Spatio-Temporal Exposure Assessment Methods for estimating the health effects of air pollution

Exposure to outdoor air pollution has been associated with increased risks of admission to hospital and death. These risks are thought to be related to both long-term exposure to air pollution measured over many months/years and to short-term changes in pollution concentrations on a day-today basis. Cohort studies use long-term pollution measurements at different geographical locations to assess the risk to health whereas time-series studies use daily average concentrations in a city related to daily counts of health events (e.g. hospital admissions for respiratory disease) in the city. As the public are exposed to air pollution over days/weeks/months/years, the extent to which estimates of the associations between long-term average and short-term variations in pollutant concentrations and health overlap is an important issue with implications for the development of policies to control pollution. To date, this issue has not been investigated in the UK.

In the past, both cohort and time-series studies have been limited by the availability of data from pollution monitors. In recent years, statistical models have been developed that incorporate data on land use, pollution emissions, and/or predictions derived from satellite data to estimate daily pollutant concentrations in small geographical areas. The objective of this study is to evaluate, and integrate, the different models and use the resulting predicted pollution concentration in a joint analyses of long- and short-term exposures for a limited number of health endpoints as a ‘proof of concept’.

King’s College London (KCL) and St George’s University of London (SGUL) have come together as a collaboration in order to carry out this work. The results of this study will be in the interests of the public and therefore the data will be processed under GDPR Article 6(1)(e) and GDPR Article 9(2)(j).

To achieve this objective, both annual and daily estimates of concentrations of particles and gaseous pollutants for lower super output areas (LSOA) within the south east of England will be derived from daily pollutant measurements at fixed site monitors, remote sensing (satellite) data and daily estimates from different modelling methods for the period 2004 to 2013. The performance of each approach in estimating the health effects associated with short and long-term exposures will be assessed using newly developed simulation methods. To demonstrate proof of concept in estimating simultaneously the health effects of both long- and short-term exposure to air pollution, estimates of pollutant concentrations from the method(s) showing the best performance will be used in separate analyses of primary care data, counts of hospital admissions for respiratory and cardiovascular disease and counts of deaths for a 5-year period (2009-2013).

The study has been funded by the Medical Research Council. The lead Principal Investigator is a senior investigator from KCL, and the co-Principal Investigator is a senior investigator at SGUL. Both KCL and SGUL have responsibility for the Hospital Episode Statistics (HES) analysis aspect of the project including the purposes and therefore KCL and SGUL are joint Data Controllers. Other collaborators for the wider project (i.e. outside this agreement) are Imperial College London, University of Athens and University of Harvard. The project includes modules undertaken at each of the institutions, the HES analysis aspect being one of these modules. SGUL and KCL are responsible for the analysis of the hospital admissions and mortality data. This application relates only to the provision of hospital admissions data provided by means of a tabulation from NHS Digital. An unsuppressed tabulation is required to ensure the data is in a comparable format to the pollution data from other sources. A separate application to the UK Data Service for access to Civil Registrations (Deaths) data was made by SGUL. There will be no linkage of the Civil Registration Data with the data provided by NHS Digital. Only the SGUL and KCL investigators will have access to the hospital admissions data under this Agreement.

The data received by St. George’s University of London or King’s College London will not be used for any purpose other than to meet objectives as stated in this Data Sharing Agreement and will not be shared with any other third party or organisation.

The Data Controllers should ensure appropriate data processing agreements with all data processors contracted to undertaking work referenced within this agreement.

The hospital admissions data extract will be stored in the data safe haven (DASH) at SGUL. The data will be accessed only by substantive employees of SGUL and KCL, and in accordance with SGUL’s DASH IT security policies and procedures. Members of the research team from KCL will only access the data via the SGUL data safe haven. No data will be stored, accessed or processed on KCL premises. The lead and co-Principal Investigators at KCL and SGUL respectively are responsible for the data processing, management and statistical analyses of the HES data. Only details of the statistical method and the interpretation of the output from the analyses will be discussed with other investigators. The admissions data will not be shared with any other institution/organisation and other collaborators within the study will not have access to the data.

SGUL/KCL are requesting daily counts of respiratory and cardiovascular emergency hospital admissions in subjects aged 35+ for each LSOA for a defined area of the south east of England (9811 LSOAs). The LSOAs will be provided to NHS Digital by SGUL. Estimated pollution concentrations for each LSOA for each day between 2009-2013 will be supplied by KCL and stored in SGUL DASH. The daily pollution concentrations will be combined with the daily counts of admissions by date and LSOA for statistical analyses by one of the SGUL investigators. The counts of admissions derived from the admissions records will not be shared with any other institution/organisation or other collaborators. No other collaborators will have access to the admission data.

The statistical analysis will consist of a regression model to assess associations between pollution concentrations and counts of admissions controlling for other factors that vary over time such as daily temperature and holiday periods. The analyses will produce risk estimates for changes in pollution levels for spatial and temporal variation in pollutant concentrations. The analysis of the admissions data is part of the final module of the Spatio-Temporal Exposure Assessment Methods Project (proof of concept) and therefore the output does not feed into other work within the project.

All organisations party to this agreement must comply with the Data Sharing Framework Contract requirements, including those regarding the use (and purposes of that use) by “Personnel” (as defined within the Data Sharing Framework Contract i.e.: employees, agents and contractors of the Data Recipient who may have access to that data).

Results from the analyses of the hospital admissions counts will be risk estimates, associating long- and short-term exposure to outdoor air pollution and admission to hospital. Summary statistics will be produced for the whole study area and study period only, not for individual LSOA or day so eliminating any small cell reporting. All outputs with be aggregated with small numbers suppressed in line with the HES Analysis Guide. Study results will be available to the project team prior to the completion date of the analysis (June 2020). The results from the study will be written up and submitted to a peer review journal for publication. It is also anticipated that the study findings will be presented at scientific conferences such as the International Society for Environmental Epidemiology.

The evidence from this study will be considered by the UK Government advisory committee on air pollution (COMEAP) when formulating advice to policy makers and ministers. COMEAP maintains a watching brief on air pollution research particularly from the UK as well as undertaking targeted work in response to specific requests for advice from Government departments. Similarly, other institutions such as WHO and the US Health Effects Institute undertake periodic reviews of the literature as part as its ongoing assessments of the health effects of air pollution. The published outputs from this study will therefore be included in such assessments.

National and international governmental agencies are responsible for the evaluation and development of policies to regulate air pollution. Part of this process is an assessment of the impact of air pollution on health. Whilst the proposed analyses of the admissions data is for ‘proof of concept’, the results from this study will provide policy makers in the UK including Defra, the Department of Health, Public Health England and Local Authorities with useful information to assess health impacts and so help plan local and national air pollution mitigation strategies in both the long- and short-term.

None to date. The agreement start date was December 2018. The data were produced in July 2019 therefore additional time is needed in order to complete the analysis.