In providing a scientific basis for the use of antipsychotic drugs in ‘acute’ schizophrenia, the NIMH/VA Collaborative Study comparing the efficacy of three phenothiazines of differing chemical type (chlorpromazine, thioridazine, trifluoperazine) cannot be bettered.⁽⁴⁾ Although its conclusions referred specifically to phenothiazines, subsequent research justifies the generalization of its findings to ‘antipsychotics’, making this elegant study relevant still.

Five main points emerged:

1 Antipsychotics produce a significantly greater improvement in patients with acute schizophrenic symptomatology than placebo. This action on the positive (or ‘acute’) symptomatology is the primary class action and is beyond doubt. Reviewing the first two decades of antipsychotic use, Davis and Garver⁽⁵⁾ found that in 86 per cent of controlled studies chlorpromazine was superior to placebo and that all 26 trials utilizing more than 500 mg/day reported definitely greater benefit than placebo. In this dose range no trials found only marginal benefits. A similar view can be drawn regarding the efficacy of licensed second-generation antipsychotics.

LEARNING POINT: The antipsychotic effect of antipsychotic drugs (or the magnitude of that effect) is not shared by other types of psychotropic agent such as sedatives, is not in need of further replication, and forms a valid basis for classification.

2 Antipsychotics of (three) different chemical types do not differ in efficacy. The conclusion that all antipsychotics have comparable ‘acute’ efficacy is not to say that clinically there is nothing to choose between them. Although sharing a primary class action, antipsychotics encompass diverse pharmacologies that in practice can overshadow the action they share in common. There are therefore many ways of viewing class members apart from the traditional subgrouping by chemical structure—e.g. clinical adverse effect profile, subdivided into general or extrapyramidal tolerability; pharmacodynamically in terms of potency, receptorbinding profile or binding characteristics (‘loose’ or ‘tight’), etc. This is a large ‘extended’ family!

3 A substantial proportion of patients show limited or no response to antipsychotic drugs. Overall, 61 per cent of those on active drug were considered ‘improved’ to ‘very much improved’.⁽⁴⁾ It has usually been agreed that standard drugs produce a satisfactory response in approximately 60-70 per cent of patients with ‘acute’ schizophrenia, while approximately 6-8 per cent do not respond at all.⁽⁶⁾ However, if one includes placebo response rates a less favourable picture emerges, suggesting that significant benefits could be more realistically expected in approximately 40 per cent of patients (Table 4.3.8.1).

Not all those non-responders could be classified as ‘treatment resistant’ (this concept, now operationally defined, long post-dated the Collaborative Study), but it does mean that in the mixture of first episode and relapsed patients presenting in routine practice, expectations for ‘acute’ (i.e. ‘positive’) symptom resolution must be realistic and cannot necessarily be improved by increasingly aggressive drug treatments alone.

LEARNING POINT: While providing a necessary foundation for clinical improvement to acute psychotic symptomatology, expectations of antipsychotic drugs on this single domain must be realistic. Treatment objectives should be correspondingly broad-based.

4 A rapid phase (over the first week) of improvement is followed by a slower, more protracted pattern of improvement (over several weeks). This finding has recently been replicated but with a somewhat different interpretation to that traditionally adopted. In a review of placebo-controlled studies, Agid et al.⁽⁷⁾ also found that standardized ratings of positive symptomatology began to decline in the first week and improvement was significantly greater in the first 2 weeks of treatment than in the subsequent two. This was interpreted as rejecting the conventional view that antipsychotic onset is delayed. In an early clinical test of the dopamine hypothesis, Johnstone et al.⁽⁸⁾ compared the two isomers of flupenthixol (flupentixol), one a dopamine blocker, the other not. They found that antipsychotic efficacy between the two groups, although evident in the first week, only became significant in the third week, often taken as supportive of the delayed onset hypothesis. A compromise between these two viewpoints comes from other data from this study, which showed that the first week of treatment
was also a time of significant improvement in non-specific symptomatology, especially anxiety, in those exposed to the antidopaminergic isomer.⁽⁹⁾

LEARNING POINT: While early improvement on antipsychotics may comprise a component of primary efficacy, such early benefits should be interpreted circumspectly in treatment planning, especially dose modifications, as they may reflect mainly non-specific changes.

5 Patients in placebo arms of antipsychotic drugs trials can—and do—show improvements in positive symptomatology. This is a standard finding in clinical trials often overlooked in routine practice. Non-specific benefits that can accrue from the environment where treatment is undertaken can contribute to maximizing outcomes.

Despite their limitations, antipsychotics are unquestionably the key element in acute treatment. In a unique study, May et al. compared response rates and outcomes in a large group of schizophrenic patients randomly assigned to one of five regimes: individual psychotherapy, antipsychotics, individual psychotherapy plus antipsychotics, ECT, and ‘milieu’ therapy (ward environment). Patients who received physical treatments did better with increased rates of discharge, reduced lengths of stay, and decreased need for additional treatments.⁽¹⁰⁾ Two years after discharge, twice as many of those treated with antipsychotics as with psychotherapy were in employment, while in the 3 years post-discharge, drug-treated patients spent less time back in hospital.⁽¹¹⁾

The issue of whether antipsychotics exert efficacy on negative schizophrenic features rose to prominence in the 1980s as a result of the implications of Crow’s Type 1/Type 2 hypothesis.⁽¹²⁾ The inference of this is that efficacy is ‘unlikely’ owing to different pathophysiological substrates proposed for ‘positive’ and ‘negative’ states. Controversy has persisted, with claims that whatever the case may be for first-generation drugs, new generation antipsychotics do possess this action. Recent reviews, however, have been increasingly circumspect.⁽¹³⁾

One consequences of testing Crow’s proposal was that ‘negative’ schizophrenic symptomatology came to be viewed as something ratable cross-sectionally. This was a radical shift for traditionally the ‘defect state’ was conceptualized as a set of phenomena evident mainly from longitudinal appraisal, relating to complex behavioural signs such as psychosocial and occupational functioning. This switch assumed that the varied states that can underlie ‘negative’ presentations can be distinguished using cross-sectional clinical means alone. This assumption has never been validated and seems to push at the limits of clinical examination.

A major confound in assessing trial data in this field is the bradykinesia of drug-related parkinsonism, a common, pervasive feature in those exposed to antipsychotics yet one whose boundaries remain poorly defined, especially for its subjective components.⁽¹⁴⁾ Changing to drugs or doses with lower extrapyramidal liability, or improving psychotic phenomena that may underlie social withdrawal is a therapeutic action of sorts, but hardly addresses the question. Likewise, utilizing elaborate statistics (e.g. path analysis) to support efficacy for second-generation antipsychotics ignores the fact that the problem is clinical attribution, not analysis.

Carpenter and colleagues⁽¹⁵⁾ raised awareness of this issue by emphasizing the varied clinical states that can present ‘negatively’, thereby introducing a differential diagnosis (Table 4.3.8.2). Furthermore, by emphasizing ‘durability’ in their definition of ‘deficit’ syndrome,⁽¹⁶⁾ they reintroduced a longitudinal component, aligning this more with the historical concept of schizophrenic negativity. Interestingly, this group found clozapine’s benefits on negative features to be confined to patients who did not conform to criteria for ‘deficit’ state.⁽¹⁷⁾

One cannot help wondering whether the quest for antipsychotic ‘therapy’ of negative states has been a wild goose chase, predicated on assessment that was reliable—but lacking validity. This view will be controversial, as will our conclusion that Crow’s hypothesis has not yet been disproven—that it remains to be shown that antipsychotics as a class exert any therapeutic benefits on primary negative schizophrenic symptomatology.

The importance of cognitive deficits underlying the overt symptomatology of schizophrenia is being increasingly highlighted. They comprise a valid endophenotype in predisposed individuals, general cognition suffers a decline in the shift to florid illness and specific cognitive impairments, especially in executive function and memory, may relate to structural changes in specific brain areas.⁽¹⁸⁾ The pharmacological question is whether drug treatments can enhance cognitive performance and thereby promote benefits in other domains.

Evaluation of cognitive actions of antipsychotics faces major methodological problems and findings remain contradictory. Standard antipsychotics initially impair aspects of attention and motor behaviour which improve following continued exposure, though working memory and long-term recall do not appear to be fundamentally affected.⁽¹⁹⁾ Paced performance tests tend to be affected while those that are untimed tend to be insensitive, perhaps reflecting subtle motor effects.⁽²⁰⁾

While new generation drugs may be associated with marginally less cognitive impairment than standard drugs, data are inadequate for firm conclusions.⁽²¹⁾ In the absence of consistent evidence of differential effects on a range of neuropsychological tests, it has been suggested that measurements of ‘social competence’ may represent a more appropriate target for study.⁽²²⁾

Furthermore, in studies evaluating cognitive-enhancing agents as ‘add on’ therapy no consistent evidence of utility has emerged.^(21,23)

Affective symptomatology is prominent in schizophrenia but the efficacy of antipsychotics on such features has received little attention.
Their utility in the treatment of anxiety and other manifestations of ‘arousal’ in acute episodes of illness has not been systematically addressed and rests largely on clinical wisdom.

There is however, a tradition in Continental Europe of attributing antidepressant actions to low-dose antipsychotics^(24,25) (e.g. flupentixol, sulpiride L-enantiomer, amisulpride) when they may exert preferential actions at presynaptic (autoreceptor) dopaminergic sites. Data remain inconclusive but it is unlikely that such an action would be clinically useful, as presynaptic selectivity is lost at doses usually required for antipsychotic efficacy.

Depression is a common feature of untreated schizophrenia and resolves as positive psychotic phenomena diminish.⁽²⁶⁾ This probably does not reflect an antidepressant action but symptom covariation. Of greater concern is what was formerly referred to as the ‘depressogenic’ action of antipsychotics.⁽²⁷⁾ This again raises the difficulty of distinguishing between similar presentations of pathophysiologically different states. Van Putten and May described a dysphoric mood state in antipsychotic-treated patients (‘akinetic depression’),⁽²⁸⁾ which resolved following administration of anticholinergic. This was most likely a subjective manifestation of bradykinesia.

Behaviour can be variously disturbed in schizophrenia, though is seldom considered other than as part of a global assessment. While certain confrontational behaviours such as hostility, belligerence, and resistiveness do improve with antipsychotics,⁽²⁹⁾ this is usually attributed to improvement in positive symptoms. There is, however, evidence that certain types of behavioural disorder correlate with negative, not positive, features and may represent a distinct domain of disorder.⁽³⁰⁾ It seems likely that certain manifestations of behavioural disorganization represent independent dimensions of pathology with their own, predominantly negative, prognostic implications.⁽³¹⁾

With mood disorders, ‘relapse’ (exacerbation of an ongoing episode) and ‘recurrence’ (emergence of a new episode) are well-defined. In schizophrenia, where full remission of acute symptomatology may not be a realistic treatment goal, the distinction is less clear. As the long-term aim is usually minimizing the likelihood of florid exacerbation in a disorder characterized by persisting symptomatology, the term ‘maintenance’ is preferable to ‘prophylaxis’.

The efficacy of antipsychotic drugs in long-term maintenance of schizophrenic illness is beyond doubt.⁽³²⁾ This applies to first-episode patients and to those who have suffered multiple episodes. Nonetheless, it remains difficult to quantify the effect as published figures vary widely. Reviewing relevant trials (covering variable follow-up intervals), Janicak et al. concluded that on average 55 per cent of those on placebo relapsed compared to 21 per cent on active medication, providing overwhelming statistical support for the maintenance effect.⁽³²⁾ In qualification, maintenance studies tend to be biased towards patients who have already shown a degree of response and it is likely that the magnitude of this effect, at least over 12-24 months, is less than trial-based analyses suggest.

A crucial question for clinicians is how long maintenance medication should be continued. The evidence is clear. Relapse rates have been shown to be similar following cessation in groups maintained well for differing lengths of time, from months to years.⁽³³⁾ Furthermore, no difference in relapse has been found in those who responded well compared to those whose response was less good.⁽³⁴⁾ The implication is that, no matter the duration or the quality of well-being on antipsychotic maintenance, relapse is inevitable following discontinuation in those with an established relapsing-remitting illness pattern (i.e. two or more episodes), which comprise the majority.

Not only is relapse a characteristic inherent to these illnesses, it seems so too is time to relapse. Davis et al. showed that placebo relapses plotted over a 2-year period occurred along an exponential line, indicating a constant rate of relapse, calculated from pooled data at a steady 11.5 per cent per month.⁽³⁵⁾

Thus, the clinician’s position is clear—for maintenance of wellbeing following second or subsequent acute episodes, ‘long-term’ antipsychotic maintenance means ‘lifelong’. Should patients decide to discontinue, past experience can offer an invaluable tool in predicting when relapse is likely.

Some patients see long-term maintenance as ‘well-being’ only of sorts, in which quality-of-life is unacceptably impaired. In such individuals, targeted intervention, where treatment is focused on prodromal relapse symptomatology, has intuitive appeal. Alas, the trial evidence does not support this as a general strategy. No controlled studies so far have found advantage in targeted intervention and in a meta-analysis, Davis et al. calculated 25 per cent relapse rates in those continuously treated, rising to 50 per cent in the targeted group.⁽³⁶⁾ While carefully selected individuals, who can work with family and psychiatrists, may prefer this approach, a further potential concern is that intermittent exposure to antipsychotics may increase liability to tardive dyskinesia.⁽¹⁴⁾

Intermittent treatment is, of course, a feature of poor compliance (also known as ‘adherence’ or ‘concordance’), itself perhaps the major contributor to relapse. In terms of major medical events, antipsychotics have a highly favourable risk:benefit ratio but in terms of medically trivial but unpleasant, intrusive adverse effects, the risk:benefit ratio is unfavourable, something often overlooked. Ensuring a maximally effective and tolerable maintenance regime is a joint exercise, from which the doctor cannot be excused.

The evidence that depot formulations enhance compliance is strong, if largely indirect. Support comes from ‘mirror image’ studies, where time in hospital is compared prior to and after starting depot. Six such studies were unanimous in showing substantial reductions in time spent in the hospital after switching to depot (average reduction: 77.8 per cent).⁽³⁶⁾ There is nothing to suggest that this reflects additional therapeutic advantage inherent to depots, whose benefit lies simply in facilitating regular administration and objective monitoring of compliance. Long-acting injectable risperidone (not a ‘depot’ in the traditional sense) has so far received favourable assessment⁽³⁷⁾ but it is too early to provide head-to-head comparative data with conventionally formulated depots. Dosages are also an important consideration in maintenance, as those compatible with maximizing long-term tolerability and well-being are likely to be considerably lower than those necessary for acute symptom control. However, the data are insufficiently clear to provide specific guidance. Kane et al. showed that relapse rates were significantly higher in patients receiving fluphenazine decanoate in a dose of 1.25 to 5 mg two weekly compared to those receiving 25 mg,⁽³⁸⁾ yet Baldessarini et al. have calculated that in long-term treatment, half-maximal effective doses (ED50) may be as low as one-fifth to one-tenth those normally employed.⁽³⁹⁾ In the absence of
specifics, general principles must suffice. Kane and colleagues also showed that while relapse was more likely on low-dose regimes, neurological tolerability and psychosocial/quality-of-life parameters were superior.⁽³⁸⁾ Gradual pursuit of the minimal effective dose is all that can be recommended. Bearing in mind the exponential pattern of relapses, with a modal point at 3-5 months,⁽⁴⁰⁾ ‘gradual’ should equate to decrements at intervals of months, not weeks.

The limitations of antipsychotic treatment in schizophrenia have been known since the 1960s but the concept of ‘treatment resistance’ only sprang to prominence with publication of the US multicentre clozapine study.⁽⁴¹⁾ Within an operationally defined framework of ‘resistance’ (failure to respond to at least two antipsychotics of different chemical type administered in adequate dose for a minimum of 8 weeks), Kane et al. showed that 30 per cent of those on clozapine improved, while only 4 per cent on chlorpromazine/ benztropine did likewise (P < 0.001).⁽⁴¹⁾

This has been interpreted as proving that clozapine possesses superior efficacy over standard agents. While this is one interpretation, it is not the only one. In this study, a chlorpromazine: clozapine dose equivalence of 2:1 was assumed, which might have disadvantaged chlorpromazine, allowing the interpretation that clozapine’s advantage lies in its unique neurological tolerability. However, many other studies have confirmed clozapine’s edge in patients who fail to respond satisfactorily to other antipsychotics and whatever the explanation, this is real added benefit. It does not mean however, that clozapine is a ‘miracle’ drug. While 30 per cent of such patients improving is welcome, the criteria for ‘improvement’ here were modest and subsequent review has failed to show substantial long-term benefits in higher level functioning, such as occupational ability.⁽⁴²⁾ Overall expectations of clozapine, a potentially difficult drug to administer, and for patients to tolerate, must be realistic.

Similar benefit has been claimed for other second-generation antipsychotics. However, there is no conclusive evidence that such advantages can be attributed to any other antipsychotic agent.

As striking as the decline in dynamic psychotherapies over the past two decades has been the rise of cognitive behaviour therapy (CBT) as a clinical and research focus across psychiatry. Especially in the UK, this has extended to advocacy in schizophrenia. With over 20 randomized-controlled trials and five meta-analysis seeming to support its use, a place in management should be beyond doubt.⁽⁴³⁾

The fact is, however, that doubt does remain. A recent Cochrane review⁽⁴⁴⁾ found that CBT did not reduce relapse and readmission compared to standard care (though it did decrease the risk of staying in hospital) and while it improved mental state over the medium term, after a year these slight benefits had disappeared. Continuous measures on mental state did not demonstrate consistent effects. Compared to supportive psychotherapy, CBT had no effect on relapse and when combined with a psychoeducational approach, no significant reduction in readmission rates relative to standard care alone could be demonstrated.

While some individual studies have shown impressive results, the powerful advocacy CBT has attracted as adjunctive management in schizophrenia seems, at this stage, disproportionate to the evidence base. Studies have been built around fundamental design flaws, most notably in relation to control conditions, allowing some to conclude that CBT can be shown to work only in poorly controlled trials and not in well controlled ones.⁽⁴⁵⁾ Even between studies showing benefit, it is difficult to discern what the most appropriate target(s) should be and what components ought to comprise an/the ideal CBT package. A further problem, specific to assessing CBT in group contexts, is inappropriate data analysis where independence of observations is universally assumed, something group interactions violate, with a resultant increase in Type 1 errors.⁽⁴⁶⁾ While CBT may hold promise, widespread endorsement of a resource intensive management would seem premature until more and better designed work reports.

A further application of CBT techniques in psychosis has been in enhancing compliance. While compliance therapy,⁽⁴⁷⁾ combining cognitive behaviour and motivational interviewing techniques, has shown promise, it has been insufficiently evaluated to support robust recommendations. A recent Cochrane review⁽⁴⁸⁾ identified only one study comparing this with non-specific counselling. No significant differences were found in overall ‘non-compliance’ rates or in mental state measures, attitudes to treatment, global functioning, or quality of life. Although at 1 and 2 year follow-ups, average number of days in hospital was reduced, this was not statistically significant. This study did not show any effect on insight, but other work has claimed that improvements can be achieved by short, insight-focused CBT interventions, but at the expense of increasing depression,⁽⁴⁹⁾ an observation also reported with non-CBT approaches targeting insight.⁽⁵⁰⁾

Cognitive remediation (or rehabilitation), in which the desired end-point is not symptom reduction per se but improved global functioning via amelioration of cognitive deficits such as impaired vigilance, attention, and planning/decision-making, has also been applied in adjunctive management. Once again, while some individual findings are encouraging, data remain inconclusive.⁽⁵¹⁾

LEARNING POINT: While present evidence does support the use of CBT led interventions in adjunctive management of schizophrenia, the research is flawed and further, well controlled studies are necessary to determine a precise role.

Unlike Freud, many analysts of the early-mid twentieth century were undaunted in pursuit of psychodynamic understanding of, and management for, schizophrenia but theories were universally unsupported by evidence. When assessed against supportive psychotherapy, no advantages could be demonstrated.⁽³⁾ While some modest revival of interest may be detected, especially amongst advocates of ‘early intervention’, a recent review provided no support for such revisionism.⁽⁵²⁾ Furthermore, May’s study^(10,11) offers a cautionary warning of the potential for harm.

As a result of the early age of onset, relapsing nature, and persistence of many clinical features, schizophrenia can potently disrupt smooth acquisition and evolution of skills essential for developing mature interpersonal relationships, occupational competence, and independent living. Social (or life) skills training evolved in the context of resettlement programmes aimed at discharging long-stay, institutionalized patients but in various forms remain widely practised. It is based on a structured learning-orientated approach to the acquisition of skills relevant to the individual and the demands of his/her environment.

Unfortunately, social skills training is difficult to evaluate, as this has become a ‘blanket term’ covering a wide range of applications and targets. While some studies have focused on rehearsal of activities of daily living, others concentrate on communication and conversational skills, and although some view improvement in symptoms as the underlying goal, for others the benefit lies with cognitive ability. Blindness of assessments is a major problem, though can be easily achieved using blinded video techniques. Thus, while individual studies have found improvements in assertiveness, general social competence, and even speed of discharge, with benefits extending to a widened social network and that generalize, a recent Cochrane review failed to find conclusive evidence of benefit.⁽⁵³⁾