Twelve Weeks, Two Hundred Pumps

The critical path runs through the drug library and the interface version, not through the pumps.

Top risks, ranked by consequence:

1. Drug library not validated - a wrong concentration or dose limit at go-live is a direct patient harm pathway and the single largest risk in this project. 340 entries rebuilt by an unassigned pharmacy team is the critical path.
2. Interface validated against the wrong EHR version - the driver is qualified at 2021.1 and production will be 2022.3 within the project window. Go-live into a version that is about to change requires revalidation, and that work is not in anyone's plan.
3. Go-live two weeks before survey - surveyors will see half-trained staff on unfamiliar pumps, and a survey finding is far harder to recover from than a delayed metric.
4. Training capacity - three vendor trainer weeks against 11 units, 22 super-users, and multiple shifts is probably insufficient; evening and night staff are usually the ones left untrained.
5. Payment milestone decoupled from readiness - a 15% milestone tied to system acceptance creates pressure to declare acceptance early.

Sequence: library build → independent pharmacy double-check → interface test in a non-production environment against the version that will be live at go-live → single-unit pilot with the vendor's driver → super-user training and competency verification → staged go-live → 30-day post-live surveillance.

Go/no-go criteria, agreed before the meeting:

• All 340 entries validated with a second sign-off by someone who did not build them.
• End-to-end interface test passed with patient-to-pump association verified using a test patient, on the production EHR version.
• 90% of each unit's nurses competency-checked, including night shift.
• Rollback plan tested: pumps can revert to manual library mode within 30 minutes.
• Downtime procedure written and rehearsed at least once.

With the CNO, do not argue against the metric - argue for the metric.

• Reframe in her terms. A failed go-live damages the medication safety metric she wants to improve, and does it in the survey window. Present the two dates as competing risks with consequences, not as your preference versus her deadline.
• Offer a staged option rather than a yes or no. If the go/no-go criteria are met, 3 to 4 units can go live at week 10 with extra super-user coverage, and the remainder go live after the survey. This gives her movement on the metric without exposing all 11 units at once.
• Make the decision explicit and documented. Bring the recommendation to the steering committee with dates and criteria; ask the CNO to either accept the staged plan or record her direction to proceed on the original date. A recorded executive decision is not bureaucracy - it is what lets you run the project safely and what protects everyone if it goes wrong.
• Confirm the payment milestone does not depend on the calendar. Have procurement confirm in writing that system acceptance means verified acceptance criteria, not a date.

With the refusing unit:

• Understand the constraint first. Grant funding may specify the device, the study protocol may require exact pump behavior, and the investigators may have a contractual obligation. That is a legitimate constraint, not obstruction.
• Do not let the unit block the project. Negotiate a documented temporary exception: the unit stays on the current pumps with a defined review date and a commitment to convert before the old fleet's support ends. If the grant-funded pumps create a second drug library, pharmacy needs to know now, because that is a maintenance burden and a safety comparison risk.
• Escalate the exception to the steering committee with the cost and risk of maintaining a parallel fleet. Exceptions that nobody approves become permanent.

Decide now, not in week 10, and write it down.

• Pre-agree the trigger. If validation is not complete and independently double-checked by the stated date, go-live does not proceed. The trigger is objective, so the conversation is about facts rather than nerve.
• Escalate immediately to the steering committee with three options and a recommendation: (a) delay go-live - cleanest, costs the metric and possibly the milestone; (b) go live on a reduced library, limited to a small number of high-use infusions with the remainder retained on the existing pumps, and expand later - viable if the interface and training are ready and the clinical leadership accepts the reduced formulary on those units; (c) go live on the units whose library subset is complete, holding the rest.
• Recommend (b) or (c) only if the interface and training criteria are genuinely met. Splitting the go-live multiplies the coordination work; it is not free.
• Never go live with an unvalidated library. Pumps with a partially built library run on manual mode, which looks like success and removes the safety benefit you are buying. Clinicians will not notice the difference until an error occurs.
• Manage the milestone. Notify the vendor and finance in writing that acceptance criteria are not met. Do not let the payment date drive the clinical decision; the contract should be read as acceptance on verified criteria, and if it is ambiguous, that is a legal question to raise now.
• Communication plan: a single message to nursing, pharmacy, and the CNO on the same day, with the new date if known and the reason stated plainly. Silence during a slip generates more damage than the slip.
• Capture the lesson for the next project: pharmacy staffing for library builds is a named, resourced task with a named owner from day one, not an assumption.