Archive for May, 2011
English waiting times improved in March, but it’s surprisingly difficult to tell from the government’s press notice. There are so many waiting time measures being tracked now, that it is hard to work out what is going on.
So what did happen in March?
As spring blossomed, Trusts sharply increased the number of patients they treated from the waiting list, and set about tackling their long-wait backlogs. As a result, the number and proportion of over-18-week waiters on the waiting list fell to pre-winter levels, which is good news for patients and the service. The number of very long waiters over 52 weeks also reduced slightly to 14,355.
But bringing in all those long-waiters comes at a price: it looks bad on the main waiting time targets, which track the patients treated, not the patients still waiting. Admissions within 18 weeks deteriorated to 89.6%, the “worst” since mid-2008. Because this “decline” is an artefact of tackling long-wait backlogs, it should “recover” as the benefits of this work feed through. (The “scare quotes” around those words indicate that I don’t believe those measures are very helpful, for reasons that are illustrated here.)
In the following chart, both lines show 90th centile referral-to-treatment (RTT) waiting times (i.e. 10 per cent of patients waited longer than the time shown by the line). The solid line shows how long admitted patients waited (this remains over 18 weeks), and the dotted line shows waiting times for those patients who are still waiting (this has peaked and is now coming down).
For the full details, a waiting time fact-checker is available for download here:
It contains the complete time series across all RTT waiting time measures. So if you hear a politician saying that waiting times went down (or up) in March, you can check the facts for yourself.
Maps of waiting time pressures
Like everything else in the NHS, waiting times vary enormously from place to place. So here is a collection of interactive maps, one for each main specialty, showing the underlying waiting time pressures around England. Each Trust has a pin stuck in it, and the pins are colour-coded according to the RTT waiting times of the longest-waiting 10 per cent of the waiting list.
Click any Trust’s pin to get more detail in a balloon. You can also click the Trust’s name in the balloon for a more comprehensive analysis with benchmarking and time trends (this is delivered via a separate website and requires free registration).
Variations between Trusts
It is fascinating how different Trusts respond to waiting time pressures in different ways. Some try to reduce waiting times by treating their long-waiters (even if that causes a breach of the headline target), and some stick to achieving their headline targets (even though that causes the backlog to grow). Although the former approach is better for patients, the penalties (which can include heavy contractual fines) make it understandable that many Trusts choose the latter.
The next exhibit is a video: a time series showing the different pressures and behaviours of all English Trusts. Each Trust is represented by a pair of dots, one red and one blue.
Both of these dots show the waiting times of the longest-waiting 10 per cent of patients; but the red dots are based on those patients who are still waiting, and the blue dots are based on those patients who were lucky enough to be admitted. Each monthly chart is sorted by the red dots, so those Trusts with the greatest underlying pressures are always on the right.
You can view the video on YouTube here: (If you are interested, the curious history of how 18 weeks was achieved is discussed here.)
The final chart in the video is:
The two outlier Trusts with very large waiting time pressures, shown by the two dots at the top right, are Kingston Hospital NHS Trust and University College London Hospitals NHS Foundation Trust. Between them, they account for over 20 per cent of the over-52-week waiters in England.
Kingston has experienced a very sharp increase in waiting time pressures since October 2010, and UCLH has been struggling with long-wait pressures for over a year. UCLH in particular illustrates how it is possible for Trusts to achieve the admission-based 18 week targets even if they have large numbers of long-waiters on their waiting lists.
Going back to the chart with the red and blue dots: the point where the red line crosses 18 weeks is an interesting indicator. It shows how many Trusts are experiencing relatively low waiting time pressures, to the extent that less than 10 per cent of their waiting list is over 18 weeks.
This indicator peaked in the month of the General Election, May 2010. Then it deteriorated until New Year, and has partially recovered since then; as the following chart shows.
So the March 2011 statistics paint a hopeful picture for English waiting times, with pressures reduced since the winter. April brought a winter of a different kind, however, as the financial pressures started to bite in earnest. We will all have to keep a close eye on waiting times.
Last week, the Board of Portsmouth Hospitals NHS Trust (PHT) considered the regular performance report from their Chief Operating Officer. The section on 18 week waits said:
Current Position: At the end of March there were 1,319 patients on the 18-week wait backlog.
Action: A plan was agreed with commissioning colleagues that PHT would focus capacity on clearing the backlog. This commenced in February and accounts for the deterioration in our performance against the 90% target for admitted patients.
That, in a nutshell, shows the dilemma caused by the government’s 18 week referral-to-treatment operating standards, as enshrined in the NHS Constitution and in law. It is clear from Portsmouth’s performance report that managers there understand the dilemma perfectly.
So what is this dilemma, exactly?
The government’s requirement for admitted patient waiting times is:
90 per cent of pathways where patients are admitted for hospital treatment should be completed within 18 weeks
…which sounds great, but is equivalent to saying:
Of every ten patients you select for admission,
only one is allowed to be a long-waiter.
Clearly, this becomes a serious problem when you have a lot of long-waiters.
To their credit, Portsmouth are tackling their long-wait backlog by actually treating their long-waiting patients, even though it makes them look bad against the headline target. In the circumstances, it would be understandable if they had chosen the alternative (as many other Trusts and commissioners do), which is to achieve the target by admitting mainly short-waiting patients, even though that allows the backlog to grow.
But it is somewhat harder to understand why the government continues to put Portsmouth, and dozens of other Trusts, in this dilemma at all. Alright, the 18 week rules were inherited from the previous administration, but waiting times are not exactly a low-profile issue, they’ve had a year to fix it, and the problem is only going to get worse as the backlog pressures grow.
Changing the rules would be straightforward: just apply the targets to those patients who are still waiting, instead of those patients who are lucky enough to be selected for treatment.
Ultimately this would mean changing the regulations that support the NHS Constitution. But the government could make a start more quickly by amending the Operating Framework, and only monitoring compliance against the 90th or 95th percentile waiting times for incomplete pathways (and ceasing monitoring against admitted and non-admitted pathways, or against the median).
That would greatly reduce the number of waiting time measures that Trusts need to monitor in their monthly performance reports, and help Trust directors focus on what is really important. And it would allow Portsmouth, along with all the other Trusts whose waiting time pressures are increasing, to concentrate on tackling the pressures properly without the risk of being penalised for their efforts.
(This post first appeared in HSJ blogs)
Want to prove that waiting times in England are going up? Pick this measure. Want to prove they’re going down? Pick that one.
There are so many different ways of measuring waiting times around at the moment, it’s easy to get confused when politicians cherry-pick the ones that suit their story (as happened recently at PMQs).
Yesterday John Appleby, Chief Economist at the King’s Fund, blogged about this very problem. He explained how the different measures work, and opened up a timely debate about which are the most meaningful.
So I thought it would be helpful to put together a simple table, showing English referral-to-treatment time trends across the whole of the available data series, and including all the different measures you might hear about. A kind of fact-checker for waiting time statistics.
You can download the table as an Excel spreadsheet here:
So the next time you hear a politician saying that waiting times have risen month-on-month under this Government, you can check if that’s true. If you hear another saying they went down last month, you can check that as well.
If you are a politician, and want to build a case that flatters your party’s record, then no doubt this assortment of measures will help with the statistical cherry-picking. But be warned, the rest of us are wising up too!
(This post first appeared in HSJ blogs)
I am glad that he mentioned waiting times, because, two weeks ago, at that Dispatch Box, he said that waiting times “have risen month on month under this Government”. That is not true. The figures, which he had at the time, show that in-patient waiting times fell from 9.1 to 9 weeks. For out-patients, they went down from 4.8 weeks to 3.5 weeks, the lowest for a year. It is important when we come to this House and make statements that are inaccurate that we correct the record at the first available opportunity.
Forceful stuff. But what do “the figures” actually mean? We’ll come to that in a moment. But first of all, imagine that, in a parallel universe, David Cameron had stood at the dispatch box and said this:
He said long-term unemployment has gone up under this Government, but that is not true. The figures clearly show that the time people remain unemployed is going down. The median time out of work, for all new job-starters, actually fell last month, and now stands at the lowest level for a year. So there.
Okay, now the alarm bells are ringing. The objections are:
- We aren’t interested in job-starters, we want to know about the unemployed.
- Most job-starters come straight from another job, so of course the median time between employments is low.
- The people we really want to hear about are the long-term unemployed. They could be out of work their whole lives and never be picked up by this statistic.
And so it is with NHS waiting times. The figures quoted by David Cameron are the median waiting times experienced by “inpatients” (called “adjusted admitted pathways” in the statistics) and “outpatients” (“non-admitted pathways”). Exactly the same objections apply:
- We aren’t interested in the patients who get treated, we want to know about patients on the waiting list.
- Many of the patients being treated have urgent clinical conditions, so of course the median waiting time is low.
- The people we really want to hear about are the long-waiters. They could be waiting for years and never be picked up by this statistic.
If PMQs were an exercise in revealing deep truths about the government of the nation, and not a political boxing match, then the Prime Minister might have looked more carefully at the figures and said something like this:
I am glad that he mentioned waiting times, because, two weeks ago he said that waiting times “have risen month on month under this Government”. I would like to set the record straight. The figures for February, to which he was referring, show that ten per cent of patients on the waiting list had waited over 19.2 weeks since referral. This was actually a slight improvement on the previous month, so “month on month” is not strictly correct.
However it is true that waiting times have worsened significantly since the record performance achieved just before the election, when ten per cent of the waiting list was above just 16.5 weeks. It is important when we come to this House that we make statements that are meaningful and honest, and I have done so today.
[Hat Tip to joefd for tweeting about analogies]
NICE launched eighteen pathways yesterday, covering everything from neonatal jaundice to dementia. If it’s your job to plan NHS capacity into the future, how should you respond when pathway changes are on their way?
The point of planning is to prepare for the future. You can’t predict everything that is going to happen, but you do your best, accounting for foreseeable changes like trend demand growth, efforts to cut waiting list backlogs, demographic drift, and foreseeable pathway changes. This is why your plans are better than assuming the status quo.
Pathway changes are often the most complicated, because they are usually not representative of the specialty and so all the performance averages (such as lengths of stay) have to be changed too. So you can’t simply deduct a quantum of demand and leave everything else the same; you need to do something a bit cleverer.
Don’t change the future – rewrite the past
The best way to model upcoming pathway changes is to rewrite the past, as if the new pathway had always been in effect. So if a particular HRG is going to be managed out-of-hospital in future, then you need to filter that HRG out of your past activity data before passing it to a query (or to Gooroo Planner) to extract the information you need about activity, length of stay, clinical urgency rates, seasonal demand profiles, and so on.
If your pathway change has more complex effects then you may not be able to capture them with simple database queries that filter patients in or out based on things like age, postcode or HRG. If this complexity is significant and needs to be modelled explicitly, then a more specialist simulation model such as Scenario Generator can be used to model the pathway flows, get an indication of the effect on waiting times, and work out the implications for capacity.
If you need to look at waiting times more thoroughly, then Scenario Generator can be used to generate the rewritten past data that you need, on the new pathway basis, before passing it to Gooroo Planner for the detailed waiting time, capacity and financial calculations.
be specific about how care will change
So when you’ve been slaving away at next year’s plans, and somebody pops up with a challenge about a pathway change, don’t mutter something about estimating the effect on demand trends. Instead, you can ask them to be specific about the characteristics of the patients affected and how their care will change. If they deliver the goods, you’ll know what to do: rewrite history, and then use that as the basis for your new plan.
The median wait will also continue to be monitored
with a view to improvement.
“Targets” have been abolished, of course. But the Department of Health nevertheless makes sure that the NHS is aware of the standards that are expected, and currently for median referral-to-treatment (RTT) waits these are:
- Admitted (inpatients and daycases): <= 11.1 weeks
- Non admitted (outpatients): <= 6.6 weeks
- Incomplete (still waiting for treatment): <= 7.2 weeks
Who could possibly object to this? The median is a kind of average, and everybody wants average waits to come down. If the waiting list backlog is being cleared, then the median will fall. So it’s an indicator of success. Isn’t it?
Before criticising the median, let’s start with its good points. Sometimes the median is a better way of describing the average, especially when large variations are involved. For instance, let’s say we are looking at how much money people earn. The mean is skewed by a small minority of millionaires: so skewed that two-thirds of people in the UK earn less than the mean. So the median is a more interesting measure of income, as it shows the income level that half the population are above and half below. (You’ll find a better discussion of this here.)
When you’re talking about income, the median tells you how you are doing relative to everyone else. But when it comes to waiting times it doesn’t work so well. Why? Because there are two kinds of patients on elective waiting lists: urgents and routines. And it matters which group you fall into.
If you are an urgent patient, you can expect to be treated as quickly as your condition requires; the long-wait statistics won’t apply to you. But all these urgent patients do affect the median: across all specialties they make up a big chunk of the 50% of patients who fall below the median waiting time, and virtually none of those above it. So if you happen to be a routine patient, then the overall median is a poor guide to the waiting time you should expect. If you are an urgent patient then it is no guide at all.
If we narrow our focus to medical specialties then the situation is reversed, because more than 50% of patients are likely to be urgent. Then the median patient is an urgent patient, and so the median tells us nothing at all about routine waiting times.
It gets worse for the poor old median. Imagine you work in a Trust where (across all specialties) 25% of elective patients are urgent. Imagine that the other 75% of patients – the routines – are being admitted in a fairly random order. Where is the median? The 50% of patients who wait less than the median comprise the 25% who are urgent, plus a further 25% who are routine. The median waiting time is at the top of these shorter routine waits.
These lucky short-waiting routine patients are jumping the queue on those less-fortunate routines who wait longer than the median. That isn’t very fair, and so you decide to sort it out. In practice it isn’t possible to admit routine patients exactly on a first-come-first-served basis, but to keep this example simple we will assume that you can. So you eliminate queue-jumping by routine patients, and make those short-waiting routines wait their turn instead. In return, the long-waiters aren’t being delayed by queue-jumping any more, so their waiting times (and the maximum waiting time) come down.
What happens to median waiting times when we achieve this perfection?
The 25% of patients who are urgent are still being treated quickly, below the median waiting time. Then the remaining 75% who are routine all wait their turn in the queue, so they all experience the same waiting time. Where is the median now? The median is now equal to the new routine waiting time, which is equal to the maximum waiting time. You have improved the management of your waiting list, everything is better and fairer, and yet (because the shortest-waiting routine patients are waiting longer) the median wait has gone up.
This is the killer argument against the median waiting time. If it goes up, it might mean that things are getting better (e.g. better scheduling), or it might mean they are getting worse (e.g. longer waiting lists). Likewise if it goes down. The position of the median does not tell us clearly whether things are good, bad, or indifferent. We always need to look at other measures, because the median on its own doesn’t help us.
That is why Trusts, Commissioners, the Department of Health, Monitor, the CQC, and everybody else should ignore the median. Instead, they should focus on the 90th or 95th centile, because that really is a guide to routine waiting times. And, as argued elsewhere, incomplete pathways are a better guide than admitted pathways (or non-admitted ones, for that matter). The statistic to watch is the 90th (or 95th) centile waiting time for incomplete pathways. The other seven measures in common usage can safely fall by the wayside.
Is there no measure of the average that could help us? Unfortunately there isn’t an easy one. The mean waiting time of patients who are still waiting is affected by the order in which patients are admitted. And the mean waiting time of admitted patients, while unaffected by the order in which patients are admitted, is only a guide to the underlying pressures in a steady-state situation (which isn’t the case here).
Having said that, we could get a reasonable indicator by dividing the number waiting by the average referral rate over some recent period, in order to back-calculate the mean sustainable admitted waiting time. And if we’re going to go to those lengths, then we may as well do the job properly and calculate the achievable waiting time, taking account of clinical priorities, removals, etc. But that is a topic for another post.