1) The study examined factors associated with compliance with follow-up appointments after psychiatric hospital discharge when system responsiveness was partially controlled. 2) They found that 33.8% of patients missed their first follow-up appointment. Patients with a primary substance-related diagnosis were most likely to miss appointments, as were uninsured patients. 3) There was a trend for patients not previously involved with outpatient care to miss appointments. Despite efforts to control system responsiveness, compliance with aftercare remained suboptimal, influenced by both client vulnerability and uncontrolled system factors.

1. Psychiatry 67(3) Fail 2004 294
Follow-Up After Inpatient Psychiatric
Hospitalization With Partial Control of the System
Responsiveness Variable
Rif S. El-Mallakh, Tina James, Tehmina Khan, Marina Katz, Bethany McGovern,
Sunil Nair, Scott Tallent, and Gregory Williams
One of the most significant predictors of prompt rehospitalization following psy-
chiatric hospital discharge is missing follow-up out-patient appointments. Previ-
ous studies have suggested that system responsiveness accounted for much of the
variance in predicting compliance with aftercare. Collaborations established at our
institution allowed us to partially control this variable, opening the way to explore
other obstacles to aftercare. All severely mentally ill subjects discharged from our
hospital are provided follow-up appointments within two weeks. We retrospec-
tively evaluated compliance with aftercare appointment and investigated factors
that were associated with compliance. Eighty-one subjects were evaluated.
Twenty-seven (33.8 %) did not attend their first follow-up appointment. Subjects
with a primary substance-related syndrome were the most likely to miss their ap-
pointment (83.3%, X2 = 17.02,p = .0045), as were uninsured patients (51.6%, X2 =
8.79, P = .003). There was a trend for individuals not previously involved with their
aftercare providers to miss their appointment (48.9%, X2 = 3.35, P = .067). Despite
partial control of the system responsiveness variable, compliance with aftercare
was suboptimal. This was due to a combination of client vulnerability variables
and uncontrollable system responsiveness factors.
For the severe and persistently men- ducing the missed appointment rate has been
tally ill, one of the major determinants of min- seen as an important step towards overall im-
imizing psychiatric hospitalizations is provement of prognosis (Byers and Cohen
ongoing outpatient care (Klinkenberg and 1979; Solomon, Davis, and Gordon 1984;
Calsyn 1996; Sparr, Moffitt, and Ward Winston, Pardes, Papernick, and Breslin
1993). This begins with the first appointment 1977).
r
after discharge from the hospital. In general, In their model of predictors of receipt
the rates of compliance with the first of aftercare and recidivism, Klinkenberg and
post-hospital discharge appointment range Calsyn (1996) defined the major contributors
from 13% (Green 1988) to 90% (Sullivan as client vulnerability, community support,
and Bonovitz 1981) and average around 50 and system responsiveness. Of these, system
to 60% (Klinkenberg and Calsyn 1996). Re- responsiveness, including such factors as
Rif S. El-Mallakh, MD, Tina James, MD, Tehmina Khan, MD, Marina Katz, MD, Bethany
McGovern. MD, Sunil Nair, MD, Scott Tallent, MD, and Gregory Williams, DO are affiliated with the
Mood Disorders Research Program, Department of Psychiatry and Behavioral Sciences, University of Lou-
isville School of Medicine, Louisville, Kentucky 40291.
Address correspondence to Dr. Rif S. El-Mallakh at the Mood Disorders Research Program;
E-mail: rselmaOl@louisville.edu

2. El-Mallakh et al. 295
proximal and convenient scheduling of after- their aftercare provider prior to admission
care appointments, were felt to be "more con- (for those patients for whom this was not the
sistent predictors of receipt of aftercare than first hospitalization), whether hospitalization
variables related to either community support was voluntary, and the use of substances in in-
or client vulnerability" (Klinkenberg and dividuals whose primary diagnosis was not
Calsyn 1996). This, in many ways, is good substance-related (e.g., a person with sub-
news, since potentially, we have more control stance-induced mood disorder would be con-
over system responsiveness than other sidered to have a primary substance-related
variables. disorder, while a person with recurrent major
In Louisville, the community mental depression who used marijuana would be
health agency (Seven Counties Services) and considered to have a primary mood disorder
the University Hospital (which serves as the and secondary substance use).
acute care facility for most of the area's se- Since all data was categorical, Chi
verely mentally ill) have collaborated to en- square was utilized for analysis. Variables
sure that the first follow-up appointment after were investigated individually because we
hospital discharge occur within two weeks of wanted to determine which of the three cate-
discharge. This occurred whether the patient gories defined by Klinkenberg and Calsyn
was insured, new to the system, or a non-resi- (1996) accounted for missed appointments.
dent of Kentucky. This setup provided an ex- For example, since community support fac-
cellent opportunity to determine, with system tors might be related to system responsiveness
responsiveness at least partially controlled, variables, a statistical method that evaluated
what additional factors might be associated relationships between the individual variables
with aftercare compliance. (such as multiple regression) would have pre-
vented us from determining the specific
category responsible.
METHOD
RESULTS
The University of Louisville Hospital
operates a 40-bed inpatient facility which
serves as the acute care inpatient facility for Eighty-one subjects were reviewed. The
severely ill individuals within the greater Lou- average age was 34.6 years (range 18 - 67, SD
isville area. Seven Counties Services provides = 11.9 years). There were 53 women (65.4%)
most of the outpatient services to these pa- and 28 men (34.6%). Table 1 summarizes the
tients. Among the several collaborations be- data as a function of whether or not patients
tween the two organizations is the effort to were compliant with their first follow-up ap-
ensure that all severely ill patients be given a pointment. Individuals admitted with a pri-
follow-up appointment within two weeks of mary substance-related disorder (specifically,
discharge. Seven Counties has several sites lo- substance-induced mood or psychotic disor-
cated throughout the city, and appointments ders) were least likely to make it to their fol-
were made in the most convenient site. .Addi- low-up appointment (16.7%, p = .0045).
tionally, case managers were available to take However, substance abuse, per se, did not
patients to their appointments if needed. predict noncompliance.
We reviewed the records of 81 patients Missed first appointments were higher
discharged in the month of February 2001 in subjects with a primary mental illness who
and collected data regarding age, gender, pri- also used substances (41.3 %) than those who
mary diagnosis for which they were admitted did not use substances (26.5%), but this dif-
(many had other secondary psychiatric diag- ference was not significant in the current sam-
noses), type of insurance, homelessness status, ple (p = .3). Similarly, there was no difference
whether this was the patient's first hospital- if follow-up was scheduled in a primary psy-
ization, whether the patient was involved with chiatric setting (30.9%) or a substance treat-

3. 296 Follow-Up after Inpatient Psychiatric Hospitalization
Table 1. Variables Examined in Subjects Who Attended or Missed Their First Appointment Following Discharge
from Psychiatric Hospitalization. Data are presented as numbers (and percentages) of patients in each category.
Analysis utilized Chi square.
Variable Attended Follow-Up Missed Follow-Up X2 r.
Diagnosis 17.02 .0045
Bipolar 14 (66.7%) 7 (33.3%)
Schizophrenia 21 (75.0%) 7 (25.0%)
Depression 11 (84.6%) 2 (15.4%)
Substance-induced 2 (16.7%) 10 (83.3%)
Insurance 8.79 .032
Medicaid 23 (85.2%) 4 (14.8%)
Medicare 13 (68.4%) 6 (31.6%)
Private Insurance 2 (66.7%) 1 (33.3%)
None 15 (48.4%) 16 (51.6%)
Previous Involvement with
Outpatient Care 3.35 .067
Yes 33 (73.3%) 12 (26.7%)
No 17(53.1%) 15 (46.9%)
First Hospitalization 1.58 .21
Yes 12 (54.6%) 10 (45.6%)
No 41 (69.5%) 18 (30.5%)
Presence Of Substance Abuse 2.44 .30
Yes 27 (58.7%) 19 (41.3%)
No 25 (73.5%) 9 (26.5%)
Voluntary Status 1.08 .30
Involuntary 15 (75.0%) 5 (25.0%)
Voluntary 38 (62.3%) 23 (37.7%)
Homelessness 0.88 .35
Homeless 4 (50.0%) 4 (50.0%)
Not homeless 48 (66.7%) 24 (33.3%)
Gender 0.002 0.96
Male 23 (65.7%) 12 (34.3%)
Female 30 (65.2%) 16 (34.8%)
ment center (50.0%, X2 == 2.62, P == .27). hospitalizations (69.5%, P == .21).
Patients without any insurance had the lowest Surprisingly, involuntary hospitalization,
rate of compliance with aftercare when this homelessness, and gender were not associated
particular variable was examined (48.4%) with aftercare noncompliance (see Table 1).
while those with Medicaid had the highest
rate (85.2%, p == .032).
DISCUSSION
There was a trend for those compliant
with preadmission outpatient care to predict
An interplay between client vulnerabil-
compliance with post-discharge aftercare ity, community support, and system respon-
(73.3% versus 53.1% for those without siveness determines the likelihood of
preadmission involvement, p == .067). This compliance with post-discharge aftercare
contrasts with the lack of significance in show (Klinkenberg and Calsyn 1996). Klinkenberg
rate between those experiencing a first hospi- and Calsyn (1996) defined client vulnerability
talization (54.6%) and those with previous to include easily measured items such as diag-

4. EI-Mallakh et al. 297
nosis, demographics, and socioeconomic sta- hospitalization did not have a significantly
tus, as well as more abstract items such as different compliance rate from those with
interpersonal skills and insight. previous hospitalizations.
Community support constitutes the liv- The only system responsiveness factor
ing situation and relationship to family mem- which could not be controlled, lack of insur-
bers and other social support. System ance, showed a nonspecific trend towards
responsiveness encompasses items that are ex- contributing to noncompliance. This may
clusively under the control of mental health seem self-explanatory, but subjects without
providers. These include issues such as conve- insurance are usually offered services at re-
nient and proximal appointments and case duced cost. Availability of insurance has not
management (defined as services to assist cli- been routinely examined in previous studies.
ents with navigating complicated procedures, Correlates of insurance, such as educational
such as applying for public housing assis- level or employment status, are generally not
tance, or simple tasks, such as shopping for correlated with aftercare compliance
food) (Klinkenberg and Calsyn 1996). (Klinkenberg and Calsyn 1996), and in-
In their review of the literature, creased state funding of outpatient services
Klinkenberg and Calsyn (1996) concluded did not reduce recidivism (Fisher, Geller,
that system responsiveness was the most sig- Altaffer, and Bennett 1992).
nificant set of variables in reducing recidivism.
Additionally, one may also argue that
To understand the relative role of other fac-
insurance availability is not under the exclu-
tors, we examined the rate of follow-up after
sive control of mental health providers, and is
discharge in a group of people in whom sys-
therefore misclassified as a system responsive-
tem responsiveness variables were partially
ness variable; rather, it is most closely related
controlled.
to socioeconomic status, and is better classi-
Our data suggest that client vulnerabil-
fied as a client vulnerability factor.
ity and community support factors appear to
playa major role in noncompliance with ini- There are several shortcomings in our
tial aftercare visits. Specifically, if the admis- study. For example, we were unable to exam-
sion was due to a primary substance-related ine several important factors in our retrospec-
syndrome, patients were unlikely to keep their tive design. Specifically, we examined only
appointments. However, substance use, per se one community support factor, homelessness.
(specifically, when it is not related to the pri- Our results are compatible with previous re-
mary reason for admission), was not predic- ports that found no relationship between
tive of poor compliance. This is consistent compliance with aftercare and homelessness.
with previous studies that those with isolated However, a homelessness outreach program
substance use disorders have a lower compli- conducted by the community mental health
ance rate (Allan 1987; Bander, Stilwell, Fein, agency may have also improved compliance
and Bishop 1983), but that psychiatric diag- among the homeless patients. The effect of
nosis was not a major predictor of noncompli- this program could not be examined in our
ance (Byers and Cohen 1979; Klinkenberg retrospective design.
and Calsyn 1996; Solomon, Davis, and Despite these shortcomings, our data
Gordon 1984). does suggest that when the system responsive-
Involvement with the outpatient care ness variable is partially controlled by ensur-
clinic prior to hospitalization was a positive ing that all appointments are made within two
predictor to keeping an appointment after weeks of hospital discharge, that compliance
hospital discharge. This appears to be specifi- remains suboptimal at 72.2%. Our design
cally related to the established therapeutic re- could not explore community support factors
lationship, since subjects with first effectively.

5. 298 Follow-Up after Inpatient Psychiatric Hospitalization
REFERENCES
Allan, C. (1987). Seeking help for drinking prob- Klinkenberg, W.D., and Calsyn, R.J. (1996).
lems from a community-based voluntary Predictors of receipt of aftercare and recidivism
agency: Patterns of compliance amongst men among persons with severe mental illness: A re-
and women. British Journal of Addiction, view. Psychiatric Services, 47, 487-496.
82,1143-1147.
Solomon, P., Davis, J., and Gordon, B. (1984).
Bander, K.W., Stilwell, N.A., Fein, E., and Discharged state hospital patients' characteris-
Bishop, G. (1983). Relationship of patient char- tics and use of aftercare: Effect on community
acteristics to program attendance by women al- tenure. American Journal of Psychiatry, 141,
coholics. Journal of Studies on Alcohol, 44, 1566-1570.
318-327.
Sparr, L.F., Moffitt, M.C., and Ward, M.F.
Byers, E.S., and Cohen, S.E. (1979). Predicting (1993). Missed psychiatric appointments: Who
patient outcome: The contribution of returns and who stays away. American Journal
prehospital, in hospital, and posthospital fac- of Psychiatry 150,801-805.
tors. Hospital and Community Psychiatry, 30,
327-331. Sullivan, K., and Bonovitz, J.S. (1981). Using
predischarge appointments to improve continu-
Fisher, W.H., Geller, J.L., Altaffer, F., and ity of care for high-risk patients. Hospital and
Bennett, M.B. (1992). The relationship between Community Psychiatry, 32, 638-639.
community resources and state hospital recidi-
Winston, A., Pardes, H., Papernick, D.S., and
vism. American Journal of Psychiatry, 149,
Breslin, L. (1977). Aftercare of psychiatric pa-
385-390.
tients and its relation to rehospitalization. Hos-
Green, J.H. (1988). Frequent rehospitalization pital and Community Psychiatry, 28, 118-121.
and noncompliance with treatment. Hospital
and Community Psychiatry, 39, 963-966.