Readmission to inpatient psychiatric hospitals within one year of discharge is approximately 40-50% (Bridge & Barb, 2004). Given that most psychiatric conditions are characterized as having a chronic, relapsing course, readmission seems quite possible. Such prevalence and course of the illness indicates that readmission is a noteworthy concern, especially related to the impact on the patient’s life, as well as the cost on the health care system (Moss et al., 2014). Additionally, readmission rates are often considered a quality of care indicator, reflecting the quality of care received while inpatient and the transition back to outpatient care (Simone, Taylor, Fung, & Kurdyak, 2013). Consequently, readmission is often perceived negatively, at times a failure, in that the goal of discharge is for the patient to successfully reintegrate into the community. Thus, understanding factors associated with readmission is vital and hopefully associated with the development of preventive measures.
This review will discuss findings from two studies, one conducted by Moss et al. (2014) in examining readmission over a period 180 days, and the other conducted by Callaly, Hyland, Trauer, Dodd, and Berk (2010) examining rapid readmission over a period of 28 days.
Moss et al. Review
Moss et al. (2014) conducted a study to determine predictors of readmission to a general psychiatry inpatient unit. They note that given the deinstitutionalization of mental health care, hospital stays are often shorter and readmission is noted to be an indicator for future admissions. In their literature review, Moss et al. indicated that history of previous admissions, length of stay, the presence of a psychiatric illness, substance abuse, personality disorder diagnosis, medical comorbidity, male gender, marital status, homelessness, unemployment, and first involuntary admission are all significant predictive factors for readmission. Also, specific to service-related factors, access to follow-up care, community support, and being discharged against medical advice were predictive of readmission.
Purpose and Methods
Moss et al.’s (2014) study conducted a retrospective review of inpatient data over a 30 month period between 2006 and 2008 at a 35 bed teaching hospital. They restricted readmission to within 180 days from the initial admission date. The assessment instrument, Minimum Data Set-Mental health (MDS-MH), was administered upon admission and prior to discharge. The MDS-MH collected data on 135 variables to capture demographic, health, and service related information. Data from patients admitted during this time period were followed for 180 days monitoring for readmissions; resulting in identification of 758 (minus exclusions) discharges, with 190 patients diagnosed (DSM-IV criteria) with Schizophrenia and related disorders and 387 diagnosed with mood disorders.
Analyses and Results
A Cox regression analysis was used to analyze variables associated with time to readmission and possible covariates. Based on the literature, Moss et al. (2014) identified variables as possible predictors: age at admission, diagnosis of schizophrenia and related disorders, level of education, marital status, length of stay, gender, diagnosis of a personality disorder, substance abuse disorders, Global Assessment of Functioning (GAF) at discharge, maximum number of alcoholic drinks in one sitting, employment, income insurance assistance, history of ER visit, vocational, history of violence, number of psychiatric admission in past two years, and receiving a pass.
Within this sample, 21% (159) were readmitted within 180 days of discharge. The sample was predominately male (45.3%), with an overall mean age of 39.6 (SD = 20.7). The mean length of stay was 19.3 (SD = 21.2) days. Covariates associated with time to readmission were receiving a pass, having one to two admissions in the past two years, and more than three psychiatric admissions in the past two years. Other variables were not found to be significant. In post hoc analyses, statistics indicate that the groups that did and did not receive a pass were not significantly different respective of diagnoses, but those with passes consisted of more men, longer lengths of stay, and higher GAF scores.
Overall, results indicate that previous admissions were associated with readmission, in that patients with one to two admissions within the past two years were 15.6 times more likely to be readmitted, and those with greater than three admissions in the past two years were 24.2 times more likely to be readmitted. Also, patients receiving a pass were 3.5 times more likely to be readmitted. Though rationale for issuing passes are variable (i.e., ease transition back into the community, assess readiness for discharge), the literature suggests that the efficacy of such a practice is not well supported (Moss et al., 2014). Unfortunately, the authors note that the use and purpose of the passes with these patients were unknown; however, Moss et al. (2014) contend that “the use of a pass does not fully mitigate the influence of these other factors” (Moss et al., 2014. P. 429). Of note, they indicate the upon discharge participation in service related treatment was also unknown, which in the past has shown to significantly influence readmission. Also, contrary to the literature, many factors previously associated with readmission were not significant in this study. Concerning these findings and compared to the literature, Moss et al. (2014) conclude that specific predictive factors are not consistently associated with readmission. Overall, previous admissions and use of passes prior to discharge were predictors of readmission at the facility sampled.
Callaly et al. Review
Purpose and Methods
Callaly et al. (2010) examined personal characteristics, characteristics of the initial admission, as well as characteristics of follow-up care after the initial admission to identify patients at high risk for readmission. Their reference period for readmission was 28 days, for which same-day admissions were excluded. The population for the study was from an integrated community and acute inpatient hospital with 20 adult beds for psychiatric admission. During the period of the study (2005-2006) there were 635 admissions, with a 12% to 13% readmission rate. Callaly et al. (2010) examined 26 variables associated with increased rapid readmission; though all variables were not listed, those selected were said to be consistent with the literature, as well as inclusive of data obtained from the Health of the Nation Outcome Scales (HoNOS) and information regarding follow-up care. The sample consisted of 54 patients with consecutive readmission and 61 patients chosen at random of whom were not readmitted. Data were analyzed through simple comparison and logistic regression.
The trend of readmission indicated that of the patients who were readmitted, 45% were readmitted within 7 days, 68% readmitted within 14 days, and 91% readmitted in 21 days. Comparison data between the readmitted group and the non-readmitted group indicated that patients who were readmitted were more likely to have had an admission the previous year, were more likely to be on Disability support, and were less likely to have had a discharge plan in place. Further, patients who received follow-up care within seven days were more likely to be readmitted. A notable data trend was that patients with Borderline Personality Disorder or who were unemployed were more likely to be readmitted. The authors note no significant difference between the groups respective of sex, discharge facility, age on onset for psychiatric care, and recent history of substance use or criminal involvement. There were also no significant differences in HoNOS scores between groups.
Of the variables examined, seven factors that were significantly (or nearly significant) different between the two groups were entered in to a multiple logistical regression analysis. Results indicate that the number of previous admissions, having no discharge plan in place or sent to the patient’s general physician, and contact with community mental health within seven days were associated with early readmission. Finally, to examine the relationship further, these results were entered into a forward stepwise regression, which again indicated that number of admissions in the previous year, no discharge plan being sent, and contact within first seven days after discharge were all significant predictors of rapid readmission.
The authors note that low treatment efficacy within the inpatient setting, poor discharge planning, and inadequate follow-up care may contribute to early readmission.
Unfortunately, the literature specific to identifying predictors for readmission are at times contradictory, and may be specific to the setting or region being studied. Overall, the most consistent and strongest predictor in Callaly et al. (2010) was the number of prior admissions. They highlight the importance of discharge planning, in that more preparation, especially for those with previous admissions, may be beneficial to the patient. Even still, the factors associated with rapid readmission are complex; and consequently, the interrelatedness of the factors associated with readmission may reflect reasons for rapid relapse. In short, the authors recommend to further examine after care of those who are not readmitted in order to identify factors associated with readmission of comparable others.
As noted by both sets of authors, readmission to a psychiatric facility is complex and multidimensional. The literature indicates many potential factors are associated with readmission, but as Moss et al. (2014) and Callaly et al. (2010) both note, the findings are at times contradictory. It appears that significant differences within the literature are related to diverse service characteristics and divergent sample characteristics being studied. Therefore, differences create unique and perhaps limited generalizability of the findings, making it difficult to act proactively regarding preventative patient care. The purpose of identifying predictive factors should be in regard to preventing readmission and providing the best care or after care for the patient. For practical use, identifying factors within a comparable setting may be useful to understanding readmission patterns for a particular type of facility.
Overall, in both studies, and consistent with the literature, having previous hospital admissions increases the likelihood of readmission. As noted by Callaly et al. (2010), examining discharge planning and after care may be necessary to direct intervention. Yet, patient treatment compliance also becomes a critical factor, in that increased efforts for after care or discharge planning may be attempted, but ultimately rests on the participation of the client. Even still, having the knowledge and awareness of potential high-risk patients for readmission may prompt practitioners to emphasize such trends upon discharge to the patient. Finally, given the diverse findings throughout the literature makes it difficult to consistently identify factors associated with readmission. Since readmission in not only detrimental to the patient but is often viewed as a quality indicator for the service provider, efforts to identify these variables should be continued.
Bridge, J. A., & Barb, R. P. (2004). Reducing hospital readmission in depression and Schizophrenia: Current evidence. Current Opinion in Psychiatry, 17, 505–511.
Callaly, T., Hyland, M., Trauer, T., Dodd, S., & Berk, M. (2010). Readmission to an acute psychiatric unit within 28 days of discharge: Identifying those at risk. Australian Health Review, 34(3), 282-228.
Moss, J., Li, A., Tobin, J., Weinstein, I. A., Harimoto, T., & Lanctoto, K. L. (2014). Predictors of readmission to a psychiatry impatient unit. Comprehensive Psychiatry, 55, 426-430.
Vigod, S. N., Taylor, V. H., Fung, K., & Kurdyak, P. A. (2013). Within-hospital readmission: An indicator of readmission after discharge from psychiatric hospitalization. Canadian Journal of Psychiatry, 58(8), 476-481.
Dannie S. Harris, M.A., M.A., M.A.Ed., Ed.S.