Factors Associated with Readmission of Patients at a University Hospital Psychiatric Ward in Iran

Objectives. Readmission has a major role in the reduction of the quality of life and the increase in the years of lost life. The main objectives of this study were to answer to the following research questions. (a) What was the readmission rate? (b) What were the social, demographic, and clinical characteristics of patients admitted to the Psychiatric Emergency Service at Nour University Hospital, affiliated to Isfahan University of Medical Sciences, Isfahan, Iran? (c) What were the effective factors on readmission? Method. This cross-sectional study was conducted on a total number of 3935 patients who were admitted to Isfahan University Hospital Psychiatric Ward in Isfahan, Iran, from 2004 to 2010. Gender, age, marital status, education, self-report history of previous admission, type of psychiatric disorder, substance misuse, suicide, and the length of the current psychiatric disorder were collected from the registered medical files of patients. The data were analysed using the negative binomial regression model. Results. We found that factors such as psychiatric anxiety disorder, bipolar I, bipolar II, psychotic disorder, depression, and self report history of previous admission were statistically significant in the number of readmissions using the negative binomial model. Conclusion. Readmission to the psychiatric ward is mainly predictable by the type of diagnosis and psychosocial supports.


Introduction
During the last three decades, the deinstitutionalization of mental health services led to faster transitions from psychiatric hospitals to the community. However, a signi�cant number of patients have had serious problems aer being discharged from hospitals which resulted in an increase in the number of psychiatric emergency referrals and readmissions [1]. e greater the number of readmissions, the greater the pressure on hospitals psychiatric wards and their limited resources.
Readmission of inpatients has been one of the most important problems in the �eld of psychiatry for the last decades [2,3]. e problem has a major role in reducing the quality of life and increasing the years of lost life [4]. About one-third of patients admitted to psychiatric services will probably be readmitted within a year [5]. In previous studies, the rate of readmitted patients has proven to be 22% in America in psychiatric patients aged over 65 years [4], 19% of children admitted at the emergency ward in Baltimore were aged 4-18 [6], 14% of patients admitted at the psychological ward of the army in Washington [7], 33.7% of patients diagnosed with schizophrenia in Turkey [8], and 43% of teenage patients admitted for psychological reasons in Northern Carolina [3]. e readmission period has been de�ned controversially in different studies, for example, readmission in a period of 90 days [9], readmission in a period of 3 years [10], admission of three or more times in a period of 30 months [11], readmission in a period of 6 months [12], and admittance of 3 or more times in a period of 2.5 years [13]. is, in turn, resulted in different rates of readmission in different studies as 14% [7], 16% [12], 20-30% [10], and 45-53% [13].
Studying the characteristics related to readmission may help to improve management programs. e factors such as demographics, social, and clinical factors have been studied for more than three decades [14]. In addition, poor adherence to treatment [8,[15][16][17], nonvoluntary �rst admission [14,17], substance, alcohol, and drug abuse [4,7,14,16] have been mentioned as related factors to the readmission of patients to psychiatric wards. Some studies have illustrated that diagnoses are related to readmission. For example, the diagnosis of schizophrenia has been reported as a major factor on readmission [4,15,[17][18][19][20].
�ength of stay at hospital has also been noti�ed as a highly related factor [15,17,19,21]. e shorter the length of the stay at a hospital, due to the lack of empty beds, the more readmission of psychiatric patients. is could decrease the quality of care and de�nitely increase the costs [10]. In some studies, the history of admission in elderly people has been pointed as a related factor [6,7,15,18,19]. e diagnosis of bipolar disorder [4,14], psychotic disorders [7,21], and mood disorders [3,7] is also mentioned as an effective factors on readmission.
From the demographic characteristics marital status, unemployment [18,19], retirement [19], and gender [8,18] are the most relevant factors to readmission. In some studies, being diagnosed with more psychiatric comorbidities [6] was also a factor associated with readmission.
Suicide [6], history of psychological problems in childhood, and �rst episode of the psychiatric disorder under age of 18 years have been revealed as related factors to readmission [7].
Regarding the mentioned �ndings, we need a better understanding of the factors related to readmission to help better management and early intervention. Understanding these factors will decrease the number of readmissions and hospital costs. It seems that those inpatients who are seeking rehospitalisation tend to repeat some of these treatmentseeking behaviors and may assist nurses in planning care [2,4,6,14,15].
In more detail, we had the following research questions. (a) What was the readmission rate ? (b) What were the social, demographic, and clinical characteristics of patients admitted to Psychiatric Emergency Service at Nour University Hospital, affiliated to Isfahan University of Medical Sciences, Isfahan, Iran? (c) What were the effective factors on readmission? We try to answer these three questions in Section 3.

Material and Method
In this cross-sectional study, data were collected from the medical �les of 3935 patients with psychiatric disorders who were admitted to the Nour Hospital Psychiatric Ward affiliated to Isfahan University of Medical Sciences, Isfahan, Iran, from September 2004 to September 2010. e patients were diagnosed to have at least one major psychiatric disorder based on Diagnostic and Statistical Manual of Mental Disorders, the fourth edition (DSM-IV) criteria. e major diagnoses were bipolar one disorder (BID), bipolar two disorder (BIID), bipolar disorder not otherwise speci�ed (BD-NOS), psychotic disorders (not mood related), anxiety disorders (all types were considered as one category diagnosis), major depressive disorder (MDD), substance induced mental disorders, and cognitive disorders. All data were gathered through a semistructured interview, which was done by a board of certi�ed attending psychiatrists and psychiatric residents. Readmitted patients were taken as the independent variable which was numeric, but for descriptive goals the number of admissions were categorised into three groups of one admission, two admissions, and three or more admissions based on the above-mentioned major diagnoses. It has been assumed that there is no possibility for the patient to change diagnosis between two admissions. Demographic characteristics such as sex, age, marital status (single, married or widowed, and divorced), and education (illiterate, primary school, high school, and higher) were also analysed. Variables such as length of the current episode before hospitalization, current substance misuse, smoking, alcohol consumption, history of admission, and history of attempt to suicide, based on the patients' self reports, were also applied. e presence or absence of each of the diagnoses, smoking, substance misuse, alcohol consumption, history of admission, and history of suicide was binary. Patients might simultaneously be diagnosed with more than one psychiatric disorder. Because the response variable (number of readmissions) was numeric without normal distribution and small spread within data, the negative binomial regression model was used to analyze the data set to determine the factors related to readmission. A value of less than 0.05 was considered as signi�cant.

Results
Out of the 3935 inpatients, 2158 (54.8%) were male. e age mean and median were 33.1 and 30 years, respectively, with a range of 14-86 years. e 95% con�dence interval (CI) of the age of the patients was 32.7-33.5. e mean (95% CI) of age in male and female patients were 33.8 (33.2-34.4) and 32.5 (31.9-33.1), respectively. ere was a statistically signi�cant difference between males and females' mean of age ( value = 0.002) (results not shown). e readmission rate of inpatients varied between 1 and 10 times in the number of previous admissions. Of the total number of inpatients, 3235 (82.3%) had only one admission, 501 (12.7%) were admitted two times, 126 (3.2%) had three times, and 75 (1.9%) were admitted four or more times.  Table 2 which is referred to the second research question. Negative binomial regression models were used to determine the variables related to the number of times of readmission which referred to the last research question. e results showed that being divorced had a statistically signi�cant correlation with a higher readmission count. e mean of the number of times of admissions for divorced patients was 1.47 times and 1.56 times more than the mean for single patients and married patients, respectively. e education status of the patients had no statistically signi�cant correlation with the readmission rate.  Regarding diagnoses, BID, BIID, psychotic disorders (not mood related), MDD, and anxiety disorders showed to be signi�cantly related to the number of times of readmission. However, BID-NOS, cognitive disorders or misusing drug was not signi�cantly related to the number of times of readmission. e mean of the number of times of readmission in patients who misused opium was about 1.3 times more than in patients who did not ( . e results showed a signi�cant relationship between self report history of a previous admission and the current admission of patients. e mean of the number of times of a current admission in patients with positive history of admission was about 2.2 times more than patients without self report of previous history of admission ( < . e results indicate that there was no statistically signi�cant relationship between the number of times of readmission and the length of the current episode before hospitalization. More information about the results of �tting the negative binomial regression model is illustrated in Table 3.

Discussion
e reported factors related to the readmission of patients with psychiatric disorders were various due to objective limitations, type of studies [2,17], or type of models used for analyzing the data. is study showed that the type of diagnosis was signi�cantly associated with the number of readmissions and all types of the above-mentioned psychiatric disorders, except bipolar NOS and cognitive disorders. Some studies indicated a relationship between schizophrenia and the number of readmissions [4,15,[18][19][20]. In our study, schizophrenia was also mentioned as a subcategory of psychotic disorders (not mood related). Other studies revealed that bipolar disorder increases the number readmissions [4,14]. In the current work, BID and BIID, but not bipolar disorder NOS, were signi�cantly related to readmission.
e length of the current episode was not identi�ed as a statistically signi�cant factor for conse�uent rehospitalization. However, in one study, length of stay at hospital was a predictor of referral to the aercare in patients who had been diagnosed for schizophrenia [12].
Some studies showed that the history of previous admission may lead to an increase in the number of readmissions [6,7,15,18,19]. In one study, patients with history of previous admission had a greater number of readmissions in comparison to patients without such history [2]. Our study revealed the same results.
In conclusion, the readmission to the psychiatric ward is mainly predictable by the type of diagnosis and psychosocial supports. e results of this study could be taken into account Psychiatry Journal 5 to help better manage the admission of patients seeking rehospitalisation. It may also lead to a better understanding of early intervention and may assist nurses in planning care.
e �rst weaknesses of our study was that the Global Assessment Functioning (GAF), as one the important predictors of outcome measures, was not determined [22]. e second, was not being able to assess accurately the outof-network hospitalization experience in the community. It means this could lead to a selection bias, and as a result it may affect the outcome of this sample. erefore, the generalization of our �ndings to other groups or communities is uncertain. As a limitation, we assumed that there is no possibility for the patient to change diagnosis between two admissions.