tetano
Editor, Senior Moderator
Influenza Other Respir Viruses
. 2024 Oct;18(10):e70026.
doi: 10.1111/irv.70026. Extrapolating Sentinel Surveillance Information to Estimate National COVID Hospital Admission Rates: A Bayesian Modeling Approach
Owen Devine[SUP] 1 [/SUP], Huong Pham[SUP] 2 [/SUP], Betsy Gunnels[SUP] 3 [/SUP], Heather E Reese[SUP] 4 [/SUP], Molly Steele[SUP] 4 [/SUP], Alexia Couture[SUP] 2 [/SUP], Danielle Iuliano[SUP] 2 [/SUP], Darpun Sachdev[SUP] 5 [/SUP], Nisha B Alden[SUP] 6 [/SUP], James Meek[SUP] 7 [/SUP], Lucy Witt[SUP] 8 [/SUP], Patricia A Ryan[SUP] 9 [/SUP], Libby Reeg[SUP] 10 [/SUP], Ruth Lynfield[SUP] 11 [/SUP], Susan L Ropp[SUP] 12 [/SUP], Grant Barney[SUP] 13 [/SUP], Brenda L Tesini[SUP] 14 [/SUP], Eli Shiltz[SUP] 15 [/SUP], Melissa Sutton[SUP] 16 [/SUP], H Keipp Talbot[SUP] 17 [/SUP], Isabella Reyes[SUP] 18 [/SUP], Fiona P Havers[SUP] 2 [/SUP]
Affiliations
The COVID-19-Associated Hospitalization Surveillance Network (COVID-NET) was established in March 2020 to monitor trends in hospitalizations associated with SARS-CoV-2 infection. COVID-NET is a geographically diverse population-based surveillance system for laboratory-confirmed COVID-19-associated hospitalizations with a combined catchment area covering approximately 10% of the US population. Data collected in COVID-NET includes monthly counts of hospitalizations for persons with confirmed SARS-CoV-2 infection who reside within the defined catchment area. A Bayesian modeling approach is proposed to estimate US national COVID-associated hospital admission rates based on information reported in the COVID-NET system. A key component of the approach is the ability to estimate uncertainty resulting from extrapolation of hospitalization rates observed within COVID-NET to the US population. In addition, the proposed model enables estimation of other contributors to uncertainty including temporal dependence among reported COVID-NET admission counts, the impact of unmeasured site-specific factors, and the frequency and accuracy of testing for SARS-CoV-2 infection. Based on the proposed model, an estimated 6.3 million (95% uncertainty interval (UI) 5.4-7.3 million) COVID-19-associated hospital admissions occurred in the United States from September 2020 through December 2023. Between April 2020 and December 2023, model-based monthly admission rate estimates ranged from a minimum of 1 per 10,000 population (95% UI 0.7-1.2) in June of 2023 to a highest monthly level of 16 per 10,000 (95% UI 13-19) in January 2022.
Keywords: Bayesian modeling; COVID‐19; Population‐based surviellance; hospitalization; sentinel surviellance.
. 2024 Oct;18(10):e70026.
doi: 10.1111/irv.70026. Extrapolating Sentinel Surveillance Information to Estimate National COVID Hospital Admission Rates: A Bayesian Modeling Approach
Owen Devine[SUP] 1 [/SUP], Huong Pham[SUP] 2 [/SUP], Betsy Gunnels[SUP] 3 [/SUP], Heather E Reese[SUP] 4 [/SUP], Molly Steele[SUP] 4 [/SUP], Alexia Couture[SUP] 2 [/SUP], Danielle Iuliano[SUP] 2 [/SUP], Darpun Sachdev[SUP] 5 [/SUP], Nisha B Alden[SUP] 6 [/SUP], James Meek[SUP] 7 [/SUP], Lucy Witt[SUP] 8 [/SUP], Patricia A Ryan[SUP] 9 [/SUP], Libby Reeg[SUP] 10 [/SUP], Ruth Lynfield[SUP] 11 [/SUP], Susan L Ropp[SUP] 12 [/SUP], Grant Barney[SUP] 13 [/SUP], Brenda L Tesini[SUP] 14 [/SUP], Eli Shiltz[SUP] 15 [/SUP], Melissa Sutton[SUP] 16 [/SUP], H Keipp Talbot[SUP] 17 [/SUP], Isabella Reyes[SUP] 18 [/SUP], Fiona P Havers[SUP] 2 [/SUP]
Affiliations
- PMID: 39440677
- PMCID: PMC11497105
- DOI: 10.1111/irv.70026
The COVID-19-Associated Hospitalization Surveillance Network (COVID-NET) was established in March 2020 to monitor trends in hospitalizations associated with SARS-CoV-2 infection. COVID-NET is a geographically diverse population-based surveillance system for laboratory-confirmed COVID-19-associated hospitalizations with a combined catchment area covering approximately 10% of the US population. Data collected in COVID-NET includes monthly counts of hospitalizations for persons with confirmed SARS-CoV-2 infection who reside within the defined catchment area. A Bayesian modeling approach is proposed to estimate US national COVID-associated hospital admission rates based on information reported in the COVID-NET system. A key component of the approach is the ability to estimate uncertainty resulting from extrapolation of hospitalization rates observed within COVID-NET to the US population. In addition, the proposed model enables estimation of other contributors to uncertainty including temporal dependence among reported COVID-NET admission counts, the impact of unmeasured site-specific factors, and the frequency and accuracy of testing for SARS-CoV-2 infection. Based on the proposed model, an estimated 6.3 million (95% uncertainty interval (UI) 5.4-7.3 million) COVID-19-associated hospital admissions occurred in the United States from September 2020 through December 2023. Between April 2020 and December 2023, model-based monthly admission rate estimates ranged from a minimum of 1 per 10,000 population (95% UI 0.7-1.2) in June of 2023 to a highest monthly level of 16 per 10,000 (95% UI 13-19) in January 2022.
Keywords: Bayesian modeling; COVID‐19; Population‐based surviellance; hospitalization; sentinel surviellance.