Examining Pregnancy-associated Drug Overdose Mortality Using Enhanced National Death Certificate Data

The last decade has shown marked increases in both maternal deaths and drug-related overdose deaths in the United States. While the epidemiology of each of these epidemics have been explored independently the intersection of maternal and overdose deaths (i.e. pregnancyassociated overdose mortality) may present unique epidemiologic patterns that can inform public health programs. Effective responses to these trends requires accurate and timely surveillance to better inform policies and programs and target prevention efforts. Death certificates are the foundation of maternal and drug overdose mortality surveillance and can provide more timely information compared with other national data sources yet valuable information can be lost in the process of coding cause of death using ICD-10 codes. To address this research gap the proposed project aims will develop strategies to examine enhanced National Death Certificate data linked to literal text information (i.e. actual information written on the cause of death section of the death certificate) to improve our understanding of the epidemiology of pregnancy-associated overdose mortality and specific drug involvement. This enhanced data source provides detailed information beyond ICD-10 code descriptions to generate more specific accurate descriptions of cause of death and circumstances related to the death. The proposed aims align with strategic research issues to eliminate health barriers and disparities Healthy People 2020 objectives to reduce drug-induced deaths and two key HRSA priority areas - maternal mortality and opioid use disorder. Additionally national vital records data will ensure more generalizable findings to direct key resources. Our team will apply methodologies to examine drug mentions based on literal text of all overdose deaths in 2016 and 2017 among female decedents between 10-54 years of age who were residents of the U.S. and were pregnant within one year of the time of death. We will estimate specific drug mentions from literal text and apply estimation procedures to account for variation in under-reported drug mentions. The type and frequency of drug mentions will be compared by key sociodemographic characteristics and timing of the death in relation to pregnancy (during or postpartum). Complete information from adjusted measures will be used to examine pregnancy-associated drug overdose mortality rates overall and by sociodemographic and geographic (e.g. rural/urban) subgroups. The denominator live births will be obtained from 2016/17 Detailed Natality files. Adjusted rate ratios will be estimated by grouped Poisson regression models.