Evaluation of Mental Health Interventions
Summary and Keywords
Mental illnesses are highly prevalent and can have considerable, enduring consequences for individuals, families, communities, and economies. Despite these high prevalence rates, mental illnesses have not received as much public policy commitment or funding as might be expected. One result is that mental illness often goes unrecognized and untreated. The resultant costs are felt not only in healthcare systems, but across many other sectors, including housing, social care, criminal justice, welfare benefits, and employment.
This article sets out the basic principles of economic evaluation, with illustrations in this mental health context. It also discusses the main practical challenges when conducting and interpreting evidence from such evaluations.
Decisions about whether to spend resources on a treatment or prevention strategy are based on whether it is likely to be effective in avoiding, reducing, or curing symptoms, improving quality of life, or achieving other individual-level outcomes. The economic evaluation question is whether the outcomes achieved are sufficient to justify the cost that is incurred in delivering the intervention.
An economic evaluation has five elements: clarification of the question to be addressed; specification of the intervention to be evaluated and with what alternative it is being compared; the outcomes to be measured; the costs to be measured (including the cost of implementing the intervention and any savings that might accrue); and finally, how outcome and cost findings are to be blended to make a recommendation to the decision-maker. Sometimes, if an evaluation finds that one intervention has better outcomes but higher costs, then the evaluation should also how one (the outcomes) might be trade-off for the other (the costs).
The article illustrates how economic evaluations have been undertaken and employed to address a range of questions, from the very strategic issue to the more specific clinical question. The purpose of the study can, to some extent, determine the type of evaluation that is needed.
Examples of evaluations are given in a number of areas: perinatal maternal mental illness; parenting programs for conduct disorder; anti-bullying programs in schools; early intervention services for psychosis; individual placement and support; collaborative care for physical health problems; and suicide prevention. The challenges of economic evaluation are discussed, specifically in the mental health field.
Impacts of Mental Illness
Consistently across the world, mental illnesses have been found to be highly prevalent, with considerable and enduring personal, societal, and economic consequences. About one adult in six will experience a “common mental disorder”—principally, depression or anxiety—in any one-week period (in the United Kingdom, but similar prevalence figures are found elsewhere), while psychotic disorders—principally, schizophrenia and bipolar disorder—have a combined annual prevalence of about 2% to 3% (McManus, Bebbington, Jenkins, & Brugha, 2016). Nearly 10% of children and young people (5–16 years) have a clinically diagnosable mental disorder (Green, Mcginnity, Meltzer, Ford, & Goodman, 2005). The mental illness most commonly associated with aging is dementia, with a prevalence of roughly 2% among 65–69-year-olds, increasing to around 20% among those aged 85–89 (Prince et al., 2014). Despite these high rates of prevalence, mental illnesses have generally not attracted the policy attention or investment of resources that might be expected.
This is emphasized by the realization that, together, mental illnesses account for a substantial proportion of the global burden of disease: 32% of years lived with disability and 13% of disability-adjusted life-years (DALYs) (Vigo, Thornicroft, & Atun, 2016). These proportions have grown over time, in part because of population ageing, but also because most mental illnesses are chronic and without cure. Moreover, a high proportion of such illnesses start in childhood, adolescence or early adulthood (half by age 14, and three-quarters by age 24) (Kessler et al., 2005), and then have sizeable personal, economic, and social impacts over the life-course. Preventive actions therefore have the potential to alleviate distress, reduce suicide rates, improve wellbeing, cut health care and other costs, and contribute to national productivity. Genes play a part in etiology, but environmental factors have stronger influences.
Mental illness is often unrecognized and untreated (or under-treated). Reasons for this—which are interconnected—include low rates of identification in primary care, low policy priority by most national governments, limited investment in effective treatments, inequalities of access to treatment, and stigma. The consequences of untreated mental illness are seen across many domains and range widely beyond the health sector; impacts will often be seen in the housing, social care, criminal justice, welfare benefits, employment, and other systems. This means that the healthcare costs of mental illness tend to be swamped by economic impacts in these other areas, as illustrated in the “Examples of Economic Evaluation” section. Personal costs for the individual with mental illness and their family can also be high because of disrupted employment.
This article sets out the principles of economic evaluation specifically in the mental health context. The broad approach is the same as would be employed in economic evaluations in other health contexts, but with some additional issues to take on board. The third section, “Uses of Evidence From Economic Evaluations,” briefly describes how evidence from economic evaluations has been employed, and the fourth section, “Examples of Economic Evaluation,” gives examples of completed studies and some of their implications. The final section finishes by discussing the practical challenges of conducting and interpreting evidence from such evaluations.
Making an Economic Case
Among the questions that need to be addressed when deciding whether to invest or deliver a particular treatment or preventive strategy are: “Does it work?” and “Is it worth it?” The first asks whether the intervention (using that term to cover medications, psychological and other treatments, care services, preventive strategies, and wider policy initiatives) is effective in avoiding, reducing, or curing symptoms of illness, or improving health-related quality of life, or achieving other outcomes such as independence in daily living or general wellbeing. The second question asks whether the outcomes achieved are sufficient to justify the cost of the intervention. The second question is what an economic evaluation aims to answer.
An economic evaluation in this healthcare context comprises five elements:
• The question addressed by the evaluation.
• The intervention to be evaluated and what it is being compared with.
• The outcomes associated with each intervention.
• The costs associated with each intervention (both the cost of implementation of the intervention itself and any savings that might accrue).
• How outcome and cost findings are blended to generate an answer to the “Is it worth it?” question.
Economic evaluations have been used to address quite broad strategic questions as well as more specific clinical questions. The purpose of the study will, to some extent, determine the breadth and type of evaluation conducted. The strongest research design for testing whether a new intervention is more effective than existing treatments is generally held to be a randomized controlled trial (RCT). This is usually the best way to examine whether one intervention makes better economic sense than another. However, it is important to note immediately that other research designs might be more suitable in some circumstances; these include observational designs and mathematical or simulation modeling. Given the chronicity of most mental illnesses and therefore the potential for quite long-term impacts of successful treatment, observational or simulation designs might be needed to capture these consequences.
The different types of economic evaluation share some common features. In particular, they share a common approach to conceptualization, definition, and measurement of costs, for example. Where they tend to differ is in relation to how they define and measure outcomes. These latter differences are primarily driven by the need to answer different resource allocation questions.
If the question is how best to treat a specific disorder—such as severe depression, and which antidepressant medication should be used, or whether to combine medication with a psychological therapy—then the outcomes that are most relevant to the clinical decision maker would be those that are specific to the condition (in this case, alleviation of depressive symptoms) as well as those relating to general quality of life. In this context, a cost-effectiveness analysis would be suitable: outcome measures used in this type of economic evaluation would be the same as the clinical researcher would be likely to choose. Costs would be measured with greater or lesser breadth, depending on the context. A cost-effectiveness analysis therefore tells decision makers what course of action (e.g., what treatment) is likely to be the most efficient way to respond to a particular set of clinical needs.
However, there is often a need to address broader questions about how to use available resources. If, for example, a decision maker with some additional or unallocated budget has to decide whether to expand treatment for depression or treatment for cancer, they will want to know how the two options compare in terms of costs and outcomes. Costs can readily be calculated and compared for interventions in the two different disease areas, but disease-specific outcomes will not be enough: one set will focus on depressive symptoms such as sadness, hopelessness, and apathy, while the other will focus on, say, tumor response rates and survival. The decision maker needs a common measure of outcome that allows comparison of the impact of depression treatment with the impact of cancer treatment. Economists have developed the utility measure (of health-related quality of life), and often operationalize it in measures of quality-adjusted life years (QALYs). How these are measured is described later, along with their pros and cons. Another potential generic measure for this kind of cross-disease comparison could be the DALY, mentioned earlier. When an evaluation uses a generic outcome measure like a QALY, it is sometimes called a cost-utility analysis. It tells the strategic health-system decision maker where they will get the biggest impact from their resources: from (in this case) depression treatment or cancer treatment.
There might be an even broader question to address, such as when a decision has to be taken about how to allocate resources between different areas of public policy: more on healthcare, or more on education or the armed forces? An economic evaluation can relatively easily compare costs between the policy alternatives, but to compare outcomes it would need a very broad measure such as the monetary value of the various outcomes or maybe some quite high-level wellbeing construct, such as happiness. In health economics, this is usually called a cost-benefit analysis.
It can therefore be seen how the question to be answered by an evaluation heavily influences the choice of outcome measure and the type of economic evaluation. We provide some illustrations later, but we can note here that a single study can usually support more than one type of evaluation.
The economic impacts of many mental health problems range across numerous services and systems, and can extend over many years. Some of these economic impacts might be seen as directly associated with the treatment of a disorder, such as money spent on buying medications and paying staff salaries. Some other impacts might be seen as more indirect, such as the lost productivity that results when someone takes time off work because of illness (absenteeism), impaired performance while at work (so-called presenteeism), early retirement, or reduced lifetime productivity because of premature mortality. Those productivity-related costs can very high for some mental illnesses (Organization for Economic Cooperation and Development, 2012). Another indirect cost could be the effect on unpaid family or other caregivers, who give up time or their own employment to provide support: these are especially high, for example, in the case of dementia (Prince et al., 2014). There may also be out-of-pocket expenses by families to pay treatments or transport to services.
An economic evaluation that aimed to address broad society-wide consequences would ideally measure a broad range of impacts. One need would be for information on services used by individuals (what services, how often, over what length of time), and then to attach a unit (average) cost to each of them. Data would also be needed on the productivity impacts of treatment, to which monetary values would then be attached, usually approximated by a sector- or job-specific wage, or maybe by an estimated average national wage rate, based on the assumption that wage approximates marginal productivity. Unpaid family support would need to be measured—how many hours on what kinds of tasks over what duration of time—and then a monetary value attached that was perhaps based on an opportunity cost assumption (what is the value of the activities given up to provide unpaid care) or a replacement cost approach (what would be the cost of bringing in a paid care worker or similar).
Perhaps the most intuitive measure of outcome is effectiveness, linked to the avoidance or alleviation of the core symptoms of the mental illness under consideration. This then leads to what health economists called a cost-effectiveness analysis (CEA). Those symptoms might include psychological experiences, behavior issues, impaired personal functioning, or poor health-related quality of life. A CEA is then available to inform decision makers who face choices over which interventions should be used to address specific health needs, such as the choice between two or more different antidepressant medications. A widely used measure of depressive symptoms is the Hamilton rating scale (Hamilton, 1960); an evaluation can rate patients before treatment and after, say 6 months, for the two or more interventions under study, and then compare changes over time for patients receiving different treatments. If one intervention achieves a significantly greater change in Hamilton score over 6 months than the other, and if was also found to be significantly less costly, then from an economics standpoint it looks cost-effective since it both improves outcomes and saves resources.
One example is a study that compared two treatments for people who experienced quite severe episodes of depression. One treatment was electroconvulsive therapy (ECT) and the other was repetitive transcranial magnetic stimulation (rTMS)—at the time, the latter was a relatively new treatment modality. After they had consented to participate in the study, patients were randomly allocated to one or other of the treatments. The primary outcome was the Hamilton rating scale, which was measured before randomization (baseline), at the end of treatment, and then again three and six months later. Costs were calculated for all services used over the same time period. ECT was significantly more effective than rTMS at the end of the treatment period (comparing Hamilton scores), but there was no difference between the treatment groups by the time of the six-month follow-up. Some secondary outcome measures were also used, and again the patients in the ECT group were doing better, in this case right up to the six-month follow-up point (Eranti et al., 2007). The cost-effectiveness study—which was embedded in the same trial—found that costs were significantly lower for patients randomized to ECT than for those randomized to rTMS. Pulling the outcome and cost results together, it was concluded that ECT was more cost-effective than rTMS (Knapp et al., 2008).
Some cost-effectiveness analyses measure more than one outcome measure. In the study of ECT and rTMS, for example, the main outcome used for the economic evaluation was the Hamilton scale, picking up the core clinical symptoms usually associated with depression. But the study also calculated QALY gains over the evaluation period, and so could also feed findings into wider resource allocation decisions. However, the study found no difference in QALY gain between the two interventions. It can be problematic for the decision maker when one outcome suggests one thing and another outcome suggests something different: in this case ECT outperforms rTMS on depressive symptoms, but the two are equally effective in terms of broader health-related quality of life (QALYs). Because ECT is significantly less costly, however, both outcome measures would still lead to the same conclusion, that ECT is more cost-effective. In some other circumstances, one intervention might be comparatively more cost-effective on one outcome measure, but less cost-effective on another. In this case, the preferred option may not be clear to the decision maker, who must weigh up the strength of evidence and perhaps consider other criteria, too (such as equity or overall budget impact).
A particularly difficult decision arises when one intervention is more effective than another, but also leads to comparatively higher costs. To decide which of the two interventions represents the better use of resources, the decision maker will need to consider under what conditions they would be willing to trade off outcomes and costs. Specifically, does the decision maker consider that the better outcomes are worth paying the higher costs? This is not a computational question but a value judgement.
When an economic evaluation finds this kind of result, the usual next step is to calculate the incremental cost-effectiveness ratio (ICER), which divides the extra cost associated with one intervention compared to the other by the additional effect that is achieved. An example can be given, from a study of computerized cognitive behavioral therapy (CCBT) for people with depression or anxiety, which was compared to treatment as usual in primary care over an 8-month follow-up period. A randomized trial found that CCBT was better than treatment as usual on a number of measures: alleviating symptoms of depression and anxiety, and improving work and social functioning (Proudfoot et al., 2004). However, CCBT was found to be the more expensive intervention in terms of costs to the health service (McCrone et al., 2004).
An ICER was calculated as the difference in cost between CCBT and usual care (with CCBT being the higher cost option) divided by difference in improvement in depressive symptoms between the two treatments. That ratio value was £21 (equivalent to the cost of achieving a 1-point gain in the depressive symptoms measure), which was equivalent to a cost of £2.50 to achieve one additional depression-free day over the period of the evaluation. (These cost figures were measured at price levels applying at the time of the study.) Does that ratio or that cost per depression-free day mean that CCBT is cost-effective? The economic analyst cannot say: it depends on whether the decision maker considers that this additional cost is justified by the better outcome (Petrou & Gray, 2011).
There are many occasions when decisions about resource allocation need comparisons across disease areas, for example when agreeing on departmental or specialty budgets within a hospital or NHS Trust, or when making higher-level strategic decisions about which conditions should be prioritized. For these kinds of decisions, the unidimensional utility construct is helpful and is usually operationalized by health economists as a preference-weighted, health-related quality of life measure. The QALY is the most widely used such measure (Petrou & Gray, 2011). Utility scores conventionally run from 1, representing perfect health, to 0, which represents death. The extra years of life that might follow from delivery of an intervention are highly relevant of course, but it is also important to understand the quality of each of those life-years.
The study of CCBT introduced earlier included calculation of the cost per additional QALY gained as a result of the computerized intervention; this ICER was found to be £2190 (at the price levels pertaining at the time). Reporting an ICER in terms of cost per QALY makes it possible for strategic decision makers to compare interventions in completely different disease areas. These kinds of comparisons are regularly made in England and Wales by the National Institute for Health and Care Excellence (NICE), a government-funded but independent body established some years ago to help decision makers in the National Health Service (NHS) to decide how to allocate their resources—how to weigh up the trade-offs involved in many scenarios.
The most commonly used tool for generating QALYs is the EuroQol (EQ-5D-3L or EQ-5D-5L) (EuroQol Group, 1990). This tool has been used in thousands of studies worldwide. The EQ-5D does not perform as sensitively as might be needed in some areas, for example in research on treatments for some severe mental health problems, or for children with mental health problems or older people with dementia (Brazier, 2010).
One recent example where the EQ-5D performed perfectly well was an evaluation of a manual-based coping strategy for family caregivers of people with dementia, which was compared in a randomized trial with usual forms of caregiver support (Knapp, King, et al., 2013). Caregivers in the intervention group had significantly better outcomes as measured in terms of symptoms of mental illness, linked to the stress of caring (Livingston et al., 2013), and QALY gains. Costs were actually slightly higher for the caregivers who received the coping strategy intervention, compared to those who were supported as usual. The cost-effectiveness ratio was calculated to be £6000 per QALY gained.
So, how is a cost-per-QALY value to be interpreted, and what decision should it encourage? The National Institute for Health and Care Excellence (NICE) in England and Wales employs a framework to help strategic decision making, particularly whether—in the wider scheme of things—the outcomes of an intervention would be considered worth the higher costs needed to achieve them. NICE uses QALYs as their preferred generic outcome measure, and to produce a threshold value for cost effectiveness: an intervention that costs more than £30,000 per QALY would generally not be considered worth paying for since the resources (represented by the cost) would (according to the NICE framework) be more productively spent (in terms of generating more QALYs per £1 of NHS expenditure) on some other health treatment. NICE uses this kind of threshold (and there are other values) only as a guide, and not as a rigid rule. What the threshold very usefully does is remind all stakeholders in a healthcare system that resources are scarce and (often tough) decisions must be taken when choosing how to use them. A number of countries other than England and Wales now also have similar health technology appraisal procedures.
I described earlier how decisions at the most strategic level of resource allocation—for example, across distinct areas of public policy—will need a higher level outcome measure to make comparisons. This is where cost-benefit analysis (as the term is used by health economists) is helpful. A cost-benefit analysis measures costs and outcomes (benefits) in the same monetary units, which makes it possible to see whether the benefits are greater or lower than the costs—if benefits exceed costs, then the intervention looks attractive, and the decision maker (or the evaluator supporting them) then needs to work out which of two or more competing interventions has the highest net benefit (the difference between benefits and costs). Cost-benefit analyses look very useful, but are difficult to do well in the health domain, principally because it is very difficult to convert a clinical measure of outcome (e.g., symptom alleviation) into a monetary value. I will give one mental health example later, where one of the outcomes (which was actually not clinical in nature) could be converted in this way.
Randomized controlled trials are considered the best way to generate robust evidence on whether a medication or other intervention is clinically effective, either in comparison to placebo (doing nothing) or to another intervention for the same underlying illness. Some of the examples given earlier provide illustrations of the utilization of such a design, and the approach can accommodate an economic evaluation without difficulty. The limitations of randomized designs are well-known: sometimes it is infeasible to randomize people or teams or localities, for example, or to persuade those people to remain in randomly allocated groups for sufficient time for long-term consequences to be assessed. Even a relatively short-term randomized trial might not be able to provide decision makers with evidence quickly enough to inform their more urgent decisions.
One alternative approach is some form of mathematical or simulation modeling such as a simple decision tree or a Markov model. Models are intended to represent what might happen in reality: they might chart care pathways for people with a particular set of needs, using data from previous trials, observational studies, or routine management information systems (secondary data) to (through the model) estimate outcomes and costs for each of the interventions under study (Briggs, Claxton, & Sculpher, 2006). Models are flexible but highly simplified simulations of reality. They are sometimes used, not to replace a randomized trial, but to try to show what would be expected to happen beyond the timescale of the trial period.
An example can be given from the child mental health area. Parenting programs have been suggested where there is a child in a family with a conduct disorder, which is the most prevalent of mental illnesses in childhood. Conduct disorders are experienced by about 5% of young people at any one time (Green, Mcginnity, Meltzer, Ford, & Goodman, 2005). While behavior problems of this kind will resolve in some children, around 50% of those children will develop antisocial personality disorder as adults (Richardson & Joughin, 2002), and behavior problems are associated with significant costs to the public purse that are likely to reach well into adulthood (Knapp et al., 2015).
There is good evidence that these programs can be effective, at least in the short to medium term. Based on data gathered by others in around 20 randomized control trials, Bonin, Stevens, Beecham, Byford, and Parsonage (2011) modeled the economic pay-offs from such parenting programs over a 25-year period. They traced the likely consequences for service use in the health, social care, and special education sectors, criminal justice system contacts, and employment patterns. Their modeling demonstrated that the economic return over the 25-year period was between 2.8 and 6.1 times greater than the cost of running the parenting program.
Uses of Evidence From Economic Evaluations
Economic evaluations have the potential to generate evidence for a range of different uses. Companies that manufacture pharmaceuticals and medical devices often use cost-effectiveness and other economic data to support the market case for their products.
More generally, economic evaluation evidence will often be used to inform commissioning decisions by hospitals and other treatment providers. Similar evidence is often used—alongside other types of evidence, of course—in formulating and monitoring policy by governments and other public bodies. A few years ago, for example, the Department of Health in England commissioned research to explore the economic case for 15 different mental health promotion and mental illness prevention interventions that had previously been shown to be effective but whose economic consequences were unknown (Knapp, McDaid, & Parsonage, 2011). The World Health Organization (WHO) generated cost-effectiveness evidence within its Choosing Interventions that are Cost Effective program, pooling evidence in order to show how, across each of its 17 sub-regions, what the costs would be, alongside evidence on health impacts and cost-effectiveness (Chisholm, van Ommeren, Ayuso-Mateos, & Saxena, 2005).
As noted earlier, a number of countries now have health technology appraisal procedures that synthesize findings on effectiveness and cost-effectiveness (including new modeling to generate new economic evidence in some cases) so as to support decisions on reimbursement and coverage within a health system. NICE in England and Wales is one example.
Examples of Economic Evaluation
There is now a fairly substantial body of evidence that well-designed, appropriately implemented interventions can be effective in preventing mental health problems emerging in the first place or, when they do arise, responding to them so as to lessen their negative impacts. However, implementation does not happen as often as it should, and one common reason is resource-related: an intervention might be seen as unaffordable in the short term or unsustainable in the longer term.
In fact, many mental health interventions have been found to be both effective and cost-effective, with many proving also to be cost-saving in relatively short periods of time. Some examples are given here.
Perinatal Maternal Mental Illness
One in five women experience some form of mental illness during pregnancy or in the year after birth, with risks for child development. Cost consequences can be high: for example, there is a cost exceeding £75,000 per woman with perinatal depression, much of it due to long-term adverse impacts on the child, including mental health impacts (Bauer, Parsonage, Knapp, Iemmi, & Adelaja, 2015). Universal and specialist interventions during the perinatal period can prevent or reduce mental illness and are cost-effective (Morrell et al., 2016). These interventions include parenting education and support, and individual and group-based psychological approaches, including those offered by health visitors or facilitated online. Many of these actually generate cost savings from a government perspective, while others look economically attractive from a wider societal perspective.
Parenting Programs for Conduct Disorder
Public sector savings can be achieved by providing the main treatments recommended by the National Institute for Health and Care Excellence (NICE) for mild-to-moderate child mental health problems. In addition, there will be other potential benefits from treatment: improved health, better relationships with family and friends, reduced social exclusion, reduced victim costs from criminal activity, and any extra effects on gross domestic product not captured as higher tax receipts. Group parenting programs for children with moderately severe conduct disorder are one example—as noted in the section “Modeling,” when summarizing the Bonin et al. (2011) modeling study.
The Incredible Years (IY) program has been implemented in some locations in the United Kingdom and has consistently generated better outcomes for children and young people (Edwards et al., 2016). Further economic analyses compared IY to no action, using evidence from previous studies in the United Kingdom and elsewhere (Gardner et al., 2017). The cost of the program is fully recovered by the public purse within 4 years, with further savings over subsequent years. Those savings are spread across numerous sectors: NHS, social services, education, and criminal justice.
Anti-Bullying Programs in Schools
Around 40% of children report being bullied (Department of Health, 2015). It can take many forms, with cyberbullying becoming a growing concern. Children who have been bullied have a higher risk of mental health problems as young people, and there are long-lasting impacts on health, employment, and earnings that last well into adulthood (Takizawa, Maughan, & Arseneault, 2014).
Anti-bullying interventions can reduce cases of bullying, and this is likely to lead to long-lasting health and economic benefits. KiVa is an evidence-based anti-bullying program that can be delivered by teachers; it focuses on enhancing empathy, self-efficacy, and anti-bullying attitudes. It has already been implemented in some schools in England and Wales, with randomized controlled trials currently underway (Clarkson et al., 2016). An exploratory economic analysis of KiVa suggested that cost of delivery is small but generates substantial projected savings in public sector spending and lost earnings, savings, and home ownership up to age 50 (McDaid, Park, & Knapp, 2017). These economic benefits arise because KiVa reduces self-harm and contact with health services. It has even greater returns on investment when long-term losses to individuals are considered.
Early Intervention Services for Psychosis
Most psychoses first appear in adolescence or young adulthood, with potentially serious disruption to school and post-school participation, educational outcomes, and longer-term employment. In turn, this disruption can negatively affect income, social roles, participation, and self-concept. The longer that psychosis remains untreated, the worse the health and other outcomes, including higher suicide risk. Other consequences of psychoses include worse (general) health and links to violence (both perpetration and victimization).
Early intervention (EI) services offer intensive multifaceted support for young people experiencing a first psychotic episode and are effective in alleviating clinical symptoms and improving vocational and social functioning. There is a compelling economic case for EI compared to standard care (McCrone et al., 2010). There are potentially both short- and long-term savings generated by EI services, accruing to the NHS (e.g., through reduced inpatient admissions) and productivity gains through better engagement with employment. Annual savings can also be demonstrated linked to reduced risk of suicide.
Individual Placement and Support
Employment rates are low for people with schizophrenia and other serious mental illnesses (OECD, 2012), not simply because of symptoms but because of negative stigmatizing attitudes among employers and employees. Paid “open” employment may not be suitable for everyone with a serious mental illness, but can be a key element of social and economic inclusion, and of recovery (in the personal sense).
Individual Placement and Support (IPS) aims to help individuals obtain competitive employment, and then support and provide on-the-job training to help them to maintain employment. Among the principles of IPS approach is the focus on finding competitive rather than sheltered employment, job searches based on personal preferences, employment specialists integrated into clinical teams, individualized time-unlimited support for both employees and employer, and advice to support transition from benefits to work (Bond, Drake, & Becker, 2012).
IPS has repeatedly been shown to be effective in achieving higher rates of competitive employment, duration of employment and hours worked compared to other approaches to employment support (Burns et al., 2007), and to be cost-effective (Knapp, Patel, et al., 2013). Indeed, the latter study is one of the few cost-benefit analyses conducted in the mental health field. Two primary outcomes were explored in this six-country randomized trial: additional days worked in competitive settings and additional percentage of individuals who worked at least one day in open employment. The former could be converted to a monetary measure by attaching a value equal to average gross wage (assumed to be equal to the value of productivity added to the economy). This revealed a large difference in net benefit in favor of IPS compared to standard vocational rehabilitation.
Collaborative Care for Physical Health Problems
People with long-term somatic health problems are at greater risk of mental illness, which in turn can affect self-management, poorer health outcomes and wellbeing, and higher health care costs. This can lead to substantial but avoidable costs to the health care system: overall health care costs might be as much as 45% higher for each individual with a long-term condition and co-morbid mental illness (Naylor et al., 2012). Multi-morbidity also increases the need for caregiver support and reduces labor force participation.
A collaborative care approach delivered in primary care (where a specially trained individual such as a nurse helps co-ordinate between health professionals and provides psychological interventions such as problem-solving therapy or CBT) for individuals living with diabetes and/or coronary heart disease as well as depression can be both effective and cost-effective (Johnson et al., 2016). Recent simulation modeling for Public Health England showed, even under conservative assumptions, a substantial return on investment from a societal perspective and a relatively low cost per QALY gained from a health and social care perspective (McDaid et al., 2017).
Although relatively rare, suicide rates are increasing in many countries, including in the United Kingdom. Suicide is an especially distressing consequence of poor mental health. Suicide prevention features prominently in national policy objectives. There are substantial personal and economic costs associated with both completed and non-fatal suicidal events: the average cost per completed suicide for people of working age in England has been estimated at £1.67m (2009 prices), including intangible costs (loss of life to the individual; pain and suffering of relatives), lost output (both waged and unwaged), police time, and the costs of coroner inquests (McDaid, 2016). There are also substantial NHS costs from recurrent nonfatal suicidal events (Sinclair, Gray, Rivero-Arias, Saunders, & Hawton, 2011).
Many different approaches can be taken to prevent suicide, and NICE (2013) guidance in England recommends a multi-component approach, including measures to restrict access to means; making transportation safer; reducing harmful drinking; addressing risks to individuals when in police custody or prison; training GPs, police and teachers to recognize potential risks; and psychosocial assessment for most individuals who present at hospital following deliberate self-harm.
Economic evidence a few years ago showed the economic case for training GPs followed by use of appropriate psychological interventions, as well as measures to reduce risks at potential suicide hotspots (bridges) to reduce the risk of self-harm and suicide (Knapp et al., 2011). More recent modeling, built on a Cochrane systematic review (Hawton et al., 2016), shows that increasing the use of psychosocial assessment when individuals present to hospital emergency departments has substantial economic benefits (McDaid et al., 2017).
When considering whether to invest in an intervention in the mental health area—whether it is a treatment for an established illness or a preventive strategy to stop some illnesses occurring in the first place—the primary question should always ask whether that intervention is likely to be effective. Will it prevent illness, reduce symptoms, improve functioning, or allow individuals to achieve personally defined goals? But decision makers will also want to know the economic consequences of any such investments.
A number of examples of interventions have been given in this article. They constitute just a small selection of those previously shown to be both effective and cost-effective. In fact, across the health sector, being cost-effective, but not cost-neutral or cost-saving, may not always be sufficient to justify expenditure, given the perennial tightness of healthcare budgets.
Decisions about whether to invest in mental health interventions can be complicated by a range of challenges. One of those challenges is that most mental illnesses have spillover effects in other domains of life, and so in other areas of public policy too. The immediate costs of responding to the needs of children with emotional or behavioral problems, for example, tend to be higher in schools than in the healthcare sector. In the longer term, failure to respond adequately to childhood behavioral problems will generate far higher costs to the criminal justice system than to the health system. By far the biggest economic impact of depression in adulthood is lost productivity because of disrupted employment, much greater even than crisis-response costs to healthcare providers.
Many of those spillover impacts are hidden from view. Lost productivity arising through unemployment or absenteeism may be visible and relatively straightforward to measure, but lost productivity through presenteeism (when someone is at work but finding it hard to perform up to their normal standard) is inherently hard to conceptualize, let alone quantify. Similarly, the unpaid support provided by family members to people with mental illnesses is largely hidden from view, and yet those caring responsibilities can damage a caregiver’s own health, wellbeing, and livelihood.
Another challenge is the long-term nature of many mental illnesses, even when treatment can be effective in alleviating symptoms in the short term. The economic benefits of an intervention might equivalently be spread over many years. However, investing in those interventions means that funds have to be found now, and the performance of decision makers is often judged within very short time horizons.
This leads to the related challenge of what can be called diagonal accounting: the costs of an intervention need to be compared with impacts (economic or otherwise) that not only may largely accrue in a different sector (different budget), but also in a different time period, perhaps many years hence. When public funds are very tight, it may not be easy to persuade decision makers to spend more from their own budgets when the main pay-offs are seen in other budgets and in future financial reporting periods. The costs of setting up a treatment program or prevention strategy in one year would need to be compared with economic and health-related gains that are not accrue in another sector but also in later years. When budgets are especially tight, it is often hard to persuade a decision maker to spend more of their own budget if the pay-offs are not going to be seen immediately and not even within their own service or system.
Another problem is that some of the savings suggested by research may not be cashable, at least immediately. An intervention that reduces hospital admissions but does not (immediately or ever) lead to the closure of inpatient beds might not generate any savings at all. Reducing burdens on unpaid family caregivers may not release any resources that are transferable to other uses with measurable economic benefits. Part of the problem, of course, is the way that economic benefits are conceptualized in terms of demonstrable contributions to the economy (such as within GDP calculations), but there is the wider difficulty of making politically awkward decisions to, for example, close a hospital.
There are no simple solutions. Many strategies have been discussed or tried, but appealing to decision makers to play the long game is probably insufficient on its own given the current state of public finances and the ways in which performance is assessed and publicly discussed. It might be possible to identify pay-offs to the funder in the short term that are sufficient to justify investment, but it is more likely that additional funds will need to be made available up front to get some interventions started.
Another approach would be to try to bring different systems together to agree a joint strategy or some form of cross-budget transfer of resources, and there are of course many local experiments now underway across the United Kingdome to explore different modes of coordinated action, or integration, involving one or more parts of the NHS and a range of other public bodies. There have also been experiments with new models of finance, such as social impact bonds.
Knapp, M., & Iemmi, V. (2016). Mental health. In Scheffler, R. (Ed.), World Scientific handbook of health economics. Singapore: World Scientific Press.Find this resource:
Knapp, M., McDaid, D., Mossialos, E., & Thornicroft, G. (Eds.). (2007). Mental health policy and practice across Europe. Maidenhead, U.K.: Open University Press.Find this resource:
Livingston, G., Sommerlad, A., Orgeta, V., Costafreda, S. G., Huntley, J., Ames, D.,. . . Cooper, C. (2017). Dementia prevention, intervention, and care. The Lancet, 390(10113), 2673–2734.Find this resource:
Lund, C., De Silva, M., Plagerson, S., Cooper, S., Chisholm, D., Das, J.,. . . Patel, V. (2011). Poverty and mental disorders: Breaking the cycle in low-income and middle-income countries. The Lancet, 378(9801), 1502–1514.Find this resource:
Mental Health Task Force (2016). The five year forward view for mental health. NHS England.Find this resource:
Organisation for Economic Cooperation and Development. (2014). Making mental health count: The social and economic costs of neglecting mental health care. Paris, France: OECD.Find this resource:
Patel, V., Saxena, S., Lund, C., Thornicroft, G., Baingana, F., Bolton, P.,. . . Herrman, H. (2018). Commission on global mental health and sustainable development.The Lancet, 392(10157), 1553–1598.Find this resource:
Prince, M., Patel, V., Saxena, S., Maj, M., Maselko., J., Phillips, M. R., & Rahman, A. (2007). No health without mental health. The Lancet, 370(9590), 859–877.Find this resource:
Saxena, S., Thornicroft, G., Knapp, M., & Whiteford, H. (2007). Resources for mental health: Scarcity, inequity, and inefficiency. The Lancet, 370(9590), 878–889.Find this resource:
Slade, M., Amering, M., Farkas, M., Hamilton, B., O’Hagan, M., Panther, G., . . . Whiltley, R. (2014). Uses and abuses of recovery: Implementing recovery-oriented practices in mental health systems. World Psychiatry, 13(1), 12–20.Find this resource:
Thornicroft, G. (2006). Shunned: Discrimination against people with mental illness. Oxford: Oxford University Press.Find this resource:
Whiteford, H. A., Degenhardt, L., Rehm, J., Baxter, A. J., Ferrari, A. J., Erskine, H. E., . . . Burstein, R.(2013). Global burden of disease attributable to mental and substance use disorders: Findings from the Global Burden of Disease Study 2010. The Lancet, 382(9904), 1575–1586.Find this resource:
World Health Organization. (2018). Mental Health Atlas 2017. Geneva, Switzerland: World Health Organization.Find this resource:
Bauer, A., Parsonage, M., Knapp, M., Iemmi, V., & Adelaja, B. (2015). The costs of perinatal mental health problems. London: Centre for Mental Health and PSSRU, London School of Economics and Political Science.Find this resource:
Bond, G. R., Drake, R. E., & Becker, D. R. (2012). Generalizability of the individual placement and support (IPS) model of supported employment outside the US. World Psychiatry, 11(1), 32–39.Find this resource:
Bonin, E., Stevens, M., Beecham, J., Byford, S., & Parsonage, M. (2011). Costs and longer-term savings of parenting programmes for the prevention of persistent conduct disorder: a modelling study. BMC Public Health, 11(1), 803.Find this resource:
Brazier, J. (2010). Is the EQ-5D fit for purpose in mental health? British Journal of Psychiatry, 197(5), 348–349.Find this resource:
Briggs, A., Claxton, K., & Sculpher, M. (2006). Decision modelling for health economic evaluation. Oxford: Oxford University Press.Find this resource:
Brimblecombe, N., Evans-Lacko, S., Knapp, M., King, D., Takizawa, R., Maughan, B., & Arseneault, L. (2018). Long term economic impact associated with childhood bullying victimisation. Social Science and Medicine, 208, 134–141.Find this resource:
Burns, T., Catty, J., Becker, T., Drake, R. E., Fioritti, A., Knapp, M., . . . White, S. (2007). The effectiveness of supported employment for people with severe mental illness: A randomised controlled trial. The Lancet, 370(9593), 1146–1152.Find this resource:
Chisholm, D., van Ommeren, M., Ayuso-Mateos, J. L., & Saxena, S. (2005). Cost-effectiveness of clinical interventions for reducing the global burden of bipolar disorder. British Journal of Psychiatry, 187(6), 559–567.Find this resource:
Clarkson, S., Axford, N., Berry, V., Edwards, R. T., Bjornstad, G., Wrigley, Z., . . . Hutchings, J. (2016). Effectiveness and micro-costing of the KiVa school-based bullying prevention programme in Wales: Study protocol for a pragmatic definitive parallel group cluster randomised controlled trial. BMC Public Health, 16, 104.Find this resource:
Department of Health. (2015). Future in mind. London: Department of Health.Find this resource:
Edwards, R. T., Jones, C., Berry, V., Charles, J., Linck, P., Bywater, T., & Hutchings, J. (2016). Incredible years parenting programme: Cost-effectiveness and implementation. Journal of Children’s Services, 11(1), 54–72.Find this resource:
Eranti, S., Mogg, A., Pluck, G., Landau, S., Purvis, R., Brown, R. G., . . . McLoughlin, D. M. (2007). A randomized controlled trial with 6-month follow-up of repetitive transcranial magnetic stimulation and electroconvulsive therapy for severe depression. American Journal of Psychiatry, 164(1), 73–81.Find this resource:
EuroQol Group. (1990). EuroQoL: A new facility for the measurement of health-related quality of life. Health Policy, 16(3), 199–208.Find this resource:
Gardner, F., Leijten, P., Mann, J., Landau, S., Harris, V., Beecham, J., . . . Scott, S. (2017). Could scale-up of parenting programmes improve child disruptive behaviour and reduce social inequalities? Using individual participant data meta-analysis to establish for whom programmes are effective and cost-effective. NIHR Public Health Research, 1(11).Find this resource:
Green, H., Mcginnity, A., Meltzer, H., Ford, T., & Goodman, R. (2005). Mental health of children and young people in Great Britain: 2004. London: Office for National Statistics.Find this resource:
Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery, and Psychiatry, 23(1), 56–62.Find this resource:
Hawton, K., Witt, K. G., Salisbury, T. L., Arensman, E., Gunnell, D., Hazell, P., van Heeringen, K. (2016). Psychosocial interventions following self-harm in adults: a systematic review and meta-analysis. The Lancet Psychiatry, 3, 740–750.Find this resource:
Johnson, J., Lier, D., Soprovich, A., Al Sayah, F., Qiu, W., & Majumdar, S. R. (2016) Cost-effectiveness evaluation of collaborative care for diabetes and depression in primary care. American Journal of Preventive Medicine, 51(1), e13–e20.Find this resource:
Kessler, R. C., Berglund, P., Demler, O., Jin, R., Merikangas, K. R., & Walters, E. E. (2005). Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry, 62(6), 593–602.Find this resource:
Knapp, M., King, D., Romeo, R., Schehl, B., Barber, J., Griffin, M., . . . Hoe, J. (2013). The START study (STrAtegies for RelaTives): Cost-effectiveness of a manual-based coping strategy programme in promoting the mental health of family carers of people with dementia: A pragmatic randomised controlled trial. British Medical Journal, 347, f6342.Find this resource:
Knapp, M., McDaid, D., & Parsonage, M. (2011). Mental Health Promotion and Mental Illness Prevention: The Economic Case. London: Department of Health.Find this resource:
Knapp, M., Patel, A., Curran, C., Latimer, E., Catty, J., Becker, T., . . . Burns, T. (2013). Supported employment: cost-effectiveness across six European sites. World Psychiatry, 12(1), 60–68.Find this resource:
Knapp, M., Romeo, R., Mogg, A., Eranti, S., Pluck, G., Purvis, R., . . . Edwards, D. (2008). Cost-effectiveness of transcranial magnetic stimulation vs. electroconvulsive therapy for severe depression: A multi-centre randomised controlled trial. Journal of Affective Disorders, 109(3), 273–285.Find this resource:
Knapp, M., Snell, T., Healey, A., Guglani, S., Evans-Lacko, S., Fernandez, J. L., . . . Ford, T. (2015). How do child and adolescent mental health problems influence public sector costs? Inter-individual variations in a nationally representative British sample. Journal of Child Psychology and Psychiatry, 56(6), 667–676.Find this resource:
Livingston, G., Barber, J., Rapaport, P., Knapp, M., Griffin, M., King, D., . . . Sampson, E. L. (2013). The START study (STrAtegies for RelaTives): clinical effectiveness of a manual based coping strategy programme in promoting the mental health of family carers of people with dementia: a pragmatic randomised controlled trial. British Medical Journal, 347, f6276.Find this resource:
McCrone, P., Craig, T., Power, P., Garety, P. (2010). Cost-effectiveness of an early intervention service for people with psychosis. British Journal of Psychiatry. 196(5), 377–382.Find this resource:
McCrone, P., Knapp, M., Proudfoot, J., Ryden, C., Cavanagh, K., Shapiro, D. A., . . . & Marks, I. (2004). Cost-effectiveness of computerised cognitive behavioural therapy for anxiety and depression in primary care: Randomised controlled trial. British Journal of Psychiatry, 185(1), 55–62.Find this resource:
McDaid, D. (2016). Making an economic case for investing in suicide prevention: Quo vadis? In R. O’Connor & J. Pirkis (Eds.), International handbook of suicide prevention: Research, policy and practice (2nd ed.). Chichester, U.K.: Wiley Blackwell.Find this resource:
McDaid, D., Hopkin, G., Knapp, M., Brimblecombe, N., Evans-Lacko, S., & Gan, C. (2017). The economic case for prevention in young people’s mental health. London: PSSRU.Find this resource:
McDaid, D., Park, A., & Knapp, M. (2017). Commissioning cost-effective services for promotion of mental health and wellbeing and prevention of mental ill-health. London: Public Health England.Find this resource:
McManus, S., Bebbington, P., Jenkins, R., & Brugha, T. (2016). Mental health and wellbeing in England: Adult psychiatric morbidity survey 2014. Leeds: NHS Digital.Find this resource:
Morrell, C. J., Sutcliffe, P., Booth, A., Stevens, J., Scope, A., Stevenson, M., . . . Stewart-Brown, S. (2016). A systematic review, evidence synthesis, and meta-analysis of quantitative and qualitative studies evaluating the clinical effectiveness, the cost-effectiveness, safety, and acceptability of interventions to prevent postnatal depression. Health Technology Assessment, 20(37).Find this resource:
National Institute for Health and Care Excellence. (2013). Self-harm. Quality Standard [QS34], London, NICE.Find this resource:
Naylor, C., Parsonage, M., McDaid, D., Knapp, M., Fossey, M., & Galea, A. (2012). Long-term conditions and mental health: The cost of co-morbidities. London: The King’s Fund.Find this resource:
Organization for Economic Cooperation and Development (2012). Sick on the Job? Myths and realities about mental health and work. Paris, France: OECD.Find this resource:
Petrou, S., & Gray, A. (2011). Economic evaluation alongside randomised controlled trials: Design, conduct, analysis, and reporting. British Medical Journal, 342, d1766.Find this resource:
Prince, M., Knapp, M., Guerchet, M., McCrone, P., Prina, M., Comas-Herrera, A., . . . Salimkumar, D. (2014). Dementia UK: Update. London: Alzheimer’s Society.Find this resource:
Proudfoot, J., Ryden, C., Everitt, B., Shapiro, D. A., Goldberg, D., Mann, A., . . . Gray, J. A. (2004). Clinical efficacy of computerised cognitive-behavioural therapy for anxiety and depression in primary care: randomised controlled trail. British Journal of Psychiatry, 185(1), 46–54.Find this resource:
Richardson, J., & Joughin, C. (2002). Parent training programmes for the management of young children with conduct disorders: Findings from research. London: Gaskell.Find this resource:
Sinclair, J., Gray, A., Rivero-Arias, O., Saunders, K. E., & Hawton, K. (2011). Healthcare and social services resource use and costs of self-harm patients. Social Psychiatry and Psychiatric Epidemiology, 46(4), 263–271.Find this resource:
Takizawa, R., Maughan, B. & Arseneault, L. (2014). Adult health outcomes of childhood bullying victimization: Evidence from a five-decade longitudinal British birth cohort. American Journal of Psychiatry, 171(7), 777–784.Find this resource:
Vigo, D., Thornicroft, G., Atun, R. (2016). Estimating the true global burden of mental illness. The Lancet Psychiatry, 3(2), 171–178.Find this resource: