As a companion to the discussions of various approaches to calculating costs, speakers offered cautions and considerations in utilizing existing methodologies. Most of the challenges are not specific to cost calculation, but plague other aspects of violence prevention research and implementation as well. Some challenges were evident in health economics and the costing of other public health problems. These challenges in both arenas increased the difficulty in enumerating costs and placing appropriate value on various preventive interventions.
The major challenge, reiterated by many speakers, was the difficulty in placing a value on nontangible, social costs. In particular, pain and suffering were two outcomes speakers cited as being unquantifiable. Some speakers suggested these costs could be indirectly measured by assessing psychological costs, but no accurate and direct assessment of them could currently be made. Speakers also felt that this was indicative of other quality-of-life issues. Thus, in placing a dollar value on some costs but not others, the true impact of violence is not being captured.
The field of violence prevention suffers from a few issues that add difficulty to estimating costs. Speaker Phaedra Corso of the University of Georgia emphasized that generating the evidence base for violence prevention is hampered by imprecise definitions and underreporting of events. Lacking clear definitions of what constitutes violence and violent events makes determining both prevalence and incidence difficult, both of which are essential to assessing true cost. Also, underreporting of events obscures true incidence and prevalence. Dr. Corso and other speakers proposed that relying on medical records and claims requires accurate coding of causes of injury, which does not always occur because universal screening is not in
place for violence. Speaker Deborah Prothrow-Stith of Spencer Stuart also highlighted the importance of definitions and their cultural context; what constitutes violence in one setting might not translate to another setting and therefore might not be captured in data analysis. Speaker Mindy Fullilove of Columbia University noted that structural violence, or violence that is institutionalized but not criminalized, is difficult to define and measure, correlates with other inequities, and is often missed when considering the cost of overt violence.
Dr. Corso, speaker Philip Cook from Duke University, and speaker and Forum member David Hemenway from Harvard University also emphasized the difficulty and importance of determining attributable risk, particularly for longer-term outcomes. Dr. Corso observed the lack of a good risk model, and Drs. Cook and Hemenway discussed the difficulty of attributing violence to specific risk factors. This provides a challenge in assessing how costs are allocated to risk factors or types of violence, making it difficult to assess how well preventive interventions are truly working.
Dr. Corso also identified general issues with the availability and generalizability of data; typically, a disparity in both the comprehensiveness and the quality of data from high-income countries and low-income countries exists. Thus, the assessment of costs in one context may not be applicable anywhere else. Dr. Waters emphasized this issue, citing that values placed on specific components do not take into account differences in cost-of-living standards across different countries. For example, productivity measured by loss of wages is markedly different in high-income and low- or middle-income countries, where average wages can differ by orders of magnitude. This human capital approach also takes into account life expectancy and age, which have other health implications and result in different values being placed on both employment and human life based on a country’s economic development status. Speaker and Forum member Michael Phillips of the Shanghai Jiao Tong University School of Medicine highlighted another downside to the human capital approach—specific types of violence pose different relative financial burdens in relation to others and thus receive lower priority than others, despite being a potentially greater social burden.
Dr. Hemenway added to the previous comments of speakers on the availability and generalizability of data by pointing out that not only is violence underreported, but also active data collection is often limited to areas of higher income or socioeconomic status. Receiving data from diverse populations is difficult when less effort is made to collect such data. Forum member Evelyn Tomaszewski made a similar statement in commenting on undercurrents of violence and effects of violence in society, such as undiagnosed psychological illness or trauma, and the impossibility of enumerating such costs. In addition, the data are often inaccurate—Dr. Hemenway stated that a recent review of the National Violent Death Reporting System
(NVDRS) indicates that reporting of mortality due to firearms has overestimated the age of victims; such victims are actually younger than previously estimated. This can have a profound effect in calculating long-term costs (especially productivity and social losses).
A few issues on the actual accounting of costs were also raised. Dr. Corso asked how pain and suffering are quantified. Ms. Tomaszewski suggested that including community costs, not just aggregated individual costs, is important as well. Dr. Waters pointed out that aggregating numbers can often yield huge ranges, because each individual cost is often displayed as a range. He also noted, and was echoed by speaker and Forum member Rodrigo Guerrero of Cali, Colombia, that these large ranges are not always helpful to policy makers. Likewise, traditional economic approaches to measuring costs can result in confusion, such as the determination of the discount rate—the deduction applied to the future value of money so as to make comparisons to current value. The discount rate can vary by region, year, time line of projection, and economic model, yielding large intervals in value. The mathematical models often used for calculating costs rely heavily on theoretical assumptions that might not always be valid—for example, models incorporating counterfactual worlds in which violence does not exist can never be assessed for accuracy.
Speakers also questioned the presentation of cost data. Often such numbers are presented in terms of dollars lost or as a percentage of the gross domestic product (GDP). The first has the disadvantage of providing less meaningful comparison across countries, and the second is often confusing to those outside of economics. The importance of context-specific denominators for such numbers is essential in showing true impact but can hamper generalizability and comparison.
Both Ms. Tomaszewski and Dr. Prothrow-Stith commented on measuring externalities such as historical trauma and discrimination. Dr. Prothrow-Stith also questioned the possibility of measuring fear and its effect on behavior, which could have enormous implications for longer-term outcomes such as chronic health effects and future perpetration of violence. Dr. Betancourt mentioned the impact of collective violence on non-state actors and infrastructure, causing widespread damage that often takes years to address. Such impact can serve as a risk factor for future violence, again bringing forth the question of measuring the cost of future violence.
Finally, some speakers questioned the negative costs of some measures of prevention that are deemed effective. In particular, Dr. Phillips and Dr. Hemenway pointed out that access to dangerous weapons is a risk factor that can be addressed by legislation, but has implications of restricting freedom in the U.S. political climate. If costs can be calculated based on what society is willing to pay to prevent such violence, certain preventive interventions will always be deemed as having too high a cost.