The United States is currently experiencing multiple, simultaneous epidemics that claim thousands of lives every week. According to the CDC, over 81,000 drug overdose deaths occurred between June 2019 and May 2020. That’s the highest number of overdose deaths ever recorded in a 12-month period. An estimated 93,000 Americans die annually from alcohol-related causes.
Many nuances are involved in measuring substance-use-related outcomes. How can we get a clearer picture of the substance-use epidemic?
Our motivations
Emergency medical services (EMS) play a huge role in the opioid crisis. Our team was interested in understanding how to use EMS data to track substance use-related outcomes, such as fatal and non-fatal opioid overdoses. Fatal overdoses are identifiable on death certificates (although there’s also considerable under-reporting of overdose deaths when causes are not known), but non-fatal overdoses aren’t easily tracked because they often occur outside the health care setting. Some overdoses that might otherwise result in death are resolved when family members or other bystanders administer an opioid antagonist like naloxone to reverse the overdose. Still, some substance-related overdoses result in 911 calls for emergency medical help. We were interested in how many EMS calls involve non-fatal and fatal substance-related overdoses and in outcomes for these calls.
EMS data
EMS data are reported by EMS agencies to their state, and most states report their data in a standardized format to the National EMS Information Systems (NEMSIS) technical assistance center. An anonymized and encrypted version of these data are available to researchers on request.
We obtained the 2019 data for this analysis and limited our sample to 911-initiated EMS activations (“calls”). We excluded calls that were canceled prior to arrival at the scene; canceled at the scene (e.g., no patient contact or no patient found); coded as “standby” (i.e., no services or support provided); or involved EMS operational support, such as transporting organs, etc.
About the sample
We studied 24.9 million total calls involving contact with a patient in 2019 in 44 states and DC (excluded: Delaware, Hawaii, Idaho, Massachusetts, Missouri, Ohio, and Puerto Rico). Not all EMS calls are included in NEMSIS; about 44% of the calls occurred in states that report less than 75% of the total calls in their state or where completeness of the data is unknown.
About our measures
First, we developed a set of substance use and outcome measures by combining information available in NEMSIS:
- ICD-10 diagnosis codes noted by emergency medical technicians (EMTs) that are associated with substance use (recorded as “symptoms” and “impressions”);
- indicators for drug or alcohol paraphernalia; indicators for specific drug overdose protocols and medications used (for example, naloxone); and
- records of direct substance use evidence (like a breathalyzer or blood test).
We further categorized calls with patient contact into calls resulting in:
- Deaths, either before the EMTs arrived or during the EMS call
- No treatment during the EMS call (e.g., not required or refused)
- On-scene EMT treatment, including those with transport
- A subgroup of those treated who were transported to a hospital emergency department (ED)
We list our main measures in the Table below. We made a number of decisions in developing our algorithms for these measures and would be happy to provide further details on request.
[wpdatatable id=1 table_view=regular]
The Figure shows the estimated prevalence of each of these measures in the 2019 NEMSIS data.
Source: Author analyses of National Emergency Medical Services Information System (NEMSIS) data, 2019. Click for full-size figure.
Comparing NEMSIS results to other sources
How do our estimates compare to other national data on acute care for substance-related causes? The CDC identified 70,630 drug overdose deaths in 2019. We identified 11,662 drug overdose deaths in NEMSIS data in 2019 — about 1 in 5 of the total number of drug overdose deaths reported by the CDC in 2019.
The Overdose Detection Mapping Application Program (ODMAP) collects real-time data on both fatal and nonfatal overdoses, but the data are not yet collected nationally, and they do not release the data to researchers. As for substance-related ED use, the CDC’s Drug Overdose Surveillance and Epidemiology (DOSE) System currently only reports percent change estimates in rates, not the rates themselves, leaving a major data gap.
Conclusions
We developed a new algorithm for identifying EMS calls involving suspected substance use and outcomes for those calls. Using a combination of factors including diagnosis and symptom codes, protocol and evidence indicators, and other variables that suggest substance use, we found that:
- About 9% of reported EMS calls with patient contact in 2019 may have involved substance use.
- Of those, 65% involved alcohol, and 41% involved drugs.
- These do not sum to 100% since some calls involved both alcohol and drugs.
- In addition, we identified around 12,000 fatal and 240,000 nonfatal drug overdoses in 2019.
This suggests the NEMSIS data may provide a useful data source for surveillance of non-fatal drug overdoses and trends in this measure. However, among calls involving drugs, we were able to identify less than half using diagnosis codes alone, suggesting the need to balance the possibilities of either under- or overcounting. Remember that for many EMS calls, there were no diagnoses reported. Moreover, since the data are at the incident level and cannot be analyzed at the level of a unique person, the actual numbers of individuals involved are unknown. Finally, we did not have access to post-EMS information, such as diagnosis codes assigned by ED clinicians.
While we can use NEMSIS to evaluate some aspects of acute care for substance use, we must acknowledge its limitations. Nevertheless, we believe EMS data can add more to the picture of the substance use epidemic in the US than is described based solely on ED visits or mortality statistics.
The authors gratefully acknowledge the NEMSIS TAC for sharing, and helping us analyze and interpret, the data.
Lisa M. Lines
Latest posts by Lisa M. Lines (see all)
Diana Zabala