Predictors of early psychiatric rehospitalization: A national case register study

Nelly Zilber, Tzipi Hornik-Lurie, Yaacov Lerner

Research output: Contribution to journalArticlepeer-review

28 Scopus citations


Background: Inconsistent results have been published on variables affecting readmission to psychiatric hospitals, in particular length of hospitalization before discharge. The objective of the present study was to develop a predictive model for 30-days readmission after discharge, as the latter is one of the performance indicators in the pending mental health reform in Israel. Method: The data were extracted from the Israeli national psychiatric case register. They concerned all patients discharged from psychiatric hospitals in Israel during a six-month period (January 1,-June 30, 2004). A follow-up since the first discharge during this period (discharge from "index hospitalization") until November 30, 2005 was performed for each patient. A Cox regression allowed constructing a multi-factorial prediction model for readmission within 30 days from discharge. Results: The readmission rate within 30 days was 13%. The variables predicting early readmission were age up to 45, history of prior hospitalization, short time between index and prior hospitalization and being discharged from a hospital in the Tel Aviv-Center district. Length of hospitalization was not a predictor of early readmission, except for the very short ones (up to eight days) which predicted earlier readmission. Conclusion: The policy of shortening hospitalizations, which potentially could lead to premature discharge, was not found to be associated with early readmission, except for extremely short hospitalizations.

Original languageEnglish
Pages (from-to)49-53
Number of pages5
JournalIsrael Journal of Psychiatry and Related Sciences
Issue number1
StatePublished - 2011
Externally publishedYes


Dive into the research topics of 'Predictors of early psychiatric rehospitalization: A national case register study'. Together they form a unique fingerprint.

Cite this