Efficacy of central nervous system stimulant treatment for cocaine dependence: a systematic review and meta-analysis of randomized controlled clinical trials Xavier Castells 1,2 , Miguel Casas 1 , Xavier Vidal 2 , Rosa Bosch 1 , Carlos Roncero 1 , Josep Antoni Ramos-Quiroga 1 & Dolors Capellà 2 Psychiatry Service, Hospital Universitari Vall d’Hebron, Universitat Autònoma de Barcelona, Spain 1 and Fundació Institut Català de Farmacologia, Clinical Pharmacology Service, Hospital Universitari Vall d’Hebron, Universitat Autònoma de Barcelona, Spain 2 ABSTRACT Aims To evaluate the efficacy of central nervous system (CNS) stimulants compared with placebo for the treatment of cocaine dependence. Methods A systematic review and meta-analysis was carried out. Bibliographic databases were searched, reference lists of retrieved studies were hand-searched and the first authors of each study were contacted. All randomized controlled clinical trials (RCCT) comparing the efficacy of any CNS stimulant with placebo in cocaine-dependent patients were included. Quantitative data synthesis was performed for each single CNS stimulant and for all CNS stimulants. Results Nine RCCT met the inclusion criteria. These RCCT included 640 patients and compared five CNS stimulants: mazindol, dextroamphetamine, methylphenidate, modafinil and bupropion with placebo. No CNS stimulant improved study retention [RR = 0.94 (0.81–1.09)] or cocaine use [RR = 0.90 (0.79–1.02)]. An exploratory analysis using indirect estimations of cocaine use showed that the proportion of cocaine-positive urine screens was lower with dexamphetamine than with placebo [RR = 0.73 (0.60–0.90)] and that all CNS stimulants pooled together also suggested a significant decrease of cocaine use [RR = 0.87 (0.77–0.99)]. Data on craving could not be meta-analysed due to heterogeneity, but no RCCT found differences in cocaine craving between active drug and placebo except one, whose outcome favoured dexamphetamine. No serious adverse event (AE) was reported. Average of AE-induced dropouts was low and was greater for CNS stimulants than placebo: 4.4% versus 1.3% (P = 0.03). Conclusion The main outcomes of this study do not support the use of CNS stimulants for cocaine dependence. Nevertheless, secondary analyses provide some hopeful results that encourage further research with these drugs, mainly with dexamphetamine and modafinil. Keywords CNS stimulants, cocaine dependence, meta-analysis, placebo, randomized controlled trial. Correspondence to: Xavier Castells, Psychiatry Service, Hospital Universitari Vall d’Hebron, Edifici escola d’infermeria 5th floor, Passeig Vall d’Hebron 119–129, Barcelona 08035, Catalonia, Spain. E-mail: xcc@icf.uab.cat Submitted 9 January 2007; initial review completed 28 March 2007; final version accepted 15 May 2007 INTRODUCTION The prevalence of cocaine dependence has been increas- ing in recent years and has become a world health problem. During 2000–01, 0.3% of the population world-wide, aged 15 years or more, had used cocaine [1]. In the European Union countries life-time cocaine use prevalence reached 3% of the adult population, with the United Kingdom (6.1%), Spain (5.9%) and Italy (4.6%) at the upper end of this range [2]. In the United States in 2004, life-time and past-year cocaine use among people aged 12 or older was 14.7% and 2.4%, respectively. Among past-year cocaine users, 27.8% were classified as having a cocaine dependence or abuse disorder [3]. A large list of drugs, comprising antidepressants, antipsychotics, dopamine agonists or mood stabilizers, has been studied for cocaine dependence, although none has proved clearly to be effective [4]. As a consequence, no drug has a Food and Drugs Administration (FDA) or European Agency for the Evaluation of Medicinal REVIEW doi:10.1111/j.1360-0443.2007.01943.x © 2007 The Authors. Journal compilation © 2007 Society for the Study of Addiction Addiction, 102, 1871–1887