Cycle Time Broken Into Segments

Total time in the building is one of the easier numbers to produce and one of the hardest to act on. The waits inside it belong to different people, and usually only one of them is the problem.

A total says there is a problem and not where it is

Cycle time, the clock from when a patient arrives to when they leave, goes on a dashboard, and when it grows everyone agrees the day is running badly. The number does not say which part of the visit grew. So the response tends to be general: move faster, room people sooner, keep the day moving.

Two visits with an identical total can be entirely different afternoons. One sat in the lobby, then moved through rooming, provider and checkout without a pause. The other was roomed on arrival and then waited alone with the door shut for nearly all of it. The dashboard cannot tell those apart, and the second is the one that produces a phone call the next day.

The break points sit where responsibility changes hands

The segments worth separating are not equal slices of the clock. They are the moments the visit changes hands: arrival to check-in complete, check-in to rooming, roomed to provider in, provider out to checkout done, and in practices with lab or imaging in the building, the detour for that. Each boundary is a queue, and a segment that grows is more often a line in front of somebody than that person working slowly.

Splitting it this way hands each piece an owner, which the total never had. Front desk, the medical assistant pool, the provider, checkout. A number belonging to all four belongs to none of them, and nobody has been told which part of the day is theirs to defend.

The timestamps already in the system record clicks

The practice management system produces most of these boundaries without being asked, and what it records is when somebody clicked something. Arrival gets marked as the front desk works through a batch rather than when the patient came through the door. Provider in is frequently the moment a note was opened, which might be in the hallway, or several minutes into the visit. The errors run in one direction per segment, so an overstated one keeps drawing attention it did not earn.

A short stretch of timing by hand settles which is which. Somebody writes the same boundaries on paper long enough to cover a Monday and a Friday, and the two versions get compared. The point is not the timing study. It is finding out which of the system's timestamps track reality, since those are the ones that can be watched afterwards without anyone standing in a hallway with a clipboard.

The longest segment and the worst one are often different

The segment that eats the most minutes and the one patients describe afterwards are frequently not the same. Waiting in a lobby with a phone, a window and a visible queue passes differently from waiting alone in an exam room where nothing is happening and there is nobody to ask. In what I have seen, complaints follow the second kind well past the point the minutes would justify.

So the measurement answers two questions. Which segment is long, which is a capacity and template question, and which segment is felt, which tends to respond to something cheaper: saying what is being waited for and roughly how long, so an unexplained gap becomes a known one.

Shortening one segment tends to move the wait, not remove it

Rooming faster is the usual first move, and it relocates the wait more often than it shortens the visit. If the provider is working at a fixed cadence, a patient roomed sooner waits in the room rather than the lobby, and a room is occupied while they do it. Where rooms rather than staff are the binding constraint, the next patient cannot be roomed at all, and the lobby line backs up into the schedule.

Which is the case for finding the constrained segment before improving any of them. A change aimed at the wrong one is not neutral. It spends a room and the patience of people asked to work faster on something that was not what was slow.

A day that falls behind is usually a template problem

Read by hour rather than as a daily average, the shape shows itself quickly. A day that starts behind is a different problem from a day that falls behind by eleven, and the second one is nearly always the template. A visit type whose real length runs past the slot it was given creates a debt that compounds through the morning. By noon it looks like a rooming failure. It was a decision about slot length that nobody has revisited.

Correcting that is a trade rather than a fix. Lengthening the slot takes appointments out of the day, which arrives as access and as revenue. Carrying the drift keeps the slot count and pays for it in the waiting room and in staff finishing late. Segment data does not settle which of those a practice would rather carry, only when the choice gets made: with the arithmetic in view, or at the end of a month of long days.