23% of COVID-19 deaths at US hospitals linked to surging caseloads

Nearly one in four deaths among inpatients with COVID-19 may be attributed to hospital strains caused by surging caseloads, according to a recent study in Annals of Internal Medicine.

“Several U.S. hospitals experienced major surges in their COVID-19 caseload during this pandemic,” Sameer S. Kadri, MD, MS, head of epidemiology in the critical care medicine department at the NIH’s Clinical Center, said in a video accompanying the study. “A key question remained, which is how much of the whopping death toll was associated with simply being admitted to surging hospitals, rather than due to the infection itself?”

Kadri and colleagues analyzed data on 144,116 individuals with COVID-19 who were admitted to one of 558 U.S. hospitals from March 1, 2020, to Aug. 31, 2020. The patients were discharged or had died by Oct. 31, 2020. The hospitals were part of the Premier Healthcare Database, which contains all payers and one-fifth of all hospitalizations in 48 states. The researchers also reviewed the dates and numbers of patients involved in COVID-19 case surges that occurred at the hospitals.

The researchers reported that of 144,116 inpatients, 78,144 (54.2%) were admitted to hospitals in the top surge index decile.

Of the entire cohort, 25,344 (17.6%) died. Although crude COVID-19 mortality dropped over time across all surge index strata, compared with nonsurging COVID-19 hospital-months (less than 50th surge index percentile), the researchers observed a correlation between mortality and increasing surge index. The adjusted OR was 1.11 (95% CI, 1.01-1.23) in the 50th to 75th surge index percentile; 1.24 (95% CI, 1.12-1.38) in the 75th to 90th surge index percentile; 1.42 (95% CI, 1.27-1.6) in the 90th to 95th surge index percentile; 1.59 (95% CI, 1.41-1.8) in the 95th to 99th surge index percentile; and 2 (95% CI, 1.69-2.38) in the greater than 99th surge index percentile.

The surge index was also linked to mortality across wards, ICUs and intubation, according to the Kadri and colleagues. The association was greater in June to August vs. March to May (slope difference = 0.1; 95% CI, 0.03-0.16), even with increasing corticosteroid use and “more judicious intubation,” the researchers wrote.

Overall, Kadri and colleagues estimated that 23.2% of the deaths were possibly attributable to hospitals strains.

“A truly humbling statistic,” Kadri said in the video. “Our findings have implications on triage, hospital preparedness, resource allocation and public health benchmarking.”

Sarah Warner, MPH, a senior data manager at NIH, said in the video that the impact of hospital surges “might have been even worse in other global regions, where existing health care infrastructure is limited at baseline.”

“We hope that our work will inform potential future threats posed by emerging variants of concern,” she said.