Referral and Patient Access Diagnostic

01

Define the unit before counting the loss

Choose referrals received during a fixed period and follow that same group to a stated cutoff date. Separate duplicates, incomplete referrals, clinical redirection, patient declines, unreachable patients, appointments still pending, and unknown status. Unknown is a finding, not permission to label someone a no-show.

Agree who can judge clinical urgency and appropriateness. Administrative staff should follow an approved escalation pathway, not improvise clinical triage. This diagnostic maps operations; it does not replace clinical protocols, the patient record, or incident reporting.

02

Walk the journey with the people doing the work

Ask a scheduler, receiving clinician, referring-team representative, and practice leader to describe the same journey. Include patient feedback through an approved channel. A process diagram is not verified until someone can show where the work is recorded and who receives it next.

• Received: Can the team confirm receipt, completeness, routing, and an accountable owner?
• Contacted: Was there two-way contact, or only an attempted call? Are language, timing, accessibility, transport, cost, and coverage barriers visible?
• Scheduled: Does the offered appointment match clinical direction and the patient's circumstances? Who acts when no suitable slot exists?
• Completed: Was the visit completed, declined, cancelled, redirected, or still pending? Keep these outcomes distinct.
• Follow-up assigned: Is the next action accepted by a named role, with a due point and an escalation path?
• Referrer informed: Has the appropriate information returned through the approved channel, with receipt or reconciliation where required?

03

Measure the transition, not just the total

For each stage, record the number eligible to reach it and the number that did. Define timestamps, exclusions, follow-up window, and data owner before comparing periods. A newer referral group has had less time to complete its journey; do not compare immature and mature cohorts as if they were equivalent.

Use a small measurement set: receipt-to-first-attempt time; two-way contact among eligible referrals; completed visits among referrals eligible for a visit; unresolved referrals by age; and confirmed loop closure. Pair these with a balancing measure such as staff time per referral or patient-reported difficulty. A higher conversion rate accompanied by worse access for complex patients is not an unqualified improvement.

04

Choose one handoff to test

Prioritize by patient consequence, repeated failure, and the team's ability to change the mechanism. Safety concerns go to the established clinical escalation process immediately; they do not wait for the 30-day experiment. For routine operational improvement, choose one change with an owner, backup, receiving role, acceptance signal, and review date.

Use the existing secure system where possible. Do not create a second patient list in this download. A shared spreadsheet can look efficient while quietly producing conflicting records and unowned work.

05

Illustrative example: a queue with no receiving owner

A hypothetical practice reviews 40 eligible referrals at the same cutoff: 30 have two-way contact, 24 are scheduled, 18 have completed visits, and 6 are scheduled for future dates. Calling all 22 without completed visits 'lost' would be misleading. The team first reconciles status and finds that several referrals are waiting for missing documents with no assigned follow-up.

The test is a daily review of that missing-document queue by one role with a backup and an explicit handoff to the referring team. At day 30, review unresolved age, completion among comparable cohorts, staff minutes, and patient barriers. Continue only if the change improves the process without hiding work elsewhere. These are invented teaching numbers, not a benchmark or claimed result.

06

Avoid the local win that harms the whole journey

• Do not reward staff only for attempts or booked slots; those counts can rise while patients remain stuck.
• Do not treat patient choice or appropriate out-of-network care as a failure to be prevented.
• Do not make technology the first intervention when the real problem is ownership or capacity.
• Do not include names, dates of birth, identifiers, clinical details, or small identifiable groups in the public worksheet.
• Do not declare success without checking the receiving team's workload and whether the referrer received the outcome.

Continue with a related resource

• Business of Medicine