Mental health disorders are very common. Last year, 120 million Americans were treated with a psychoactive prescription drug. The use of psychoactive prescription drugs is increasing despite their untoward side effects, some of which are very serious and can result in hospitalization and death. These side effects may be substan- tially amplified by concomitantly used medications, i.e., a harmful drug interaction (DI). This is concerning since the use of multiple drugs is common and increasing among persons with mental health disorders, including in older adults with multiple chronic conditions. In fact, harmful DIs are projected to result in nearly 5 million hospitalizations and 150,000 premature deaths of older adults in the next decade. Unfortunately, nearly all existing evidence on DIs with psychoactive prescription drugs comes from case reports or pharmacokinetic studies examining drug concentration changes rather than clinical endpoints. Thus, DI compendia and clinical decision support software surely fail to warn against unrecognized harmful psychoactive DIs, and falsely warn against safe combinations that prevent appropriate use and contribute to alert fatigue. Of the very few studies of effects of psychoactive DIs on clinical endpoints, most examined injurious falls. None have systematically examined venous thromboembolism (VTE) or serious bleeding, despite well documented associations between VTE and antipsychotics and benzodiazepines, and between serious bleeding and antidepressants. Biologically plausible mechanisms support both VTE and bleeding risks, including antipsychotics’ dopamine antagonism increasing platelet aggregation and antidepressants’ serotonin reuptake inhibition increasing gastric acid secretion, inhibiting platelet activity, and increasing bleeding time. Effects of psychoactive DIs on VTE and bleeding have been hypothesized by many, but remain unstudied. Given high rates of VTE and serious bleeding in older adults, it is imperative to identify drugs that increase patients’ risks of these serious clinical endpoints when co-administered with commonly used psychoactive prescription drugs. We propose a triphasic strategy to: a) generate signals of psychoactive DIs resulting in VTE and/or serious bleeding, via screening of longitudinal healthcare data; b) prioritize signals via multidisciplinary expert panel focus groups; and then c) test, in an independent population, these prioritized signals in a series of rigorous cohort studies built on a causal inference framework designed to emulate target trials. These complementary phases ensure achievement of our broad objective to generate clinically-actionable psychoactive DI evidence to prevent iatrogenic harm in patients. We will apply each of the three phases to each of the following specific aims: 1) generate signals for antipsychotics (by class and for individual agents) resulting in emergency department treatment ± hospitalization for acute VTE (phase a), prioritize DI ...