It has long been known that most patients with bipolar disorder have a course marked by multiple recurrence of major depressive and manic episodes. More recently, many epidemiological studies have shown that at least 50% of unipolar depressive patients have one or more subsequent episodes of depression in their lifetimes. Likewise, relapse following successful short-term treatment of depression is so common that it is now recommended to prolong treatment in the form of maintenance pharmacotherapy. Interest in preventing the recurrence of depression has been stimulated by the results from long-term trials involving antidepressant drugs and lithium. However, these pharmacological studies suffer somewhat from methodological deficits, making difficult any clear and complete interpretation of their results. While some methodological recommendations have now been established to compensate for previous deficits (eg, definition of relapse and recurrence, duration of acute and maintenance treatments, statistical analysis of efficacy), unfortunately, many key points still remain to be resolved (eg, selection of patients, dosage in maintenance and prophylactic phases, maximum length of a course of treatment, etc).

In randomized clinical trials in patients with major depression quality of life is considered as an important dimension of treatment outcome in relation to clinical efficacy and safety. The internal validity, reliability, as well as external validity of quality of life scales have been analysed. It is concluded that such scales have their most appropriate applicability in medicine and long-term trials with antidepressants.

Studies on long-term efficacy and safety should be a main concern in the evaluation of novel antipsychotics. The present paper is a review of important issues related to the design and performance of such trials. The variability of the natural course of schizophrenia, the variability in treatment response, and the variability of actions of different neuroleptics need to be considered. Long-term studies need to address maintenance of efficacy, prevention of relapse and recurrence, health economics, quality of life and a large number of safety issues in order to get approval for new drug applications and reimbursement on expensive new medicines. Assessment instruments for efficacy, tolerability and safety, need to be supplemented with rating scales for various aspects of psychopathology (positive, negative, deficit symptoms), social and work functioning, drug induced mental side effects, quality of life and accounts of total treatment costs.

This paper discusses the need for a health economics perspective and some of the central methodological issues raised by economic appraisals, including cost-effectiveness and cost-utility analyses. It illustrates how cost-effectiveness studies can inform policy and practice decisions concerning the treatment of schizophrenia, focusing on five generic evaluation questions commonly raised. Examples are drawn from recent and current UK research.

Long-term (minimum six months) safety and efficacy studies with an anti-anxiety agent are aimed at identifying the emergence of any clinical effects not observed during shorter periods of treatment including the identification of late occurring adverse events, loss of therapeutic benefit, or post-treatment withdrawal effects. There is the need to establish diagnostic criteria for persistent anxiety for patients who truly benefit from long-term therapy. Risk factors predisposing to drug dependence should be assessed prior to continuous use. Ideally, long-term continuous studies should be blinded, comparative and placebo-controlled, with an adequate duration of the withdrawal phase. The treatment response of the elderly in comparison to the overall patient sample should be emphasised.

Development of treatments for dementia is beset by special problems in defining the diagnosis, establishing efficacy criteria, and specifying the necessary duration of study. There is need for agreement among clinicians and scientists on diagnostic subgroups of dementia. Similarly, there is a need for harmonization of the regulatory guidelines in Europe, Japan, and the United States regarding the decision set of variables on which to base efficacy claims. The duration of trials must be based upon the intended claim: transient symptomatic benefit, maintained symptomatic benefit, or a therapeutic effect on disease progression. Claims other than transient benefit require long-term trials, suggested to be of at least six months in duration. Problems with long-term studies include slow patient accrual, high dropout rates, changing milieu, low return on investment, and lack of unanimity regarding regulatory requirements. Regulatory authorities must come to some accord, consonant with current clinical/scientific wisdom and consensus, regarding diagnosis, efficacy criteria, and feasible study duration, in order to attract continued sponsor investment in the development of antidementia treatments.

This paper provides a brief overview of issues to consider in evaluating patient benefit from health care. Different types of measures are described and it is noted that the choice of measure is dependent upon the purpose of the study. The cost-utility approach to evaluation is discussed in more detail. This approach is often criticised but it raises important issues in medical decision-making and has considerable potential in the evaluation of patient benefit from health care.

This paper considers some of the theoretical and practical problems of conducting cost-utility analyses alongside clinical trials. In order to measure utilities of different health states in a clinical trial a number of critical assumptions have to be made. Some of these assumptions are questionable on a theoretical level, others empirically invalid. The practical problems of measuring utilities are discussed. The standard gamble is shown to be the most validated method of utility measurement, but still based on very strong assumptions. The standard gamble instrument is also costly and difficult to administer in clinical trials. Other instruments are found to be less valid than the standard gamble. It is concluded that although cost-utility analysis seems relevant in some instances, investigators should avoid this assessment of utility and instead measure cost-effectiveness, cost-benefit and quality of life.