States’ Methods for Capturing and Reporting Local Responses to Suspected Nonfatal Drug Overdoses

Abstract

This study examines the available data used to surveil state-level drug overdoses and characterizes them based on their timeliness, drug involvement, and specification of nonfatal outcomes.

Nonfatal overdose is a key predictor of future drug overdose death.¹ State nonfatal overdose data can inform local communities about the dangers of the illicit drug supply and provide data-driven insights to inform the distribution of lifesaving overdose-reversing drugs (eg, naloxone), harm-reduction resources, and treatment services.² Current state-level surveillance methods track encounters with emergency medical services, visits to emergency departments (EDs), and hospital admissions to understand overdose trends; however, not all states publish data on nonfatal outcomes.³ This study identified the types of publicly available data used to surveil state-level drug overdoses and characterized these efforts based on their timeliness, drug involvement, and specification of nonfatal outcomes.

Methods

A manual review of published reports and surveillance dashboards on local responses to suspected drug overdoses from all US state and district government websites was performed in July through August 2022. Data sources used for identifying overdoses were evaluated and categorized based on responses occurring in prehospital (by first responders or community bystanders) and within-hospital (EDs or inpatient hospitalizations) care settings. Two analysts independently reviewed each website and together adjudicated any differences.

Data were assessed for confirmation of drug involvement based on suspicion (from syndromic surveillance), clinical impression (indicated in discharge data or by first responders reporting symptom improvement following naloxone administration), or toxicology testing. Responses were considered nonfatal if states specified that they were nonfatal or if they published detailed outcome information that differentiated between fatal and nonfatal. Data sources were characterized by availability of demographic stratifications, geographic granularity, whether data were reported in real time (within 2 weeks of the overdose), and whether states performed deduplication efforts to capture unique events. Detailed definitions for characteristics are available in Supplement 1.

Results

The Table summarizes published state drug overdose data by reviewed characteristics. ED visits were the most common data type (n = 45), followed by first responders (n = 34), inpatient hospitalizations (n = 22), and community responses (n = 15). Two states did not publish any data. Among states reporting ED data, 33 published overdose estimates involving any drug, 39 involving opioids, and 23 involving other nonopioids. For states reporting first responder data, 21 published estimates involving any drug, 19 published estimates involving opioids, and 2 published estimates involving other nonopioids. Three states reported using toxicology testing to confirm drug involvement, while the remaining were based on suspicion or clinical impression. Nonfatal outcomes were specified by 22 states reporting ED data, 12 states reporting inpatient data, 16 states reporting first responder data, and 7 states reporting community data. Among states reporting ED data, 34 published stratifications by age, 31 published stratifications by sex, 22 published stratifications by race, and 15 published stratifications by ethnicity. For states reporting first responder data, 16 stratified by age, 14 stratified by sex, 4 stratified by race, and 3 stratified by ethnicity. County was the most common geographic unit reported across data types. Arizona, Michigan, and Rhode Island reported data in real time, and only Maine reported deduplicating events across settings.

Table. Number of State and District Governments That Publish Data on Local Responses to Suspected Drug Overdose by Reviewed Characteristics^a.

Characteristics	Prehospital care setting		Hospital care setting
	First responder	Other community setting	ED visit	Inpatient	Other
State publishes data type
Yes	34	15	45	22	7
No	17	36	6	29	44
Drug typeb
Any drug	21	6	33	16	4
Opioids	19	10	39	22	7
Other nonopioids	2	3	23	13	3
Naloxone delivery
Yes	30	10	0	0	0
No	4	5	45	22	7
Drug involvementb
Suspected	27	9	23	1	0
Clinical impression	7	6	22	21	7
Toxicologic confirmation	0	0	3	1	1
Specified as nonfatal
Yes	16	7	22	12	4
No	18	8	23	10	3
Reporting characteristics
Static report (most recent year)
2022	7	2	11	0	1
2021	6	3	5	4	0
2020	5	3	3	2	1
2019 or prior	5	0	9	5	2
No static reports	11	7	17	11	3
Interactive dashboard (most recent year)
2022	11	2	11	1	0
2021	4	5	8	4	0
2020	4	1	8	5	4
2019 or prior	2	1	7	6	1
No dashboard	13	6	11	6	2
Real-time data reporting
Yes	3	1	2	1	0
No	31	14	43	21	7
Available stratifications
Demographicsb
Age	16	2	34	17	4
Race	4	1	22	15	1
Ethnicity	3	1	15	9	0
Sex	14	2	31	16	3
Geographic granularity
State	11	8	8	4	0
Public health region	1	0	1	1	0
County/parish	18	7	33	16	5
City, zip code, neighborhood, or geographic coordinates	4	0	3	1	2

Abbreviation: ED, emergency department.

^a Among states publishing any type of data, 9 reported performing any deduplication efforts to capture unique overdose events and 1 state reported performing deduplication efforts across the prehospital and within-hospital care settings

^b Counts are not mutually exclusive within this data characteristic. Other community responses include responses by poison control centers or naloxone administrations delivered by a community grant program, layperson, or bystander prior to emergency medical services responding. The other health care setting group includes drug overdoses that resulted in health care utilization more broadly, such as aggregate counts of hospitalizations (ED and inpatient) or Medicaid claims. Geographic granularity was based on the most granular unit published by states. Real-time data reporting was defined as publication of data within 2 weeks of the overdose.

Discussion

Nearly all states reported overdose data from EDs, two-thirds published first responder data, and less than one-third tracked community responses. These findings indicate that not all states comprehensively report nonfatal overdoses and current estimates may be undercounted given that not all people who experience an overdose connect with medical services or community programs.

Methods for reporting drug overdose data varied from state to state and within data types. Given variation in states’ methods for capturing nonfatal overdose, states and the federal government should continue to work together to expand and standardize reporting of nonfatal drug overdose data in a comprehensive, timely, and geographically precise manner.^4,5 Such data can be used to predict and prevent fatal drug overdoses and to target and evaluate drug policies and programs in an equitable manner.⁶

This analysis was limited to data published on government websites and may not reflect the entirety of states’ overdose surveillance capabilities. States may restrict analyses for internal use or report data to federal agencies but not publish it. Publication of estimates by demographic and geographic stratifications may also be limited in states with small populations.

Section Editors: Jody W. Zylke, MD, Deputy Editor; Kristin Walter, MD, Senior Editor.

Supplement 1.

Definitions and characteristics of data published on State and District government websites to identify responses to suspected drug overdose

Supplement 2.

Data Sharing Statement