Inclusion and Exclusion Criteria
Types of Participants
Human participants of any age from studies taking place in high income countries. These studies needed to look at individuals with a primary diagnosis of moderate or severe mental illness, including schizophrenia, bipolar disorder, moderate and major affective disorder and delusional disorder. We only included individuals with alcohol or substance abuse disorders where these were part of a dual diagnosis with one of these mental health problems. We also excluded diagnoses of Alzheimer's Disease and related disorders, eating disorders and intellectual disabilities, from our analysis.
As our focus here was on disease prevention and health promotion; we excluded studies that looked at the economic benefits of treating physical health problems in people with pre-existing mental and physical co-morbidities. Nor did we look at the economic literature on interventions to prevent mental health problems in people with chronic physical illness such as diabetes or cardiovascular disease. There were no other restrictions on socio-economic or clinical characteristics.
Types of Intervention
Any non pharmacological interventions specifically targeted at promoting the physical health of individuals living with mental health problems. Interventions could for instance include exercise programmes, nutritional advice, smoking, alcohol and drug cessation and infectious diseases prevention. Our analysis was restricted to interventions delivered in countries defined by the World Bank as being high income.
Types of Outcome Measure
We included all of the standard economic evaluation methods that have been defined in health economics. In brief, all measure costs in the same way but differ in how they measure economic outcomes. Cost-effectiveness (CEA) and cost-consequences (CCA) analyses report outcomes using natural clinical units e.g. changes in symptom free days; cost-utility analyses (CUA) studies measure outcomes in a common metric either in terms of quality or disability adjusted life years gained; while cost-benefit analyses (CBA) elicit monetary values from the public for different health outcomes. In addition we also included cost-offset analyses (COA), which highlight the potential resource savings of an action without reporting on changes in health outcomes. Studies that looked at economic incentives such as monetary rewards or changes in the price of goods to promote behaviour change were also included within the scope of our review.
Types of Study
All of the economic evaluations described above that bring together cost data with outcome data were eligible for inclusion regardless of whether they were performed prospectively alongside a controlled trial or other quasi-experimental study design or retrospectively using data from a previous effectiveness study. Economic modelling studies that estimated the potential cost effectiveness of interventions by synthesising evidence from trials and other data sources on effectiveness and costs were also eligible. We also recorded study protocol papers reporting current evaluations in order to help identify the current interest in this area, and the way in which the evidence base is likely to develop further.
Search Process
We designed detailed specific systematic search strategies for several health and social science bibliographic databases: PubMed, PsycINFO, CINAHL, Francis, SocIndex and EconLit, covering the period January 1990 to December 2012, with searches run for the period until June 2012 on 7 September 2012, and updated searches for the final six months of the 2012 run on 2 January 2013. Our search strategy for the PubMed database is provided in Additional file 2 as an example of search strategies used. We combined a wide range of phrases for mental disorders with health promotion and public health terms along with terms for economic evaluations and/or specific economic terms/phrases such as cost-benefit. Papers could be in any language, but they needed to have an English language abstract to be potentially eligible for inclusion.
Our electronic search was supplemented by hand-searches of a small number of relevant journals. A limited search of Google Scholar was also undertaken, alongside scrutiny of relevant websites including think-tanks, university economic research groups and some government departments. We also looked at actual examples of programmes and guidelines for the promotion of physical health that were identified separately within our project.
References were initially imported into Endnote X5 and duplicates eliminated. Remaining references were initially checked independently by two reviewers, based on their title and abstract to decide if they met our inclusion criteria. In the case of disagreement, further discussions about inclusion/exclusion were made. Full texts of potential matching articles were retrieved and then assessed to determine whether they looked at both the costs and effectiveness of interventions Articles were excluded if full texts did not provide cost or resource data. Data on bibliographic information, the intervention and comparator, duration of study, target population, economic evaluation methods used, empirical study design, cost and resource findings, effectiveness results and synthesis of costs and effects were then extracted from eligible studies into a bespoke Excel data extraction form. We conducted a narrative analysis and review of these studies. We did not plan or conduct statistical meta-analysis or other formal, aggregative synthesis of the results of included studies. All costs were converted to 2010 International Dollars, as well as being reported in their original currency and price year in Table 1 .