A few words about this work
I shared results from some of my work as a CDC PHIFP Fellow in the 2018 Public Health Informatics conference in Atlanta, Georgia, USA. I presented an assessment of novel non-hospital-based data as complementary sources of information for surveillance of prescription and illicit opioid adverse events in Denver. You can download the presentation slides here: JTPrieto PHI Conference 2018
Abstract
Date: August 22, 2018
Authors: José Tomás Prieto (Public Health Informatics Fellowship, Division of Scientific Education and Professional Development, Centers for Disease Control and Prevention [CDC], first author, presenter, host site: Denver Public Health); Dean McEwen (Denver Public Health, presenter); Kenneth Scott (Denver Public Health); David Edwards (Denver Health Paramedics); Judith Shlay (Denver Public Health); Mark Messonnier (Division of Scientific Education and Professional Development, CDC); Sridhar R. Papagari Sangareddy (Division of Scientific Education and Professional Development, CDC); Arthur Davidson (Denver Public Health)
Background: Official statistics describing the U.S. opioid epidemic are typically based on deaths involving prescription and illicit opioid poisonings. Estimates originate from traditional data sources, including death certificates and emergency department and hospital discharge databases, but are often significantly delayed for coding before aggregation and dissemination by health departments. Although alarming, overdose death rates only hint at a more complex and nuanced problem, because they omit adverse events that include nonfatal overdoses treated outside the hospital by emergency medical services or community members.
Objectives: To evaluate novel, nonhospital-based data as complementary sources of information for timely surveillance of prescription and illicit opioid adverse events in Denver.
Methods: We used two data sources as follows: the Rocky Mountain Poison and Drug Center (RMPDC) call center log for Denver, and trip reports in which naloxone was documented as administered from the Denver Health Paramedics Division which is the primary provider of emergency medical services in Denver. We applied regular expression matching to data to identify prescription and illicit opioid adverse events occurring out of hospital in Denver during July 2013–February 2017. Pearson’s coefficient (r) assessed the correlation between monthly number of cases of opioid adverse events identified in the novel datasets and monthly number of opioid-related overdoses reported at Denver healthcare facilities, available through the Center for Disease Control and Prevention’s National Syndromic Surveillance Program (NSSP).
Results: During the study period, NSSP reported 1,170 opioid-related overdoses in Denver, 496 of which (42.4%) involved heroin. Paramedics recorded 2,946 encounters involving naloxone administration, 255 (8.7%) of which explicitly mentioned heroin in the chief complaint. In total, 296 encounters (10.0%) did not result in patient transport to a healthcare facility; 276 of these encounters (93.2%) involved a fatality at the scene. RMPDC’s call center log registered 568 calls related to opioid poisonings in Denver, 43 (7.8%) of which mentioned heroin. In 159 calls (28.0%), persons contacting the call center were treated outside of a healthcare facility. We found a weak correlation (0.2 < r < 0.4) between monthly number of paramedics interventions involving naloxone administrations and monthly number of opioid-related overdoses in NSSP. We found a negative correlation (r = -0.2) between monthly number of calls to RMPDC involving opioids and monthly number of opioid-related overdoses reported in NSSP.
Discussion: Results indicate that events captured by RMPDC or paramedics were not necessarily associated with those captured by NSSP, and that RMPDC and paramedics combined captured more opioid adverse events than NSSP in Denver. The combination of novel and near-real-time data sources, such as poison center calls and paramedic trips reports, might be complementary tools for monitoring the true extent of the problem, provide timely data for surveillance of the epidemic, and lead to more accurate epidemic burden estimates. It might also help produce early and rapid interventions by increasing the situational awareness of public health departments. Our work is an ongoing public health informatics attempt to link Denver’s unconnected data systems and provide actionable surveillance information to improve population health.