Where a service begins and a risk signal gets lost
7 min read
Lesson objectives
- Apply where a service begins and a risk signal gets lost to a service case with evidence and decision limits.
Please sign in to use all the tools
Once you sign in, the reading tools are yours: larger or smaller text, typeface, line spacing, focus mode, highlighting and bold, and notes saved to your account and gathered in My study.
Imagine an outpatient unit promising that a result will reach the responsible team within a day. The results team, however, believes that its follow-up duty begins only when a clinician opens the message. A risk signal can disappear in the space between those promises. A quality manager cannot decide the person's diagnosis or treatment from this description. The manager can examine who owns the transfer, what demonstrates completion, and what happens when completion cannot be demonstrated. We begin with the boundary of a service system. Here a system means the linked work, people, data, tools and decisions that deliver a defined service, rather than a department name on an organization chart. A map that starts at the clinic door and ends when the person leaves may omit consequential work after the visit, including delivery of results or resolution of a pending referral. Set the start and end from the service promise and its associated risk, then trace accountability across organizational boundaries. A useful map makes a handoff testable. An exhaustive diagram that hides the decision point is less useful to a manager.
First define the unit of analysis. It might be one outpatient visit, one referral request, or one result requiring acknowledgement. These units must not share a denominator without explanation: visits, results and referrals are different populations. State an observable start event, such as registration of a referral request, and an end event that shows the promise was met, such as acknowledgement by the receiving service and assignment of follow-up responsibility. List the intervening steps as observable verbs: create, review, send, receive, acknowledge, and follow up. For each transfer name the sender, receiver, channel, receipt evidence, expected time, and exception route. A sent message is not proof of reading. Reading is not proof that a suitable professional action followed. Each state needs different evidence. If a local escalation interval has not been established, mark it for confirmation against local policy instead of inventing an official deadline. The administrative question concerns whether the system reliably prevents an orphaned task. The appropriateness of a clinical decision belongs to the qualified clinical team.
Build a compact decision map with three columns: event, evidence and response owner. At referral approval, evidence is a reference number, timestamp and destination, owned initially by the sending team. At handoff, evidence is a receipt tied to the same reference, owned by the receiving team. When the locally agreed interval passes without receipt, do not close the task merely because software shows Sent. Move it to an exception queue with an owner and review time. If the teams use different systems, test whether their identifiers match before calculating a delay. Resending a request may create a duplicate, so distinguish a retry from a new referral. Exception counts can initially rise after surveillance improves; that may reflect better visibility rather than worse service. Compare rates with a stable denominator, inspect a sample of underlying records, and note cases attributable to delayed synchronization. This combination of operational data and direct checking is more informative than a single dashboard figure. It also prevents a confident explanation from outrunning the available evidence.
When presenting the map to service leaders, ask more than whether the written steps look right. Have both a sender and a receiver describe a busy day, a shift change, and the absence of the usual coordinator. That comparison exposes the difference between the stated process and work as actually done. In the referral case, a procedure may say acknowledgements are automatic while the event log shows they depend on one person opening a shared inbox. The improvement choice then changes. Training everyone to press Acknowledge might help a little, but a shared work queue with a named backup during absence addresses the single-person dependency more directly. Do not prescribe that solution before testing the explanation. Compare absence days with ordinary days and ask staff which exceptions the software cannot display. If an automatic acknowledgement is generated before actual receipt, a higher acknowledgement rate may offer false reassurance. Leaders must test the validity of evidence, not merely notice that a field is populated.
There is another boundary tradeoff. A broad map reveals a whole journey but can dilute accountability for the first test. If the observed gap concerns referrals from one clinic to one receiving department, begin with that interface and state its links to the wider journey. Do not announce that every referral in the organization has the same failure. If the investigation shows that the receiving department acknowledges requests but allocation within that department is delayed, extend the end point to assignment of follow-up and name its owner. This revision follows new evidence; it is not a cosmetic redraw. Keep a dated record of the boundary and denominator definition so that the team knows why a measure changes later. When comparing two months, recalculate the earlier month under the new definition if the underlying records permit. Otherwise, show a visible break in the series. Such discipline stops a leadership group from rewarding a numerical improvement produced solely by redefining the work.
Lesson summary
Apply where a service begins and a risk signal gets lost to a service case with evidence and decision limits.