Four Insulin Errors, a 22% Barcode Override Rate, and a Demand for Discipline
A 260-bed community hospital has had four insulin infusion errors in eight weeks. Two involved U-500 concentrated insulin stored on the same shelf as U-100 vials; one reached the patient and produced a glucose of 28 mg/dL, treated with dextrose and no lasting harm. The barcode medication administration override rate on the medical unit is 22% against a system target of 5%, and the override-reason log is dominated by "scanner will not read." The involved night nurse has 14 months' experience and was floated from a surgical unit with a two-hour orientation. A pharmacy technician edited a label template two months ago without change control. Risk management wants the nurse referred to peer review and the CNO wants disciplinary action to "send a message." The patient's family has not yet been told the event was preventable, and the hospital's event reporting system shows the four events were reported under three separate categories.
1. Apply a just culture classification to the night nurse, the pharmacy technician, and the unit leadership, and state what action follows for each.
Read the worked answer
Separate the behavior from the outcome. The outcome here was serious; that does not determine the response.
Night nurse. Pull the override log and the workflow. If 22% of scans on the unit are overridden and the reasons are dominated by "scanner will not read," the nurse was working in a system that routinely forced bypass - at-risk, system-induced behavior. The response is to remove the barrier, coach, and verify competency, not to discipline. If the record shows the nurse knowingly administered a product that did not match the patient or the order, that moves toward reckless behavior and warrants a formal process. Peer review is for clinical judgment questions; it is the wrong venue for a scanner failure.
Pharmacy technician. A template edit outside change control is a process failure. Determine whether the technician knew a safety review was required; most such edits are made to solve a local annoyance. Fix change control regardless; individual action only for a knowing bypass.
Unit leadership. Accountability belongs here: override rates were unmonitored, and a nurse was floated with two hours of orientation to a unit handling concentrated insulin. That is a leadership system failure and needs expectations plus support, not a scapegoat.
Separately: the family must be disclosed to, and the four events must be reclassified under one category so the pattern is visible.
2. Which controls would you put in place at 72 hours, 30 days and 90 days, and how would you know they worked?
Read the worked answer
72 hours. Physically separate U-500 from U-100 - different storage location, distinct shelf labeling, auxiliary warning labels. Remove U-500 from floor stock, or require pharmacy verification plus an independent double-check before administration. Huddle every shift with the actual event story. Begin 100% audit of insulin administrations for two weeks. Merge the four events into one reporting category.
30 days. Survey barcode scanner coverage on the medical unit to find dead zones and failed devices, and fix them. Stand up a weekly override dashboard by unit with reason codes and a named reviewer. Write a change-control policy that covers label and order-set templates, with a safety review gate. Require documented float orientation and insulin competency before a floated nurse takes an assignment.
90 days. Run an FMEA on the full insulin process, review smart pump library guardrails, and deliver just-culture training to managers so the next event is classified consistently.
Measures: U-500-related events (target zero), barcode scan compliance ≥95%, override rate below 5% with a shift in the reason mix, audit findings per 100 administrations, and staff-reported barriers. Sustainability depends on hard controls - physical separation, forced functions, EHR hard stops - rather than education alone, and on a control plan with an owner and a monthly review. Also sweep the other units for look-alike storage before assuming this is a single-unit problem.