当前位置： SCI文献检索 > Epidemiology and Psychiatric Sciences期刊下所有文献 > Exploring the predictors of early readmission to psychiatric hospital.

Exploring the predictors of early readmission to psychiatric hospital.

Abstract：

BACKGROUND:Aims of this study are to explore the associations of readmission to psychiatric hospital over time, to develop a statistical model for early readmission to psychiatric hospital and to assess the feasibility of predicting early readmission. METHOD:The sample comprised 7891 general psychiatric discharges in South London, taken from a large anonymised repository of electronic patient records. We initially explored time to readmission using Cox regression - this included investigation of time-dependent effects. Subsequently, we used logistic regression to create a predictive model for 90-day readmission. We investigated the effect on readmission of a set of variables that included demographic variables, diagnosis and legal status during the index admission, previous service use, housing variables and individual item scores on the Health of the Nation Outcome Scales (HoNOS) at admission and at discharge. RESULTS:Fifteen per cent of those discharged were readmitted within 90 days. Cox regression demonstrated that the estimated baseline hazard of readmission declined steeply after discharge and that the effects of several predictors, especially diagnosis, changed over time - most notably, personality disorder was associated with increased readmission relative to schizophrenia at the time of discharge, but did not significantly differ by 1-year postdischarge. In the logistic regression, increased readmission was associated with personality disorder diagnosis; shorter length of the index admission (excepting zero length admissions); number of discharges in the preceding 2 years; and having a high score at discharge on the HoNOS overactive and aggressive behaviour item, cognitive problems item or hallucinations and delusions items. Detention under Section 3 or a forensic section of the Mental Health Act during the index admission was associated with reduced readmission. The coefficient of discrimination for the logistic regression, which is equivalent to r 2, was 0.04 and the estimated area under the receiver operating curve was 0.65. CONCLUSIONS:The association found between early readmission and personality disorder diagnosis merits further investigation, as does the possible trade-off between reduction in length of stay and increased readmission. Other novel findings such as the associations found with HoNOS item scores also merit replication. As with previous studies, we found that the rate of readmission declines steeply after hospital discharge, so that the period immediately subsequent to discharge is a period of comparatively high risk. However, prediction of early readmission within this high-risk group remains challenging - it seems most likely that many unmeasured influences operate subsequent to the time of discharge.

journal_name

Epidemiol Psychiatr Sci

journal_title

Epidemiology and psychiatric sciences

authors

Tulloch AD,David AS,Thornicroft G

doi

10.1017/S2045796015000128

subject

Has Abstract

pub_date

2016-04-01 00:00:00

pages

181-93

issue

eissn

2045-7960

issn

2045-7979

pii

S2045796015000128

journal_volume

pub_type

杂志文章

在线工具

Exploring the predictors of early readmission to psychiatric hospital.