Schizophrenia Research Forum - A Catalyst for Creative Thinking

Antipsychotic Maintenance Treatment: Must It Be Forever?

30 May 2007. Antipsychotic drugs are highly effective treatments for the positive symptoms of schizophrenia, but the serious and sometimes permanent side effects of these medications make successful long-term maintenance treatment challenging for both patient and clinician. Although newer “atypical” medications such as risperidone appear to present less risk of tardive dyskinesia and other extrapyramidal symptoms than first-generation drugs, second-generation agents have troublesome side effect profiles of their own, including disturbances in glucose metabolism, weight gain, and a heightened risk of diabetes (see American Diabetes Association et al., 2004; Haddad, 2004).

These side effects have led to the development of low-dose maintenance treatment protocols, but they have also prompted some researchers to ask whether there may be subgroups of people with schizophrenia who can eventually do well without taking any antipsychotic medications—and if so, whether this population has distinctive characteristics that would make these individuals easier to identify. The ethics of placebo-controlled studies of schizophrenia treatment remain highly controversial (see SRF related news story), but three recent prospective studies used other research designs to address these questions.

Who Succeeds Without Antipsychotics?
It is well-known that many people with schizophrenia stop taking antipsychotic medications after acute hospital treatment (see Fenton and McGlashan, 1987; Lieberman et al., 2005; also see SRF related news story). Martin Harrow and Thomas H. Jobe of the University of Illinois College of Medicine undertook a prospective 15-year follow-up study of 145 psychosis patients, including 64 who eventually received a diagnosis of schizophrenia, to see what types of schizophrenia patients discontinued medications—whether on their own initiative or with the guidance of a physician—and how they fared over the long term.

In particular, Harrow and Jobe were interested in whether patients’ past developmental achievements and their prognostic potential, as assessed during the index hospitalization, were predictive of their success off medication. In addition, the researchers used locus-of-control (LOC) and self-esteem scales at 4.5-year follow-ups to determine whether these personality measures correlated with patients’ medication status and well-being in later years.

As reported in the May 2007 issue of the Journal of Nervous and Mental Disease, Harrow and Jobe found in follow-up assessments at 2, 4.5, 7.5, 10, and 15 years that up to 41 percent of the patients with schizophrenia in the study sample had stopped taking antipsychotic medications, and as many as 25 percent were receiving no mental health treatment whatsoever. However, from the 4.5-year mark to the conclusion of the study, the patients with schizophrenia not taking antipsychotic drugs scored significantly better than those still taking medications on measures of global functioning and adjustment (p <.001). At 15 years, 64 percent of those on antipsychotic medications had psychotic activity, versus 28 percent of the group off medications. Moreover, 83 percent of the subjects with schizophrenia (19 of 23 patients) with a uniformly poor outcome at the 15-year follow-ups were on antipsychotic medications.

Harrow and Jobe found a strong correlation between the 4.5-year LOC and self-esteem scores and the likelihood that a patient would be using antipsychotic medications at each subsequent follow-up. Those patients whose results indicated an “internal” LOC—reflecting a belief that the course of their life is mostly influenced by their own skills and efforts rather than by chance, fate, or other powerful individuals—and those who scored higher in self-esteem were far less likely to be taking antipsychotic drugs at 15 years. Patients who had more significant developmental achievements prior to hospitalization and those deemed during hospitalization to have good prognostic potential were also far less likely to use antipsychotic drugs.

The researchers stress that, overall, schizophrenia has a poor outcome: compared to the control group of patients who had been treated for psychosis but did not have schizophrenia, those with schizophrenia functioned less well at all stages of the study. However, Harrow and Jobe say that their data provide evidence that specific subgroups of patients with schizophrenia with measurable personality traits and prognostic factors may not immediately relapse and may function quite well for a considerable period of time without antipsychotic drugs.

Harrow and Jobe point out that a trend toward better functioning without medications only began to emerge 4.5 to 7.5 years after their patients were discharged from the hospital, and that their long-term, “naturalistic” study design may avoid the inevitable observer bias of clinically based studies. “The controlled trials data on clinic populations of patients suggest that among the patients with schizophrenia who stay in clinic treatment settings for years after the acute phase there is increased risk of relapse when going off antipsychotics,” the authors write. “However, the current data suggest that for the select subgroup of patients with schizophrenia who are not in clinic settings, who have gone off antipsychotics and did not immediately relapse, and stayed off them for a period of time, a surprising number experienced periods of recovery and continued to function well for a considerable period without antipsychotics. Clearly, the present longitudinal data suggest that not all patients with schizophrenia need to use antipsychotic medications continuously throughout their lives.”

If at First You Don’t Succeed . . . .
Another naturalistic follow-up study by Tadashi Nishikawa and colleagues at Seiwakai Nishikawa Hospital in Hamada, Japan, and at the National Institutes of Health in Baltimore, Maryland, also provides evidence that discontinuation of antipsychotic medications is a viable option for some patients with schizophrenia.

In an analysis accompanied by eight case reports in the spring 2007 issue of Psychiatry, Nishikawa and colleagues report that eight out of 30 remitted patients (26.7 percent; selected from a total population of 300 +/- 20) followed for an average of 10.7 years in a five-step drug withdrawal program were able to forego antipsychotic drugs for more than 2 years, even after multiple psychotic episodes. In one case, a patient did not take antipsychotic drugs for 15 years, until a relapse required a brief course of drug therapy.

Remission was defined as follows: "(a) total scores in each of nine clinical dimensions of Positive and Negative Syndrome Scale, including the positive and negative syndrome and depression, were either 1 or 2, (b) patients kept up normal social and occupational functioning, and (c) the same condition lasted at least 3 months under neuroleptic medication."

The team emphasizes that in half of these eight cases patients needed two or more attempts at antipsychotic withdrawal to reach a drug-free state, an outcome that could be obscured by more standard study designs. “These cases would be counted as ‘relapse’ or ‘dropouts’ in short-term randomized control studies,” the Nishikawa team writes. The researchers argue that with intensive follow-up care, minor relapses and repeated withdrawal programs may help patients to achieve a drug-free state in the long term, especially if withdrawal is accompanied by social-skills training.

The Nishikawa group’s data show that younger patients with acute onset of symptoms, as well as those who receive antipsychotics at lower doses or for a shorter period of time, are most likely to successfully withdraw from the drugs for extended periods. The authors argue that high-dose maintenance regimens may induce neuronal adaptations that make withdrawal more difficult (see Lieberman et al., 1994). “Even though this is an open study with a limited number of patients,” the Nishikawa team writes, “our results suggest that approximately one fourth of remitted schizophrenics could discontinue maintenance neuroleptic medication even after multiple episodes.”

Relapse: Too High a Risk?
A more cautious conclusion emerged from a shorter-term clinical study published in the May issue of the Journal of Clinical Psychiatry by Lex Wunderink and colleagues at the University Medical Center in Groningen, The Netherlands. In this work, researchers compared antipsychotic maintenance treatment to “guided discontinuation” of the drugs in 131 schizophrenia patients in remission after a first episode of psychosis. For the purposes of the study, the researchers defined remission as a sustained improvement in positive symptoms for 6 months; negative and disorganizational symptoms were not considered in admission criteria.

The patients, who were randomly assigned to receive maintenance treatment or guided discontinuation, were followed for 2 years following their first response to drug treatment, and were assessed using the Positive and Negative Syndrome Scale (PANSS) after the required 6-month remission, at 15 months, and at 24 months. Measures of drug side effects, social functioning, and quality of life were also taken at some of these same intervals.

The 63 patients in the maintenance group received treatment according to American Psychiatric Association guidelines, with preference given to low doses of second-generation antipsychotics, while in 65 patients the dosage was gradually tapered off and discontinued if possible. If signs of relapse or positive symptoms occurred, clinicians were instructed to restore the dosage to its previous effective level.

The Groningen team found that relapse rates were twice as high in the discontinuation group (43 percent vs. 21 percent, p = .011), and only about 20 percent of the group was successful in discontinuing the drugs. The PANSS scores, hospitalization time, and side effects were virtually identical in both groups, leading the researchers to conclude that the risk of relapse outweighs any other benefit that might come from universally tapering off medication in first-episode patients: “Given the relatively small number of patients who were successfully discontinued (21.5 percent), the twice-higher relapse rates with the discontinuation strategy, the selected patient sample of stably remitted and cooperative patients included in the trial, and the lack of substantial advantages of the discontinuation strategy over maintenance treatment, the discontinuation strategy does not seem to offer sufficient benefits over maintenance treatment to implement the strategy in regular practice for all remitted first-episode patients,” Wunderink and colleagues write.—Peter Farley.

Harrow M, Jobe TH. Factors involved in outcome and recovery in schizophrenia patients not on antipsychotic medications: a 15-year multifollow-up study. J Nerv Ment Dis. 2007;195(5):406-414. Abstract

Nishikawa T, Hayashi T, Koga I, Uchida Y. Neuroleptic withdrawal with remitted schizophrenics: a naturalistic follow-up study. Psychiatry. 2007;70(1):68-79. Abstract

Wunderink L, Nienhuis FJ, Sytema S, Slooff CJ, Knegtering R, Wiersma D. Guided discontinuation versus maintenance treatment in remitted first-episode psychosis: relapse rates and functional outcome. J Clin Psychiatry. 2007;68(5):654-661. Abstract

Comments on News and Primary Papers
Comment by:  Sarah Yates
Submitted 1 June 2007
Posted 1 June 2007

Discontinuation studies like Wunderink's are problematic because it is difficult to tell whether relapse is due to genuine recurrence reflecting original natural history, or drug withdrawal. There is little evidence to show what length of tapering protocol might be most appropriate.

To illustrate the difficulty, consider what happens in Tourette syndrome:

"Rapid discontinuation from drugs such as haloperidol, pimozide, and fluphenazine may lead to severe withdrawal effects. In general, discontinuation of medication may lead to 2 to 3 months of increased symptoms. Thus, if those medications are withdrawn, it cannot be expected that the patient's "real" status will be visible for quite a while. Some patients may improve for a few weeks after neuroleptic discontinuation and then worsen after an additional week or so, remain worse for a while, and then gradually improve. Side effects such as cognitive blunting, troubles with memory, feelings of dullness, poor motivation, school and sociable phobias, excessive appetite, and sedation may lift rather quickly over days to several weeks, while emergent tic symptoms remain or become worse. Thus, the decision to discontinue neuroleptics, particularly haloperidol, may be harder than their initiation. Withdrawal must be planned so that the patient's life is disrupted as little as possible. Often, patients and their families will have great difficulty in tolerating the discontinuation and will need a good deal of support from the physician."

(From Guide to the Diagnosis and Treatment of Tourette Syndrome. Ruth Dowling Bruun, M.D., Donald J. Cohen, M.D., James F. Leckman, M.D. Tourette Syndrome Association. 1984. Published online at Internet Mental Health, a website by Canadian psychiatrist Philip Long.)

If this was schizophrenia and the individual was in a trial, he or she would be back on meds like a shot, rather than helped through what might be a temporary exacerbation.

I have to say I find it impossible to believe there would be no difference in side effect profile between individuals on and off atypical antipsychotics. There is a flaw in the protocol.

View all comments by Sarah YatesComment by:  Marvin Herz
Submitted 4 June 2007
Posted 4 June 2007
  I recommend the Primary Papers

It is clear that some patients with schizophrenia can do well off antipsychotic medication most of the time. The challenge is how to identify them and develop strategies to minimize the possibility of full relapse. The Harrow study is very helpful in that regard. A small percentage of first break patients will not have another relapse after recovering from the episode even if they are off medication. The benefits and risks should be discussed with the patient if he/she wants to be off medication.

In my studies, I have identified patient characteristics that help identify which patient might succeed off medication, including first break. They should have insight into their illness and not lose it if they begin to relapse; be capable of recognizing their early warning signs of impending relapse and be willing to take APDs when this occurs (having a cooperative family helps); no history of violence or suicidality during prior psychotic episodes; no history of rapid decompensation in prior episodes; and not a drug abuser. Patients should be taught to recognize early signs of relapse and coping skills to deal with them. See Herz et al. (2000) describing a strategy for relapse prevention. In this study, patients were on maintenance APDs.

View all comments by Marvin HerzComment by:  Sarah Yates
Submitted 6 June 2007
Posted 8 June 2007

Marvin Herz writes, "A small percentage of first break patients will not have another relapse after recovering from the episode even if they are off medication. The benefits and risks should be discussed with the patient if he/she wants to be off medication."

But surely the bigger issue is that patients can have relapses whether on antipsychotics or not. The real question is who will, taking into account serious physical side effects, significant increased mortality, and cognitive effects, benefit overall and long term from antipsychotics.

Well, maybe not the 20 percent or so who show little or no functional response; they are getting a heap of side effects for minimal benefit. And for those in the middle, i.e., excluding those 20 percent who would have gotten better anyway—leaving 60 percent of the total—well, we do not know, because discontinuation effects fatally flaw the maintenance studies. It is impossible to accurately judge the impact of discontinuation on relapse rates (see previous post). Discontinuation studies also miss out those who are not stable, read not doing well on antipsychotics; the trials only include those who are stable on antipsychotics. So the take-home efficacy data is nothing like as sound as presented. And we know, do we not?, that there is hard evidence of biased trial design, selective reporting, fraudulent use and analysis of data and ghost writing. Sad but undeniably true.

The Harrow study shows 40 percent off meds, and they are, as a cohort, doing better than those on antipsychotics. Some have psychosis, but as is very, very clear, so do many on antipsychotics. It is a fair bet that those reported as doing very badly on antipsychotics would do better off antipsychotics.

Look at 5-year outcomes from a service-based model built around antipyschotic restriction, e.g., Seikkula in Finland. It's flawed, but it is real-world stuff, and outcomes are very good with 29 percent on antipsychotics. I see no evidence the specific Open Dialogue approach is itself contributing to the good outcomes, but if you're looking for what happens to those off antipsychotics, this is as good a place as any.

So where does all this leave us? In my view, it is time for an independent and unbiased review of all the available literature, including claims that early intervention with antipsychotics prevents cognitive decline. Maybe the majority of reviewers should be eminent researchers with expertise in relevant disciplines, rather than psychiatrists.

Folks, you need to build a credible evidence base for current practice, because you do not have one. Many of you are acting like you're giving out insulin for diabetes; indeed, some of you make that analogy. But the data do not support you.


Seikkula J, Aaltonen J, Alakare B, Haarakangas K, Keranen J, Lehtinen K. Five-year experience of first-episode nonaffective psychosis in open-dialogue approach: Treatment principles, follow-up outcomes, and two case studies. Psychotherapy Research, March 2006; 16(2): 214.

Dixon LB, Lehman AF, Levine J. Conventional antipsychotic medications for schizophrenia. Schizophr Bull. 1995;21(4):567-77. Review. Abstract

Hellewell JS. Treatment-resistant schizophrenia: reviewing the options and identifying the way forward. J Clin Psychiatry. 1999;60 Suppl 23:14-9. Review. Abstract

View all comments by Sarah YatesComment by:  Marvin Herz
Submitted 12 June 2007
Posted 13 June 2007

Sarah Yates dismisses the importance of maintenance medication in preventing relapse, because patients relapse on or off medication. It is known that the majority of patients alternate between acute episodes (relapse) and periods of full or partial remission. There have been randomized controlled studies comparing an intermittent (targeted) medication strategy with maintenance medication which have all found that intermittent medication results in much higher relapse rates over 2 years. In our study, it was 30 percent for intermittent and 16 percent for those on maintenance (Herz et al., 1991). Other studies have found even larger differences favoring maintenance.

Relapses are highly undesirable for patients and their families. After each relapse, re-hospitalizations are lengthier and patients may not recover to their previous level of symptoms and functioning. However, a German study found no benefit in maintenance medication for first break patients (Gaebel et al., 2002). My basic point is that identifying characteristics of patients who might benefit from being off medication most of the time is important, but we must develop strategies to protect them from the likelihood of having relapses.

View all comments by Marvin HerzComment by:  Patricia Estani
Submitted 13 June 2007
Posted 18 June 2007
  I recommend the Primary Papers

I think that the question of this research news—must the administration of antipsychotic drugs be forever?—is one of the most important questions that really began an excellent debate among the scientific community.

The article by Dr. J. Bola (Bola, 2006; also see SRF related news story) is a really excellent article in this regard, not only for a clinical discussion but also for a research discussion. I think that the scientific community must take these ideas as one of the most important issues for future discussion in the field of schizophrenia research.

View all comments by Patricia EstaniComment by:  Vittorio Di Michele
Submitted 20 June 2007
Posted 20 June 2007
  I recommend the Primary Papers

In my opinion, these papers confirm and disclose, with stringent scientific methodology, a wide range of phenomena, until now not adequately investigated and understood, observed in countries with a very strong commitment to community-based mental health care, like Italy. The following phenomena listed are typical of a community-based mental health system, which is specifically organized by rules and recommendations of the National Plan of Italian Government:

1. A large proportion of schizophrenia-schizoaffective patients isn’t ever admitted in a psychiatric inpatient unit.

2. The antipsychotic mean doses are lower in Italy than in the U.S., and many patients show a “satisfactory” response at dosage much lower than expected.

3. Many patients and their caregivers (usually parents or partner) manage their antipsychotic daily dosage by themselves, adjusting the dosage according to their needs. Surprisingly, very often this management works!

4. Psychosocial interventions reduce the need for antipsychotics and are currently delivered in a special unit of Mental Health Department termed the “Daytime Center.”

5. The prevalence of schizophrenia “at the service” is systematically lower than expected, suggesting thus that many patients are adequately treated in the community by private practice or GP.

6. Experimental studies conducted by my colleagues and me suggest that the main variable predicting admission in a long-term residential rehabilitation facility is the occurrence of turbulent and violent behavior at least once in the course of illness. Such patients take antipsychotics at higher dosage than controls (Di Michele et al., 2007), show an inadequate control of their positive and negative symptoms, and are admitted at least once to a hospital psychiatric unit.

7. Patients in care by mental health centers have invariably poor prognoses: they do not work, have attempted suicide at least once, are unmarried or divorced (Di Michele and Bolino, 2004). The younger the age of first contact, the poorer is the prognosis.

In conclusion, even though we have much more information than in the past about the outcome and prognosis of patients admitted in a psychiatric unit, information about patients who attend community care is scarce and dispersed, but potentially valuable and of great interest for research. This approach will reduce the heterogeneity of the disease in terms of outcome and social disability. In my opinion, the perspective of investigation of drug treatment in schizophrenia should be expanded because many potential confounders influence the therapeutic response and prognosis, especially in a naturalistic setting.

View all comments by Vittorio Di MicheleComment by:  Sarah Yates
Submitted 22 June 2007
Posted 22 June 2007

I am not dismissing the importance of maintenance medication; I am saying that the randomized controlled studies quoted by Marvin Herz, which show advantages of maintenance over intermittent medication, are so fundamentally flawed by inadequate tapers and potential discontinuation effects that it is difficult to interpret the results (see my previous posts).

In addition, design of the studies gives an apparently distorted effect, because significant numbers of patients, up to 40 percent, who continue to experience psychotic symptoms whilst taking antipsychotics are not included in the trial, so one is examining a subset of “good” drug responders, giving a false picture of antipsychotic efficacy. The evidence shows that a significant proportion of “good” responders would have recovered without drug treatment in the first place. No drug treatment is not equivalent to treatment and withdrawal.

For these reasons the clearest way to truly assess the benefit of long-term maintenance medication is to use placebo-controlled RCTs (Bola, 2006; see SRF related news story. Available trials do not show benefits from automatic long-term antipsychotic use. In the current climate it is difficult to plan such trials. We must therefore look to naturalistic studies such as Seikkula's to indicate likely outcome of restricted use of antipsychotics.

What I am saying is that the results of such trials seem at face value to be at odds with results from discontinuation trials; in other words, outcome is much better than one would predict, and this is because discontinuation trials tend to bias outcome. Quotes based on data such as presented by Marvin Herz suggest a straightforward scenario whereby relapse rate for those maintained on antipsychotics is half that for those off antipsychotics. This is very misleading, indeed, plain wrong, which is why there is surprise that the 40 percent of patients with schizophrenia off antipsychotics in the Harrow study are doing much better as a cohort than those on antipsychotics. Would results have been even better if discontinuation effects had been minimized by seeking to avoid drug use, wherever possible, in the first place? Given the implications for forcible medication, a scientific analysis of all the available data is long overdue. In my opinion, severely restricted use of antipsychotics, as in Seikkula's study, is the right answer. At the very least, the evidence base for current “orthodox” practice requires more research to plug glaring holes.

View all comments by Sarah YatesComment by:  William Carpenter, SRF Advisor (Disclosure)
Submitted 22 June 2007
Posted 22 June 2007
  I recommend the Primary Papers

I have been interested in this issue for a long time, first as a treatment issue and later as an ethics problem. Key aspects of the discussions are often misconceptualized. My view is as follows:

1. The issue is never drugs or no drugs, but rather how to integrate therapies.

2. Continuous medication is not the only active pharmacotherapy strategy. Targeted antipsychotic treatment is effective (less so than continuous for relapse prevention, but perhaps better for negative symptoms and similar to overall outcome measures). A targeted approach may be optimal for patients who refuse continuous medication, for patient subjects in off-medication protocols for research, and for a subgroup with good prognostic indicators who want to consider managing their recovery process off medication. The targeted approach addresses relapse prevention by intervening early in exacerbation and by assuring continuity of clinical care (see Buchanan and Carpenter, 1996, for overview).

3. Off-medication research is feasible and ethically defensible. The Helsinki Declaration was a flawed guideline and had not been accepted as binding in most countries. The recent "Clarification" brings it into line with ethical justification for off-medication research. We have addressed this change and how to proceed with off-medication studies in schizophrenia (Carpenter et al., 2003). Key points are strong scientific merit, inclusion/exclusion criteria to select patients to maximize safety, and early detection of exacerbation or failure to improve with a targeted antipsychotic drug intervention.


Buchanan RW, Carpenter WT. Targeted maintenance treatment in schizophrenia: Issues and recommendations. CNS Drugs. 1996; 4:240-245.

Carpenter WT, Schooler NR, Kane JM. The rationale and ethics of medication-free research in schizophrenia. Arch Gen Psychiatry. 1997;54:401-407. Abstract

Carpenter WT, Appelbaum PS, Levine RJ. The declaration of Helsinki and clinical trials: A focus on placebo-controlled trials in schizophrenia. American Journal of Psychiatry. 2003;160:356-362. Abstract

View all comments by William CarpenterComment by:  Shitij Kapur (Disclosure)
Submitted 28 June 2007
Posted 28 June 2007

Have followed the comments above with great interest. I come to this from the perspective of someone who studies the mechanisms of antipsychotics—usually focusing on the acute and the shorter-term, but find myself increasingly getting interested in the longer-term issues.

I think there can be little denying that antipsychotics administered to those acutely psychotic lead to less distress for those treated and better ability for the rest of us to work with them. The short-term mechanism (D2 blockade) and short-term gains are rather undeniable.

The question is, what are the longer-term trade-offs? This remains a complex question. From a biological perspective the introduction of any drug, including antipsychotics, is an external challenge for the body. The body responds by trying to establish homeostasis and compensate for this challenge. We have shown both in animals (Samaha et al., 2007) and in humans (Silvestri et al., 2000) that, in certain situations and certain individuals, the system responds to resist drug action by producing new receptors. Exactly how this comes about is not known. However, it seems reasonable to suggest that in this longer-term counter-response of the brain to the drug (and not just in the initial effect of the drug on the brain) may lie the clue to the longer-term outcomes: why some do well and others don't; why some relapse on the meds and others don't; and why some can tolerate withdrawal and others can't.

A few years ago I read an article by Whitaker, an author who does not generally write in this field (Whitaker, 2004). I cite and bring that article to the attention of this dialogue. The abstract says: "Although the standard of care in developed countries is to maintain schizophrenia patients on neuroleptics, this practice is not supported by the 50-year research record for the drugs. A critical review reveals that this paradigm of care worsens long-term outcomes, at least in the aggregate, and that 40 percent or more of all schizophrenia patients would fare better if they were not so medicated. Evidence-based care would require the selective use of antipsychotics, based on two principles: (a). no immediate neuroleptisation of first-episode patients; (b). every patient stabilized on neuroleptics should be given an opportunity to gradually withdraw from them. This model would dramatically increase recovery rates and decrease the percentage of patients who become chronically ill."

I do not agree with this rather radical position—but it does provide a very different perspective from mainstream thinking at the moment. I do often share it with my residents and fellows, not to endorse this position but to remind them that the jury is still out on the really long-term (years) effects of antipsychotics.

View all comments by Shitij KapurComment by:  James C. Matthews
Submitted 28 June 2007
Posted 29 June 2007

The whole idea of "to drug or not to drug" is rather new to me. As a schizoaffective patient for many years who has been mostly compliant with my psychiatrist's direction, I can only say that this might only add confusion to the issue. I attend a lot of discussion and support groups for people with bipolar disorder and schizophrenia. The message I hear the most is to keep taking one's meds. The reason that this message is so often repeated is simply that it is so frequently rejected or ignored by so many patients. The biggest reason for relapse and continued problems is noncompliance. Once patients learn that other patients are experimenting with discontinuing medication, you will simply have more cases of noncompliance. Most patients don't live in a vacuum and most don't necessarily do what they are told. While many lack insight, they do not lack free will.

View all comments by James C. MatthewsComment by:  Sarah Yates
Submitted 2 July 2007
Posted 2 July 2007

Thanks to all for an interesting debate.

William Carpenter writes, "The issue is never drugs or no drugs, but rather how to integrate therapies."

The therapeutic issue remains drug or no drug if one assumes appropriate social and emotional support will be available to all from diagnosis. At any given point in management the decision that really matters, and is directly influenced by the psychiatrist, is drug or no drug. All other specific therapies are of unproven utility (with the possible exception of cognitive behavior therapy), and can be administered concurrently with drugs. So what therapy is it one is integrating drug use with? I think the issue is whether the long-term goal is to maximize drug use, or minimize drug use, both in terms of numbers of individuals treated, and dose (including number of antipsychotics prescribed concurrently), and duration once treated.

My fundamental position is that careful assessment of long-term data suggests that the goal should be to minimize use of antipsychotics for psychosis including schizophrenia, and in most cases this is exactly the opposite of current professional practice.

I would argue that Harrow's study lends weight to the good clinical service outcomes found in circumstances where the issue is precisely antipsychotic or no antipsychotic at initiation of medical involvement. In Seikkula's study 71 percent of non-affective psychotic patients had no antipsychotic administered for the duration of the 5-year study (Seikkula et al., 2006). The other 29 percent had antipsychotics administered at some stage, but not necessarily continuously. In such a service there is no absolute clinical rule that dictates drug or no drug; rather, individual circumstances are weighed in each case. At any given decision point, however, the issue is drug or no drug, and the emphasis is self-evidently on minimizing use of antipsychotics. In the short term, benzodiazepines may be utilized to this end. If William Carpenter is saying the issue is never (or shouldn't be) universal use of drugs or universal condemnation of drugs, then I heartily agree. If he is saying that everyone with psychosis or schizophrenia must necessarily be medicated with an antipsychotic at some point, I think I disagree. (Where is the evidence this is in the best long-term interest of all patients? What evidence there is suggests otherwise; Bola, 2006; Whitaker, 2004). If he is saying that those started on antipsychotics at any point should have an opportunity to come off again, and there is a pressing need to determine how best to manage this, I agree.

Withdrawal studies involving cessation of antipsychotic over days or weeks are likely a mistake. Factors likely to compromise success in studies such as Wunderink et al. are the fact that the studies are open, and criteria for relapse are vague.

As it is hard to withdraw, maybe impossible for some, this ought to be taken into account when initiating patients to treatment; if initiation can be avoided, it should be avoided.

Considering Harrow and Jobe

Does long-term use of antipsychotics protect against poor outcome? Consider the following quotes from Harrow and Jobe:

It is reasonable to argue that had these individuals not been on antipsychotics, the long-term outcome would have been even worse. But, has anyone ever proven this properly? What if the opposite is true? It is my understanding that overall outcomes are no better, maybe worse, than they were in pre-neuroleptic times (Hegarty et al., 1994). The Harrow study does not assess tardive dyskinesia, tardive dysmentia, heart disease, stroke, respiratory illness, obesity, metabolic disturbance, diabetes, suicide rate, death rate. If long-term side effects are added to the balance, what is overall cost-benefit? Okay, the worse-outcome patients may be easier to manage, though by no means all would be disruptive or violent without treatment, but is it ethical to treat with dangerous and debilitating drugs to this end? Was physical lobotomy justified for management purposes?

Do those who choose to take themselves off drugs do worse in the long run than they would otherwise have done? Or do they do considerably better? How do so many manage to get off drugs and stay off drugs? Is it because they are not in a clinical setting, choose their own pace for withdrawal, and are not put back on drugs at the first sign of relapse/psychosis? It is not clear from the Harrow study whether the 40 percent off drugs were monitored; the suggestion is many were no longer seen in a clinic and so made their own judgment regarding need or cost/benefit of medication.

Harrow and Jobe also write, "Of the schizophrenia patients not on any medications at the 15-year follow-up, 29 percent were on antipsychotics at the 2-year follow-up." In other words, 70 percent of the 40 percent off antipsychotics had come off at a relatively early stage.

Further quotes from Harrow and Jobe:

It is not clear what the full pattern of antipsychotic use was over the 15-year period, although there is a clear correlation between drug status at 2 years and at 15, suggesting those off drugs tended to stay off drugs. There is evidence of increasing benefit off drugs over time. This is itself important; short-term results do not necessarily translate into long-term outcome.

The trouble with the Harrow study is that it is chicken and egg—do individuals do well long-term off antipsychotics because they had a milder variant of schizophrenia to start with, or because personality traits which increase likelihood of proactive drug cessation translate into better outcome due to less exposure to toxic and habituating effects of drug, or because personality traits can improve ability to cope with psychosis and negative effects? Or a combination of the three? Would overall outcome be better if everyone in the study were forced to take antipsychotics long-term, or if no one took antipsychotics at any stage (assuming the choice was one extreme or the other)? A very big question if you think about it enough, and I would be interested to hear the answer. Would outcomes be better overall if professionals rather than patients always decided who should go on, come off, and when? Or would they be better if patients were allowed to determine cost/benefit for themselves, including those with uniformly poor outcome who may feel side effects are awful, but are not permitted to come off drugs? How dependent is patient perception on professional advice? Is there evidence that automatic medication with antipsychotics on evidence of psychotic activity is beneficial in the long term; how big would the cohort off antipsychotics at 15 years be if such automatic medication was enforced; and what would the impact be?

Discussions of the Harrow paper have tended to assume that all of the 60 percent left on antipsychotics were benefiting overall, and so the challenge is to identify the small proportion who can remain free of psychosis with no drug. This totally ignores the fact that 83 percent of the very worst outcome cases were on antipsychotics, and around 70 percent on antipsychotics had psychotic activity. In other words, the drugs do not work well long-term, if at all, in a significant number of cases. In fact, this study could be interpreted as suggesting they might be detrimental long-term. The neat premise that if someone has psychotic activity the solution is necessarily to put them on antipsychotics, and, hey, presto! they will no longer experience psychosis, is not supported by the evidence in the paper.

In crude terms the Harrow study could be read as suggesting a third of patients with schizophrenia (worst-outcome cases) were not benefiting from antipsychotics (indeed maybe harm exceeds benefit), at least a third were managing reasonably well and avoiding side effects off antipsychotics, and the cost/benefits of antipsychotic use for the remaining third are debatable.

Even the Wunderink study suggests discontinuation should be tried. Twenty percent discontinued successfully, and only 25 percent of those in the discontinuation group experienced relapse, loosely defined but not severe since in fact the maintenance treatment group spent slightly longer in hospital than the discontinuation group. To suggest there are no benefits to being off medication is shortsighted (see below).

What Are the Costs of Antipsychotic Treatment?
Why does all this matter so much? If antipsychotics benefit some, then isn't it best to play on the safe side and treat everyone who might benefit?

The side effects of the antipsychotics are continuously downplayed. These are terrible drugs, arguably second only to chemotherapy, possibly worse in some respects. I won’t produce an exhaustive list of side effects, pyramidal effects, but will concentrate on key issues. All the available evidence suggests the drugs not only cause significant morbidity, but also mortality, although for a variety of reasons (which could be discussed elsewhere), the available data likely underestimate the problem. (See Joukamaa et al., 2006; Osborn et al., 2007 for death-rate studies). Joukamaa found a 2.5 increase in death rate for each neuroleptic administered, a sixfold increase for those on three or more concurrent antipsychotics. Osborn found very significant increases in death rates from coronary heart disease and stroke. Both studies were weighted for confounding factors, and the Osborn study has a systematic bias, which means seriously mentally ill people registered as being off antipsychotics may have been on antipsychotics or other harmful psychotrophics for long periods, or may have been in hospital on high doses.

Life expectancy for those with schizophrenia is reduced by over 25 years, with main causes of death cited as coronary heart disease and diabetes. Some deaths are unexplained (Colton et al., 2006). The fact that there have been few systematic long-term studies of either total morbidity burden or mortality is scientifically and ethically unacceptable, and undermines the concept of evidence-based medicine, particularly when medication can be enforced. The big problem with increased mortality associated with common disorders is it is easy to ignore. If a significant number of patients die of drug-related coronary heart disease, it's no one's problem. If far fewer patients (a handful) die of a rare problem (e.g., white blood cell abnormalities, as with clozapine), suddenly there is a huge ethical problem. Is it ethics? Or is it that because death is directly attributable to drug, it looks bad, and someone has to carry the can. No one forces anyone to stay on clozapine if there are white cell changes, even if that individual is benefiting from the drug, and it is the only drug that appears to work. There is weekly monitoring at huge cost, to pick up a few rare cases. Why? Because, horror, someone might die. Yet when I raise the question of morbidity and mortality I am told, by some psychiatrists, "All drugs have side effects." What is that supposed to mean? If the rare problems caused by clozapine were attributable to anything but drug, no one would bat an eyelid because in any individual case it would be difficult to prove it was drug-related (even though in some it would be), and no psychiatrist would ever need to feel culpable; more important, he or she would not get into trouble. Something is very wrong.

At root there are significant numbers of people who do not care whether those with schizophrenia are dying due to drug or not, let alone suffering unacceptable side effects. At the very least, they are not looking. Some claim the Osborn study shows a protective effect of antipsychotics at low doses, but he himself writes (personal communication) that the data collection methodology justifies no such conclusion. Within the limitations of the data, the conclusion is antipsychotics are associated with mortality in a dose-dependent manner. Osborn also writes (personal communication) the study was incapable of showing effects of atypicals because there were so few on atypicals, and in addition no one was on atypicals long enough to study the two parameters the researchers looked at, i.e., death from coronary heart disease and stroke, both of which tend to be long-term effects. The atypicals, with all their increasingly well-understood metabolic effects, though many individuals are not monitored, are likely worse than the typicals, though the long-term data are not available.

The drugs also cause tardive dyskinesia in over 40 percent of patients long-term, with evidence atypicals will have similar outcome (de Leon, 2007) and tardive dysmentia is associated with this, i.e., there is clear evidence of irreversible brain damage affecting motor skills and mental faculties. Imaging studies indicate dose-related drug-induced loss of brain volume; properly controlled experiments with monkeys suggest significant reductions in brain volume (e.g., 10 percent in parietal lobe) and reduction in glial cells after only 1 year on atypicals or typicals (Konopaske et al., 2006). As discussed previously in this thread there is evidence the drugs cause upregulation of dopamine receptors, with potential for worsening of the condition they are designed to treat.

In addition, and by no means least, the effect of antipsychotics is frequently a kind of death of the personality—they cause numbing, avolition, and compromise higher thinking skills, as normal volunteers attest. Doctors were unable to work after administration (Belmaker and Wald, 1977), and there are numerous accounts of detrimental cognitive and emotional effects on volunteers. I can attest to the effects of atypicals firsthand, as I have an organic illness which is thought to have triggered a psychosis and (possibly) necessitated use of a low-dose antipsychotic for a relatively brief period. Chemical lobotomy is an excellent description, interestingly no longer in use. Presumably it is no longer considered ethical, and no one wants to consider the full implications of treatment. It took several weeks for the worst effects to appear. Normal life was only possible once I was off the drug. Does this make me an antipsychiatrist? Certainly not. Would it make me likely to worry about those subjected to enforced use of such drugs, particularly since the literature accompanying the drugs gives little indication of cognitive effects? Well, of course.

If I have the professional expertise to read primary data objectively (I have a Ph.D. in inherited neuromuscular diseases, was Research Director of a large medical charity for over 15 years, instigated the European Neuromuscular Centre, and chaired a Primary Care Trust on strategic development and provision of clinical service) and question the evidence base for current use of antipsychotics and for overall cost/benefit, is it reasonable to do so? Absolutely. Does it serve my purpose to approach the literature with any intrinsic bias? Well, the interesting thing is my original bias was by and large in favor of antipsychotic use, but after ploughing my way through hundreds of key papers, including all those underpinning the U.K. National Institute of Clinical Excellence (NICE) guidelines, and seeking the advice of scientific and clinical colleagues as questions arose, I became more and more concerned. So far no one has been able to produce the evidence needed to alleviate those fears.

If anything, NICE guidelines are probably biased by over-reliance on short-term studies and poorly designed withdrawal studies. It hardly helps that despite NICE guidelines suggesting discontinuation of antipsychotics 1 to 2 years after a psychotic episode, based on expert assessment of all available evidence, the majority of U.K. psychiatrists and GPs fail to instigate this. Despite the NICE recommendation that multiple antipsychotics should not be prescribed, and atypicals and typicals should not be used concurrently except for crossovers, this is common practice. A recent Health Commission survey which sent specialist pharmacists into U.K. psychiatric hospitals found a third of patients were being given more than the recommended dose, and more than 70 percent of prescriptions were considered inappropriate, often because of inadvisable concurrent prescribing, including more than one antipsychotic (see BBC news story).

Consequences of Forced Intervention
Here is a big issue. What about scope for (in effect) indefinite forcible treatment in the community following one admission to hospital for mental illness, as the current U.K. mental health bill proposes? A Community Treatment Order (CTO) could be issued without court involvement, and will permit anyone deemed to be noncompliant with treatment to be returned to hospital for treatment, probably a depot injection in many cases. There is no appeal, and the Order can be renewed at 6-month intervals. Most psychiatrists believe all but a very small minority of individuals with schizophrenia will do better long-term if maintained on antipsychotics, and it is more ethical to err on the side of caution, i.e., over-treat rather than under-treat. Where is the real evidence? The CATIE study fits with the Harrow study in that it suggests in the real world patients do come off medication, either because side effects are intolerable, or because they see no clinical benefit (Lieberman et al., 2005). Are we risking abusive situations because psychiatrists’ perceptions will be able to override reality? In other words, those who would have done much better long-term off drugs, or at least by choosing for themselves when and how to take drugs, will now be forced onto drugs.

Some patients may do better in a regimen of long-term forcible medication, but who will these be? The current U.K. bill makes no attempt to even define this population; as it stands, the legislation can apply to anyone hospitalized for mental illness.

Good-quality, long-term randomized placebo controlled trials are needed to address the questions in this post, or at least matched service cohorts, and despite William Carpenter's argument, the current climate precludes this. Even if the ethical argument is won (and it is debatable it only applies to good prognosis patients), who is going to fund the large well-designed unbiased trials needed to provide a definitive answer? One, or a series, which will overturn current orthodoxy. The pharmaceutical companies? Do the pharmaceutical industries have a vested interest in minimizing use of antipsychotics, or drawing attention to the fact there may be an issue? Okay, say we've found the money, and we've found the impartial scientists and clinicians committed to answering the right questions, with the intelligence, breadth of knowledge, insight, and freedom to design the studies so that they will in fact answer those questions ethically. How long will the studies take? Ten years? Fifteen years? In the meantime, I believe it morally wrong to act as if the drugs are "safe and effective," and choosing not to take them is almost by definition a sign of insanity. Implementation of poorly defined CTOs is just one example of behaving as if there is a guaranteed benefit, at acceptable cost. Increasing use in (often nonpsychotic) children, who have no choice, and in presymptomatic adolescents, who may be making poorly informed choices, are other examples....

This goes back to the original argument: what treatment regimen does the balance of available long-term data suggest? Seikkula's 5-year study indicates the advisability of restricting initiation of antipsychotics wherever possible. Harrow's study could be read as a manifesto for patient choice. The CATIE study suggests given the choice, patients tend to come off drugs. Bola, at the very least, highlights the need for more placebo-controlled data. But, I hear readers exclaim, what about those with no insight, i.e., who fail to recognize they are ill. If a patient never regains insight, then are the drugs working, and is it ethical to continue them? If the patient does regain insight, then what is wrong with an advance directive? Is “insight” always taken to mean recognition of illness, or is it often taken to mean recognition that drug treatment is beneficial to the individual, i.e., benefit outweighs cost? If the patient thinks treatment isn't beneficial, is the patient by definition lacking in insight? Is benefit to the patient always the goal, or is social management and protective medicine also a consideration? In the current climate almost anything can happen to a patient on antipsychotics, no questions asked, but should something go wrong while the patient is not medicated, the psychiatrist is considered culpable.

Who has some decent evidence showing that there is a causal relationship between duration of non-drug-treated psychosis and outcome? In other words, demonstrated by a long-term randomized placebo-controlled trial comparing immediate treatment of acute-onset psychosis with delayed or no treatment? The prognosis for slow-onset schizophrenia has long been understood to be poorer than for acute-onset schizophrenia. and slow-onset cases are more likely to have delayed treatment. In addition if you delay treatment of schizophrenia, let alone acute psychosis, at least 20 percent, and often many more, will recover spontaneously (Dixon et al., 1995). If you compare those who do not recover well with a cohort with very recent onset of psychosis, there is obviously going to be a difference in favor of the recent-onset cases.

This whole topic is crucial. The advent of CTOs means that psychiatrists' perceptions can and will override the reality of the current situation. The reality is once in the community, patients make their own judgments regardless of orthodoxy, and many come off drugs. Most are not supported in this (according to patient surveys conducted by MIND, the U.K.’s foremost mental health charity).

The Harrow study suggests that long-term, and on balance, the patients are more likely to get it right than anyone else. If they didn't, the outcomes would have been reversed. Listening to most psychiatrists, one would predict, indeed be certain, that the outcomes would have been reversed. There are plenty of psychiatrists, and even more GPs who would argue 100 percent of the patients in the Harrow study should have been on antipsychotics continuously; either that or they would enforce or recommend long periods on antipsychotics, creating a situation where discontinuation was effectively impossible. The Harrow study is counterintuitive to many, as is the finding that outcome (based on identical diagnostic criteria across countries) is better in countries with very low use of antipsychotics (15 percent; Jablensky, 1992). Common criticisms of the second WHO study are unfounded (Jablensky, 1994).

The tragedy is that there is no overall evidence CTOs are even beneficial for those targeted few they are designed to serve. And the potential for negative long-term impact on health has not been thoroughly assessed. Yet they have the potential to destroy the lives of many, particularly if broadly applied as will be feasible in the U.K. The Institute of Psychiatry, King's College London, has published a new report, “International Experiences of Using Community Treatment Orders” by Dr Rachel Churchill, Professor Matthew Hotopf, and associated coauthors. The report explains that there is no robust evidence as yet to prove whether community treatment orders (CTOs) are beneficial or harmful to patients (see BBC news story). This report represents the most comprehensive and thorough review of research into international experiences of using compulsory treatment orders, summarizing evidence from 72 databased empirical studies undertaken in six countries over the last 30 years.

It is only possible for such legislation to come into force when lay politicians are fed the false idea that antipsychotics are "safe and effective." Unfortunately, the pharmaceutical companies are involved in this both directly via lobbying to government and indirectly by just about any means possible. Does this worry anyone?

Evidence-based medicine? First, the evidence base is flawed, and from what we know of proven practice by pharmaceutical companies, deliberately distorted on a number of levels; second, many psychiatrists feel free to do whatever they wish, regardless of any evidence. Fancy forcible long-term maintenance, without appeal, on a doubled risperidone depot, haloperidol chaser, and four other concurrent psychotrophics anyone? Think it couldn't happen? Think it advisable?

So, should we be minimizing or maximizing use of antipsychotics?


Belmaker RH, Wald D. Haloperidol in normals. Br J Psychiatry. 1977 Aug 1;131():222-3. Abstract

Bola JR. Medication-free research in early episode schizophrenia: evidence of long-term harm? Schizophr Bull . 2006 Apr 1 ; 32(2):288-96. Abstract

Colton CW, Manderscheid RW. Congruencies in increased mortality rates, years of potential life lost, and causes of death among public mental health clients in eight states. Prev Chronic Dis. 2006 Apr 1;3(2):A42. Abstract

Dixon LB, Lehman AF, Levine J (1995) Conventional antipsychotic medications for schizophrenia. Schizophr Bull 21:567–577. Abstract

Hegarty JD, Baldessarini RJ, Tohen M, Waternaux C, Oepen G. One hundred years of schizophrenia: a meta-analysis of the outcome literature. Am J Psychiatry. 1994 Oct 1;151(10):1409-16. Abstract

Assen Jablensky, Psychological Medicine, suppl. 20 (1992), 1-95.

Joukamaa M, Heliövaara M, Knekt P, Aromaa A, Raitasalo R, Lehtinen V. Schizophrenia, neuroleptic medication and mortality. Br J Psychiatry. 2006 Feb 1;188():122-7. Abstract

Konopaske GT, Dorph-Petersen KA, Pierri JN, Wu Q, Sampson AR, Lewis DA. Effect of chronic exposure to antipsychotic medication on cell numbers in the parietal cortex of macaque monkeys. Neuropsychopharmacology. 2007 Jun;32(6):1216-23. Epub 2006 Oct 25. Abstract

Lieberman JA, Stroup TS, McEvoy JP, Swartz MS, Rosenheck RA, Perkins DO, Keefe RS, Davis SM, Davis CE, Lebowitz BD, Severe J, Hsiao JK, . Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N Engl J Med . 2005 Sep 22 ; 353(12):1209-23. Abstract

Osborn DP, Levy G, Nazareth I, Petersen I, Islam A, King MB. Relative risk of cardiovascular and cancer mortality in people with severe mental illness from the United Kingdom's General Practice Rsearch Database. Arch Gen Psychiatry. 2007 Feb 1;64(2):242-9. Abstract

de Leon J. The effect of atypical versus typical antipsychotics on tardive dyskinesia: a naturalistic study. Eur Arch Psychiatry Clin Neurosci. 2007 Apr 1;257(3):169-72. Abstract

Seikkula J, Aaltonen J, Alakare B, Haarakangas K, Keranen J, Lehtinen K. Five-year experience of first-episode nonaffective psychosis in open-dialogue approach: Treatment principles, follow-up outcomes, and two case studies. Psychotherapy Research, March 2006; 16(2): 214.

Whitaker R. The case against antipsychotic drugs: a 50-year record of doing more harm than good. Med Hypotheses. 2004 Jan 1; 62(1):5-13. Abstract

View all comments by Sarah YatesComment by:  Wolfgang Gaebel
Submitted 13 July 2007
Posted 13 July 2007
  I recommend the Primary Papers

Efficacy of Treatment Has to Be Related to Safety
The three articles and their summary provided by the Schizophrenia Research Forum address very important questions in long-term treatment of schizophrenia. Due to the efficacy of (first- and second-generation) antipsychotics (FGAs, SGAs) florid psychotic symptoms mostly remit under initiated and maintained treatment in the acute and stabilization phase. However, this mental illness is in most cases (about 80 percent) characterized by symptom recurrence or relapse, and long-term (low-dose) maintenance treatment is indicated to prevent relapse and maintain or improve patients' level of functioning. Like nearly all drugs, antipsychotics can also cause side effects, like extrapyramidal reactions, tachycardia, hypotension, lethargy, impotence, or hyperprolactinaemia (mainly by FGAs), agranulocytosis (Clozapine), weight gain, or metabolic effects (mainly discussed for the newer SGAs). However, contrary to the SRF summary, serious side effects are rather infrequent and uncommon. Nevertheless, to avoid (potential) harm for patients, antipsychotic treatment should be kept as short (and in the lowest dose) as possible. Hence, the respective questions arise: How long should long-term treatment after remission of schizophrenic symptoms and stabilization be maintained? Are there alternative long-term treatment strategies to continuous treatment? Are longer periods without medication indicated? Which kind of patient is eligible?

Guidelines recommend evidence-based treatment
These questions are not new and have all (admittedly not sufficiently) been addressed by former studies. Evidence has been summarized by different national and international psychiatric associations and integrated to guideline recommendations (for a review see Gaebel et al., 2005). Contrary to Sarah Yates, they can be seen as ("credible") evidence-based recommendations to assist practitioners in treatment decisions. Based on our review, the three guidelines with highest (quality) scores are the ones from NICE (2002), APA (2004), and RANZCP (2002; revised in 2005). In addition, the German guideline has been recently revised based on the highest methodological standard (Gaebel et al., 2005).

According to these guidelines, (long-term) treatment should be maintained after a first episode in schizophrenia for at least 1 year; after a relapse for 2-5 years and for multiple-episode patients indefinite maintenance treatment is recommended.

To minimize side effects, especially tardive dyskinesia (TD) to some extent caused by FGAs, and to deal with frequent partial or complete noncompliance, targeted intermittent treatment has been provided in the late 1970s. In stable patients, drug treatment is gradually and stepwise discontinued and restarted in case of early warning signs (EWS) or prodromal symptoms of an impending relapse (requiring teaching patients and relatives to recognize EWS; see Marvin Herz's contribution). However, the risk for relapse was shown to be noticeably higher for targeted intermittent compared to maintenance treatment for schizophrenia patients in general (Kane, 1999), and intermittent treatment appears to increase rather than decrease the risk for TD (APA, 2004). In addition, the accuracy of relapse prediction based on prodromal symptoms or early warning signs is limited (Gaebel and Riesbeck, 2007). Reflecting these results, guidelines recommend (reservedly) targeted intermittent treatment as an alternative long-term treatment strategy after gradual and stepwise drug discontinuation and under close monitoring of early warning signs for an impending relapse. Against this background, the discontinuation strategy described by Nishikawa et al. (2007; increasing drug breaks and afterwards dose reduction) seems less appropriate.

Differential indication for targeted intermittent treatment?
On the other hand, research suggests that intermittent treatment seems more feasible in special groups of patients, particularly in first-episode patients. In a post-hoc analysis, maintenance treatment was found to be superior to targeted intermittent treatment regarding relapse prevention only in multiple-episode patients; however, both long-term treatment strategies were shown to be equally effective in first-episode patients (Gaebel et al. 2002). In recent years several new research programs on first-episode schizophrenia have been implemented, some of them focusing on the indicated duration of long-term (maintenance) treatment and the efficacy of targeted intermittent treatment. The (prospective) study of Wunderink et al. (2007), however, resulted in a twofold enhanced risk for relapse under intermittent compared to continued drug treatment, also in first-episode patients. Since continued maintenance treatment is recommended for the first post-acute year, their design—treatment was withdrawn already after a 6-month remission period—is debatable. A comprehensive program on optimization of long-term treatment for first-episode schizophrenia was conducted within the German Research Program on Schizophrenia (GRNS; Wölwer et al., 2003), which was funded by the German Ministry of Education and Research. In the first post-acute year, antipsychotic treatment was maintained (randomized controlled trial comparing low-dose haloperidol with risperidone; Gaebel et al., 2004; 2007), combined with different psychological strategies (2 months’ psychoeducation vs. 12 months’ cognitive behavioral therapy). In the second post-acute year, antipsychotics in patients sufficiently stable were randomly either continued or (stepwise) discontinued. Under both conditions, an early intervention was initiated in case of prodromal symptoms or other early warning signs as guided by a comprehensive decision algorithm (Gaebel and Riesbeck, 2007). Similarly to Wunderink and colleagues’ (2007), first (and preliminary) results indicate a significantly higher risk for relapse under targeted intermittent treatment. On the other hand, about 50 percent of patients having been discontinued remain stable in the second treatment year. However, final analyses have to be awaited.

All these data indicate that targeted intermittent treatment bears a higher risk for relapse also in first-episode patients. On the other hand, there are patients who remain stable although discontinued from antipsychotics. How can these patients be identified? The results from Harrow and Jobe (2007) demonstrate that patients with a favorable prognosis, as indicated by a higher premorbid level of functioning and favorable personality factors (e.g., internal locus of control), are more likely to remain stable over longer periods off medication. The respective prognostic analyses from the Wunderink et al. study, as well as our own data, will be provided in the near future. Nevertheless, expectations have to be scaled down, since replication of prognostic factors rarely succeeds. Even if factors can be replicated, applicability to an individual patient is still difficult. Appropriate cut-off scores have to be defined based on prognostic/diagnostic parameters (mainly sensitivity and specificity). In addition, according to the comprehensive aetiopathogenetic Vulnerability-Stress-Coping-model for schizophrenia, symptom (re-)exacerbation depends on (high biological) vulnerability, the occurrence of stressful life events, and (maladaptive) coping strategies. Accordingly, a better prognosis could be expected, if (markers of) all these factors are considered in addition to other relevant course indicators. As data from the first-episode long-term study within the GRNS indicates, the highest risk for relapse or deterioration is given if all three factors are present simultaneously.

Shared decision-making based on empirical evidence and practice
Bringing all issues together on a decision for the appropriate long-term treatment strategy after recovering from an acute episode in schizophrenia has to consider different factors regarding premorbid prognostic factors, indicators of vulnerability and illness course, coping abilities, and (future) stress exposure. Last but not least, one essential factor is added by the individual needs and preferences of the patient and his/her social environment. Accordingly, based on best knowledge, mainly provided by the treating psychiatrist and the individual needs expressed by the patient, pros and cons of the different treatment strategies have to be discussed leading to a shared decision. Nevertheless, more research regarding different long-term treatment strategies and their differential indication is needed to provide the evidence base for individual decisions.


Lehman AF, Lieberman JA, Dixon LB, McGlashan TH, Miller AL, Perkins DO, Kreyenbuhl J; American Psychiatric Association; Steering Committee on Practice Guidelines. American-Psychiatric-Association (APA): Practice guideline for the treatment of patients with schizophrenia, second edition. Am J Psychiatry, 2004; 161(Suppl): 1-56. Abstract

Gaebel W, Falkai P, Weinmann S, Wobrock T. Behandlungsleitlinie Schizophrenie. In: S3 Praxisleitlinien in Psychiatrie und Psychotherapie. edited by Deutsche Gesellschaft für Psychiatrie, Psychotherapie und Nervenheilkunde (DGPPN). Darmstadt: Steinkopf Verlag, 2005.

Gaebel W, Jänner M, Frommann N, Pietzcker A, Köpcke W, Linden M, Müller P, Müller-Spahn F, Tegeler J: First vs. multiple episode schizophrenia: Two-year outcome of intermittent and maintenance medication strategies. Schizophrenia Research, 2002; 53, 145-159. Abstract

Gaebel W, Möller HJ, Buchkremer G, Ohmann C, Riesbeck M, Wölwer W, von Wilmsdorff M, Bottlender R, Klingberg S: Pharmacological long-term treatment strategies in first episode schizophrenia. Study design and preliminary results of an ongoing RCT within the German Research Network on Schizophrenia. Europ Arch Psychiatry Clin Neurosci, 2004; 254:129-140. Abstract

Gaebel W, Riesbeck M, Wölwer W, Klimke A, Eickhoff M, von Wilmsdorff M, Jockers-Scherübl MC, Kühn K, Lemke M, Bechdolf A, Bender S, Degner D, Schlösser R, Schmidt LG, Schmitt A, Jäger M, Buchkremer G, Falkai P, Klingberg S, Köpcke W, Maier W, Häfner H, Ohmann C, Salize HJ, Schneider F, Möller HJ Maintenance treatment with risperidone or low-dose haloperidol in first-episode schizophrenia. One-year results of a randomized controlled trial within the German Research Network on Schizophrenia. J Clin Psychiatry, 2007 (in press).

Gaebel W, Riesbeck M. Revisiting the relapse predictive validity of prodromal symptoms in schizophrenia. Schizophrenia Research, 2007 (in press).

Gaebel W, Weinmann S, Sartorius N, Rutz W, McIntyre JS. Schizophrenia practice guidelines: international survey and comparison. Br J Psychiatry. 2005;187:248-55. Abstract

Harrow M, Jobe TH. Factors involved in outcome and recovery in schizophrenia patients not on antipsychotic medications: a 15-year multifollow-up study. J Nerv Ment Dis. 2007, 195(5):406-14. Abstract

Kane JM. Management strategies for the treatment of schizophrenia. J Clin Psychiatry; 1999, 60 (Suppl 12): 13-17. Abstract

Leucht S, Barnes TR, Kissling W, Engel RR, Correll C, Kane JM. Relapse prevention in schizophrenia with new-generation antipsychotics: a systematic review and exploratory meta-analysis of randomized, controlled trials. Am J Psychiatry; 2003, 160:1209-1222. Abstract

National Institute for Clinical Excellence (NICE). Guidance on the use of newer (atypical) antipsychotic drugs for the treatment of schizophrenia. Technology Appraisal Guidance 2002; 43 (

Nishikawa T, Hayashi T, Koga I, Uchida Y. Neuroleptic withdrawal with remitted schizophrenics: a naturalistic follow-up study. Psychiatry. 2007;70(1):68-79. Abstract

Royal Australian and New Zealand College of Psychiatrists (RANZCP). Royal Australian and New Zealand College of Psychiatrists clinical practice guidelines for the treatment of schizophrenia and related disorders. Aust NZ J Psychiatry, 2005; 39:1-30.

Wölwer W, Buchkremer G, Häfner H, Klosterkötter J, Maier W, Möller HJ, Gaebel W. German research network on schizophrenia - bridging the gap between research and care. Europ Arch Psychiatry Clin Neurosci; 2003, 253:321-329. Abstract

Wunderink L, Nienhuis FJ, Sytema S, Slooff CJ, Knegtering R, Wiersma D. Guided discontinuation versus maintenance treatment in remitted first-episode psychosis: relapse rates and functional outcome. J Clin Psychiatry, 2007, 68:654-61. Abstract

View all comments by Wolfgang GaebelComment by:  William Carpenter, SRF Advisor (Disclosure)
Submitted 14 July 2007
Posted 16 July 2007
  I recommend the Primary Papers

Sarah Yates provides an interesting and informative comment touching on many relevant issues—please read. My thoughts on a couple of these issues are as follows:

1. Minimizing or maximizing antipsychotic drug administration may be a public health issue, but it is not the concept for treating an individual. The issue is optimizing drug therapy (as well as other therapies). In this regard, the evidence base is very weak and practice is probably very bad. Drugs do not come to market with good evidence for optimal dosing even in the acute situation. The doctor's responsibility will be to develop recommendations at each phase of illness based on relatively little information. And the available data will be on groups of patient subjects in illness phase and other circumstances different from the individual being treated.

2. Optimizing will depend on many things, but decisions on how to approach this are developed in the doctor/patient relationship in the broad context of the recovery process. Here an appreciation of risks and benefits is ascertained with attention to the patient's value system.

3. As a clinical concept, the "no drug" proposition is applicable only if based on an anti-drug ideology. The usual situation should be when and how to use drugs and when and how to be off medication. I would also view the "continuous medication always" as based on ideology.

4. The symptom reduction and relapse prevention effects of dopamine antagonists are valid, and may apply to virtually all patients with schizophrenia and psychosis. The concept of non-responders is flawed, for the patients who have a very inadequate response may be more symptomatic if off drug. And there are not alternative, robust antipsychotic therapies that exclude dopamine antagonists. There is no evidence that those persons doing poorly on drugs would do better without (even adverse effects included in the consideration).

5. Classifying drugs as first- and second-generation or typical and atypical is not useful for therapeutic or adverse effect considerations. Generally, efficacy and effectiveness is similar across these drugs (clozapine excepted), but adverse effects vary widely and need to be defined at the individual drug (rather than drug class) level. Minimizing risk by careful selection of drug and dose is the most important effect the doctor's decision can have on the risk/benefit equation.

6. Clinicians need to develop effective collaboration with patients who do not (or will not) consume medication in the recommended manner. Poor adherence is common, and for many reasons. In general, I think clinicians have not effectively engaged in developing alternative strategies, let alone identifying persons who may be good candidates for reducing risks by being off medication (and on a targeted drug strategy). Herz provides some guidelines in his comments in this forum.

Yates brings up many other issues, but I'll leave this discussion with the assertion that key issues in clinical care cannot be anchored in compelling evidence at this time. But the primary question concerning antipsychotic drug treatment for schizophrenia is not yes or no, but how.

View all comments by William CarpenterComment by:  Sarah Yates
Submitted 17 July 2007
Posted 18 July 2007

I welcome much of what William Carpenter has to say, though, as he points out, all of it is up for debate, as there are no certainties. However, I think it a little harsh to state, "As a clinical concept, the ‘no drug’ proposition is applicable only if based on an anti-drug ideology." No drug ever, for anyone under any circumstances, yes: this is an ideology. To raise the legitimate concern that once an individual is started on drugs it may be very hard to ever stop them, at least for some individuals, is not an ideology. It is a pragmatic concern, from which follows the premise that if it is possible to avoid drugs, it might well be a good idea to do so, and the long-term outcome might be better. As discussed previously, there is evidence to suggest this is not such a silly idea as many suppose. Seikkula is not an ideologist in the sense that patients sometimes do go on antipsychotics, on a needs basis.

In my own case, I tapered from 1.5 mgs risperidone to 0.25 mgs every third day, over nearly a year, although clearly I was on a very low dose for most of that time. (You could say most of my treatment was one long taper.) I still experienced marked somatic symptoms on total withdrawal, and this lasted for around 2 weeks. Given the effects of low dose, I am likely a poor metabolizer. Genotyping tests can classify individuals as poor, moderate, or extensive metabolizers, and this can lead to huge differences in therapeutic effect and toxicity of psychotrophics ( Gillman, 2007; Bondy and Spellmann, 2007). Yet another complicating factor to add to the mix.

To my knowledge there is no satisfactory study to show whether poor responders would do better long term off drugs or not. Therefore, it remains a legitimate question.

May I add that I thought Wolfgang Gaebel's post very well referenced and a useful contribution to the discussion. I do not necessarily agree with him on all issues. but it is good to get the data (even if it is flawed to a greater or lesser extent). How he can make his assertions regarding tolerability and safety I do not know. The data do not support him. Omission is not evidence. I referenced some of the recognized problems.

Antipsychotics have terrible side effects. I agree that NICE guidelines are probably the best of the bunch; I do not want to throw the baby out with the bathwater. However, if you look at the discontinuation studies that underpin them, there is almost certainly a biasing of results due to withdrawal effects. Determining the extent of this bias, and how much appropriate tapering might affect it, is largely guesswork. The recommended 2- to 5-year maintenance on drug after a second or subsequent psychotic episode is pretty much a guess. Which is it to be, 2 or 5? At least the guidelines raise the question, presumably answered by an individualized review.


Bondy B, Spellmann I. Pharmacogenetics of antipsychotics: useful for the clinician? Curr Opin Psychiatry. 2007 Mar;20(2):126-30. Review. Abstract

Gillman, K (2007). Cytochrome P450 Enzymes:, accessed 17 July 2007.

View all comments by Sarah YatesComment by:  Cenk Tek
Submitted 26 July 2007
Posted 26 July 2007
  I recommend the Primary Papers

I believe the jury is out to decide if long-term antipsychotic treatment is hazardous or not in terms of brain biology. The new generation of antipsychotics clearly increases the risk for other medical problems, thus the re-emerging interest in targeted antipsychotic treatments.

Nobody likes to use a medication all the time. I found most of my patients to be reluctantly compliant on their medications after many self-trials of no medication periods with fairly dramatic psychosocial consequences.

In my experience, the main determinant of success of targeted antipsychotic treatment is not the severity of symptoms such as delusions, but if and how fast insight is lost. There are clear subgroups where the first sign of a relapse is quick loss of insight, while others may keep partial insight throughout relapse of other symptoms. If the desire, and origins of the desire to go off medications, are well addressed in therapy, the second group does enjoy medication-free periods, if not complete medication-free life, albeit almost always with residual symptoms.

Please note that for younger patients the desire to go off medications often reflects a desire to wish a nasty, lifelong illness away rather than a side effect problem.

Also, perhaps Drs. Carpenter and Kapur can comment on older papers that examined long-term outcomes of depot medications to oral variants, which is integral to this discussion.

View all comments by Cenk Tek

Comments on Related News

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Daniel Weinberger, SRF Advisor
Submitted 18 October 2005
Posted 18 October 2005

The Lieberman et al. CATIE study is a landmark large-scale clinical trial of antipsychotic drug therapy and will generate considerable discussion in the coming months. It offers important insights about real-world treatment of individuals with the diagnosis of schizophrenia, in the sense of typical practices in clinics around the country and the clinical experience of many practitioners. It probably comes as no surprise that the response to available antipsychotic agents is suboptimal and that differences between drugs are not dramatic in many cases.

One of the questions that comes to my mind about the results is whether and to what degree they are generalizable. Do the results of this study accurately characterize the effects of these drugs across the spectrum of patients with chronic schizophrenia who are treated with them? In other words, are the patients in the CATIE trial representative of the patients with chronic schizophrenia who are in need of these medications? I believe there are several indicators to suggest that they may not be. First, of the subjects in this trial, most of whom (75 percent) were male, 40 percent had been or were married. Second, the mean age at first antipsychotic treatment was 26 years. Third, 30 percent of the subjects were on no medication when they entered the trial. These are all somewhat atypical characteristics in my experience, especially for a predominantly male sample.

In the NIMH schizophrenia genetic study that I direct, we have extensively evaluated over 600 subjects with schizophrenia from around the country. In our sample, the mean age at first antipsychotic treatment is 21 and the ever-married rate is 15 percent, and our sample is one-third female. Moreover, less than 10 percent of our sample is unmedicated at the time that they are evaluated. The finding that a mean dose of 20 mg of perphenazine was as effective as other medications also is somewhat surprising in my experience, as having used this drug for many years, I have rarely seen chronic, actively symptomatic patients respond well without dosing around 32 milligrams and above. Is it possible that the CATIE trial inadvertently enrolled patients more in the schizophrenia spectrum end of the distribution of patients receiving these drugs who may tend not to show as clear benefit? Or maybe the size and breadth of the CATIE trial obscured the signal from the more classic patient with schizophrenia for whom antipsychotic treatment is essential.

It will be interesting to see whether other academic schizophrenia centers concur with the demographics of my experience as noted above or those of CATIE. Multicenter studies—and CATIE involved 57 centers each contributing relatively small samples over a 2-year period—are susceptible to dilution effects and to the possibility that the sample is clinically "noisy." It will be interesting to see, when data analyses from the next stages appear, whether differences are found in the results from different centers who participated in the trial. Will CATIE have told the story of how these drugs work in patients who receive them, or will it have failed to identify the signal from the noise?

View all comments by Daniel Weinberger

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Scott Hemby
Submitted 19 October 2005
Posted 19 October 2005
  I recommend the Primary Papers

Related News: CATIE Comes to Surprising Conclusions

Comment by:  David Lewis, SRF Advisor
Submitted 19 October 2005
Posted 19 October 2005
  I recommend the Primary Papers

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Max Schubert
Submitted 19 October 2005
Posted 19 October 2005
  I recommend the Primary Papers

I also have not seen the response at that dose of perphenazine and even the atypical antipsychotics in chronic schizophrenics. In fact, the only medication that seemed to have an adequate "real-life" dose was olanzapine.

View all comments by Max Schubert

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Iulian Iancu
Submitted 20 October 2005
Posted 20 October 2005
  I recommend the Primary Papers

It seems that the doses used are not equivalent, and the researchers have used somewhat lower doses of perphenazine and risperidone (in favor of olanzapine). Thus, it is obvious that perphenazine and risperidone have showed smaller efficacy.

View all comments by Iulian Iancu

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Xiang Zhang
Submitted 20 October 2005
Posted 21 October 2005
  I recommend the Primary Papers

There is evidence that the Chinese traditional medicines may be an alternative approach in the treatment of schizophrenia. Our recent studies indicate that the extraction of gingko biloba may increase the effectiveness of antipsychotic drugs, but reduce their side effects. This finding may provide a new clue to develop a novel therapeutic drug for treatment of schizophrenia.

1. Zhang XY, Zhou DF, Zhang PY, Wu GY, Su JM, Cao LY. A double-blind, placebo-controlled trial of extract of Ginkgo biloba added to haloperidol in treatment-resistant patients with schizophrenia. Journal of Clinical Psychiatry. 2001; 62(11):878-83. Abstract

2. Zhang XY, Zhou DF, Su JM, Zhang PY. The effect of extract of ginkgo biloba added to haloperidol on superoxide dismutase in inpatients with chronic schizophrenia. Journal of Clinical Psychopharmacology 2001;21(1):85-88. Abstract

View all comments by Xiang Zhang

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Alonso Montoya
Submitted 21 October 2005
Posted 21 October 2005
  I recommend the Primary Papers

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Alexander Miller
Submitted 21 October 2005
Posted 21 October 2005
  I recommend the Primary Papers

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Marvin Swartz
Submitted 26 October 2005
Posted 26 October 2005

Reply to Dr. Weinberger's questions about the generalizability of the CATIE sample, by Marvin Swartz, for the CATIE investigators
As CATIE investigators, we have been mindful of concerns about the generalizability of the CATIE sample. In response to a similar concern, our colleague Jeffrey Swanson at Duke compared CATIE participants to a quasi-random sample of 1,413 patients enrolled in the Schizophrenia Care and Assessment Program (SCAP), an observational, non-interventional study of schizophrenia treatment in usual care settings in the United States. The two samples were similar in demographic characteristics, e.g., gender (70 percent male in SCAP, 74 percent male in CATIE), age (mean age = 43 years in SCAP, mean age = 41 years in CATIE), and education (36 percent of SCAP participants had a high school education and 28 percent attended college; in CATIE these percentages were 35 percent and 39 percent, respectively). The CATIE study had a lower proportion of participants from racial minority backgrounds (40 percent vs. 54 percent). The samples also resembled each other in clinical characteristics. Nearly one-third of the patients in both studies had recently been hospitalized. The CATIE sample had slightly higher average scores on psychotic symptom severity than the SCAP patients (mean PANSS total score = 75 vs. 71), and also slightly higher scores on functioning and quality of life (mean Heinrichs-Carpenter QLS score = 63 vs. 57) (Haya Ascher-Svanum, Ph.D., Senior Research Scientist, Eli Lilly and Company; personal communication). These similarities provide some confidence that CATIE’s RCT design did not result in a biased selection of patients.

Thanks for your comments on the CATIE study.

View all comments by Marvin Swartz

Related News: CATIE Comes to Surprising Conclusions

Comment by:  William Carpenter, SRF Advisor (Disclosure)
Submitted 26 October 2005
Posted 26 October 2005

The antipsychotic drugs mainly treat psychosis (in contrast to cognition impairments and primary negative symptoms). In the CATIE study, the drugs tested share the same mechanism of action (D2 antagonism). Clozapine aside, the second-generation drugs (SGA) have not established superior efficacy over first-generation drugs (FGA). The FDA has granted no such claim, and the Cochrane reviews do not support superior antipsychotic efficacy. The appearance of superiority, including the terrific organization of data in the Davis meta-analyses, may be extensively based on last observation carried forward, excessive dose of the FGA, failure to pretreat with anti-parkinsonian drugs, sponsor bias, and a number of other methodological problems including the fact that most study subjects are doing poorly on FGA when recruited into comparative studies. "Atypical antipsychotic" means only low extrapyramidal symptoms at therapeutic dosing. In this regard, the CATIE findings are not surprising, but simply point to the considerable shortfall in effectiveness associated with current treatments. The drugs will vary considerably along side effect liabilities, and matching patient to side effect profile is the key to individualizing drug choice at the moment.

As to time on drug, there was not a long-acting depot arm to the study, and this method should probably be considered in substantially more patients than is the practice in the U.S. Olanzapine did a little better on the time on drug measure, and risperidone was second. This may relate to the fact that these were the two most common drugs used at study onset, so more patients with known tolerability to these drugs began the trial. In any case, concern with weight and the metabolic syndrome will drastically cut the time on drug for olanzapine in current practice.

It is almost impossible to have a level playing field in comparative drug studies, since optimal dosing and individualized dosing parameters are simply little known with most antipsychotic drugs. In this regard, we don't know if quetiapine and ziprasidone would have done better at higher dose; or if risperidone being yoked to olanzapine led to suboptimal dosing in many cases. In Rosenheck's JAMA report, he observed that pretreatment with an anti-parkinsonian drug led to similar effectiveness comparing olanzapine with haloperidol. Would perphenazine have been even better with anti-cholinergic pretreatment?

In my view, this is a critically important study in that it reasonably represents an effectiveness study in typical settings [probably more representative than the Weinberger data set (see Weinberger commentary)] without sponsor bias. As such, it has succeeded in calling public attention to the relative lack of progress associated with "me-too" dopamine blocking antipsychotic drugs. This conclusion is reinforced by the U.K. study reported by Peter Jones at the ICOSR where SGA did not beat FGA on the primary endpoint (quality of life) or on many secondary measures. Another head-on comparison study with public support.

My hope is that industry will devote discovery resources to the challenging problems of novel treatments with new molecular targets addressing problems with impaired cognition and primary negative psychopathology. Refining antipsychotic drugs has not advanced therapeutics much since the introduction of chlorpromazine. Reducing the neuroleptic adverse effects of FGA is a real advance, especially considering the excessive dosing. But significant new liabilities are associated with some of the SGA. We now need to meet the efficacy challenge for the components of schizophrenia that mainly cause poor functional outcomes.

View all comments by William Carpenter

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Daniel Weinberger, SRF Advisor
Submitted 29 October 2005
Posted 30 October 2005

Dr. Swarz's comment providing data from the SCAP study is helpful in confirming that CATIE patients are similar in many phenomenological respects to other patients in schizophrenia treatment programs. Indeed, in terms of PANSS ratings, sex ratios, age at enrollment in the study, and history of recent hospitalizations, CATIE patients are not substantially different from patients we see at the NIH in Bethesda, Maryland and we saw when our program was located at St. Elizabeths Hospital in Washington, D.C. In my comment, I asked specifically about three CATIE characteristics that seemed atypical to me: age at first antipsychotic treatment (26), precentage of patients who were or had been married (40%), and percentage of patients who were unmedicated at the time they volunteered for the study (30%). It would enlighten this discussion if Dr. Swarz would report these data from the SCAP study.

View all comments by Daniel Weinberger

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Robert McClure (Disclosure)
Submitted 31 October 2005
Posted 1 November 2005
  I recommend the Primary Papers

It would be interesting to learn from Dr. Swartz and the CATIE investigators (a) the age at first antipsychotic treatment, (b) the percentage of patients who were or had been married, and (c) the percentage of patients who were unmedicated at the time they volunteered for the study in the SCAP sample. I suspect these three variables, if available, will more closely resemble those of the CATIE trial sample than the CBDB sibling study sample.

Dr. Weinberger has suggested that the CATIE trial inadvertently enrolled patients more in the schizophrenia spectrum end of the distribution, or maybe the size and breadth of the CATIE trial obscured the signal from the more classic patient with schizophrenia, so the results may not be generalizable. I suspect that differences in criteria for recruitment and retention between the CBDB sibling study and the CATIE study explain the differences among the demographic variables of the samples.

The clinical characteristics of the CBDB sibling study sample are what one would expect in a study whose purpose is to find associations between genetic variation and neuroimaging/neuropsychological phenotypes, among affected and unaffected family members. The usual patient included in the CBDB sample probably: had onset of active symptoms in late adolescence or early adulthood (i.e., high school or college age, before many people marry); was started on medications earlier in life; and had more intact nuclear families (parents, siblings, etc.) than the usual CATIE subject. Patients with later onset of illness or milder symptoms (who are more likely to be or have been married) and who did not start on medications once psychotic symptoms occurred, were less compliant with their medications, and/or had fewer intact family relationships were unlikely to successfully travel to Bethesda and complete two full days of research testing. The CATIE recruitment strategy did not exclude the unusual patient with treatment of symptoms later in adulthood, require intact nuclear family, or require compliance with medications at time of study entry.

The CBDB sample better represents a "textbook case" of schizophrenia. Many patients who do meet DSM-IV criteria for schizophrenia may not be good candidates for a genetics study, but may still have schizophrenia and are appropriate candidates for a large clinical study. This would suggest that the findings can be generalized to other groups of patients with the illness, though perhaps not the "classic" cases of schizophrenia gathered in the CBDB study.

View all comments by Robert McClure

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Captain Johann Samuhanand
Submitted 7 November 2005
Posted 7 November 2005

Is there any published evidence that gingko biloba could be useful in containing the side effects of clozapine and other atypicals, or are there studies in progress?

View all comments by Captain Johann Samuhanand

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Xiang Zhang
Submitted 8 November 2005
Posted 9 November 2005
  I recommend the Primary Papers

Reply to comment by Johann Samuhanand
To our best knowledge, there is no published evidence that gingko biloba could be useful in reducing the side effects of clozapine and other atypicals. However, using the same group of patients with schizophrenia as we reported previously (Zhang et al., 2001), our recent study has shown that chronic patients with schizophrenia demonstrated significantly lower CD3+, CD4+, and IL-2 secreting cells, together with CD4/CD8 ratio, than did healthy controls at baseline. After a 12-week treatment, EGb added to haloperidol treatment increased the initially low peripheral CD3+, CD4+, and IL-2 secreting cells, together with CD4/CD8 ratio. There was only a significant increase in CD4+ cells in the placebo plus haloperidol group. These findings suggest that ginkgo biloba may improve the decreased peripheral immune functions in schizophrenia (Zhang et al., 2006).

As we have known, although clozapine is superior over the other drugs in terms of efficacy, it can severely deplete white blood cells, leading to limitations on its use. If gingko biloba may indeed produce beneficial effects on the immune system in schizophrenia, there is a possibility that ginkgo biloba may be useful in reducing the side effects of clozapine, at least in regard to immune function.

On the other hand, a limitation of the design of our previous study (Zhang et al., 2001) is the use of haloperidol as the antipsychotic treatment at a time when atypical antipsychotic drugs are the standard of care. Therefore, a further study is warranted to investigate whether ginkgo biloba shows similar benefits in augmenting the atypical antipsychotics, which already have the capacity to improve the positive and negative symptoms and have better profiles in terms of extrapyramidal side effects.

Zhang XY, Zhou DF, Zhang PY, Wu GY, Su JM, Cao LY. A double-blind, placebo-controlled trial of extract of Ginkgo biloba added to haloperidol in treatment-resistant patients with schizophrenia. Journal of Clinical Psychiatry. 2001; 62(11):878-83. Abstract

Zhang XY, Zhou DF, Cao LY, Wu GY. The effects of Ginkgo biloba extract added to haloperidol on peripheral T-cell subsets in drug-free schizophrenia: a double-blind, placebo-controlled trial. Psychopharmacology 2006 (in press)

View all comments by Xiang Zhang

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Patricia Estani
Submitted 25 November 2005
Posted 25 November 2005
  I recommend the Primary Papers

I recommend this clear and well-written paper for students to understand the basis of the CATIE studies.

I agree with Dr. Weinberger about the variables that could obscure the results in the target population or the schizophrenic population. His remarks about the control conditions or the dissection of the variables in the study are important. The difference between typical and atypical drugs is clear in these data.

New drugs, diferent from the typical and atypical drugs, based on new genetics research and new genetic routes must be developed in order to achieve new successes in the treatment of schizophrenia.

I think that atypical antipsychotics do not mean only low extrapyramidal symptoms at therapeutic doses. Several studies have demonstrated that atypical drugs(especially olanzapine) are better than typical drugs in important characteristics such as cognitive functioning.

View all comments by Patricia Estani

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Mike Irwin
Submitted 29 November 2005
Posted 29 November 2005
  I recommend the Primary Papers

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Patricia Estani
Submitted 13 December 2005
Posted 13 December 2005
  I recommend the Primary Papers

The most important current development of new antipsychotic drugs is focused on two mechanisms, the α7-nicotinic receptor agonists that are good new candidates for the management of the disease (Martin et al., 2004) and, most recently (and I think probably the closest to development), is the one that focuses on glutamatergic neurotransmission (Coyle and Tsai, 2004).

On the other hand, I think that behavioral and cognitive therapy, as well as family support and family management given by a professional in this area of health, are important to ensure an excellent result in schizophrenic patients.

Martin LF, Kem WR, Freedman R. Alpha-7 nicotinic receptor agonists: potential new candidates for the treatment of schizophrenia. Psychopharmacology (Berl). 2004 Jun ;174(1):54-64. Abstract

Coyle JT, Tsai G. The NMDA receptor glycine modulatory site: a therapeutic target for improving cognition and reducing negative symptoms in schizophrenia. Psychopharmacology (Berl). 2004 Jun ;174(1):32-8. Abstract

View all comments by Patricia Estani

Related News: CATIE Comes to Surprising Conclusions

Comment by:  Robert Fisher
Submitted 24 December 2005
Posted 28 December 2005
  I recommend the Primary Papers

[Disclosure: R. Fisher was Study Coordinator, Recruiter, and Diagnostician for the Byerly Group at UT Southwestern CATIE site, the second-largest enrollment site in the study.]

The CATIE study is likely the best designed and implemented research project ever conducted regarding schizophrenia and relevant psychopharmacology. The extensively collected data will have an enormous heuristic value in the study and evaluation of this disorder in all aspects of schizophreinia. I found Drs. Lieberman and McEvoy to be true professionals in this study design.

View all comments by Robert Fisher

Related News: New York Times Reports on Medication Conundrum in Early Episode Schizophrenia Research

Comment by:  Jim Botta
Submitted 24 March 2006
Posted 27 March 2006

In some of the countries where these studies of withholding medication were conducted, there are comprehensive health systems that provide organized support for the individual. The USA does not have this type of coordinated mental health support, even in our large metropolitan areas. Lacking this, it may be unethical to conduct this type of study in this country.

View all comments by Jim Botta

Related News: New York Times Reports on Medication Conundrum in Early Episode Schizophrenia Research

Comment by:  Ron Unger
Submitted 1 April 2006
Posted 3 April 2006

It seems bizarre to me to question the ethics of research utilizing non-medication interventions, when medications themselves are responsible for such incredible damage. Whether it's obesity, diabetes, movement disorders, or just general numbing of the mind, "antipsychotic" medications might best be characterized as a poisoned life raft—possibly useful in a pinch, but nothing we should be relying on if we can possibly avoid it. Jim Botta wants to use the lack of psychosocial support in the U.S. as an excuse to not even research alternatives to medication; of course, failing to do the research will only reinforce the continued over-reliance on medications and absence of alternatives. A wiser suggestion would simply be to insure that participants in such research be provided with adequate support. After all, research with good outcomes for people not utilizing medications (such as Soteria) have already been done in the U.S.; why shouldn't there be more such studies?

Long-term studies show that most people who make strong recoveries cease taking medication at some point, though quitting itself is often a difficult process. It makes sense to look for ways to help many people recover without ever getting dependent on medications.

View all comments by Ron Unger

Related News: Less Is More: Antipsychotic Dose After First-Episode Psychosis

Comment by:  Robin Emsley (Disclosure)
Submitted 2 August 2013
Posted 7 August 2013
  I recommend the Primary Papers

The article by Wunderink et al., like the original paper, is thoughtful and informative. However, the results are a bit surprising and will hopefully not encourage clinicians to discontinue maintenance antipsychotic treatment. My own view is that we should be taking relapse very seriously, as it may play a critical role in disease progression, in addition to the obvious associated psychosocial risks. An earlier Dutch cohort study with a 15-year follow-up found accruing morbidity and identified relapse as a critical factor. After each relapse, a subset of patients experienced persistence of positive and negative symptoms (Wiersma et al., 1998).

Given the very strong association between treatment discontinuation and relapse (Robinson et al., 1999), our inability to accurately identify early warning signs of relapse, and the failure of rescue medication to abort relapse, together with additional evidence for emergent treatment refractoriness in a subset of patients after relapse (Emsley et al., 2013), long-term maintenance treatment seems indicated.


Wiersma D, Nienhuis FJ, Slooff CJ, Giel R. Natural course of schizophrenic disorders: a 15-year followup of a Dutch incidence cohort. Schizophr Bull . 1998 ; 24(1):75-85. Abstract

Robinson D, Woerner MG, Alvir JM, Bilder R, Goldman R, Geisler S, Koreen A, Sheitman B, Chakos M, Mayerhoff D, Lieberman JA. Predictors of relapse following response from a first episode of schizophrenia or schizoaffective disorder. Arch Gen Psychiatry . 1999 Mar ; 56(3):241-7. Abstract

Emsley R, Chiliza B, Asmal L. The evidence for illness progression after relapse in schizophrenia. Schizophr Res . 2013 Aug ; 148(1-3):117-21. Abstract

View all comments by Robin Emsley

Related News: Less Is More: Antipsychotic Dose After First-Episode Psychosis

Comment by:  Martin HarrowThomas JobeRobert Faull
Submitted 14 August 2013
Posted 14 August 2013

While the empirical data on short-term studies of antipsychotics has produced considerable positive results, there is very little evidence on the long-term (four years or longer) efficacy of antipsychotics. Leucht and colleagues have noted “nothing is known about the effects of antipsychotic drugs compared to placebo after three years….” (Leucht et al., 2012).

Since multiyear (four years or longer) double-blind, drug-placebo studies of schizophrenia cannot be done, the relative lack of firm, long-term evidence on antipsychotic medications makes it more important that evidence from other types of studies be considered more carefully. This includes the long-term outcome of lower-dose and non-medicated schizophrenia patients from the seven-year Wunderink study and our 15- to 20-year naturalistic research finding more favorable long-term outcomes for unmedicated schizophrenia patients than those continuously prescribed antipsychotics (Harrow and Jobe, 2007; Harrow et al., 2012; Harrow and Jobe, 2013). The Wunderink research involves comparison of maintenance antipsychotic treatment with a randomly assigned dose reduction/discontinuation sample of patients with schizophrenia. The findings in favor of the lower dose/discontinuation sample from this research raise major questions about the efficacy of antipsychotics when they are used for long-term treatment.

Similar results emerged from multiple follow-ups across 15-20 years comparing our longitudinal sample of schizophrenia patients continuously prescribed (over the 15 to 20 years) antipsychotics with those not on antipsychotics for the last 15-20 years. In this research, the non-medicated patients functioned significantly better after the first two-year follow-up, and this continued thereafter, with these results also raising major questions about the longitudinal treatment of schizophrenia.

In addition, for schizophrenia patients on continuous antipsychotic medication, we found low rates of recovery (both functional and complete symptomatic recovery for a year or longer were required for the rating of recovery), a number of rehospitalizations, poor work functioning, and a surprising frequent presence of disrupted functioning. The differences between our sample of unmedicated patients with schizophrenia and our continuously medicated patients were significant starting at the 4.5-year follow-up and continuing up to 20 years. In general, prognostic measures we used indicted our unmedicated schizophrenia patients had a better prognosis, but even when this was controlled for, our continually prescribed patients had poorer outcomes. This longitudinal research raises issues of whether long-term treatment of schizophrenia with antipsychotics facilitates recovery and whether all, some, or very few schizophrenia patients benefit from a treatment plan involving continuous use of antipsychotics throughout their lifetimes.

In regard to the long-term evidence on the efficacy of antipsychotic medications, both the lower dose/discontinuation comparison with maintenance antipsychotics in the research of Wunderink and colleagues and our research comparing schizophrenia patients prescribed antipsychotics for 20 years with those not on antipsychotics for 20 years show a striking similarity. Both of these longitudinal programs found that the largest and most significant differences occurred on a long-term basis after the first two years.

While both of the above long-term research programs show samples of unmedicated or low-medicated schizophrenia patients in recovery, they are not unique. Research on other long-term samples from the Chestnut Lodge Study by Fenton and McGlashan (Fenton and McGlashan, 1987), the Vermont Long-Term Study by Harding (Harding et al., 1987), the World Health Organization Study and the DOSMED by Jablensky and many others (Jablensky and Sartorius, 2008), and M. Bleuler’s longitudinal sample (1978) all point to long-term samples of schizophrenia patients who show recovery. In our long-term 20-year sample, the periods of complete recovery varied in length, with some unmedicated schizophrenia patients showing complete recovery for a few years and others for a much longer period. In addition, schizophrenia patients who were continuously prescribed antipsychotics showed significantly fewer periods of complete recovery than those not prescribed antipsychotics at any follow-up (p <.001).

Overall, based on multiple samples of schizophrenia patients from different countries, the positive evidence on some unmedicated samples of schizophrenia patients showing long-term favorable outcomes was overwhelming. The questions, of course, are what types of schizophrenia patients show favorable outcomes while unmedicated, how large a percentage show this, and, as McGorry and others have noted, what other types of interventions besides medications are helpful. In non-psychiatric areas, the original effectiveness of some medications continues for many years, but for other medications the body readjusts and they become less effective over time. The above studies alert us to the issue of whether multiyear treatment of schizophrenia patients with antipsychotics is helpful or harmful to many of these patients and to which ones. This important issue is still to be resolved with studies lasting longer than two to three years.


Bleuler M. 1978. The Schizophrenic Disorders: Long-Term Patient and Family Studies. New Haven: Yale University Press.

Fenton WS, McGlashan TH. Sustained remission in drug-free schizophrenic patients. Am J Psychiatry. 1987 Oct; 144(10):1306-9. Abstract

Harding CM, Brooks GW, Ashikaga T, Strauss JS, Breier A. The Vermont longitudinal study of persons with severe mental illness, II: Long-term outcome of subjects who retrospectively met DSM-III criteria for schizophrenia. Am J Psychiatry. 1987 Jun; 144(6):727-35. Abstract

Harrow M, Jobe TH. Factors involved in outcome and recovery in schizophrenia patients not on antipsychotic medications: a 15-year multifollow-up study. J Nerv Ment Dis. 2007 May; 195(5):406-14. Abstract

Harrow M, Jobe TH, Faull RN. Do all schizophrenia patients need antipsychotic treatment continuously throughout their lifetime? A 20-year longitudinal study. Psychol Med. 2012 Oct; 42(10):2145-55. Abstract

Harrow M, Jobe TH. Does Long-Term Treatment of Schizophrenia With Antipsychotic Medications Facilitate Recovery? Schizophr Bull. 2013 Mar 19. Abstract

Jablensky A, Sartorius N. What did the WHO studies really find? Schizophr Bull. 2008 Mar; 34(2):253-5. Abstract

Leucht S, Tardy M, Komossa K, Heres S, Kissling W, Davis JM. Maintenance treatment with antipsychotic drugs for schizophrenia. Cochrane Database Syst Rev. 2012; 5:CD008016. Abstract

View all comments by Martin Harrow
View all comments by Thomas Jobe
View all comments by Robert Faull

Related News: Less Is More: Antipsychotic Dose After First-Episode Psychosis

Comment by:  Stephen Marder, SRF Advisor
Submitted 16 August 2013
Posted 16 August 2013

This is an important contribution that has a number of important implications:

First, outcomes such as relapse and remission seem very crude to me, and they tell very little about what is of interest to many people with schizophrenia. An advantage of this study is that the authors used a sophisticated functional measure. I agree with Cenk Tek that patients who have a recurrence when the dose of antipsychotic is too low or medications have been stopped can become better managers of their illnesses (see Cenk Tek's full comment). I also believe that one of the goals of early treatment should be to educate patients to self-manage symptoms when they occur so that a minor exacerbation can be successfully managed.

Second, this study reminds us of an older literature from the 1980s, including my work (Marder et al., 1984), which documented that living with excessive dopamine blockade can be a personal burden that is associated with increased ratings of anxiety and depression that interfere with functioning. Other burdens of higher antipsychotic doses can include weight gain and sedation, which may interfere with adjustment.

Third, I agree with Will Carpenter that treatment of these individuals is likely to be optimal when patients and their treatment providers make decisions about drug doses and drug strategies in a collaborative atmosphere (see William Carpenter's full comment). According to Wunderink and co-authors, optimal treatment was more likely to occur in the drug reduction group.


Marder SR, Van Putten T, Mintz J, McKenzie J, Lebell M, Faltico G, May PR. Costs and benefits of two doses of fluphenazine. Arch Gen Psychiatry . 1984 Nov ; 41(11):1025-9. Abstract

View all comments by Stephen Marder

Related News: Less Is More: Antipsychotic Dose After First-Episode Psychosis

Comment by:  Clive AdamsStephanie Sampson
Submitted 14 August 2013
Posted 20 August 2013

One swallow does not necessarily indicate summer has arrived.

Shortly after publishing a Cochrane review on intermittent drug techniques for schizophrenia (Sampson et al., 2013), we were pleased to see Lex Wunderink and colleagues publish these new findings from their five-year follow-up. This current review includes the initial data of Wunderink et al. from the two-year study demonstrating double the relapse rates in their dose-reduction strategy relative to the maintenance treatment group. These data must be considered within the totality of all similar evidence, or there is risk of overextrapolating results from small trials. However, along with five other important and pioneering trials, the two-year findings by Wunderink et al. are entirely consistent (I2 = 0 percent) with other results (RR Relapse 2.46 95 percent CI 1.7-3.5), with Wunderink and colleagues' data contributing 42 percent of the weight to the finding. At around two years, maintenance treatment does seem to avoid more relapses than a variety of intermittent drug strategies.

The review also suggests that intermittent treatment is significantly better than placebo in terms of relapse rates (2 RCTs, n = 290, RR 0.37 95 percent CI 0.24-0.58). This relates to Dr. Carpenter’s valid point on the earlier report from Wunderink et al. (see Carpenter's comment) that for many reasons, poor adherence to antipsychotic medication is common and that greater consideration of the types of people who may well benefit from a "targeted drug strategy" is required between clinicians and their patients. Although NICE guidelines state that low-dose prescribing and use of intermittent dosing strategies may well minimize side effects in the long-term use of antipsychotic medication, the risk of symptom worsening and relapse is stated to outweigh any benefits of the intermittent techniques "in the long term" (NICE, 2010).

Wunderink and colleagues' welcome new data, however, help bridge a gap in understanding the longer effects of such intermittent techniques on recovery after first-episode psychosis. It is good to see this long-neglected important area of research active after so long in hibernation. But, as Gaebel et al. and Wunderink et al. help us emerge from the permafrost, it is important to remember that Wunderink and colleagues' remarkable achievement in following people for so long is still results based on only 103 people. Psychiatry, blinking in the sunlight, has so often followed a good result full of hope. If replication is impractical, caution in consideration is vital—or the first swallow of summer may turn out to be the last hawk of winter.


NICE 2010. The NICE guideline on core interventions in the treatment and management of schizophrenia in adults in primary and secondary care. National Clinical Guideline: 82. Updated edition. London, Leicester: The British Psychological Society and The Royal College of Psychiatrists, British Library, Stanley Hunt, 2010. Accessed August 14, 2013.

Sampson S, Mansour M, Maayan N, Soares-Weiser K, Adams CE. Intermittent drug techniques for schizophrenia. Cochrane Database Syst Rev. 2013;7:CD006196. Abstract

View all comments by Clive Adams
View all comments by Stephanie Sampson