There is work running in your program this week whose cost you cannot see separately. Your staff are doing it, and your program is paying for their time.

You can put a number on it this quarter. Two of the records are already in the building: the OIS and the staffing schedule. The third is the service record, and only half of it already exists. None of the three gives you hours. The hours come from the people doing the work.

Heather Turner ran the floor-level version of this on this list in June. The fifteen-minute gap an administrator reads as open capacity, already consumed by a transport delay, a change of clothes, a bladder that was not ready. She priced one slice of it: two therapists thirty minutes late on one machine, at around $73 an hour in overtime, roughly $18,000 a year. That was the therapist's hour, in her voice. This is the program's ledger in mine.

Start with where the cost lands, because that decides what kind of number you are building. Some of it is absorbed by people already on payroll, inside hours you have already paid for. Some turns into additional spending: overtime, a differential, an agency invoice. And some displaces other work that did not get done. Those are three different things and they do not add up. Keep the value of existing staff time separate from additional spending. Record displaced work alongside them without counting the same hours again.

Where a cost lands tells you where to start looking, not why it is there: salary absorption might be a capacity problem or a workflow problem, and overtime might be scheduling or a vacancy nobody has posted. Which of the three it landed in picks the question you ask first.

The OIS gives you the events. It is already an event ledger, with a timestamp and an owner on nearly everything that happens to a patient between consult and first fraction. Export a quarter and count the events behind the activities on your list. The export tells you how often, and who was attached.

The service record gives you the incidents. The vendor's portal holds the ticket history: when a fault was called in, and when the vendor logged a response. What it does not hold is when an engineer actually walked in and when the machine went back to treating. Those two are yours to write down, which is what the June issue asked you to start: a shared sheet with four columns. If you started it in June you already have a quarter of it. If not, start it this week. Your agreement defines how response time and uptime are measured, so these timestamps reconstruct the incident rather than price it. Ask the staff involved what work it created and when. Rescheduling patients. Calling them back. Replanning around a machine. Standing at a console waiting on an answer. Those definitions are written by people whose job is to define a reportable obligation. I did that job, and they are not dishonest. What the quarter cost you is a separate number, and nobody on their side was ever asked to produce it.

The schedule gives you the roles. The roster says who was on, in what role, and against what template the day was built. Cross it against the OIS events and the service incidents and you know which roles stood in the path of each activity. That is as far as the records reach.

Use the records to identify the events and the roles involved, then have the people doing the work validate an average duration for each activity, and count each person's time once. A dosimetrist and a physicist sitting with the same replan is one event and two people, not two events. That validation step is the difference between a number the department will stand behind and a number somebody built in a spreadsheet.

For the rate, go to your own finance office rather than to a survey. They already carry a loaded hourly cost for every role in the building: salary, benefits, taxes, the whole number. The minute a survey number goes on the page, the conversation is about the survey number. Your own loaded rate does not give anybody that exit.

Then the arithmetic is one line. Event count, times average staff time by role, times the loaded hourly rate for that role. Per activity, in three columns, and anything you annualize gets labeled an estimate, because that is what it is.

One thing this changes about how a program reads an automation claim. When a step gets automated, the hours usually do not leave the building. They move, and they land on somebody whose time was never counted, which is why it looks like a saving. So before you accept that a step got cheaper, find out where the rest of it went.

None of this produces a bill. Nobody is going to pay it. It produces one page: the activities, how often each happened, whose time it took, the rate, and which column it landed in. That page is useful in two conversations you are already having. Staffing, where "we are short" becomes an estimate with a method attached instead of a feeling. And service, where the vendor's figure and yours describe the same events from two sides.

This issue's move, and it starts this week. Pick one activity. Re-simulation is the usual place to begin: the OIS already counts them, and everybody has an opinion about them. Then ask the person doing the work how long one takes, start to finish, including the part before they open the chart. One export and one conversation. If the answer surprises anybody, the quarter is worth spending.

Weeks two to four, write the list. Five to eight activities that consume clinician or staff time in your building and whose cost you cannot currently see on its own. Re-simulations. Plan revisions after approval. Peer review preparation. Payer documentation cycles and peer-to-peer calls. Physicians and physicists pulled out of a scheduled task into the CT suite. Coverage during equipment downtime. Patient callbacks and reschedules. Keep it to the ones you can actually count, and put a named owner beside each one who can say where the record lives.

The rest of the quarter, count it and price it. Export the events, total the service incidents, pull the roster, then sit with each activity's owner and agree an average duration per role. That last step is the one nobody can export, and it is what makes the page hold up when the department reads it. Three columns, not one total, a source beside every number, an owner beside every gap. Then take it to whoever owns the staffing plan and ask which column they want first.

The OIS already counted those events. The sheet, if you kept one, has the incidents. The roster knows who was on. What none of them holds is how long it took, and the only people who have that are the ones doing it. Ask one of them this week.

Putting a number on work a program is already absorbing and has never costed is operational benchmarking, and it is work Medsolve does for programs.

On the response-time number.

The Pitch: "We responded within four hours. Uptime ran 97 percent for the quarter."

The Reality: Both figures can be accurate, and both are calculated under definitions written into your own coverage schedule. Take one incident from last quarter and walk it: the hour it was called in, the hour the portal recorded a response, the hour an engineer was on site, the hour it closed. Then ask the therapists who were on what they did during that incident. That is one row of the log and one conversation, and it is the same exercise this issue runs over a quarter.

With Heather Turner, RT(T), PMP

The Simulation Pipeline Leak: Why your treatment machine is waiting on your CT scanner.

Many operations teams focus heavily on linear accelerator metrics, tracking beam on time, table turnover, and slot utilization. But one of the biggest hidden drivers of unstable LINAC schedules lives further upstream in the simulation suite.

Let's be realistic: in radiation oncology, unexpected delays will happen. A patient has a hard time filling their bladder, a vac-lok takes longer to mold than expected, or a complex inpatient transfer arrives late. Clinical curveballs are part of the job, and that is okay. We will never eliminate those variables entirely. But that is precisely why we must aggressively control the operational items we can handle in advance. When you fix the predictable administrative and workflow delays ahead of time, your department gains the breathing room it needs to absorb those unavoidable clinical curveballs without the whole day falling apart.

When the simulation process is erratic or poorly coordinated, the damage doesn't stay in the CT room. It creates an immediate ripple effect that drags key personnel off the treatment floor when they are needed most. If a therapist has to call a physician to the simulation suite to verify an uncomfortable setup, review scan limits, or confirm a complex immobilization strategy before letting the patient off the table, that physician gets pulled away from signing off on a complex SBRT verification on the LINAC. If a physicist gets pulled into the CT room to troubleshoot a motion management tracking issue, they are forced to abandon their chart checks and routine QA tasks. The treatment therapists at the console are left standing around waiting for approvals, the waiting room backs up, and the schedule descends into delays.

If you want a calm, predictable treatment floor, you have to run your simulation suite with the exact same operational discipline as your primary treatment machines:

Standardize Simulation Slots by Anatomy: A straightforward prostate simulation takes a very different amount of time than a complex stereotactic case requiring custom immobilization, contrast administration, or surface tracking. If your CT schedule treats all simulation appointments as standard 30 minute blocks, you are building delays into the system before the patient ever sees a treatment machine.

Eliminate Unplanned Staff Pulls: Track your re-simulation rates and classify the root causes, whether it is positioning shifts, anatomical changes, or missing clinical documentation. Every time you eliminate an unnecessary re-sim, you keep your physicians at the treatment console for critical verifications and keep your physicists focused on chart approvals and QA workflows.

Synchronize Intake with Simulation: A patient should never sit on the CT table while staff wait for an updated physician order or an insurance authorization clarification. Ensure all pre-simulation requirements, including orders, consent, and initial authorization, are 100% complete before the patient enters the dressing room.

Your linear accelerator can only treat as smoothly as your simulation suite feeds it. By stopping leaks in the simulation pipeline, you protect your clinical resources, keep your team where they belong, and ensure your treatment machines run on time.

From Yoel. If you run the one-activity version this week and get stuck on which record holds it, reply and tell me where you are stuck. I read these myself and I answer them myself.

Yoel