Abstract
Background and Aims: Chronic liver diseases (CLD) represent a significant healthcare burden, mostly due to late diagnosis and numerous co-morbidities. We evaluated the effect of co-morbidities, cirrhosis, and disease etiology on hospitalization duration. Methods: Hospitalizations due to alcohol-related, viral, autoimmune, and overlapping liver disease in Belgrade, Serbia (2016-2022), were identified using pre-defined discharge codes. We investigated the hospitalization trend descriptively by plotting the relative mean change in the hospitalization length against time. Assuming the covariate relationship in the directed acyclic graph, we estimated the direct causal effect of the diagnosis type on the length of stay (LOS) by fitting pre-specified Bayesian distributional lognormal models based on domain knowledge. We conducted a post hoc analysis of the impact of cirrhosis on LOS per primary diagnosis. Results: The empirical data show a decrease in the estimated average LOS (8.25-5.51 days). For the same period, the median LOS decreased (4 days (IQR 0-12) to 1 day (IQR 1-7)). In 2021, the share of short-term hospitalizations rose to 46.94%, while the median long-term hospitalization peaked at 11.5 days (IQR 7-21). The expected LOS was the highest for the primary diagnosis of autoimmune liver disease (15.89, 95% CI [14.74, 17.2] days), followed by alcohol-related liver disease (14.22, 95% CI [13.68, 14.79] days). The largest impact of cirrhosis on LOS was observed among patients hospitalized due to viral disease (4.19, 95% CI [2.29, 6.33] days). Conclusions: The presence of co-morbidities and cirrhosis significantly affects LOS. In order to provide better treatment and reduce healthcare costs, there is the need to detect liver disease at earlier stages and better manage its associated co-morbidities.