NADD Bulletin Volume II Number 2 Article 2

Complete listing

Antipsychotic Drugs in Mentally Retarded Persons: A Critical Review

Arthur Rifkin, M.D.


This article reviews all published studies comparing an antipsychotic (AP) to placebo in persons with mental retardation. APs are the most commonly used psychotropic drug in this population, and very large numbers of such persons take psychotropics; according to one survey 57% in institutions, 41% in community based programs, and 22% in school-based settings (Baumeister, Todd et al. 1993) This review, will focus on the use of APs for disruptive behavior, by far the most frequent indication, and not review more specific indications, such as Autistic Disorder. The intent is to present a capsule description of each study, rather than compress the data into a table, because the studies differ greatly, and the quality of most is less than adequate, which is discussed for each example. The reader should form an opinion based on an assessment of the quality of each study. If more studies had adequate methodology, perhaps a meta-analytic approach would have been preferable. Only statistically significant differences are reported in the article.


D’Hollander et al (1967) in 46 institutionalized females, ages 5-17 years, compared dixyrazine (an AP never marketed in this country), chlorpromazine (CPZ)(12-125 mg/d) and placebo, each given for 4 weeks in a cross-over design. All were chosen because of behavior problems. Dixyrazine and CPZ were superior to placebo on measures of overactivity, loquacity, disobedience, aggression, and irritability. However, the authors did not analyze for order or period effects, a necessity in cross-over studies, ie, before combining the results of each drug, whether given first, second, or third, you should analyze whether the sequence (eg, CPZ before or after placebo) or period (eg, 2nd vs 3rd) showed differences. Only if they do not, should the authors combine the results. The absence of such an analysis makes these data uninterpretable. They used low doses of CPZ making it difficult to demonstrate efficacy.

Lindholm(1967) also compared dixyrazine to placebo. They studied 35 subjects selected for symptoms of restlessness, poor concentration, aggressiveness, irritability and nervousness. The drug showed superiority to placebo on several measures, but their statistical analysis and outcome measures were not clearly presented, and they used a 10-week cross-over design without analyzing for an order effect, making interpretation of results unclear.

Weir et al(1968) did two 12-week cross-over studies; one comparing pericyazine (N=22)(an AP not marketed in this country), the other CPZ (N=21) to placebo, in subjects described as behavior problems. They reported no differences between either drug and placebo, but these data, also, are uninterpretable because they did not present confidence intervals or discuss power, so we cannot know how to assess a finding of no difference. All future studies that report no difference between active drug and placebo failed to do this as well. As before, they did not analyze for the order of cross-over.

Alexandris and Lundell (1968) treating children with hyperactivity and aggression, assigned 7 subjects to thioridazine (30-150 mg/d), 6 to amphetamine (7.5-75 mg/d) and 8 to placebo. The study lasted 6 months. They did not do statistical analyses but report thioridazine superior to placebo on 11 of 14 measures, such as hyperkinesis, attention, and aggressiveness. The lack of statistical analysis makes this study uninterpretable.

Burk and Menolascino (1968) compared haloperidol to placebo in 50 institutionalized subjects with moderate retardation with “secondary emotional symptoms”, who were 5 to 21 years old, for 8 weeks. Their only outcome measure was a global scale based on observations at the end of 4 and 8 weeks, “on the ward or in the school, and during a psychiatric examination”. They report one statistical analysis, a significant chi square of the global outcome, favoring haloperidol. This early study lacks adequate methodology to justify any conclusions, such as no report of the symptoms, the dose, or using validated outcome measures with adequate interrater reliability.

LeVann(1970) studied 59 hospitalized children. They assessed symptoms of manageability, hostility toward self, senseless overactivity, overall behavior, appetite, and sleep, comparing chlorprothixene to thioridazine in a cross-over design with a 2-week placebo period between drugs. This is not a true placebo-controlled trial, and there was no analysis of an order effect. This study, as well, is uninterpretable.

Goldberg and Kurland(1974), in 30 male hospitalized adolescents, compared pimozide (mean=6.5 mg/d) to placebo. The were chosen for “anxiety, uncooperativeness, hyperactivity, inattentiveness and irritability”, and treated them for 12 weeks. They found no difference, without presenting confidence intervals or a discussion or power to detect a clinically important difference.

Väisänen et al(1975) did a triple cross-over study with sulpiride (an AP not marketed in this country), CPZ (75-150 mg/d) and placebo, each period lasting 4 weeks, in 60 institutionalized subjects diagnosed as “restless”. Their only finding was that sulpiride was better than placebo for a global rating. They did not analyze for order, making their study uninterpretable.

McConahey et al (1976) studied 22 institutionalized women, with an age range of 20 to 65 years, who had moderate to profound retardation, comparing chlorpromazine ( mean dose = 88 mg/d) to placebo using an ABAB crossover design, with each period lasting 4 weeks. Using a 23 item behavior scale, they found no significant difference on any item. The low dose used, as well as the absence of an analysis for an order effect, makes it difficult to interpret this study.

Elie et al (1980) compared an investigational drug to thioridazine and placebo in 51 inpatient adult subjects for 4 weeks, who displayed aggressive behavior. At three time periods, with 2 rating measures, there was a trend level finding (p<.1) for thioridazine to be worse than placebo.

Singh and Aman (1981) in 19 severely retarded institutionalized patients, with IQs below 31, who had received thioridazine for at least 6 months for aggressiveness, were randomized to 3 treatments, given in a crossover design: an individualized dose of thioridazine found best, according to clinical judgment during the pre-study period; a fixed dose of thioridazine of 2.5 mg/kg/day; and placebo. The subjects had a mean age of 16 years (SD=4.06). On 10 outcome measures, one (hyperactivity) showed thioridazine superior to placebo. This possibly chance finding, together with the absence of an analysis for an order effect, makes interpretation of results problematic.

Lynch et al(1985) compared a long-acting depot antipsychotic drug, pipothiazine palmitate, to placebo in 28 adult subjects whose aggressiveness made management difficult. They received each drug for 13 weeks in a crossover design. The investigators report that active drug superior to placebo on a global scale and on a target symptom scale of aggressiveness, but they did not publish any statistical analysis, precluding considering this study adequate.

Heistad et al (1979) in 100 institutionalized subjects selected because they received thioridazine (mean dose=235, range 30-800 mg/d) were randomized to a cross-over study, duration 4 weeks for each period comparing, thioridazine with placebo. Thioridazine proved superior on the global rating, and, in a subset of 27 subjects, on a scale of daily monitoring of symptoms. Here, too, they did not analyze the cross-over data for an order effect, so we cannot adequately interpret this data.

White and Aman (1985), in only 8 subjects (mean age=15.7 years, SD=3.42), chosen for symptoms of aggression, hyperactivity and stereotypic behavior, compared pimozide (6 mg/d) to placebo, in a 4-week cross-over design. On 12 measures of behavior, on 1 (hyperactivity) showed a difference, possibly a chance event. Unlike previous cross-over studies, these authors did analyze for an order effect and found one symptom (irritability) did show an effect of order. Given the small sample size, they could have detected only large effects.

Aman et al (1989) studied 20 moderate to profoundly retarded institutionalized subjects in a crossover study with three 3-week periods of haloperidol .025 mg/kg/d, .05 mg/kg/d, and placebo. All the subjects had problem behaviors of aggression, screaming, stereotypies or bizarre behavior. Their significant findings were limited to improved stereotypies with haloperidol on one scale, but not on another; greater motor activity on the higher dose of haloperidol (which might have been akathisia) but for this item they had a significant order effect (making it impossible to ascribe the change to the effect of haloperidol); and less inactivity on haloperidol.

Singh and Owino (1992) did a placebo-controlled discontinuation study of zuclopenithixol (a drug not marketed in this country) in 43 adult subjects with “behavioral disorders”. At the endpoint, one of the 2 global items “Improvement” showed a significant effect for the active drug, while the other global item, “Severity” did not. On a behavior rating scale the 2 groups showed a significant difference at endpoint, and at the 12th week, favoring the active drug.

Vanden Borre et al(1993) compared risperidone (4 - 12 mg/d) versus placebo in a crossover study (3 weeks for each period) in 37 subjects (15 to 65 years old) in 6 Belgian institutions who had “persistent behavioral disturbances”. They continued to receive previously prescribed medication, usually antipsychotics and benzodiazepines. On a global rating risperidone was significantly superior to placebo, but there was a significant effect of the order they received the drugs which would make the effect of treatment uninterpretable. On a rating scale of behavioral symptom risperidone did significantly better but there appears to be a large order effect. The investigators did not report a statistical analysis of order for this measure. A visual analogue score for the chosen most important target symptoms showed no significant treatment effect. On weekly measures of global improvement, risperidone showed a significant difference from placebo. The investigators did not analyze their results by site, although they had six different sites for only 37 subjects, which would make it difficult to detect any but very large differences among sites.

Aman and White (1988) compared 2 doses of thioridazine (1.25 and 2.5 mg/kg/d) in 11 subjects who had received thioridazine. At the conclusion of the comparison all received placebo. They found less hyperactivity and self-injurious behavior on the higher dose compared to the lower, excluding the placebo group. Their other statistical analysis, that included the final placebo period, was, as they acknowledged, not a true controlled trial because the order of placebo was not varied. This study, then does not provide any clear data on the comparison of an AP to placebo.

DeSmedt and Van Bellinghen (1998) compared risperidone (.05 mg/kg/d, range=.03-.06) to placebo in 13 subjects (6-18 years) for persistent behavioral disturbances. They did not exclude patients with axis I disorders. Risperidone was superior to placebo on 1) irritation, 2) hyperactivity, 3) appropriate speech, 4) the visual analogue scale for the target symptom, and 5) on 4 subscales (adaptation, peers’ attitude, social relationship and occupational attitudes on a rating scale. Finding differences in such a small sample indicates a large drug effect, but, this, the most methodologically adequate study, is not a pivotal study because of its small sample.

Discussion and Summary

This comprehensive critical review of APs in mentally retarded subjects demonstrates a clear conclusion: We do not know if they work. Most studies show negative results and/or gross methodological problems. Only the most recent data, DeSmedt and Van Bellinghen(1998), not yet published, presents convincing evidence of a drug effect in a small sample of 13 subjects. Surely, this represents an embarrassing position. The most commonly used drug for behavior problems in mentally retarded patients, a drug used to a great extent and which has, at times, serious side effects, lacks validity. The incidence of mental retardation equals that of schizophrenia, yet the drug research data base shows great disparity of effort to validate the use of APs.

The only justification for continued use of APs for disruptive behavior in mental retardation is the absence of adequate evidence for any other drug. Absent from this discussion has been the importance of valid diagnoses. The uncertainty of valid diagnostic distinctions in those showing disruptive behavior makes a bad situation worse. We are very far from achieving an adequate data base of the efficacy of drugs for validated diagnostic states in mentally retarded persons.


Alexandris, A. and F. W. Lundell (1968). “Effect of thioridazine, amphetamine and placebo on the hyperkinetic syndrome and cognitive area in mentally deficient children.” Canadian Medical Association Journal 98: 92-96.

Aman, M. G., C. J. Teehan, A. J. White, S. H. Turbott and C. Vaithianathan (1989). “Haloperidol treatment with chronically medicated residents: dose effects on clinical behavior and reinforcement contingencies.” American Journal of Mental Retardation 93: 452-460.

Aman, M. G. and A. J. White (1988). “Thioridazine dose effects with reference to steriotypic behavior in mentally retarded residents.” Journal of Autism and Developmental Disorders 18: 355-366.

Baumeister, A. A., M. E. Todd and J. A. Sevin (1993). “Efficacy and specificity of pharmacological therapies for behavioral disorders in persons with mental retardation.” Clinical Neuropharmacology 4: 271-294.

Burk, H. W. and F. J. Menolascino (1968). “Haloperidol in emotionally disturbed mentally retarded individuals.” American Journal of Psychiatry 124: 1589-1591.

DeSmedt, G. and M. Van Bellinghen (1998). Risperidone for treating behavioral disturbances in mentally retarded children and adolescents: A double-blind placebo-controlled trial. American College of Neuropsychopharmacology, Las Croabas, Puerto Rico.

d’Hollander, L., P. J. Fontaine and T. Stassart-Deryck (1967). “Evaluation clinique de l’action de la dixyrazine sur les troubles du comportement chez des enfants traites en internat medico-pedagogique.” Acat Neurol Psychiatr Belg 67[suppl]: 5-35.

Elie, R., Y. Langlois, S. F. Cooper, G. Gravel and J.-M. Albert (1980). “Comparison of SCH-1279 and thioridazine in aggressive mental retardates.” Canadian Journal of Psychiatry 25: 484-491.

Goldberg, J. and A. Kurland (1974). “Pimozide in the treatment of behavioral disorders of hospitalized adolescents.” J Clin Pharmacol 14(2): 134-139.

Heistad, G. T. and R. L. Zimmermann (1979). “Double-blind assessment of Mellaril in a mentally retarded population using detailed evaluations.” Psychopharmacology Bulletin 15: 86-88.

LeVann, L. J. (1970). “Clinical experience with Tarasan and thioridazine in mentally retarded children: A comparative double blind study.” Applied Therapeutics 12: 30-33.

Lindholm, O. (1967). “A clinical study of Esucos in a mentally retarded clientele.” International Journal of Neuropsychiatry 3: 209-218.

Lynch, D. M., C. L. S. Eliatamby and A. A. Anderson (1985). “Pipothiazine palmitate in the management of aggressive mentally handicapped patients.” British Journal of Psychiatry 146: 525-529.

McConahey, O. L., T. Thompson and R. Zimmermann (1976). A token system for retarded women: behavior therapy, drug administration, and their combination. Behavior Modification of the Mentally Retarded. T. Thompson and J. G. Grabowski. New York, Oxford University Press: 167-234.

Singh, I. and W. J. E. Owino (1992). “A double-blind comparison of zuclopenithixol tablets with placebo in the treatment of mentally handicapped in-patients with associated behavioural disorders.” Journal of Intellectual Disability Research 36: 541-549.

Singh, N. N. and M. G. Aman (1981). “Effects of thioridazine dosage on the behavior of severely mentally retarded persons.” American Journal of Mental Deficiency 85: 580-587.

Vaisanen, K., P. Kainulainen, M. T. Paavilainen and M. Viukari (1975). “Sulpiride versus chlorpromazine and placebo in the treatment of restless mentally subnormal patients - a double-blind cross-over study.” Current Therapeutic Research 17: 202-205.

Vanden Borre, R., R. Vermote, M. Buttiens, P. Thiry, G. Dierick, J. Geutjens, G. Sieben and S. Heylen (1993). “Risperidone as add-on therapy in behavioural disturbances in mental retardation: a double-blind placebo-controlled cross-over study.” Acta Psychiatrica Scandinavica 87: 167-171.

Weir, T. W. H., G. A. Kernohan and D. N. MacKay (1968). “The use of pericyazine and chlorpromazine with disturbed mentally subnormal patients.” British Journal of Psychiatry 114: 111-112.

White, T. J. R. and M. G. Aman (1985). “Pimozide treatment in disruptive severely retarded patients.” Australian and New Zealand Journal of Psychiatry 19: 92-94.

For further information contact:

Arthur Rifkin, M.D.
Dept. of Psychiatry
Hillside Hosptial
Glen Oaks, NY 11004