New research from the University of East Anglia reveals the most effective ways of tracking the spread of Covid at the height of the pandemic.
From the Zoe app to hospital admissions, the ONS coronavirus infection survey in England (ONSCISE) data and Google searches – the study shows how different tracking systems fared when it came to publishing infection estimates.
While the ONS sample of random households was the most accurate tracker, the survey was more costly than others, and there was often a lag in the publication of its estimates.
Other alternative epidemic trackers were producing data either the next day, or at least within three days. These included data from the Zoe app, hospital admissions, virus counts in community wastewater, NHS 111 calls and GP consultations.
Infectious disease expert Prof Paul Hunter from UEA’s Norwich Medical School, said: “At the very start of the pandemic, existing surveillance methods from NHS usage were adapted to track cases of Covid.
“But no one way of tracking Covid in 2020 and 2021 could give a timely or complete picture of the information needed for the epidemic response.
“We wanted to better understand the strengths and weaknesses of all the Covid trackers, taking into account things like how quickly they were able to provide information and how closely they tracked the ONS estimates of incidence which are considered the best indicator of total infections.”
The research team studied daily rises and falls in ONSCISE numbers from September 2020 to November 2021 in England – and compared them to 11 other data sources including swab test results, GP consultations, NHS 111 calls, emergency department attendances, hospital admissions, Google trend phrases (for covid or coronavirus), and estimates generated after self-reported test results via the Zoe app.
Lead researcher Dr Julii Brainard, also from UEA’s Norwich Medical School, said: “We compared each time series for when Covid cases seemed to be rising or falling in England.
“The most accurate epidemic tracker was derived from the ONS sample of random households who agreed to be tested for Covid, so we especially compared the other sources to this.
“We found that the cases detected by swab testing because of medical need or occupational risk, as well as emergency department attendances with a covid diagnosis, were very good independent trackers of how the epidemic was developing.
“The Zoe app estimates were also highly correlated with the ONSCISE estimates of covid cases in England but as the Zoe app used ONSCISE data to help calibrate its estimates it was not independent of other ways of counting covid cases.
“Also, sometimes the Zoe app estimates were too high and, at times, it delayed spotting downturns in case counts.
“Some trackers such as hospital admissions were sometimes very good indicators of community prevalence but became much less indicative of community prevalence after the vaccination programme in early 2021 - which almost broke the link between infection and need for hospitalisation.
“Google search frequency for the phrases ‘coronavirus’ or ‘covid’ and calls to the NHS 111 advice line and NHS 111 website visits appeared to strongly indicate public interest and anxiety rather than case status.
“Wastewater samples over all England, and GP appointments for patients thought to have covid were not highly indicative of all-England prevalence.
“Our findings are important because it’s useful to know which systems could be developed quickly, or at relatively low cost, or that met other information demands, were about as good as the best epidemic tracker - ONSCISE.
“Early on in the pandemic, existing NHS and UK Health Security Agency surveillance methods were adapted to track cases. Of these, visits to emergency departments most mirrored the national infection prevalence.
“We can also observe that if a testing system like the PCR and lateral flow swabbing programme in England is sufficiently comprehensive, these data can be fairly reliable at indicating what is happening to community infection levels.
“The approach of the innovative Zoe app, which used recent self-reported data to update the most recent ONSCISE estimates of community infection, could be adopted for integration with other surveillance systems to estimate whole community case counts.
"All-England Google Trends search ranks for specific phrases or virus counts in wastewater samples, were not very correlated with national case counts.
“Overall, this research highlights the advantages of using multiple tracking systems, incorporating both NHS and non-NHS data sources, to track the community spread of Covid.
This research was funded by the National Institute for Health Research (NIHR) Health Protection Research Unit (HPRU) in Emergency Preparedness and Response at King’s College London in partnership with the UK Health Security Agency (UKHSA).
‘Comparison of UK surveillance systems for monitoring COVID-19: Lessons for disease surveillance’ is published in The Lancet Public Health.
Artificial Intelligence has helped scientists at the University of East Anglia reveal a new form of aggressive prostate cancer which could revolutionise how the disease is diagnosed and treated in the future.Read more
A research project co-led by Prof Paul Hunter, from UEA’s Norwich Medical School, has been awarded more than £3 million to carry out research into improving business continuity for health services following extreme weather events.Read more
Psychotherapy is an effective treatment for adults PTSD following exposure to multiple traumatic events, according to new research from the University of East Anglia.Read more