Every metric runs on the correct subset. Layer 1 = all emergency-response encounters (eResponse.05). Layer 2 = TAD transports: a Layer-1 encounter transported by this unit (eDisposition.30) to an authorised alternate-destination facility. Layer 3 = everything else. Layer 1 is the denominator for transport rate.

A transport whose destination facility code is on the approved list. The destination type (Mental Health or Sobering Center) comes from that code. A VA-style facility only counts when the destination type (eDisposition.21) equals the configured value — otherwise it was a regular hospital transport. The facility list and qualifier are editable in tad_config.py.

TAD transports as a percentage of all emergency encounters (Layer 2 ÷ Layer 1). Non-transports and their refusal/release reasons are tracked separately.

A patient who had a TAD transport and any EMS encounter within 72 hours afterward. Confirmed = exact LAST|FIRST|DOB match. Probable (QA review) = DOB exact plus name similarity at or above the configured threshold. Patient-identifiable detail lives only in the QA sheet, never in the dashboard.

TAD transports that may not have met alternate-destination criteria. Flag A: altered patient at any TAD destination (impression, symptom, OR acuity too high — all three checked independently). Flag B: overdose routed to the sobering center (belongs at an ED). Flag C: dystonic reaction at any TAD destination. Value sets are editable in tad_config.py.

Median minutes for dispatch→scene, scene→destination, transfer of care, and total unit time, computed on TAD transports. Intervals outside the configured sanity bounds are dropped.

Primary symptom grouped into Behavioral Health, Substance Use, or Other, plus the initial patient-acuity distribution. Together they describe who is being routed to alternate destinations.

Age groups and sex for TAD patients, for equity review across the alternate-destination population.