Predictors of Hospital Readmission for Children with Psychiatric Illness who have Received an Initial Course of Treatment

Copyright: © 2016 Kagabo R, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. *Corresponding author: Robert Kagabo, Department of Family and Preventive Medicine, Division of Public Health, University of Utah, Salt Lake City, Utah, USA, E-mail: robert.kagabo@hsc.utah.edu


Introduction
The report that one out of every five children in any given year suffers from a mental disorder in the United States [1] is alarming and signifies that mental health among children is a challenge resulting in several public health concerns.Recurrent hospital readmission of children with psychiatric illness has been identified as one of these public health concerns.To begin to research the public health concerns or develop any interventions, it is necessary to identify factors that predict such readmissions.In this study, we used data of children hospitalized for treatment of psychiatric illness at the University of Utah's Neuropsychiatric Institute between 1999 and 2010.We looked at several factors, such as demographics, insurance type, primary diagnosis at the time of admission, and hospital length of stay to see if they are possible rehospitalizaton predictor variables.multiple readmissions and this recurrent hospitalization is a public health concern.In a prospective cohort study done in New York State, where 109 children were followed, research found that 37 of them were readmitted within the first year of discharge [2].Managed care imposed restrictions on length of stay have been cited as possible explanations for possible increases.A study in 1995 which focused on managed care restrictions found that in Massachusetts, the 30-day readmission of children and adolescents was 7.5% before imposing restrictions and increased to 10.1% after the restrictions, while in Tennessee, the 30-day readmissions went from 9.2% in 1995 to 12.2% in 2000.A retrospective cohort design study in Maryland followed 522 Medicaid eligible adolescents who were admitted to psychiatric hospitals between July 1997 and June 1998 and found that out of the 522 adolescents, 198 (37.9%) had at least one readmission encounter within a year of discharge [5].
In a different study, an increase in admission rates was compared to discharges among children and adolescents compared to adults and the elderly.This study used the yearly U.S. National Hospital Discharge Survey data and found that discharges of child and adolescent psychiatric patients had increased significantly from 155 per 100,000 in 1996 to 283 Open Access 2 per 100,000 in 2007 compared to other age groups, yet the total inpatient days also increased significantly [6].If there was no increase in admission or readmission, an increase in discharge would indicate a decrease in total inpatient days.The one-year window of observation after discharge from pediatric psychiatric inpatient has revealed a number of outcomes such as: suicide attempts, high mortality risk, and readmissions [2,7,8].The reports vary but the limited research done indicates that between 25 and 33% of children and adolescents are readmitted within the first year of discharge with most of them occurring within the first 3 months [2,7,9].Blader [2] followed 109 children, ages 5 to 12, for one year after their discharge from acute inpatient care.His main outcome variable was time to re-hospitalization and he found that 81% of the rehospitalizations occurred within 90 days following discharge.The variables in the study that predicted rehospitalization risk were severe conduct problems, harsh parental discipline, and disengaged parent-child relations.Lyons et al. [10] conducted a study where they followed 255 patients admitted to any of the seven psychiatric hospitals in a regional managed care program in the Chicago area.These patients were followed for a 6-month period and the investigators found that 17.6% of the sample was readmitted within the 6-month period, with 7.1% of the readmissions occurring within 30 days of discharge.
In Taiwan, the relationship between psychiatrist caseload and length of inpatient stay was investigated with a study sample size of N=66959.Thirty-two percent of the sample was readmitted within 30 days.Analysis was done to compare a 30-day readmission rate according to psychiatrists' caseload categorized as: low, medium, high, and very high.The readmission rate was on average higher for psychiatrists with higher volume of caseload.Another result was the decrease in the length of inpatient stay (LOS) observed with increase in psychiatrist's caseload.Inpatients had lengths of stay on average that were 1.22, 2.03, and 7.59 days shorter for medium, high, and very high psychiatrists' caseload volume, respectively, compared to psychiatrists with low caseload volume [11].These results seem to indicate an association between psychiatrists' caseload and readmission; and they remained true for the different disorder categories, namely schizophrenia, major depressive disorder, bipolar disorder, and others.In 2010, in another study in Taiwan, the researchers compared readmission rates and predictors in 14 days, one year and five years after discharge.The study consisted of 44,237 firsttime psychiatric inpatients discharged in 2000 and then followed for five years.The readmission cumulative incidences reported were 6.1% at 14 days, 22.3% at one year, and 37.8% at five years.Although not completely understood, male gender was found to be highly associated with high readmission rates.Length of stay in days less than 15 was also found to be highly predictive of readmission [12].
In the United States, researchers have reported that there has been a decrease in number of days spent in hospitals as inpatients.Observations of admission to psychiatric inpatient care dropped between 1990 and 2000 from 833 to 714 persons per 100,000, but this was reversed in 2004 when the observed psychiatric admissions were 910 persons per 100,000 according to data from Substance Abuse and Mental Health Services Administration [6].The dramatic increase in admissions was observed among adolescents and yet this was the same group with most discharges [2,6].The rates of readmission reported in the range from about 10% within 1 month to as high as 86% for a seven-year observation [12,13].The introduction of managed care appeared to fuel a significant increase in psychiatric readmissions.Before managed care, about one-quarter of the youth were reported to experience readmission within one year of discharge, but recent results show readmission ranging from 30 to 50% [14,15].Some studies have found that there has been a trend suggesting a move away from long-term psychiatric hospitalization, but such a move has co-occurred with many more psychiatric readmissions [16][17][18].The readmissions, however, have been documented and studied more among adults and less among adolescents.In one study using data from the youth inpatient unit (YIU) in Christchurch, New Zealand, investigators studied adolescent readmissions; they compared adolescents with a single admission with those with more than one admission in a 12-month period.They found that readmissions were not associated with psychiatric diagnosis, but were associated with child sexual abuse and tended to occur more in the young ages [16].
While the limited research has consistently estimated that about one third of children and adolescents are readmitted within the first 12 months of discharge, [16,17,19,20] the difficult question seems to be the identification of factors predicting such readmissions.In South Wales Sydney, researchers did a chart review study of 112 patients admitted over a one-year period to identify some modifiable factors associated with pediatric psychiatric readmissions.They found that 31% of the patients were readmitted.Males were slightly more likely to be readmitted than females.These same researchers reported a trend of patients with bipolar disorder being more likely to be readmitted and a trend of patients with adjustment disorders being less likely to be readmitted [20].

Methods
A retrospective cohort study was performed where data were collected from 7724 patients between age 3 and 17 who were seen for inpatient care at the University of Utah Neuropsychiatric Institute.These patients were admitted between the years 1999 and 2010 and ethical approval for the study was obtained from the University of Utah Institutional Review Board.We used STATA 12 to analyze data and perform crude and adjusted or multivariate logistic regressions.Participants were retrospectively followed up at 30, 90, 180, and 365 days following discharge.
In the data preparation, 245 participants were eliminated because they had clinical length of stay days (LOS) at the hospital of more than 365 days, which we assumed would put them in a residential treatment.In the early years of the 1990s, part of UNI was a residential setting and patients would stay longer than 365 days.An additional 214 patients who had missing variable information we needed were also eliminated.The exclusion criteria allowed for at least 12 months follow up for all participants after their initial discharge.The final sample included 7,265 participants, all of whom received at least a one-year retrospective observation.
We divided the participants in four age groups as follows: 3-6, 7-11, 12-14, and 15-17 years old.This age categorization follows a similar pattern used in a number of psychological studies.In a children's pain perspectives study of children 4 to 14 years old, the age categorizations were: 4-6, 7-11, and 12-14 years old.This age categorization follows the stages of development described by Jean Piaget while showing that children have differences in the way they behave, understand, and describe pain at different stages of development [21].Only patients who were younger than 18 years of age at the time of their inpatient admission or readmission for psychiatric treatment were included in the study.

Results
Table 1.1 shows a total of 7,265 patients included in the study where 50.70% of them were female.Of the total participants, 1620 (22.30%) of them were readmitted within one year of discharge.The age groups 12-14 and 15-17 had the most contribution to the participants of 27.23% and 54.66%, respectively, and still made up the largest portion of those readmitted with 28.21% and 51.23%, respectively.Although these two age groups had the most number of participants, age group was not found statistically significant in predicting readmissions (Table 1.2).For the bivariate analysis, those patients in the age group 12-14 were 13% more likely to be readmitted compared to age group 3-6, but this was not The patients were predominantly white at 61.53% of the population and with the largest contribution of readmission of 70.86%.The largest percentage of patients (74.08%) had their care paid by Utah Commercial Insurances followed by Utah Medicaid (23.73%).The majority of the readmits 71.11% and 26.98% were also from those patients paid for by these two insurance types, respectively.The results show the biggest percent of patient 82.08% and 17.45% were discharged to home or self- This was a significant result at p=0.001 (Table 1.2).
For the one-year observation after discharge, patients were 1.01 times more likely to be readmitted for every additional day of stay as inpatient when only considering the bivariate analysis.The result is significant at p=0.002 with a 95% confidence interval of (1.004-1.016).When controlled for the other variables involved including diagnosis at admit, still patients were 1.0003 times more likely to be readmitted with each additional day of stay, but the results were not significant with p=0.923.with psychiatric illness.Additional research that investigates how other predictor variables interact with the diagnosis may be useful in efforts to understand readmission among pediatric psychiatric patients.

Limitations and strengths
This study was the first of its kind to utilize the large data set spanning 11 years to investigate readmissions.The use of a number of diagnoses as defined by DSM IV and ICD 9 codes was unique to this study, as well as a strength.Use of a large data set from a regional psychiatric hospital of the nature of UNI is a big advantage to the study; however, at the same time, lends great disadvantages because individuals who go back to their local hospitals may not be counted as readmissions.The study was not designed to allow us to see what post discharge treatment variables or considerations may be available to determine whether they could serve as predictor variables.Dual diagnoses or comorbid illnesses such as diabetes, that may interact with psychiatric illness, may be crucial in the prediction of whether a patient is readmitted or not.Our data did not provide information on dual diagnosis or comorbid illness, so we were not able to assess their influence on readmission.This study utilized diagnoses at the time of admission and with psychiatric illness, a change in diagnosis is possible after being an inpatient.It is likely that the results could be different if diagnoses at discharge were available and considered.

Conclusion
When considering bivariate analysis, some of the suspected predictor variables were significant in predicting readmission, but mostly such prediction disappeared when controlling for other variables.For bivariate analysis, patients with the Utah Medicaid type of insurance were more likely to be readmitted compared to those with commercial insurance and other types of insurance, but such prediction again disappeared with the introduction of control variables.The hospital length of stay was also found insignificant in multivariate analysis.Readmission was highly statistically dependent on type of diagnosis.Those most common in the prediction were schizophrenia disorders, bipolar disorders, and unspecified psychosis.Other facility as a discharge disposition was more likely to predict readmission compared to discharge home, with the greatest influence in the first few months following discharge.referral to psychiatric may prevent rehospitalization and lead to better mental health outcomes.Future efforts should be directed at using these results to reduce these readmission rates, and investigate the effect of early referral.
n H U B f o r S c i e n t i f i c R e s e a r c h Citation: Kagabo R, Kim J, Hashibe M, Kleinschmit K, Clark C, et al. (2016) Predictors of Hospital Readmission for Children with Psychiatric Illness who have Received an Initial Course of Treatment.J Psychiatry Ment Health 1(2): doi http://dx.doi.org/10.16966/2474-7769.109

Table 1 .
with P=0.527.For the multivariate analysis, the results were similar with P=0.733, as shown in table 1.2.For the oldest age group 15-17, in the bivariate analysis, the results showed that this age group was 0.03% less likely to be readmitted compared to the 3-6 age group.These results were still not significant with p=0.986 in the bivariate and p=0.829 in the multivariate analysis, as shown in table 1.2.
2: Predictors of inpatient readmission within One Year of discharge (1999-2010).Note: Bivariate analyses are same as crude analyses and in multivariate analyses all independent variables were included.significant

Table 1 . 3 :
Predictors of inpatient readmission within 30 Days of discharge (1999-2010).Note: Bivariate analyses are same as crude analyses and in multivariate analyses all independent variables were included.careand other facility categories, where still most of the readmission of 79.20% and 20.31% came from respectively, as seen in table 1.1.For the year follow up following discharge and in the bivariate analysis, patients with Utah Medicaid insurance were 1.25 times more likely to be readmitted compared to those with Utah Commercial insurance type.