top of page

ThinkWicker

Stored Delay

Writer: Marcus Hale
Marcus Hale
21 minutes ago
6 min read

A long wait for an appointment looks like proof of unmet demand, but much of it is backlog that generates its own extra work, and a schedule booked to the last slot keeps that backlog in place.


A health center reports that the next routine appointment is four weeks out. The leadership team reads the number in one of two ways. Either demand has outgrown the providers, so the answer is recruitment, or the center is attracting more patients than it can serve, so outreach should slow down. Both readings treat the wait as a measure of demand. Much of it is stored delay: backlog the system carries forward every day, which generates work of its own.


The paradox is that the wait can persist even when daily capacity roughly matches daily demand. A practice that sees about as many patients as ask to be seen each day can still have a month-long wait, because it is working through last month's requests while today's join the end of the line.


This matters for marketing decisions as much as for operations. A health center that mistakes stored delay for excess demand will cut outreach it needs, and one that mistakes it for a shortage will spend on recruitment before fixing the queue.



Where the idea of stored delay comes from


Mark Murray, a family physician, and Catherine Tantau worked in primary care at Kaiser Permanente in northern California, where the average wait for an appointment had reached 55 days. Their 2000 article in Family Practice Management, published by the American Academy of Family Physicians, described an approach they called advanced access, also known as open access or same-day scheduling: work down the existing backlog once, then follow "one very simple yet challenging rule: Do today's work today."


The Agency for Healthcare Research and Quality later summarized the approach in its CAHPS improvement guide. It reports that the wait for routine appointments at Kaiser Permanente in Roseville, California, fell from 55 days to one day in less than a year, and that the share of patients seeing their own physician rose from 47 percent to 80 percent. At the Alaska Native Medical Center, routine waits in family medicine and pediatrics fell from 30 days to one day.


Murray and Donald Berwick set out the method in JAMA in 2003, arguing that delays often stem from unplanned scheduling and poor resource allocation rather than unavoidable resource limits.


AHRQ is careful about what the evidence shows. The guide says open access "has not yet been formally evaluated with systematic controlled studies" and that the support for it is anecdotal. It also states that "no scheduling system works effectively if demand is greater than capacity on a permanent basis," and it names the main barriers as psychological rather than logistical, since clinicians and staff often doubt that existing resources can meet demand.


A wait list measures how much yesterday the system is still carrying, which can be very different from how much demand exists today.


How a backlog creates its own demand


Stored delay grows because a backlog generates extra work. Four mechanisms account for most of it.


  • No-shows rise with lead time. A patient booked four weeks out is more likely to have recovered, found care elsewhere or forgotten. Each empty slot is capacity lost while the queue stays full.

  • Calls multiply. Patients who can't get through on the callback ask to be added to the cancellation list and check on their place. Front desk time goes to managing the queue instead of serving patients.

  • Problems get handled twice. A patient who can't see their own provider is seen elsewhere, sometimes in the emergency department, then still needs the original visit.

  • Continuity breaks. When the next open slot is with whoever happens to be free, visits get shorter and less complete, and follow-up visits multiply.


Each of these consumes capacity the practice could have used to clear the backlog. The queue protects itself.



Why a full schedule makes it worse


Leaders often take a fully booked schedule as a sign of good management. Queueing research points the other way. A 1999 simulation study by Adrian Bagust, Michael Place and John Posnett in the BMJ simulated bed use in a hypothetical English acute hospital facing variable emergency admissions. It reported that "risks are discernible when average bed occupancy rates exceed about 85%," and that a hospital operating at 90% or more could expect regular bed shortages and periodic bed crises.


The reason is variability. Patient requests arrive unevenly. A system with no slack has nowhere to put a busy Monday, so the overflow becomes tomorrow's backlog, and the backlog becomes next month's wait. Spare capacity that appears as waste on a utilization report absorbs surges.


A primary care schedule is a different system from a hospital ward, and the 85 percent figure should not be carried over as a target. The shape of the curve is the lesson. Waits grow faster than utilization as a schedule approaches full.



Stored delay or true shortage


The model separates two conditions that produce the same wait.

Indicator

Stored delay

True shortage

Daily requests vs. daily visits

Roughly equal

Requests consistently exceed available visits

Wait times

Stable or slowly growing

Rising steadily

Work created by the queue

Higher: repeat calls and rescheduled missed appointments

Lower: more of the demand reflects new appointment needs

What fixes it

A one-time effort to clear the backlog, followed by planned spare capacity

More providers, expanded hours or additional sites


The first condition can be fixed with a defined project. The second needs investment. Treating one as the other wastes the effort.


Michael Lipsky's work on frontline public services explains why the difference often goes unnoticed. In Street-Level Bureaucracy, he described how workers facing more demand than they can meet adopt practices such as rationing resources and routinizing their interactions with clients. In a clinic, those routines include triage rules, callback lists and unofficial ways of fitting patients in. They keep the day running. They also hide the backlog from anyone reading the summary report.


When the queue creates its own work, adding demand and adding providers both miss the cause.


The measurement problem


Most health centers track the third available appointment, which tells leaders how long the wait is. It does not tell them why. Stored delay shows up in three numbers few centers put side by side.


  1. Daily requests against daily completed visits for the same provider team, over at least a month. If they are close, the backlog is stored delay.

  2. No-show rate by lead time. If visits booked more than two weeks out miss far more often than visits booked within a few days, the queue is producing its own empty slots.

  3. Queue-generated work. Count calls about existing appointments, rebooked no-shows and visits that repeat an unresolved problem. This is capacity spent on the backlog.


Health centers with many uninsured patients and limited provider supply may find an underlying shortage. The point of measuring is to find out which problem they have.



The decision stored delay puts in front of leaders


Marketing, outreach and recruitment decisions all rest on a reading of demand. A long wait is the number most leaders use for that reading, and stored delay distorts it.


The decision comes before any campaign or hiring plan. Leaders have to know whether the wait is a storage delay or a true shortage, and they can find out from the requests, no-shows and queue work they already record. If it is a storage delay, the first investment is a one-time effort to clear the backlog and a schedule that keeps some slack. Demand work comes after.


Before deciding whether to grow demand or add providers, find out how much of the wait the system is making for itself.


References



Some ideas are worth discussing in the context of your organization.



 
 

Need more than insights?

bottom of page