2. Background
ü A substantial proportion of psychiatric inpatients are being readmitted
after discharge.
ü Readmissions are disruptive for psychiatric patients and their families;
contribute to rising costs of mental health care and readmission rate is
a commonly used indicator of the quality of care.
ü Among pre-discharge factors, the role of inpatient care has been less
frequently assessed, with the exception of LoS.
ü In the pre-discharge period an extensive number of patient-based
factors, such as clinical and socio-demographic variables, have been
examined as possible direct predictors of readmission or mediators of
other health process factors.
3. The objective of this systematic review is to describe pre-discharge
predictors of readmission after discharge from psychiatric or general
health in-patient care with a psychiatric diagnosis.
Aims
all the ones which referred to the admission period until
discharge (including the discharge phase) or to the period
before index admission
pre-discharge predictors:
4. Methods
Included:
ü Studies on the association between pre-discharge variables and inpatient readmission
after discharge were considered;
ü Patients > 18 years old and with a main psychiatric diagnosis;
ü Quantitative studies with some quantitative measures of association;
ü Papers published in English, German, Spanish, Italian and French.
Excluded:
ü The interest of the review is on the risk of being readmitted, for this reason papers
reporting only analyses on other kinds of outcomes, even if connected to readmission
(i.e. related to time to readmission or cumulative Los or number/frequency of
readmissions) were excluded.
Comprehensive literature searches were conducted in the following electronic bibliographic
databases: Ovid Medline, PsycINFO, ProQuest Health Management and OpenGrey. In addition,
Google Scholar was utilized.
Relevant publications published between January 1990 and June 2014 were included.
5. Main Results
734 unique articles identified
§ 313 excluded in the first stage after the screening of abstracts
§ 299 excluded checking the full text of the papers
122 remaining papers:
§ 14 had only outcomes related to frequency, cumulative LoS or intensity of
readmissions
§ 49 only to time to readmission
à 59 included
q 7 case-control, 5 intervention studies, 47 cohort studies
q The majority conducted in USA (61%)
q Methodology: comparison between readmitted versus not readmitted is typically
performed.
q Follow-up period: a medium time-span (between 1 month and 1 year) in
around two third of cases, with 8 papers with short (up to 30 days) and 12 with
long (more than 1 year) follow-up time
7. Age
Analysed in the majority of papers,
turning out as non-significant in the
majority of them and different direction
• risk of readmission associated with
younger age (9 papers)
• non-monotonic behavior (2 papers)
• higher risk for older age (4 papers)
Gender different directions
Marital status Being married (including also cohabitee/partner in few studies) resulted somehow
protective (9 papers)
Living situation Analysed in around 15 papers, turning
out as non-significant in the majority of
them
• living alone or not in care never
protective
• a protective role for education (3
papers)
Educational level
Employment In no case lack of employment was found as a protective factor
Ethnical group Being black and Hispanic considered, but results in different directions
Other variables resulted associated to readmission risk but analysed in few
papers:
Variables related to income Residential instability
Number of cohabitants Immigration status
Disability Support Pension Military situation
Service-connected disability Criminal justice situation
Patients’ demographic,
social and economic
characteristics
8. Diagnosis §Primary psychiatric diagnosis, is the main clinical characteristic of the patients analysed, but
results are not always consistent and turned out to be not significant in many cases
§Different direction and categories analysed, difficult to synthesize the results
§Specific diagnostic groups considered in same papers
§Psychiatric comorbidity for other psychiatric diagnosis is also explicitly examined in some
papers: the presence of a comorbid substance abuse or dependence diagnosis and
psychiatric diagnosis seems increase the risk of readmission in 11 papers; number of
psychiatric diagnosis and personality disorder in comorbidity analysed in very few paper and
different directions
Other variables resulted associated to readmission risk but analysed only
in some papers:
ü Global Assessment of Functioning, Brief Psychiatric Rating Scale and other scales
ü A history of suicide attempt
ü Problems evaluated at discharge using DSM Axis IV
ü Premorbid level of dysfunction
ü Quality of life
ü Cognitive impairment in patients with dementia
ü Proxy of severity: poor versus fair or good prognosis or requiring extensive assistance
ü Patient’s satisfaction on different aspects of hospital treatment
ü Patient’s attitude towards illness and care
Patients’ clinical
characteristics
9. Other variables resulted associated to readmission risk but analysed only
in some papers:
ü Index admission corresponding to the first onset of illness
ü Duration of illness
ü Previous use of services
ü Age at onset
ü Non-hospital pre-admission contacts
Patients’ clinical
history
Admission
history
§ It is the variable most often and consistently analysed
and associated with readmission risk.
§ In particular, it turned out to be significantly associated with
a higher risk in 29 out of 35 cases.
§ In 20 papers on readmission such relationship was found
in all the multivariate analysis performed.
10. Environmental factors
v Hospital location: comparison across national regions; urban vs rural areas
v Variables related to neighborhood environment characteristics
Health system factors
v Physician experience (using age as a proxy)
v Payment/reimbursement mechanisms and insurance coverage
Social context
v Relationship with caregivers and family system dynamics and involvement in care
Contextual factors
à health system variables evaluated at aggregated level were described in another
review of the CEPHOS-LINK project
Variables resulted associated to readmission risk, but they are
analysed only in some papers:
11. Admission and discharge
characteristics
Length of
stay
§ It was analysed as a predictor in about half of the papers; in the
majority of these, association with LoS was non-significant
§ Results were not consistent across studies
Legal
status
§ Legal status of the index admission is considered among the predictors
in 9 papers; with a higher risk for voluntarily admitted patients
found only in 2 cases
Other variables resulted associated to readmission risk but analysed only
in some papers:
ü Type of discharge (escape from hospital, discharge against medical advice, discharge
referral to other centers due to remission, discharged on medical advice)
ü Discharge destination/referral (relatives, a service in community, outpatient
commitment group, social welfare services)
ü Complications during hospitalization
ü Variables related to treatment and clinical practice during admission and at discharge
(Intervention papers)
12. Adjustment for confounding factors:
Representativeness
of the target
population
Participation
rate > 90%
Generalizability
Lost to follow-
up < 10%
Readmission to
all hospitals
previous
admissions
age diagnosis other
Fulfils
criterion
18 (30.51 %) 39 (66.10%) 48 (81.36%) 42 (80.77%) 25 (42.37%) 35 (59.32%)
46
(78.0%)
46
(79.31%)
45
(76.27%)
Does not fulfil
criterion
39 (66.10%) 13 (22.03%) 11 (18.64%) 10 (19.23%) 21 (35.59%) 23 (38.98%)
12
(20.3%)
12
(20.69%)
14
(23.73%)
Unclear 2 (3.39%) 7 (11.86%) 13 (22.03%) 1 (1.69%)
1 (1.7%)
Quality assessment
The selected studies were assessed for quality using a set of questions broadly based on the
CONSORT criteria for intervention studies and on the STROBE criteria for observational studies
13. Discussion
Some limitations have to be considered:
ü associations are not straightforward and the interactions between
factors (such as for patients severity) complicate the reading of the
real effects
ü low representativeness of the included papers
ü almost all studies used multivariate analytical methods, but only
around 60% adjusted for the number of previous hospitalizations
ü we have reported results also of bivariate analysis, and many of the
variables resulted statistically significant only at this level
14. Conclusion
ü A wide range of studies on the association between pre-discharge variables
and readmission emerged:
ü a lot of sociodemographic and clinical characteristics of the patients studied
ü less variables analysed as regard admission characteristics, clinical events or treatment
at pre discharge level and in particular during admission
ü Results difficult to summarize, as different psychiatric populations from
different countries, types of inpatient services and health systems, and the
variables examined varied largely among studies
ü different historical periods (studies conducted in the 90’ are different from the more
recent ones)
ü different mental health systems (different hospital-community balance among
countries, different readmission rates)
ü The existence or the number of previous admission were the most consistent
predictors of readmission rates, and only in few cases it resulted not
significantly associated to readmission
Editor's Notes
and changes in number of beds and the pressure of LoS reduction, according to cost containment, as well as discharge planning have to be accounted for
For example, has received much and increased attention over the past few years within the hospital process to reduce readmission; nevertheless the time available to a healthcare team to adequately prepare patients for discharge has been reduced.
Non escludiamo il paper se ha anche analisi su binary, escludiamo in caso solo le analisi specifiche
È per dire che teniamo il binary
Papers reporting only analyses on other kinds of outcomes, even if connected to readmission in inpatient care (i.e. related to time to readmission or cumulative Los or number/frequency of readmissions) - results on analyses of these outcomes in the included papers were disregarded as well-.
The analysed variables were classified according to the following categories: patients’ demographic, social and economic characteristics; patients’ clinical characteristics; patients’ clinical history; ; ; . The sections below report in detail the results for the significant variables.
Ho messo qui quelle della percezione
A variable reflecting health system characteristics may be measured at the individual level, and in this case it was included in this review; on the contrary
Relationship with caregivers and family system dynamics and involvement in care l’ho messo qui anche se forse è sociodemo ma visto che è in qualche modo legato ad aspetti di un conteto di supporto mi pareva avesse più valore qui m
Support and variables related to caregiver and social support
The majority of papers are not representative of the general psychiatric population discharged from an inpatient service.
Nearly all the datasets were from general psychiatric hospitals or inpatient psychiatric units in a general hospital (also depending on the organization of the health system in each country) and only few papers had bad generalizability as the studied settings were diagnostically specialised units
The majority of papers reported a participation rate over 90% of the selected population.
Almost all studies used multivariate analytical methods, i.e. the association between predictors and readmission was assessed controlling for confounders, but only around 60% adjusted for previous hospitalization which is the variable consistently considered associated to readmission in the literature.
Commenti da dire a voce:
This area, and for example the discharge planning process, is not been really focused in the reviewed papers, also if it could be deeply connected to the continuity of care. Therefore, the high heterogeneity of the studies included in the review do not allow a clear comparison of their results.
However, in the majority of these cases there could be two reason for these particular result: other covariates associated with previous psychiatric admission were included (such as cumulative LoS, age at onset, etc.); or authors have selected a particular sample of ‘high or low users’’ patients.
For other quality aspects, instead, almost all studies used multivariate analytical methods, controlling for confounders the association between predictors and readmission, but we noted that only around 60% adjusted for the number of previous hospitalizations which is the variable consistently considered associated to readmission in the literature.
Most differences also regard the outcome variable considered (“admission rates” versus “time to readmission”). Moreover, when admission rates were considered, they could be calculated with a short time period (early readmission) or using longer periods. In this review only “readmission rates” studies have been included while studies on “time to readmission” were excluded.
.
Most of the differences among studies are also probably due to different historical periods (studies conducted in the 90’ are different from the more recent ones), and to different mental health systems (countries where the balance between hospital and community is different will show different readmission rates and different influencing factors).