NHS gooroo blog

Posts Tagged ‘disruption’

Previously I outlined why the ‘PTL’ approach to waiting times management is a labour-intensive way of achieving sub-optimal performance, and why it should be replaced with a rules-based approach to patient scheduling that optimises waiting times continuously without needing regular intervention from senior managers.

Here we will illustrate the point, taking a real NHS waiting list and (in a simulator) converting its management to the rules-based approach. We will see the 92nd centile waiting time fall from 14 weeks to 6 weeks,  even though the size of the waiting list has not reduced. We will see how the biggest improvements are seen in the first month and how gains are sustained indefinitely (so long as the waiting is not allowed to grow).

So let us start by introducing our real waiting list. Here it is, and it’s the new outpatient list for one consultant orthopaedic surgeon.

Let’s take a quick tour of this image. The vertical grey stripe shows the date of this analysis, which in this example is the week beginning Monday 20th August 2012. To the left we have the waiting list; each blob is one patient, and they are sorted according to their referral date (so the longest-waiting patient at the far left has been waiting 36 weeks). If a patient does not have an appointment date (like the longest-waiting patient) then they appear hollow. When they are given an appointment they turn solid, and a copy of the patient is placed on the appointments diary; the appointments diary goes off to the right and into the future.

There are 121 patients on this waiting list, and the 92nd centile waiting time (i.e. how long the 112th patient has waited) is 14 weeks. So that is our starting point.

How are this consultant’s outpatients being scheduled?

Looking at the appointments diary (the grid to the lower right), there are plenty of empty clinic slots in the current week. Then for the next two weeks all slots are fully-booked, and after that there are plenty of empty slots again. So it is not clear, looking at the appointments diary, what the rules are for offering slots to patients.

Looking at the waiting list (to the top left), we can see that a majority of short-waiting patients are unbooked, then nearly all the longer-waiting patients are booked, and the few very-long-waiters are mostly unbooked.

So it looks on the face of it as if this is loosely a partially-booked system, booked about 3-4 weeks in advance, with efforts being made to book patients in date order. However slots are not being fully utilised, many short-waiting patients are being booked ahead of longer-waiters, and the very longest-waiting patients are not booked. So there should be scope for improving the management of this consultant’s list if we used all the available clinic slots and booked strictly in date order.

To mimic the possibility that patients are genuinely choosing not to take up offers of early appointments, we will assume that some 10 per cent of patients rearrange their appointments at short notice and put themselves back in the queue. (It is of course possible that the very-longest patients, who are unbooked, are choosing to delay their appointments by several months, but the delays are so long that it is surely legitimate to ask whether they should have been referred in the first place.)

So let’s hand this waiting list over to the computer and let it do the booking. (If you want to do this kind of analysis on your own waiting lists, the actual waiting list was loaded up into Gooroo SimActive, and then after clicking Save you can hand it over to the rules-based booking simulator.)

What’s changed in this picture? We’ve filled all available routine appointment slots (the blue ones) up to 4 weeks ahead, by booking patients in strict date order. Also, more patients have been referred, including a couple of urgent patients (shown in red); those urgent patients have been booked into urgent slots (the red ones). So we can already see how things are going to get better: over the next few weeks all our longest-waiting patients are going to be seen (unless they cancel).

Fast-forward a few weeks, and things are looking very different:

Most of the long-waiters have been seen already, and there’s just one patient from May still waiting. We did have a couple of cancellations who were put back on the list (they are shown with a double letter, to show that they’ve been cancelled). When we hold this week’s clinic, we’ll clear most of the remaining long-waiters and things should look a lot better.

One week later:

This is the new normal shape for our waiting list, under rules-based partial booking up to 4 weeks in advance. If we chose to run a fully-booked service instead, offering patients a choice of slots in up to 3 different weeks (which is compatible with Choose and Book rules), waits would be a week or two longer but otherwise in a similar position.

This is sustainable, so long as the size of the waiting list does not grow. Here is the same list a year later:

We’ve achieved a dramatic reduction in waiting times by implementing some relatively simple rules for booking patients, continuously as they come in. No weekly PTL meeting, no senior managers fire-fighting, just good clerk-level administration.

If you have access to YouTube (i.e. probably not from an NHS computer) you can see this scenario running as a video here:

 

Also new in Gooroo SimActive: there is now a download button that downloads all the data including the meta-data about additions, removals, etc. So now you can build up a library of files for your subspecialties and consultants, without having to type in the meta-data every time you upload something.

In a parallel post I explain why PTLs should now change, and evolve into “active PTLs” which work continuously to minimise waiting times for all patients. This blog post explains how in a bit more detail, describing the rules for operating active PTLs.

I’ll also take the opportunity to sketch out briefly the origins of PTLs, because they were a tremendous achievement in their day. It is easy to forget just how unmanaged the NHS’s waiting lists were in the 1980s, and the originators of PTLs deserve credit for their roles in making today’s shorter NHS waiting times possible.

Let’s start with the active PTL rules.

There are only five rules, and they aren’t particularly complicated. The difficult part was excluding all the alternatives, and quantifying the behaviour of the system to allow the calculation of booking rules and waiting times; this took two years of PhD-level research, and the study of over a billion simulated patient bookings. If you want to find out more about the simulator research, you can download the research papers here, and you can try the simulator by logging in here and clicking SimView (registration and use is free to NHS).

The purpose of laying out the rules in this blog post is to stimulate interest in the next stage, which is to take the active PTL rules beyond the simulator and into the real world. If you are interested in joining those hospitals who have already expressed an interest then you can email me at rob.findlay@nhsgooroo.co.uk

Getting ready

Before implementing an active PTL, you will first need to:

a) know, at subspecialty and stage-of-pathway level, the size of waiting list that is consistent with your waiting times targets;

b) ensure that enough slots will be delivered through your available capacity to achieve and sustain a waiting list that is no bigger than that; and

c) carve out the right number of slots for urgent and cancelled patients.

A free booking rules calculator that helps with all this is available after login at nhsgooroo.co.uk.

The active PTL rules

The rules work differently for fully-booked and partially-booked services. In a fully-booked service, which should include all services using direct Choose & Book, all patients are invited to make an appointment. In a partially-booked service, which only works when the provider has control over all appointments, slots are only available a limited number of weeks ahead (typically 6 or 4 weeks) to minimise disruption caused by staff taking leave. The rules work for both clinics and theatres.

The active PTL rules are driven by five different events:

1) An urgent patient needs booking

Find out how long the patient can safely wait because of their clinical condition. Book them into the latest empty urgent or routine slot within that time. If no empty slots are available, create one by cancelling the routine patient who will be least inconvenienced.

2) In a fully-booked service: a routine patient has had their appointment cancelled and needs rebooking

Offer the patient a choice of any empty urgent or routine slot in the first three weeks in which empty slots are available.

3) In a fully-booked service: a new routine patient is added to the waiting list

Offer the patient a choice of any empty routine slot in the first three weeks in which empty routine slots are available.

4) In a partially-booked service: empty routine slots become available

Select routine patients for booking in the following order: cancelled patients first (starting with the longest-waiters), then new patients (again starting with the longest-waiters). Book each patient into the soonest empty routine slot, until all available routine slots are filled.

5) There is an empty urgent or routine slot at very short notice which is at risk of being wasted

Fill the slot, ideally with an urgent patient or by bringing forward a long-waiting patient, or alternatively with a new routine patient.

Tactics that are not in the rules

Avoid holding extra slots in reserve. Avoid running services that neither offer bookings to all patients (if fully-booked), nor fill all available routine slots (if partially-booked). Avoid “rippling”.

A short history of PTLs

According to Anthony McKeever (who was there at the time), PTLs all came about in the mid to late 1980s.

The thought leader was Professor John Yates who studied in great detail the influences that led to long waiting times. By analysing the available data he identified that if you increased the focus on the back of the queue then long waits could be greatly reduced.

Mersey RHA, under Sir Duncan Nichol and Sir Donald Wilson, turned this into a policy to achieve 2 year maximum inpatient waits, which sounds long today but was ground-breaking at the time.

This policy was developed into practical methods by Kevin Cottrell and Anthony McKeever. First they developed the concept of Personal Treatment Plans, which were individualised for each long-waiting patient and agreed with their consultant. These developed into provider-led Patient Treatment Lists, and these were the first PTLs. As other NHS organisations picked up the techniques, the PTL abbreviation stuck but came to stand for a variety of different words.

First published at HSJ blogs

“Cancellations push up waiting times. So does a big waiting list. Common sense, isn’t it? You don’t need a model to tell you that.”

All true. But what common sense doesn’t tell you is how much they push up waiting times. This matters when you’re looking at a 10-week outpatient wait and trying to get it down to 8 weeks – you need to know how big an effort you’re going to need.

So what does increase waiting times? And is there anything you can do about it? If you can answer those questions, then you’re well on the way to cutting waits at almost no cost, and without wasting valuable time trying to fix things that don’t actually matter.

This post reports on simulation research by Gooroo that investigates the (sometimes surprising) effects of different kinds of disruption on waiting times. (It does not consider the effects of good and bad practice when booking patients; that will be left for a later post.)

Things that make a big difference

Urgency of casemix

Urgent patients, by definition, need to jump the queue because of their clinical conditions. And the more queue-jumping there is, the longer other patients wait, and the longer the longest waiting times.

Similarly, just how quickly these “urgent” patients need to come in also makes a significant difference to waiting times. The quicker they come in, the more other patients are pushed back, and the more the longest waits go up.

All this is well-established, and the routine waiting time you would expect as a result can easily be calculated.

But something else happens as well. Urgent patients also cause a lot of general disruption when booking patients. The more urgent patients there are, the more disruption there is. Both the number of longwaits (after taking into account the queue-jumping effect), and other undesirable events (like delayed urgent patients and rebooked routine patients), increase steadily as the urgency of the casemix goes up.

You can’t alter the clinical urgency of the patient. But you can increase the care taken to describe their clinical urgency accurately. If urgency is being systematically over-declared, then waiting times are being needlessly increased. So it is worth paying close attention to the process and criteria used to declare the urgency of patients.

Number of patients waiting

If a list gets bigger (relative to activity) then waiting times go up. Again this is well-known, easily quantified, and the expected routine waiting time can be calculated with the size of the waiting list taken into account.

But again, there are additional effects. Bigger lists suffer much more disruption than smaller ones. Again the disruption increases the number of longwaits who exceed the expected routine waiting time, though it does not greatly affect urgent patients or rebookings.

Controlling the size of the waiting list is already a priority throughout the NHS, so this finding merely adds emphasis.

Removals

Here “removals” means those patients who are removed from the waiting list well before they come in for their appointments. If a patient already has an appointment booked, then this is cancelled and reused for another patient.

You might expect removals to cause some disruption, but not much. But that expectation does not account for the sheer numbers of patients removed from NHS waiting lists. For admitted patients (inpatients and daycases) it is quite typical for 15% of patients to be removed, and often as many as 25%. At such high volumes of removals, the level of disruption is huge, with longwaits and rebookings being the main consequence.

So any action that might reduce the number of removals, by ensuring that patients are only added to the waiting list if they are likely to proceed with their appointments, would be beneficial.

(Suspending patients is not currently practised in England, but is used elsewhere, and this practice has a smaller effect on longwaits and rebookings.)

Modest effects, but still worth looking at

The following effects would be worth looking at too, but more for reasons other than reducing disruption.

If patients are cancelled on the day of their appointment and subsequently rebooked, this causes a moderate increase in disruption, in addition to the waste of capacity caused by their unused appointment slot. (Patients who are cancelled but not rebooked, like many Did Not Attend (DNA) patients, waste capacity but do not cause disruption.) If the main motivation for tackling cancellations is to reduce wasted capacity, then deliberately over-booking every session (and accommodating the resulting variability in session length) would be worth considering as a compensating tactic.

If significant numbers of patients are pooled between several clinicians, this helps to even out waiting times (and therefore reduce the maximum waiting time for the service). This levelling of waiting times should be the main motivation for this tactic. As a useful side-effect, it also happens to reduce the amount of general disruption and so produces a further benefit on waiting times.

Likewise if sessions are combined (e.g. running one all-day operating list, instead of two half-day lists a few days apart) then there is a small benefit for longwaits and rebookings. This benefit is partially offset because there is a slight increase in the delay to cancelled urgent patients as a result of the less-frequent sessions. Again, combining sessions would normally be considered because of other motivations, namely making it easier to schedule long procedures flexibly, which is why all-day sessions are beneficial in specialties like orthopaedics that have common, long procedures.

For full details of our research into the causes and effects of disruption, see Research White Paper 3. For quantification of their effects on waiting times, see Research White Paper 5, or use Gooroo’s Booking Rules Calculator for individual services (this appears just after login) or Gooroo Planner for bulk analysis.