Two-year prospective case-controlled study of a case management program for community-dwelling individuals with schizophrenia.

BACKGROUND: A community-based rehabilitation program is an essential element of the comprehensive treatment of individuals with schizophrenia.
OBJECTIVE: Assess the long-term effects of a community-based case management program for providing rehabilitations services to individuals with schizophrenia.
METHODS: A total of 730 community-residing participants who met ICD-10 diagnostic criteriafor schizophrenia were enrolled, 380 in the case management group and 350 in the control group from two districts in Shanghai. Case management involved monthly training visits with patients and their co-resident family members that focused on encouraging medication adherence. Participants were assessed every three months for 24 months with the Camberwell Assessment of Need (CAN), Positive and Negative Syndrome Scale (PANSS), WHO-Disability Assessment Scale (WHO-DAS), and the Quality of Life Scale (QOLS). Level of discomfort due to side-effects was also assessed every three months. Individuals who discontinued their antipsychotic medication without physician approval for one month or longer at any time during follow-up were classified as 'self-determined medication discontinuation'.
RESULTS: Compared to the treatment as usual group (i.e., follow-up management every 3 months), by the end of the two-year follow-up those who participated in the case management program had significantly lower rates of medication discontinuation, significantly less severe negative symptoms, lower relapse rates and lower rehospitalization rates. Other factors that had an independent effect on discontinuation of medication included educational level (those with more education had higher discontinuation rates), lack of family supervision of medication, higher dosages of medication, and greater medication-related discomfort.
CONCLUSIONS: Case management is a feasible and effective long-term method for improving the rehabilitation outcomes of community residents with schizophrenia. Our results highlight the need to involve family members in the management of patients' medication, to use the minimum effective dosage of medication, and to aggressively manage all side-effects.

Introduction

As part of the three-tier mental health service delivery system established in China in the late 1970s, in the early 1980s the Shanghai Municipality set up a service network at the municipal, district (19 districts) and sub-district (220 sub-districts) levels that came to be known as the ‘Shanghai Model’ of psychiatric rehabilitation.[1] The municipal-level (tertiary) services were coordinated by the Shanghai Mental Health Center, the district-level (secondary) services were provided by district psychiatric hospitals, and the sub-district level (primary) community-based services were provided by community mental rehabilitation centers. Most of the hospitals were run by the Bureau of Health but some were run by the Bureau of Civil Affairs and the Bureau of Public Security; the rehabilitation centers were primarily administered by local authorities (e.g., the neighborhood committee). By the mid-1990s a number of serious problems had surfaced within this service delivery network because of a rapidly increasing patient load and because the community rehabilitation centers were not financially viable.[2] To address these problems over the last 20 years the number of beds at the district level hospitals has more than doubled (currently more than 8000 beds); the community rehabilitation centers have been replaced by community ‘Sunshine Centers’ centrally managed and financed by the Disabled Persons Federation; and the Shanghai Center for Disease Control (CDC) has developed a three-tier monitoring system of persons with serious mental disorders at the city, district and sub-district levels – the ‘Shanghai Management System for Patients with Serious Mental Disorders’ – that coordinates 3-monthly follow-up visits (by public health staff) and the provision of free medication to the 11,376 individuals with schizophrenia currently (as of the end of 2013) registered in the system.

Despite this expansion of services, serious problems remain: (a) only 17% of the Sunshine Centers and the district and sub-district CDC mental illness management groups have any staff members trained in rehabilitation; (b) the Sunshine Centers, the CDC monitoring system and the psychiatric hospitals are run by different government departments and institutions; and (c) the CDC center at the tertiary psychiatric hospital responsible for coordinating follow-up and services to all seriously mentally ill individuals in the city does not have the manpower to provide technical support and supervision to CDC centers situated in the 19 district-level psychiatric hospitals or to the CDC working groups in the 220 sub-districts that provide the direct community-level supervision of the patients. As a result of these problems, by the end of 2013 only 20% (2285/11,376) of the patients with schizophrenia registered in Shanghai’s CDC management system received community-based rehabilitation services at the Sunshine Centers.

Clearly some changes are needed to resurrect the ‘Shanghai Model’.[3] Based on the experiences of other countries that use a variety of case management models to provide community-based services for individuals with severe mental illnesses,[4]-[6] we developed a case management protocol suitable for community-dwelling individuals with schizophrenia in Shanghai and assessed its effectiveness over a two-year follow-up period.

Methods

Sample

As shown in Figure 1, Individuals with schizophrenia were recruited from the Xuhui and Hongkou Districts of Shanghai. There are eight sub-districts in Hongkou District with 5231 individuals with schizophrenia registered in the CDC management system. There are 13 sub-districts in Xuhui District with 3827 registered individuals with schizophrenia. Patients with schizophrenia registered in the CDC system in these two districts have all received monthly follow-up visits and free medication for the last five years. The inclusion criteria for the current study were as follows: (a) ≥18 years of age; (b) met ICD-10 diagnostic criteria for schizophrenia; (c) had impairment in daily functioning due to the mental illness; and (d) the participants’ family members or guardians provided written informed consent.

Figure 1.

Flow chart of the study

A convenience sample of 380 subjects who met the inclusion criteria was recruited for the case management intervention group (study group) by the administrators of the district-level CDC monitoring system (who knew the patients well because they regularly follow-up patients each month). Controls matched for age (±5 years), gender, and duration of illness (±5 years) were identified from the same sub-districts for the first 350 subject in the study group. Due to time constraints we were unable to identify controls for the last 30 subjects recruited for the study group.

Case management

The implementation of case management followed six steps. (1) Eight case management teams (four case management teams for each district) were established. Each team had three or four members including at least one community psychiatrist, one nurse and one community health worker. Each team was responsible for managing 30 to 50 cases. (2) Baseline evaluations were conducted by each team to assess the health condition, recovery status, daily functioning, employment status, and social activities of participants in both the study group and the control group. (3) The needs of the participants were assessed using the Camberwell Assessment of Need[7](CAN). And (4) using information from the needs assessment, patient-specific case management plans were developed for all participants in the study group based on the content of the ‘Handbook of Rehabilitation for Individuals with Schizophrenia’,[8] which includes information on drug adherence training, daily skills training, family psychological intervention and so forth. (5) Throughout the two-year follow-up period the case management team visited each participant in the study group monthly (for about 30 minutes) to implement the training specified in the patient-specific management plan. (6) Patients in the study group also participated in eight group training sessions that covered the same material at the local Sunshine Center during the first year of enrollment.

Participants in the control group received regular monitoring follow-up every three months by the staff members of the district and sub-district monitoring system.

Evaluation

The Positive and Negative Syndrome Scale[9] (PANSS) was used to assess symptom severity; it contains a positive subscale, a negative subscale, and a general psychopathology subscale. The WHO Disability Assessment Scale[10] (WHO-DAS) was used to assess social functioning; it has 12 items that assess family functioning, employment functioning and social functioning. The Quality of Life Scale[11] (QOLS) includes 40 items divided into six domains: daily life, family life, social relationship, financial status, working status, and sense of security (higher scores represent a higher quality of life). Assessments were conducted at the baseline, and at 3, 6, 9, 12 and 24 months after recruitment. Team members were trained before the study and their inter-rater reliability for the various scales was good (Kappa=0.82-0.93). The study was conducted from August 1, 2008 to December 31, 2011. At each of these follow-up assessments clinicians also assess the degree of discomfort the individual experienced in the prior month due to side effects of the medication (rated ‘none’, ‘mild’, ‘obvious’ and ‘serious’ discomfort).

The main outcomes used to assess the efficacy of the case management method of supervising community-dwelling individuals with schizophrenia were medication adherence, relapse, and re-hospitalization. Medication discontinuation was defined as self-discontinuation of antipsychotic medication for one month or longer at any time during the 2-year follow-up (discontinuation by the treating clinician due to side effects was not counted). Relapse was defined as (a) an increase of at least 25% in PANSS total score, (b) an episode of self-harm, or (c) severe suicidal ideation that persists for at least one week.

Statistical analysis

SAS 9.0 statistical software was used for analysis. Repeated-measures analysis of variance (ANOVA) was used to compare between-group and within-group differences. Chi-square tests were used to compare categorical variables. A backward stepwise logistic regression was used to identify factors associated with medication adherence; the factors entered into the model included group membership, gender (male=0, female =1), age, three levels of education (middle school or less, high school, and college or higher), duration of illness, supervision of medication by family members (yes or no), type of antipsychotic medication (traditional [including clozapine] or new atypical), mean dosage over the follow-up period (chlorpromazine-equivalent), and mean rating over the follow-up period of the 4-category discomfort due to side-effects variable. We used the intention-to-treat method with the Last Observation Carried Forward (LOCF) to address missing data. The level for statistical significance was set at p<0.05.

The study was approved by the Shanghai Mental Health Center.

Results

As shown in Table 1, there were no statistically significant differences between the study group in any of the baseline demographic and treatment variables. All patients were on stable dosages of a single antipsychotic medication that were provided free of charge. Most of the patients were taking oral perphenazine, chlorpromazine, clozapine, risperidone, or olanzapine. Throughout the 24-month follow-up only nine patients had to change their medication and the method of managing medication (either by family members or by the patient himself or herself) did not change.

Comparison of illness characteristics between treatment and control groups

The proportions of individuals who dropped out of the study, discontinued their medication, relapsed, and were rehospitalized over the 24-month follow-up period are shown in Table 2. There were no differences in the dropout rates between the two groups but after one year of case management the rates of discontinuing medication, relapse and rehospitalization were significantly greater in the control group than in the case management group.

Psychiatric symptoms, social functionality, and quality of life scale score

Table 3 shows the results for PANSS, WHO-DAS and QOLS. Repeated measures ANOVA analysis indicated statistically better outcomes over the 24-month period in the total PANSS score, the PANSS negative syndrome score, the WHO-DAS total score, and in the daily life, family life and sense of security subscales scores of the QOLS. In most cases these differences became more evident over time and were greatest at the time of the 12-month and 24-month follow-up evaluations.

Analysis on the factors related to the discontinuation of medication

Some patients need to change medications because of serious side effects but most patients discontinue their medication for other reasons. This is universally recognized as one of the most important determinants of relapse so it is helpful to understand the factors that are associated with self-determined drug discontinuation. To determine the factors associated with self-determined discontinuation of medication (defined in this study as stopping antipsychotic medication without physician approval for one month or longer at any time during the 2-year follow-up), and to determine whether or not case management has an independent effect of discontinuation, we conducted a logistic backward stepwise regression with self-determined medication discontinuation as the dependent variable and group membership, gender, age, education level, duration illness, type of medication (i.e., typical or new atypical), high (versus low) mean chlorpromazine-equivalent dosage of medication (dichotomized at 400mg/d), family member supervision of medication, and medication tolerance (overall assessment over the 2 year follow-up by the evaluating clinician) as the independent variables. As shown in Table 4, five variables remained in the final model, indicating that they independently affected discontinuation of medication. (1) Individuals participating in the case management program were significantly less likely to discontinue medication than those in the control group. (2) Patients with higher level of education were more likely to discontinue than those with less education. (3) Patients who supervised medication themselves were more likely to discontinue medication than those for whom coresident family members supervised the use of medication. (4) Patients on higher dosages of medication were more likely to discontinue medications than those on lower dosages. (5) Patients who experienced greater medication-related discomfort were more likely to discontinue medications than who experienced less discomfort; there was a clear stepwise increase in risk of discontinuation as the level of discomfort increased.

Discussion

Key findings

This two-year follow-up study in a relatively large sample of community dwelling individuals with schizophrenia found that compared to treatment as usual a case management program involving monthly training visits with patients and their co-resident family members resulted in significantly lower rates of medication discontinuation, significantly less severe negative symptoms, lower relapse rates and lower rehospitalization rates. Other factors that had an independent effect on discontinuation of medication over the two-year follow-up period included educational level (those with a high educational level were more likely to discontinue medication), lack of family supervision of medication, higher dosage of medication, and greater medication-related discomfort (i.e., side effects). These results highlight the need to involve family members in the management of patients' medication, to use the minimum effective dosage of medication, and to aggressively manage side-effects – even those side effects that pose no medical risks for the patient. The reasons individuals with higher levels of education were more likely to discontinue medication are unknown; further research will be needed to understand this relationship and to modify individualized case management strategies accordingly.

Limitations

The major limitations of this study are that group assignment was not random and that evaluation of the outcome measures was conducted by team members who knew the group assignment of the participants (i.e., it was not a ‘blind’ evaluation). The matching of intervention group subjects with control group subjects we used to balance potential confounders was not complete (we did not have the time to identify matched controls for the last 30 subjects enrolled in the intervention group). However, there were no significant differences in the main demographic and clinical measures at baseline between groups so we do not believe that the failure to randomize participants had a substantial effect on our outcome measures. Two of the main outcome measures – medication discontinuation and rehospitalization – were easily observable events, so it is unlikely that the lack of blinding of raters would have affected these outcomes. However, other measures such as the PANSS, QOLS, and WHO-DAS scale require clinician judgment, so the possibility that the scores were biased in favor of the case management treatment cannot be definitively excluded. Other limitations are that the selected sample only included individuals who were already registered in registry of mentally ill individuals operated by the Center for Disease Control who resided in two districts of Shanghai that have high-quality health care services, regular follow-up of all individuals with serious mental illnesses every three months and free medication. The results may be different in community residents with schizophrenia who are not in the formal registry, for those who live in communities with less developed health care services, and for those living in communities where free medications are not available for persons with serious mental illnesses.

Implications

Family intervention, a key component of case manage-ment during rehabilitation, plays a vital role in the improvement of social functioning and reintegration of individuals with schizophrenia. Case management involves an iterative process of assessing each patient's strengths and weaknesses, providing regular and ongoing training (typically targeted on medication management, activities of daily living and social functioning), periodically re-assessing patients' functioning, and making mid-course corrections in the training package.[12]-[15] Given that discontinuation of medication is the main reason for relapse and rehospitalization,[16],[17] training patients and their co-resident family members about adherence and providing them with psychological support for adherence (even in the presence of discomfort and side-effects) is the cornerstone of successful case management interventions. Many factors affect the adherence to medication, including the psychosocial characteristics of the patient and the patient's co-resident family members, fluctuations in disease symptomatology, the type and dosage of antipsychotic medication used, and the patient's tolerance of the medication.[18],[19] It is important to make ongoing assessments of these factors during the follow-up of each patient and to actively address them as part of each patient-specific case management plan.

Case-management of persons with serious mental illness requires a substantial investment of professional personnel. But it also results in substantial improvement of patients' functioning and a resultant decrease in the usage of inpatient services. Cost-benefit analyses are needed to determine the potential costs and benefits of up-scaling this type of service to other urban and rural parts of China.

Table 1.

Social and demographical characteristics of the participants

Characteristic	Case management(n=380)	Control(n=350)	statistics	p
Mean (sd) age	39.5(7.9)	40.7 (9.3)	t=0.56	0.683
Mean (sd) duration of illness in years	11.6 (6.1)	10.4 (7.6)	t=1.15	0.312
Gender [n (%)]
Male	279 (73.4)	243 (69.4)	χ2=1.43	0.233
Female	101 (26.6)	107 (30.6)
Education level [n (%)]
Middle school or lower	138 (36.3)	135 (38.6)	χ2=0.68	0.713
High school	191 (50.3)	174 (49.7)
Junior college or above	51 (13.4)	41 (11.7)
Marital status [n (%)]
Not married	226 (59.6)	235 (67.1)
Married	129 (34.0)	90 (25.6)	χ2=5.90	0.052
Separated	25 (6.4)	25 (7.3)
Employment [n (%)]
Unemployment	53 (13.9)	50 (14.3)	χ2=0.32	0.957
Employed	50 (13.2)	45 (12.9)
Medical leave	180 (47.4)	160 (45.7)
Retired	97 (25.5)	95 (27.1)
Main caregiver [n (%)]
Parents	63 (16.6)	61 (17.4)
Spouse	120 (31.6)	88 (25.1)	χ2=3.77	0.152
Children or other relatives	197 (51.8)	201 (57.5)
Number of previous hospitalizations [mean (sd)]	1.8 (1.7)	1.8 (2.0)	t=0.03	0.956
Family member supervises medication [n (%)]	276 (72.6)	245 (70.0)	χ2=0.62	0.432
Uses new atypical antipsychotic [n (%)]	142 (37.4)	139 (39.7)	χ2=0.42	0.515
Mean chlorpromazine equivalent dose over 2-year follow-up (mg/d) [mean(sd)]	216.2 (146.8)	237.5 (135.3)	t=1.18	0.129
Uses relatively high mean dose of antipsychotic over 2-year follow-up (Chlorpromazine equivalent dose>400mg/day) [n (%)]	62 (16.4)	53 (15.1)	χ2=0.22	0.642
Tolerance of medication over 2-year follow-up [n (%)]
No discomfort	67 (19.5)	51 (14.6)	χ2=4.27	0.234
Mild discomfort	188 (48.3)	172 (49.1)
Obvious discomfort	82 (21.1)	77 (22.0)
Serious discomfort	43 (11.1)	50 (14.3)

Table 2.

Comparison of rates of loss to follow up, medication discontinuation, relapse, and rehospitalization between the case management group and the control group at five follow-up points (n, %)

Time of follow-up	Case management(N=380)	Control(n=350)	χ2	p
3 months
Lost to follow up	0(0)	0(0)	--	--
Discontinuation of medicine	7 (1.9)	10 (2.9)	0.82	0.367
Relapse	2 (0.5)	2 (0.6)	Fisher's exact test	1.000
Rehospitalization	2 (0.5)	3 (0.9)	Fisher's exact test	0.675
6 months
Lost to follow up	4 (1.1)	5 (1.4)	Fisher's exact test	0.744
Discontinuation of medicine	14 (3.7)	14 (4.1)	0.05	0.824
Relapse	4 (1.1)	7 (2.0)	Fisher's exact test	0.368
Rehospitalization	6 (1.6)	11 (3.1)	1.96	0.162
9 months
Lost to follow up	10 (2.6)	11 (3.1)	0.17	0.680
Discontinuation of medicine	37 (9.7)	43 (12.3)	1.21	0.271
Relapse	19 (5.0)	25 (7.1)	1.47	0.224
Rehospitalization	17 (4.8)	22 (6.3)	1.18	0.279
12 months
Lost to follow up	17 (4.8)	16 (4.6)	0.02	0.878
Discontinuation of medicine	43 (11.3)	58 (16.6)	4.22	0.039
Relapse	30 (7.9)	45 (12.9)	4.87	0.027
Rehospitalization	25 (6.6)	39 (11.1)	4.74	0.029
24 months
Lost to follow up	36 (9.5)	37 (10.6)	0.24	0.621
Discontinuation of medicine	50 (13.2)	80 (22.9)	11.71	<0.001
Relapse	40 (10.5)	60 (17.1)	6.75	0.009
Rehospitalization	34 (8.9)	53 (15.1)	6.67	0.009

Table 3.

Mean (sd) results for Positive and Negative Syndrome Scale (PANSS), WHO Disability Assessment Scale (WHO-DAS) and Quality of Life Scale (QOLS) for 380 patients in the case management (intervention) group and 350 patients in the control group at baseline and at five follow-up points

Scale/subscale	Group	Baseline	3 month	6 month	9 month	12 month	24 month	Fa	p
PANSS total score	intervention	51.5 (16.8)	50.4 (16.3)	50.0 (16.1)	49.2 (15.7))	47.9 (17.8)b	45.1 (17.1)b	4.81	<0.001
	control	52.3 (15.9)	52.3 (16.3)	51.5 (16.7)	51.7 (16.8)	51.7 (16.9)	51.6 (18.8)
	F	0.07	0.29	0.07	0.28	2.64	2.97
	p	0.899	0.636	0.892	0.607	0.044	0.038
PANSS positive syndrome score	intervention	11.5 (4.9)	11.5 (4.7)	11.3 (5.3)	11.2 (5.4)	11.3 (5.2)	11.0 (5.1)	0.07	0.740
	control	11.5 (4.9)	11.4 (5.7)	11.5 (5.2)	11.4 (5.5)	11.6 (5.5)	11.8 (5.1)
	F	0.02	0.03	0.03	0.02	0.03	0.04
	p	0.930	0.923	0.916	0.924	0.915	0.911
PANSS negative syndrome score	intervention	17.9 (6.0)	17.7 (5.7)	17.5 (6.1)	16.2 (6.3)	14.3 (6.6)	12.2 (6.1)c	4.32	<0.001
	control	15.4 (6.2)	15.3 (6.5)	15.9 (6.8)	16.1 (7.3)	16.3 (7.5)	16.2 (7.1)
	F	1.88	1.82	1.87	0.02	1.81	3.79
	p	0.1039	0.1721	0.1267	0.9281	0.1904	0.001
PANSS general psychopathology score	intervention	25.4 (8.4)	25.2 (8.2)	25.4 (8.3)	25.1 (8.5)	25.1 (8.1)	24.9 (8.2)	0.89	0.390
	control	25.3 (7.7)	25.6 (8.1)	25.8 (8.4)	25.6 (8.4)	25.4 (7.9)	25.5 (7.4)
	F	0.02	0.02	0.03	0.03	0.02	0.31
	p	0.919	0.921	0.904	0.913	0.930	0.573
WHO-DAS	intervention	6.5 (0.4)	6.3 (0.5)	6.3 (0.3)	5.8 (0.4)	4.5 (0.3)b	4.3 (0.2)b	2.79	0.033
	control	6.1 (0.4)	6.1 (0.5)	6.4 (0.7)	6.4 (0.9)	6.7 (1.1)	6.5 (0.9)
	F	0.04	0.03	0.02	0.29	2.95	2.87
	p	0.888	0.914	0.916	0.620	0.038	0.040
QOLS daily life subscale score	intervention	14.4 (3.8)	14.6 (3.7)	14.6 (3.7)	15.9 (3.9)	17.2 (4.4)	18.4 (4.8)b	3.79	0.001
	control	15.8 (3.9)	15.7 (4.0)	16.1 (4.3)	16.7 (3.9)	16.3 (4.1)	16.6 (4.1)
	F	0.27	0.16	2.32	0.26	0.26	2.94
	p	0.599	0.701	0.079	0.681	0.681	0.037
QOLS family life subscale score	intervention	12.0 (3.0)	12.3 (2.8)	12.9 (3.3)	12.6 (3.1)	13.6 (3.2)	14.8 (3.5)b	3.22	0.015
	control	12.1 (2.9)	12.7 (3.2)	11.4 (3.7)	12.6 (3.6)	12.3 (3.1)	12.4 (3.1)
	F	0.02	0.05	0.21	0.02	0.18	3.10
	p	0.933	0.893	0.646	0.905	0.642	0.034
QOLS social relationships subscale score	intervention	7.2 (3.0)	7.2 (3.1)	7.6 (3.7)	7.4 (3.5)	7.4 (3.3)	7.5 (3.2)	0.04	0.875
	control	7.3 (3.1)	6.8 (3.4)	6.6 (3.8)	6.9 (3.5)	7.2 (3.6)	7.1 (3.5)
	F	0.03	0.05	0.15	0.26	0.03	0.06
	p	0.925	0.812	0.602	0.675	0.914	0.834
QOLS financial status subscale score	intervention	7.5 (1.7)	7.3 (1.6)	7.5 (1.4)	7.2 (2.1)	7.6 (1.8)	7.5 (1.9)	0.02	0.930
	control	7.2 (2.1)	6.8 (1.7)	7.0 (2.3)	6.9 (2.1)	7.1 (2.5)	7.0 (2.6)
	F	0.07	0.28	0.06	0.04	0.07	0.07
	p	0.789	0.625	0.828	0.872	0.805	0.799
QOLS working status subscale score	intervention	8.9 (1.9)	9.3 (2.1)	9.1 (2.2)	9.3 (1.9)	9.2 (2.3)	9.4 (1.7)	0.47	0.174
	control	9.1 (2.1)	8.9 (2.3)	9.0 (2.5)	8.9 (2.1)	9.0 (1.8)	8.7 (1.9)
	F	0.03	0.03	0.02	0.05	0.04	1.03
	p	0.925	0.917	0.934	0.833	0.916	0.144
QOLS sense of security subscale score	intervention	3.1 (0.2)	3.5 (0.7)	3.6 (0.5)	4.2 (0.7)b	4.4 (0.4)	4.9 (1.0)b	3.12	0.025
	control	3.1 (0.4)	2.9 (0.7)	3.0 (0.7)	2.4 (0.6)	2.6 (0.5)	2.1 (0.3)
	F	0.01	1.17	0.06	1.99	2.22	2.99
	p	0.988	0.219	0.807	0.106	0.085	0.032

aF value comparing within-group difference (over time) of the overall model from repeated-measures ANOVA

bp<0.05 - within group difference (vs. baseline)

cp<0.01- within group difference (vs. baseline)

Table 4.

Factors associated with ‘self-determined’ discontinuation of antipsychotic medication during two years of treatment in 380 patients with schizophrenia who received case management and 350 patients who received treatment as usuala

	Odds ratio	95% Confidence Interval	p
Received case management	0.20	0.07-0.62	0.005
Educational level
middle school or below	1.00	---	---
high school	2.04	1.27-7.28	0.032
college	2.83	0.99-6.21	0.052
Medication not supervised by family member	5.05	1.31-12.17	0.026
High mean chlorpromazine-equivalent dosage (>400mg/day)	6.31	1.98-21.99	0.029
Tolerance of medication
no discomfort	1.00	---	---
mild discomfort	1.17	0.94-12.60	0.053
moderate discomfort	3.10	1.56-16.63	0.038
severe discomfort	7.76	2.33-37.23	0.010

aFactors considered in the backward stepwise regression include group membership, gender, age, education level, duration of illness, type of medication (i.e., typical or new atypical), high (versus low) mean chlorpromazine-equivalent dosage of medication, family member supervision of medication, and medication tolerance (overall mean assessment or the 2-year period by treating clinician).