The nurses on a ward keep a paper list alongside your system because they do not trust its task list. What has gone wrong, and what do you do?

Ask to photocopy the list

The first move is not a design discussion. It is to obtain a few of the actual lists at the end of a shift and to watch two handovers. A ward list that has survived long enough for you to hear about it is a stable artefact that busy people maintain by hand under time pressure, which means every column on it is earning its keep. It is the most honest requirements document you will ever be handed, and nobody had to ask for it.

What you tend to find is not a duplicate of the system. It is four or five things the system holds badly or not at all. There is an ordering, and the ordering is neither alphabetical nor by bed but by who the nurse is worried about, which is a judgement no field in the record contains. There are tasks with no requester and no order behind them: chase the radiology report, ask the family to bring the medicines in, find out whether the referral was accepted. There is a short note per patient carrying the reason this patient is being watched, which is the most useful sentence at handover and the hardest to model. There is a record of what has already been chased, so the next person does not chase it again. And there are the patients who are not on the ward's list in the system at all — the one in a side room, the one boarded from another speciality, the one still on a trolley in a corridor.

So the diagnosis has a shape before you write anything. The system holds orders, results and documentation; the ward is holding work. Those are different objects, and most electronic records were built for the first.

What a shared work list has to do that a private one does not

Once you accept that the ward needs a work list, the requirements turn out to be unusual, and a generic task feature will not survive contact with a shift.

A clinical task has to be able to exist without an order behind it, because a great deal of ward work is chasing, arranging and confirming rather than doing something that was prescribed. It has to carry a due time that is clinically meaningful — observations due within the hour, a dose due at a fixed time, a review before the ward round — because a task whose due time is a rounded guess is ignored within a week. It has to be assignable to a role on the current shift rather than to a named individual, since the individual changes every twelve hours and a task assigned to someone who has gone home is a task nobody holds. It has to be completable with a reason for non-completion that is not treated as a failure, because "not done, patient was in theatre" is normal and needs to be sayable. And it has to be visible to the whole team at once, which is exactly what the paper version is not.

That last property is where the argument for change actually lives, and it is worth reaching for early.

Why the private list is a safety problem and not a compliance one

The tempting framing is that staff are working outside the system, the system's data is therefore incomplete, and the fix is training plus an audit. That framing loses the question, because keeping the list is very often the safer thing to do given what the software offers, and everyone on the ward knows it.

The hazard is not that the list is on paper. It is that it is held by one person. If a nurse is holding the knowledge that bay four needs watching and that a result is still outstanding, and that nurse is pulled into an emergency, sent to another ward or simply hands over at speed, the work becomes invisible at precisely the moment nobody has spare attention. The consequences are the familiar ones: the abnormal result nobody acted on because the person waiting for it was not there, the escalation that was going to happen and did not, the job done twice, and the patient who moved ward and left their entry behind. A shared list degrades gracefully when a person disappears. A private one does not degrade, it vanishes.

Framing it this way also changes what you propose. You are not trying to stop the paper. You are trying to make the shared version better than the paper, at which point the paper retires itself, and if it does not you have learned something you did not know.

Deciding what to change, and in what order

The credible answer is incremental and starts with the cheapest of the missing capabilities rather than with a new module. Give the ward a list that is genuinely theirs, including boarded and outlying patients, because a list that omits three of your patients cannot be the list of record. Allow a task to be created without an order. Let the row carry a one-line clinical concern with an author and a timestamp, and treat that as a first-class field rather than as a comment. Make the sort order reflect urgency, and let the ward change it. Then look at whether the paper survives.

Some of what is on the list should never move into the system, and saying so is a mark of judgement. Personal reminders, shorthand and the coordinating chatter of a shift do not need to become a permanent part of a clinical record that a patient can request and a court can read. The test to apply column by column is whether the next person on shift needs it in order to work safely. If they do, it belongs somewhere shared. If it is one person's memory aid, leave it alone.

The other half of the change is not software at all. Handover needs a structure and a protected place to happen, the ward round has to produce tasks rather than intentions, and someone has to own the list on each shift. A perfect task feature with no agreement about who reads it at eight in the morning changes nothing. And you will know whether it worked by watching another handover, not by asking whether people liked it.

When the system disappears entirely

In May 2017 the WannaCry ransomware spread through parts of the NHS and left staff without access to clinical systems, cancelling appointments and operations and diverting emergency patients elsewhere. Wards with a workable paper process kept going; wards without one stalled.

Hold that next to this question, because it cuts against the instinct to eliminate paper on principle. The ward list, the printed summary and the pre-printed order form are also the substrate the workflow runs on when there is no screen at all. The mature position is neither to tolerate a private list nor to abolish paper, but to make the shared electronic list the working tool while keeping a deliberate, current and practised paper fallback for the day the tool is not there.

A workaround is a design signal with a timestamp on it. Ask what the list can do that your software cannot, build that, and the paper will retire itself. Ban it first and you have removed the evidence without removing the need.