Standard image of the new scientist

Official British statisticians had concerns about the trend in the Covid-19 poll

By Adam Vaughan

The ONS survey tracks the prevalence of Covid-19 in parts of the UK

James Veysey / Shutterstock

The UK’s Office for National Statistics (ONS) privately discussed concerns about the risk of its flagship Covid-19 infection survey, which documents show a biased picture of the country’s epidemic between May and July.

The survey is considered to be one of the best insights into the prevalence of the disease as it polls random households across the four UK countries. However, New Scientist previously reported that the response rate fell as the survey expanded.

Of the households invited to participate in England, the response rate fell from 51 percent in April to 5 percent in October, raising concerns that only a certain group of people taking the survey are less representative of Covid- The actual spread of 19 in the parish. In October, the ONS announced to New Scientist that the survey had been weighted and continued to provide reliable estimates of the spread of the virus.


Now documents released under the Freedom of Information Rules show that the government agency was sufficiently concerned about the risk of bias from falling response rates that it discussed measures to address the problem.

“The lockdown has subsided and public attention has begun to return to work and the new normal. In this context, both the response rates of households and the average number of respondents in a household have decreased, which increases the risk of non-response bias, ”says an appendix to the contract between ONS and IQVIA, the contractor for the survey.

The annex discusses the need for action to reverse the falling response rate and reduce the bias caused by households’ failure to respond. “Particularly interested in instant quick wins,” it says. Meetings between the ONS and IQVIA were proposed for late July and August to discuss measures such as nudge letters and texts, as the officials called this an “urgency”.

The ONS was contacted by New Scientist about the documents and stated that only some initial work was carried out as the risks did not materialize. This shows an October goal of 150,000 unique 14-day tests achieved ahead of schedule. After hitting a low of 5 percent in October, the response rate in England rose to 12 percent by December 11th.

That’s still exceptionally low for an ONS survey, with response rates ideally more than 80 percent, says Sheila Bird of the University of Cambridge. She says the survey is “a big undertaking” but the ONS should be “much clearer” about how it addresses the risk of biasing the rebalancing of the results to reflect the wider population.

Elsewhere in the documents, key performance indicators for IQVIA are listed, including “Time to Book Appointments”, “Completion of Appointments” and missed calls returned within a business day.

However, participants in the survey showered IQVIA with complaints in September and October. Attendees couldn’t get through to call centers, calls weren’t returned, and no-shows for booked appointments. These issues seem to have continued into November and December, and people checked out on social media to report ongoing problems contacting IQVIA and missed appointments.

According to the ONS, the survey continued to provide reliable estimates of the epidemic. “The Coronavirus Infection Survey has met or exceeded all key performance targets and remains an important source of updated data on Covid-19 infections and antibodies across the UK,” said a statement.

The documents also show when the ONS predicts its survey will become unnecessary as the country’s epidemic recedes. Tables show that by March 2022, two years after the survey began, the body is unlikely to have any unique subjects per two weeks.

Sign up for our free Health Check newsletter for a round-up of all the health and fitness news you need to know every Saturday

More on these topics: