Fairness · 10 min read · 2026-09-14

How to distribute on-call shifts fairly in a medical team.

I am not an FMG manager. I spend my weeks sitting beside the people who are, setting up clinics and looking at their spreadsheets. The on-call argument comes up in every team, and almost always in the same words: "it is always me who works the weekends." The teams where it finally stopped got there with a few lines written on one page, not with a better calculation.

By Kamel Gorieze

Co-founder · product and user experience

In short

One column per kind of call in your file: weekday, evening, weekend, statutory holiday. You check one line at a time instead of arguing about a total.
Shifts picked up as replacements count. That is where the person who always says yes hides, and it is the most common omission.
A written maximum, along the lines of "no more than one weekend in three". It can be checked without opening a file, and it protects whoever is having a bad year right away.
The holiday rotation written three years ahead, in June. The dates of the holiday period never move, so the order need not either.
The swap rule written once: who approves, who the shift counts for, and no swapping from one column to another.
With part-timers on the team, divide each column by the number of half-days worked before comparing anything.

Friday, 4:40 p.m.

The scene I see most often starts with a call late on a Friday afternoon. A physician will not be in on Monday. The manager hangs up and starts dialling.

The first one does not pick up. The second is on leave. The third hesitates, then says no. Twenty-five minutes later, the sixth person says yes. And in just about every clinic I visit, that sixth person is the same one every time.

That yes gets written down nowhere. Not in the schedule, which only shows who works Monday. Not in the shift count, because the shift was never hers to begin with. Nowhere, except in the head of the person who said yes.

Three months later she says she has had enough, and everyone is surprised. An unfairness nobody can put a number on is the one that lasts: the person on the receiving end has nothing to show, goes quiet, and eventually leaves. It is the most silent of the signals that show up in a grid before a departure.

"Everybody works twelve"

The first thing people hand me when I ask is the shift count: everybody works twelve, so it is fair. It is the only number a spreadsheet gives you without being asked, so it is the one that ends up being the argument.

Except twelve shifts means nothing until you know which ones. A Saturday shift takes away a whole weekend, a Tuesday shift takes away an evening, and a night shift takes away the next day too, because the clinic opens anyway.

Here are four physicians from one team over a year. The numbers are made up for the example, but the shape of the table is one I see everywhere.

Physician	Weekday, day	Evening or night	Weekend	Holiday	Total	Half-days per week
Dr. A	3	4	4	1	12	8
Dr. B	7	3	2	0	12	8
Dr. C	2	2	3	1	8	4
Dr. D	5	3	3	1	12	6

The "Total" column says three of the four did the same thing. The columns on the left say otherwise: Dr. A worked twice as many weekends as Dr. B. And Dr. C, who works half as much as Dr. A, worked almost as many weekends as she did. Sort by shift count and Dr. C comes out as the most spared of the four, when she is in fact the one giving up the most.

Add the Friday-afternoon replacements, which appear in none of these columns, and you have the table people bring to a meeting to prove everything is fine.

What teams where it stopped actually do

Faced with a table like that, almost everyone proposes the same fix: give each kind of call a value and balance a total. It starts from a good idea. In practice the meeting turns into a debate about the value of a Saturday, runs for three sessions, and settles nothing.

One column per kind of call

Not a total, columns: weekends on one line, nights on another, statutory holidays on a third, and you check the balance one line at a time. It is cruder than a well-built points system, and it works better.

A weekend column can be checked in ten seconds. Three is three, and the discussion is over. The value of a Saturday cannot be checked at all: it gets negotiated, and there is no right answer waiting at the end, only one people eventually accept grudgingly.

Columns also fix the Friday 4:40 p.m. problem. A replacement goes in the column of the shift it replaced, at the moment the person accepts, and the yes stops being invisible. It is one more line in the file, and it changes what the team sees about itself.

A maximum, not an average

The rules that actually exist inside teams are maximums. No more than one call a week. No more than one weekend in three. Never a clinic day after a night. Short sentences anyone can check without opening a file.

And there is nothing above you that will set it for you. The law says nothing about on-call frequency, according to the Fédération des médecins spécialistes du Québec, and the duty to draw the roster up equitably falls on whoever draws it up. If your team does not write its maximum down, it does not have one, and the real limit becomes whatever the most accommodating person is willing to put up with.

A maximum also does something an annual average never will: it protects whoever is having a bad year right away. The person who said yes six times in a row in March does not need a promise about December.

Christmas gets settled in June

Christmas does not balance inside a year. There is only one of it, and it lands on one person. The teams that solved it solved it between years: whoever worked it this year does not work it next year, the order is written down, and it is published several years ahead.

The dates never move, which makes it easy. The Programme de financement et de soutien professionnel pour les GMF sets the two holiday periods at 22 to 28 December and 29 December to 4 January, under section 4.6.1.1 of the version in force from 1 April 2026. A three-year rotation therefore takes one evening to write, and the month to write it is June. By November it is no longer an organizational discussion, it is a negotiation between people who already know what they want.

The swap rule, written once

When I dig into an on-call argument, I almost always find a swap argument underneath. Who got called first. Whether the swapped shift counts for the person giving it up or the person taking it. Whether you are allowed to trade a weekend for a Tuesday. Three questions to settle once, ten minutes to write down, and the answer to the last one is usually no, otherwise you have just rebuilt the exchange rate the columns were there to remove.

The page that comes out of the meeting

On-call rules, team of 14 physicians, reviewed June 2026.

Columns tracked: weekday, evening or night, weekend, statutory holiday. A replacement counts in the column of the shift it replaced.
Maximums: one weekend in three, never two calls in the same week, no clinic the day after a night.
Holidays: 2026, Christmas to A and New Year to B. 2027, B and C. 2028, C and D.
Swaps: approved by the manager, the shift counts for whoever works it, no swapping from one column to another.
Part-time: each column scaled to the number of half-days before comparison.
Review: in June, by the manager and the responsible physician.

Part-time changes the ranking

A family medicine team is almost never four people working full time. You have someone at four half-days, someone at six, someone coming back gradually from a leave. Comparing raw totals between those people means nothing.

Go back to the table above. Dr. C works four half-days a week and three weekends a year. Dr. A works eight and four weekends. For the time she works, Dr. C is doing one and a half times what her colleague does, and she is the one with the smallest shift count in the table.

The fix costs one column. If nobody has done it where you are, it is almost certainly because everyone's practice time is written in a different file from the schedule.

Points, yes, but afterwards

I am not saying points systems are worthless. They measure unfairness across several years, once a team gets too big for the columns to be read at a glance. And there are numbers on this, not just opinions.

An American neurosurgery program kept its own long enough to be judged: 94 residents tracked over 22 years, before and after implementation. The spread in load between them fell and duty-hour violations went from 34 % to 11.1 %, according to The American Journal of Surgery in July 2026.

Another team measured what automating schedule generation does, in Neurosurgery in June 2024: variation between calls fell by 70 %, and the share of residents who found the schedule fair went from 43 % to 95 %. The schedule became more equal, and the sense of fairness moved about twice as fast as the measurement did. People could finally see the numbers, and that counted as much as the numbers.

One caution before you quote that in a meeting: both studies cover neurosurgery residents, not an FMG. A resident cannot refuse call and their hours are regulated, which is not the situation of a physician in your group.

We also checked how a points system goes wrong, with a small optimization model: twelve physicians, an instruction to even out the totals, and the best possible answer gives four shifts to two physicians and a single one to a third. The arithmetic is perfectly correct and the result is impossible to publish. It took adding a per-person maximum to get back to a normal schedule. The code is in our article on the subject.

So use the columns and the maximums to decide, and points to measure, if you want them.

Two things that snag anyway

The one who wants the weekend. There is always someone who prefers Saturday, for daycare or just because it suits them. Count their shift at full value anyway. A discounted shift because it was welcome is the best way to discourage the one person doing you a favour, and you will need them next year.
Incomplete years. The physician who arrives in September does not owe a full year, the one who leaves in March leaves columns nobody knows what to do with, and the one coming back from a long absence wonders whether to catch up. Decide all three calmly, in advance, and write them on the same page as the rest. That is so many fewer negotiations when the day comes.

The one-hour meeting

All of this gets decided in a single meeting. Here is the agenda.

List your kinds of call. Three or four, no more. Weekday, evening or night, weekend, statutory holiday. Past that your columns become unreadable.
Pull the last twelve months, one column per kind. You already have the information, it is in your old schedules. Count the replacements along with the rest, or you start again with the same blind spot.
Write down your maximums. No more than one weekend in three. Never a clinic day after a night. Sentences anyone can check without opening a file.
Write the holiday rotation for three years. The dates never move, so the order can be written ahead. Do it in June, while the subject interests nobody.
Write the swap rule. Who approves, who the shift counts for, and whether you can swap across two columns. The answer to the last one is usually no.
Put part-timers on the same footing. Divide each column by the number of half-days worked before comparing anything. The ranking will change.
Date the page and name who reviews it. An undated rule gets renegotiated at every disagreement. A dated, signed one gets reread.

What comes out of that meeting is worth more than the software that will apply it. A team that has agreed on its columns and its maximums tracks this in a spreadsheet without trouble. No software, ours included, is going to produce that agreement for you.

Where Synchro fits in

Synchro applies the rules the group has written instead of imposing its own. Kinds of call and maximums are configured per team, everyone's columns stay visible to the whole team with their date, and a shift that comes free is offered to every eligible physician at once, with a record of who accepted and when. That last part is what fixes the Friday 4:40 p.m. from the start: the yes is recorded the moment it is given. Pricing is published, $10 CAD per user per month billed annually, first month free, and our on-call scheduling page covers the rest.

The limits, while we are being clear. Everything I just described can be done in a spreadsheet, and plenty of teams do it very well. What eventually gives way there is the update after every swap. Synchro holds no patient data, does not replace your EMR and does not calculate your attendance rate. Our figures are self-reported and come from a single pilot FMG. And no software is going to decide for you how many weekends a year is too many.

Frequently asked questions

How do you distribute on-call shifts fairly in a medical team?

By counting each kind of call in its own column instead of computing one total. A column for weekends, one for nights, one for statutory holidays, and you check the balance one line at a time. Add a written maximum, for instance no more than one weekend in three, and a holiday rotation that runs from one year to the next. Those rules can be checked without anyone having to agree on what a Saturday is worth, and that is exactly the discussion that never concludes.

Do last-minute replacements count toward fairness?

Yes, and this is the most common omission. A shift picked up as a favour on a Friday afternoon rarely makes it into the annual count, because it was not on that person's schedule to begin with. The result is that whoever agrees to help out most often is also the one who looks least loaded on paper. Record the replacement in the column of the shift being replaced, at the moment it is accepted.

Do you need a points system for on-call shifts?

It measures rather than prevents, and it comes after the columns. Points systems work when a team keeps them up for a long time: an American neurosurgery program tracked 94 residents over 22 years with its own and brought duty-hour violations down from 34 % to 11.1 %, according to The American Journal of Surgery in July 2026. The trouble with a single total is that it lets the kinds of call trade against each other: four weekday shifts are worth as much as two weekend shifts, so two people can show the same number when only one of them lost their weekends.

Is there a rule on on-call frequency in Quebec?

No. The law says nothing about on-call frequency, according to the Fédération des médecins spécialistes du Québec, and the duty to draw the roster up equitably falls on whoever draws it up. Nobody above you is going to set your maximum for you. If your team does not write its own, it does not have one, and the real limit becomes whatever the most accommodating person is willing to put up with.

How do you account for part-time physicians?

Put everyone on the same footing before comparing. A physician working four half-days a week with three weekends a year is doing twice as much, for the time they work, as a colleague at eight half-days who also works three. Without that correction, the most loaded person on your team can look like the most spared one, because they have the smallest shift count overall.

Is a spreadsheet enough to track this?

For the arithmetic, yes. One column per kind of call, one division for part-time, and you have what you need. What eventually gives way in a spreadsheet is the update after every swap and every last-minute replacement. Columns that are right on publication day and wrong three weeks later are worth less than nothing, because they give a real unfairness the look of a settled file. If your columns are current after every change, your file does the job.

Public sources cited: Fédération des médecins spécialistes du Québec, "Fréquence de garde" page, consulted 2026-09-14, for the silence of the law on on-call frequency; Adegbenro TM, Bangash AH, Fluss R et al., "The resident call points system: a quantitative framework for equitable workload distribution in neurosurgical training", The American Journal of Surgery, vol. 257, July 2026, article 116996, for the 94 residents tracked over 22 years and duty-hour violations falling from 34 % to 11.1 %; Porche K, Mohan A, Dow J et al., "Automated and optimized neurosurgery scheduling system improves resident satisfaction", Neurosurgery, vol. 94, no. 6, June 2024, pp. 1273-1281, for the 70 % reduction in variation and perceived fairness rising from 43 % to 95 %. Both studies cover neurosurgery residency programs, not FMGs. Programme de financement et de soutien professionnel pour les groupes de médecine de famille, MSSS, in force from 1 April 2026, section 4.6.1.1, for the dates of the two holiday periods. The table of four physicians, the page of rules and the scenes described are examples built to illustrate the point, drawn from what we see in clinics, and not identifiable real cases. The optimization results come from our own runs with OR-Tools 9.15 on 2026-09-08. Product names belong to their respective owners and this site is not affiliated with any of them.

About the author

Kamel Gorieze