Racial Disparities in Suicide Data Collection: Why Data Quality is Key for Suicide Prevention

As a public health researcher who studies mental health, I lead studies that seek to identify the factors that impact a person’s risk of suicide over the course of their life. I also aim to further understand how these risk factors change over time, vary by location, and affect people differently based on individual characteristics such as race and ethnicity, income, and education. To investigate these questions, my research team and I highly value resources that collect data on suicide at the population-level, such as national mortality registries or large cohort studies (i.e., those that follow individuals and collect information from them over a period of time). These types of data give us the ability to comprehensively describe risk factors for suicide as well as identify patterns of risk at the community, state, or national levels. But can we always depend on how this data is collected and reported? For instance, is suicide data reported differently for different groups of people? With these questions in mind, I am always looking for ways to support continual improvement of these data resources so that we can most effectively and equitably study suicide.

In suicide prevention research, a key resource that we use to obtain this type of comprehensive data is the National Violent Death Reporting Systems (NVDRS). In brief, the NVDRS is a nation-wide system that tracks the occurrence of violent deaths across the country, including suicides. It was launched in 2002 by the Centers for Disease Control (CDC) and, as of 2018, reports mortality data for all 50 states and US territories. The NVDRS collects valuable data on the ‘who, when, where, how, and why’ of suicide deaths in the U.S. by gathering detailed information about anything that could have been related to the incident, including the person’s history of mental health conditions or treatment, health problems, or stressors related to finances, health, relationships, and work. Therefore, the NVDRS is an important tool for studying suicide risk and using research to target prevention. It has been previously used by researchers to identify unique trends in suicide mortality, including increased risk among military personnel, victims of intimate partner violence, and older adults that are transitioning to long-term care.

The NVDRS also contains an additional data resource that can be utilized to study suicide: suicide case narratives. For each case in the NVDRS, there is a narrative that records additional details from the suicide event in a brief, textual format. For instance, it might describe the victim’s history of depression, or their relationship with their partner, in more depth than the other quantitative variables in the system. Therefore, these narratives are a valuable source of information for researchers who want to study specific risk factors for suicide, such as a history of mental health condition or intimate partner violence. As such, these narratives are a unique feature of the dataset that have tremendous potential as a resource for identifying new avenues for suicide prevention.

However, like any data source, these texts may be subject to measurement error, biases, or other issues that negatively impact their utility in suicide prevention research. Notably, the NVDRS and other mortality registries have been found to have inconsistencies in how suicide data is reported for certain groups of people, such as underreporting for people of color. 

 With this in mind, my team and I wanted to see if these patterns were present in the narrative data by examining if the narratives tend to change based on the individual characteristics of the suicide victim (e.g., age, gender, race or ethnicity). Mainly, we were curious whether there are patterns that exist in the NVDRS narrative data that need to be considered when using these texts for research. In a recent study, we reviewed a large sample of NVDRS narratives (almost 24,000!) and investigated these questions with the aim to foster improvement of the dataset.

 In our investigation, we found that there are significant inequalities in the narratives based on the individual characteristics of the suicide victim. We observed that being older, less educated, and a person of color was associated with having a missing narrative. Even when a narrative wasn’t missing, we found that people with these traits had significantly shorter narratives than their younger, more educated, and white counterparts. Therefore, our findings suggest that there are systematic biases in these narratives that may limit what we can learn about suicide prevention for older adults, racial and ethnic minorities, and those with less education.

 All this being said, I believe that the NVDRS – including its case narratives – is an invaluable tool for researchers, advocates, and policymakers to further suicide prevention. Although our study has identified a gap in the quality of the narrative data, this is something that can be addressed through collaboration between the NVDRS administrators and researchers. For example, the NVDRS could increase the training of abstractors (i.e., NVDRS staff that write the narratives) or create additional technical supports that would decrease the amount of variability between narratives. Overall, we should strive for continual improvement of the dataset by decreasing the impact of bias, thereby ensuring that this resource can live up to its potential.

 As we reflect on the importance of suicide prevention, I want to emphasize that good data leads to good advocacy and improving data quality and equitability will strengthen our suicide prevention efforts in both research and advocacy. By doing so, we can ensure that our programs and policies are evidence-based, and that that evidence is representative of all groups and communities.