Modelling pathways that bypass outpatients

Modelling pathways that bypass outpatients

04/12/2015
by Rob Findlay

In a typical surgical pathway, if we see a lot of extra new outpatients, what happens to the demand for elective inpatients and daycases? It goes up, because some of those extra outpatients turn out to need treatment and are added to the waiting list for an operation.

The simplest way to handle this in Gooroo Planner is to assume that if new outpatient activity goes up by (say) 5 per cent, then elective demand will also go up by 5 per cent. This is equivalent to assuming that the conversion rate is steady, and it’s a useful method because it works automatically without any need for manual adjustments.

Usually this logic works pretty well, but it breaks down if significant numbers of patients somehow manage to get elective care without going through new outpatients. Common examples are direct access endoscopy patients who are referred by their GPs directly onto the gastroenterology daycase waiting list, or post-trauma patients who may be added to elective orthopaedic waiting lists directly from fracture clinic.

For instance, let’s say that last year we saw 1,000 new outpatients and 5,000 new fracture clinic attendances. And let’s say demand for elective surgery was 1,000 patients, of whom half originated in new outpatients (with 50 per cent of 1,000 new outpatients converting) and the other 500 originated in fracture clinic (with 10 per cent of 5,000 new fracture patients converting).

Modelling naively

If we just modelled this orthopaedic service in the simple way, we might put the new outpatients at the start of the pathway (StreamPosition = 1) and the elective inpatients/daycases next in the pathway (StreamPosition = 2), with the fracture clinic patients being outside the pathway (StreamPosition = 0). This tells Gooroo Planner to interpret this pathway as 1,000 new outpatients converting to 1,000 electives – a conversion rate of 100 per cent – with the fracture clinic happening independently.

So things are going to go wrong if we then plan an outpatient waiting list initiative of 100 extra appointments, because Gooroo Planner would use a conversion rate of 100 per cent to produce extra demand of 100 elective patients (instead of a 50 per cent conversion rate to produce extra demand of 50 electives).

How can we adjust our model so that it reflects reality?

Method 1: Splitting the whole pathway

We could split the outpatient and fracture pathways completely, so that we have separate pathways for:

1,000 new outpatients becoming 500 ordinary electives (a conversion rate of 50%), and
(as a separate pathway) 5,000 fracture patients becoming 500 post-trauma electives (a conversion rate of 10%).

By splitting the pathway like that, all the conversion rates are calculated correctly and the usual logic for working out the knock-on effects works out fine.

The logic works, but in practice it can be tricky to split the data like this. The difficulty is at the elective stage of treatment: how can we distinguish between patients who started their pathways as new outpatients, from those who started in fracture clinic? The answer depends on how those elective patients are coded in PAS. If we’re lucky, it won’t be an issue because all the post-trauma patients will be coded as emergencies or have a trauma specialty code. Or the source of referral might reveal whether they came from fracture clinic rather than outpatients, or there might be some other clue such as not being on an RTT waiting times pathway.

But what if we don’t have any of those clues and can’t tell them apart at all?

Not to worry – there is still a way of handling this.

Method 2: Weighting the conversion rates

There is a dataset field in Gooroo Planner called StreamWeighting, which we can use to tell the model that fracture clinic patients don’t convert as often as new outpatients.

So in this example, we can put both the new outpatients and fracture clinic at StreamPosition = 1, meaning that patients can start their pathways in either setting. The electives go at StreamPosition = 2 as before. And for fracture clinic we set StreamWeighting = 0.2 to tell Planner that fracture clinic patients are one-fifth as likely to convert as new outpatients. (Note that we are not entering the actual conversion rate here – only how often fracture clinic patients convert relative to new outpatients.)

If you want to model this yourself in Gooroo Planner, the complete dataset looks like this (including the extra 100 new outpatients and 50 electives that we are going to do):

After running the model, based on delivering the specified activity plan FutActiv, the results are:

The 100 extra outpatients convert correctly to extra elective demand of 50 patients, as shown in the “Future total new demand from upstream knock-ons” line, thanks to the StreamWeighting that we put in for Fracture clinic.

This method is simpler, although it does rely on you being able to describe (or at least estimate) the relative conversion rates for different pathways.

Return to Post Index