Psychiatric outcome studies (POS) of treatment with psychotherapy as well as pharmacotherapy have the following three psychometric measures in common: (a) The rating of core symptoms; (b) the global clinical assessment; and (c) the rating of social functioning or quality of life.

In the psychometric evaluation of these three outcome measures, effect size is the clinically most meaningful statistic, both in placebo-controlled and in dose-response trials.

The clinically most valid outcome scales and the use of effect size statistics will be shown with reference to the literature on POS.

The present study, conducted in collaboration between the Departments of Psychiatry in Swiss Universities and the World Health Organization, had two main goals: to develop assessment methods which could subsequently be used in the Swiss centres in a standard manner; and to make arrangements for continuing collaboration between the centres in Switzerland and the acquisition of new knowledge about the distinctions between depression and cognitive impairment. For this aim, three different groups of elderly patients of either sex were selected during the period of November 1989 to July 1991 for inclusion in the study. The first two groups included the first ten patients of either sex over 60 years of age consecutively contacting the participating institutions and showing depression with or without clinically significant symptoms of cognitive impairment; the control group included patients showing no depression or clinically significant symptoms of cognitive impairment. A total of 125 patients were included in the initial evaluation, 69 of which were reassessed at a seven-month follow up (on average). Each patient was administered a number of clinician-rated or self-report instruments for the assessment of depression, cognitive impairment, disabilities, physical status and onset of disorders. The study has shown that a variety of instruments can be used for the reliable assessment of depression or cognitive impairment in the elderly; but the instruments for the assessment of depression differentiate only poorly between patients with or without cognitive impairment. Because of the importance of identifying both depressed and cognitively impaired patients among the elderly, different assessment instruments targeted at the different symptom clusters need to be administered simultaneously.

Among the treatment modalities in the acute therapy phase of a major depressive episode, ECT (electroconvulsive therapy) has the highest response rate (90%), but also the highest relapse rate in the continuation phase over the next 6 months (65%). Placebo has the lowest response rate (45%), but the highest pharmacological relapse rate (50%). The SSRIs seem to have the lowest relapse rate compared to imipramine (12% versus 30%). Both the SNRIs and mirtazapine (the “dual action” drugs) have higher relapse rates (20%) although they have a higher response rate than the SSRIs (70% versus 60%).

Quality of life became a paradigm in the medical setting two decades ago. Although ‘humanistic’ psychologists still maintain that what can be specified and measured is precisely what is not quality of life, time has shown that the major components of quality of life in health and disease can be measured with an acceptable internal as well as external validity. Quality of life in randomized clinical trials is now the third dimension to be measured, the other dimensions are efficacy of the drug and adverse drug reactions. Among the major areas of external validity quality of life scales (eg the PCASEE questionnaire and the Psychological General Well-Being Schedule) have discriminating validity (the discrimination between treatments), predictive-validity (the prediction of relapse of illness), and teological validity (defining goal of treatment).

Data from six randomised, controlled, comparative studies of risperidone and conventional neuroleptics (haloperidol, zuclopenthixol and perphenazine) in the treatment of 911 patients with chronic schizophrenia were analysed to estimate the benefits and risks of treatment. Efficacy and risk of treatment were assessed by means of the Positive and Negative Syndrome Scale (PANSS) and the Extrapyramidal Symptom Rating Scale (ESRS). An effect size (difference between treatments) was calculated from the PANSS and ESRS scores. The effect size of antipsychotic effect favoured risperidone (sizes ranged from 0.22 to 0.37 on the PANSS subscales), and the effect size of safety also favoured risperidone (sizes ranged from 0.18 to 0.36 on the ESRS). An analysis of the proportions of patients requiring antiparkinsonian medication produced an effect size of 0.37 (20% of risperidone patients and 38% of patients receiving a conventional neuroleptic required antiparkinsonian medication). These data indicate that risperidone is a more effective antipsychotic than the conventional agents and causes less severe extrapyramidal symptoms.

The algorithms for the demonstration of shared phenomenology of psychiatric syndromes in DSM-III are resistant to quantification. In contrast, the rating scale approach quantifies clinical target syndromes in psychiatry. The two most useful statistical models for quantifying shared phenomenology by symptom rating scales have been reviewed; namely factor analysis and latent structure analysis. Results have shown that factor analysis has demonstrated dimensions of dementia, delirium, schizophrenia, mania, outward aggression, depression and anxiety. Latent structure analysis has confirmed that the items of brief rating scales (such as the Melancholia Scale) are additively related implying that their total scores are sufficient statistics for the measurement of these factors or dimensions. Latent structure analysis should be considered as a psychometric “glasnost” compared to algorithm-resistant logic of quantification in DSM-III.

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.

Major depression is one of the medical disorders in which quality of life instruments have most often been applied. It is important to distinguish between disease-dependent scales (ie, scales measuring the symptoms of major depression, such as the Beck Depression Inventory) and generic scales (ie, scales measuring self-reported states without reference to a specific disorder, such as the WHO well-being scale). The empirical use of generic scales for quality of life major depression has indicated that they can predict relapse and compliance. In the future, more trials comparing the different treatments with cost-effectiveness vs quality of life in major depression are required.

In patients with chronic idiopathic pain disorders we have analysed the construct validity of the Melancholia Scale as compared to the results with the scale in primary depression. The patients (n= 253) were treated in a placebo controlled trial with either clomipramine or mianserin independently of the Melancholia score. The construct validity of the Melancholia Scale was further analysed by the testing of the intensity model of depression versus anxiety using the Beck Depression Inventory, the Hamilton Anxiety Scale, the Spielberger State-Trait Anxiety Scale, and the Melancholia Scale. The construct validity in terms of scale homogeneity was analysed by Loevinger coefficients which can be considered as a latent structure evaluation. The Melancholia Scale showed acceptable homogeneity, while the Hamilton Anxiety Scale lacked sufficient homogeneity. In total, 33% of the patients had a score of 10 or more on the Melancholia Scale (corresponding to 13 or more on the Hamilton Depression Scale). The predictive validity of the Melancholia Scale was evaluated using active treatment versus placebo response after 6 weeks of therapy. It was shown that in patients with a Melancholia Scale score of 10 or more (corresponding to “less than major depression”) 72% had full recovery when treated with clomipramine, while 36% of the placebo treated patients obtained a full recovery (P≤0.05). The patients treated with mianserin obtained a full recovery in 52%. The group of patients with a Melancholia Scale score of 10 or more scored higher also on the anxiety scales indicating that the relation between depression and anxiety is a matter of severity. The depressed patients had significantly lower imipramine binding sites than the non-depressed patients.