Methods for early characterisation of the severity and dynamics of SARS-CoV-2 variants: a population-based time series analysis in South Africa

Lancet Microbe. 2022 Oct;3(10):e753-e761. doi: 10.1016/S2666-5247(22)00182-3. Epub 2022 Aug 31.

Abstract

Background: Assessment of disease severity associated with a novel pathogen or variant provides crucial information needed by public health agencies and governments to develop appropriate responses. The SARS-CoV-2 omicron variant of concern (VOC) spread rapidly through populations worldwide before robust epidemiological and laboratory data were available to investigate its relative severity. Here we develop a set of methods that make use of non-linked, aggregate data to promptly estimate the severity of a novel variant, compare its characteristics with those of previous VOCs, and inform data-driven public health responses.

Methods: Using daily population-level surveillance data from the National Institute for Communicable Diseases in South Africa (March 2, 2020, to Jan 28, 2022), we determined lag intervals most consistent with time from case ascertainment to hospital admission and within-hospital death through optimisation of the distance correlation coefficient in a time series analysis. We then used these intervals to estimate and compare age-stratified case-hospitalisation and case-fatality ratios across the four epidemic waves that South Africa has faced, each dominated by a different variant.

Findings: A total of 3 569 621 cases, 494 186 hospitalisations, and 99 954 deaths attributable to COVID-19 were included in the analyses. We found that lag intervals and disease severity were dependent on age and variant. At an aggregate level, fluctuations in cases were generally followed by a similar trend in hospitalisations within 7 days and deaths within 15 days. We noted a marked reduction in disease severity throughout the omicron period relative to previous waves (age-standardised case-fatality ratios were consistently reduced by >50%), most substantial for age strata with individuals 50 years or older.

Interpretation: This population-level time series analysis method, which calculates an optimal lag interval that is then used to inform the numerator of severity metrics including the case-hospitalisation and case-fatality ratio, provides useful and timely estimates of the relative effects of novel SARS-CoV-2 VOCs, especially for application in settings where resources are limited.

Funding: National Institute for Communicable Diseases of South Africa, South African National Government.

MeSH terms

  • COVID-19* / epidemiology
  • Communicable Diseases* / epidemiology
  • Humans
  • Middle Aged
  • SARS-CoV-2 / genetics
  • South Africa / epidemiology
  • Time Factors

Supplementary concepts

  • SARS-CoV-2 variants