Developing A National Learning Health System For Asthma

Talk Code: 
Colin Simpson
Colin R Simpson12, Ireneous N Soyiri1, Aziz Sheikh1, Stefan Reis3, Kimberley Kavanagh4, Massimo Vieno3, Tom Clemens5, Edward J Carnell3, Jiafeng Pan4, Rachel C Beck3, Hester JT Ward6, Chris Dibben5, Chris Robertson4
Author institutions: 
1 Victoria University of Wellington. 2 Asthma UK Centre for Applied Research, The University of Edinburgh. 3 Centre for Ecology and Hydrology. 4 Strathclyde University. 5 Administrative Data Research Centre - Scotland. 6 National Services Scotland, NHS


Scotland has the highest prevalence of asthma in the world and some of the poorest health outcomes from asthma. Although ambient pollution, weather changes and sociodemographic factors have been associated with asthma attacks, it remains unclear whether modelled environment data and geospatial information can improve population-based asthma predictive algorithms. We aim to create the afferent loop of a national learning health system for asthma in Scotland. We will investigate the associations between ambient pollution, meteorological, geospatial and sociodemographic factors, and asthma attacks.


We will develop and implement a secured data governance and linkage framework to incorporate primary care health data, modelled environment data, geospatial population and sociodemographic data. Data from 75 recruited primary care practices (n=500,000 patients) in Scotland will be used.. Scottish population census, education, weather and pollution databases will be incorporated into the linkage framework for analysis. We will then undertake a longitudinal retrospective observational analysis. Asthma outcomes include asthma hospitalisations and oral steroid prescriptions. Using a nested-case control study design, associations between all covariates will be measured using conditional logistic regression to account for the matched design, and to identify suitable predictors and potential candidate algorithms for an asthma learning health system in Scotland.Our objectives are to: 1. Gain the necessary permissions required to compile and use longitudinal disparate datasets from clinical and health services databases, environment, geospatial, population and sociodemographic and education databases; 2. Model potential meteorological and pollution predictors of asthma at higher resolutions; 3. Investigate geospatial locations and mobility, exposure at work place, schools and residence; 4.Understand the associations between environmental and health covariates and clinically relevant asthma outcomes; and 5. Investigate the feasibility of creating a daily pollution and meteorological forecast for use in an operational learning health system.


We have gained permissions from privacy and governance committees and recruited 73 general practices. Modelled near surface concentrations on key air pollutants at a horizontal resolution of 5 km x 5 km at hourly time steps have been generated using the EMEP4UK atmospheric chemistry transport model for the datazones of the primary care practices' populations. Health datasets have been compiled.


Findings from this study will contribute to the development of predictive algorithms for asthma outcomes and be used to form the basis for our learning health system prototype.

Submitted by: 
Colin Simpson
Funding acknowledgement: 
This work was supported by the Natural Environment Research Council [NE/P011012/1] and the Asthma UK Centre for Applied Research (AUK-AC-2012-01).