Causes of death in people with liver cirrhosis in England compared with the general population: a population-based cohort study.

Abstract

There is a need for unbiased estimates of cause-specific mortality by etiology in patients with liver cirrhosis. The aim of this study is to use nationwide linked electronic routine healthcare data from primary and secondary care alongside the national death registry data to report such estimates. We identified from the linked Clinical Practice Research Datalink (CPRD) and English Hospital Episode Statistics adults with an incident diagnosis of liver cirrhosis linked to the Office for National Statistics between 1998 and 2009. Age-matched controls from the CPRD general population were selected. We calculated the cumulative incidence (adjusting for competing risks) and excess risk of death by 5 years from diagnosis for different causes of death, stratified by etiology and stage of disease. Five thousand one hundred and eighteen patients with cirrhosis were matched to 152,903 controls. Among compensated patients, the 5-year excess risk of liver-related death was higher than that of any other cause of death for all patients, except those of unspecified etiology. For example, those of alcohol etiology had 30.8% excess risk of liver-related death (95% confidence interval (CI): 27.9%, 33.1%) compared with 9.9% excess risk of non-liver-related death. However, patients of unspecified etiology had a higher excess risk of non-liver-related compared with liver-related death (10.7% vs. 6.7%). This was due to a high excess risk of non-liver neoplasm death (7.7%, 95% CI: 5.9%, 9.5%). All decompensated patients had a higher excess of liver-related mortality than any other cause. In order to reduce associated mortality among people with liver cirrhosis, patients' care pathways need to be tailored depending on the etiology and stage of the disease.