Abstract / Summary
Hospital readmission is used in cirrhosis research as if a return to hospital meant the disease had worsened: it is a trial outcome, a quality indicator and a prediction target. We asked what those readmissions were. Sixty-one adults with alcohol-related cirrhosis, selected because they had been hospitalized more than once, contributed all 326 later admissions to one department. A clinician read each full record and classified it for acute decompensation and for admission mode, blinded to diagnostic codes and structured flags. Fewer than two in five admissions (125, 38.3%) were decompensations. One in three (114) had been booked in advance; none of these was a decompensation, yet they used 29.1% of all bed-days. Among 124 returns within 30 days, planned admissions outnumbered decompensations, 48 to 43. Counting all admissions tracked a patients decompensations only moderately (correlation 0.50) whereas counting urgent admissions tracked them closely (0.93). In crude analysis the sicker patients had fewer admissions, not more, because planned care goes to those well enough to receive it. A liver severity score, computable for 165 admissions, did not separate planned from urgent admissions or decompensations from other stays, even within one patient. Readmission was not the same thing as decompensation.