Corticosteroids (‘steroids’) are taken by 3% of the population. This high prevalence represents a major public health challenge because of the known side-effects of steroid therapy, including diabetes, osteoporosis and heart disease. Steroids also increase the risk from infections. These risks may be caused in part by adrenal crisis, a condition in which life-threatening steroid deficiency occurs as a result of a failure to increase steroid dosing appropriately during illness. However, patient and clinician awareness of the importance of dose adjustment is suboptimal. Furthermore, it is unclear whether the risks of adverse outcomes from infection are universal or confined to subgroups based on patient, disease or treatment characteristics. Understanding this risk would allow better targeted intervention for patients in greatest need, with the potential to improve self-management and disease outcomes. Using data from routine clinical care in the NHS, we aim to determine if patients taking steroid treatment are at increased risk of death, major medical problems (diabetes, cardiovascular, liver and kidney disease) and adverse hospital outcomes (adrenal crisis, need for hospitalisation, length of stay and need for intensive care). We will explore which patients are at greatest risk, comparing steroid dose exposure, duration of treatment, mode of administration (oral, inhaled, topical, intranasal or injectable) and co-existing disease. In so doing, we hope to generate a clinical prediction score for risk of adverse hospital outcomes. This will allow resources for steroid management during illness to be targeted towards individuals at the highest risk of adverse events.
Presently there are few data on the risks of major adverse hospital outcomes in patients on established glucocorticoid (GC) therapy, with attendant uncertainties with respect to risk stratification, advice for shielding and management in the community and hospital settings. Considering these uncertainties, we seek to explore the risks of adrenal crisis, development of medical comorbidities, hospitalisation, intensive care admission and mortality in patients taking GC therapy using a population-based retrospective cohort design embedded within CPRD GOLD and Aurum. Work package 1 (WP1) will examine the risk of GC exposure on outcomes, with work package 2 (WP2) identifying risk factors and development and validation of a clinical risk prediction model.
Patients exposed to GCs (oral, injectable (intra-articular, intramuscular), inhaled, intra-nasal, topical) will be matched with up to 5 linkage-eligible non-exposed comparators, thus controlling our analyses for the potentially confounding effects of age, gender, and general practice. Data will be linked to ONS data for mortality and to HES admitted patient care (APC) data (exploring number of admissions, need for intensive care, and duration of stay). We will compare the incidence rates of adverse events and relative risks will be estimated using stratified Cox regression models.
We will identify relevant risk factors for the composite outcome of mortality and adverse hospital outcome in the patients prescribed GCs in CPRD Aurum. These will include GC-related variables (mode of administration, total dose, duration of exposure), co-morbidity and patient-related factors (age, sex, indices of social deprivation. We will use the identified risk factors to develop a risk prediction model which will be validated internally (using cross-validation and/or bootstrapping) and externally using data from CPRD GOLD. The performance of the model will be assessed using discrimination and classification measures.
Primary outcome measure: Composite outcome of all-cause mortality and adverse hospital outcome (defined by need for hospitalisation, need for intensive care, hospital-acquired infection).
Secondary outcomes:
• Mortality (all-cause, death from adrenal crisis, infection)
• Number of hospital attendances
• Number of hospital admissions
• Duration of inpatient stay
• Hospital-acquired infection
• Need for intensive care admission
• Risk of developing new medical comorbidity (diabetes, cardiovascular, hepatic, renal disease)
Darren Ashcroft - Chief Investigator - University of Manchester
Matthew Carr - Corresponding Applicant - University of Manchester
Aled Rees - Collaborator - Cardiff University
Anthony Avery - Collaborator - University of Nottingham
Claire Higham - Collaborator - The Christie NHS Foundation Trust
David Williams - Collaborator - Cardiff University
Eleni Domzaridou - Collaborator - University of Manchester
Matthew Sperrin - Collaborator - University of Manchester
Tjeerd van Staa - Collaborator - University of Manchester
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation;Practice Level Index of Multiple Deprivation