Findings

Virus Watch preliminary findings on 14 December 2023

Virus Watch is a research study run by University College London and the NHS. The study aims to identify how COVID-19 spreads, and how to stop it. Findings so far presented on this page are early, preliminary results and should be interpreted with caution as they have not all yet been peer-reviewed by scientists external to our research collaborators. We are presenting these early findings for participants, the general public and policy makers. A detailed description of the Virus Watch study and our research questions can be found in our study protocol. Virus Watch began recruiting people in June 2020 and since then, 58,628 people in 28,527 households across England and Wales have joined the study. You can find a detailed description of the cohort in a profile of the study here.

Table of contents

  1. Summary of findings
  2. Tracking COVID-19 in England and Wales
  3. Long-Term Outcomes of SARS-CoV-2 Variants and Other Respiratory Infections
  4. Examining immune response to different SARS-CoV-2 variants and its association with the risk of SARS-CoV-2 infection

1. Summary of findings:

Tracking COVID-19 in England and Wales:

  • Conducting large-scale epidemiological studies to accurately estimate infection rate trends can be resource-intensive and costly. Many studies have been set up in the UK to monitor COVID-19 but variations in study designs may produce different results. It is therefore important to assess the validity of our study results to ensure their reliability.
  • We estimated the incidence rates of COVID-19 in England and Wales using Virus Watch data, and compared them with estimates from the ONS COVID-19 infection survey (CIS), the largest regular survey of COVID-19 infections and antibodies in the UK.
  • Virus Watch- and ONS- estimated incidence rates showed similar trends for England and Wales. However, Virus Watch-estimated peaks of infection during the Omicron BA.1 and 2 waves were lower than estimates from the ONS.
  • Our analysis indicates that the Virus Watch research approach is a low-cost and reliable method for COVID-19 surveillance even in the absence of national testing in the UK. Similar approaches can also be utilised by low-resource settings.  

Long-Term Outcomes of SARS-CoV-2 Variants and Other Respiratory Infections:

  • To understand Long Covid, we need to investigate which features of the SARS-CoV-2 virus impact the likelihood of developing the condition.
  • Using data from Virus Watch, we compared the likelihood of developing long-term symptoms after infection with different SARS-CoV-2 variants, and compared to other non-Covid infections and to people in the same time period who did not experience an infection.
  • People who were infected with earlier variants (Wild Type, Alpha, Delta, and Omicron BA.1) were more likely to develop long Covid (27-34%) compared to those who were infected with later sub-variants of Omicron (11-14%).
  • People who had Covid-19 or other non-Covid respiratory infections (8-23%) were more likely to develop long-term symptoms than people who did not have an infection during the same periods (1-3%). 
  • Variant is an important factor influencing how likely someone is to develop Long Covid, with the condition less likely following recent Omicron sub-variants. Having any respiratory infection increases the likelihood of developing long-term symptoms.

Examining immune response to different SARS-CoV-2 variants and its association with the risk of SARS-CoV-2 infection

  • With the proliferation of new SARS-CoV-2 variants, estimating antibody levels to gauge population-level immunity against such variants can play an important role in informing public health policy (we call this variant-specific serosurveillance).
  • Previous research focused on levels of antibodies that bind to the spike protein part of the virus as a measure of protection against infection, but these levels may not accurately predict infection risk with new variants. 
  • We used finger prick blood samples from Virus Watch participants to measure how well their antibodies disable the virus instead (we refer to this as virus inhibition). 
  • We then investigated the association between variant-specific virus inhibition and subsequent SARS-CoV-2 infections.
  • We conducted this study during the transition from the Delta to Omicron BA.1 variant dominance, allowing us to investigate this association within each variant’s dominance period separately.
  • We found that greater inhibition of Omicron BA.1 virus correlated with reduced infection risk during both Delta and Omicron BA.1 periods.
  • Delta virus inhibition was linked to reduced infection risk during its dominance period, but not during the Omicron BA.1 period.
  • These results align with earlier findings, suggesting that monitoring people’s antibody levels using home-based testing could be very useful in informing public health policy. Specifically, results of such testing could be used for targeted vaccination campaigns by identifying groups that are at the highest risk of infection.

You can find our previous findings on the links below.

You can also find a list of our publications here.

2. Tracking COVID-19 in England and Wales

The ONS CIS is a large longitudinal study which included a randomly selected and representative cohort of 227,797 households. Participants were incentivised to provide nose and throat swabs weekly, irrespective of symptoms, which were then tested for SARS-CoV-2. Between May 2021 and Match 2022, the study conducted around 390,300 PCR tests monthly. The study, which lasted three years (2020-2023), incurred costs totalling £988.5 million. 

Virus Watch is a community cohort study conducted in England and Wales with a spend of £4.89 million. Study participants provided us with dates and results of their COVID-19 test results via weekly surveys.  In order to minimize any missing results, we also linked to the NHS Test and Trace programme. Test results from both sources were from free PCR or lateral flow tests (LFTs) that became widely available through the national testing programme. Unlike the ONS CIS, Virus Watch study participants were not incentivised to take nose and throat swabs, and therefore, the majority of the testing conducted was when someone had symptoms of COVID-19.

Given the key differences in study design, we compared incidence rate estimates from Virus Watch to ONS CIS to evaluate the validity of Virus Watch results in capturing infection rates.  

Virus Watch identified a total of 30,031 COVID-19 cases in England and 570 in Wales between June 2020 and February 2023. We found similar trends between Virus Watch and ONS CIS COVID-19 incidence estimates for England and Wales, both with and without the incorporation of linked national testing data into the Virus Watch study. In particular, the magnitude and trend of Virus Watch- and ONS-estimated rates for England were generally consistent, although Virus Watch-estimated peaks of infection during the Omicron BA.1 and 2 waves were found to be lower than estimates from the ONS.

The Virus Watch approach, can serve as a low-cost and supplementary surveillance method during periods of national testing, while also allowing the consideration of multiple risk factors. In the absence of a comprehensive national testing programme, our study approach could produce acceptably accurate results with considerably lower costs compared to larger studies like the ONS CIS. Moreover, our approach is adaptable for implementation at a relatively low cost in both developed settings without routine testing and in low-resource settings. This will facilitate monitoring and response capabilities not only for COVID-19 but also other acute respiratory infectious diseases in the future.

Figure 1. ONS CIS and Virus Watch estimates of COVID-19 incidence rates in England from June 2020 to February 2023 with the dominant variant of concern of each period labelled. Virus Watch data in A) incorporates SGSS and Pillar 2 linked data, while B) excludes SGSS and Pillar 2 linked data. Shaded regions indicate the 95% confidence interval for incidence rate estimates. Dotted vertical lines indicate the periods of time when different COVID-19 variants were most common in the UK, and the red line indicates the ending of national free COVID-19 testing.

A graph of data on a white background

Description automatically generated

Figure 2. Virus Watch and ONS estimates of COVID-19 incidence rates in Wales from June 2020 to February 2023 with the dominant variant of concern of each period labelled. Shaded regions indicate the 95% confidence interval for incidence rate estimates. Dotted vertical lines indicate the periods of time when different COVID-19 variants were most common in the UK, and the red line indicates the ending of national free COVID-19 testing.

A graph of data source

Description automatically generated with medium confidence

Read the full analysis here.

3. Long-Term Outcomes of SARS-CoV-2 Variants and Other Respiratory Infections

Long Covid – long-term, often disabling symptoms that develop after SARS-CoV-2 infection – affect approximately 10-30% of people in the community with mild-moderate Covid and over 50% of people with severe Covid. Understanding aspects of the virus that influence the likelihood of developing long Covid is an important aspect of understanding this condition. 

Using data from Virus Watch, we investigated (1) whether different SARS-CoV-2 variants were associated with the likelihood of developing long Covid, and (2) whether there were differences in how likely someone was to develop new long-term symptoms after non-Covid respiratory infections and amongst people who did not have an infection (i.e., just affected by the pandemic stressors and other events at the time) compared to people with Covid infections. We accounted for the impact of different factors that might influence this relationship, including people’s demographic characteristics, health conditions, and vaccination status. This study focused on people’s first episode of Covid-19, because repeat infections are likely to interact with the immune system in a different way. 

We classified variants by the time period during which people were infected, and compared people who had Covid with people who had non-Covid infections and no infection. Periods for each variant were defined by national region, and ranged as follows: Wild Type (February-December 2020), Alpha (December 2020-May 2021), Delta (May– December 2021), Omicron BA.1 (December 2021 – March 2022), Omicron BA.2 (March– June 2022), Omicron BA.5 (July-November 2022), and Omicron other (other periods before March 2023, when multiple Omicron sub-variants were circulating).   Long Covid was defined as long-term symptoms that developed within three months of Covid infection and lasted at least two months, which is the World Health Organisation definition.

People who were infected during the time periods when the Wild Type (28%, 95% confidence interval (CI) 14-43%), Alpha (28%, 95% CI 14-42%), Delta (34%, 95% CI 25-43%), and Omicron BA.1 (27%, 95% CI 22-33%) variants were dominant were more likely to develop Long Covid compared to people infected during periods when later Omicron sub-variants were dominant (range between 11% (95% CI 8-15%) and 14% (95% CI 10-18%) (Figure 1). The risk of long-term symptoms in people who had non-Covid respiratory infections varied across different periods from 8% (95% CI 4-11%) to 23% (95% CI 18-28%). For the later forms of Omicron, the risk of long-term symptoms was similar whether you had Covid or another respiratory infection. If someone did not have any infection, their chance of having long-term symptoms was consistently low (range across time 1% (95% CI 0-2%) to 3% (95% CI 1-6%). 

Figure 1. Predicted Probabilities of Developing New Long-Term Symptoms by Variant Period and Infection Status

A screenshot of a graph

Description automatically generated

Figure 1 Note: The row at the top shows the predicted probability of developing long-term symptoms without accounting for any factors other than variant and infection type (SARS-CoV-2, other acute respiratory infection, or no infection), the one in the middle shows the likelihood after accounting for demographic and health-related factors, and the final row shows the likelihood after account for these and vaccination status. 

The findings from this study suggest that the variant that someone is infected with is an important factor that influences their likelihood of developing Long Covid. This relationship remains even after accounting for other factors like someone’s pre-existing health and their vaccination status. Encouragingly, more recent Omicron sub-variants seem to be associated with lower likelihood of developing Long Covid, though a notably proportion of people still develop long-term symptoms after both Omicron and other non-Covid respiratory infections. Having any respiratory infection was associated with a greater likelihood of new long-term symptoms, indicating an important role of infections themselves in the development of these symptoms. Further research into features of the SARS-CoV-2 virus and other respiratory viruses that influence how and why people develop long-term symptoms after infections is needed.

Read the full analysis here.

4. Examining immune response to different SARS-CoV-2 variants and its association with the risk of SARS-CoV-2 infection

Due to high population vaccination rates in the UK (up to 88% vaccinated with two doses), most COVID-19 cases are now likely caused by post-vaccine infections. In England and Wales in October 2023, over 41% of COVID-19 cases were reinfections. This is attributable to the spread of new SARS-CoV-2 variants, which can evade individuals’ immune defenses built up against the previous variants of this virus. Gauging population-level immunity against such variants and estimating the associated risk of infection can play an important role in informing public health policy (we call this variant-specific serosurveillance).

We used finger prick blood samples from Virus Watch participants to measure how well their antibodies disable the virus (we refer to this as virus inhibition). This offers a more precise and up-to-date measure of protection against infection with specific variants, compared with the routinely offered SARS-CoV-2 antibody tests. We then investigated the association between participants’ variant-specific virus inhibition and their subsequent SARS-CoV-2 infections. 

Crucially, we conducted this study during the transition from the Delta to Omicron BA.1 variant dominance in England, allowing us to investigate this association within each variant’s dominance period separately.

We found that greater inhibition of Omicron BA.1 virus correlated with reduced infection risk during both Delta (Odds Ratio = 0.56, 95% confidence interval 0.34-0.90) and Omicron BA.1 periods (OR = 0.80, 95% CI 0.69-0.92). Delta virus inhibition was linked to reduced infection risk during its dominance period (OR = 0.69, 95% CI 0.50-0.95), but not during the Omicron BA.1 period (OR = 0.91, 95% CI 0.80-1.02).

These results agree with earlier findings that immunity against Omicron BA.1 is protective against Delta, but not vice versa. This suggests that monitoring people’s antibody levels using inexpensive home-based testing could be very useful in informing public health policy. Specifically, results of such testing could be used for targeted vaccination campaigns by identifying groups that are at the highest risk of infection with a specific new variant.

Read the full analysis here.