• By Dartington SRU
  • Posted on Friday 21st October, 2011

Speaking the language of policy

Effective interventions can prevent or reduce violence against women, particularly intimate partner violence (IPV), and they can also improve the health of survivors, research shows. But when it comes to persuading policymakers to invest, is evidence on the effectiveness of interventions enough – or do we need evidence of cost effectiveness, too? In a recent study, several health economists from Australia and the United Kingdom argue that economic evaluation – that is, evidence of cost effectiveness – must be a critical component of IPV prevention and intervention research, if researchers want their work to speak the language of policy.In some countries, a first step toward showing decision-makers the policy relevance of IPV interventions has already been made, in that an estimate of the total cost of IPV to society is already available. IPV results in much mental, physical, and social suffering, and it comes with financial costs to the wider economy, through the health, justice, child protection, and welfare services, and through victims’ reduced ability to engage in the economy themselves as workers and consumers. But to make the policy argument clear, the authors say, researchers need to take the next step. They need to assess the cost effectiveness of an intervention by assessing the change in outcomes compared to the amount of inputs. The economists, led by Lisa Gold of Deakin University in Australia, also point out the strengths and weaknesses of the two main approaches to economic evaluation: evaluation alongside randomized controlled trials (RCTs), and economic modeling. Considering that the strongest research design for establishing causal links between an intervention and an outcome is an RCT, economic evaluation alongside an RCT provides the strongest evidence on cost effectiveness and value for money. Cost effectiveness and cost-benefit analysisAn estimate of cost effectiveness weighs costs against outcomes. In the case of IPV interventions, Gold and colleagues explain, this means assessing “the resources used by, and affected by, the intervention. Any change in outcomes due to the intervention can then be assessed relative to any increased investment required, to estimate not only effectiveness (e.g., change in mental health due to the intervention) but also cost effectiveness (e.g., additional cost per point improvement in mental health scale).”A cost-benefit analysis (CBA) is one way to evaluate cost effectiveness. Most health economic evaluations present the costs in monetary terms and the benefits in medical terms – such as a $700 cost to bring about a two-point change in the mental health scale. By contrast, CBA measures both the costs and the benefits in monetary terms.Of course, translating health outcomes into monetary terms is not straightforward. What is the financial value of better mental health or less fear of violence? However, health economists have developed a raft of economic valuation techniques to estimate the “monetary value of an intervention to relevant stakeholders (women, families, and/or the general community) and therefore to directly compare the additional costs of an intervention to the monetary value of the additional benefits gained.” The authors point out that although economic valuation techniques have been used extensively in women’s health, this has not been the case in evaluations of IPV interventions. Resource costs and savingsResources required to provide an intervention in the short term – such as health professionals’ time, office space, and equipment – tend to be fairly easy to measure. However, predicting the impact of the intervention on future resource use is more difficult. For example, “women’s use of health services may go up in the short term if an intervention successfully engages women in the service system, but fall in the long term if health and social outcomes improve, reducing women’s need for services in the future,” the authors say. Most trials, however, have only short-term follow-ups, so long-term benefits often have to be predicted with economic modeling techniques rather than with direct data. What to measure? So if IPV researchers commit to economic evaluation, what should they measure? According to Gold and her team, although it is important to measure IPV-specific outcomes such as Composite Abuse Scale (CAS) scores, such measures are often not very useful to policymakers who want to compare the cost effectiveness of spending limited resources in one public health area or another. For instance, which is more valuable: an IPV intervention, measured in cost per point change on CAS; or a smoking-reduction intervention, measured in cost per quitter? To make it easier for policymakers to take these decisions, generic quality of life measures can be included in economic evaluation. For example, much work has been done recently to develop measure of QALY – Quality Adjusted Life Years – which aim to make direct comparisons possible even when the health interventions are very different. RCTs or economic modeling?Because randomized controlled trials (RCTs) are relatively expensive to conduct, they are usually designed to be just large enough to detect differences in outcome measures such as mental health – but this also means that the sample size is often too small to detect a policy-relevant difference in cost. The researchers suggest that economic modeling can be used in economic evaluation when no direct long-term data, such as those from follow-ups of an RCT, are available. One such type of economic model is a “Markov model,” which can be used to represent states in a health condition that evolves over time. In the context of an IPV intervention, a Markov model could use a set of states such as “abuse identified” or “no abuse.” Then, using a wide range of information including data from studies and expert opinions, the model is constructed: the probability that a woman will transition from one state to another is assigned, and estimates of resource use and health outcomes are attached to each state and transition. Finally, total costs and outcomes for interventions are estimated. Economic modeling sounds good, but it has a major drawback: the results Markov models produce are inherently uncertain, because every step in the construction of the model involves introducing a new set of assumptions. However, economic modeling may be the best option – or the only option – when real-world data are not available. If IPV researchers do want to include economic evaluation in their study design, Gold and colleagues advise them to consider two questions. First, will the outcome measures make sense to policymakers? Second, will researchers be able to make reasonable estimates of all the resources invested in the program, and all the resources that may be affected by the program? If the answer to both question is yes, then evaluating IPV interventions in economic terms may help researchers to speak the language of policy.ReferenceGold, L., Norman, R., Devine, A., Feder, G., Taft, A. J., & Hegarty, K. L. (2011). Cost-effectiveness of health care interventions to address intimate partner violence: What do we know and what else should we look for?. Violence against Women, 17, 3, 389-403.

Back to Archives