Returning Disposition Data From 911 to 988: Building Feedback Loops That Improve Continuity, QA, and System Learning

Most 988–911 interfaces are built as one-way routes: 988 escalates, 911 responds, and the story ends. In reality, the story continues—and the system needs feedback. If 988 does not learn whether dispatch succeeded, whether the outcome was transport, stabilization, or no contact, and whether law enforcement was used, it cannot refine triage thresholds or reduce repeat crises. A structured disposition feedback loop returns essential outcome data from 911 to 988 in a controlled, minimal, and auditable way. This article is part of 988–911 crisis routing and interfaces and supports broader crisis response models that emphasize continuity and measurable improvement.

Why disposition feedback is a system requirement, not a “nice to have”

In crisis systems, learning depends on outcome visibility. Without feedback, 988 centers may over-escalate because they cannot see when mobile crisis would have been sufficient, or under-escalate because they never learn that a “low risk” case became an ED admission two hours later. Commissioners and state oversight bodies increasingly expect proof of continuous improvement: documented QA loops, corrective actions, and measurable changes in diversion and repeat-contact rates.

They also expect that data sharing is purposeful and proportionate. A disposition loop is not a request for full 911 records. It is a controlled minimum dataset that supports continuity, QA sampling, and system governance while respecting privacy constraints and local legal frameworks.

Define a minimum disposition dataset that is operationally useful

High-functioning systems agree a minimum dataset that 911 can return reliably without overburdening call takers or dispatch. Typical fields include: shared incident identifier, dispatch initiated (Y/N), responder type (EMS, law enforcement, co-responder, fire), contact achieved (Y/N), disposition category (stabilized on scene, transported to ED, transported to crisis receiving center, refused care, unable to locate), and timestamp of closure. The goal is to enable 988 to do three things: close the loop with the caller when appropriate, learn whether escalation thresholds are working, and trigger QA review when outcomes indicate risk or system failure.

Operational Example 1: Same-day continuity call after ED transport

What happens in day-to-day delivery: A 911 disposition feed flags that a 988-initiated escalation resulted in ED transport and that contact was achieved. Using the shared incident identifier, the 988 center schedules a same-day continuity call (or next-available window) to the caller or approved support person, depending on consent and local workflow. The call focuses on immediate stabilization: confirming discharge instructions if already released, clarifying follow-up appointments, reinforcing coping strategies used successfully, and confirming how to re-access support without re-entering crisis escalation. The interaction is logged against the original incident identifier.

Why the practice exists (failure mode it addresses): Without feedback, 988 may assume “911 handled it,” leaving a gap after ED contact—often the highest-risk period for repeat calls. The continuity call reduces fragmentation and supports step-down stabilization.

What goes wrong if it is absent: The caller returns from the ED with no structured follow-up, confusion about medications or safety planning, and no clear pathway back to community support. Repeat contact occurs within days, and escalation becomes the default cycle.

What observable outcome it produces: Reduced short-window repeat contacts, improved documentation of continuity actions, and clearer system evidence that escalation is followed by stabilization work rather than abandonment.

Operational Example 2: QA trigger when law enforcement-only response occurs

What happens in day-to-day delivery: The disposition feed identifies a law enforcement-only response on a 988-initiated escalation. This automatically triggers QA sampling within 24–48 hours. Reviewers assess whether the escalation tier was correctly applied, whether a mobile crisis option was available but not used, and whether the minimum dataset transferred to the PSAP supported proportionate dispatch. Findings are categorized (clinically justified, capacity-driven, protocol gap, documentation gap) and routed to a joint improvement log.

Why the practice exists (failure mode it addresses): Law enforcement-only outcomes can reflect necessity, but they can also reflect defaulting behavior under uncertainty. The trigger prevents normalization and forces the system to review whether thresholds and routing logic are working as intended.

What goes wrong if it is absent: Police involvement becomes routine without scrutiny. 988 staff cannot tell whether their escalation decisions are driving over-response. Community trust erodes, and the system becomes more defensive over time.

What observable outcome it produces: A measurable reduction in avoidable law enforcement-only dispatch, better alignment between escalation criteria and response type, and stronger audit trails for oversight review.

Operational Example 3: Detecting “no contact achieved” patterns to prevent silent failures

What happens in day-to-day delivery: The disposition feed flags “unable to locate” or “no contact achieved” for a subset of escalations. The 988 center applies a reliability workflow: confirm location data quality captured during the call, verify whether jurisdiction routing created delays, and conduct a structured recontact attempt where permitted. If patterns emerge (for example, a specific boundary area repeatedly failing), the issue is escalated to interagency governance for routing redesign or location-handling updates.

Why the practice exists (failure mode it addresses): “No contact” is often treated as an endpoint, but operationally it may signal routing flaws, location uncertainty, or handoff breakdown. The loop converts silent failure into actionable system learning.

What goes wrong if it is absent: The system assumes response occurred when it did not. The caller recontacts in a worse state, or harm occurs without the system understanding the root cause. Leadership lacks evidence to fix chronic interface defects.

What observable outcome it produces: Reduced recurrence of no-contact outcomes in known hotspots, improved location capture practices, and clearer governance action items linked to measurable defect patterns.

Governance, privacy, and implementation expectations

Two practical expectations tend to govern implementation. First, funders and state agencies expect documentation of the minimum dataset, decision rights (who can access it), retention rules, and QA sampling approach. Second, they expect that feedback is used—shown through change logs, updated training, and measurable movement in performance indicators (repeat contacts, diversion, and escalation appropriateness).

Implementation succeeds when the feedback loop is simple enough to run daily and rigorous enough to withstand audit. The shared incident identifier is the anchor: it allows continuity actions and QA review without requiring broad data exchange.

Measuring whether the loop is working

Systems commonly track: percentage of escalations with received disposition within a target window, percentage of law enforcement-only outcomes reviewed, repeat contact rates within 72 hours and 7 days, and “no contact achieved” rates by geography and time of day. These measures show whether the interface is improving or merely moving workload between agencies.