The Scottish Government

« Previous | Contents | Next »

Listen

ANNEX 3 ASSESSING THE COST EFFECTIVENESS OF AREA-BASED SUICIDE PREVENTION STRATEGIES: PRACTICAL AND METHODOLOGICAL CHALLENGES.

Introduction

Economics is concerned with how we allocate resources between competing activities in society and what the consequences are of such resource allocation decisions. It is based on the fundamental tenet that there is a finite level of resources and therefore most decisions to use resources in one way will mean that resources are not available for other activities.

It can be difficult to identify the levels of expenditure on national suicide prevention activities, due in part to a lack of earmarked funding, and also because strategies may be delivered across many sectors, by many different public and private agencies, often funded in completed different manners. Nevertheless it is clear that substantial levels of funding may be allocated towards such strategies - for instance £19.5 million ($A48 million) was invested in suicide prevention in Australia between 2000 and 2004.

Economic evaluation, which compares both the effectiveness and costs of one or more programmes or individual interventions, can be a useful aid to policy makers in assessing whether such an investment in suicide prevention activities represents value for money. Ultimately, for Choose Life and other national suicide prevention strategies, we would want to be able compare the cost of implementation with their impact on the rate of completed suicides and non fatal deliberate self-harm events.

Aims

This paper has a number of aims. Firstly it briefly defines different approaches to economic evaluation. It then looks at the extent to which these different methods of economic evaluation have been applied to the assessment of area based suicide prevention strategies drawing on evidence from a literature review. We then consider what practical and methodological challenges there may be to the greater use of economic evaluation. A practical approach is outlined that, in the absence of robust information on the effectiveness of suicide prevention strategies, may be helpful to policy makers in informing initial decisions about the potential cost effectiveness of suicide prevention strategies. Finally we suggest some key steps to facilitate the potential inclusion of economic evaluation in any future evaluation of Choose Life or similar programmes.

A brief primer on economic evaluation

It is perhaps helpful to start by briefly outlining the concept of economic evaluation. Useful guides some which deal with these issues in more depth, including some with a relevant focus on health promotion and complex interventions are available (Byford et al. 2003; Drummond et al. 2005; Hale et al. 2005).

There are several different approaches; they all estimate costs in the same way but differ in they way they treat outcomes. Cost minimisation analysis only compares the costs of interventions; it can only be undertaken where there is strong evidence that interventions are equally effective. The most frequently used approach, cost effectiveness analysis ( CEA) measures effects using a natural measure, in this case the number of suicides or deliberate self-harm event averted.

While intuitively easy to understand, this approach makes it is difficult to compare investment in mental health interventions with that of investment in other areas of health policy as no common unit of outcome is used. In respect of prevention of suicide, there may also be additional outcomes that may be of importance, so focusing on one measure of effectiveness alone may not be helpful. One variant on CEA that can be seen in mental health economic evaluations, the cost-consequence analysis ( CCA) seeks to address this problem by presenting the costs and a range of outcomes for interventions without making any judgement as to whether one outcome is of more importance than another.

Cost utility analysis ( CUA) uses a common outcome measure where benefits are measured in utilities, i.e., the individual's preference for a specific level of health or a specific health outcome, with examples being the Quality Adjusted Life Year ( QALY) and the Disability Adjusted Life Year ( DALY). This would allow the benefits of investing in suicide prevention to be compared with other health programmes. Many different instruments can be used to identify different levels of states of health. These can include generic instruments such as the EuroQol, which assesses health status along five dimensions including anxiety /depression, as well as disease specific instruments such as the McSad health state classification system for depression (Bennett et al. 2000). Different techniques can then be used to get individuals to estimate the utility (or their level of satisfaction) associated with these health states (See (McCulloch 2003) for a description of some of these approaches). In the case of the DALY - different levels of disability e.g. level of mobility and their relative weights have been determined by a small expert group rather than the general population. The use of CUA for suicide prevention is complex; some individuals experiencing suicidal thought may place a greater value on death compared to life (Chisholm et al. 2006).

In cost benefit analysis all costs and benefits are valued in the same (monetary) units. With two or more alternatives, the intervention with the greatest net benefit would be deemed the most efficient. CBAs are thus intrinsically attractive, but conducting them can be problematic because of the difficulties associated with valuing outcomes in monetary terms and their use in mental health has been limited (Healey & Chisholm 1999). If the methodological and practical challenges facing CBA can be overcome it has the potential to allow decision makers to consider the merits not only of allocating resources within health care but also whether it would be more appropriate to invest in other sectors such as housing, education or defence. This is clearly important given that the benefits of suicide prevention go well beyond the health care system.

Literature review - methods

To what extent has economic evaluation of suicide prevention strategies taken place? An extensive review of the literature was conducted. A number of bibliographic databases were searched including Medline (Pubmed version incorporating HealthStar), Econlit, and Psychinfo. Searches combined key economic terms such as the Medical Subject Heading ( MESH) term "costs and cost analysis" with various terms for suicide, attempted suicide and deliberate self-harm. No language restrictions or time limits were employed. Evaluations of drug therapies were explicitly excluded from this analysis.

A hand search of a number of journals to identify relevant papers during the four years 2002 to 2005 inclusive was also conducted. Journals searched include Crisis, Journal of Socio-Economics, and Acta Psychiatrica Scandinavica. National websites for suicide prevention strategies were also examined, firstly to identify information on the level of investment in these strategies where available, and secondly to identify any economic studies or plans to conduct economic evaluation.

Results of review

Over 100 published papers, book chapters, governmental documents and grey literature that potentially were of relevance on the basis of their abstracts and titles were retrieved. Our review suggests that there have been few economic evaluations generally in the area of mental health promotion and mental illness prevention. Studies that do exist focus on secondary prevention or targeted interventions for specific individuals or sub population groups.

Given the limited knowledge on the effectiveness at programme level of national strategies, (a necessary prerequisite to economic evaluation) it might be considered unsurprising that no economic evaluations of national suicide prevention strategies were found. Similar observations have been seen elsewhere, for instance the US Preventive Services Task Force in its review of the evidence on the effectiveness of screening for suicide risk, failed to identify any studies that included an economic evaluation (nor for that matter any evidence that screening for suicide reduces suicide attempts or mortality) ( US Preventive Services Task Force 2004).

Yet despite this lack of cost effectiveness studies, what was also apparent from our review is that many agencies and reviews have for some significant time argued for the greater use of cost effectiveness analysis in reviewing suicide prevention strategies. Most notably the US Surgeon General stressed the importance of economic evaluation when setting out his public mental health strategy for suicide prevention ( US Public Health Service 1999)

This is not to say that no economic evaluations of area based suicide prevention strategies have been conducted, but they are rare and modest in scope. Perhaps the most complete, albeit still with many limitations, is a retrospective analysis of an area based multi- intervention suicide prevention programmes for the Western Athabaskan Native American Tribe living in a reservation town in New Mexico . This mental health status of this indigenous population in the US is poor; the rate of suicide is high, around 1.5. times higher than the general population.

The New Mexico prevention strategy primarily focused on young people between the ages of 15-19 but had the whole community as a secondary target group. The initiative included the training of between 10 and 25 youths per annum to respond to young people in crisis and to refer individuals to the appropriate mental health services. This was one element of a many different elements within the programme including: postvention outreach, community education on suicide prevention, and suicide-risk screening in health and social care programmes.

Both a cost benefit analysis and a cost utility analysis were performed. Costs in the analysis included direct costs associated with the use of emergency services as well as public health and social care services. Broader costs to insurers, local government and the general public were also included. The evaluation of outcomes took the form of a before and after study. Rates of suicidal acts (completed and attempted suicides)
8 years before the introduction of the intervention were compared with the rate over the subsequent 10 years.

As figure A1 indicates, the rate of suicidal acts (completed suicide and suicide attempts) fell markedly after the introduction of the programme from 59 per 1,000 to under 11 per 1,000 ten years later. Direct costs of $120,000 were avoided and it was estimated that the value of suicides averted was some $1.7 million. The cost per QALY saved was just $419, while the value of benefits gained were 43 times greater than costs incurred. While this study might suggest that such a programme can be highly cost effective there are many limitations. Most notably no comparator group was included in the evaluation; cyclical increases and decreases in the suicide rates in the tribe had in fact been observed every six years between 1957 and 1987. Moreover the analysis by its nature makes it difficult to determine which individual elements of the strategy were effective.

Figure A1 Long term change in suicide rates following introduction of suicide prevention initiative

Another US based study is a hypothetical cost benefit analysis of two suicide prevention programmes targeted at university students in Florida (de Castro et al. 2004). One was a five lesson general suicide education programme and the other a peer support group programme. Data on the effectiveness of the two interventions were obtained from a review of the literature and costs estimated of delivering the two interventions to all university students in the state. In fact this study was a cost-offset analysis as the potential costs of delivering the intervention were compared with the lifetime costs that would be avoided if suicides were prevented - nevertheless both interventions the study concluded would be cost saving as these cost offsets would be far greater than the costs of implementation. The peer support programme had a lower cost and potentially greater level of costs avoided, however the costs of implementation are underestimated as the authors themselves admitted that they had not put a value on the significant time input of volunteers in delivering peer support programmes.

Although our brief here is not to look systematically for evidence of cost effectiveness studies targeted at the individual rather than population level, it is worth noting that other than evaluations of antidepressants and other medications which sometimes consider suicide as an outcome, few such evaluations appear to have been conducted. One exception is an analysis in England of a home based social work intervention targeted at children who had deliberately poisoned themselves.(Byford et al. 1999) This concluded that family-based social work intervention for children and adolescents who have deliberately poisoned themselves is as cost-effective as routine care alone. Suicidal ideation was lower in a sub group of children without depression at a six month follow up, with no difference in costs.

A more recent study looked at the use of cognitive behavioural therapy with people with a history of deliberate self-harm in centres in both England and Scotland. This study suggested that manual cognitive behaviour therapy was likely to be cost effective in reducing the number of deliberate self-harm events, but it did not look at suicide as a potential outcome.

The potential costs and consequences of suicide prevention centres in the US were assessed in the 1980s (Medoff 1986). These relied on the provision of a 24 hour telephone service whose primary action was to initiate crisis intervention services. A regression model was used to estimate the impact of suicide prevention centres on suicide rates. The economic analysis suggested that the value of human lives saved was at least five time greater than the costs of providing suicide prevention centres. Again however little evidence on the effectiveness of these centres was provided.

An economic analysis of an educational programme for general practitioners, so as to improve their ability to recognise and treat the symptoms of depression, was conducted on the Swedish island of Gotland in the late 1980s (Rutz et al. 1992). Although called a cost benefit analysis, in fact this study actually reported costs avoided, including those associated with suicide, rather than using a formal cost benefit analysis technique such as assessing willingness to pay for this programme. It was concluded that the programme was cost saving, although evidence on the effectiveness of the intervention was limited by the lack of a comparison group.

While no studies of the cost effectiveness of screening strategies for suicide in the population were identified we did identify one study that had evaluated the costs of delivering a school based mental health screening and treatment programme in New York, one of whose stated objectives was the prevention of suicide . This study not only included the direct costs of delivering the programme but also resource inputs in-kind such as the time of student volunteers. Elsewhere there has been some economic appraisal of various safety measures (such as safety nets and barriers for bridges) and restriction of access to means such as firearms and poisons, the latter usually in the context of injury or violence prevention rather the suicide per se.

Macroeconomic factors, economic behaviour and suicide

It is worth noting that while little work has been done to look at interventions. Economists and those from other disciplines have sought to use to economic theory to explain the rationale for suicide and suicide attempts and thus identify potential ways in which to influence individual behaviour. (Hamermesh & Soss 1974; Yaniv 2001; Marcotte 2003). Other analysis has focused more on the impact of, and changes in the socio-economic circumstances of the individual and their community. For instance much research has looked at the links between unemployment and socio-economic deprivation and suicide or suicide attempts e.g. (Platt & Hawton 2000; Rodriguez-Andres 2005). One study looked at attempted suicide trends in Helsinki, during an economic recession that Finland experienced during most of the 1990s (Ostamo & Lonnqvist 2001). This particular study observed that suicide attempt rates remained high but stable, rather than increasing during the recession, with the authors suggesting that this indicated that there were individuals at risk of suicidal behaviour regardless of social and economic change in society.

Increasingly complex econometric analysis has been used to examine the interaction of a whole range of socio-economic factors on suicide rates e.g. (Lucey et al. 2005; Berk et al. 2006). There remains much scope for further work to identify the specific impacts of individual factors or combinations of factors on the risk of suicide. The nature of the relationship between various socio-economic factors and suicide is complex; there is conflicting evidence from studies as to whether factors such as poverty or the level of interest rates etc are risk or protective factors for suicide.

These factors might also vary depending on the age and gender of the individual for example. The different conclusions reached by studies are also likely to reflect differences both in methodologies and in the size and type of study populations and their environs e.g. access to social welfare support. One recent review of the association between the socio-economic status of an area and completed suicide sought to disentangle some of these factors. It came to the conclusion that suicide prevention resources might best be concentrated in 'communities of a relatively lower socio-economic level, in particular those at high levels of concentrated disadvantage' (Rehkopf & Buka 2006).

Practical and methodological challenges

Why has there been so little economic evaluation of suicide prevention strategies to date, despite the many calls for such evaluations? There do not appear to be any insurmountable problems in applying conventional methods of economic evaluation to this area. It might be argued that economic analysis focusing on a single outcome such as the rate of suicide or quality adjusted life years saved may be too reductionist for the evaluation of any public health or health promoting intervention (Kelly et al. 2005) - but we have noted pragmatic approaches such as cost consequences analysis, or the more theoretically robust cost benefit analysis method, can be used to take account of multiple outcomes of interest.

One additional problem however for cost benefit analysis may potentially be the undervaluing by the general public in willingness to pay studies of interventions intended to promote and improve mental well-being, because of negative attitudes towards poor mental health. While we are unaware of any empirical evidence testing this assertion, some public surveys have indicated that mental health is seen as a low priority when it comes to determining how to allocate health system funds (Matschinger & Angermeyer 2004).

Instead the primary reason for the limited application of economic evaluation appears to be the lack of evaluation of the effectiveness of area based suicide prevention strategies; evaluation has concentrated on the impact of individual interventions such as primary care physician training programmes or restricting access to means.

We do not seek here to discuss the challenges of outcome evaluation, but raise two key issues for illustrative purposes. Firstly attributing changes in the rate of suicide to the delivery of complex multi-faceted programmes are difficult to determine. For national strategies, the possibilities of having comparator group may be limited, and instead analysis tends to focus on changes in long term suicide trends. This makes it more difficult to determine what change in suicide rates would have occurred in the absence of any strategy. Programme level evaluation may also mask the effectiveness of individual components of the strategy, again adding an additional layer of complexity to any evaluation.

It would be interesting to compare the relative success across different local areas that may deliver a different mix of interventions as part of a national suicide prevention strategy. Yet despite the profound and terrible consequences of suicide, in absolute terms the number of suicides in many areas can be very small - this can make it difficult to identify any significant impact of programmes. Other potential measures, such as levels of suicidal ideation at suicide attempts may be used instead, but the relationship between these measures and the level of completed suicides is not necessarily straightforward.

What then of the use of economic evaluation as part of the evaluation of individual components of a suicide prevention strategy? Again there is little evidence suggesting that much analysis has been conducted. Again methodologically there do not appear to be any great challenges, although there may practical issues in data collection, such as both obtaining accurate information, and subsequently valuing the level of inputs, from volunteers to suicide prevention activities.

In some respects the lack of studies may reflect the relatively low interest, until recently at least, in funding economic evaluations (of non drug therapies)in the area of mental health. It is also consistent with the limited number of economic evaluations of public health and health promoting interventions; again an area where funding for economic evaluation (in comparison to evaluation of health care interventions) has been more limited. Another reason may be the limited amount of multi-disciplinary work between health economists and those in the public health community. Clearly the evaluation of national suicide prevention strategies requires the input of individuals from a number of disciplines, including economics. Commissioners of evaluations can play an important role by including an economic dimension in study specifications.

Illustration of potential cost effectiveness of Choose Life using an economic threshold analysis

We have argued that the primary reason for the limited used of economic analysis would appear to be the challenge in obtaining data on effectiveness rather than difficulties in using existing methods of economic evaluation.

This does not mean that we cannot provide an input into the policy making process. A practical approach is outlined below, in the absence of robust information on the effectiveness of suicide prevention strategies, that still may be helpful to policy makers in informing initial decisions about the potential cost effectiveness of suicide prevention strategies. If the level of improvement in outcomes required in order for the strategy to be considered cost effective is modest, then the risks in implementing the strategy may be low. Implementing the strategy, including an evaluation, to help answers questions on effectiveness may then seem sensible. On the other hand, if the level of improvement in outcomes required is substantial then it may be prudent to consider different approaches. This approach is known as threshold analysis and we illustrate here how it can be used taking data from our analysis of Choose Life.

Threshold analysis

This threshold simply refers to what different societies consider to be cost-effective. This is a subjective judgement depending on many factors, including the level of resources in a country. For instance the National Institute for Health and Clinical Excellence in England and Wales generally consider that if the cost per year of full quality year of life gained is under £30,000 then this represents value for money. Of course decisions never are (nor should they be) made on the basis of cost effectiveness alone; other factors such as fairness, as well as ethical and political considerations will also be important considerations. We may for instance be willing to sacrifice some efficiency in how we can allocate resources in order to reach a sub group of the population who might for instance have very poor levels of health, or in this case a greater risk of suicide. The technique can also be used to crudely estimate how many lives would need to be saved in order for the strategy to represent value for money or even become cost saving, that is where the net benefits from investing in Choose Life outweigh the costs of the strategy.

Assumptions

Box A1 sets out the key components of threshold analysis. In order to complete the analysis we will make use of information that we have collected on the level of resources invested in Choose Life as well on the estimated costs of suicide in Scotland discussed in annex 1.

Initially we will set our threshold for this analysis at a level of £30,000 per life year saved. (We assume that all years saved would be spent in full quality health.)

The next step in the analysis is to obtain evidence on the potential effectiveness of the strategy. In this case, we have little empirical evidence on which to draw, and so instead consider the implications for potential cost effectiveness, by varying the rate of effectiveness between a 1% and 30% reduction in the annual rate of suicide in Scotland.

Our previous analysis which has estimated the lifetime costs of all suicides occurring in Scotland in 2004 can be used to estimate any cost offsets associated with a reduction in suicides.

Information (albeit incomplete) taken from our analysis of the resources invested in Choose Life can be used to provide an estimate of the total cost of investing in Choose Life. This includes not only funding for national programmes, core funding distributed to all 32 CPPs, but also the additional monetary and in-kind resources necessary to deliver the programme.

We also assume here that the comparator against which we could compare this investment is simply to maintain the status quo prior to the introduction of Choose Life - i.e. no new interventions are developed nor additional resources invested in Choose Life.

Costs are discounted at a rate of 3.5% per annum. The discounting of future life years saved is controversial - we present findings here using both non discounted as well as discounted life years saved.

We have taken average life years lost per suicide in 2004 (34 years) as the potential number of years to be gained for each suicide averted.

All results are presented using 2005 price years

Results

Potential years of life that could be saved

For suicides in 2004, and using data on life expectancy from the General Actuary's Office, approximately 28,345 (non-discounted) lifetime years would be lost as a result of suicide in Scotland across all age groups. If discounted at the same base rate used for future costs of 3.5% per annum then the number of years of life lost is reduced to 11,893. In this simple analysis we have assumed that a suicide successfully averted in any one year is permanently averted, i.e. there will be no future suicidal attempts by the individual concerned. More complex models could be built in future to take account of this and other simplifications.

Estimated current annual investment in Choose Life

The total costs of investment in Choose Life currently identified are £4m in core funding per annum, plus an identified £0.52m per annum in additional monetary funding and £0.09m in in-kind investment. We have noted evidence from case studies suggests that these additional investments are underestimated. We have also calculated total investment assuming that the same share of in-kind and monetary investment in the case studies is seen across the whole programme. This would raise the total annual cost of investment in Choose Life to £6.01million with in kind investments accounting for £0.9 million and monetary funding for £1.11 million.

Potential level of acceptable investment in Choose Life

Table A10 provides information on the maximum level of investment that would be consistent with our threshold under different circumstances. Even when future years of potential life saved are discounted at a rate of 3.5% per annum, and the success rate of Choose Life in reducing the annual rate of suicide is just 1%, the value for money of investing in Choose Life appears highly promising. In total, some 160 (discounted) or 284 (non discounted) life years would be saved. Our threshold analysis would indicate that if we were willing to pay £30,000 for each additional year of life saved as a result of Choose Life then we could invest up to £15.6 million per annum, taking account of the £10.79 million in cost offsets generated. This is well in excess of the current level of investment.

If the annual rate of suicide were to fall by 20% then the programme would be highly cost saving. We would save more than 5,600 life years (undiscounted) or 3,200 (discounted) and generate cost offsets to society of almost £216 million. We would have to be spending more than £386 million per annum before the cost per life year saved was above our threshold! Only if the reduction in the rate of suicide was less than 0.4 of one percent would our current level of investment mean that the cost per life year saved was above our threshold.

Even if what society is prepared to pay to save one additional year of life is reduced, this makes little difference to the maximum level of investment. Keeping all our baseline assumptions, reducing this threshold by one third to £20,000, the maximum we could invest would be only be reduced by 11% to just over £14m.

Table A10 Maximum levels of investment in Choose Life to have a cost per life year saved of £30000 or less (2005 prices)

Another way of looking at this is to look at how varying the number of potential lives saved impacts on the potential cost per life year saved (see table A11) Again we use our baseline assumptions on total investment per annum in the programme and potential costs avoided. In the base case scenario only 3.2 lives would need to be saved for the strategy to cost less than £30,000 per life year saved; the threshold below which interventions are generally considered to be cost effective by NICE. This analysis also suggests that only five lives would need to be saved in any one year for the strategy to be cost saving (dominant) compared with no action, that is for the value placed on suicides averted to be in excess of the annual investment in the strategy. As the number of suicides averted increases this benefit: cost ratio would begin to rise considerably.

Table A11 Potential cost per life year saved, varied by number of suicides averted (2005 prices)

* CL strategy is dominant compared to no action, with both lower costs and additional lives saved

Varying perspective

Much economic analysis is conducted not from a societal perspective but from a very narrow public sector perspective. In this case we would only be interested in the direct public costs of investing in Choose Life - £4 million per annum, while the only cost offsets of interest are any costs to the emergency and health care services etc that can be avoided as a result of not having to respond to a suicide.

In this case where only a 1% reduction in suicide was achieved, with life years saved discounted at our base rate of 3.5 %, the threshold for investment would still be £4.88 million, which is still in excess of our current £4 million level of investment. If our analysis also included productivity losses avoided then the threshold for investment would rise to more than £7.87 million. Even if it was assumed that the total level of direct costs attributed to suicide was just 0.67 of one percent of indirect costs (Clayton & Barceló 2000) and with a threshold of just £20,000 per life year saved, society would still be willing to invest £3.2 million in the strategy.

If we look at how this impacts on the cost per life year saved, then 11 lives would have to saved for the strategy to be considered cost effective. It would however take more than 700 lives to be saved for the strategy to become cost saving because of the very low level of direct costs avoided per life saved. If the analysis also included indirect costs avoided, i.e. productivity losses from lost opportunities for paid and non-paid work, then the number of lives per annum that would need to be saved for the strategy to be cost saving would be just over 17.

Implications

Threshold analysis cannot tell us whether investing in Choose Life represents value for money, nor the appropriate level of investment to make. However it does suggest that if the Choose Life initiative does indeed achieve even a very modest reduction in the rate of suicide of just one per cent, then the cost per life year saved is likely to be well below £30,000. This generally would suggest that investing in the programme would represent value for money and the level of success required by the strategy modest. This is also reinforced by looking at the very modest number of lives that would have to be saved in order for the net benefits from the strategy - that is the value placed on lives saved to be greater than the costs of investing in Choose Life. In our base scenario here, this is just five lives per annum. If our perspective was more restricted and we excluded the intangible value placed on life then, we would need to save 17 lives in order to have a positive net benefit. Looking at direct costs only, saving 11 lives per annum is likely to mean that the strategy has a level of cost effectiveness comparable to that of many other interventions recently recommended for funding by the NHS. However it will only when evidence of effectiveness is available that we can more definitively state that investing in Choose Life represents value for money.

There are important limitations of this analysis to flag up - firstly we have not adjusted future years of life to take account of their quality - this would reduce the value of life years saved (it might however be argued that value places on the intangible benefits of life foregone pick this up on the cost side of the equation). Second, we have here compared investing in Choose Life with taking no other action (over and above what is already in place) to tackle suicide. There may be other alternative models or programmes of suicide prevention, i.e. those that are more closely controlled and delivered centrally, that might be better options against which to compare Choose Life. Again, the challenge here is to identify the effectiveness of alternative models and consider whether these could work in a Scottish context.

We have however been highly conservative in not including the potential added benefits of avoiding non fatal deliberate self-harm events in this analysis. Our analysis looks at what would happen if we can reduce the current rate of suicide in Scotland; it may be the case that this rate might naturally fall (or rise) substantially in future years. This might also have implications for the value for money of investing in Choose Life.

A final observation is that it that simply looking at the cost per life year gained, or in this case life year saved, is somewhat simplistic. As this is a ratio between costs and expected benefits, it is quite possible to have an intervention which appears highly cost effective, but also has substantial budgetary implications. This does not appear to be the case in the scenario mapped above, but it will always be important to also consider the budgetary impact of any public investment. However we have also made a crude estimate of the costs and benefits of investing in the Choose Life strategy; although the value of the intangible benefits of lives saved used in our analysis has had to make use of data related to road traffic accidents rather than death by suicide. One potential area for future work would be to elicit specific suicide related value for the intangible benefits of lives saved.

Bridging the gap - including economic evaluation in future analysis of suicide prevention programmes

We have illustrated the potential use of economic methods to provide some sense of the economic case for any programme or intervention. Any future economic evaluation of the Choose Life strategy, would almost certainly be one of the first (if not the first) evaluations worldwide undertaken of a national strategy. How might we facilitate the potential inclusion of economic evaluation in any future evaluation of suicide prevention strategies of similar programmes? We set out some steps below.

Identify and collect data on relevant health and non-health outcomes

Information from any previous review of effectiveness studies may identify some outcomes of importance to the evaluation (in this case suicides averted). There may also be additional outcomes that merit attention. In all prospective evaluations (not just economic evaluation) consultation should take place at an early stage to agree on health and non-health outcomes of importance to the analysis. The not inconsiderable challenge then is to ensure that it is feasible to collect data on these outcomes and possible to attribute them to the implementation of Choose Life.

Collecting data on the cost and uptake of components of a suicide prevention strategy

Currently there is only limited information on the uptake of different interventions at local level within the Choose Life programme. For some interventions it is important to have a basic understanding of how many and how frequently individuals use a service. Basic socio-demographic data are also important to help address questions of equity and the effective targeting of Choose Life. Is it in fact being delivered to those intended? The challenge is then how to collect and record such information in a non obtrusive fashion; for some activities it might for instance be easy to use an attendance register - for others this may actually dissuade participation. Many activities will not be done on a face to face basis and there will also be issues of confidentiality and anonymity.

Equally it is vital that more information is collected on the contributions of volunteers. Not only is this important for economic evaluation, it is also essential to building a case for the continued use of the partnership model used by Choose Life. Our case study analysis has indicated that volunteer contributions can be substantial; but they are almost certainly not being fully captured. A simple monitoring form might periodically ask those delivering projects to record the number of volunteers and estimate the aggregate number of hours they donate to an intervention. This might be backed up by some limited validation exercises, for instance observing the use of volunteer time in some project, or by asking volunteers to keep diaries etc. Practical approaches to quantify this significant input have been developed, e.g. (Gaskin & Dobson 1997).

Information on the use of resources provided in kind by host organisations, as well as additional monetary investment is required. As we have noted, there have been challenges in obtaining this information through the initial template that local CPPs were asked to complete. In the case of in-kind resource use and time given by paid staff, a variant of a data collection tool such as the Client Service Receipt Inventory, widely used in social care and in mental health evaluations, might be considered (Beecham & Knapp 1992).

Consider the type of economic evaluation that is most appropriate

If the focus of evaluation is narrowly on suicides averted then simple cost effectiveness analysis or perhaps CUA using quality or disability adjusted life years would be practical. If additional outcomes of importance are also identified (for instance reduction in non fatal suicide events) a cost consequences analysis ( CCA) - similar to CEA, but comparing costs with more than one outcome may be appropriate.

The case for cost benefit analysis

Any future evaluation of a complex suicide prevention strategy may well wish to consider using cost benefit analysis. Theoretically this is the most robust of the approaches to economic evaluation, although it can have a number of practical difficulties (Healey & Chisholm 1999).

Members of the public might, for instance, be presented with a range of information (taken from the effectiveness evaluation) on the impact that Choose Life has on the rate of suicide (and other outcomes of interest). Using a technique called 'contingent valuation', they may be asked to express their 'willingness to pay' for public investment in Choose Life. This use of this technique in relation to suicide and other injuries has, to a limited extent, been used in the United States; currently work is also underway in Ireland.

An alternative approach, increasingly used in health economics, may be 'conjoint analysis' e.g. (Sassi et al. 2005) whereby individuals are presented with a range of different scenarios which can include information on the effectiveness of Choose Life as well as the costs of delivering the programme and other issues of interest that may impact on the willingness of individuals to use Choose Life interventions. The ranking of these scenarios might then be used to estimate willingness to pay for an intervention. The major drawback of these two approaches however, is the need to interview what may be several thousand people, preferably on a face to face basis. For an evaluation of a national programme this may however be practical.

Putting any future economic evaluation into context

Our phase one evaluation has emphasised the importance of understanding the context in which interventions are delivered. This use of contextual information in economic evaluation has been rare, although a few economists have argued for the use of ecological or holistic economic evaluation methods (Jan et al. 2004). These would have to be tailored to suicide prevention but box A2 illustrates how this has been used alongside an economic evaluation of a maternal health intervention in Australia. In the case of Choose Life the wealth of information emerging from phase one of the evaluation could play an important role in describing this context. Further such context setting work would be helpful to any evaluation of effectiveness and cost effectiveness in phase two.

Conclusions

Many methodological papers and policy documents recognise the importance of economic evaluation as one element of the assessment of suicide prevention strategies. Our review of the literature suggests that despite this acknowledgement, there have been very few economic evaluations of national or area-based suicide prevention strategies. This is perhaps unsurprising given that there are few effectiveness evaluations of such strategies. Even when looking at specific individual interventions that have been suggested for suicide prevention such as primary care physician training, there appears to be little economic evaluation.

The human and economic costs of suicide are profound, suggesting that the economic case for investment into effective prevention strategies may be strong. We have illustrated here also how even in the absence of full information, economic methods can play a modest role in informing the policy making process. There do not appear to be any major conceptual reasons as to why a full economic evaluation could not be used; cost benefit analysis may well be the most appropriate approach.

While economic evaluation may be an integral component of future evaluations; there may also be some scope for retrospectively making some assessment of cost effectiveness of a strategy in a particular country or context, if robust information on effectiveness and resources required to deliver this programme are available from effectiveness reviews.

While our analysis would suggest that, even if only moderately effective, investment seems merited because of the many years of life saved, there will also be many other potential benefits gained. In addition to having a potential impact on the number of deliberate self-harm events, other benefits may include the promotion of positive mental well-being and thus improved physical health, as well as the acquisition of new skills by those participating in delivering interventions.

References

Beecham, J. and Knapp, M. (1992). Costing psychiatric interventions._ Measuring mental health needs. G. Thornicroft, C. R. Brewin and J. Wing, Eds. London: Gaskell: 163-83.

Bennett, K. J., Torrance, G. W., Boyle, M. H. and Guscott, R. (2000). Cost-utility analysis in depression: the McSad utility measure for depression health states. Psychiatric Services, 51(9): 1171-6.

Berk, M., Dodd, S. and Henry, M. (2006). The effect of macroeconomic variables on suicide. Psychol Med, 36(2): 181-9.

Byford, S., Harrington, R., Torgerson, D., et al. (1999). Cost-effectiveness analysis of a home-based social work intervention for children and adolescents who have deliberately poisoned themselves. Results of a randomised controlled trial. Br J Psychiatry, 174: 56-62.

Byford, S., McDaid, D. and Sefton, T. (2003). Because It's Worth It. A practical guide to conducting economic evaluations in the social welfare field. York: Joseph Rowntree Foundation, York Publications.

Chatterji, P., Caffray, C. M., Crowe, M., Freeman, C. and Jenson, P. (2004). Cost assessment of a school based mental health screening and treatment programme in New York City. Mental Health Services Research, 6(3): 155-66.

Chisholm, D., Salvador-Carulla, L. and Ayuso-Mateos, J. L. (2006). Quality of life measurement in the economic analysis of mental health care_ Quality of Life in Mental Disorders. H. Katschnig, H. Freeman and N. Sartorius, Eds. Chichester: John Wiley & Sons Ltd.

Clayton, D. and Barceló, A. (2000). The cost of suicide mortality in New Brunswick, 1996. Chronic Diseases in Canada, 20(2): 89-95.

de Castro, S., Newman, F., Mills, G. and Sari, N. (2004). Economic evaluation of suicide prevention programs for young adults in Florida. Business Review Cambridge, December: 14-20.

Drummond, M., O'Brien, B., Stoddart, G. and Torrance, G. (2005). Methods for the Economic Evaluation of Health Care Programmes. Oxford.: Oxford Medical Publications.

Gaskin, K. and Dobson, B. (1997). The economic equation of volunteering: a pilot study. Loughborough, Centre for Research in Social Policy, Loughborough University.

Hale, J., Cohen, D., Ludbrook, A., Phillips, C., Duffy, M. and Parry-Langdon, N. (2005). Moving from evaluation into economic evaluation: a health economics manual for programmes to improve health and well-being. Cardiff, National Assembly for Wales.

Hamermesh, D. and Soss, N. (1974). An economic theory of suicide. Journal of Political Economy, 82(1): 83-98.

Healey, A. and Chisholm, D. (1999). Willingness to pay as a measure of the benefits of mental health care . Journal of Mental Health Policy and Economics, 2: 55-58.

Jan, S., Conaty, S., Hecker, R., Bartlett, M., Delaney, S. and Capon, T. (2004). An holistic economic evaluation of an Aboriginal community-controlled midwifery programme in Western Sydney. J Health Serv Res Policy, 9(1): 14-21.

Kelly, M. P., McDaid, D., Ludbrooke, A. and Powell, J. (2005). Economic appraisal of public health interventions. Briefing Paper. London, Health Development Agency.

Lucey, S., Corcoran, P., Keeley, H. S., Brophy, J., Arensman, E. and Perry, I. J. (2005). Socio-economic change and suicide. A time series study from the Republic of Ireland. Crisis, 26(2): 90-94.

Marcotte, D. E. (2003). The Economics of Suicide, Revisited. Southern Economic Journal, 69(3): 628-43.

Matschinger, H. and Angermeyer, M. (2004). The public's preference concerning the allocation of financial resources to health care: results from a representative population survey in Germany. European Psychiatry, 19(8): 478-82.

McCulloch, D. (2003). Valuing Health in Practice. Priorities, QALYs and Choice. Aldershot: Ashgate.

Medoff, M. H. (1986). An evaluation of the effectiveness of suicide prevention centers. Journal of Behavioral Economics, 15: 43-55.

Ostamo, A. and Lonnqvist, J. (2001). Attempted suicide rates and trends during a period of severe economic recession in Helsinki 1989 - 1997. Social Psychiatry & Psychiatric Epidemiology, 36: 354-60.

Platt, S. and Hawton, K. (2000). Suicide behaviour and the labour market._ The international handbook of suicide and attempted suicide. K. Hawton and K. van Heeringen, Eds. Chichester: Wiley.

Rehkopf, D. and Buka, S. L. (2006). The association between suicide and the socio-economic characteristics of geographical areas: a systematic review. Psychological Medicine,36(2): 145-58.

Rodriguez-Andres, A. (2005). Income inequality, unemployment and suicide: a panel data analysis of 15 European countries. Applied Economics,37: 439-51.

Rutz, W., Carlsson, P., von Knorring, L. and Walinder, J. (1992). Cost-benefit analysis of an educational program for general practitioners by the Swedish Committee for the Prevention and Treatment of Depression. Acta Psychiatr Scand, 85(6): 457-64.

Sassi, F., McDaid, D. and Ricciardi, W. (2005). Conjoint analysis of preferences for cardiac risk assessment in primary care. Int J Technol Assess Health Care,21(2): 211-8.

US Preventive Services Task Force (2004). Screening for suicide risk, Agency for Healthcare Research and Quality.

US Public Health Service (1999). The Surgeon General's Call to Action to Prevent Suicide. Washington, DC, US Public Health Service.

Yaniv, G. (2001). Suicide Intention and Suicide Prevention: An Economic Perspective. Journal of Socio-Economics,30(5): 453-68.

Zaloshnja, E., Miller, T. R., Galbraith, M. S., et al. (2003). Reducing injuries among Native Americans: five cost-outcome analyses. Accid Anal Prev, 35(5): 631-9.

« Previous | Contents | Next »