Investigating the neighbourhood-level effects of air pollution exposure on antidepressant prescriptions in Greater London

I was a lead supervisor of an MSc in Epidemiology student summer dissertation in 2024/25 academic year with Prof. Marta Blangiardo. The dissertation was titled Investigating the neighbourhood-level effects of air pollution exposure on antidepressant prescriptions in Greater London. The dissertation proposal is below.

Dissertation Proposal

Background

In 2021, the Health Effects Institute attributed 8.1 million deaths attributable to air pollution exposure globally, with increasing evidence of the negative relationship between environmental exposures and mental health and well-being. Greater London has some of the largest health disparities between regions, and some of the highest rates of depression and anxiety in the country. Antidepressants are predominantly prescribed to treat clinical depression, along with other mental health conditions such as anxiety and PTSD. Over 20% of adults in England are prescribed antidepressants each year, contributing to a large burden of health and increased need to targeted medical services.

This project investigates the role of air pollution exposure on monthly area-level antidepressant prescriptions, which are commonly used for treating depression, as well as other mental health conditions, including anxiety and PTSD. The NHS antidepressant prescription data is available for each GP surgery across Greater London from 2014 - 2018, mapped to admin areas and linked to neighborhood-level characteristics.

Air pollution exposure estimates come from a new monthly air pollution model for Greater London. There is also scope to investigate other relevant exposures, including local greenspaces, nearby roads, and noise pollution.

Aims

The primary aim of this project is to investigate the effects of air pollution exposure on monthly, area-level antidepressant prescriptions:

a. To model the relationship between air pollution and antidepressant use, capturing trends over time and any geographic patterns between areas.

b. To consider the effects of neighbourhood-level covariates, including population density, age distributions, and unemployment.

c. To investigate local greenspace as a potential effect modifier in the air pollution and prescription relationship.

Proposed methods

All prescription data for London from 2014-2018 are already retrieved as monthly data by GP surgery and processed into Lower Tier Local Authority (LTLA) areas. The air pollution data has been produced at a 1km x 1km spatial resolution and will also be aggregated into LTLAs.

We will start by using a Bayesian regression model to explore the relationship between air pollution exposure and antidepressant prescriptions, considering covariates such as age structure and unemployment for each LTLA. From this model, we will test the residuals for any remaining spatial autocorrelation and will motivate further modelling and conclusions. Finally, we will consider greenspace areas within each LTLA, and we will investigate its effect in relationship to antidepressant prescriptions.