Measuring the effect of Non-Pharmaceutical Interventions (NPIs) on mobility during the COVID-19 pandemic using global mobility data
Evaluating the efficacy of government interventions used to mitigate the spread of COVID-19 has been challenging. Mobility data from phones can be used as a low-cost and standardised mechanism to observe the change in the aggregate movement of populations in response to interventions stimuli.
The pandemic caused by the coronavirus disease 2019 (COVID-19) has had a huge impact on global health and economies, with costs of $11 trillion projected (as of June 2020) by the International Monetary Fund (IMF). In the lead-up to the development and deployment of vaccination programmes, the implementation of governmental Non-Pharmaceutical Interventions (NPIs) has been the primary means of controlling the pandemic. One of the intended effects of these NPIs has been to reduce population mobility and thereby the opportunity of COVID-19 spreading. Due to the huge costs of implementing these NPIs, it is essential to have a good understanding of their efficacy. NPIs with the same intention and thus classification may have different effects in different countries, as substantiated and reported by previous works. It is therefore imperative to evaluate the variation in the effectiveness of NPIs in different countries.
In our paper ‘Measuring the effect of Non-Pharmaceutical Interventions (NPIs) on mobility during the COVID-19 pandemic using global mobility data’, we investigated the proportional contribution of NPIs to the magnitude and rate of mobility changes at a multi-national level (Figure1). We hypothesised that the magnitude of mobility loss may describe the capacity for a given country's population to self-isolate and the rate of the mobility loss may reflect some degree of urgency at which NPIs are enacted.
Figure 1 An example of population mobility from Spain (Apple dataset).
We examined the freely available aggregated phone-derived mobility data from Apple and Google and their relationship to NPIs. Apple mobility data were derived from requests for directions in Apple Maps for driving, walking and train transit; while Google mobility data shows movement trends by country across different categories of places, like retail and recreation (RAR), grocery and pharmacy, transit and stations (TS), parks, workplaces and residential. To better understand the relationship between NPIs deployed, we performed chi-square and cluster analysis on the NPIs provided by the ACAPS (https://www.acaps.org/) database.
Our analysis revealed that NPIs with the greatest impact on the magnitude of mobility change were: lockdown measures; declaring a state of emergency; closure of businesses and public services and school closures. NPIs with the greatest effect on the rate of mobility change were the implementation of lockdown measures and the limitation of public gatherings. Furthermore, separately recorded NPIs like school closure and closure of businesses and public services were closely correlated with each other, both in timing and occurrence. This finding suggests that the observed significant NPI effects are mixed with and amplified by their correlated NPI measures. Finally, we observed direct and similar effects of NPIs on both Apple and Google mobility data. Although Apple and Google data were obtained by different methods they were strongly correlated indicating that they are reflecting overall mobility on a country level.
To conclude, by objectively evaluating the efficacy of NPIs using publicly available datasets from Google and Apple, we identified that lockdown measures and restrictions of gatherings had the largest impact on reducing mobility, hence the spreading opportunity of COVID-19. The availability of this anonymised data, putatively with an increased spatial resolution and standardised analytical tools, provide the opportunity for governments to build timely, uniform, targeted, and cost-effective mechanisms to monitor COVID-19 or future pandemic countermeasures. This includes initial primary effects and habituation to those measures exemplified by more recent observations of increasing mobility despite maintaining or re-issuing the same NPI measures.
Snoeijer, B.T., Burger, M., Sun, S., Dobson, R.J.B., Folarin, A.A. Measuring the effect of Non-Pharmaceutical Interventions (NPIs) on mobility during the COVID-19 pandemic using global mobility data. npj Digit. Med. 4, 81 (2021). https://rdcu.be/ckCMF