New Methods in Drug Surveillance Show Promise in Reducing Suicide

Mental health, suicide, and how to address these issues are increasingly salient parts of the national health discussion in the United States. For well over a decade, the suicide rate has increased and is now the 10th leading cause of death in the United States. The majority of suicides are associated with a psychiatric disorder. The use of psychiatric medication is widespread in dealing with many of these disorders, but the best practices remain hotly debated.

Dr. Robert Gibbons and his colleagues studied a new medication surveillance technology, iDEAS—High Dimensional Empirical Bayes Screening—to identify both potential risks and benefits across hundreds of medications simultaneously. Instead of using the traditional method of self-reported data on side effects, this study used commercial claims data. The commercial claims data that were collected were “suicidal events” which include suicide attempts and intention to self-harm that led to a medical claim. In total, the data spanned over 150 million people between 2003 and 2014 and assessed the effect of 922 prescription drugs on the risk of suicide events.

The authors point out that the traditional drug screening process falls short in several important ways. First, it focuses on harmful effects in isolation from one another. iDEAS aims to evaluate the comprehensive risks as well as the benefits of all medications. Second, it is unknown which population is at risk of harm from the medications, “so the rate of adverse drug reactions is inestimable” (Gibbons et al. 2019, 2). Third, there is little information on the history of diagnoses and treatments for patients when making assessments of the medication. Fourth, without demographic information, it is difficult to generalize the results. Fifth, it is difficult to make causal inferences. For example, both suicide and the use of antidepressants are associated with major depressive disorder. This implies there is also an association between antidepressants and suicide, but that association could be explained by the underlying illness and not the antidepressants. This emphasizes the importance of improving the data and methods used to guide policies related to drug surveillance.

After implementing iDEAS, the authors found that it produced significant results that highlighted a drug’s effect on increasing or decreasing the risk of suicide. Of the 922 prescription drugs, the study found 10 drugs with increased rates of suicide events and 44 drugs with decreased rates. Not all of those 44 drugs are intended to treat psychiatric disorders, but still have positive effects on reducing the risk of suicide events.

Paradoxically, drugs approved to treat mental health disorders that were accompanied by a warning about an increased risk of suicidal ideation or behavior were discovered to be overrepresented in the study’s list of drugs that significantly reduce the risk of suicide. The authors claim that this highlights the issue with the current methodology’s focus on potential harms to evaluate the overall risk of medications. Being able to weigh the risk of harm with the probability of benefits among all possible therapies is important in optimizing the use of prescription drugs from the perspectives of the physicians and the patients.

Updating the FDA’s methodology for assigning risk to medications has the potential to reduce the risk of suicide. Providing more detailed information about psychiatric drugs to doctors would allow them to better tailor their treatment plans for their patients. iDEAS gives an estimate of both potential harms and benefits as it analyzes across multiple drugs and drug types. This form of comparative analysis has both statistically and clinically significant results. While promising, the authors note that these results need to be confirmed by a formal randomized controlled trial or another study design that can provide causal inferences. If confirmed, iDEAS may play a critical role in improving the FDA’s evaluation methodology and ongoing drug surveillance of adverse effects.

---

Gibbons, Robert, Kwan Hur, Jill Lavigne, Jiebiao Wang, and J. John Mann. 2019. “Medications and Suicide: High Dimensional Empirical Bayes Screening (iDEAS).” Harvard Data Science Review 1, no. 2: 1-32. https://doi.org/10.1162/99608f92.6fdaa9de.