Evaluation of the impact of glucose-lowering treatment with pioglitazone on stroke outcomes

Date of ISAC Approval: 
29/03/2017
Lay Summary: 
Type 2 diabetes (T2DM) is a chronic condition characterized by elevated blood sugar levels (hyperglycaemia), which can result in an increased risk of a variety of conditions, including heart disease, strokes, kidney failure, blindness and amputation. Different treatments are available to treat T2DM depending on the stage of the disease and the characteristics of the patient. One drug used to treat T2DM is pioglitazone. Pioglitazone has been shown to have a beneficial impact on the number of people having strokes and the severity of the stroke. In this study we wish to use the CPRD database to compare stroke outcomes for patients prescribed pioglitazone with those prescribed other diabetes treatments. To do this we wish to match patients treated with pioglitazone to other patients with diabetes so that they are as similar as possible. We then wish to compare how many patients in each group develop strokes and, of these, how many die as a result. We also wish to compare how long people stay in hospital after a stroke and how many are transferred for rehabilitation.
Technical Summary: 
The aim of the study is to compare stroke outcomes for patients with diabetes treated with either pioglitazone or non-pioglitazone inclusive regimens. Patients treated with pioglitazone will be selected from the Clinical Practice Research Datalink and matched to controls by demographic, treatment and morbidity criteria. HES inpatient, outpatient and ONS data will be required for this study, so the population will be limited to English patients participating in the linkage scheme. Time to incident stroke event and subsequent events will be compared using Cox proportional hazards models. The proportion of patients treated with and without pioglitazone and those previously treated with pioglitazone but subsequently discontinued who have a stroke event and die within 28 days following an index event and the proportion of patients that require hospitalised rehabilitation will be compared by logistic regression. Linear regression will be used to compare length of stay relating to stroke events.
Health Outcomes to be Measured: 
- Incident stroke - Morbidity due to stroke - stroke recurrence - mortality due to stroke
Collaborators: 

Dr Christopher Morgan - Chief Investigator - Human Data Sciences
Dr Christopher Morgan - Corresponding Applicant - Human Data Sciences
Dr Sara Jenkins-Jones - Collaborator - Human Data Sciences
Sarah Holden - Collaborator - Human Data Sciences

Linkages: 
HES Admitted;HES Outpatient;ONS;Practice IMD (Standard)