Harmonizing Multiple Data Sources And Psychological Autopsy To Characterize Suicides Among Opioid-Related Deaths

点击翻译按钮获取中文摘要

Project Summary/ Abstract This Mentored Patient-Oriented Research Career Development Award is designed to provide the applicant with the advanced training necessary to establish an independent program of research in the epidemiology of overdose mortality and suicide. A comprehensive training program is proposed, combining formal coursework, mentoring, and hands-on training experiences designed to develop expertise in data linkage/ harmonization, latent class modeling (LCA), qualitative psychological autopsy (PA) and mixed methods research. As overdose deaths increase, they continue to be treated as accidents resulting from changes in opioid use. However, epidemiologic research suggests that many of these deaths are likely suicides, This has important implications for appropriately targeting interventions. We will measure the magnitude of this misclassification of manner of death (MOD; intentionality) and identify factors that may guide medical examiners to more accurately classify suicide decedents. We hypothesize that approximately one third of opioid related deaths of undetermined manner are truly suicides, and that LCA can distinguish subgroups of decedents with greater likelihood of suicidal intent. We will use PA in a subset of previously undetermined intent decedents to test predictive ability of empirically-derived classes and characterize the diverse paths to overdose. We propose to analyze all opioid overdose deaths in Maryland from 2006-2019 (n=13,861) using demographic, social, and clinical data which we will link from the Maryland Suicide Data Warehouse to mortality data from the Office of the Chief Medical Examiner (OCME). First, taking one third of this sample, we will compare cases classified as suicidal (n=115) from accidental (n=756) and undetermined intent (n=3,748) using a three-way multinomial logistic regression. Next, variables found to be most salient comparators will be used in LCA of the remaining cases, agnostic of OCME MOD class (n=9,286). Comparing the empirically derived classes with the OCME designations, we will assess the proportion of designated suicides and accidents in each class. Finally, from each of these latent classes, we will select 40 decedents designated by OCME as ‘undetermined manner’ to be further examined by multiple collateral interview PA, to corroborate these latent classes. By generalizing findings from LCA and PA, more accurate suicide rate estimates can be made. Findings would impact future MOD designation and potentially, how prevention interventions target accidental overdoses and suicides. Training and mentorship plans will leave the candidate well positioned to become an independent physician- epidemiologist, able to utilize both qualitative and quantitative methods for the validation of large linked data sets describing the interrelated suicide and overdose crises. His long term career goals include the elucidation of mechanisms of self injury mortality, the use of mixed methods for the generation and investigation of novel hypotheses regarding pathways to suicide, and the ability to translate these findings into suicide prevention.