Nothing in this story happened. It is a composite assembled from the way point-of-administration diversion is typically found in an intensive care unit, written so that the sequence of signals is visible.

The setting

A medical intensive care unit in a mid-sized community hospital. Twelve beds, two automated dispensing cabinets, a night shift that runs with four nurses and a charge nurse. Hydromorphone injection is the unit's most-used opioid, drawn from the cabinet in single-use vials, with a standing policy that partial doses are wasted with a witness at the cabinet within thirty minutes of removal.

The first signal

The signal was not a shortage. The cabinets balanced every shift. It was a complaint. A patient's daughter told the day charge nurse that her mother had been “in agony all night” and that the night nurse “kept saying she'd already given something.” The charge nurse looked at the record: two doses of hydromorphone removed overnight for that patient, both charted as given, both followed by a documented pain score that had not changed. She filed the complaint under the unit's patient-experience process and, because a recent in-service had asked staff to do so, also sent a note to the pharmacy's diversion lead.

The look

The diversion lead, a pharmacist who spends part of each week on surveillance, ran the monthly cabinet report early. For the previous eight weeks she compared each night nurse on the unit on three measures: hydromorphone removals per shift worked, the share of removals followed by a documented waste, and the time between removal and waste. One nurse stood apart. His removals per shift were higher than his peers', which on its own meant little; he often took the sickest assignment. But nearly every removal was followed by a waste, and the waste was documented almost immediately: two minutes, three minutes, occasionally under a minute after the removal. His peers' wastes were scattered across the thirty-minute window and were far less frequent.

Two further things stood out. The witness on his wastes was, more often than chance would suggest, the same colleague, a newer nurse who worked the same rotation. And his patients' pain scores, charted after each dose, moved less than other nurses' patients' scores did.

The confirmation

None of that was proof, and the diversion lead treated it as a reason to look harder rather than a conclusion. She asked security to preserve the cabinet-area camera footage for the previous two weeks before it was overwritten. She pulled the medication administration record for every patient the nurse had been assigned in the window and lined up, for each hydromorphone removal, the order, the charted administration time, the waste record, the witness, and the pain score before and after. In a real dose that chain is consistent. In roughly a third of his, it was not: the waste was documented before the charted administration time, which is physically backwards.

The footage showed the pattern the data had suggested. The nurse would remove a vial, walk to the medication room rather than the bedside, and return to the cabinet with his witness to document a waste. The witness, interviewed later, said she had been asked to “just confirm” on the screen while she was charting and had not watched the syringe emptied. She had not thought of it as a problem.

Pharmacy also retrieved the sealed sharps container from the cabinet area. It was sent, chain of custody intact, for testing of the wasted syringes recovered from it. That result came later and confirmed what the other evidence already showed: the syringes contained less drug than the waste records claimed.

What the hospital did

On the day the record review was complete, the diversion lead, the pharmacy director and the nursing director met. The nurse was removed from patient care pending investigation. The pharmacy director made the written notification to the local DEA office within one business day of that meeting, treating the discovery date as the date the review established a loss, and filed the Form 106 within the 45-day window once the quantity could be estimated from the records. The loss was reported to the chief executive under the hospital's policy. Patient safety reviewed every patient the nurse had cared for during the window, with particular attention to pain management and to any procedure where a tampered syringe could have been used. The nursing board report followed the state's rules. The interview with the nurse came after all of this, with the records on the table.

What would have surfaced it sooner

  • Waste timing as a routine report. Waste documented within a minute or two of removal, every time, is a recognizable pattern. It was in the data for weeks before anyone ran the report.
  • Witness pairing review. The same two names on most wastes is visible in a monthly pairing report. Here it was found only because someone was already looking.
  • Charted administration time versus waste time. A waste recorded before the dose was given is an impossibility the system can flag automatically.
  • Pain-score response by caregiver. Patients whose documented pain does not move after documented doses, concentrated under one nurse, is a clinical quality signal as much as a diversion signal.
  • Witness training. The witness had been taught to confirm a screen, not to watch a syringe. The blind witnessing guide covers the gap.

The complaint channel worked because a recent in-service had told the charge nurse where to send it. The data channel worked because someone ran a report early. Neither was luck, and neither was a system. The hospital made both routine afterwards.

This narrative is an illustrative composite. No real person, patient, facility or event is described.