We will estimate the health effects and healthcare system burden of cold waves and heat waves in England.
We will use England patient data from CPRD to look at the impact of changes in ambient temperature on the full healthcare chain (GP visits, secondary referrals, hospitalisations and deaths) in England. The analysis will also include a healthcare cost estimate based on cost information about primary and secondary care. We will produce a thorough statistical analysis, breaking down the effect of ambient temperature by age, gender, medical preconditions, type of disease, as well as social characteristics (e.g. deprivation).
The study will also assess the effect that climate change would have on health and healthcare. For that purpose, we will be using the data from climate models (such as the UK Climate Impacts Programme Projections, released in 2018) to infer changes in healthcare patterns. We will look at net effects of climate change on, as well as changes in the timing of these outcomes (since there is a risk of hospitalization during heat waves) and its structure (e.g. fewer consultations, more A&E hospitalizations).
We will pay particular attention to policy changes (e.g. Universal Credit, cuts in disability benefits) which may have affected the temperature-health relationship. We will also correlate the available health outcomes with energy and housing prices to analyse the effect that a restriction in energy delivery and access to housing could have had on temperature-induced health outcomes.
This study will correlate health outcomes (e.g. consultations per 100,000 inhabitants) with temperature data (e.g. daily average temperatures) and estimate dose-response functions.
The dose-response functions will be estimated with daily data. The panel variable will be the geographical areas available in CPRD (practice region). Therefore, we will use CPRD individual-level data to produce an aggregated dataset of daily rates (for consultations, prescriptions, hospitalizations, mortality, etc.) to be as comprehensive as possible. The outcome variables will include:
- A large spectrum of health outcomes recorded in CPRD (consultations, medical tests, prescriptions, hospitalisations and deaths)
- Breakdown of the outcome variables by subgroups to define vulnerable populations (according to age, preconditions, cause of morbidity) and potential mechanisms linking temperature to human health (e.g. the impact of smoking on increasing the impact of cold weather on pulmonary diseases).
- Estimate a cost for healthcare by associating each intervention to a cost estimate, and look at the dose-response function of the aggregate healthcare cost to changes in temperature.
The econometric specification will control for seasonality and location-specific unobservables. It would also include temperature lags to account for short-term dynamics, and specifications may include confounders such as precipitation and pollution. We would furthermore estimate models at monthly and yearly levels to account for medium-term adaptation to cold and heat. We will provide estimates of climate change impacts with general circulation models.
Finally, the project aims to select exogenous policy changes (e.g. changes in electricity and housing prices, the introduction of Universal Credit, the Two-Child limit or the cuts introduced on disability benefits) and see if these have had an impact on the temperature-morbidity/mortality relationships.
The outcome variables will be expressed in daily rates per inhabitant and per area. We will be comprehensive of all main cost entries for the NHS and patients, even though we may expect some to be more weather sensitive (e.g. consultations for respiratory diseases) than others: consultations (e.g. with GPs, nurses, phone consultations); prescriptions (by type of drug); medical tests; referrals to specialists (or consultations of specialists); hospitalisations (including admissions, but also daily attendance); and mortality (all-cause and cause-specific). We would furthermore produce several breakdowns for these outcomes, e.g. by age, gender, deprivation index, region (e.g. North, South of England), for urban/rural areas; by pathology. We will give a monetary value to each outcome (e.g. the cost of a consultation) to provide an NHS cost estimate.
Rafael Perera - Chief Investigator - University of Oxford
Francois Cohen - Corresponding Applicant - University of Oxford
Anant Jani - Collaborator - University of Oxford
Brian Nicholson - Collaborator - University of Oxford
Clare Heaviside - Collaborator - University of Oxford
Louisa Chenciner - Collaborator - University of Oxford
Marshall Burke - Collaborator - Stanford University
Mateo Petel - Collaborator - Not from an Organisation
Mudith Jayasekara - Collaborator - University of Oxford
Patrick Patrick - Collaborator - University of Oxford
Radhika Khosla - Collaborator - University of Oxford
Sam Bickersteth - Collaborator - University of Oxford
Sarah Lay-Flurrie - Collaborator - University of Oxford
2011 Rural-Urban Classification at LSOA level;HES Accident and Emergency;HES Admitted Patient Care;HES Diagnostic Imaging Dataset;HES Outpatient;ONS Death Registration Data;Patient Level Index of Multiple Deprivation