Background

Background

Background

SORT

• August 2021: SORT is launched as a collaboration between DPH + SFFD
• Team includes a Community Paramedic, a Street Medicine Clinician and other specialists and peer counselors.
• Respond to people immediately after an overdose, and again within 72-hours,
• They operate citywide, 12 hours per day, 7 days per week.

How did I get involved?

Data Sources

OCME data

SORT initial visits

SORT follow-up visits

List of Hotspots

These addresses are the top locations where fatal overdoses have occurred in 2020-21, according to OCME records.

Fatal ODs per month

Firelog Data

• Earliest and latest Firelog reports in this data: 08/02/2021, 02/27/2022
• Number of unique firelog encounters: 1112
• Numer of unique patients in firelog: 666
  • This is limited by the large number of Firelog encounters w no MRN: 344 encounters (31%)

Below is a map showing where the Firelog visits have occurred.

Timeline of ODs in hotspot locations

Sort Team Follow up Data

• Total number of visits in SORT follow up data: 324
• Total number of unique patients in SORT follow up data: 177

Are all SORT team follow up records accounted for in the firelog data set?

• Interestingly, 145 of the MRNs included in the SORT follow-up encounters are NOT represented in the initial firelog data. Are there patients being added to the list for follow up that did not come from the Firelog?

Patient demographics in SORT Follow-up Data

Encounter information

Standard reporting metrics (Bupe, Naloxone, HR kits,…)

Buprenorphine data from SORT Follow-up

Service referral

Subsequent overdose

• Number of people w at least one additional Firelog encounter for OD (fatal and nonfatal): 67
  • This represents 10 percent of all people who have had at least one Firelog encounter.

Below is an example of data showing subsequent overdoses, grouped by patient MRN. It lists the total number of encounters a person has in the firelog, as well as the unique number of each encounter listed. It also lists the disposition, followup narrative, and whether encounters were overdoses or not.

Subsequent fatal overdose

The OCME Data I have is anonymous, so it is not possible to match these data w firelog or Epic data. One possibility would be to use OME as the disposition, but this would be a huge under-representation of true fatal overdose deaths.

Questions

• For demographic data, do you want to use the patients in the firelog data, or in the SORT Follow up data set?
• How are we determining fatal vs non-fatal overdoses? Disposition = OME doesn’t return that many results, and may be unreliable given many patients may die en route to hospital.

Conclusions

• Opioid overdoses have hotspots; these seem to represent a mix of SROs, SIPs, and shelters
• There are also spikes in timing; this warrants further investigation
• The number of fatal ODs is still high, but it is decreasing. This trend started before SORT
• The SORT Team is reaching many people, but significant barriers still exist:
  • Most of the time, we can’t find people
  • Getting people to fill out spreadsheet data for every encounter isn’t working for all metrics
• Measing the specific impact of SORT is challenging, since many large forces impact OD rates

Next Steps

• Look into hotspots
• Connect w CBOs (DOPE project, Drug users union, and others)
• Try to get non-anonymous OCME Data, to track from non-fatal ODs to fatal ODs.
• Ongoing work on SORT program evaluation