Using risk tables to assess cardiovascular risk in type 2 diabetes has drawbacks

EDITOR—Hall et al describe the use of primary prevention risk tables in type 2 diabetes.1 We agree that targeting cardiovascular risk in diabetes is a priority, but their proposals imply an overreliance on risk scores as the sole determinant of cardiovascular risk assessment in diabetes. This approach might lead to withholding treatment in some people whose risk is underestimated by risk scores.

Unfortunately, the decision at what level of cardiovascular risk to start lipid lowering treatment in diabetes is not straightforward. The authors' oversimplistic approach, although convenient, is unscientific and flies in the face of epidemiological evidence which suggests that type 2 diabetes should be regarded as a disease group for secondary rather than primary prevention.2

Using the Framingham equation to evaluate cardiovascular risk in diabetes entails caveats. These include a low baseline prevalence of diabetes in the Framingham cohort and the omission from the equation of triglyceride concentration, an important determinant of cardiovascular risk in type 2 diabetes.3 The low prevalence of diabetes in the cohort leads to wide confidence intervals in the predicted risk. Thus, in a diabetic man with average risk factors, the upper 95% confidence interval crosses the 15%, 10 year threshold from the age of 40 onwards.3

People with risk scores below a chosen cut-off point may therefore have a higher true risk. The proposals by Hall et al would lead to a rigid prescribing protocol whereby all patients with scores above a threshold (for example, 15%) would receive treatment, whereas those below would not. Furthermore, an overemphasis on the risk score might be at the expense of ignoring other key factors not represented by the Framingham risk equation, such as ethnic group, family history, microalbuminuria, and triglyceride concentration.

Authors' reply

EDITOR—The purpose of our paper was to identify whether there was clinical value in having cardiovascular risk scores. We wanted to identify whether having an integrated single score of macrovascular cardiovascular risk highlights to clinicians that a clinical issue needs addressing, which influences their prescribing habits. Our paper indicates that this is the case and that in the setting of a busy clinic having to assess a multitude of individual risk factors may result in cardiovascular risk being overlooked. Having established that an integrated score is useful to practising clinicians, the next challenge is to identify an appropriate risk score to use.

Smith and Corrall correctly indicate that the New Zealand risk score, and others based on Framingham, all underestimate the cardiovascular risk in diabetes, although data from Tayside indicate that this is probably not to the extent suggested by Haffner et al. 1 2 Also, the level of risk chosen to start treatment is an arbitrary cut-off point which can be adjusted—for example, to 15% from 20%—if thought desirable. The real answer is to define the epidemiology of cardiovascular risk in diabetes more accurately so that more accurate tables can be developed.