Opioid Overdose Surveillance

Despite progress in reducing the prevalence of deaths attributable to prescription opioid use, drug overdose continues to heavily affect our nation. In 2019, nearly 71 000 drug overdose deaths occurred in the United States, ¹ almost 71% of which involved an opioid. ² For every drug overdose death, many more nonfatal overdoses occur, which has a substantial emotional and economic impact. Nearly 1 million (n = 967 615) nonfatal drug overdoses were treated in emergency departments (EDs) in 2017, 32% (n = 305 623) of which were known to be opioid related. ³

Since 2015, the Centers for Disease Control and Prevention (CDC) has awarded millions of dollars in funding to support overdose prevention (eg, through the Prevention for States program) and surveillance (eg, through the Enhanced State Opioid Overdose Surveillance [ESOOS] program) in health departments. ^4,5 The ESOOS program was established to provide more timely and comprehensive data on nonfatal and fatal opioid overdoses than available through existing data sources. Twelve states and the District of Columbia were initially funded in 2016, and funding was expanded to 32 states and the District of Columbia in 2017. Participating states shared data with CDC quarterly on nonfatal overdoses and biannually on fatal overdoses, providing a more complete picture of the drug overdose landscape in the United States than that available through vital statistics data alone.

During the 3 years of the ESOOS program (2016-2018), both CDC and funded health departments learned important lessons. This supplement of Public Health Reports contains 11 articles that describe innovative overdose surveillance activities funded by the ESOOS program, including overdose case definition development and validation, data linkage, improvement in toxicology testing and enhanced biosurveillance, and use of data to inform localized response. The expertise developed and knowledge gleaned from both the nonfatal and fatal overdose surveillance activities in the ESOOS program and described in these articles have the potential to advance overdose epidemiology in other jurisdictions. These articles highlight the importance of timely, comprehensive local data to inform real-time public health response to address the opioid overdose epidemic.

Overdose Surveillance Data

States used ED, syndromic, hospital discharge, and emergency medical services (EMS) data to report nonfatal opioid overdoses to the ESOOS program. Hospital discharge data are considered the gold standard for capturing confirmed opioid overdoses and, thus, for assessing overdose prevalence, but the time lag in the data presents challenges for monitoring and response. ⁶ ED syndromic surveillance data are collected in near-real time—within 24-48 hours of an ED visit—and are helpful in tracking outbreaks, ⁷ including outbreaks of drug overdose. ^8,9 Trends in ED syndromic surveillance data can inform a more localized response. EMS encounter data, which are a timely option for detecting suspected drug overdoses, are especially important because they may detect overdoses that do not present to the ED. ¹⁰

CDC worked with states to develop ED- and EMS-specific case definitions for suspected drug overdoses. ED case definitions were created using chief complaint search terms and discharge diagnosis codes (eg, International Classification of Diseases, Tenth Revision, Clinical Modification [ICD-10-CM] codes). EMS case definitions primarily used text found in the primary or secondary impression fields of the encounter, whether naloxone was administered, and whether a response to naloxone was recorded (if available). Because naloxone is commonly and appropriately used to rule out the occurrence of a drug overdose in patients with altered mental status, severe respiratory depression, or apnea, administration of naloxone alone is not always an accurate indicator of overdose. Response to naloxone is therefore included to improve the sensitivity of the case definition. ¹¹

Within the ESOOS program, enhanced data on fatal drug overdoses were reported through the State Unintentional Drug Overdose Reporting System. The State Unintentional Drug Overdose Reporting System collects detailed information on toxicology, death scene investigations, route of administration, and other risk factors that may be associated with a fatal overdose that are not available through the National Vital Statistics System. To further increase the comprehensiveness of fatal overdose reporting, the ESOOS program also funded expanded toxicology testing of fatal overdoses by directly supporting medical examiners and coroners.

Overdose Case Definition Development and Validation

Three articles in this supplement illustrate the process of developing and validating new case definitions for the detection of opioid overdoses. Using ED syndromic surveillance data from North Carolina, Brathwaite et al ¹² compared the CDC-developed opioid overdose case definition with a state-based case definition to determine whether the state definition should include additional chief complaint text and/or ICD-10-CM codes from the CDC definition. A comparison of the definitions identified text and codes that increased the sensitivity of the state definition and that decreased false positives in North Carolina ED data.

Hallowell et al ¹³ described the development of a case definition to identify suspected opioid overdose–related EMS encounters using the primary and secondary impression field, naloxone administration, medication response, and keyword search of the report narrative. The article describes the importance of considering naloxone administration with other factors to avoid misclassification of opioid overdoses. After validation using an iterative process of random medical record review, the case definition accurately identified 90.0% of opioid overdoses and accurately excluded 83.3% of opioid overdoses. Use of the case definition enabled analyses that identified key patterns in overdose locations and people who experienced repeat overdoses. Another article from North Carolina by Sivaraman et al ¹⁴ described the development of a case definition for opioid overdose in EMS data. The authors compared a previously published definition with a new case definition generated using feedback from subject matter experts and a rule-based algorithm with innovative machine-learning techniques and found the new case definition had a higher positive predictive value. These articles highlight the importance of rigorous testing of syndromic surveillance definitions and demonstrate how programs can optimize their identification of opioid overdoses based on the intricacies of their own data.

Data Linkage

Improving linkages between disparate data systems is invaluable in informing interventions, and several articles in this supplement describe novel linkages used to provide more complete coverage and granularity of local overdoses than provided by either data source alone. Two articles examined linkage methods for EMS encounters and ED visits from syndromic surveillance data. In Massachusetts, Rahilly-Tierney et al ¹⁵ linked EMS ambulance trips and ED visits of people experiencing an opioid overdose from syndromic surveillance data through a deterministic match process. An article by Fix et al ¹⁶ described a novel iterative linkage approach that resulted in 91.1% of EMS encounters with naloxone administration linking with an ED visit in North Carolina. These linkages allowed for more accurate accounting of overdoses and a more complete picture of each event than provided by either data set alone, including initial patient presentations, field interventions, and ultimate diagnoses. Furthermore, improved EMS and ED data linkage could enhance surveillance activities, inform emergency response practices, and improve quality of care for drug overdose.

Although not a traditional data linkage, another article in the supplement compared data sources by evaluating time-series relationships between data sources. ¹⁷ As an ESOOS recipient, Kentucky supplemented its ED discharge data–based opioid overdose morbidity surveillance with EMS data and ED syndromic surveillance data to obtain more rapid data on opioid overdose. Rock et al determined how well the 2 rapid systems’ time series for monitoring opioid overdose trends aligned with the time series captured by traditional discharge data. Their findings indicate that both the EMS and ED syndromic surveillance data could be used to monitor nonfatal opioid overdose trends in Kentucky to inform public health action in near-real time.

Improvements in Toxicology Testing and Enhanced Biosurveillance

Drug overdose mortality data from 2017 and 2018 demonstrated that the types of opioids and combinations of drugs contributing to deaths are changing. Although the prevalence of overdose deaths involving prescription opioids and heroin decreased from 2017 to 2018, the prevalence of overdose deaths involving synthetic opioids other than methadone (primarily illicitly manufactured fentanyl) increased. ^18,19 Identifying fentanyl and fentanyl analogs in postmortem samples requires specialized toxicology testing that traditional laboratory settings often lack. Clinton et al describe how ESOOS funding was used to expand toxicology testing to detect novel synthetics in overdose deaths that would not have been identified through basic toxicology testing. ²⁰ Adopting expanded forensic toxicology testing is necessary to track changing trends in the overdose epidemic.

The Minnesota Department of Health implemented a pilot system to better understand the toxicology of nonfatal overdoses. ²¹ Although toxicology testing panels in hospitals may be limited in scope, such as panels used for forensic testing, many overdose-related ED patients do not receive toxicology testing, particularly if results would not lead to a modified clinical treatment plan. ^22,23 The Minnesota Drug Overdose and Substance Use Pilot Surveillance Activity used ED syndromic surveillance data, enhanced toxicology testing of clinical specimens, and medical record abstraction to describe trends in ED visits attributable to drug overdose and substance misuse. The pilot also found frequent patient exposure to substances not reported by patients or suspected by clinicians, demonstrating the usefulness of toxicology testing for nonfatal overdoses in characterizing the true epidemic. ²¹

Using Data to Inform Action for Localized Response

Ultimately, the goal of increasing the timeliness and comprehensiveness of nonfatal and fatal overdose data through the ESOOS program was to inform prevention and response efforts for opioid-involved overdoses. Three articles in the supplement describe how surveillance data from the ESOOS program were used to inform localized responses to the opioid overdose epidemic. Acharya et al ²⁴ described a data-driven approach developed to address increasing law enforcement actions in Maryland targeting opioid overprescribing. Using data from Maryland’s Prescription Drug Monitoring Program along with ED and EMS data, they responded to 12 actions during a nearly 2-year period to ensure that abrupt closures of facilities where opioids were prescribed did not result in an increase in the number of overdoses or in disruptions in care.

To provide rapid intervention and limit opioid overdose–related harms, Connecticut created the Statewide Opioid Response Directive. An article by Canning et al describes the development of the program based on rapid reporting of EMS data on suspected opioid overdoses and how the program was successfully used to identify and mitigate an outbreak associated with the use of crack cocaine mixed with fentanyl. ²⁵ Lasher et al ²⁶ similarly described their ability to respond to an increase in nonfatal opioid overdoses attributable to cocaine and fentanyl in Woonsocket, Rhode Island, detected by an alert in its ED syndromic surveillance data. Community response actions included naloxone distribution, deployment of peer recovery support services to connect people to treatment services, and targeted messaging about the outbreak.

Conclusion

Although strides have been made in recent years to strengthen public health data reporting to improve the timeliness, specificity, and comprehensiveness of opioid overdose data, more work remains as the epidemic evolves to ensure real-time response with the best local data available. Issues with data completeness and reporting, workforce capacity, and local variation in overdoses present challenges to prevention and response. Addressing these issues is the focus of CDC funding to health departments through the Overdose Data to Action cooperative agreement initiated in 2019. ²⁷ Funds awarded as part of this agreement support the expansion of data collection beyond opioid overdose to all drug overdose, increased frequency of nonfatal and fatal overdose data reporting, and innovative surveillance activities to support interventions. These activities help further increase the comprehensiveness of surveillance data and allow jurisdictions to tailor their surveillance efforts to specific needs. The articles in this supplement and future dissemination of Overdose Data to Action activities can be used to inform public health professionals, clinical care providers, and policy makers of the value of improved overdose surveillance and can provide surveillance practitioners with practical advice and lessons learned to monitor and direct prevention activities for overdose morbidity and mortality in their jurisdictions.