Barcode Scanning, a Cluster of Wrong-Patient Events, and a Manager Who Wants to Fire Someone
A 320-bed hospital went live on a new EHR and barcode medication administration system 11 weeks ago. Pharmacy reports that the reported medication administration error rate rose from 0.9 to 2.6 per 1,000 doses, and three wrong-patient administrations occurred in a single week, all on the 5 West night shift. The scanner override rate is 23% systemwide against a vendor benchmark of 5%; override reasons are dominated by "scanner malfunction" and "patient armband unreadable." Two of the nurses involved in the wrong-patient events were placed on administrative leave by the nurse manager, and staff have since stopped filing incident reports - reporting volume fell 41% in two weeks. The CNO is inclined to terminate one nurse with a prior documented error.
1. Is termination defensible under a just-culture framework? Explain your answer to the CNO.
Read the worked answer
Termination is defensible only if this is recklessness, and the facts point the other way.
Run the classification:
- Human error - an unintentional slip. Console and fix the system.
- At-risk behavior - a shortcut taken because the system makes the safe path harder. Coach, remove the incentive to drift.
- Reckless behavior - conscious disregard of substantial and unjustifiable risk. Manage through sanction.
A 23% override rate against a 5% benchmark is the signature of at-risk behavior driven by the environment: if nearly one in four scans is overridden, the scanner and armband hardware are not working, and the override is becoming the routine workflow. Sanctioning individuals for behaving the way the system rewards is both unjust and strategically self-defeating.
The reporting collapse is the more urgent signal. A 41% drop in voluntary reports means the organization has just lost its sensor network; that is a patient safety emergency in its own right. Terminating a nurse now will be read by staff as confirmation that reporting is dangerous.
Recommended: rescind the administrative leave absent evidence of recklessness, address the prior documented error through the performance process separately, and publicly separate the system fix from any individual discipline.
2. What analysis would you commission, and which corrective actions come first?
Read the worked answer
Aggregate the three wrong-patient events rather than analyzing them individually - a cluster in one week on one unit suggests a shared cause.
Analysis:
- A multidisciplinary aggregate review (RCA-style, or apparent-cause analysis if the facts converge quickly) covering the three events, plus a review of all override events on 5 West for the same period.
- A failure mode and effects analysis on the full barcode loop: admission armband printing, armband durability, printer placement per unit, scanner charging and docking, wireless dead zones, and the override reason taxonomy.
Sequencing by risk reduction per unit of effort:
- Eliminate the two dominant override drivers - put an armband printer on 5 West, replace or repair scanners, and stop using "scanner malfunction" as a free-text catch-all; replace with structured reasons that trigger a work order.
- Redesign the override screen so it requires a second identifier and captures the reason as discrete data.
- Reinforce two-patient verification during the night shift, when staffing and cross-coverage patterns create the highest interruption load.
- Only then consider policy changes or individual performance plans.
3. How will you know whether the interventions worked, and what would tell you they did not?
Read the worked answer
Use a balanced set, not a single number.
Leading and process measures:
- Override rate, overall and by reason code (target trajectory toward 5–8% within 90 days).
- Armband audit: percentage of patients with a scannable, correct armband on daily rounds.
- Scanner uptime and mean time to repair.
Outcome measures:
- Wrong-patient administrations per 1,000 doses, with statistical process control charts rather than month-to-month comparison.
- Reported errors per 1,000 doses - but read this carefully. Reporting volume should rise as trust returns. A falling error rate alongside falling reporting is not improvement; it is a blind spot.
Balancing measures:
- Medication administration time per nurse, to detect whether the fix has created an impossible workload that will regenerate workarounds.
- Overtime and night-shift turnover on 5 West.
The intervention has failed if the override rate falls only because staff have learned to enter a different reason code, or if reporting volume stays depressed while error counts drop.