, Volume 172, Issue 4, pp 393–399

Fine-tuning risperidone dosage for acutely exacerbated schizophrenia: clinical determinants


  • Hsien-Yuan Lane
    • Department of PsychiatryChina Medical University and Hospital
  • Yue-Cune Chang
    • Department of MathematicsTamkang University
  • Chih-Chiang Chiu
    • Laboratory of Biological PsychiatryTaipei City Psychiatric Center
  • Sue-Hong Lee
    • Laboratory of Biological PsychiatryTaipei City Psychiatric Center
  • Cher-Yeang Lin
    • Department of PsychiatryChina Medical University and Hospital
    • Department of PsychiatryTzu-Chi University
Original Investigation

DOI: 10.1007/s00213-003-1685-6

Cite this article as:
Lane, H., Chang, Y., Chiu, C. et al. Psychopharmacology (2004) 172: 393. doi:10.1007/s00213-003-1685-6



Risperidone doses for acute schizophrenia were rather high in most recent studies.


We tested a hypothesis that fine-tuning risperidone dosage to relieve side effects still yields efficacy. Clinical factors influencing the dosing were also determined.


One hundred and forty-six schizophrenia inpatients with acute exacerbation entered a prospective, 6-week, repeated measures study. Risperidone doses were titrated to 6 mg/day (if tolerable) within 7 days, but were lowered thereafter if adverse reactions appeared. Efficacy and safety were measured biweekly.


Forty-eight patients tolerated the 6-mg/day target dose well, while the other 98 received lower final doses (mean±SD=3.4±0.9 mg/day) to curtail adverse effects. At endpoint, 64.3% of the low-dose patients and 43.8% of the high-dose subjects responded to treatment [≥20% reduction in the Positive and Negative Syndrome Scale (PANSS) total score] (P=0.018). In detail, the low-dose individuals were significantly superior in percentage changes in the PANSS total and general-subscale scores at endpoint. The low-dose group also tended to improve more (albeit statistically insignificantly) in the PANSS positive and negative subscales and other efficacy measures. Compared to the patients with undifferentiated subtype, those with disorganized subtype received higher dosage by 0.90 mg/day, after controlling for other variables (P=0.008). Paranoid subtype was similar to undifferentiated subtype in drug doses. Patients with longer illness duration also showed a trend to use higher dosage (P=0.078).


These findings suggest that dosage adjustment to diminish side effects does not compromise risperidone response and that disorganized patients and perhaps patients with longer illness duration are prone to receive larger doses.


Diagnosis subtypeDisorganized subtypeDoseGenderRisperidoneChinese


Risperidone is a serotonin/dopamine antagonist with efficacy for both positive and negative symptoms of schizophrenia (Chouinard et al. 1993; Marder and Meibach 1994; Bouchard et al. 2000). Although it is now the most frequently prescribed atypical antipsychotic agent, its optimal dosing constitutes a challenge for clinicians. The pivotal studies (Chouinard et al. 1993; Marder and Meibach 1994) concerning chronic patients identify a dosage of 6 mg/day (on average) to produce optimal efficacy. Recently, risperidone treatment for general patients (Gutierrez-Esteinou and Grebb 1997; Collaborative Working Group on Clinical Trial Evaluations 1998; Kasper 1998) or first-episode patients (Kopala et al. 1997; Lane et al. 2001) has involved efforts to use lower doses. It has been suggested that unduly high doses may be detrimental to risperidone’s efficacy (Gutierrez-Esteinou and Grebb 1997; Kopala et al. 1997; Collaborative Working Group on Clinical Trial Evaluations 1998; Kasper 1998; Lane et al. 1999, 2001). However, doses of 8 mg/day or higher were still utilized in certain chronic patients (Tran et al. 1994; Klieser et al. 1995; Lindenmayer et al. 1998; Bouchard et al. 2000; Csernansky et al. 2002; Volavka et al. 2002). For acutely exacerbated individuals, the mean endpoint doses in most recent studies (Hoyberg et al. 1993; Huttunen et al. 1995; Blin et al. 1996; Pappas et al. 1997; Peuskens et al. 1999) were 8–12 mg/day; another study utilized a fixed dose of 6 mg/day (Zhang et al. 2001).

Since the interindividual variability of risperidone pharmacokinetics is greater than 40-fold (Anderson et al. 1994; Darby et al. 1997), the blood drug level rather than the dose per se may more directly affect clinical efficacy. Risperidone’s therapeutic blood levels, unfortunately, are unclear (Bondolfi et al. 1998; Lane et al. 2000). Furthermore, significant dissociation of brain and plasma kinetics with risperidone was reported recently (Tauscher et al. 2002). Although there is a biochemical model of dopamine receptor occupancy (Kapur and Seeman 2001; Lane et al. 2002), it is just not clinically useful. Clinically practical and individualized dosing strategies thus need to be established. In a pilot study for acutely ill patients (Lane et al. 2000), we found that up to 6 mg/day risperidone was efficacious. More importantly, reducing the dose to relieve side effects yielded possibly better efficacy. Among the 30 patients, 17 tolerated a 6-mg dose well, while the other 13 received a lower mean dose of 3.6 mg to curtail side effects. As a result, the 13 low-dose patients tended to excel, but statistically insignificantly, in terms of percentage changes in the scores of the Positive and Negative Syndrome Scale (PANSS) (Kay et al. 1987) and three subscales (positive, negative, and general psychopathology) than the high-dose patients (Lane et al. 2000). This preliminary finding, if substantiated by further evidence, could assist clinicians in optimizing risperidone effectiveness with fewer side effects. The current study aimed to verify this dosing hypothesis and to examine clinical determinants for the dosage in a larger schizophrenia population.

Materials and methods

This prospective, open-labeled, and repeated measures study was conducted in the inpatient unit of the institution. The protocol was approved by the facility’s institutional review board and has therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. All patients signed the written informed consent prior to their inclusion in the study.


All newly hospitalized, Han Chinese, schizophrenic patients with acute exacerbation were screened and evaluated by experienced psychiatrists. The Structured Clinical Interview for DSM-IV (American Psychiatric Association 1994) was conducted for the diagnosis. Han Chinese patients in Taiwan entered into this study if they 1) were physically healthy and had all laboratory parameters within normal limits, 2) aged 18–60 years, 3) satisfied DSM-IV criteria for schizophrenia, 4) had a minimum baseline total score of 60 on the PANSS, 5) had no DSM-IV diagnosis of substance (including alcohol) abuse, 6) were non-smokers, 7) had not received depot antipsychotics for the preceding 6 months, 8) had no history suggesting that antipsychotics treatment would be contraindicated, and 9) had never received atypical antipsychotics before.

Study design and clinical assessments

Our dosing strategy was to avoid treatment-emergent side effects (excluding weight change). The subjects initially received a placebo for up to 7 days to curtail adverse effects (if any) from previous drugs as far as possible. Thereafter, risperidone was titrated up from the starting dose of 2 mg/day on day 1 to the target dose of 6 mg/day (or lower in case of side effects) within 1 week. That is, the only reason for not increasing the dose was the presence of side effects, no matter whether the patients responded clinically to lower doses. In details, risperidone dosage was adjusted after systematic safety assessment on day 2 (dose: 3 mg/day, or lower if side effects), day 3 (4 mg/day, or lower if side effects), day 5 (6 mg/day, or lower), and day 7 (the same dose as that used on day 5, or lower if side effects). From day 8 to day 42 of risperidone treatment, the dose remained the same as that used on day 7, or was reduced on days 14 or 28 if drug side effects were found. To sum up, side effects were evaluated on days 0, 2, 3, 5, 7, 14, 28, and 42, by means of routine physical examination, the Extrapyramidal Symptom Rating Scale (ESRS) (Chouinard et al. 1980), and the Udvalg for Kliniske Undersogelser (UKU) Side Effect Rating Scale (Lingjaerde et al. 1987). The ESRS was designed to evaluate three types of extrapyramidal side effects (EPS): parkinsonism, dystonia, and dyskinesia (Chouinard et al. 1980). The parkinsonism subscale also includes a item to evaluate akathisia. Side-effect profiles other than EPS were determined by the UKU scale. Laboratory tests were conducted on day 0 and endpoint (day 42, or day 28 or day 14 for the patients who dropped out prematurely).

Efficacy assessments were conducted on day 0, day 14, day 28, and day 42. The main efficacy instrument was the PANSS. Categorical response was defined a priori by 20% or more reduction in the PANSS total score from baseline. Other measures included the Clinical Global Impression (CGI) scale (Guy 1976), the Global Assessment of Functioning (GAF) (DSM-IV axis V), and the Nurses’ Observation Scale for Inpatients Evaluation (NOSIE) (Honigfeld and Klett 1965). The NOSIE, with 30 items, captures social functioning, daily activities, and personal hygiene of hospitalized seriously ill patients. Symptom or behavior frequency is scored from 1 (=never) to 5 (=always) for each item. This scale thus complements the PANSS (that focuses more on psychopathology) and has been useful in the clozapine trial conducted by Kane et al (1988). An experienced research psychiatrist (H.-Y.L.) administered all psychopathology and side-effect rating instruments and blindness was therefore not achieved.

Regarding the concomitant medication, lorazepam was allowed as needed for insomnia (PO) or agitation (IM), and benztropine, for EPS. No other centrally acting drugs or cytochrome P450 inducers (or inhibitors) that might interfere with risperidone’s metabolism (Lane and Chang 1998) were permitted.

Statistical analyses

At endpoint, the subjects receiving the target dose of 6 mg/day were defined as the high-dose patients and the others who did not attain that dose, the low-dose patients. The first aim of this study was to compare the two dose groups in terms of (1) demographic characteristics, including gender, age, body weight, diagnosis subtype, age at the onset of psychosis, duration of illness, number and total duration of prior hospitalizations, and (2) performances in various drug-efficacy instruments from baseline to endpoint. For the between-group comparisons, two-tailed Mann-Whitney U statistics tests were used for dimensional data, while two-tailed χ2 tests were used for categorical data. Since most of the dimensional data in this population appeared to be asymmetrically distributed, Mann-Whitney U-tests rather than t-tests were utilized.

Another main purpose of this study was to identify clinical correlates of risperidone dosage. We postulated that patients’ demographic characteristics and possibly baseline psychopathology (or certain aspects of functioning) might influence their susceptibility to adverse drug reactions. We collected the doses as the response values. Potential prognostic factors included risperidone treatment duration (weeks 0–6), patients’ demographic characteristics (shown above), and any baseline assessment scale score that was found to be different between the two dose groups. Since our dosing strategies relied on drug safety rather than efficacy, the prognostic factors did not include the post-treatment drug-efficacy scores. To evaluate the possible effects of the prognostic factors on the response values, the multiple linear regression model, which is capable of adjusting other prognostic factors, was used. Furthermore, due to the repeated assessments of each subject, multiple linear regression with the generalized estimating equation (GEE) method (Zeger et al. 1988) was utilized to take into account the within-subject dependence. The “PROC GENMOD” procedure under the SAS/STAT V8.1 system (SAS Institute Inc. 1999) was thus employed. Statistical significance was defined as P<0.05.


Patient disposition and drug dosing

Twenty-nine patients were excluded by the screening procedure. One hundred and forty-six patients completed at least the assessments on day 0 and on day 14 of risperidone treatment and were eligible for data analysis. The number of patients active on days 28 and 42 were 136 and 117, respectively. The reasons for premature discontinuation were concurrent somatic illness (n=4), non-compliance (n=4), poor response (n=5), and good response but willing to be discharged earlier (n=16). Since our dosing strategy was to abate side effects (excluding weight gain) as far as possible, no patients dropped out due to adverse events. Side effects other than weight change were minimal (if any) and resolved after dose reduction. The mean±SD body weight, however, rose gradually: 60.8±11.8 kg at baseline, 62.1±12.0 on day 14, 62.9±12.2 on day 28, and 63.9±11.8 on day 42 (Lane et al. 2003).

The mean±SD risperidone doses were seen to be reasonably stable: 4.4±1.5 mg/day on day 14, 4.4±1.4 mg/day on day 28, and 4.3±1.4 mg/day on day 42. At endpoint, 48 (33%) patients received the target dose of 6 mg/day well, while the other 98 (67%) patients tolerated only lower doses of risperidone (mean±SD=3.4±0.9 mg/day). Regarding the concomitant medications, 16 (16%) subjects in the low-dose group and 6 (13%) in the high-dose group received lorazepam injections IM for agitation during the trial (NS). At endpoint, the mean PO dose of lorazepam (0.9±1.1 mg/day versus 0.7±0.9 mg/day) or benztropine (0.5±1.0 mg/day versus 0.4±0.9 mg/day) also did not differ significantly between the low-dose and the high-dose groups.

Patient characteristics and treatment response between the two dose groups

No statistically significant differences were found between the two dose groups with respect to most demographic variables; however, the high-dose group had a higher percentage of disorganized-subtype patients than the low-dose group (Table 1). Besides, the mean duration of prior hospitalizations in high-dose patients was almost double that in the low-dose subjects; the between-group difference, however, did not reach statistical significance possibly due to the rather large standard deviations in both groups (Table 1).
Table 1

Demographic characteristics (mean±SD) of schizophrenic patients treated with low (<6 mg/day) or high (6 mg/day) doses of risperidone

Low-dose group (n=98)

High dose group (n=48)


Age (years)




Body weight (kg)




Age at onset of psychosis (years)




Duration of illness (months)




Number of hospitalizations




Duration of hospitalizations (weeks)




Male gender (n, %)

53/98, 54.1

32/48, 66.7


Diagnosis subtype (n, %)


67/98, 68.4

29/48, 60.4



24/98, 24.5

7/48, 14.6


7/98, 7.1

12/48, 25.0

aTwo-tailed Mann-Whitney U-tests

bTwo-tailed χ2 tests

As shown in Table 2, the high-dose recipients also had higher baseline NOSIE scores (and thus poorer social function and personal hygiene). In contrast, baseline performances in the GAF, another functioning assessment scale (but not particularly designed for acutely ill inpatients as the NOSIE), did not differ between groups. There were also no significant differences between the two dose groups in other baseline clinical assessments, including the CGI, the PANSS total, and three PANSS subscales scores (Table 2).
Table 2

Change (mean±SD) in the Clinical Global Impression (CGI) scale, the Global Assessment of Functioning (GAF), the Nurses’ Observation Scale for Inpatients Evaluation (NOSIE), and the Positive and Negative Syndrome Scale (PANSS) of schizophrenic patients treated with low (<6 mg/day) or high (6 mg/day) doses of risperidone

Low-dose group (n=98)

High dose group(n=48)




Baseline score




Percent change at endpoint





Baseline score

36.1 ±8.6

37.9 ±6.4


Percent change at endpoint

36.8 ±61.3




Baseline score




Percent change at endpoint




Positive PANSS

Baseline score




Percent change at endpoint




Negative PANSS

Baseline score




Percent change at endpoint



0.120 a

General PANSS

Baseline score




Percent change at endpoint





Baseline score




Percent change at endpoint




Responders (≥20% improvement in total PANSS score) (n, %)

63/98, 64.3

21/48, 43.8


aTwo-tailed Mann-Whitney U-tests

bTwo-tailed χ2 tests

As for drug-efficacy comparison, a statistically significantly higher portion of low-dose patients showed response (≥20% reduction in the PANSS total score) at endpoint (Table 2). Low-dose individuals were also significantly superior in percentage changes in the PANSS total and general-subscale scores from baseline to endpoint. In addition, the low-dose group tended to excel (albeit statistically insignificantly) in terms of percentage changes in the PANSS positive and negative subscales, the CGI, the GAF, and the NOSIE scores (Table 2). Overall, compared to the high-dose counterparts, the low-dose participants performed at least equally in the efficacy measurements.

Determinants of risperidone doses

For evaluating the effects of patients’ background characteristics on risperidone dosage, the GEE method’s multiple linear regression model was utilized. Since the two dose groups varied in baseline NOSIE performance (Table 2), this variable, as well as patients’ demographic data, was used for the potential prognostic factors. Table 3 shows that diagnosis subtypes could influence risperidone doses even after controlling for treatment duration and other patient-related variables. Compared to patients with undifferentiated subtype, those with disorganized subtype received higher dosage by 0.90 mg/day on average (P=0.008). Paranoid and undifferentiated subtypes were comparable in drug doses. Patients with longer duration of illness also showed a trend to use higher dosage; each month increment in the duration predicted a 0.0082-mg increase in the daily dose (P=0.078). Risperidone treatment duration, gender, age (between 18 and 60 years old), body weight, age at illness onset, the number and duration of prior hospitalization, and the baseline NOSIE score, however, did not significantly affect the doses after adjusting the effects of other variables (Table 3).
Table 3

Effects of risperidone treatment duration, patient-related variables, and baseline Nurses’ Observation Scale for Inpatients Evaluation (NOSIE) score on risperidone doses: multiple linear regression analysis with the generalized estimating equation method under the SAS/STAT V8.1 system


Estimated coefficient

Standard error of the estimated coefficient


Treatment duration(week)




Male gender, compared to female




Age (years)




Body weight (kg)




Age at onset of psychosis (years)




Duration of illness (months)




Number of prior hospitalizations




Duration of prior hospitalizations (weeks)




Paranoid type, compared to undifferentiated type




Disorganized type, compared to undifferentiated type




Baseline NOSIE score





The results of this study lend support to our hypothesis that dose reduction for curtailing risperidone-related side effects still yields favorable efficacy in acutely exacerbated schizophrenia patients. Two-thirds of our subjects could not tolerate the target dose of 6 mg/day and received lower doses (mean±SD=3.4±0.9 mg/day). Interestingly, the low-dose patients, in comparison with the high-dose (6-mg/day) recipients, improved more in the PANSS total and general-subscale scores (Table 2). The low-dose participants also showed a trend (without statistical significance) to surpass the high-dose individuals in other efficacy assessments. Nonetheless, multiple comparison false positives still should be taken into accounts. In addition, caution should be exerted when interpreting the findings because the two dose groups differed in background characteristics: the high-dose group had more disorganized-subtype patients and higher baseline NOSIE scores (and, consequently, inferior social and self-care functioning). If the two groups had been similar prior to treatment, whether the low-dose group would still excel in treatment response remains unknown. Studies using random-assignment to fixed doses of risperidone are required for the purpose. Nonetheless, our study design and findings may provide a safe, efficacious, and feasible risperidone dosing strategy for acutely symptomatic patients in “real world” clinical settings.

Another main finding of the present study is that, after adjusting other variables, schizophrenia patients with disorganized subtype could tolerate higher risperidone dosage by around 0.9 mg/day on average than those with paranoid subtype or undifferentiated subtype (usually with mixed paranoid and disorganized manifestations in our subjects). Paranoid symptoms have been found to be associated with higher plasma prolactin levels in neuroleptic-treated schizophrenics (Csernansky et al. 1986). Investigators (Seeman and Lang 1990; Szymanski et al. 1995) have also assumed that individuals with higher prolactin levels may have poorer tolerance to dopamine antagonists. It is thus theoretically possible that individuals with paranoid or undifferentiated subtype (hence with paranoid symptoms and possibly higher prolactin levels) could sustain only lower dosage of risperidone. Further studies are warranted to elucidate the role of prolactin levels in determining the tolerance to risperidone or other antipsychotics (particularly those with moderate or high dopamine blockade activity). In an earlier retrospective study (Young and Meltzer 1980), paranoid symptoms did not significantly influence the doses of conventional antipsychotics in 61 patients. The dosing tactic, perhaps not exclusively based upon side-effect diminution, may be different from that of the present study.

Previous studies (Tsuang and Winokor 1974; Zalewski et al. 1998) also indicated that patients with disorganized (formerly hebephrenic) subtype had earlier age of onset, more motor symptoms, inappropriate or flat affect, formal thought disorder, more disruption of familial and social relationships, and perhaps poorer cognitive functioning than paranoid-subtype patients. Etiopathogenesis (Stober et al. 1999) and neurophysiological performances (Bruder et al. 2001) may also differ between the subtypes. In a similar manner, our high-dose group (with more disorganized patients) showed poorer baseline social and personal-care function than the low-dose group.

In our small pilot study (Lane et al. 2000), patients who attained only lower doses of risperidone had fewer prior hospitalizations. In the present larger-sized study, number and duration of prior hospitalizations, however, did not significantly impact the dosing, especially after controlling for other variables. Instead, longer illness duration tended to raise the dose tolerability (P=0.078) (Table 3). More subjects are needed to confirm this issue. Similarly, when taking classical antipsychotics, multi-episode patients are less likely to experience side effects than first-episode individuals (Chakos et al. 1992).

Regarding gender effects, male patients show poorer response to classical antipsychotics than females (Szymanski et al. 1995) and require higher dosage when the doses are adjusted according to therapeutic efficiency (Melkersson et al. 2001). In the current study, gender effects on risperidone dosage were insignificant with the dosing strategies based on safety. As for clozapine, the prototype of atypical antipsychotics, males may tolerate larger doses because they have more rapid metabolism for clozapine (Lane et al. 1999).

The limitations of this study were obvious. First, this was an open trial; further double-blind studies are necessary. Second, whether the results could be applied to populations other than acutely ill patients remains unknown. Third, in comparison to white individuals, Chinese have been reported to have one-third higher plasma levels of risperidone active moiety (risperidone plus 9-hydroxyrisperidone) (Lane et al. 2000). Therefore, a portion of white patients may require higher dosage such as 8 mg/day or so. The present study also had several strengths. It enrolled a large cohort of consecutively hospitalized patients and applied novel statistics (GEE methods) to control for other potential variables and the effects of repeated measurements. In addition, owing to our dosing policy, drug side effects were minimal (if any) and thus unlikely to interfere with efficacy assessments.

In summary, this study found that two-thirds of Han Chinese patients with acutely exacerbated schizophrenia did not tolerate a 6-mg daily dose of risperidone. Adjusting dosage to 3.4±0.9 mg to relieve side effects yielded at least equal efficacy in comparison with those tolerant of the 6-mg regimen. Patients with disorganized-subtype and perhaps patients with longer illness duration were more likely to receive higher doses. That is, in general, one would expect higher doses to be needed in a more chronic, non-responsive population. Further rigorously designed studies in chronic patients and in other ethnic populations are needed.


This work was supported by grants NSC 91-2314-B-320-005 and NSC 92-2314-B-039-017 from the National Science Council; NHRI-EX-92–9134PI from the National Health Research Institutes; and CMU92-TH-03 form the China Medical University, Taiwan.

Copyright information

© Springer-Verlag 2003