Research on Youth Suicide and Sexual Orientation is Impacted by High Rates of Missingness in National Surveillance Systems

Abstract

Objective The sexual orientation of youth who die by suicide in the United States is usually unknown. This study assessed how observed patterns of unknown sexual orientation are likely to affect research findings. Methods We analyzed the National Violent Death Reporting System (NVDRS) Restricted Access Dataset to assess whether sexual orientation among youth suicide decedents is disproportionately known for different demographics. We then assessed the degree to which estimated sexual minority rates would be affected if researchers were to assume either (a) that sexual orientation data is missing completely at random, or (b) that orientation information is missing at random after accounting for observed demographic patterns. Results <10% of the sample had known sexual orientation. Sexual orientation was more frequently known for females, white people, and older people, and missingness varied by geography. The choice between modeling the data as missing completely at random versus at random conditional upon demographics had a > 2-fold impact on estimated sexual minority rates among youth suicide decedents. Conclusion Research on sexual orientation and youth suicide is strongly impacted by how researchers account (or do not account) for missingness.