Non-steroidal anti-inflammatory drugs (NSAIDs), such as ibuprofen, are widely prescribed by doctors to relieve pain, reduce inflammation and reduce fever. About 15% of the UK population have or will be prescribed one. NSAIDs are, however, associated with many emergency hospital admissions from adverse drug reactions, in particular gastro-intestinal (GI) bleeds. The risk of such bleeding can be 3 to 12 times greater than for non-users, depending on the NSAID, other drugs taken in combination with them, and co-morbidities. Acute GI bleeds are serious and one in five can be fatal. NSAIDs also increase risks of cardiovascular problems and impaired renal function, especially in older patients with higher baseline risks, but about 9% of patients over 70 will receive an NSAID prescription of over 3 months duration. National Institute for Health and Care Excellence (NICE) guidelines recommend taking into account patient characteristics and comorbidities when prescribing NSAIDs, but, while observational studies have identified and estimated the risks of GI bleeding associated with very many factors, few have tried to incorporate these into risk prediction models. Existing prediction models are limited in the factors used and the types of patients covered. Currently, no general risk prediction model for GI-bleeding exists to assist prescribing in heterogeneous NHS primary care settings, where most NSAID prescribing takes place. The aim of this study is to develop one, focusing on patients attending a GP for the first instance of medium to severe pain control and with no recent history of analgesic prescribing.
Aim: To develop and validate a risk prediction model for estimating risk of a major gastrointestinal bleed in patients requiring pain and inflammation control.
Study Design: Prospective open cohort study
Setting: UK General Practice
Patients aged 18 or older with prescriptions of analgesics between 1 Jan 2010 and 31 Dec 2020, but without any record of a previous major GI bleed, and having no prescriptions for analgesics within the prior 24 months.
Primary outcome: First GI bleed requiring hospital admission or causing death (HES/ONS/GP records with specific to GI bleed codes).
Methods: CPRD AURUM will be used for derivation and internal validation and CPRD GOLD for external validation. Migrated practices will be will be removed from the derivation cohort and retained as GOLD practices. Only linked to ONS and HES practices will be included. Patients will be followed from the date of the first record of pain until their treatment is changed or until their last record on the database. A Cox regression model will be used to assess the probability of the outcome based on established risk factors including comorbidities, life-style factors, NSAIDs, other analgesics, and other medicines used. The model will be evaluated in a separate external validation cohort for model-fit, discrimination and calibration accuracy. Sensitivity, specificity, positive predictive value and negative predictive values will be estimated at different risk thresholds. A decision-curve analysis will also be conducted.
Outputs: A clinical model for predicting risk of major GI bleed in patients recently unexposed to NSAIDs to inform its prescribing.
First GI bleed requiring hospital admission or causing death. We will use hospital and ONS records with specific to GI bleed ICD-10 codes or GP records with GI bleed READ codes associated with hospital admission with a plausible ICD-10 code. If GI bleed is recorded by a GP within a month prior to a GI-bleed-associated hospital admission or death a date of GP record will be use as an outcome date.
Anthony Avery - Chief Investigator - University of Nottingham
Yana Vinogradova - Corresponding Applicant - University of Nottingham
Amelia Taylor - Collaborator - University of Nottingham
Barbara Iyen - Collaborator - University of Nottingham
Brian Bell - Collaborator - University of Nottingham
Colin Crooks - Collaborator - University of Nottingham
Darren Ashcroft - Collaborator - University of Manchester
Rachel Elliott - Collaborator - University of Manchester
Ralph Kwame Akyea - Collaborator - University of Nottingham
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation;Practice Level Index of Multiple Deprivation