Reduce patient waiting time in a small clinic

Four things, in this order. Measure your own mean consult length and count arrivals per hour, because if arrivals exceed what the room can serve, nothing the front desk does will save the session. Close the gap between when the desk starts registering and when the doctor actually starts. Make the serving order visible, and hold an early arrival only while somebody who waited longer is still ahead. Then put up a screen, knowing it buys a calmer room, not a shorter wait.

First the arithmetic, because some sessions are impossible

Linda Green's queueing chapter states the rule that governs every waiting room: the higher the average utilization, the longer the waits, and the relationship is not linear. Delay rises at an increasing rate, there is an elbow past which small increases in load produce large increases in delay, and average delay approaches infinity as utilization approaches one. Unless average utilization stays strictly below 100 percent the system is unstable and the queue keeps growing all session.

The capacity check comes first, and Green's own worked example shows it is not a bare division. An emergency room taking 10 patients an hour, where each provider treats 2 an hour, needs a minimum of 6 providers, not the 5 the ratio suggests. Run the same check on your evening session using your own measured mean. If that mean is 8 minutes, two hours of consulting has room for about 15 patients at full stretch, and Green's next line is the one that matters: a system held at 100 percent utilization is unstable, so the number you can actually see without the queue growing all evening sits below that ceiling. When more people than that walk in, the overflow is arithmetic, not a front desk failure.

Two more results from the same chapter explain why hospital advice does not transfer. The smaller the system, the longer the delays at any given utilization, because queues have economies of scale: a large hospital can run at a higher utilization than a small one and still hold similar delays, and one consult room cannot. And the greater the variability in service time, the longer the delays at any given utilization, so a clinic seeing a broad mix of illnesses waits longer than a clinic of the same size, at the same doctor utilization, doing one narrow procedure.

Measure your own mean rather than borrowing one. A review of 179 studies covering 28,570,712 consultations in 67 countries found average primary care consultation length ranging from 48 seconds in Bangladesh to 22.5 minutes in Sweden, with 18 countries holding about half the world's population at 5 minutes or less. Its Indian entries are 1.9 minutes (1979, stopwatch), 2.3 minutes (2013, stopwatch), 2 minutes (2015) and 1.5 minutes (2005), mostly public sector measurements, so do not assume them for your own room. Time what Bailey called gross consultation time: all the time a patient claims the doctor's attention or otherwise prevents the doctor from seeing the next patient, which takes in the writing you finish after the patient has left, because that stops the next one going in too.

What the measured Indian numbers say

Numbers on this topic circulate in India without sources: the two pages ranking for this query that we checked publish about fourteen figures between them and cite a study for none of them. The figures below each have a source, and each carries the setting it was measured in, because not one of them was measured in a small private clinic in a tier 2 city.

In every one of these that split the visit into waiting and work, waiting was the larger half: 17 minutes 25 seconds of registration against 6 minutes 2 seconds of counter work, 38 minutes of a 66 minute Monday, 9 of the 10 minutes a counter registration took. And patients notice. In nationally representative data on older adults in India, waiting time was the aspect of an outpatient visit most often rated negatively of the six patient experience domains measured.

Two changes that move the clock before any software

Start time first. A simulation built on recorded arrivals, service times and flow for about 375 patients at an orthopedic clinic of a Tehran teaching hospital measured a base case wait of 55.36 minutes, with arrivals a mean 4.2 minutes apart. Changing the physicians' work time reduced waiting more than increasing the number of physicians did, and doing that together with a later start to patient admission cut waiting by about 71 percent, a figure the same paper's abstract puts at about 73 percent. The authors attribute much of the base case delay to the gap between admission opening and the physicians actually arriving. The scheduling literature treats a doctor's lateness the same way, as a modelled input: a late start puts its delay into every patient in that session.

Second, find the peak in your own day book rather than assuming somebody else's. Arrivals clustered into the morning at the North India emergency OPD above, about 26.3 percent of them between 9 and 12, and the anti rabies OPD's longest day was Monday, 66 minutes against 53 later in the week. Neither of those is your clinic, so count your own arrivals by hour and by weekday for a fortnight first. Whatever you can then shift off your own peak, a report collection, a repeat visit whose paperwork is already done, is worth more than the same load shifted off a quiet hour, because delay rises at an increasing rate rather than in step with load.

Tokens, appointments, and the patient who arrives two hours early

Maister warns that appointment systems are troublesome in practice: people book and do not turn up, and the spacing is a trap, since slots too far apart leave the doctor idle while slots too close together make the session run cumulatively late, so the doctor falls further and further behind. If you do book slots, the oldest usable rule is Bailey and Welch, 1952: for one doctor, 2 patients at the start of the session and 1 per interval after that, with the interval set equal to the mean service duration. Later work found it still performs well once no shows and imperfect punctuality are added to the model.

Fairness decides what you do with a 4 pm booking who walks in at 2 pm. Maister quotes Sasser, Olsen and Wyckoff: one of the most frequent irritants customers mention is the prior seating of those who arrived later, and the feeling that somebody has cut in front turns even the most patient customer furious, so great care to be equitable is vital. Where a queue has no visible order, each person stays nervous about whether their place is being kept, which is why many facilities issue numbers, serve in strict numerical order, and display the number being served so people can estimate their own wait. A priority rule is tolerated only when people can see why it is fair. So an early arrival should displace nobody who has already waited, and should not leave the doctor sitting idle either.

The other half of Maister is expectation. A patient told the doctor is delayed thirty minutes feels a moment of annoyance and then settles into accepting it; a patient told the doctor will be free soon spends the whole time in nervous anticipation, afraid to step out and come back. Once an appointment time has passed, even a ten minute wait grows steadily more annoying, because the wait no longer has a knowable limit. Announce a number, not a reassurance.

The display board fixes the room, not the clock

One Indian intervention here has been measured properly. A quality improvement project in a high volume internal medicine OPD of a tertiary public hospital in South India put televisions in the waiting areas showing curated family friendly content and short health awareness videos in the local language. Reported verbal altercations fell from a mean of 39 per week to 7.25, about 81 percent, over a period when weekly attendance was essentially unchanged, 2,228 patients against 2,253. Average waiting times remained unchanged. That is the honest case for a screen: fewer arguments at the desk, the same minutes on the clock.

What nobody can tell you, us included

• No measured before and after study of a token or queue system in a small Indian private clinic of 1 to 4 doctors was found in the open literature, so no minute saving can honestly be attributed to running one.
• No credible India specific outpatient no show rate was found, so the case against booking every slot rests on the scheduling literature and on Maister, not on an Indian percentage.
• Two pages ranking for this query, from Adrine and DocTrue, publish about fourteen numbers between them, including a fall from 45 minutes to 12 and a 50 percent cut in perceived wait, and neither cites a study or a URL for any of them. They are quoted here as claims those pages make, and should not be read as measurements of anything.
• Elaior has zero paying customers, so this page carries no before and after number of its own.

How Elaior's queue is built, stated as mechanism

The rank order is fixed: called first, then critical, then an appointment whose slot has arrived, then walk ins, then a parked early appointment, then a booking that has not arrived, then a patient who stepped out. An appointment keeps its slot priority for 15 minutes past the slot. The board, the doctor's list and the server side gate on calling the next patient read one comparator, so all three agree on who is next.

An early arrival is parked, which is what stops a 4 pm booking arriving at 2 pm from displacing people who have sat there since noon. The parking is conditional: they are held only while the lane holds somebody ahead of them, in the room, at the door, waiting their turn, or another early arrival with a sooner slot, and they become an ordinary waiting patient the moment it does not, so no doctor sits idle while a patient sits outside waiting on a clock. A walk in who registers in the meantime parks them again. The code is queuerank.HeldForSlot, with a verbatim twin in the web client.

The waiting room display makes the token the hero, with each doctor's speciality, a derived state of in consult, on a break or ready, next token chips, a live countdown between patients, and a wait estimate from a rolling 30 day average per doctor. The patient portal shows the same queue as a live card with the number of people ahead of you.