Crisis discharges and readmission risk in acute psychiatric male inpatients
© Niehaus et al; licensee BioMed Central Ltd. 2008
Received: 07 December 2007
Accepted: 17 June 2008
Published: 17 June 2008
Severe pressures on beds in psychiatric services have led to the implementation of an early ("crisis") discharge policy in the Western Cape, South Africa. The study examined the effect of this policy and length of hospital stay (LOS) on readmission rates in one psychiatric hospital in South Africa.
Discharge summaries of adult male patients (n = 438) admitted to Stikland Psychiatric Hospital during 2004 were retrospectively examined. Each patient's clinical course was then analysed for the period between January 1st, 2004, and August 31st, 2006.
Although shorter LOS was associated with decreased readmission rates, the effect of crisis discharges was far more powerful. Patients discharged as usual had a far lower risk of readmission than those discharged due to bed pressures (i.e. crisis discharge).
Increased risks associated with the early discharge policy necessitate the urgent review of the current management of bed shortages in this inpatient facility. The strengthening of community initiatives, particularly assertive outreach could be a way forward.
One of the global trends in psychiatric care in recent decades has been large scale deinstitutionalization, leading to a reduction in number of beds available and shorter length of hospital stay (LOS) for most patients [1, 2]. Data from the US shows that psychiatric length of stay in all types of hospitals continued to decrease between 1988 and 1992 with it being most noticeable in psychiatric hospitals where length of stay declined from 75 to 56 days. In the same period there was a reduction of 12.5 million inpatient days in psychiatric hospitals in the US alone . In the UK there is a similar trend with the number of in-patient psychiatric beds in England having fallen dramatically over the past four decades . In addition to shortening of length of stay, Lay et al  also suggests an overall redistribution of treatment resources with decreased inpatient treatment for people with schizophrenia and an increase for affective disorders. This decrease in inpatient care for specific diagnostic categories is supported by at least some data from the US, that also suggests the decline in inpatient treatment has not led to an increase in outpatient visits for people with schizophrenia .
The shortening of length of stay seemed to be supported by early studies in the 1970's which showed no difference in readmission rates when comparing short vs. long term hospitalizations [6, 7]. This was seen as evidence for the support of the deinstitutionalization process as it was suggested that longer hospitalization leads to difficulties for patients to re-enter the real world. The data examining the relationship between length of stay and rate of readmission have however not been unequivocal. For example a Cochrane meta-analysis could demonstrate no effect on readmission rates by planned short stay admissions, but the authors also stressed the need for more, large, well-designed trials, especially in the developing world .
Most clinically stable patient in the ward
Not pose an immediate threat to him-/herself or others
Less ill than the patient that needs urgent admission
Most practical follow-up arrangements have been put into place prior to discharge
Despite the fact that changes in governmental policy and legislation are usually aimed at addressing shortcomings in the health care system, the impact of these changes are often not predictable. For example, Bauer et al. 2007 reported how changes in Israeli mental health policies ultimately led to an increase, rather than a decrease, of involuntary admissions .
Our aim was to determine the effects of the crisis discharge policy on readmission rates of acute psychotic male inpatients.
The crisis discharge policy
The 'crisis' discharge policy was adopted by the Associated Psychiatric Hospital management team in the Western Cape Province, South Africa, to deal with severe acute bed shortages in the state psychiatric sector .
This retrospective study (N05/03/047) was approved by the Committee for Human Research at the University of Stellenbosch, and patient confidentiality was protected by using only a study number linked to the particular patient. Data was captured by examining the discharge summaries of all male patients between the ages of 18 and 60 (extremes included) admitted and discharged from the acute wards at Stikland Psychiatric Hospital (SPH) in the year 2004. SPH is a state psychiatric facility with specialized wards for patients with acute psychosis. The genders are segregated and the facility has 80 such beds for male patients. As the bed shortage in the female wards is less pronounced, the emergency discharge policy is used much less and this study therefore focused on male patients only. It serves a mixed urban and rural community of 1.5 million people in the Western Cape Province of South Africa. Patients who were transferred to chronic psychotic or non-psychotic wards, to other hospitals, who self-discharged or died, were excluded from the analysis. Readmissions up until 31 August 2006 were included in the dataset. Data were captured using the patients' hospital folder number as reference and included variables for patient demographic characteristics, five axis DSM IV diagnosis, length of hospital stay (LOS) and discharge status (crisis discharge or discharge as usual).
Marital status was divided into 2 broad groups (either single or not single) and the diagnoses were divided into patients that had a co-morbid substance disorder diagnosis (HR = high risk) and those not having a co-morbid substance disorder diagnosis (LR = low risk). The time to readmission was calculated as the time between the first discharge and second admission, and was used as the outcome variable in the analysis. Crisis discharge and LOS were the main predictors.
Summary statistics for index and readmission data
Time to Readmission (days)
Median (n = 438)
Crisis discharge group
Non-crisis discharge group
*LOS <= 39 days
LOS > 39 days
Re-admission Yes (n = 163)
Re-admission No (n = 275)
Total (n = 438)
Crisis discharge group
Non-crisis discharge group
Marital status: Single
Marital status: Not single
**Substance use disorder
High risk group
Substance use disorder
Low risk group
LOS <= 39 days
LOS > 39 days
Results of Cox Proportional Hazards Regression
Estimated Hazard Ratio
95% Confidence Interval
Wald test = 26.6 on 7 df, p = 0.000391
The participants' mean age on admission was 32.9 (s.d. = 10.4) years. Most patients (n = 356 or 77%) belonged to the group made up of single, divorced or widowed people. The majority of the patients were primarily Afrikaans speaking (n = 359 or 82.3%). Of the DSM IV axis I diagnoses, n = 307 (70.0%) were accounted for by just three diagnoses; Schizophrenia (n = 207 or 47.2%), Bipolar Disorder (n = 67 or 15.3%) and Schizo-affective disorder (n = 33 or 7.5%). Of these, 119 (38.9%) exhibited a co-morbid substance-related disorder. The majority of patients (n = 339 or 77.2%) were involuntary admissions and 15% (n = 67) were assisted users (user not refusing admission but currently regarded as not competent to judge treatment needs).
Readmissions and Crisis Discharge
Of the 438 admissions included in the analysis, 180 patients (41.0%) were crisis discharges on their first discharge, whilst 254 (58.0%) were discharged as usual (missing data for four patients). The mean LOS for all admissions was 43.9 (s.d. = 39.4) days. For the crisis discharges, the mean LOS was 40.6 (s.d. = 32.7) days, while for the non-crisis discharges it was 46.4 (s.d. = 43.7) days. The median time to readmission was longer (688 days) for the non-crisis discharge group than for the crisis discharge group (628 days).
During the entire study period, 163 (37.2%) of the 438 index admissions were readmitted to hospital. This means that more than half the observations, 275 (62.8%) were censored. Of the readmissions, 81 (50.6%) were crisis discharge patients. 45% of the crisis discharge group were readmitted, while the same was true of 31% of the non-crisis discharge group.
Time to readmission
If we refer to Fig. 1, we see that there is (visually) a difference between the survival functions of the crisis and non-crisis discharge groups. The non-crisis discharge group has a higher survival time, where survival time is the time to readmission. The graph includes 95% confidence intervals which generally support the result that the two groups are different.
The Cox regression results in Table 2 show that crisis discharge had a significant influence on the time to readmission (Hazard ratio = 1.646, P = 0.002, CI 1.200, 2.260). A hazard ratio of 1.646 implies that the hazard for the crisis group is 1.6 times that of the non-crisis discharge group, so the crisis discharges will be readmitted sooner than those that received a complete treatment. The table also shows that marital status and income had no influence on time away from the hospital until readmission. As expected, LOS had a marginally significant effect on time away from the hospital (Hazard ratio = 1.004, P = 0.022 CI 1.001, 1.010). LOS and crisis discharge did not interact significantly with each other, nor did any of the other factors interact significantly.
Impact of crisis discharge
Our study showed that so-called "crisis discharge" was associated with a significantly increased risk of readmission and shorter time until readmission and this was independent of the impact that LOS had on readmission rates. The link between increased risk of readmission and crisis discharges was not unexpected, considering that such discharges are, by definition, sudden and unplanned. In South Africa and internationally, maintaining patients in the community, prevention of relapse and reduction of risk of readmission are often the responsibilities of the community psychiatric services [14, 15]. Patients who are discharged as usual are more likely to be better engaged and more likely to follow up with outpatient care . Furthermore, a pre-discharge program makes it easier to address issues such as co-morbid substance use or abuse , which was highly prevalent in our sample (27.1%) although it did not increase readmission risk.
Length of hospital stay
In this study, we found that decreased LOS led to longer time until readmission. However, the results in Tables 1 &2 are at best borderline and therefore inconclusive. Figueroa et al found a direct relationship between decreasing LOS and readmission rates when examining private psychiatric inpatients. Similarly, Appleby et al found higher readmission rates in schizophrenic patients with a LOS of less than 30 days compared to those with a length of stay of more than 30 days in a public psychiatric hospital. Heeren et al as well as Wickizer et al found a positive correlation between shorter LOS and readmission rates in a geriatric and child and adolescent unit respectively. It has therefore been postulated that too short a LOS does not allow for a resolution of the patient's clinical condition nor allow adequate preparation for the patient's discharge, thereby contributing to a revolving door effect [9, 21].
However, a number of other studies [2, 8, 22] did not find any adverse outcomes for short hospital stays. It is not surprising that studies examining length of stay and readmission rates have not always reported consistent findings. There is a large variability in reported lengths of stay in different studies, with private institutions reporting a mean LOS of 7.1 and 6.7 days [14, 16] and psychiatric hospitals a mean LOS of 50.0 and 63.2 days [23, 24]. The latter compares more closely with ours of 44 days. The results from these studies are also influenced by factors such as different hospital types (general versus psychiatric), sources of funding (state versus private), differing study populations (adult, older and young adults) and the large variability in psychiatric diagnoses, natural course and response to treatment. Importantly, in many of these studies, short hospital stays were often pre-planned [22, 25, 26]. Also, in the case of private psychiatric hospitals the length of stay is usually predetermined at the time of admission which might stress the importance of discharge planning irrespective of the LOS. This differs from the crisis discharge, which is, by definition, sudden and therefore unplanned.
Although our one-year readmission rate of 15.5% can be favourably compared to Lyons et al's  six-month readmission rate of 17.6% and Segal et al's  one-year readmission rate of 29%, we present evidence that the crisis discharge policy may exacerbate a revolving door effect in Stikland psychiatric hospital. Such frequently-returning patients may contribute significantly to costs and bed-occupancy, thereby counteracting the intended cost reductions that were the motivation for, inter alia, shorter LOS. In addition, the clinical management of frequently admitted patients may be adversely affected through demotivation of staff and therapeutic nihilism  if readmitted patients are viewed as 'regulars' who have familiar, unchanging repetitive issues and patterns of admissions within an already struggling and short-staffed mental health system.
Strengths and limitations
A major strength of this study was that all admissions were from one hospital and data/participants were evaluated in a standardized fashion. This yielded a sizeable study sample that was followed up over a long period, comparing very favourably to sample sizes and study period of other studies [e.g ]. The use of one treatment facility lessened the impact of physician variability on the use of the crisis discharge policy and the retrospective nature of the trial reduced any potential bias regarding LOS. The impact of patient variables were also lessened due to the large variability of patients within our drainage area, which included both involuntary and voluntary, state and medical aid patients from different ethnic and socio-economic backgrounds.
Staggered admissions may have had an impact on readmission rates and a longitudinal follow-up design may uncover valuable data obscured by our current analysis. Admission data prior to the implementation of the crisis discharge policy, which could have strengthened this study substantially, was not available. This is especially significant if the total number of admissions over the last few years had increased without a concomitant increase in the number of inpatient beds, as this would have a direct bearing on the number of crisis discharges for the year. Finally, readmissions would have been missed if patients were admitted to a private or other psychiatric hospital, but this would have introduced a positive bias to our results.
This is the first study to address the possible impacts of the Western Cape Province's APH crisis discharge policy on patient rehabilitation and readmission. Further research is clearly needed on the implications of these findings as well as ways of reviewing crisis discharge policies and its adverse outcome on readmissions. LOS and the crisis discharge policy seem to exacerbate the revolving door effect in this psychiatric hospital. Readmission is often used as quality indicator for inpatient psychiatric services, and could be seen as a failure of the earlier hospital admission especially when it occurs within a relatively short-time after a previous discharge. Since the main factor influencing the crisis discharge policy is inpatient bed availability, provision for increasing the number of available acute psychiatric beds needs to be seriously considered in an effort to reduce the incidence of crisis discharges. Additionally, or alternatively, strengthening community based services, particularly outreach initiatives, would be important and we believe this to be the way forward.
Dr Aliza du Toit is acknowledged for her contribution in revising the drafted manuscript.
- Lay B, Nordt C, Rossler W: Trends in psychiatric hospitalisation of people with schizophrenia: a register-based investigation over the last three decades. Schizophr Res. 2007, 97: 68-78. 10.1016/j.schres.2007.07.006.View ArticlePubMedGoogle Scholar
- Johnstone P, Zolese G: Length of hospitalisation for people with severe mental illness. Cochrane Database Syst Rev. 2000, CD000384-Google Scholar
- American Hospital Association: AHA Hospital Statistics, 1993–1994. 1993, Chicago, Ill, AHAGoogle Scholar
- Thompson A, Shaw M, Harrison G: Patterns of hospital admission for adult psychiatric illness in England: analysis of hospital episode statistics data. British Journal of Psychiatry. 2004, 185: 334-341. 10.1192/bjp.185.4.334.View ArticlePubMedGoogle Scholar
- Leslie DL, Rosenheck R: Changes in inpatient mental health utilization and costs in a privately insured population, 1993 to 1995. Med Care. 1999, 37: 457-468. 10.1097/00005650-199905000-00005.View ArticlePubMedGoogle Scholar
- Glick ID, Hargreaves WA, Drues J, Showstack JA: Short or long hospitalization for psychiatric disorders? Two-year results [proceedings]. Psychopharmacol Bull. 1977, 13: 48-49.PubMedGoogle Scholar
- Glick ID, Hargreaves WA, Drues J, Showstack JA, Katzow JJ: Short vs long hospitalization: a prospective controlled study. VII. Two-year follow-up results for nonschizophrenics. Arch Gen Psychiatry. 1977, 34: 314-317.View ArticlePubMedGoogle Scholar
- Johnstone P, Zolese G: Systematic review of the effectiveness of planned short hospital stays for mental health care. BMJ. 1999, 318: 1387-1390.View ArticlePubMedPubMed CentralGoogle Scholar
- Segal SP, Akutsu PD, Watson MA: Factors associated with involuntary return to a psychiatric emergency service within 12 months. Psychiatr Serv. 1998, 49: 1212-1217.View ArticlePubMedGoogle Scholar
- Herring LM: Management of full acute admission services in the Associated Psychiatric Hospitals. Associated Psychiatric Hospital Circular Number 3/2006. 2006Google Scholar
- Bauer A, Rosca P, Grinshpoon A, Khawaled R, Mester R, Yoffe R, et al: Trends in involuntary psychiatric hospitalization in Israel 1991–2000. Int J Law Psychiatry. 2007, 30: 60-70. 10.1016/j.ijlp.2006.02.002.View ArticlePubMedGoogle Scholar
- 2008, [http://www.r-project.org/]
- SPSS for Windows. [9.0]. 1999, Chicago, SPSS Inc
- Lyons JS, O'Mahoney MT, Miller SI, Neme J, Kabat J, Miller F: Predicting readmission to the psychiatric hospital in a managed care environment: implications for quality indicators. Am J Psychiatry. 1997, 154: 337-340.View ArticlePubMedGoogle Scholar
- Koch A, Gillis LS: Non-attendance of psychiatric outpatients. S Afr Med J. 1991, 80: 289-291.PubMedGoogle Scholar
- Figueroa R, Harman J, Engberg J: Use of claims data to examine the impact of length of inpatient psychiatric stay on readmission rate. Psychiatr Serv. 2004, 55: 560-565. 10.1176/appi.ps.55.5.560.View ArticlePubMedGoogle Scholar
- Haywood TW, Kravitz HM, Grossman LS, Cavanaugh JL, Davis JM, Lewis DA: Predicting the "revolving door" phenomenon among patients with schizophrenic, schizoaffective, and affective disorders. Am J Psychiatry. 1995, 152: 856-861.View ArticlePubMedGoogle Scholar
- Appleby L, Luchins DJ, Desai PN, Gibbons RD, Janicak PG, Marks R: Length of inpatient stay and recidivism among patients with schizophrenia. Psychiatr Serv. 1996, 47: 985-990.View ArticlePubMedGoogle Scholar
- Heeren O, Dixon L, Gavirneni S, Regenold WT: The association between decreasing length of stay and readmission rate on a psychogeriatric unit. Psychiatr Serv. 2002, 53: 76-79. 10.1176/appi.ps.53.1.76.View ArticlePubMedGoogle Scholar
- Wickizer TM, Lessler D, Boyd-Wickizer J: Effects of health care cost-containment programs on patterns of care and readmissions among children and adolescents. Am J Public Health. 1999, 89: 1353-1358.View ArticlePubMedPubMed CentralGoogle Scholar
- Segal SP, Akutsu PD, Watson MA: Involuntary return to a psychiatric emergency service within twelve months. Soc Work Health Care. 2002, 35: 591-603. 10.1300/J010v35n01_13.View ArticlePubMedGoogle Scholar
- Glick ID, Hargreaves WA, Goldfield MD: Short vs long hospitalization. A prospective controlled study. I. The preliminary results of a one-year follow-up schizophrenics. Arch Gen Psychiatry. 1974, 30: 363-369.View ArticlePubMedGoogle Scholar
- Gastal FL, Andreoli SB, Quintana MI, Almeida GM, Leite SO, McGrath J: Predicting the revolving door phenomenon among patients with schizophrenic, affective disorders and non-organic psychoses. Rev Saude Publica. 2000, 34: 280-285.View ArticlePubMedGoogle Scholar
- Spiessl H, Hubner-Liebermann B, Binder H, Cording C: [Heavy users in a psychiatric hospital – a cohort study on 1811 patients over five years]. Psychiatr Prax. 2002, 29: 350-354. 10.1055/s-2002-34659.View ArticlePubMedGoogle Scholar
- Mattes JA, Klein DF, Millan D, Rosen B: Comparison of the clinical effectiveness of "short" versus "long" stay psychiatric hospitalization. IV. Predictors of differential benefit. J Nerv Ment Dis. 1979, 167: 175-181.View ArticlePubMedGoogle Scholar
- Salokangas RK: Hospital and outpatient care for psychotic patients during the last three decades. Subsequent hospital and outpatient treatment of psychotic patients hospitalized for the first time in 1949–50, 1959–60 or 1969–70. Acta Psychiatr Scand. 1980, 62: 47-62. 10.1111/j.1600-0447.1980.tb00592.x.View ArticlePubMedGoogle Scholar
- Bernardo AC, Forchuk C: Factors associated with readmission to a psychiatric facility. Psychiatr Serv. 2001, 52: 1100-1102. 10.1176/appi.ps.52.8.1100.View ArticlePubMedGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-244X/8/44/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.