Health Trade-offs from Policies to Alter Fish Consumption

Abstract

Fish consumption is associated with several health effects. Episodes of large-scale methyl mercury poisoning caused neurodevelopmental harms, particularly to those with prenatal exposure. Because methyl mercury is found to varying degrees in most fish, there has been concern about the consequences of smaller exposures. Conversely, fish contain polyunsaturated fatty acids (PUFAs) that have been associated with salutary effects, most notably on the cardiovascular system. These and other potential health effects yield a classic risk‐risk trade-off. Decision makers must consider how policy and educational initiatives affect fish consumption patterns and resulting risk trade-offs. Although there is a substantial body of literature on the health effects of fish, methyl mercury, and PUFAs, there are also significant gaps in our understanding. Nonetheless, policy decisions are made on a regular basis and they must be informed by the best available scientific information, supplemented where necessary by expert judgment. Decision analysis is an important tool that can quantitatively synthesize disparate information and assess the impact of uncertainty on our conclusions. The Harvard Center for Risk Analysis undertook a project to create a decision model to evaluate policies that influence population fish consumption. It convened an expert panel to help guide the process and advise on evaluation of the scientific literature. The overall approach was to identify the important health effects, assess the dose‐response relationships between fish consumption (or a constituent) and each health outcome, and then to synthesize these dose‐response relationships into an overall model of health effects. To accomplish this, all the health effects had to be converted to a standard unit, quality-adjusted life years in this case. Based on the overall model, we assessed the potential consequences of educational and other policy interventions. The panel identified the health effects that would have the greatest influence on the overall health impact of fish consumption—stroke, heart disease, and prenatal cognitive development. Four of the papers that follow present the estimation of dose‐response relationships for fish consumption and stroke, fish consumption and coronary heart disease, maternal PUFA intake and prenatal cognitive development, and methyl mercury exposure and prenatal cognitive development. 1‐4 Each of these papers is based on a literature synthesis performed by us or a high-quality review performed by others. We relied on the panel to ensure the overall coherence of the analysis and to provide required judgments, such as which studies to include or exclude, how those studies should be weighted, the suitability of health measures, and the assignment of values to parameters. Within each article, we indicate how those judgments were made. The project’s final paper, 5 also in this issue, draws on the four dose‐ response papers to estimate the aggregate impact of hypothetical shifts in population fish consumption that might result from government advisories and education campaigns. Although it is impossible to overcome all limitations of analyses like those presented here, we believe it is important to inform decisions with the best available information. Nonetheless, the limitations must be addressed through transparency so that the models and parameter estimates can be criticized. In particular, sensitivity analyses must be conducted to identify those assumptions whose uncertainty most influences the findings of the assessment. As additional information becomes available, these models merit refinement. We believe that decision-analytic approaches such as this one are an important way to inform policy decisions.