Letters

Prediction of cardiovascular risk

BMJ 1999; 318 doi: https://doi.org/10.1136/bmj.318.7195.1418 (Published 22 May 1999) Cite this as: BMJ 1999;318:1418

Program is not suitable for diabetic patients

  1. Andrew Zambanini, Senior registrar in clinical pharmacology,
  2. Martin R Smith, Specialist registrar in diabetes and endocrinology,
  3. Michael D Feher, Senior lecturer in clinical pharmacology
  1. Section of Clinical Pharmacology, Imperial College School of Medicine and Beta Cell Diabetes Centre, Chelsea and Westminster Hospital, London SW10 9NH rnkk003{at}cxwms.ac.uk
  2. Clinical Pharmacology Unit, EA 643, Claude Bernard University, BP 3041, 69394 Lyons Cedex 03, France
  3. Centre for Clinical Pharmacology, Department of Medicine, University College London, Rayne Institute, London WC1E 6JJ

    EDITOR—Hingorani and Vallance describe a clinic based computer program for the management of cardiovascular risk factors.1 However, the program has limitations for diabetic patients.

    This program and previous methods, together with the recent joint British recommendations,2 are based on algorithms derived from the Framingham heart study of 5573 people aged 30-74 years originally screened in 1968. However, only 4% (237 subjects) of this cohort had diabetes, as defined by a random glucose concentration >9 mmol/l or the use of diabetic treatment. Importantly, diabetes was not classified according to type, and the presence of proteinuria or microalbuminuria (important cardiovascular risk factors in diabetes) were not included.

    The prevalence and characteristics of other risk factors further highlight that people with diabetes are not average patients3; “diabetic dyslipidaemia,” increased prevalence of hypertension, and ethnic origin (for example, south Asian) contribute to an increased atherogenic potential. Despite the strong link between triglyceride concentration and coronary heart disease in diabetes, this important risk factor was omitted from the program.

    The glycaemic state was included in the program as a risk factor both as “diabetes” and as a blood glucose concentration. The impact of glycaemia on cardiovascular events is not dichotomous, but, like blood pressure, has a continuous relation.4 The inclusion of blood glucose as a variable is an important addition to risk assessment.

    The present program and other methods derived from Framingham data have inconsistencies for the primary prevention of cardiovascular disease in diabetes. This is especially true for patients with type 2 diabetes, who have a 2 to 5-fold increase in age adjusted mortality from cardiovascular disease. The emerging evidence is that patients with diabetes should be treated as recommended for secondary prevention.5

    References

    Hypothesis of program is flawed

    1. François Gueyffier, Research fellow,
    2. Jean-Pierre Boissel, Head
    1. Section of Clinical Pharmacology, Imperial College School of Medicine and Beta Cell Diabetes Centre, Chelsea and Westminster Hospital, London SW10 9NH rnkk003{at}cxwms.ac.uk
    2. Clinical Pharmacology Unit, EA 643, Claude Bernard University, BP 3041, 69394 Lyons Cedex 03, France
    3. Centre for Clinical Pharmacology, Department of Medicine, University College London, Rayne Institute, London WC1E 6JJ

      EDITOR—Hingorani and Vallance advocate using a computer program to predict the cardiovascular risk of individuals and estimate the benefit expected from treatment.1 They assume that the risk ratio corresponding to a decrease of a risk factor in a given individual is the same as the risk ratio between two people with the same difference in risk factors.We call this assumption the isotropy hypothesis.2 The isotropy hypothesis is unlikely to be true for several reasons. Firstly, for a given individual, the reduction in risk will not be proportional to the reduction in blood pressure for the range of blood pressures that treatment could produce. In other words, we do not believe that a person with an initial systolic blood pressure of 170 mm Hg would obtain a relative risk reduction of stroke of about 50% if his blood pressure reached 155 mm Hg with treatment, about 75% if it fell to 140 mm Hg, and 88% if it fell to 115 mm Hg. Secondly, for coronary events, randomised controlled trials have reported differences between relative risks observed and expected from a same blood pressure reduction,3 which raises doubts about isotropy. Thirdly, a recent trial showed that using drugs to reduce blood pressure further than with conventional treatment has no effect on cardiovascular risk.4 We believe this approach may be dangerous and misleading. For example, it is simple to show a smoker what his risk would have been if he did not smoke However, this lower risk may not be the value he would achieve by giving up smoking, and is certainly not what a doctor can expect, on average, from interventions against smoking.

      Hingorani and Vallance's proposal to determine patients' targets for blood pressure or cholesterol concentration based on the desired risk reduction may lead to unrealistic expectations and in some cases dangerous behaviour by prescribers or patients.

      We favour the approach proposed by Jackson.5 A patient's absolute risk reduction is derived by multiplying the predicted risk by the relative risk reduction estimated from a systematic overview. This estimate of relative risk reduction could be modulated according to the results of a search for modifiers of treatment effect, ideally performed on an individual patient database from all available clinical trials. Such modifiers may include the intensity of reduction of the risk factor.

      References

      Authors' reply

      1. Aroon D Hingorani, British Heart Foundation intermediate fellow,
      2. Patrick Vallance, Professor
      1. Section of Clinical Pharmacology, Imperial College School of Medicine and Beta Cell Diabetes Centre, Chelsea and Westminster Hospital, London SW10 9NH rnkk003{at}cxwms.ac.uk
      2. Clinical Pharmacology Unit, EA 643, Claude Bernard University, BP 3041, 69394 Lyons Cedex 03, France
      3. Centre for Clinical Pharmacology, Department of Medicine, University College London, Rayne Institute, London WC1E 6JJ

        EDITOR—Any program designed to calculate cardiovascular risk is inevitably limited by the original data gathering process used to derive the risk equations. Despite Zambanini et al's concerns about the limited number of diabetic patients included in the original Framingham cohort, the Framingham risk equation is proving to be a robust tool for estimating risk in the United Kingdom.1 One advantage of a computer based approach is that as new risk factors are identified and estimates of risk further refined, the appropriate changes can be included in the predictions. We hope that further data on both diabetic and non-diabetic patients will be included in the program as they become available. All risk calculations on patients stored in the database will then be refined automatically.

        We also considered it important to design a system which not only provided an assessment of risk but also indicated the likely maximum benefits of intervention. Gueyffier and Boissel raise doubts about the validity of such calculations, but our comparison of computer predictions with data from the West of Scotland coronary prevention study and recently published meta-analyses2 support the view that, for cholesterol lowering at least, individuals adopt a new level of risk predicted from the Framingham cohort study. As we explain in our paper, this is also likely to be true for stopping smoking, although the effects of blood pressure reduction on risk of coronary heart disease may be overestimated. Gueyffier and Boissel's proposal that the benefits of risk factor intervention be estimated from a systematic overview of intervention trials is a reasonable alternative, and our program could be adapted to perform such a task. Either way, the predictions for blood pressure intervention would be tested against trial data.

        In designing this program, our aim was to take forward theoretical discussions about multiple risk factors, risk calculations, and the implementation of thresholds for statin and antihypertensive treatment. In particular, we felt there was a need to develop a tool which would aid the practising doctor and “operationalise” the extensive available evidence on risk factors and intervention.3 The level of interest which our paper has generated, particularly inquiries about the availability of the program, seems to endorse our view that there is a great clinical need for a program such as ours.

        References

        View Abstract

        Sign in

        Log in through your institution

        Free trial

        Register for a free trial to thebmj.com to receive unlimited access to all content on thebmj.com for 14 days.
        Sign up for a free trial

        Subscribe