Decreasing abandonment of calls to the 988 Suicide and Crisis Lifeline

2023-07-25

2023-08-17

- fraction of calls connected to a counselor.

- likelihood that the call is routed
- time elapsed from start of the call until disconnection or connection with a counselor
- use of the 988 text line within 1 hour of the initial call
- likelihood of follow-up calls back to the 988 line within 24 hours of the initial call
- likelihood that a caller connects with a counselor via phone or text within 24 hours of the initial call
- likelihood that someone selects IVR option “0” during the call
- likelihood that someone selects IVR option “1” during the call
- likelihood that someone selects IVR option “2” during the call
- likelihood that someone selects IVR option “3” during the call
- likelihood that someone selects IVR option “4” during the call
- time point during the IVR message when a call disconnects

Some callers to the 988 system will hear the existing 988 IVR messages and content and waiting messages and content; others will hear revised messages and content designed to reduce dropped calls. Which version of the messages and content a caller hears will be based on a randomized allocation of all possible last-4-digit sequences of 10-digit phone numbers. Then we will compare outcomes for incoming callers who hear the two versions.

The Department of Health and Human Services (HHS), Substance Abuse and Mental Health Services Administration (SAMHSA) and Vibrant Emotional Health (who implements and administers the 988 system) aim to make changes to the integrated voice response (IVR) system and messages that callers hear when calling 988 and attempting to reach a crisis counselor. The intervention involves changing the message script and voice recording that callers hear during the initial IVR (before a call is routed) and the script, voice recording, and waiting music that callers hear while waiting to be connected to a counselor. We will evaluate whether changes to the IVR and waiting message content increase call connections.
To do this, the evaluation will compare outcomes for incoming callers who are randomly assigned to hear either the current version of the 988 IVR and queue messages (the control condition) or a revised version of the messages (the treatment condition).
Control condition: Callers initially hear an automated message with options the caller can choose by pressing the appropriate number. The full initial IVR message callers hear if they do not choose any IVR menu option is approximately 45 seconds long. If the caller remains on the line to the end of the IVR message they will be routed to a queue while the system finds an available counselor. After a brief pause the caller hears music after the call is routed and hears cut-in messages at 18-second intervals.
Treatment condition: Callers initially hear an automated message with the same options as the control condition. The message has been revised to be shorter and increase clarity. If the caller remains on the line they will be routed to a queue while the system finds an available counselor. After a brief pause the caller hears a message indicating that the system is checking for available counselors and they will hear music while this happens. Music plays after an initial post-routing message of what to expect, and then the caller hears cut-in messages at 30-second intervals. The messages are recorded with voice talent that was selected to sound more like someone a potential caller would want to have a conversation with. Minor changes have been made to the wording of the messages to reduce length. Post-routing cut-in messages have been revised to include affirmative messaging encouraging callers to stay on the line and highlight other resources available to callers (including chat 988, text 988, and 911 if necessary). The wait music has been changed.
The intervention does not change the messages or content that callers hear if they select any of the IVR menu options. Those messages and content after selecting an IVR option will remain the same for both the control and treatment conditions.
Callers will be assigned to either the control or treatment condition based on the last four digits of their 10-digit phone number. The randomization of 4-digit-sequences is stratified by the 8th, 9th, and 10th digits to ensure balance across treatment and control for any patterns in assignments of these digits. Once the study period begins, calls from numbers with any of the 5000 control-assigned last-4-digits will be hear the control version of the IVR and queue message, and those with the other 5000 will be assigned to the treatment version.

Randomization done in office by a computer

Last 4 digits of phone numbers

Yes

10000

Approximately 315,000 calls

5000 intervention (new IVR), 5000 comparison (status quo IVR)