Much to improve: A survey of controlled Nordic schizophrenia trials KRISTIAN WAHLBECK, CLIVE ADAMS, BEN THORNLEY Wahlbeck K, Adams C, Thornley B. Much to improve: A survey of controlled Nordic schizophrenia trials. Nord J Psychiatry 2000;54:105 – 108. Oslo. ISSN 0803-9488. The Cochrane Schizophrenia Group’s Register of Trials was surveyed for reports on trials of Nordic origin. Quality assessment of reports was undertaken in accordance with the Jadad scale. Two hundred and twenty reports covering 181 trials from four Nordic countries were found, with most originating from Sweden. The Nordic trials were smaller than those from other countries and with a few exceptions had inadequate statistical power to produce any precise conclusions. Interventions applied were almost entirely biologic. Even though about half of the Nordic trial reports were of low quality, the mean quality score of these reports was significantly better than that of the rest of the world (U =147260, df =2035, P =0.017). In the future trials should be designed to recruit adequate sample sizes to be able to provide clinically useful data, and trial reporting must be improved. The use of prospective con- trolled trials should be extended to evaluation of psychosocial treatments of schizophrenia. • Controlled clinical trials, Random allocation, Research design, Sample size, Schizophrenia. Kristian Wahlbeck, Department of Psychiatry, University of Helsinki, Lappviksva ¨gen, PB 320, FIN-00029 HUCS, Finland; Accepted: 6 May 1999. T he basis of the evaluation of the treatments of schizophrenia lies in clinical trials, of which ran- domized controlled trials constitute the gold standard (1). The Cochrane Collaboration aims to promote the accessibility of trial-derived clinical evidence and pub- lishes maintained systematic reviews within the Cochrane Library (2). As a part of this process the Cochrane Schizophrenia Group has established a regis- ter consisting of all available relevant controlled trials. This register is an extensive source of trials for those preparing systematic reviews. The Cochrane Schizophrenia Group’s Trials Register also offers the possibility of surveying both the quantity and the quality of schizophrenia trials (3). The aim of this paper is to do this for schizophrenia trials of Nordic origin in the hope of improving trial quality by drawing attention to the strengths and weaknesses of current trial practice. A special emphasis will be put on factors that empirically are known to influence esti- mates of trial treatment effects. These factors include randomization, blinding procedures, and reporting of treatment withdrawals (4). Materials and Methods At the time of the survey (December 1998) the Trials Register contained 3933 reports of published and un- published controlled clinical trials focusing on the care of those with schizophrenia and other non-affective psychoses. Uncontrolled studies, within-patient com- parisons, and studies using historical controls are not included in the register. It has been compiled by identi- fying and hand-searching key journals from 1948 to the present. Conference proceedings were also hand- searched. Comprehensive searches of Biological Ab- stracts, CINAHL, The Cochrane Library, EMBASE, LILACS, PsycLIT, PsynDEX, MEDLINE, and Socio- file have been undertaken. The exact search terms have been reported elsewhere (5). The resulting 30,000 elec- tronic records were checked for duplicates before 2 trained selectors highlighted studies that were possibly relevant. Full copies were then acquired (6000) and, if meeting the inclusion criteria, added to the register. A trial was defined as originating from a Nordic country when the address of the first author was in Denmark, Finland, Iceland, Norway, or Sweden. Searching the address field in the Register for the Nordic countries identified a group of trials. Then the register was searched again, using surnames of authors identified in the first search. All records were then manually checked to ensure that they were of Nordic origin. Details of country of origin, publication year, trial size, and type of trial intervention were registered. The Jadad scale, a measure of methodologic quality (6), was used for assessing the Nordic trials and a sample of © 2000 Taylor & Francis 105 Nord J Psychiatry Downloaded from informahealthcare.com by Prof. Kristian Wahlbeck on 12/10/12 For personal use only.