A Case Control and Cohort Design Study, using Multi-state Models, Analysis of Treatment Patterns in Chronic Obstructive Pulmonary Disease and Asthma.

Date of Approval: 
2021-03-18 00:00:00
Lay Summary: 
A description of the treatment patterns observed in patients with Chronic Obstructive Pulmonary Disease (COPD) and Asthma from diagnosis until the most recent data available. We will analyse the probability of switching from one treatment to another over time and compare this between controlled and uncontrolledpatients. Patients will be grouped into Asthma or COPD sufferers. We aim to show the proportions of patients who follow each treatment pathway to guide future research and show where support can be given to patients to access better care. Retrospective treatment data from CPRD Aurum will be analysed at a class level, alongside hospitalisation data from Hospital Episode Statistics (HES). The study period will begin 1st November 2018 so as to include patients being treated under the most recent, established set of the National Institute of Clinical Excellence (NICE) guidelines.
Technical Summary: 
This project uses linked primary and secondary care data from England to model the probability of transition from one treatment to the next given the current treatment and time of the transition. Due to the lack of pilot data, we do not put any hypothesis about the estimated transition, occupation probabilities or of the dependence of these probabilities on any baseline features.Therefore, we do not plan to adjust for baseline variables. The purpose of this study is to estimate probabilities. Our secondary objective will compare the evolution of these transitions’ probabilities between controlled and uncontrolled groups. We will employ both a cohort and case-control study designs. We will use multistate statistical models to model the probability of the next treatment given the current prescribed treatment and the time of treatment transition. The possible dependence on the history will be handled via multinomial modelling. Using the same multi-state methodology, we are going to analyse the transition probabilities between the controlled and uncontrolled groups and compare the evolution of these probabilities over time. We will use the most recent AURUM data available as the end of study date. We will include patients who have a diagnosis of COPD/Asthma in primary care on or after 1st November 2018, who are >18 years old and have >12 months follow-up data. We will use data available 12 months prior to index to describe the patient demographics. The date of first diagnosis will be defined as ‘index date’. We will define the treatment pathway at a class level from treatment records in primary care. Controlled and uncontrolled patients will be defined using exacerbations and hospitalisations relating to Asthma or COPD. Hospitalisations will include admissions to Accident and Emergency (A&E) using HES A&E data. ‘Hospitalisations’ will be defined using HES Admitted Patient Care (APC) data.
Health Outcomes to be Measured: 
Outcomes to be measured will include. • From HES APC: Hospitalisation spells due to indication (Asthma or COPD); Bed days due to indication • From HES A&E: Admissions to secondary care via A&E; All cause visits to A&E • From CPRD AURUM: Prescriptions of OCS; exacerbations as recorded in primary care Controlled and uncontrolled status will be algorithmically defined. A ‘spell’ is a period of hospitalisation.Discharge codes are organised with the primary code being the main reason for admission. This may not be what the patient first presents with when arriving at hospital.
Application Number: 

Abi Mawer - Chief Investigator - Twist Health
Abi Mawer - Corresponding Applicant - Twist Health
Amy Mulick - Collaborator - Twist Health
James Jordan - Collaborator - NHS Scotland
Lina Eliasson - Collaborator - Twist Health
Magdalena Murawska - Collaborator - Twist Health
Rebecca Sarson - Collaborator - Twist Health
Stephanie Best - Collaborator - Twist Health

HES Accident and Emergency;HES Admitted Patient Care