De-identified data and analysis code for: Urgent rather than total admission count tracks chart-confirmed acute decompensation in alcohol-related cirrhosis

De-identified data and analysis code underlying a single-centre retrospective cohort study of 61 adults with alcohol-related cirrhosis who had been hospitalized more than once, and the 326 subsequent hospitalizations that followed. Every admission was adjudicated against the full medical record by physicians blinded to the hospital's structured decompensation flags. Recurrent hospitalization was an entry criterion rather than an outcome, so these data describe the composition of recurrent hospital use. They cannot be used to estimate readmission risk or predictors of frequent admission. The package contains a patient-level table (61 rows), an episode-level table (326 rows) carrying both the blinded adjudication and the final classification with an explicit provenance column for each, the independent duplicate ratings for the 62-episode reliability sample, and a Python script that reproduces every estimate reported in the accompanying paper. Free-text chart evidence is deliberately withheld because clinical narrative describing an individual admission could permit re-identification in a cohort of this size. Access is controlled rather than open: the ethics approval covering this analysis did not extend to public dissemination of individual-level records.