Evidence-based treatment is today’s clarion call for psychiatry. At its best, this standard leads to rejection or acceptance of treatments on the basis of scientific documentation of efficacy. In schizophrenia, this standard has almost eliminated treatment with psychotherapy to the exclusion of pharmacotherapy, and exploratory/insight-oriented psychotherapy and treatment based on psychogenic etiology have given way to psychosocial treatments with scientific documentation of efficacy (1–3). But the evidence-based standard has not created the political clout (e.g., service system finance) or professional clout (e.g., critically informed mental health professionals) to assure broad application of proven treatments (4, 5).
Clinical trials, the primary source of data for evidence-based therapeutics, are routinely available for pharmacotherapy. Food and Drug Administration regulatory procedures assure that drugs approved for the treatment of schizophrenia have documentation of efficacy. What follows is complicated. The physician must decide which drug, at what dose, in what combination, for what period of time, for which aspects of illness, whether to use off-label, and at what risk and cost. This decision is case specific, is modified over time, and must weigh patient wishes and alternative approaches. Empirical data are inadequate.
There is an especially large gap between clinical trial data and treatment recommendations for first-episode schizophrenia. In this issue, Gitlin et al. now add important descriptive data to the three clinical trials addressing treatment following an initial episode of psychosis (6–8). Research on medication withdrawal is crucial, for virtually all first-episode patients will have time without medication. Rates of patient nonadherence are very high, even with new-generation antipsychotic drugs, and it is unwise to base clinical care on the presumption that a young person with an initial psychotic experience will receive continuous drug therapy for the next 50 years or so. Presently there are far too few data from clinical investigations informing doctors and patients of the consequences and best management procedures for medication withdrawal. In addition to a paucity of data, another problem is that clinical trials are concerned with group effects, and what is best for the group may not be best for the individual. Clinical trials rely on narrowly defined outcome measures in selected patients, while the treating doctor must consider all aspects of each case without exclusion of complicated cases or those with uncertain diagnoses. Drug beats placebo for relapse prevention in cohorts of first-episode patients, but the physician may, for example, have a patient whose psychosis responded rapidly to treatment and who has remained clinically stable for 6 months. This patient dislikes weight gain, attributes emotional dullness to antipsychotic medication, feels depressed, and does not wish to continue taking medication. Diagnosis is difficult in single-episode cases. What is the evidence-based recommendation?
Not long ago, experts would argue along two lines. First, psychosis is devastating and may be neurotoxic, and relapse prevention trumps all other considerations. Or, second, the minority of patients who do not have a relapsing form of illness must be identified and not exposed to long-term risks of medication. The other patients will relapse sooner or later in any case, and we have better treatment for psychosis than we do for dyskinesia. The former position never dealt with the implication of life-long antipsychotic drug treatment for some individuals who may not have a relapsing illness, and the latter position did not deal with medication withdrawal in cases for which clinical monitoring was inadequate. Neither dealt with the fact that one patient may have insight, be cooperative, and respond quickly to drug treatment while another may have had a rapid onset, be hostile and aggressive, and resist treatment. This state of discourse led to the equivocal 1997 APA practice guideline stating that "a patient who has had only one episode of positive symptoms and has had no symptoms during the following year of maintenance therapy may be considered for a trial period without medication" (9, p. 41).
The report by Gitlin et al. advances knowledge on pharmacotherapy of first-episode patients with schizophrenia.
1. Long-term follow-up studies (10–15) previously suggested that 15%–25% of single-episode patients would not relapse within 5 years. Gitlin et al. used a sensitive indicator of exacerbation and found that all patients showed indications of recurrence. There are two profoundly important conclusions. First, of patients clearly meeting criteria for schizophrenia, there does not appear to be a substantial minority who would maintain recovery indefinitely without medication. Therefore, the benefit of continued drug treatment appears to apply to most patients. Second, mild exacerbation can be routinely detected and effectively treated, preventing severe relapse in most cases. Therefore, the risk side of medication withdrawal is substantially reduced in most cases.
2. Antipsychotic drugs are effective for psychosis but are not broadly antischizophrenic in therapeutic action. The new generation of drugs are less likely to cause negative symptoms and cognitive impairments (compared to haloperidol in most studies), but efficacy for primary negative symptoms has not been documented (16), and controlled-study evidence for pro-cognitive efficacy (rather than reduced adverse effects) is modest (17). These aspects of schizophrenia mainly account for reduced quality of life and functional outcomes. The benefit of maintenance drug treatment is relapse prevention, not comprehensive treatment of schizophrenia.
3. New-generation antipsychotic medications provide greater variation in risks and benefits, increasing the importance of individualized therapeutic recommendations. On the benefit side, the course of depression or of hostility/aggression appears more favorable with the new-generation antipsychotic medications. On the risk side of these medications, tardive dyskinesia liability is low, but weight gain and diabetic and cardiovascular risk are often greater. The doctor must tailor her or his recommendation to the very particular circumstances of each case. In the preceding example, a new-generation drug at a low dose might be selected to reduce negative subjective experience and tardive dyskinesia risk and to improve the course of depression. Given smoking and a sedate lifestyle, selecting a drug that is not likely to increase weight or diabetic and cardiovascular risk is important.
4. The risk of prolonged and serious adverse effects of psychosis per se (e.g., neurotoxicity) has not been substantiated in medication withdrawal studies (18). The study by Gitlin et al. demonstrates the effectiveness of clinical monitoring and intervention, substantiating a clinical method for minimizing risk from medication withdrawal.
Despite an exacerbation/relapse rate of 96% within 2 years following medication withdrawal, Gitlin et al. do not recommend indefinite antipsychotic maintenance for all patients. Their rationale overlaps with the considerations listed here and also calls attention to social stigma, financial cost, and the patient’s attitude toward the illness and drug treatment. They wisely note that clinically supervised medication withdrawal is safer than covert nonadherence, and a return of symptoms without medication may establish a stronger doctor-patient collaboration for drug treatment in the long term. The choice is not between drug or no drug, but between continuous prophylactic and targeted, rapid-intervention drug treatment (19, 20).
The description of medication withdrawal effects by Gitlin et al. for this cohort of first-episode patients is generously informative, and their thoughts regarding the implications for clinical care are complex and wise. Clinicians make decisions for each individual patient. Group results from study cohorts are invaluable for primary treatment efficacy questions, but evidence-based therapeutics depend on knowledgeable and wise clinicians to translate data into individualized treatment.
Address reprint requests to Dr. Carpenter, Maryland Psychiatric Research Center, P.O. Box 21247, Catonsville, MD 21228-0747; firstname.lastname@example.org (e-mail). Supported by NIMH grant MH-40279.