Key emerging themes for assessing the cost-effectiveness of reporting incidental findings

To the Editor: We congratulate Bennette et al. for an innovative first step to addressing a challenging issue—how to assess the cost-effectiveness of reporting incidental findings (IFs) discovered with sequencing technologies—as described in their article, “The Cost-Effectiveness of Returning Incidental Findings From Next-Generation Genomic Sequencing.”¹ At the University of California at San Francisco Center for Translational and Policy Research on Personalized Medicine, we are conducting related analyses that further inform these issues.² We would like to highlight key emerging themes and suggestions for future work and to discuss the importance of some assumptions made by Bennette et al. that could greatly impact the findings of cost-effectiveness analyses.

Of particular importance for future work is the need to examine the likely cost-effectiveness in real-world settings. Bennette et al.¹ assumed that individuals would remain at risk but not be detected through any other means during their lifetime (other than for familial hypercholesterolemia). However, particularly for the two most prevalent conditions they examined (hereditary breast and ovarian cancer; Lynch syndrome), many individuals with the conditions will be identified even if there is no sequencing. By assigning all benefits to the detection of IFs, the cost-effectiveness of reporting IFs in real-world settings will seem better. We thus suggest that future analyses consider including a background rate of detection rather than using “nothing” as a comparator, which would enable the findings to be more comparable with those of other analyses that follow the standard approach of comparing an intervention to current practice.³

Other simplifying assumptions that Bennette et al.¹ acknowledge could affect their results. First, aggregating results of different models is a reasonable first step, but it is unclear how sensitive the results are to the assumptions of individual models and whether it is reasonable to aggregate possibly heterogeneous findings—with different populations and modeling approaches—into an overall cost-effectiveness ratio. Many previous studies (e.g., Vegter et al.⁴) have noted the heterogeneity found across cost-effectiveness analyses of genetic testing. In future research, it would be helpful to develop a transparent means of aggregating results so that they can be readily replicated. Second, future analyses could take into account interactive effects, namely, the differences in life-expectancy from finding one result when evaluating the potential effects of another result. The likelihood of finding more than one IF in a given person is very small in the current analysis but will increase as more returnable IFs are identified in the future. Third, in real clinical practice, it is possible that unproven and potentially costly management strategies could be used in a fraction of individuals receiving a given IF result. Not accounting for this may miss an important determinant of downstream clinical effectiveness and cost.

In sum, the approach of Bennette et al.¹ provides an important initial approach for analyses that can continue to refine approaches to defining and measuring the value of new genomic testing technologies that return multiple results. It should be noted that the results to date suggest that reporting IFs may be cost-effective in certain scenarios but are not generally cost-saving as some have claimed or hoped, and that many simplifying assumptions may overestimate the cost-effectiveness of reporting IFs. Future research can continue to refine the modeling approaches and estimates used.

Disclosure

The authors declare no conflict of interest.