Predicting risk of type 2 diabetes in England and Wales: prospective derivation and validation of QDScore
BMJ 2009; 338 doi: https://doi.org/10.1136/bmj.b880 (Published 18 March 2009) Cite this as: BMJ 2009;338:b880
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
I would like to make two points on the excellent attempt by Hippisley
-Cox and colleagues (1) at deriving a tool to estimate risk of developing
type 2 diabetes incorporating ethnicity and deprivation. The design of a
web based calculator (www.qdscore.org) available for use by patients for
self assessment of their risk is an interesting concept.
Firstly, I would like to request the authors to clarify the reason
for using the Townsend deprivation score. This has been replaced for a
while with the Index of Multiple Deprivation(2) which is a better and more
refined measure of deprivation and is widely used to assess and address
health inequalities in England and Wales.
Secondly, all variables used in the algorithm except the deprivation
score are measured at an individual level and therefore suitable for
individual risk assessment. Deprivation indices based on postcode are
useful to estimate population risk but can cause confusion when used to
predict individual risk. The same individual can change their risk by
simply moving from one post code to another. For example, a 43 year old
Indian female with family history of diabetes and BMI of 24kg/m2 could
reduce her risk from 5% to 3% by moving from London (SW3 2AS) to Suffolk
(IP33 2QZ).
Risks for individuals should be based on personal indices of
socioeconomic status such as educational level or occupation or income
rather than postcode deprivation. It is therefore important that one or
more of such data are collected routinely to enable individual risk
estimations.
1. Hippisley-Cox J, Coupland C, Robson J, Sheikh A, Brindle P.
Predicting risk of type 2 diabetes in England and Wales: prospective
derivation and validation of QDScore
2.
http://www.communities.gov.uk/communities/neighbourhoodrenewal/deprivati...
Competing interests:
None declared
Competing interests: No competing interests
Trying out my own data on the risk calculator (I had to lie about my
age) produced a 10 x 10 table of smileys- 13 of which had frowns, placed
at random in the 100 squares. I think this is a well meaning, but doomed,
attempt to transfer a legitimate statistical interpretation, ie of a large
number of people, to the risk of an event to an individual. This hooker
has been recognised for years, and no one has solved it. There is also
Hume's problem: there can be no certainty that what happened under certain
circumstances in the past will happen again in the future ( a mantra much
employed by financial advisers.) The changing risks associated with the
Framingham equation over the years and in different places testify to the
truth of this. Finally, we dont know either whether ameliorative measures
will work (unless you already know the glucose tolerance) nor at what risk
level they might usefully be employed.
All this is a pity because the study per se is jolly good!
Competing interests:
None declared
Competing interests: No competing interests
Even though the QDScore is the first algorithmn to calculate the 10
year risk for type 2 diabetes, how will individuals or GPs respond to a
positive result ? Does the knowledge lead to an effective treatment ? If
so, how does the treatment relate to a pathophysiology ?
A major problem is that the authors failed to recognise that there
was a common feature in most of the risk factors. Thus,aging, high
BMI,smoking status, hypertension and cardiovascular disorders are all
associated with an increase in blood viscosity and a reduction in red cell
deformability.
Furthermore, the selected age range (25 to 79 years) appears to have
been treated as a continuum, but during the aging process, from about 50
years of age, blood viscosity rises and red cell deformability is reduced
in concert with an increase in fibrinogen levels. And the selected range
excludes children and adolescents who have been diagnosed with type 2
diabetes. In childhood obesity, a high BMI has been shown to be
associated with similar changes of fibrinogen levels, blood viscosity and
red cell deformability as occurs in the aging process. (1) Others have
published similar findings. Ernst et al (2) have reported that in grossly
obese individuals on a low calorie diet, blood viscosity was reduced and
red cell deformability was increased.
The effects of smoking were assessed as smokers or non-smokers, but
the increase in blood viscosity is related directly to the numbers of
cigarettes smoked. So the risk from smoking 4 cigarettes daily is greatly
different from that of smoking 40 cigarettes daily. Cessation of smoking
reverses the smoking-related changes.
In 1930 attention was drawn to the fact that blood pressure was
related directly to blood viscosity and subsequent studies have confirmed
that relationship, although it has failed to gain clinical recognition.
There is a large literature which documents the association of increased
blood viscosity with cardiovascular disorders. Although it was recognised
that social deprivation was associated with other risk factors (diet,
obesity, smoking) it was not recognised that the effects of all three
factors would be cumulative, which would explain the poor outcome. It is
of some relevance that there are published reports which show that type 2
diabetes is associated with increased blood viscosity and poorly
deformable red cells. Those changes may be exacerbated by similar changes
in risk factors.
Therefore of the 9 risk factor used in the development of the
QDScore, 7 factors share the common feature of altered blood rheology.
How does this impinge on the QDScore ? What benefits will flow from
calculating a QDScore, in comparison with simply recognising that altered
blood rheology is the major factor even in the pre-diabetic state ? An
important fact is the blood rheology changes are potentially treatable.
Huang et al (3) reported that Gingko biloba extract (Egb 761) lowered
blood viscosity and improved red cell deformability and, "...it
effectively improved retinal capillary flow rate in type 2 diabetic
patients with retinopathy." Because of the beneficial effects of the
omega-3 fatty acids in fish oil in other disorders, it is surprising that
this has not been shown in type 2 diabetes. A 1994 report (4) concerning
the consumption of seal oil and salmon by Alaskan Natives, concluded,
"Consumption of seal oil and salmon, high in omega-3 fatty acids, appears
to lower the risk of glucose intolerance and is a potentially modifiable
risk factor for NIDDM in Alaskan Natives."
While it is uncertain to what extent the QDScore would be affected by
recognising that 7 of the 9 risk factors shared a common feature, it does
raise questions about the utility of the algorithm. For that reason, when
faced with pre-diabetic patients, GPs could suggest that they explore the
potential of Gingko biloba extract (Gbl761) at 180mgs daily or fish oil at
6 to 10 grams daily, with the objective of normalising the flow properties
of their blood.
References.
1. Cacciari E, Balsamo A, Palareti G, et al. Haemorheologic and
fibrinolytic evaluation in obese children and adolescents. Eur J Pediatr
1988; 147: 381-4.
2. Ernst E, Weihmayer T, Matrai A, et al. Changes in blood rheology of
grossly obese individuals during a very low calorie diet. In J Obesity
1989; 13 (Suppl 2):167-8.
3. Huang SY, Jeng C, Kao SC, et al. Improved haemorrheological
properties by Gingko biloba extract (Egb761) in type 2 diabetes mellitus
complicated with retinopathy. Clin Nutr 2004; 23: 615-21.
4. Adler AI, Boyco FJ, Schraer CD, et al. Lower prevalence of impaired
glucose tolerance and diabetes associated with daily seal oil or salmon
consumption among Alaska Natives. Diabetes Care; 1994: 1498-501.
Competing interests:
None declared
Competing interests: No competing interests
This is an interesting and innovative paper by Hippisley-Cox et al
(1). The QDScore algorithm suggested by the author to predict the
diagnosis of type 2 diabetes mellitus has focussed on important risk
factors such as age, family history, obesity and ethnicity.
Central obesity is a stronger risk factor for type 2 diabetes than
overall obesity. In Harvard study, waist circumference (WC) is a better
predictor of type 2 diabetes and cardiovascular complications than body
mass index (BMI). This was shown in a 13 year follow up study published in
2005 (2). Although WC is not routinely recorded in the clinical computer
system in the UK, it would be interesting to see if there is a change in
the 10 year risk when WC measurements are used instead of BMI.
Women with previous gestational diabetes are more likely to develop
type 2 diabetes. Feig et al discussed that the risk of developing type 2
diabetes after gestational diabetes increases over time (4.9 % at 15
months postpartum, 13.1 % at 5 years postpartum, 18.9 % at 9 years
postpartum) (3). Surely, the significant increase in risk means previous
gestational diabetes should be included in the risk stratification
algorithm.
References:
1. Hippisley-Cox J, Coupland C, Robson J, Sheikh A, Brindle P.
Predicting risk of type 2 diabetes in England and Wales: prospective
derivation and validation of QDScore. BMJ 2009;338:b880, doi:
10.1136/bmj.b880
2. Wang Y, Rimm EB, Stampfer MJ, Willett WC, Hu FB. Comparison of
abdominal adiposity and overall obesity in predicting risk of type 2
diabetes among men. Am. J. Clinical Nutrition, Mar 2005; 81: 555 - 563.
3. Feig, DS, Zinman, B, Wang, X, Hux, JE. Risk of development of
diabetes mellitus after diagnosis of gestational diabetes.CMAJ 2008;
179:229.
Competing interests:
None declared
Competing interests: No competing interests
Dear Sir,
The report by Hippisley-Cox et al is timely, given the twin epidemics
of diabetes and obesity sweeping the world and causing much suffering.
While I await further validation of the model with great interest, it is
already a potentially useful tool for risk stratification and patient
education.
I have one major concern about the online calculator
(www.qdscore.org) as it currently stands. While most of the questions
clearly describe risk factors for the development of diabetes, the wording
of the question asking 'Are you on blood pressure treatment' strongly
suggests that it is the treatment rather than the blood pressure that
increases the risk of diabetes. This misinformation is potentially
compounded if a user of the calculator tries the calculation with and
without this check-box ticked, where the risk of diabetes appears lower if
the person is not receiving blood pressure lowering agents. I believe
there is a substantial risk that the calculator as it stands could
therefore cause some individuals to cease their required blood pressure
treatment in the mistaken belief that this might reduce the risk of
diabetes, when it will instead potentially increase the risk of
cardiovascular and other complications related to their blood pressure
level.
I am sure this is not the intention of the authors, and expect that a
minor change to the wording will clarify the issue (an alternative
phrasing could be: Do you have high blood pressure requiring treatment?),
and urge the authors to make this change as soon as possible.
Yours sincerely,
Dr Vlado Perkovic
Competing interests:
None declared
Competing interests: No competing interests
Major limitations with QFracturesScores
I read with interest the paper on the prediction of fracture in
England and Wales by Hippisley-Cox and co-workers [1]. There are several
major limitations with the analyses. Risk factors were only assessed at
baseline (e.g. patient’s registration date), not taking into account any
changes in risk factor status over 8 years of follow-up. A patient, who
developed incident cardiovascular disease in the first year of follow-up,
would be incorrectly classified as a patient without for the total
duration of follow-up. Furthermore, it should be well known that GP
medical records record a higher rate of disease in the first year of
registration, reflecting recording of prevalent conditions [2]. In the
analyses by Hippisley-Cox et al, such records would be incorrectly
classified as incident events. In contrast, most researchers would
typically start follow-up, at least one year after registration.
Assessment of risk factors in the time before registration (at baseline),
likely leads to underrecording (e.g., prescriptions are not recorded when
the patient is not registered in the practice). The choice of risk factors
for QFracturesScores was inconsistent with literature. As examples, asthma
was selected rather than chronic obstructive pulmonary disease [3,4] (more
prevalent than asthma in elderly). Medical conditions such as dementia and
use of psychotropic drugs such as anticonvulsants, selective serotonin
receptor inhibitors, anxiolytics/hypnotics, and antipsychotics, or the
daily dose of oral glucocorticoids, were not considered. Previous work on
a fracture risk score in GPRD was not discussed [5]. In my opinion, given
these limitations, the application of QFracturesScores and its clinical
calculator on www.qfracture.org, should be limited, until these
methodological concerns have been tested.
References
[1] Hippisley-Cox J and Coupland C. Predicting risk of osteoporotic
fracture in men and women in England and Wales: prospective derivation and
validation of QfractureScores. BMJ 2009;339:b4229
[2] Lewis JD, Bilker WB, Weinstein RB, Strom BL. The relationship
between time since registration and measured incidence rates in the
General Practice Research Database. Pharmacoepidemiol Drug Saf.
2005;14(7):443-51.
[3] de Vries F, van Staa TP, Bracke MS, Cooper C, Leufkens HG,
Lammers JW. Severity of obstructive airway disease and risk of
osteoporotic fracture. Eur Respir J. 2005 May;25(5):879-84.
[4] Vestergaard P, Rejnmark L, Mosekilde L, Fracture risk in patients
with chronic lung diseases treated with bronchodilator drugs and inhaled
and oral corticosteroids Chest 2007;132(5):1599-1607
[5] van Staa TP, Geusens P, Kanis JA, Leufkens HG, Gehlbach S, Cooper
C. A simple clinical score for estimating the long-term risk of fracture
in post-menopausal women. QJM. 2006 Oct;99(10):673-82.
Competing interests:
All authors confirm that they are not involved in any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in this manuscript.
Funding:
The department of Pharmacoepidemiology and Pharmacotherapy, Utrecht Institute for Pharmaceutical Sciences has received unrestricted funding for pharmacoepidemiological research from GlaxoSmithKline, Novo Nordisk, the private-public funded Top Institute Pharma (www.tipharma.nl, includes cofunding from universities, government, and industry), the Dutch Medicines Evaluation Board, and the Dutch Ministry of Health.
Competing interests: No competing interests