Building on-call rosters: rest, fairness and traceability

Of all schedules, on-call rosters are the hardest to hold together. Not because of volume: one on-call line means one person per period. The difficulty is that on-call duty produces effects you cannot see at the moment you plan it.

Someone on call is not at work, but neither are they free. If called out, their intervention time becomes actual working time, which shifts their rest, which invalidates the assignment planned for the next day. The schedule was correct when you wrote it; it becomes wrong overnight, and nobody notices before morning.

An example makes the mechanism obvious. A practitioner finishes at 6.30 pm and goes on call. They are called at 11 pm and work until 2 am. If eleven consecutive hours of rest must follow that intervention, they are not available before 1 pm the next day, while the schedule had them starting at 8.30 am. Nobody made a mistake: the next day’s assignment was decided weeks earlier, the call came at night, and the two pieces of information never met.

A current example: the 2026 overhaul in French public hospitals

A recent text shows how quickly this subject grows complicated. Instruction DGOS/RH5/2026/57 of 6 May 2026 overhauls the compensation regime for home on-call duty of medical, pharmaceutical and dental staff, together with teaching and hospital staff, in French public health establishments.

What it sets out:

• On-call flat rates are set by each establishment within a band running from 70 to 280 euros. A half on-call period is paid at half the rate, and each establishment must define at least three distinct rates.
• Classifying a line into one rate or another takes account of average activity intensity, call-out frequency, the share of telemedicine, and territorial or regional referral activity.
• For teaching and hospital staff, the rate may be increased since 1 May 2026, up to 500 euros for a night or two half-days. The increase follows call-out frequency and duration, decided by the director on the proposal of the care continuity committee.

One reading caution, because it can waste time: the text mentions the flat-rate system taking effect on 1 November 2025, a date earlier than its own signature, without explanation. If you work on this, check the version in force in the official bulletin rather than relying on a summary. Several online commentaries on this text carry obvious date errors.

On scope: the instruction covers medical and equivalent staff in public health establishments. It does not govern on-call duty for paramedical staff, the private sector, or emergency services, which fall under separate frameworks.

What the reform changes for whoever builds the roster

Beyond the amounts, the structural point lies elsewhere: line classification and the increase now rest explicitly on call-out frequency and duration.

Those data stop being comfort statistics and become the basis of a compensation calculation. If you cannot say, line by line, how many times practitioners were called out and for how long, you cannot properly build your rate schedule. The roster becomes supporting evidence.

It is a quiet but important shift. Many establishments run their rosters rigorously yet record call-outs only informally, when they record them at all.

On call, on site, on duty: a costly confusion

The terms get used as synonyms and they are not. In hospital settings the distinction has direct consequences for how working time is counted.

• On-site duty requires physical presence. It is effective working time from end to end, whether or not anything happened.
• On-call duty requires availability without presence. Only intervention time, travel included, counts as effective working time.
• Care continuity is the overall arrangement guaranteeing the service, made up of the two above.

Modelling on-call duty as on-site duty inflates the counters and distorts displayed workload. The opposite error is worse: it makes genuinely worked time disappear. If your tool does not separate the two, you will sooner or later make one error or the other.

Three constraints every on-call roster must hold

Rest. The principle of eleven consecutive hours of daily rest structures the whole thing, and the applicable regimes vary by status. What matters organisationally is that a night call-out has consequences for the following day’s schedule. A tool that does not connect the intervention to tomorrow’s assignment will let you produce invalid rosters in perfect good faith.

Fairness. An on-call line is rarely attractive, and distribution plays out over time. Weekend nights, public holidays and school holiday periods are the friction points. With no counter visible to everyone, each person keeps their own mental tally, and the gap between actual and perceived distribution becomes a source of conflict. A shared counter, even an imperfect one, defuses most of those conversations.

Replacement. An on-call rota tolerates no gaps. When someone declares themselves unavailable a fortnight out, you need to identify immediately who holds the required qualification, who is already in mandatory rest, and whose counter is lowest. All three criteria together, otherwise you solve one problem by creating another.

What you need to be able to produce

• The rota by line and period, with its change history.
• Call-outs, timestamped, with their duration.
• Totals per person over a reference period, to make fairness objective.
• Replacements made and the reason for them.

None of this is exotic. But these four items rarely sit in one place, and it is precisely their scattering that makes the exercise painful when something has to be justified.

In practice

An on-call line models simply as a dedicated project to which people are assigned by period, kept separate from day-to-day activity schedules so the totals stay readable. Team management lets you partition departments, each keeping control of its own rota without losing the overall view. The various planning display modes matter more than they appear here: the per-person view checks rest and totals, the per-period view spots gaps. What is covered is detailed in the feature list.

One closing piece of advice: start by recording call-outs, even by hand, before trying to optimise distribution. Fairness is argued from figures; without figures, it is argued from memories.

One way to settle it without losing a week: 30 days of free trial, long enough to test the tool against a case you know inside out.

Source: instruction DGOS/RH5/2026/57 of 6 May 2026, French official health bulletin. This article sets out a general framework and is not legal advice.