1362 www.thelancet.com Vol 383 April 19, 2014 The Global Mental Health (GMH) movement has played a pivotal part in bringing to attention the unmet needs of patients with mental disorders, particularly in lowincome and middle-income countries. Sch izophrenia is of primary concern in view of the high level of associated disability and stigma, and the risk that, without treatment, patients will experience prolonged institutionalisation, neglect, and abuse. Sudipto Chatterjee and colleagues’ multicentre, randomised controlled COmmunity care for People with Schizophrenia in India (COPSI) trial, in The Lancet, represents a milestone by showing the benefi ts of a collaborative community-based care plus facilitybased care model compared with conventional facility-based care alone for treatment of moderate to severe schizophrenia. However, implementation of collaborative community-based care in low-income and middle-income countries has several issues that need further consideration, such as ensuring continuity in supervision of community workers, safeguarding the physical health of patients, and embedding services within the local context and culture. Collaborative community-based care makes sense: physical facilities (eg, clinics and hospitals) are not needed, demand on professional skills is low, and the family remains the core unit of care. COPSI is the fi rst trial to test collaborative community-based care rigorously in a developing country, India. 187 participants were Challenges in rolling out interventions for schizophrenia of psychotherapy is undoubted, the evidence base for its eff ect size is less solid than that for antidepressants. The main reason for this weaker evidence is the diffi culty in defi nition of valid control groups and the fact that therapists, patients, and often even raters are not masked. Outcome in psychotherapy control groups has even been found to be signifi cantly worse than that in pill placebo groups (the so-called nocebo eff ect), because patients are fully aware of their study situation. Testing psychotherapy against a nocebo condition could therefore lead to artifi cially large group diff erences and eff ect sizes. In summary, the present approach to estimation of the benefi ts of antidepressant treatments is likely to underestimate the clinical signifi cance of antidepressants and overestimate that of psychotherapy. At the same time, we are experiencing an increasing tendency to medicalise individuals who have emotional reactions to diffi cult life circumstances but without any clinical signs of depression, and to off er them antidepressants or psychotherapy which might not be appropriate to their needs. We should be careful not to off er our treatments to the wrong patients, but to provide them consistently to the right patients.