Development of an identification tool for patients over 65 years at risk of death or emergency hospital admission during cold weather.

Date of ISAC Approval: 
18/05/2016
Lay Summary: 
Cold weather poses challenges to primary care, with increased risk of deaths and hospital admissions. Around 24,000 deaths are attributed to cold weather each year in England and Wales. It was recently recommended that GPs flag patients living in hard to heat homes, but GPs are already over-stretched and do not have the tools to determine which patients are at highest risk. Using three datasets, including CPRD, we will develop a tool to help primary care teams flag people at risk of death or emergency admission to hospital during or just after cold spells. The CPRD dataset comprises over 2 million patients from general practices across Great Britain. We will link the data with information on weather conditions during the period of study to ascertain cold weather spells. We will then use information in CPRD to find out which aspects of people's backgrounds and their medical history predict their risk of dying or going to hospital during or just after a cold weather spell. We will also find out whether people have much contact with primary care before a cold spell, so we can know whether primary care teams have any opportunity to intervene.
Technical Summary: 
Cold weather poses primary care challenges both for prevention and treatment of resulting morbidity. Approximately 24,000 extra deaths occur every winter in England & Wales, chiefly in older people: figures which will rise as the population ages. In March 2015, NICE recommended GPs should flag patients living in cold or hard to heat homes, but GPs currently have neither capacity to visit everyone nor tools to determine those at highest risk. This proposal will address this second issue, to help primary care teams focus resources towards those most likely to benefit. We will use CPRD, linked with Hospital Episode Statistics, to investigate predictors of emergency hospital admission and mortality following cold spells, and examine primary care contacts around such events. We will also link local weather data over the period of follow up to identify cold weather spells. This will allow precise estimation of risks associated with cold weather. We will investigate the degree and nature of contact with primary care during cold spells, both for patients who die or are admitted to hospital within 14 days, and other patients. Prediction equations will be built using a training set (randomly chosen half of the data) and tested in the remainder.
Health Outcomes to be Measured: 
1 - Death (with cause). Although all-cause mortality will be the primary focus, we will also assess death of cardiovascular disease (ICD10 codes I20-I25 and I60-I69) 2 - Emergency admission to hospital. Each emergency admission during the follow up period (ending at first emergency admission or 31st March 2014, whichever is earlier) to be ascertained for each patient.
Collaborators: 

Professor Richard Morris - Chief Investigator - University of Bristol
Alastair Hay - Collaborator - University of Bristol
Claudio Sartini - Collaborator - University College London (UCL)
Dr Daniel Lasserson - Collaborator - University of Birmingham
Dr Peter Tammes - Collaborator - University of Bristol
Professor Richard Morris - Corresponding Applicant - University of Bristol

Linkages: 
HES Admitted;ONS;Other;Patient IMD;Practice IMD (Standard)