A quantitative assessment of epidemiological parameters required to investigate COVID-19 burden

Abstract

Solid estimates describing the clinical course of SARS-CoV-2 infections are still lacking due to under-ascertainment of asymptomatic and mild-disease cases. In this work, we quantify age-specific probabilities of transitions between stages defining the natural history of SARS-CoV-2 infection from 1965 SARS-CoV-2 positive individuals identified in Italy between March and April 2020 among contacts of confirmed cases. Infected contacts of cases were confirmed via RT-PCR tests as part of contact tracing activities or retrospectively via IgG serological tests and followed-up for symptoms and clinical outcomes. In addition, we provide estimates of time intervals between key events defining the clinical progression of cases as obtained from a larger sample, consisting of 95,371 infections ascertained between February and July 2020. We found that being older than 60 years of age was associated with a 39.9% (95%CI: 36.2-43.6%) likelihood of developing respiratory symptoms or fever ≥ 37.5 °C after SARS-CoV-2 infection; the 22.3% (95%CI: 19.3-25.6%) of the infections in this age group required hospital care and the 1% (95%CI: 0.4-2.1%) were admitted to an intensive care unit (ICU). The corresponding proportions in individuals younger than 60 years were estimated at 27.9% (95%CI: 25.4-30.4%), 8.8% (95%CI: 7.3-10.5%) and 0.4% (95%CI: 0.1-0.9%), respectively. The infection fatality ratio (IFR) ranged from 0.2% (95%CI: 0.0-0.6%) in individuals younger than 60 years to 12.3% (95%CI: 6.9-19.7%) for those aged 80 years or more; the case fatality ratio (CFR) in these two age classes was 0.6% (95%CI: 0.1-2%) and 19.2% (95%CI: 10.9-30.1%), respectively. The median length of stay in hospital was 10 (IQR: 3-21) days; the length of stay in ICU was 11 (IQR: 6-19) days. The obtained estimates provide insights into the epidemiology of COVID-19 and could be instrumental to refine mathematical modeling work supporting public health decisions.

Keywords: Contact tracing data; Disease burden; Epidemiological parameters; Risk outcomes; SARS-CoV-2.

Fig. 1

A Schematic representation of transition probabilities characterizing possible disease outcomes after SARS-CoV-2 infection. These include the symptomatic ratio (SR), the ratio of critical cases (CR), the case (CFR) and infection fatality ratios (IFR) and similar quantities that could be estimated using ascertained symptomatic infections (asCR, asCFR) as the set of exposed individuals. B Schematic representation of transition probabilities characterizing the hospital (HR) and ICU (IR) admission among infected individuals, and of similar quantities that could be estimated using ascertained symptomatic infections (asHR) or hospital patients (hCFR, hIR) as the set of exposed individuals. C Schematic representation of time to key events defining the temporal clinical progression of cases. D Schematic representation of the differences in the ascertainment rates associated with SARS-CoV-2 infections and symptomatic cases in the community and among close contacts of identified cases, with the latter representing individuals who were all tested for SARS-CoV-2 infection and daily monitored for symptoms during their quarantine or isolation period.

Fig. 2

A Comparison between the age distributions of critical cases as obtained when applying estimated risk outcomes to available serological records with the one observed in Lombardy during the first COVID-19 wave. B Comparison between the age distributions of deaths as obtained when applying estimated risk outcomes to available serological records with the one observed in Lombardy during the first COVID-19 wave and the one associated to deaths occurred in Italy between February 2020 and April 2021, as reported by the Integrated National Surveillance System (NSS).

Fig. 3

A Age-specific case hospital admission ratios among ascertained symptomatic cases (asHR). B Age-specific ICU admission ratios among hospitalized cases (hIR). Bars of different colors represent crude percentages observed across different epidemic periods; vertical lines represent 95% confidence intervals computed by exact binomial tests. Numbers shown in each panel represent the age-specific number of events observed in the data among exposed COVID-19 cases.