Cirrhosis is the most serious form of liver disease. It occurs when the liver has been damaged so badly that it can no longer repair itself and stops working properly. This can result from drinking too much alcohol, a fatty diet or a long-term liver infection such as hepatitis; a variety of conditions which lead to liver damage. The liver plays an important role in the function of other body organs and systems, such as the heart and immune system. Therefore, examining the performance of these may help to indicate how ill the patient is due to cirrhosis. The aim of this study is to observe the time to death and develop a formula to predict survival in cirrhosis patients, taking into account the different organs and body systems affected by the liver disease. Since cirrhosis is often diagnosed late, many patients die soon after they are diagnosed with the disease; therefore, the study will be split into two parts, looking at death within three months of diagnosis and in the longer term.
We aim to model mortality in patients with cirrhosis while adjusting for a number of contributing factors. From a population of research-quality patients in CPRD eligible for linkage to HES data, we will identify patients with cirrhosis using Read and ICD-10 diagnostic codes. Admissions indicating decompensating events will be identified from HES using ICD-10 codes. In order to model mortality in cirrhosis patients, two statistical models will be created, a logistic regression and a time-dependent survival model. The logistic regression will predict death within three months of diagnosis of cirrhosis while adjusting for a number of laboratory results and prior events (including events that would indicate failure or deterioration of bodily organs). A time-dependent model will be run for patients that survived more than three months adjusting for laboratory results and cumulative admissions for decompensating events on a monthly basis.
Patients with cirrhosis will be sampled from the CPRD clinical and referral tables by Read code, and from HES data using ICD-10 codes. Mortality will be ascertained from presence of death date in the CPRD patient table.
Craig Currie - Chief Investigator - Cardiff University
Ellen Hubbuck - Corresponding Applicant - Pharmatelligence Limited t/a Human Data Sciences
HES Admitted Patient Care;HES Admitted Patient Care;ONS Death Registration Data