate (HFR) Estimated avoidable COVID-19 deaths in hospitals
Total Deathsa Lowest (%)b Highest (%)b Fold increaseb Assuming the lowest HFR in each city (%)c Assuming the lowest HFR across all 14 cities (%)c
Belo Horizonte 06/04/20–26/07/21 42,113 7,692 7.7 12.7 1.64 29.1 (24.6–33.7) 29.1 (24.6–33.7)
Curitiba 02/03/20–26/07/21 33,015 7,535 8.1 18.1 2.23 39.6 (34.2–44.7) 47.3 (43.9–50.8)
Florianópolis 09/03/20–26/07/21 4,032 893 7.8 17.1 2.20 22.9 (12.5–32.6) 44.5 (40.9–48.1)
Goiânia 16/03/20–26/07/21 20,044 6,246 11.6 25.9 2.23 44.7 (39.0–50.0) 61.2 (58.6–63.7)
João Pessoa 09/03/20–26/07/21 10,552 3,636 15.3 29.7 1.94 18.8 (13.0–25.1) 65.9 (63.7–68.1)
Macapá 30/03/20–26/07/21 3,169 1,011 11.0 41.5 3.78 41.6 (30.5–51.8) 68.4 (66.2–70.6)
Manaus 24/02/20–26/07/21 26,260 10,168 16.4 33.4 2.04 39.3 (35.9–42.3) 70.9 (69–72.9)
Natal 16/03/20–26/07/21 9,344 3,512 11.4 32.9 2.87 35.4 (30.2–40.5) 66.9 (64.8–69)
Porto Alegre 02/03/20–26/07/21 16,106 5,266 8.6 27.1 3.16 41.8 (37.7–45.7) 59.6 (57.1–62.3)
Porto Velho 30/03/20–26/07/21 6,795 2,473 11.5 32.2 2.79 39.1 (32.2–45.5) 70.3 (68.3–72.4)
Rio de Janeiro 16/03/20–26/07/21 73,139 28,053 19.2 26.0 1.35 10.2 (8.5–11.9) 66.9 (64.8–69)
Salvador 16/03/20–26/07/21 26,964 8,509 9.7 26.9 2.76 20.8 (16.2–25.2) 61.3 (58.8–63.9)
São Luís 24/02/20–26/07/21 8,545 2,547 8.2 26.5 3.25 35.7 (28.4–42.4) 60.5 (57.9–63.1)
São Paulo 20/01/20–26/07/21 182,288 42,769 8.8 19.8 2.24 36.2 (33.9%-–38.4) 50.1 (46.9–53.4)
All 14 cities 20/01/20–26/07/21 462,366 130,317 29.8 (28.7–30.9) 57.1 (54.3–59.9)

aObserved deaths plus expected deaths in COVID-19-attributable hospital admissions with unreported outcome.

bAge-specific COVID-19-attributable in-hospital fatality rates were estimated from linked data on underreporting-adjusted deaths in COVID-19-attributable hospital admissions. Non-parametric loess estimates were obtained and weighted by the population age composition across cities. Lowest fatality rates were calculated in the period before Gamma’s first detection in each location, and highest fatality rates were calculated including the time after Gamma’s first detection. The lowest fatality rates in the period before Gamma’s first detection agreed with those observed across the entire study period for all cities except Belo Horizonte. See the main text for further details and analyses.

cEstimates are based on hypothetical scenarios evaluated under the Bayesian multi-strain fatality model, assuming the lowest observed in-hospital fatality rates seen in the periods before Gamma’s first detection in each location.