Intended for healthcare professionals

Rapid response to:

News Roundup [abridged Versions Appear In The Paper Journal]

Diabetes, not obesity, increases risk of death in middle age

BMJ 2006; 333 doi: (Published 28 September 2006) Cite this as: BMJ 2006;333:672

Rapid Response:

Identifying patients with diabetes in the QOF - two steps forward one step back

The introduction of the Quality and Outcomes Framework for General
Practice (QOF) in April 2004 has resulted in a step change in the recorded
quality of care for patients with diabetes in the UK1. For the first time,
national data on the prevalence of diabetes in almost every practice in
England is available via QMAS.

Patients are included on the QOF diabetes register if they have a
specific set of Read codes (diagnostic codes) recorded within their
primary care electronic health record prior to the search date. In April
2006, however, the definition for identifying patients with diabetes
changed from using a high level general Read code (C10% – diabetes
mellitus), to one of two more specific Read codes (C10E and C10F)2. The
rationale was to allow the future development of QOF indicators more
closely aligned to NICE guidance which distinguish type one and type two

The aim of this study is to determine the effect of the change in
definition on the recorded national prevalence of diabetes and the likely
implication for patient care.


We used version 11 of the QRESEARCH database for this analysis3. We
included UK practices where the last upload date was on or after 01 June
2006 to ensure data transmission was complete. We identified all patients
currently registered on the 01 June 2006 and used this population as the
denominator for our diabetes prevalence rates.

We compared two methods to identify patients with diabetes based on
the presence of diagnostic codes within their primary care electronic
record. Method one was based on the original QOF definition i.e. patients
with a C10% read code recorded in their electronic health records prior to
01 Jun 2006. Method two was based on the new QOF definition for diabetes
i.e. two more specific read codes, C10E or C10F. We compared the overall
and age specific prevalence rates per 100 registered patients using both
definitions. We estimated the number of patients in the UK who would no
longer be identified as having diabetes according to the new QOF
definition using ONS mid year population estimates for 2004.


480 QRESEARCH practices were included in this analysis. There were
3.63 million patients registered on 01 Jun 2006 of whom 128,421 patients
met the original QOF definition for diabetes giving a crude prevalence of
3.54%. However, applying the new QOF definition, we only identified
100,231 patients giving a crude prevalence of 2.76%. This is equivalent to
a 22% drop in prevalence.

Our analysis demonstrates that the largest fall occurred at the
extremes of age with a 27% drop in patients aged 90 plus and a 26% drop in
those aged under 5 years. Table 1 shows the estimated numbers of patients
with diabetes using the old QOF definition and the new definition.
Overall, unless action is taken, an estimated 457,000 patients with
diabetes in the UK will be lost from the existing general practice
diabetes registers.


The change in the QOF definition of diabetes, which came into effect
on 01 April 2006, is likely to result in almost half a million patients in
the UK with a genuine diagnosis of diabetes may no longer recalled for
diabetes care by their GP. This failure is most likely to affect patients
at the extremes of age.

Some patients with diabetes will remember to consult their GP for
their annual checkups. However, there is a risk that a significant number
may not. Therefore there is an urgent need to consider reverting to the
original diagnostic codes in QOF used to identify patients with diabetes
so that the disease registers remain complete. A separate indicator for
the proportion of all diabetes patients with a more specific diagnostic
code would achieve the desired goal without risking patient falling off
the register altogether.

Alternatively each GP needs to identify every patient with diabetes
who no longer appear on the QOF recall lists and update their electronic
records with one of the two more specific diagnostic codes. The GP will
need to keep the date of the original diagnosis (to avoid corrupting data
needed to determine the incidence of diabetes).

There has been a global and sustained increase in the prevalence and
burden of disease due to diabetes4, exacerbated by the ageing population4
and the concurrent epidemic in obesity5. However at a national level,
unless appropriate action is taken it will appear that the prevalence of
diabetes has fallen and yet this will just be an artefact due to a recent
change in definition. Policymakers and those who review QOF need to
anticipate unintended consequences of changes in read code definitions on
patient care.


We acknowledge the contribution of Dr David Stables, co-founder of
QRESEARCH and clinical director of EMIS (the leading supplier of general
practice computer systems in the UK). We acknowledge the EMIS practices
who contribute data for free.

  Table 1 Ageband	estimated number of patients with diabetes in UK 
(method 1)estimated number of patients with diabetes in UK (method 
2)estimated number  of patients lost from diabetes registers
0-4	1,111	819	292
5-9	5,098	3,996	1,101
10-14	9,918	7,610	2,308
15-19	14,798	11,094	3,704
20-24	16,939	13,571	3,368
25-29	19,890	15,508	4,381
30-34	33,751	26,613	7,138
35-39	55,081	43,057	12,024
40-44	82,229	64,700	17,529
45-49	114,760	90,875	23,886
50-54	157,887	124,704	33,184
55-59	216,915	171,608	45,307
60-64	228,708	178,987	49,720
65-69	283,563	220,473	63,090
70-74	302,968	237,100	65,868
75-79	260,543	203,030	57,512
80-84	180,942	139,722	41,220
85-89	72,625	54,978	17,647
90 plus	30,329	22,062	8,267
total	2,088,053	1,630,508	457,546

Competing interests:
JHC is a general practitioner, professor of general practice and an unpaid director of QRESEARCH which is 50% owned by the University of Nottingham and 50% by EMIS. SOH is an unpaid director of QRESEARCH and a full time employee of EMIS. QRESEARCH undertakes research commissioned by government organisations including the Health and Social Care Information Centre, National Audit Office, Disability Rights Commission, Health Protection Agency and the Department of Health

Competing interests: Table 1 Ageband estimated number of patients with diabetes in UK (method 1)estimated number of patients with diabetes in UK (method 2)estimated number of patients lost from diabetes registers0-4 1,111 819 2925-9 5,098 3,996 1,10110-14 9,918 7,610 2,30815-19 14,798 11,094 3,70420-24 16,939 13,571 3,36825-29 19,890 15,508 4,38130-34 33,751 26,613 7,13835-39 55,081 43,057 12,02440-44 82,229 64,700 17,52945-49 114,760 90,875 23,88650-54 157,887 124,704 33,18455-59 216,915 171,608 45,30760-64 228,708 178,987 49,72065-69 283,563 220,473 63,09070-74 302,968 237,100 65,86875-79 260,543 203,030 57,51280-84 180,942 139,722 41,22085-89 72,625 54,978 17,64790 plus 30,329 22,062 8,267total 2,088,053 1,630,508 457,546

03 October 2006
Julia Hippisley-Cox
Shaun O'Hanlon
University of Nottingham