Insulin Order Set: Two Near-Misses and a Hard Stop by Friday
Three weeks after go-live of a redesigned insulin order set, two near-misses occurred. In the first, a nurse accepted the preselected sliding-scale option on a patient whose provider had ordered a basal-bolus regimen. In the second, two insulin orders displayed simultaneously on the MAR because the prior order was not discontinued when the new set was activated; the nurse caught the duplication at the bedside. An analyst proposes a quick fix: a hard stop that blocks the MAR until the provider discontinues the old order. Nursing leadership wants it deployed by Friday. The change control board does not meet until next month, and the next scheduled build window is in nine days.
1. Analyze the two near-misses. What design and process failures does each represent, and would the proposed hard stop have prevented either?
Read the worked answer
Event 1 is a default-value problem. A preselected sliding-scale option makes the wrong choice the path of least resistance: the nurse must notice and actively change a field that the system has already answered. Defaults are a form of decision support whether or not anyone designed them as such, and the safe default is either the option most likely to be correct or no selection at all, forcing deliberate choice. Secondary contributors include whether order verification requires the nurse to read back the full regimen rather than the first line, and whether the order set visually distinguishes the regimen types.
Event 2 is an order-lifecycle problem. Activating a new set did not discontinue the prior orders, and the MAR faithfully displayed both. This is a build logic and reconciliation failure, not a nurse failure; the bedside catch is the safety net working, not the control working.
The hard stop addresses neither fully. It would not touch Event 1 at all. For Event 2 it would block administration until discontinuation - but by converting a duplicate-display problem into a possible delay of a time-critical medication, it introduces a new failure mode (missed or late insulin) while leaving the underlying activation logic unfixed. A hard stop is also a workaround that will persist long after the real defect is corrected, and hard stops tend to be removed only after they cause an incident of their own.
2. Recommend an action plan for the next five days and for the next 90 days.
Read the worked answer
Next five days - interim controls, not the fix:
- Flag every patient with an active insulin order on a daily pharmacy review report; pharmacist verifies one active regimen per patient.
- Require an independent double-check of the regimen type at the first dose after any order set activation, communicated as a safety alert rather than a policy change.
- File safety reports for both near-misses so they enter the event system and reach risk management.
- Send a targeted communication to prescribers and nurses on activating the new set and discontinuing prior orders.
- Notify the CMIO, pharmacy, and the order set owner in writing the same day.
Ninety days - the actual fix:
- Remove or neutralize the sliding-scale default; if the vendor permits, force an explicit selection.
- Correct the activation logic so prior insulin orders are discontinued or flagged for reconciliation.
- Add duplicate-therapy decision support for insulin as a backstop, designed with override-reason capture.
- Run usability testing with bedside nurses on the redesigned set - both events are the kind of finding a five-nurse think-aloud session surfaces before go-live.
- Track balancing measures: hypoglycemia events, near-miss reports, override rates, and time from order to first dose.
3. How do you deploy a change this urgent without simply bypassing change control and creating a new risk?
Read the worked answer
Compress governance rather than skipping it. The goal is a documented emergency change with the same evidence standard applied faster.
Do: initiate the organization's emergency change procedure if one exists - most change control policies have one and it is simply underused. Provide a written risk assessment comparing the risk of waiting nine days against the risk of deploying an untested hard stop. Obtain explicit approval from the CMIO, CNO, and pharmacy, recorded. Test in a non-production environment even if abbreviated; a hard stop that misfires on all insulin orders is a worse event than the one you are preventing. Define a rollback plan and the person authorized to execute it. Stage the deployment on one unit for 24 hours with active monitoring before enterprise release. Schedule retrospective review by the change control board as a condition of approval, not an afterthought.
Do not: deploy an untested hard stop on a Friday, deploy without a back-out, or let the interim control quietly become the permanent solution.
Also name the tradeoff explicitly to leadership: the fastest safe option is likely the default removal plus the pharmacy double-check, with the ordering-logic fix following in the next build window. If leadership still wants the hard stop on Friday, ask them to own the decision in writing, and say what you would monitor for as the early sign it is harming patients.