Residents of South Asian cities know the air quality is poor: It’s often possible to see the smog blanketing the skyline, or feel that it’s hard to breathe. The air gets particularly bad from around November to February, when weather conditions conspire to trap pollution such that it can’t escape. Conditions have worsened as cities have rapidly grown, with construction, traffic, industry, and power plants all producing harmful air pollutants.
Governments and other organizations track overall air quality levels, but Rest of World wanted to explore the impact of polluted air specifically on gig workers who make deliveries on bikes or motorcycles. South Asia has a huge gig economy workforce, working for international platforms including Uber and inDrive, as well as local giants such as Swiggy and Zomato in India, Careem and Bykea in Pakistan, and RedX and Pathao in Bangladesh. The nature of this work — outside, in the densest urban areas — means these workers are particularly exposed to pollution.
To measure just how exposed they are, we asked a delivery rider in three of the most polluted South Asian cities — New Delhi in India, Lahore in Pakistan, and Dhaka in Bangladesh — to carry an air quality monitor with them on a typical shift. Here’s how we did it.
Using the sensors
We chose to use the Atmotube Pro air quality monitor, made by U.S. company Atmo. Our criteria were that it had to be wearable, designed for outdoor use, have enough battery to last for a whole shift, be relatively low-cost, and track air quality in real time. We wanted a sensor that tracked key pollutants associated with negative health impacts, such as PM2.5 (particulate matter under 2.5 micrometers in diameter), and which connected real-time measurements with both time and location. This would allow us to compare any notable disparities in air quality with our rider’s experience of their day — for example, if a particular area or time of day was associated with greater or less pollution.
We shipped an Atmotube Pro to three reporters, one in each city. While reporting for this story, they each identified a gig worker who was willing to help with our experiment by wearing the monitor for a day. The riders in Lahore and New Delhi carried the monitor on January 9; owing to disruptions around the Bangladesh elections, the rider in Dhaka carried the monitor a little later, on January 12.
After testing the Atmotube devices, our reporters helped the gig workers set them up. We set the devices to “always on,” meaning they would gather air quality data continuously, in real time. We also selected the “high accuracy GPS mode.” The Atmotube device was connected to the gig rider’s smartphone, and data recorded in the Atmotube app. The gig workers clipped the device where it would have continuous airflow over the sensor.
We wanted the gig riders to capture a regular “day in the life” of their work, so we asked them to conduct their shift as usual — starting and finishing as they normally would, and taking breaks whenever they wanted. At the end of their workday, we exported the air quality data collected by the device as a CSV file.
Analyzing the data
The data gathered by the air quality monitors included time, VOC ppm (volatile compounds in parts per million), temperature, humidity, pressure, PM1 (particulate matter below 1 micrometer in diameter), PM2.5, PM10, and geographic coordinates. It also included an Air Quality Score (AQS), a measurement created by Atmo. This is because the Air Quality Index (AQI), a commonly used measurement of air quality, relies on averages of different pollutants over a period of hours, and so isn’t designed to reflect differences in real-time air quality. Inversely to AQI, a higher AQS value means air is cleaner and a lower value means air is more polluted.
This data was recorded in minute averages (however, for at least one rider, PM data was only recorded every 10 minutes during some periods, which we believe may have occurred due to the device losing connection with the phone).
By analyzing this data, we could track the changing air quality throughout a rider’s day. We soon found that the AQS value was not much use for our purposes, as this value stayed at 0 — the lowest possible score — for all three riders, for the majority of the time period. While this clearly supported our thesis that delivery riders are consistently exposed to highly polluted air, we wanted to see their daily experience in more detail.
We focused mainly on the PM measurements, as we were particularly interested in the health impacts of poor air quality. Particulate matter is associated with many health issues, including severe conditions such as cancer, stroke, and heart disease; medical experts we spoke to highlighted PM2.5 in particular as a risk to health. Our results showed that the riders were consistently exposed to PM2.5 values between 100 and 300 micrometers per cubic meter. To put that into perspective, in 2021, the World Health Organization suggested that 24-hour average exposure levels to PM2.5 should not exceed 15 micrograms per cubic meter more than a few times a year.
By comparing the data with notes from our reporters and photographers who followed the riders for part of their day, we were able to correlate notable dips in the PM level with specific moments in the riders’ day — for instance, when they went indoors. (These relative dips were nevertheless well above the WHO guidelines for PM exposure.)
Our VOC measurements were less concerning, remaining below 0.3 in New Delhi and Dhaka, and below 0.4 (aside from one anomaly) in Lahore. Guidelines for VOCs vary and usually focus on specific individual compounds, but in general, VOCs are more of a concern for indoor pollution.
Temperature and humidity can affect air quality, but small fluctuations across the day did not appear to correlate with changes in air quality in our data. We also did not observe any particular changes in air quality associated with location within each city.
Caveats and limitations
Tests have shown that the Atmotube Pro has a high level of accuracy, especially for measuring PM1 and PM2.5, although several factors could affect accuracy levels, including wind direction, high humidity or mist, and any blockage in airflow.
We only collected data from a single rider, on a single day in each city. Our data therefore only shows a snapshot of one gig worker’s experience on a given day, and cannot necessarily be taken to represent gig workers’ exposure to pollution on a broader scale. It is intended to be more illustrative than scientific, aiming to give insight into an individual’s daily life.
We spoke with many more delivery workers, as well as health and environmental experts, to place this data into context and get a broader understanding of how air quality impacts gig workers in South Asia. You can read the full story here.
