Skip to content
Home bread crumb About the data

About the data

How to use Healthier Lives

We’ve created a video tutorial explaining how to get the best out of Healthier Lives. This video shows you how to use the online tool and what the website can tell you about health care in your area.

The data presented in Healthier Lives: Diabetes have been drawn from two major sources in the NHS:

These data are not new, they have been available in the public domain but this is the first time they have been published in this form making information easy to access, view and compare CCGs, local authorities and general practices.

The data can tell us about:

The QOF data are for the financial year 2014/15. Data covering complications of diabetes are for the period April 2010 to March 2012. The tool will be updated as new data becomes available, at least annually. We have not used the more usually published achievement scores, but the intervention rates which look at the extent to which practices meet targets in ALL people with diabetes. For more information see the Quality and Outcomes Framework (QOF) – 2014/15 Frequently Asked Questions.

QOF data is only collected for people age 17 and over. It includes people with type 1 and type 2 diabetes. Ninety percent of people with diabetes have type 2 diabetes. The audit data used in the tool refer both to type 1 and type 2 diabetes and covers all ages.

The data covers the whole of England and are broken down by:

We have excluded 21 practices which have a list size which is not larger than 900 patients.

How the data is organised

We have grouped the data into 4 themes

The colour scheme and how to interpret it

We have used a colour scheme for the maps and tables to denote where local authorities, CCGs or practices are statistically significantly different from the national average. To do this we have used funnel plots which is standard methodology for making comparison between institutions or areas. (D. J. Spiegelhalter, 2005; D. Spiegelhalter, 2003).

Funnel plots measure the expected variation about a target value (often the national average) for the range of sample sizes of the areas or institutions to be compared. This allows us to distinguish between variation due to chance because of differences in sizes of general practices or areas and variation unlikely to be random and worthy of investigation.

The colours are as follows.

For CCGs and local authorities:

For general practices:

The diagram above gives an example (in this case for urine testing). The grey lines define a typical funnel shape (on its side) which shows the expected level of uncertainty around the national average for the range of population sizes covered by the data (e.g. people with diabetes). The funnel shape arises because there is more uncertainty when basing estimates on smaller sample sizes. The coloured blobs are the areas being compared. Those which sit above the upper grey line are coloured green; those below the lower coloured line are coloured red; and this between the lines i.e. in the funnel, are coloured amber. The colours derived from the funnel plots are then plotted on the map to represent the geographical variation in the data for the particular indicator being shown. Improvement is measurable on a funnel plot (and therefore maps) by a reduction in the number of “red” areas and an increase in the number of “greens”.

For complication rates we compare the 95% confidence interval for excess complications in people with diabetes with the average for the general population to be (95%) certain that the result was not found by chance. We have applied the same method (confidence interval (95%) overlapping the England value) to the indicator ‘Physical inactivity’. The indicators ‘Deprivation score’, ‘People over 65’, ‘Diabetes prevalence’, and ‘Percentage of population who identify their ethnicity as Asian or Asian British’ are displayed by quintile.

For all complications, people with diabetes have higher rates than people without diabetes. On average people with diabetes are 1.34 times as likely to have a stroke and 3 times as likely to have a major lower limb amputation as people without diabetes.

In this case red represents a significant excess complication rate compared to the excess complication rates in England as a whole; amber represents the national average excess complication rate; and green represents a better than average complication rate. This does not however mean that people with diabetes in areas coloured green have complication rates lower than the general population but they have fewer complications than people with diabetes in other areas.

An area may appear grey on the map if there is no data published, or numbers of events are too small or where there are concerns about data quality.

Why does Healthier Lives present QOF data without taking account of exceptions ?

There are two valid methods to calculate QOF indicators. We have chosen to show the proportion of patients receiving the intervention because:

From a public health perspective we are more interested in the actual proportion of patients receiving the intervention, i.e. the proportion of all patients with this condition who were treated. The HSCIC (QOF FAQs, p.16-17) states "Percentage of patients receiving the intervention, gives a more accurate indication of the rate of the provision of interventions as the denominator for this measure covers all patients to whom the indicator applies, regardless of exception status."

We consider this to be the better comparable indicator because, while there are very good reasons why a patient might not be treated (such as terminal illness), a generous interpretation of exception rules can also be used to improve practice performance.

The tool is intended to highlight variation and encourage conversation about the causes of variation. We are not suggesting that every practice should, or can, achieve a 100% intervention rate for every indicator - clearly there are patients it would not be desirable to be included - however it is clear that there is unwarranted variation in exception rates and the data is not available for us to make adjustments.

Triangulation with other sources of primary care data such as the National Diabetes Audit (NDA) support this approach. For those QOF indicators which match NDA indicators, a higher degree of correlation was found with intervention rates than with achievement scores, so intervention rates seem a better measure of true performance.

Action to tackle diabetes 

The quality of care and outcome for people with diabetes can be improved. There are good examples across the country, for example the work in Tower Hamlets, a case study for which can be accessed here.

  1. Ensure that patients are educated about their condition at a level that suit their needs and also delivered by trained staff on a regular basis. (1,2)  
  2. Work on prevention of disease and limiting the risk of complications by
  1. Risk assessment:
  1. Keep updated records of people's level of risk and create a recall system which will allow patients to be contacted and invited for regular reviews. (5)
  2. Improving monitoring (i.e. feet check, kidney function, urine dip)
  1. Involve patients more in the planning of their care, therefore encouraging patients to engage more and be better motivated to achieve better control of their health. (1,2)
  2. Encourage patients to have their Health check reviews particularly targeting the Asian and Afro-Caribbean population. (4,5)
  3. Improving access to healthcare for people who do not routinely use them. This can range from the homeless to people who are carers or disabled. (6)
  4. Engage with the public and patients to find out about needs and demands, obtaining feedback on quality of care so as to improve the current system and engage patients more. (6,7)
  5. Regularly perform searches on the GP database to look at patients to frequently fail to collect repeat prescriptions or attend follow-up appointments. (7)

Once those patients have been identified, maybe phone them or send a letter to invite them for an appointment. This will also be an opportunity to check for compliance with medication.



3 sigma = 3 standard deviations (SD). This statistical term means that if the size of the population that we are looking at is large we would only expect approximately 0.15% (1 in 370) values to lie higher than the average plus 3SD by chance and only approximately 0.15% (1 in 370) to lie below the average minus 3 SD.