Accuracy of online surveys in predicting COVID-19 uptake and demand: A cohort study investigating vaccine sentiments and switching in 13 countries from 2020 to 2022.

During pandemic conditions, efficient use of resources is essential, but vaccine hesitancy threatens to render vaccine deployment ineffective and inefficient. This study investigated the sensitivity and specificity of a straightforward ex-ante online survey question in predicting ex-post vaccine uptake, and the sociodemographic characteristics associated with individual changes (e.g., false positive or negative responses) in vaccine preferences. We conducted a pooled analysis on stated vaccine preferences and self-reported uptake of a convenience sample of 5213 individuals who were sampled in 13 countries in 2020/21 and 2022. We found online survey response of vaccine intention to be a strong determinant of vaccine uptake. The survey sensitivity was 80.5 %, while specificity was 51.8 %. Though we found evidence of stronger consistency between ex-ante intention and ex-post action in university educated respondents, and weaker consistency amongst those with right leaning political tendencies, further work is required to generalize these results by using representative datasets within each country. In situations where health technologies need to be rolled out urgently, simple and rapid surveys linked to individual health profiles have potential to increase the speed and efficiency of deployment (e.g., Personal Level Integrated Data Asset (PLIDA) in Australia), though low specificity threatens to undermine efficiency gains and sensitivity requirements in risk-averse medical decision making settings may be beyond the capabilities of simple surveys.