Why the Contractor Population Index matters, and how they got it wrong.



The Contractor Population Index, like most things that are working well, has not attracted much attention until now. Its purpose has been to make sure that the cash value of a QOF point takes into account the size of the practice. A practice with ten thousand patients should, all other things being equal, get twice as much for each QOF point as a practice with five thousand. Changes and mistakes in how this is calculated will cost practices thousands of pounds.

Until April 2013 an “average” practice in England was considered to have 5891 patients ( the three other countries used their own values). Effectively QOF was paid in blocks of 5891 patients. In the first year of QOF a point was worth £120 per 5891 patients, or just a shade over 2p per patient. This value gradually increased over the following years. It didn’t actually matter what an actual average practice list was, we were effectively being paid per patient.It simply made the contract easier to present when larger numbers were used.

Autumn 2012 probably represented the lowest point in relations between the GPC and the Department of Health. The contract imposition at the time was wide ranging. In paragraphs 33-35 of the letter from Richard Armstrong in December 2012 it was proposed to change the 5891 figure based on the actual average list size. This was stated as requiring an increase of 16% - 6834 patients. The actual figure was to be calculated on the first day of January before the QOF year began. In the long term the trend is for average lists to rise if the population increases or smaller practices close. Of course if the cash value of a point stays the same it is spread between a larger number of patients and the value per patient will fall. For one year only there was to be an increasein the value of a QOF point. Mr Armstrong’s letter stated that “This would be a cost neutral change in 2013/14.”

In the end the average list size on 1st of January was 6911 - an increase of 17.3%

When we received the calculation from CQRS last week there was another problem. Whilst both the imposition letter and the Statement of Financial Entitlement (the document that lays out what practices should be paid in exhaustive detail) say that the average list size before the start of the QOF year should be used (i.e. 1st January 2013 for the year just gone) the HSCIC have instead used 1st January 2014. As the trend to larger lists has continued they have used an average list size of 7052 - a 19.7% increase on the old figure. That sixteen percent increase in the value of a point is being stretched even further. For a 10,000 patient practice this would represent more than £3,500 loss.

This loss is simply an incorrect interpretation of the rules and was sorted within a week.

There is, unfortunately, a further sting in the tail. That figure from the first of January 2014 will be used for QOF payments in April 2015. Thanks to HSCIC we now know there will be 2% less money per patient. Even with the new, slimmer, QOF this decrease will wipe out the majority of the 0.28% uplift to GP payments this year.

Letter from the HSCIC now amended to the correct date..

Statement of Financial Entitlement 2013

The "Cost Neutral" promise

Charts can now be copied and downloaded

I had an email a few weeks ago asking why the charts on the site could not be copied or downloaded. The simple answer is that that in order to have interactivity on the charts as the mouse hovers over them and to keep loads relatively low on the server the charts are generated in your browser by javascript. All of the actual work is done by the really easy to use Google Charts library. Because of the way that this is done - it never really exists as a "picture" - it seemed downloading would be impossible. This is a pity because they can be handy to use in documents and presentations. If you have seen any presentations that I have done I use screenshots of the site and these can be a bit of a faff to produce.

On looking a little further I was pleasantly surprised to see that those nice folk at Google had made a print and copy option for the charts. This only happened at the end of January and the site is currently on an even newer (and possibly less stable version) as there was a small bug which Google quickly squashed.

So if you want to download or print a chart just click on the link underneath to get a fully usable print or embeddable png. I would be grateful if you could mention the source when you use this in presentations or documents but otherwise there are no conditions on what you do with them.

2012/13 QOF Data now on the site

All of the QOF data for 2012/13 is now on the main website and will be available for download in a few days. Due to the timings of it being published this year and the time I had available the data for all four countries is going on at the same time.

There are a few small changes this year. Firstly there were new smoking indicators which now applied to all patients who had reached their fifteenth birthday. As there is both an indicator showing how many people were asked, and a second indicator about giving advice to people who said that they smoked we can used the numerator of the first and the denominator of the second to give an idea of the proportion of patients who smoke. This is a little limited as it only applies to patients who were asked but may be useful or at least interesting. This appears at SMO PREV ALL in the numbers. It is entirely a pseudo indicator made up by myself.

I had an interesting email during the year about the prevalence figures on the site. In several disease areas, such as diabetes, osteoporosis or rheumatoid arthritis the register is restricted according to the age of patients. For instance diabetes is 17 or over and nobody goes on the osteoporosis register until they are fifty years old. At the moment the denominator on the site is the whole practice list, rather than just those patients who could possibly qualify for the register. This would tend to underestimate the prevalence in the qualifying population.

There are two sides to this. Using the adjusted denominator makes it easier to compare populations. It is a more realistic measure of disease prevalence although in many cases the actual disease is relatively rare outside these age groups. The vast majority of diabetes, for instance, occurs in later life. Osteoporotic fracture is rare before 50.

There are snags though. Only England produces the adjusted figures and they are listed on the spreadsheets as estimates. It is also easier to see the relative disease burden on practices from the whole list figures. Comparison with earlier years is also easier.

My compromise is to keep the main prevalence figures the same but to include the adjusted figures for English practices. If you click through on diabetes you will see DM 32 Adj which has the same numerator but the adjusted denominator. Not so much a compromise as a fudge - I do both things!

Downloads should be available in a few days. As always the Access version is a bit painful to set up but I will try and push my increasingly aged laptop through it! General QOF stuff (and some stuff that comes to mind but absolutely no pictures of my lunch) please do follow me on twitter. You can see recent tweets to the right of this post.

The year 2013/14 will, for the first time have different indicators in each of the four contries. I have a year to figure out how to deal with this! In the meantime I hope to get a better API up and running. Also we should get some idea of what a new DES for admissions avoidance in England. It may be possible to model some of this from QOF data, but we await the details.

Windows XP in the NHS.

According to a pdf page of statistics Chrome has overtaken IE as the most used browser in the UK and USA and Windows XP is only used by about 10% or web browsers.

Well that is in the real world. I wondered what the NHS looked like. Back in December I did some log file analysis from this site and found that it was very much IE and WinXP that were the dominant browsers from within the NHS. The output from Awstats in still online for NHS operating system and the total operating system use. I also looked at NHS browser and all browser use. I am still on XP at my practice although I use Firefox as my browser. Frankly the NHS has been a bit busy on other things over the last six months so I was not expecting much change.

The new figures are based on visits to this in June. There are figures for all visits and visits from the NHS. The NHS figures are taken from IP addresses of the form 194.176.105.xxx - which is just a quick grep through the logfile.











All VisitorsNHS
December June December June
Windows XP34.9%39.5%84.6%78.9%
Windows Vista7.3%3.4%0.3%0.3%
Windows 728.9%26.3%7.3%15.2%

The figures outside the NHS are, in general, lower as the Windows use is a bit diluted by mobile and Apple operating systems which are unheard of in the NHS. There is an odd rise in Windows XP outside the NHS and this highlights sampling problems. This is the equivalent of conducting a survey of people who happen to walk past your house. It is not truly random.

Nevertheless the NHS is moving off XP but it seems very slow. The countdown to the end of support is coming. You can download a clock from Microsoft, but it doesn't work on XP.

I will try to remember to do the same again in December.

GPPAQ not as fiddly as they would have you believe

Updated with correction 17th May

In this new QOF year there are two indicators added the the hypertension area about the assessment of physical activity. You can see the argument here - physical activity is generally good for the cardiovascular systems and specifically can have a beneficial effect on the blood pressure. There is a second indicator for a brief intervention to encourage exercise.

The chosen method to do this by the GPPAQ form with associated guidlines. The method of calculating this in the paper through that link is a bit fiddly. It involves cross referencing across two tables and is simply slow to do. There is an online version of the questionnaire at Patient.co.uk. Only the first three questions are used, the answers to rest are ignored, which seems a poor repayment of effort.


On closer inspection it appears that this is a needlessly complicated way of working out the result and that it is much easier to just allocate each answer a numerical value and calculate the result at the end. It should be no more complicated than working out the result of one of the "Smash Hits" personality analyses of my youth.

Is your occupation :

  • None or sedentary (0 points)
  • Standing (1 point)
  • Physical Handling (2 points)
  • Heavy lifting work (3 points)

How much physical exercise (jogging/gym/football/swimming but not cycling) do you do weekly?

  • None (0 points)
  • Less than an hour (1 point)
  • One to three hours (2 points)
  • More than three hours (3 points)

Home much cycling do you do a week?

  • None (0 points)
  • Less than an hour (1 point)
  • One to three hours (2 points)
  • More than three hours (3 points)

Just add up the points

  • Zero points - inactive
  • 0.5 to one point - moderately inactive
  • 1.5 to two points - moderately active
  • 2.5 points or more -active

Simple! Of course you may be thinking to yourself that you are sitting in front of a computing device of such power that it would have you burnt as a witch as late as the 1990s and you should not be asked to add up numbers yourself. In that case I have a javascript GPPAQ calculator which is no more or less accurate than the Patient.co.uk on but is less cluttered and I am happy for you to use the code in your own systems under BSD license.

If you use EMIS PCS I have written a protocol (with thanks to EMIS Support for squashing the bug) which asks you the questions and inserts the Read code automatically. You can download it here and then import it through protocol designer.


Are the new indicators worth doing?

Readers of the Autumn issue of GP Business may have seen my calculations of the amount of money payable for meeting each of the current QOF indicators. Although there were huge variations the majority of indicators seem to have been worthwhile from a financial point of view.

In a letter to GPs in England the GPC has urged practices to review whether it will be worthwhile taking part in all of the new indicators.

Where some of these areas, such as rheumatoid arthritis, are completely new it is very difficult to predict the potential rewards.

There are a couple of areas that are easier to estimate, and the headline seems to be that these are greatly underfunded.


Three points are proposed for an annual dietary review in all patients with diabetes. Special training may be needed to allow a GP or nurse to conduct this review. In Scotland it has been confirmed that no further training is required.
The payment for this review will be around £1.55 if the practice reviews ninety percent of patients with diabetes. Even using the costings used by NICE there is only three minutes or so of nursing time available - even less if extra training is required. There is very little detail about how long that this expected to take, but this would seem quite a short time.

The estimates are similar for the annual exercise questionnaire for patients with hypertension. GPs in Scotland will be glad to know that these indicators do not apply to them, having been declared unworkable in negotiation with the Scottish government.

The General Practice Physical Activity Questionnaire (GPPAQ) is fiddly to score and will have to be applied to every patient with hypertension less than 75 years old. If we fairly conservatively assume that this would mean half of all patients on the hypertension register (it is closer to 70% in my practice) then this will work out at around a pound per patient. This is half of the payment for the much simpler depression screening questions and would not cover the cost of a mailshot or telephone survey.

There is a follow up indicator. Patients whom the GPPAQ has assessed as “less than active” should have a brief intervention. Even saying fairly generously that half of all patients are active there are only two pounds available to deliver the brief intervention. If fewer patients are “active” then this amount falls. This would have to be very brief to make this a worthwhile activity.

In one of the strange quirks which make QOF so complicated that less success in getting patients to complete the questionnaires will increase the cash available for each brief intervention.

Practices have tended to chase every possible QOF point in the past. The very high levels of point scoring - over 97% in the last few years - suggest that practices have aimed at QOF in its entirety. The lowest levels of achievement have been where evidence has been weak and hassle has been greatest - the PHQ9 indicators.

Where there is a poor business case the GPC is right that practices should consider carefully whether it is reasonable to work towards these new indicators. Picking and choosing will be a considerable change to the way that practices approach the QOF.

More on QOF contract changes

The full details of the English government's enforced contract changes were released today. You can read my annotated version of the letter. Most of the changes were what was expected, certainly in relation to the QOF. There is plenty in the medical media about the changes as well. Here are a couple of bits you may have missed.

There is a plan to renumber the indicators to make them neater. So this years diabetes 31 may be next years diabetes 5. Seems a recipe for confusion and generating extra work at the Commissioning Board to me. Actually seems like a bit of work reprogramming the website to me too!

More serious is the CPI in QOF. For practices with average prevalence there is a certain amount of cash per patient for each point. When the QOF was originally devised this was not really a headline amount so it became £120 or so per typical practice of 5891 patients (5885 in Wales, 5095 in Scotland and 4937 in Northern Ireland)

£120 per 5891 patient is simply a way of saying 2p(ish) per patient. As the value of points rose over the years so did this payment per patient. Practices got bigger on average (whatever the pros and cons, smaller practices are simply not in fashion) but that did not matter. It was still 2p per patient. The government was paying no more, and no less than it planned to for each registered patient.

In the letter the government plans to increase the average list size by 16% and the payment per point by 16%. They say this will be cost neutral in 2013/14, and it will. They are still paying 2p per patient.

In each future year the average will likely rise again. However the value per
point will be by negotiation. Essentially there is not guarantee that the payment per patient will not fall as the average practice size rises. It will depend on negotiations producing a similar rise in point value.

I can see no reason to change this other than as a little time bomb in the future (but then I can't see why you would want to renumber all the indicators. I may not be in the DH mindset.)

This is an obscure part of the calculation although it can't really be said to be hidden - it is there three times- it is easily missed.