Emergency Medical Service (EMS) data has been successfully employed in the past to complement opioid overdose surveillance data provided by the traditional routes. While EMS data still has the limitation of being non-standardized and dependent on the documentation of the EMS provider; it has the advantage of providing information on non-fatal overdoses that would and could not be captured elsewhere. While EMS data in reality starts from a 911 dispatch with a Computer Aided Dispatch (CAD) and in many communities, a call taking set of protocols and guidelines using ProQA or other Emergency Medical Dispatch (EMD) system, moving to an Electronic Patient Care Record (ePCR) and finishing in some cases with Emergency Department Hospital records, we rarely are able to trace the entire path that a single 911 call generates within data sources. Correctly identifying an Opioid Overdose continues to be a challenging task. Its clinical features are non-specific and EMS providers can easily be deceived on the true underlying cause of the present illness. Identifying a set of indicators of overdose that provide key elements throughout the entire EMS data sources can provide a more robust understanding of opioid overdose surveillance within the EMS context.
As EMS data is noted to be a typically “dirty” data source of information, in order to mine it for overdose data, one must look in non-standard locations. With that in mind, we have created a set of clinical indicators of overdose that illustrate the acknowledgment of key features through the entire EMS data. From the CAD system we have used a Chief Complaint or Problem of Overdose/Poisoning ingestion, along with a free text search in the comment section for overdose related keywords. To complement the CAD information, from the EMD we used a ProQA determinant of Protocol 23: Overdose/Poisoning (Ingestion). From the ePCR we obtained a Primary and Secondary Impression of Overdose along with a free text search in the Narrative for any opioid overdose related keyword. We used the identical set of keywords in both the CAD-Comments and ePCR-Narrative free text search. An intervention of Narcan/Naloxone administration as well as vital signs where included. From the vital signs we used an altered mental status determined by a Glasgow Coma Score (GCS) of less than or equal to 13 and a decreased Respiratory Rate (RR) rated of less than or equal to 12 per minute.
The following table depicts the data obtained within each of the EMS data sources:
To build our 2 indicators of overdose OD1 and OD2 we used a combination of “and” and “or” statements. OD1 included a CAD-Chief Complaint of Overdose, or an EMD-ProQA code of 23, or a Primary/Secondary Impression of Overdose, and at least 1 keyword present in either the CAD-Comments or the ePCR-Narrative. OD1 also included an intervention of Narcan and a documented improvement of either the GSC or the RR after the medication was administered. OD2 was built with a CAD-Chief Complaint or ProQA Determinant, or Primary/Secondary Impression of Overdose along with a Narcan intervention or a GSC ≤ 13 and a RR ≤ 12. Although we have created OD1 and OD2 with filtering conditions through the entire EMS data sources, individual criteria can be built within the context of a single data source. Our next steps would be to determine validation metrics for each individual data source and filtering conditions we used to build OD1 and OD2.
We participated in the information swap meet at the International Society for Disease Surveillance (ISDS) Conference, January 29th – February 1st, 2019, in San Diego, California where we presented our findings. Click here for current poster.
Please check back with us on this page, or you may alternatively sign up for our Health Intelligence updates and be notified when this page is updated with the detailed recommendations and the poster is available. You may visit our Health Intelligence page for further resources related to awareness and mitigation of opioids and overdoses in response to this nationwide but often local crisis.