Linking Inpatients

With Schizophrenia
to Outpatient Care
Mark Olfson, M.D., M.P.H.
David Mechanic, Ph.D.
Carol A. Boyer, Ph.D.
Stephen Hansell, Ph.D.

P
Objective: This study focused on inpatients with schizophrenia or atients with schizophrenia com-
schizoaffective disorder who were scheduled to begin outpatient care monly fail to continue in treat-
with clinicians who had not previously treated them. The authors eval- ment after hospital discharge.
uated the effects of communication between the patients and their As many as one-third to one-half of
outpatient clinicians before discharge on patients’ referral compli- hospitalized patients with schizophre-
ance, psychiatric symptoms, and community function at follow-up nia and related disorders miss their
three months after discharge. Methods: A total of 104 adult inpatients first scheduled outpatient appoint-
with schizophrenia or schizoaffective disorder who were scheduled to ment after hospital discharge (1–3). A
receive outpatient care from clinicians who had not previously treated failure to follow up with outpatient
them were evaluated at hospital discharge and again three months lat- care after leaving the hospital greatly
er. Comparisons were made between patients who had telephone or increases the risk of relapse and re-
face-to-face contact with an outpatient clinician before hospital dis- hospitalization (4,5).
charge and patients who did not have such contact. Results: About half A number of discharge planning
(51 percent) of the inpatient sample communicated with an outpatient strategies have been developed to
clinician before leaving the hospital. Compared with patients who had help smooth the transition from inpa-
no communication, those who spoke with an outpatient clinician were tient to outpatient care. Examples in-
significantly more likely to complete the outpatient referral. After clude scheduling appointments be-
baseline scores and other covariates were controlled for, predischarge fore hospital discharge (5,6), mini-
contact with an outpatient clinician was associated with a significantly mizing the period of time between
lower total Brief Psychiatric Rating Scale score at follow-up and less discharge and the first scheduled out-
self-assessed difficulty controlling symptoms. Nonsignificant trends to- patient appointment (7,8), providing
ward improved medication compliance and a lower rate of homeless- telephone reminders and transporta-
ness were also found. The two patient groups did not significantly dif- tion to outpatient appointments (9,
fer in the proportion who were readmitted to the hospital or who 10), and, whenever possible, referring
made a psychiatric emergency room visit during the follow-up period. patients back to the same clinician
Conclusions: Direct communication between inpatients and new out- who treated them before the hospital-
patient clinicians may help smooth the transition to outpatient care ization (11).
and thereby contribute to improved control of clinical symptoms. (Psy- In some cases, it is not possible to
chiatric Services 49:911–917, 1998) refer inpatients back to the outpatient
clinicians who have previously treated
them. Patients move, refuse to return
to their previous outpatient clinicians,
Dr. Olfson is associate professor in the department of psychiatry at Columbia Univer-
sity College of Physicians and Surgeons at New York State Psychiatric Institute, 722
or are referred to a level of care, such
West 168th Street, New York, New York 10032 (e-mail, [email protected] as day hospital treatment, that is dif-
bia.edu). He is also research associate at the Institute for Health, Health Care Policy, ferent from what they received be-
and Aging Research at Rutgers University in New Brunswick, New Jersey, where Dr. fore the hospital admission. Under
Mechanic is director and Dr. Boyer is associate director. Dr. Hansell is associate pro- such circumstances, the risks of refer-
fessor of sociology at Rutgers University. ral noncompliance are believed to be
particularly great. One strategy for fa-
cilitating the referral is to arrange for
the inpatient to meet the new outpa-
PSYCHIATRIC SERVICES ♦ July 1998 Vol. 49 No. 7 911
tient clinician before the patient is care in general hospitals and patient 120 days, or withdrew their consent.
discharged from the hospital (11). outcomes for Medicaid patients with The baseline inpatient assessment
This meeting may help prepare the schizophrenia and related disorders. was administered to 323 patients and
patient for the transition by reducing was completed by 316 patients.
the patient’s apprehension about the Subjects The sample of 316 screened and se-
first scheduled outpatient appoint- Eligible subjects were English-speak- lected patients and the 694 screened
ment. ing, newly admitted psychiatric inpa- but nonselected patients did not sig-
Social anxiety is common in schizo- tients, between 18 and 64 years old, nificantly differ in age, gender, eth-
phrenia (12). Fear of ordinary social who were Medicaid enrolled or eligi- nicity, marital status, or recent work
situations may contribute to the com- ble and had an admitting clinical di- history. In addition, a similar propor-
mon tendency among people with agnosis of schizophrenia or schizoaf- tion of the selected and nonselected
schizophrenia to social isolation (13). fective disorder. Subjects were subse- samples reported active drug or alco-
Longitudinal research has revealed quently entered in the study if they hol use before admission (42 percent
that adults with schizophrenia com- provided written informed consent versus 38 percent for drug use and 38
monly fail to establish close relation- and met criteria for schizophrenia or percent versus 39 percent for alcohol
ships, seldom initiate new social con- schizoaffective disorder according to use). However, blacks were overrep-
nections, and tend to withdraw from the Structured Clinical Interview for resented in the selected sample (58 per-
existing relationships (14,15). The DSM-III-R (SCID) (17), updated to cent versus 42 percent in the nonse-
general tendency of patients with include DSM-IV criteria, as adminis- lected sample), and whites and Asians
schizophrenia to retreat from new so- tered by a trained research assistant. were underrepresented (whites, 40
cial interactions strengthens the clini- Patients with a severe and highly dis- percent versus 49 percent in the non-
cal rationale for introducing inpa- abling general medical condition selected sample; Asians, 2 percent
tients with schizophrenia to their out- were ineligible for the study. Subjects versus 9 percent in the nonselected
patient clinicians before hospital dis- who had stays longer than 120 days, sample) (χ2=34.7, df=2, p<.001). Pa-
charge. who were discharged against medical tients in the selected sample were
In the study reported here, we ex- advice, or who were transferred to an- also significantly more likely than
amined the effects of predischarge other inpatient psychiatric facility those in the nonselected group to re-
communication between outpatient were also excluded from the study. port at least one previous psychiatric
clinicians and inpatients with schizo- Subjects were recruited in several hospitalization (93 percent versus 86
phrenia or schizoaffective disorder on phases. A total of 1,328 prescreened percent; χ2=10.9, df=1, p<.01).
patients’ referral compliance, psychi- patients consecutively admitted to Of the 316 patients who entered
atric symptoms, and functional out- four general hospitals in New York the study, 117 were scheduled to be-
comes. We compared the three- City during the period from October gin outpatient care with a clinician
month posthospital adjustment of pa- 1994 to April 1996 met the age, payer who had not previously treated them.
tients who received telephone or status, and diagnosis eligibility crite- We located 104 of those patients
face-to-face contact with an outpa- ria. Based on medical records and dis- (88.9 percent) for a three-month fol-
tient clinician before discharge with cussions with inpatient staff, we elim- low-up assessment. The group who
that of a similar patient group who inated 4 percent of the screened sam- was lost to follow-up did not signifi-
did not receive such contact. We hy- ple due to severe general medical cantly differ from the completer
pothesized that as a consequence of conditions, 4 percent who lived out- group in age, gender, race, or score on
predischarge contact with an outpa- side of New York City, and 9 percent the Brief Psychiatric Rating Scale
tient clinician, patients would be who did not speak or understand (BPRS) (18) or Global Assessment
more likely to complete referrals for English. Scale (GAS) (19) at the baseline in-
outpatient care. We further predicted A total 1,010 screened patients (76 terview.
that as a consequence of improved percent) were therefore assessed as
linkage to outpatient care, predis- eligible to receive the diagnostic in- Assessments
charge communication with an out- terview. Of this group, 57 percent Within 72 hours before hospital dis-
patient clinician would be associated (N=576) agreed to be interviewed, 31 charge, patients completed a struc-
with improved symptom control and percent (N=310) refused, and 12 per- tured assessment intended to collect
improved community functioning. cent (N=124) were not approached. data on clinical symptoms, social
Of the 576 patients who consented to functioning, medication compliance,
Methods the diagnostic interview, 68 percent mental health service utilization, and
Data were drawn from the longitudi- (N=394) met DSM-IV criteria for other outcome domains. Clinical
nal patient outcome phase of the Rut- schizophrenia or schizoaffective dis- symptoms were assessed by trained
gers hospital and community survey, order. Of the patients who met the di- research assistants using the BPRS,
which has been described in detail agnostic criteria, 71 did not complete GAS, and the Center for Epidemio-
elsewhere (16). Briefly, a primary aim the baseline assessment because they logical Studies—Depression Scale
of the Rutgers study, conducted be- left the hospital against medical ad- (CES-D) (20). Life satisfaction was
tween 1991 and 1996, was to examine vice, were transferred to another in- assessed with items from the Quality
the relationship between psychiatric patient facility, had a stay longer than of Life Interview (21). In addition,
912 PSYCHIATRIC SERVICES ♦ July 1998 Vol. 49 No. 7
several self-report items were modi- Table 1
fied from the National Health Inter- Demographic and clinical characteristics of inpatients with schizophrenia who did
view Mental Health Supplement to and did not have predischarge contact with outpatient clinicians, in percentages1
assess difficulties with psychosocial
functioning (22). Predischarge No predis-
Three months after hospital dis- contact charge con-
Characteristic (N=53)2 tact (N=51)2 χ2 df p
charge, patients were reinterviewed
by trained research assistants using Gender .4 1 .51
the same clinical instruments. At the Male 58.5 64.7
time of discharge, each patient’s pri- Race 1.2 2 .55
White 51.9 41.2
mary inpatient clinician was asked to Black 46.2 56.8
complete an Inpatient Treatment Asian 1.9 2.0
Survey, which was developed to Marital status 5.1 3 .17
probe various aspects of inpatient ser- Never married 80.4 61.2
Married3 5.9 8.2
vice delivery. Clinicians were asked Separated or divorced 13.7 28.6
whether the patient was scheduled to Widowed 0 2.0
begin treatment with a new outpa- Education (years) 2.2 3 .52
tient clinician after hospital dis- Less than 12 33.9 38.8
12 43.4 30.6
charge. Other items ascertained 13 to 15 18.9 22.4
whether an outpatient clinician came 16 or more 3.8 8.2
to the unit to evaluate the patient, DSM-III-R substance use
whether an outpatient clinician talked disorder
Alcohol 20.8 19.6 0.2 1 .76
to the patient by telephone, or Drug 26.4 29.4 0.1 1 .64
whether the patient went for an inter-
1 Analysis included only inpatients with referrals to outpatient clinicians who had not previously
view or started the outpatient pro-
treated the patients.
gram while still in the hospital. In the 2 Ns vary for patients with predischarge contact from 51 to 53 and for patients with no predischarge
analyses described below, patients contact from 49 to 51.
who were scheduled to see a new 3 Includes married and living together as though married.
clinician were stratified by whether
they received at least one of these
three services, which we collectively
refer to as contact with an outpatient outpatient care. The two groups’ ex- Results
clinician before hospital discharge. perience during the three-month fol- General characteristics
low-up period was then compared, Approximately half of the inpatients
Analytic strategy and including the proportions of each who were referred to new outpatient
statistical methods group who completed scheduled out- clinicians (53 patients, or 51 percent)
Our primary goal was to study the ef- patient referrals, made a psychiatric communicated with them before
fect of predischarge outpatient clini- emergency room visit, were hospital- hospital discharge. Most of these pa-
cian contact on the short-term course ized, became homeless, or reported a tients (43, or 81.1 percent) went for
of patients with schizophrenia. The period of one week or more of com- an interview at an outpatient pro-
two study groups were first compared plete cessation of their antipsychotic gram while they were still in the hos-
on sociodemographic characteristics, medications. Separate analyses were pital. The remainder either spoke
hospitalization history, length of index made of the proportion of patients in with an outpatient clinician on the
inpatient stay, and presence of DSM- each group who reported increased telephone (five patients, or 9.4 per-
III-R drug or alcohol use disorders as difficulties from the baseline to the cent), met an outpatient clinician on
determined by the Mini International follow-up assessment in four areas of the inpatient unit (three patients, or
Neuropsychiatric Interview (23). Stu- psychosocial functioning and symp- 5.7 percent), or both (two patients, or
dent’s t test was used for comparisons tom scores at three months after dis- 3.8 percent).
involving continuous variables, and charge. Multiple linear regression The two study groups did not differ
the chi square test was used for com- equations were used to examine asso- in their racial or gender distribution.
parisons involving categorical vari- ciations between predischarge outpa- The mean±SD age of the groups was
ables. Comparisons were considered tient clinician contact and symptom 33.4±8.7 years for patients with pre-
statistically significant at the 5 per- scores at follow-up, controlling for discharge contact with outpatient
cent level (alpha=.05, two-tailed). age, gender, race, hospital, legal sta- clinicians and 34.4±9.9 years for pa-
We also compared the patients with tus of admission, and baseline symp- tients with no contact, not a signifi-
and without outpatient contact before tom scores. Selected comparisons in- cant difference. A majority of patients
discharge on aspects of inpatient volving patients who were scheduled in each group had never married and
treatment, including legal status, sat- to return to the outpatient clinicians had a relatively low level of education
isfaction with treatment, and per- who had previously treated them are (see Table 1). Substantial and similar
ceived likelihood of continuing in also presented. proportions of each group met crite-
PSYCHIATRIC SERVICES ♦ July 1998 Vol. 49 No. 7 913
Table 2 care. Remarkably, only one patient
who met or spoke with an outpatient
Treatment characteristics of inpatients with schizophrenia who did and did not
have predischarge contact with outpatient clinicians, in percentages1 clinician before leaving the hospital
failed to complete the scheduled out-
Predischarge No predis- patient referral.
contact charge con- The two study groups did not sig-
Characteristic (N=53)2 tact (N=51)2 χ2 df p
nificantly differ in the proportion who
Previous psychiatric made a psychiatric emergency room
hospitalizations 2.0 3 .58 visit, required readmission to a psy-
None 13.7 12.2 chiatric hospital, or reported a period
One 13.7 6.1
Two 15.7 14.3 of medication noncompliance of one
Three or more 56.9 67.3 week or more during the three-
Involuntary admission 52.8 70.6 3.5 1 .06 month follow-up period (see Table 3).
Refused outpatient referral 20.4 25.3 1.1 1 .28 However, there was a nonsignificant
Satisfied with inpatient treatment 84.6 84.0 0.0 1 .91
Inpatient staff available to help trend toward an increased rate of self-
after discharge 73.6 65.3 0.8 1 .36 reported medication noncompliance
Patient is very certain he or she among the patients who did not have
will complete referral 82.4 83.7 0.0 1 .86 predischarge contact with an outpa-
Staff predicts patient will definitely
complete referral 42.3 34.0 0.7 1 .39 tient clinician.
Received treatment during the
three months before admission 56.6 52.9 0.1 1 .71 Psychiatric symptoms
Received antipsychotic medication At follow-up, the group who had pre-
during the three months before
admission 50.9 51.0 0.0 1 1.00 discharge contact with an outpatient
clinician had a significantly lower
1 Analysis included only inpatients with referrals to outpatient clinicians who had not previously mean total BPRS score than the
treated the patients.
2 Ns vary for patients with predischarge contact from 51 to 53 and for patients with no predischarge
group with no contact. After the fol-
contact from 49 to 51. low-up BPRS score was adjusted for
age, gender, race, hospital, admitting
legal status, and baseline BPRS score,
outpatient clinician contact was asso-
ria for a DSM-III-R alcohol or drug pital, there was a nonsignificant trend ciated with a 5.5 point decrease in to-
use disorder at the time of hospital toward involuntary admission among tal BPRS score (see Table 4). An an-
discharge. patients who did not have contact alysis using BPRS subscale scores
with an outpatient clinician before (24) revealed that outpatient clinician
Inpatient treatment hospital discharge (see Table 2). The contact was associated with a signifi-
The proportion of inpatients who had two groups did not significantly differ cant decrease in the hostility-suspi-
contact with an outpatient clinician in mean length of inpatient stay cion and anxiety-depression subscale
before discharge varied across the (40±24.9 days for patients with pre- scores. In a separate model, comple-
four participating hospitals (χ2=20.8, discharge contact with their outpa- tion of the outpatient referral was as-
df=3, p<.001). At three of the hospi- tient clinicians and 34.5±22 days for sociated with a lower total BPRS
tals, a majority of the patients (rang- patients without such contact). score at follow-up after baseline
ing from 60.2 percent to 65.2 per- The vast majority of the inpatients BPRS score, age, gender, and race
cent) had predischarge outpatient with and without predischarge outpa- were controlled for (beta=–6, p=.04).
clinician contact. However, at one tient clinician contact were very cer- Because only one patient met with an
hospital only two of 22 patients (9.1 tain that they would keep their sched- outpatient clinician before hospital
percent) had such contact. uled outpatient appointments (see discharge but did not receive outpa-
Most of the patients in both study Table 2). In this regard, inpatient staff tient care, we could not explore the
groups had three or more previous were considerably less optimistic. interaction between inpatient contact
psychiatric hospitalizations (see Table The staff predicted that only a minor- with outpatient clinicians, referral
2). In addition, a majority in each ity of the patients in each group compliance, and symptom control.
group viewed their inpatient care as would definitely attend the arranged Inpatients with no predischarge
satisfactory and believed that if they outpatient mental health programs. contact reported more psychosocial
needed help after hospital discharge, difficulties in several areas. In the
there was a staff member on the unit Service utilization outcomes area of symptom control, this differ-
whom they could contact. A similar As Table 3 shows, patients who com- ence reached the level of statistical
proportion of both groups refused municated with an outpatient clini- significance (see Table 3). The two
one or more outpatient referrals dur- cian before discharge were signifi- study groups did not significantly dif-
ing the course of their inpatient stay. cantly more likely than those who did fer in mean scores on the GAS or
Although a majority in each group not communicate to complete their CES-D or global QLI scores at fol-
were admitted involuntary to the hos- scheduled referral for outpatient low-up (see Table 4).
914 PSYCHIATRIC SERVICES ♦ July 1998 Vol. 49 No. 7
Other outcomes Table 3
As Table 3 shows, there was a non- Three-month postdischarge outcomes for service use and impairment among pa-
significant trend toward reduced tients who did and did not have contact with outpatient clinicians while in the hos-
homelessness among patients who pital, in percentages1
communicated with an outpatient
clinician before hospital discharge. Predischarge No predis-
This trend became statistically signif- contact charge con-
Outcome (N=53)2 tact (N=51)2 χ2† p
icant when the sample was expanded
to include patients who were sched- Psychiatric outpatient visit 98.1 62.7 20.9 .001
uled to return to clinicians who had Psychiatric emergency room visit 18.9 26.0 .8 .38
previously treated them (5.4 percent Psychiatric hospitalization 17.3 25.5 1.0 .31
Medication noncompliance3 20.8 34.7 2.5 .11
versus 15 percent, χ2=6.4, df=1, p= Homeless 3.8 13.7 3.3 .07
.01). At the follow-up interview, a Self-reported psychosocial
similar proportion of patients in each difficulties4
group reported having worked for pay Symptom control 12.0 30.6 5.1 .02
Control of anger 13.5 18.0 .4 .58
or having sought paid employment. Concentration 22.0 28.6 .6 .45
Recognition of symptoms 12.2 17.0 .4 .51
Inpatients returning Occupational functioning
to outpatient clinicians Had paid employment 13.2 14.3 .2 .84
Sought paid employment 26.1 25.6 0 .96
The study also provided follow-up data
on 135 patients scheduled to return to 1 Analysis included only inpatients with referrals to outpatient clinicians who had not previously
outpatient clinicians who had previ- treated the patients.
2 Ns vary for patients with predischarge contact from 49 to 53 and for patients without predischarge
ously treated them. Within this patient contact from 47 to 51, except for data on whether patients sought paid employment; this question
group as well, predischarge contact was asked only of unemployed patients (46 with predischarge contact and 42 without predischarge
with an outpatient clinician was associ- contact).
3 Analysis was limited to patients prescribed oral antipsychotic medications.
ated with successful completion of the 4 Patient reports of greater difficulty in these areas at follow-up than at baseline
outpatient referral. Of the 68 patients † For all comparisons, df=1
who had predischarge contact with an
outpatient clinician, 62, or 91.2 per-
cent, completed their referral, com-
pared with 49 of the 67 patients, or more likely to have had a previous in- sion (80.7 percent versus 51 percent;
73.1 percent, who did not have such patient psychiatric admission (96.3 χ2=23.9, df=1, p<.001). Patients who
contact (χ2=7.5, df=1, p=.006). percent versus 87.5 percent; χ2=6.5, had no previous psychiatric admis-
Compared with the 104 patients df=1, p=.01) and to have taken anti- sions improved from baseline to fol-
starting with new clinicians, the 135 psychotic medications during the low-up in mean BPRS score signifi-
returning patients were significantly three months before the index admis- cantly more than those who had

Table 4
Mean scores on measures of symptoms three months after discharge among patients who did and did not have contact with
outpatient clinicians while in the hospital
Predischarge No predischarge Estimated
contact (N=53) contact (N=51) difference
between
Measure Mean SD Mean SD t df p groups1 p

Brief Psychiatric Rating Scale2

Total 34.5 10.3 41.0 11.5 3.0 96 .004 –5.5 .02
Subscale
Hostility-suspicion 5.1 2.9 7.1 3.2 3.2 97 .002 –1.6 .01
Anxiety-depression 9.9 4.7 12.1 5.6 2.1 99 .03 –2.6 .01
Psychosis 7.8 4.1 9.0 4.3 1.4 98 .15 –.4 .66
Activation 4.5 1.7 5.0 2.4 1.3 100 .20 –.2 .62
Anergia 7.8 3.9 8.1 3.1 .5 100 .62 –.1 .85
Global Assessment Scale3 56.6 20.0 52.0 19.4 1.2 100 .24 1.6 .65
Center for Epidemiological
Studies—Depression Scale2 25.3 8.6 26.6 10.0 .7 87 .51 –.4 .83
Global Quality of Life4 4.8 1.7 4.5 1.7 1.0 100 .33 .4 .21
1 Change scores were controlled for age, gender, race, hospital where patient was treated, admitting legal status, and baseline score.
2 Higher scores indicate more extensive symptoms.
3 Higher scores indicate better global functioning
4 Higher scores indicate greater satisfaction with living circumstances.

PSYCHIATRIC SERVICES ♦ July 1998 Vol. 49 No. 7 915

previous admissions (t=2.1, df=236, turning to clinicians who have previ- ate outpatient services and ex-
p=.04). Similarly, patients who were ously treated them. Because patients pressed little interest in the patient’s
not taking antipsychotic medications referred to new clinicians are less course of recovery after hospital dis-
before admission improved more than likely to have been on medication charge.
patients who were on medications be- than continuing patients, becoming Surveys of inpatient staff have indi-
fore admission (t=1.9, df=235, p=.05). involved in active treatment may be cated that psychiatric inpatient units
In a correlation that controlled for especially helpful in promoting symp- vary considerably in how they ap-
baseline BPRS score, discharge to a tom improvement. proach the inpatient treatment of
new clinician was correlated with a It is possible that inpatients who schizophrenia, with some units plac-
significant improvement in BPRS have greater interest in or motivation ing a greater emphasis than others on
score (beta=–4.1, p=.007). However, for participating in their own recovery practices likely to improve continuity
this relationship was attenuated when are particularly amenable to meeting of care (27). The findings reported
the correlation was first controlled for their new outpatient therapist. In oth- here support strategies for strength-
whether the patient was taking med- er words, inpatient communication ening treatment in this area by ex-
ication before the index admission with a new outpatient clinician may panding the clinical responsibilities of
(beta=–3.8, p=.02) and then by serve as a proxy for treatment motiva- inpatient staff and outpatient clini-
whether the patient had been previ- tion, and it may be this motivation cians to include introducing selected
ously hospitalized (beta=–3.4, p=.04). rather than the meeting per se that severely ill inpatients to their outpa-
explains the improved outcomes. tient clinician before hospital dis-
Discussion and conclusions However, we found no evidence that charge. Such an added responsibility
We found that inpatients with schizo- the inpatients who communicated runs counter to the trend to shorten-
phrenia who communicated with with an outpatient clinician were ing the length of inpatient stays (28)
their new outpatient clinicians before more favorably predisposed to treat- and narrowing the scope of inpatient
they left the hospital were substan- ment than the inpatients who did not. treatment (29) and may therefore be
tially more likely to complete their re- For example, a similar proportion of difficult to implement in some inpa-
ferrals for outpatient care compared each group refused an outpatient re- tient settings.
with inpatients who did not have such ferral or expressed dissatisfaction Inpatients in the study reported
contact. The inpatients who met or with their inpatient treatment. The here had relatively long inpatient
spoke with a new outpatient clinician association between predischarge stays. At many units, financial consid-
were also less symptomatic at follow- outpatient clinician contact and im- erations have driven the average
up, and fewer of them reported diffi- proved symptoms also remained sig- lengths of inpatient stay for patients
culties controlling their symptoms, nificant after the analysis controlled with schizophrenia below two weeks
presumably because they were more for legal status at admission, which is (30). Shortening the length of hospi-
likely to continue in outpatient care. presumably related to treatment mo- talization reduces the time available
These findings are consistent with tivation. However, because the study to stabilize patients and arrange for
earlier research on the value of coor- was observational rather than experi- them to meet with an outpatient clin-
dinating the referral process (2,5, mental in design, we cannot rigorous- ician. Some payers explicitly restrict
11,25) and underscore the impor- ly exclude the possibility that differ- payment for outpatient services for
tance of the interpersonal dimension ences in motivation or other unmea- inpatients as a form of double billing.
of the referral process. sured patient characteristics or treat- Such restrictive reimbursement poli-
Introducing severely mentally ill ments account for the observed out- cies further reduce the clinical oppor-
inpatients to new outpatient clini- comes. tunity to familiarize inpatients with
cians while they have the support of Marked differences in the rate of outpatient care and may thereby im-
inpatient staff may ease patients’ outpatient clinician contact between pede continuity of care.
anxieties over meeting the clinician the participating hospitals suggest We did not find a significant effect
for the first time. If the first contact that hospital treatment routines may of inpatient contact with outpatient
between the outpatient clinician and play an important role in determin- clinicians on improved occupational
the inpatient is successful, it may ing whether inpatients meet their functioning, quality of life, rehospital-
form a basis for developing a work- new outpatient clinician before hos- ization, or emergency room visits.
ing alliance. Patients with schizo- pital discharge. At the hospital that However, we did find a nonsignificant
phrenia who form strong alliances only rarely established predischarge trend toward reduced homelessness
with their outpatient clinicians dur- contact between inpatients and their and a weak trend toward improved
ing the first few months of treatment new outpatient clinicians, we found medication compliance. The relative
have been found to achieve superior little evidence during a site visit that independence of clinical symptoms
outcomes (26). the inpatient staff were actively in- from level of community functioning
Inpatients who start with new out- volved in promoting continuity of (31) may help explain our failure to
patient clinicians are likely to be ear- care. Inpatient staff reported that find significant differences in func-
lier in their illness trajectories and they seldom made special efforts to tional outcomes despite improve-
less compliant with treatment pro- ensure that more difficult-to-place ments in clinical symptoms. It is also
cesses than are patients who are re- patients were connected to appropri- possible that a larger sample, fol-
916 PSYCHIATRIC SERVICES ♦ July 1998 Vol. 49 No. 7
lowed with more sensitive measures use of aftercare: effect on community measuring overall severity of psychiatric
tenure. American Journal of Psychiatry disturbance. Archives of General Psychia-
over a longer time period, might have 141:1566–1570, 1984 try 33:766–771, 1976
yielded different results.
4. Green JH: Frequent rehospitalization and 20. Radloff LS: The CES-D scale: a self-report
Several other caveats should be depression scale for research in the general
noncompliance with treatment. Hospital
noted. First, no information was and Community Psychiatry 39:963–966, population. Applied Psychological Mea-
available on the posthospital adjust- 1988 surement 1:385–401, 1977
ment of the 13 patients referred to 5. Stickney SK, Hall RWC, Gardner ER: The 21. Lehman AL: A Quality of Life Interview for
new clinicians who were lost to fol- effect of referral procedures on aftercare the Chronically Mentally Ill—Shortened
low-up. Although this group resem- compliance. Hospital and Community Psy- Version. Baltimore, University of Mary-
chiatry 31:567–569, 1980 land, Department of Psychiatry and Behav-
bled the completer sample in base- ioral Sciences, 1992
line psychopathology, we have no 6. Sullivan K, Bonovitz JS: Using predischarge
appointments to improve continuity of care 22. Mental Health, United States, 1992. Edited
means of assessing the effects of pre- by Manderscheid RW, Sonnenschein MA.
for high-risk patients. Hospital and Com-
discharge communication with new munity Psychiatry 32:638–639, 1981 DHHS Pub (SMA) 92-1942. Washington,
outpatient clinicians among these DC, US Government Printing Office, 1992
7. Axelrod S, Wetzler S: Factors associated
patients. Second, as previously not- with better compliance with psychiatric af- 23. Sheehan DV, Lecrubier Y, Sheehan KH, et
ed, the observational design of the tercare. Hospital and Community Psychia- al: Comparison of the Mini International
try 40:397–401, 1989 Neuropsychiatric Interview with the
study made it impossible to exclude SCID-P and CIDI: a validity study. Psy-
effects of unmeasured treatment or 8. Frances A, Docherty JP, Kahn DA: The ex- chopharmacology Bulletin 31:616, 1995
patient variables that could have pert consensus guideline series: treatment
of schizophrenia. Journal of Clinical Psychi- 24. Guy W: ECDEU Assessment Manual for
contributed to the observed out- Psychopharmacology, rev ed. Washington,
atry 57(suppl 12B):1–58, 1996
comes. Third, the study sample rep- DC, US Department of Health, Education,
resented a highly impaired inpatient 9. McIntosh J, Worley N: Beyond discharge: and Welfare, 1976
telephone follow-up and aftercare. Journal
population who were referred to of Psychosocial Nursing and Mental Health 25. Caton C, Koh S, Fleiss JL, et al: Rehospi-
outpatient clinicians who had not Services 32:21–27, 1994 talization in chronic schizophrenia. Journal
of Nervous and Mental Disease 173:139–
previously treated them, and the re- 148, 1985
10. Bostelman S, Callan M, Rolincik LC, et al:
sults may not generalize to other pa- A community project to encourage compli-
26. Frank AF, Gunderson JG: The role of the
tient groups. ance with mental health treatment after-
therapeutic alliance in the treatment of
The period immediately after psy- care. Public Health Reports 109:153–157,
schizophrenia: relationship to course and
1994
chiatric hospital discharge is especial- outcome. Archives of General Psychiatry
47:228–236, 1990
ly hazardous. During this time, pa- 11. Chen A: Noncompliance in community
psychiatry: a review of clinical interven-
tients are at increased risk for treat- tions. Hospital and Community Psychiatry
27. Olfson M, Boyer CA, Hansell S, et al: To-
ment dropout (25), clinical relapse ward an integrated approach to the study of
42:282–287, 1991 inpatient treatment of schizophrenia. New
(32), homelessness (33), and suicide Directions for Mental Health Services, no
12. Penn DL, Hope DA, Spaulding W, et al:
(34). When inpatients with schizo- Social anxiety in schizophrenia. Schizo- 73:25–38, 1997
phrenia are referred to clinicians phrenia Research 11:277–284, 1994 28. Wickizer TM, Lessler D, Travis KM: Con-
whom they do not know, the risks may 13. Strauss JS, Rakfeldt J, Harding CM, et al:
trolling inpatient psychiatric utilization
be particularly great. Introducing in- through managed care. American Journal
Psychological and social aspects of negative
of Psychiatry 153:339–345, 1996
patients to their outpatient clinician symptoms. British Journal of Psychiatry
155(suppl):128–132, 1989
before patients leave the hospital may 29. Persad E, Kazarian SS, Joseph LW: The
Mental Hospital in the 21st Century. Mid-
help improve their short-term clinical 14. Isele R, Merz J, Malzacher M, et al: Social
dletown, Ohio, Wall & Emerson, 1992
outcome. ♦ disability in schizophrenia: the controlled
prospective Burghölzi study. European 30. Olfson M, Mechanic D: Mental disorders
Archives of Psychiatry and Neurological in public, private nonprofit, and proprietary
Acknowledgments Sciences 234:348–356, 1985 general hospitals. American Journal of Psy-
This research is supported by the Robert chiatry 153:1613–1619, 1996
Wood Johnson Foundation and grants 15. Hamilton NG, Ponzoha CA, Cutler DL, et
43450 and 01042 from the National Insti- al: Social networks and negative versus pos- 31. Green MF: What are the functional conse-
tute of Mental Health. itive symptoms of schizophrenia. Schizo- quences of the neurocognitive deficits of
phrenia Bulletin 15:625–633, 1989 schizophrenia? American Journal of Psychi-
atry 253:321–330, 1996
References 16. Boyer CA, Olfson M, Kellermann SL, et al:
Studying inpatient treatment practices in 32. Caton CLM, Goldstein JM, Serrano O, et
1. Miner CR, Rosenthal RN, Hellerstein DJ, schizophrenia: an integrated methodology. al: The impact of discharge planning on
et al: Prediction of compliance with outpa- Psychiatric Quarterly 66:293–320, 1995 chronic schizophrenic patients. Hospital
tient referral among inpatients with schizo- and Community Psychiatry 35:255–262,
phrenia and psychoactive substance use 17. Spitzer RL, Williams JBW, Gibbon M: 1984
disorders. Archives of General Psychiatry Structured Clinical Interview for DSM-III-
54:706–712, 1997 R—Patient Edition (SCID-P, Version 1.0). 33. Caton CLM: Mental health service use
Washington, DC, American Psychiatric As- among homeless and never-homeless men
2. Bogin DL, Anish SS, Taub HA, et al: The sociation, 1990 with schizophrenia. Psychiatric Services
effects of a referral coordinator on compli- 46:1139–1143, 1995
ance with psychiatric discharge plans. Hos- 18. Overall JE, Gorham DP: Brief Psychiatric
pital and Community Psychiatry 35:702– Rating Scale. Psychological Reports 10: 34. Geddes JR, Jusczak E: Period trends in rate
706, 1984 799–807, 1962 of suicide in the first 28 days after discharge
from psychiatric hospital in Scotland,
3. Solomon P, Davis J, Gordon B: Discharged 19. Endicott J, Spitzer RL, Fleiss JL, et al: The 1968–92. British Medical Journal 311:357–
state hospital patients’ characteristics and Global Assessment Scale: a procedure for 360, 1995

PSYCHIATRIC SERVICES ♦ July 1998 Vol. 49 No. 7 917

918 PSYCHIATRIC SERVICES ♦ July 1998 Vol. 49 No. 7