The ethics of medication-free research protocols in adults with schizophrenia have been increasingly examined. In adult studies, the diagnosis of patients who have been chronically ill for years is not equivocal, and there have been objections to medication withdrawal periods that have not been adequately supervised and lasted longer than a few weeks or both (1). Potential risks associated with medication-free periods include the prolongation and reemergence of psychosis, the risk of injury to self or others, emotional suffering, and the increased burden on caregivers (1, 2). In addition, it has been proposed that rapid withdrawal of maintenance antipsychotic medications during research protocols may represent an iatrogenic, time-limited pharmacologic stress factor (2) and that interruptions in treatment and/or sustained nontreatment may have deleterious consequences (3). The decision-making capacity of adult patients with schizophrenia to understand these issues and give initial and continuing informed consent has been appropriately questioned, and this has led to ethical concerns regarding the inclusion of these patients in research involving medication withdrawal (4). Given these concerns, the strongest support for medication-free research may be with studies of patients with treatment-refractory schizophrenia who have already been responding poorly to treatment and for whom improvement is as likely as further deterioration (5).
Children and adolescents are recognized as a vulnerable population, and special ethical and regulatory considerations are in place for reviewing research involving these subjects (6). By virtue of their ongoing development, pediatric subjects may differ from adults in their pharmacokinetics, clinical response, and adverse reactions to psychotropic medications (7–11). Thus, there is an ethical imperative to conduct research into childhood psychiatric disorders (12). However, from the vantage point of protecting pediatric research subjects, the potential risks of research should be offset by significant direct benefits to the individual children participating (6, 13).
Controlled treatment trials of atypical neuroleptic drugs that have been shown to have superior efficacy compared to standard neuroleptics in adults (14) and neurobiological studies are a major part of the childhood-onset schizophrenia project at the National Institute of Mental Health (NIMH). For this research, children and adolescents diagnosed with childhood-onset schizophrenia have undergone medication-free periods lasting up to 4 weeks. This report reviews the diagnostic and treatment benefits and adverse events of a medication-free lead-in period in pediatric patients with childhood-onset schizophrenia who have participated in studies at NIMH during the past 5 years. The data reported herein are intended to aid the development of humane research protocols involving children and adolescents with psychotic disorders.
Recruitment and diagnostic procedures are described in detail elsewhere (14–17). Subjects who had been diagnosed with DSM-III-R or DSM-IV criteria for schizophrenia were sought by means of national recruitment through professional and patient advocacy groups for an inpatient study involving treatment trials of atypical and typical neuroleptics. Inclusion criteria included at least two unsuccessful previous neuroleptic trials, onset of psychotic symptoms by age 12, a premorbid IQ of at least 70, and the absence of any significant medical or neurological disorders.
After a 2-week observation period during which they received admission medications, all patients had their medications tapered over 1 to 2 weeks, followed by a 2- to 28-day drug-free interval, as clinically tolerated, before randomization into the trial.
Between April 1993 and March 1998, 34 (29%) of 119 patients screened in person were diagnosed with definite or probable childhood-onset schizophrenia. All of the patients included in this study were seen by one of two psychiatrists (S.K. and L.K.J.) and the head nurse of the unit (C.B.). The interrater reliability, based on structured and clinical interviews for childhood-onset schizophrenia, was good (kappa=0.77) (17). Thirty-one of these patients (and their parents) agreed to participate in biological research or medication trials that involved a drug-free interval. Three patients already responding well to clozapine declined to participate.
All treatment protocols were approved by the Institutional Review Board of NIMH, and each child or parent gave his or her respective written assent or permission for each study. The charts of these 31 patients were reviewed (S.K. and C.B.), and information regarding admission and discharge diagnoses, length of drug-free interval, and (during the medication-free period) significant adverse events was gathered.
Standardized ratings included the Scale for the Assessment of Positive Symptoms (18), the Scale for the Assessment of Negative Symptoms (19), the Subjective Treatment Emergent Symptoms Scale (20), and the Children’s Global Assessment Scale (21).
All patients with a discharge diagnosis of childhood-onset schizophrenia (N=24) have been contacted by telephone (M.L.) at regular intervals since discharge from the study. In-person follow-up was obtained for 15 (88%) of 17 children eligible for follow-up (those who had been discharged from NIMH for at least 2 years). Follow-up information is based on data collected at either 2-year (N=9) or 4-year (N=6) assessments by using clinical interviews, selected portions of structured interviews (22), and all available clinical and school records. Outcome criteria based on the Children’s Global Assessment Scale (21) were adapted from a naturalistic follow-up study of 18 outpatients with childhood-onset schizophrenia conducted at the University of California, Los Angeles (UCLA), Neuropsychiatric Institute (23) who had a longer follow-up interval than those in the present study (t=3.5, df=24, p=0.0005). As suggested by Asarnow et al. (23), children were classified into the following four categories on the basis of their Children’s Global Assessment Scale scores at the end of their follow-up intervals: 1) deteriorating (scale scores were lower at the end of follow-up than at initial entry into the project and fell below 40 at final follow-up), 2) minimal improvement (final scale scores between 30 and 50), 3) moderate improvement (final scale scores between 50 and 60), and 4) good outcome (final scale scores of 60 or above).
Between April 1993 and March 1998, 31 patients who were rigorously diagnosed as having definite or probable childhood-onset schizophrenia were admitted to the child psychiatry research unit at NIMH. At the completion of a medication-free interval of 2 to 4 weeks, seven (23%) of the patients were diagnosed as having a disorder other than childhood-onset schizophrenia for diagnostic reasons and because schizophrenic symptoms were not confirmed. The final diagnoses of these patients were posttraumatic stress disorder (N=1), a subtype of atypical psychosis provisionally labeled by us as "multidimensionally impaired" (N=4), and personality disorder (N=2). At the 2- to 4-year follow-up, three of seven patients remained free of neuroleptic therapy.
All of the patients with childhood-onset schizophrenia enrolled in this protocol were felt to be treatment refractory. However, because eight (26%) of the patients experienced rapid and severe deterioration of their schizophrenic symptoms and increased aggression, the drug washout period was curtailed and antipsychotic treatment was initiated.
At the 2- to 4-year clinical follow-up at NIMH, for 15 patients who had a final diagnosis of childhood-onset schizophrenia, there was no evidence of harm from the medication-free interval as judged by clinical symptoms and overall functioning. According to the Asarnow et al. (23) outcome criteria, four patients (27%) were classified as deteriorating, seven (47%) as minimally improved, one (7%) as moderately improved, and three (20%) with a good outcome. These outcome data are not significantly different from those of the 18 UCLA patients with childhood-onset schizophrenia (23) who did not undergo a drug washout (p=0.09, Fisher’s exact test). The 18 UCLA patients were classified as follows: three (17%) deteriorating, five (28%) minimally improved, five (28%) moderately improved, and five (28%) with a good outcome.
At follow-up, 14 of 15 NIMH subjects with childhood-onset schizophrenia were being treated with an antipsychotic medication. Ten of the 14 patients remained on a regimen of clozapine. Only one patient was in clinical remission and drug free.
To our knowledge, this is the first report to examine the potential benefit-to-risk ratio for children and adolescents diagnosed with childhood-onset schizophrenia who have participated in a drug discontinuation study. An ethical question raised by these data is whether the short-term suffering of the eight patients (26%) whose washout was abruptly terminated was offset by the rediagnosis of seven patients (23%) who, as a result of the washout, were found to have a probable wrong diagnosis (although some were indeed psychotic, and not all remained neuroleptic free thereafter) and who were saved the potential risks of a trial of clozapine.
Overall, given that the treatment conditions were maintained as basically safe and that the subjects selected were children and adolescents with a poor response to treatment and with a potentially lifelong illness and chronic disability, it could be argued that the benefits of improved diagnosis and treatment compensated for the risks of the medication-free period and nontherapeutic studies for this group. However, the small group size, the variance in the developmental range of the subjects, and the unique qualities of this study limit the generalizability of these findings.
Potential clinical benefits of participation in drug withdrawal protocols include a lower risk for adverse drug interactions, the disentanglement of adverse behavioral effects of medication from manifestations of disease, the identification of early signs of tardive dyskinesia that might be masked by antipsychotic drugs and clarified by their removal, more accurate assessment of treatment effects, and improved diagnostic assessment (1, 24, 25). It should be acknowledged that it is possible that for the seven patients we rediagnosed, the screening diagnosis of schizophrenia was, in fact, correct and the comorbid conditions of these patients became more prominent after their psychosis was in remission. Consistent with this hypothesis, 40% of adults diagnosed as having schizophrenia were found to remain stable without medication for several years after abrupt removal of oral neuroleptic treatment (2).
On the other hand, several concerns regarding the scientific rationale of a drug-free lead-in period (24, 26) and the rapid rate of drug discontinuation in research protocols have been raised (2). An incomplete washout, the iatrogenically provoked clinical worsening of schizophrenic symptoms, and withdrawal syndromes all limit the value of a putative "drug-free" study design (2, 26). Currently, the optimal length of an antipsychotic-free lead-in period remains unclear, since the elimination half-life of antipsychotics may be several weeks or more, thus making the attainment of a physiological and pharmacokinetic drug-free status impractical (2, 26).
A related matter is the rate of drug withdrawal. Slower (safer) discontinuation protocols are costly, although probably desirable clinically, and result in longer inpatient stays than abrupt discontinuation protocols. Slow discontinuation of some psychotropic drugs (e.g., lithium carbonate, sedative-anxiolytics, and anticonvulsants) is clearly important; however, the data on this point remain inconclusive for antipsychotics (2). Approximately one-quarter of the patients with childhood-onset schizophrenia in this study became symptomatic when not taking medication. It is possible that although this group was "treatment refractory," some patients were receiving a benefit from their antipsychotic medications, and/or these patients experienced a rebound psychosis or withdrawal syndrome from abrupt withdrawal of psychotropic medications (2, 27).
The question of whether additional clinical improvement might have occurred for the patients with childhood-onset schizophrenia who participated in this study had they not undergone a drug washout period cannot be answered. However, all of the patients with childhood-onset schizophrenia who were able to complete a 6- to 8-week atypical neuroleptic trial were rated at least minimally improved compared to their admission condition. Preliminary evidence from controlled trials of adult schizophrenic patients indicates that brief, closely monitored, drug-free periods do not contribute significantly to a poor long-term clinical course (1, 28, 29).
It is of interest that the clinical outcome of the NIMH group at 2- to 4-year follow-up is similar to that of a clinically referred study group of a less severely ill (not all patients were treatment refractory) cohort of patients with childhood-onset schizophrenia studied at the UCLA Neuropsychiatric Institute who did not undergo a drug washout (23). It should be stressed that a comparison between these two studies is tentative at best because of the small group size of both studies and the confounding differences between patient populations and their treatments during the follow-up interval (i.e., the UCLA patients with childhood-onset schizophrenia were diagnosed with DSM-III criteria and were not treated with atypical neuroleptics). To more adequately evaluate the effects of drug discontinuation, a future controlled trial of patients with early-onset schizophrenia might involve first converting subjects to a standardized treatment (one agent and a limited milligram-per-kilogram daily dose range), restabilizing them, removing the drug at more than one rate with graded placebo substitution, and then following up with a standardized retreatment as well as systematic short-term follow-up (particularly at 1, 3, 6, and 12 months).
On the basis of our experience, we advise institutional review boards to consider the following issues when examining research protocols that involve drug discontinuation in children and adolescents with schizophrenia. First, criteria should be specified in the protocol for the termination of the drug washout period with patients who experience a worsening of their symptoms. In contrast to research involving adult schizophrenic patients, where there is less assurance of adequate decision-making capacity to give informed consent, pediatric patients have guardians who are capable of giving informed consent. However, to ensure that guardians (and study participants) fully understand the risks associated with drug discontinuation, they should be referred to the child’s physician to discuss the risks and benefits involved in research participation before admission. In addition, before the drug washout period, education for patients and their guardians should be provided to help prepare them for this procedure.
It is debatable whether children and adolescents with a history of aggression or self-injurious behavior and psychotic symptoms should be included in drug-free research. If it is necessary to involve these patients, then additional safeguards are indicated, such as physical restriction of these patients to hospital grounds, additional structure to their daily routine, a tight behavior modification and reward system, a physical environment that is free of dangerous items, and adjunctive benzodiazepines, as needed. For nursing and physician staff, routine meetings to formulate treatment plans and the involvement of outside consultants are helpful for children whose continued participation in research poses complex ethical dilemmas.
In our experience, a short drug-free evaluation for diagnostically complex children with psychotic symptoms on an inpatient unit may be beneficial to more fully understand the factors that contribute to a child’s treatment-refractory illness. The important information gained from these diagnostic evaluations justifies the need for clinicians who may not necessarily have the benefit of adequate information before admission to have the opportunity to observe severely disturbed children and adolescents without medication.
Received April 30, 1998; revisions received Nov. 16, 1998, and Jan. 12, 1999; accepted Jan. 13, 1999. From the Child Psychiatry Branch, NIMH; and Yale University School of Medicine, New Haven, Conn. Address reprint requests to Dr. Rapoport, Child Psychiatry Branch, NIMH, Bldg. 10, Rm. 3N202, 10 Center Dr., MSC 1600, Bethesda, MD 20892-1600, firstname.lastname@example.org (e-mail). The authors thank NIMH Clinical Center staff, the clinicians who referred patients to this study, and Drs. F. Miller, A. Wichman, R. Wyatt, J. Volavka, R. Baldessarini, S. Frangou, D. Rosenstein, E. Arnold, and F.X. Castellanos for their suggestions.