Improving the representativeness of UK's national COVID-19 Infection Survey through spatio-temporal regression and post-stratification

Abstract

Population-representative estimates of SARS-CoV-2 infection prevalence and antibody levels in specific geographic areas at different time points are needed to optimise policy responses. However, even population-wide surveys are potentially impacted by biases arising from differences in participation rates across key groups. Here, we used spatio-temporal regression and post-stratification models to UK's national COVID-19 Infection Survey (CIS) to obtain representative estimates of PCR positivity (6,496,052 tests) and antibody prevalence (1,941,333 tests) for different regions, ages and ethnicities (7-December-2020 to 4-May-2022). Not accounting for vaccination status through post-stratification led to small underestimation of PCR positivity, but more substantial overestimations of antibody levels in the population (up to 21 percentage points), particularly in groups with low vaccine uptake in the general population. There was marked variation in the relative contribution of different areas and age-groups to each wave. Future analyses of infectious disease surveys should take into account major drivers of outcomes of interest that may also influence participation, with vaccination being an important factor to consider.

Fig. 1. Post-stratified modelled estimates of swab PCR positivity by region over time.

Modelled estimates are post-stratified for age, sex, CIS area (116 sub-regions within the 9 administrative regions shown), ethnicity and vaccination status. Estimates are presented as posterior medians (solid lines) with shading representing a 95% credible interval. Crude numbers from the underlying data are labelled as raw data. Vertical black lines indicate from left to right the start (at the national level) of the Alpha, Delta, Omicron BA1, and Omicron BA2 dominant period, respectively.

Fig. 2. Crude cumulative vaccination uptake based on survey participants and admin data.

The red and green lines almost perfectly overlap reflecting the fact that the assumptions we made about vaccination uptake among children too young to be present in the 2011 Census are of negligible importance given that during the study period very few children aged 9–11 years of age – the ages from which we extrapolated to younger children assuming I) no vaccinations under the age of 5 and half the coverage of that observed among 9–11 years old children for those aged 5–8 years of age (admin data main), or II) no vaccinations under the age of 9 (admin data sensitivity).

Fig. 3. Post-stratified estimate of swab PCR positivity by CIS area over time.

Estimates are post-stratified for age, sex, ethnicity and vaccination status.

Fig. 4. Impact of post-stratifying for vaccination status (yes/no and interaction with time) on modelled estimated antibody positivity at the 100 BAU/ml threshold by region over time.

Modelled estimates are post-stratified for age, sex, CIS area, ethnicity and vaccination status. Estimates are presented as posterior medians (solid lines) with shading representing 95% credible intervals. Crude numbers from the underlying data are labelled as raw data. BAU: binding antibody units.