Validity of Air Quality as a Measure of Human Mobility in Uganda. The COVID-19 Context.

Abstract

Background Mobility patterns are valuable in identifying transmission patterns for infectious diseases and in deriving contact matrices that are used to parametrize mathematical models. Aggregated location data from mobile phones have been the main means of measuring human mobility on a population level. However, these data come with several limitations related to individual privacy, access and restriction of the GPS location by the user that limit their use. Methods We explored the viability of using ground monitored air quality data as an alternative to aggregated location data from mobile phones, as a measure of human mobility in two cities in Uganda. We determined associations between air quality and human mobility; and the effect of mobility restrictions on mobility and air quality using Pearson correlation (R), multivariate regression and visualized these relationships using scatter plots. Results Daily mean levels for PM2.5 in both cities were consistently higher than the WHO guideline limit, with a mean of 77.0µg/m3 (Range = 22.0–309) for Kampala and 60.0µg/m3 (Range = 18.2–331) for Wakiso. PM10 levels had a mean of 84.6µg/m3 (Range = 25.0–318) in Kampala and 67.9µg/m3 (Range = 21.0–340) in Wakiso. PM2.5 was negatively correlated with the government response stringency index for Kampala (R = -0.31, p < 0.001) and Wakiso (R = -0.21, p < 0.001). In Kampala, PM2.5 was positively associated with movement in grocery and pharmacy (R = 0.24, p < 0.001), parks (R = 0.25, p < 0.001), retail and recreation (R = 0.24, p < 0.001), transit stations (R = 0.3, p < 0.001) and work places (R = 0.2, p < 0.001); and negatively correlated with movement in residential places (R = -0.3, p < 0.001). Only associations between PM2.5 and movement in workplaces and residential places were statistically significant in Wakiso (R = 0.14, p < 0.001 and R = -0.19, p = 0.003 respectively). Conclusions These findings suggest that air quality data are linked to human mobility data and could thus be used to monitor human movement patterns. This work represents a pioneer study to empirically and quantitatively assess the value of air quality data as a surrogate for human mobility in Uganda.