Estimating vaccine efficacy against COVID-like illness from omicron and other COVID-19 variants

By Colin Lightfoot, M.Sc. Infection and ImmunityReviewed by Aimee MolineuxDec 21 2021

Since the omicron variant of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was initially identified in November 2021, new illnesses have increased substantially in South Africa, despite increased usage of the coronavirus disease 2019 (COVID-19) vaccination.

The Gauteng province has seen the majority of confirmed cases. Omicron carries several mutations that could boost transmissibility while also lowering immune defense. Initial data indicate that when exposed to the omicron variant of the SARS-CoV-2 virus, there is a higher chance of reinfection.

These early claims, however, are based on limited sample sizes. Furthermore, a recent large-scale review of COVID-19 test findings in South Africa reveals that the current COVID-19 vaccines may not be as effective against the omicron variant as previously thought.

In the case of novel variations like omicron, preliminary estimates of vaccination efficacy as a contributor to variant emergence and dominance are needed, especially in the early stages of national and international transmission.

*Important notice: medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.

In a new study, researchers combined real-time syndromic surveillance data from daily cross-sectional surveys with a COVID-like illness (CLI) prevalence-based vaccine efficacy estimator to better explain the early onset of the omicron variant in Gauteng, South Africa (V ECLIP ).

A preprint version of this study, which is yet to undergo peer review, is available on the medRxiv* server.

The study

The authors initially identified self-reported symptoms that looked to be rising during the delta and omicron waves to characterize the constellation of symptoms that can be utilized to identify COVID-19 in Gauteng. These symptoms included cough, fever, and muscle aches and pains. These symptoms were accompanied by a loss of smell or taste, which has been persistently linked to COVID-19 infection throughout the pandemic and was a consistently strong predictor across countries, time, and syndromic surveillance platforms.

Stringent CLI (anosmia with fever, cough, and/or myalgias), classic CLI (cough with anosmia, fever, cough, and/or myalgias), and broad CLI (myalgias with anosmia, fever, cough, and/or myalgias) were the three criteria used. The authors used 7-day moving averages to show how the prevalence among respondents to UMD-CTIS cross-sectional surveys using these CLI definitions correlates with COVID-19 instances from official case count data.

The findings imply that the conventional CLI criteria (cough plus at least one of the following symptoms: fever, muscle pain, or anosmia) is consistent with reported cases. This was true during the delta wave, when the Delta variant was dominating, and during the first two weeks of December, when Omicron infections were on the rise. Furthermore, the authors discovered that the strict CLI definition, which is intended to be more specific to COVID-19 but relies on a less commonly reported symptom (anosmia), has a weaker signal during the omicron wave period of infection than both the classic and broad CLI definitions, which include more commonly reported symptoms. Because Omicron cases are on the rise in Gauteng right now, the signal-to-noise ratio is shifting.

The authors compared CLI among the vaccinated and unvaccinated, during the delta and omicron wave periods, because the three syndromic surveillance signals based on the CLI definitions tracked with COVID-19 cases and were thus suitable proxies for COVID-infection. Vaccinated persons were defined as those who self-reported receiving two doses of vaccination, whereas unvaccinated individuals were defined as those who self-reported receiving no vaccine. Using the CLI prevalence among vaccinated and unvaccinated survey respondents to derive V ECLIP, the authors may use traditional cohort study theory to approximate the attack rate of COVID-19 for both vaccinated and unvaccinated persons.

It is important to note that the V ECLIP estimate is not equivalent to the true vaccine efficacy because the key assumptions are made up: (1) the authors assumed that self-reported prevalent CLI is a valid proxy for incident COVID-19 infection, but breakthrough infections can be asymptomatic, and not all CLI is test-positive COVID19; (2) the authors assumed a simplified model of unvaccinated and vaccinated sub-cohorts, but they did not have information on natural immunity among the unvaccinated, nor on vaccine formulation or timing for the vaccinated; (3) the authors assumed that the delta and omicron variants were the only variants during their respective study periods, and did not account for other variants or viral co-circulation; (4) the authors assumed that the survey sampling of the study base (i.e. active Facebook users) was consistent across study periods, non-differential in terms of symptoms, vaccination status, and/or potential vaccine efficacy modifiers.

If these assumptions are incorrect, the estimator may be biased either upwards or downwards (e.g. omicron infections asymptomatic or cough unrelated to COVID-19, respectively). As a result, these V ECLIP estimates should be treated as preliminary assessments that can be used to compare relative changes in V ECLIP, thereby alerting to a potential loss in vaccination efficacy with the development of the Omicron form in Gauteng, South Africa, or elsewhere.

Implications

While the syndromic surveillance informed proxy of vaccine efficacy may not be the gold standard for estimating vaccine efficacy, the authors have demonstrated that using remote, real-time CLI trends survey respondents over time, analysis of regional data from UMD-CTIS can quickly point to high-level changes. Nonetheless, this preliminary research only outlines a small number of questions that can be posed utilizing UMD-CTIS to better understand the epidemiological profile of Omicron in Gauteng, South Africa, and globally. As we continue to deal with SARS-CoV-2 variants, as well as seasonal and emerging infectious illness in general, it's critical that we combine lab-based findings with early findings from syndromic monitoring data streams.

*Important notice: medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.