Across the UK the planning round for next financial year is getting underway, while in England some commissioners are still signing off this year’s contracts (with nearly half the year already gone). Could there be any better time to ask: how could the annual planning round make better progress? Or, to put it another way, how can we stop it from going around in circles?

The task facing us is this: to develop a plan that will keep up with demand and achieve the cancer and other waiting times standards, but do so within the limited money and capacity available. With flat money, it’s a bigger ask than ever. How can we structure the whole process, to agree a suitable plan as quickly and easily as possible?

1) Organising the data and methods

At the very outset we need to agree on the planning model that all parties will use throughout discussions. This model will show the implications of the emerging plan, so that all parties can see whether the current position is consistent with their objectives, and work up their own proposals properly before bringing them to the table. So the model needs to be neutral (which in England means it must favour neither commissioner nor provider), and it needs to be both sophisticated enough to account for everything that matters, yet straightforward enough to be useful in practice.

What if we don’t agree a model at the outset? The main alternative is for one party (usually the hospital’s analysts) to build a home-grown spreadsheet, and continually adapt and extend it as discussions progress and new considerations are thrown into the mix. But the build-as-you-go approach can lead to the model becoming a bit of a battleground in its own right. These spreadsheet models become complex very quickly, so a lot of time is spent explaining and understanding them, and correcting the errors that spreadsheet models are prone to. Also, any party that is not involved in building the model cannot test its own proposals properly before bringing them to the table, and may worry that the model’s construction in some way favours its creator.

We also need to reach agreement about the level of detail we are going to use in our model. For instance we might model by specialty (or subspecialty), and split between non-elective, new outpatient, etc (and we need to decide how to handle things like fracture clinic, non-consultant led clinics, and diagnostics).

The level of detail is not quite a free choice. Our data must be internally consistent, and that can get especially tricky when it comes to data about the waiting list. It may not be clear, for instance, which subspecialty a patient will eventually fall under when they are still on the outpatient waiting list, and a crude split by consultant is often a good workaround. In England there are further complexities, because the split between specialist commissioning (NHS England’s responsibility) and the rest (the CCG’s responsibility) is a complicated one that runs right through the middle of many specialties. It is often impossible to tell which waiting list patients will eventually be paid for by which organisation. One option is to simplify matters by modelling a catchment population with CCG and NHS England patients all lumped together, and then split specialist from non-specialist commissioned activity pro rata right at the end.

The next task is to agree at the outset how clinical urgency is going to be handled, and in particular the time limits for treating different urgent patients. This should be relatively straightforward, for instance: cancer patients within 2 weeks, and other urgent patients within 4 weeks, at each stage of treatment.

A common cause of awkwardness is whether or not to update the data as time goes by (often accompanied by a little peeking to see which point of view this would favour). So it is worth agreeing at the outset whether, and when, the data should be refreshed; for instance: that initially the model will be based on a recent 12-months-worth of data (e.g. August 2012 to July 2013), but refreshed in January as soon as data is available for the calendar year 2013.

2) Historical facts

With the ground rules set out, the time has come to gather some historical facts about our chosen past period: that’s waiting list and non-waiting list activity, and other waiting list movement data such as additions, other removals, and growth in the waiting list.

All this data should be a matter of indisputable record, but it is a fact of life that records vary and we need to take care that all parties agree at the outset what the correct numbers are (which in England includes comparisons between the provider’s data and any alternative sources such as SUS and SLAM). This is a technical and unexciting exercise, but confidence relies on it and it needs doing properly. Once done, the numbers should be frozen (except when the data undergoes a planned refresh).

3) For early agreement

At this stage we can start getting into some early discussions, by seeking agreement on some relatively uncontroversial or technical points that could cause trouble if they aren’t resolved early. These are:

• Plans should always start from demand, but how should we measure it? Using the “activity plus growth in list size” method (usually more reliable because of better data), or the “additions minus removals” method?
• When working out waiting times, are we assuming a fully-booked service (more likely for new outpatients) or a partially-booked service (more likely for admitted patients)?
• What opening list sizes are we assuming for the 1st of April 2014? We could simply assume they’re the same size as they are now, but this is an important assumption and if a good plan exists for the rest of this year then that may help.
• Are we assuming that new outpatient activity has an automatic knock-on effect on elective demand? Usually yes, but there may be exceptions that we need to carve out, such as services where most outpatients happen at another provider.
• What should we assume about the week-by-week seasonality of demand? For instance, we could use the average of the last three calendar years, or decide that some recent 12 month period is a good enough baseline.
• Finally, if we are modelling money alongside activity, what average price are we assuming for each unit of activity in each service line? This is a bit of a moving target in England because new guidance will come out half-way through the process, but the important thing is to agree a starting assumption, and a methodology where tariff and locally-agreed prices are mixed.

4) Performance assumptions for discussion

Now we can get into the nitty-gritty. Along with detailed pathway changes and proposed re-provision of services, here are the key performance assumptions that should be the main focus of discussions:

• How fast will demand grow (and is there any one-off extra demand expected)?
• What waiting times are we aiming for at each stage of treatment (allowing a suitable margin for suboptimal waiting list management and imperfect subspecialty resource allocation)? What proportion of patients need to meet the target, and is it based on waiting times as patients are treated or on snapshots of the waiting list? Do we want to cap the list size too, to stop it from growing?
• What proportion of patients should fall into each urgency category, based on the urgency definitions agreed at the outset? Do the measured proportions reflect reality, or are they too high and pushing up routine waiting times? How should urgency be adjusted, to allow for patients who have a diagnostic step and arrive on the admitted patient list with just a few weeks left on the clock?
• What proportion of patients will be removed from the waiting list? Removals cause scheduling disruption and push up waiting times, so could the removal rate be reduced?

Hospitals will also want to consider carefully things like DNA rates, capacity performance, and capacity costs. Although these are not always of direct interest to English commissioners, they strongly affect the activity that providers can deliver, both in total and at different times of year.

5) The answer we are working towards

As both sides work towards agreement, the emerging plans will be expressed as future activity (and how it varies week-by-week through the year), and the money that implies. Although these are the answers we are working towards, they can only be reached by working through all the other things listed above.

At first we might allow our model to be driven by the waiting times targets, but as we converge on a solution it may be helpful to flip over to modelling based on a specified amount of activity instead. That way, we have the benefit of being able to fine-tune activity, with a watchful eye on the implications.

At the start of discussions the gap may look impossible. But as they progress, a workable solution should emerge, without resorting to the kind of big last-minute deals that wreck the carefully-built involvement of stakeholders along the way. And with a well-structured approach, it should be possible to reach that agreement long before the deadline.

If you are using Gooroo Planner as your planning model, you may find this spreadsheet useful. It groups all the possible data entry fields under the above headers, to help you consider them at the right stage of your negotiations. In practice you probably won’t use most of these fields, which isn’t a problem because you can start with a minimal dataset and extend it gradually, but they are all included for completeness.