Analysis of predicted coronary heart disease risk in England based on Framingham study risk appraisal models published in 1991 and 2000BMJ 2002; 325 doi: https://doi.org/10.1136/bmj.325.7357.194 (Published 27 July 2002) Cite this as: BMJ 2002;325:194
- Kiran Nanchahal, lecturer in medical statistics ()a,
- ohn R Duncan, statisticianb,
- Paul N Durrington, professor of medicinec,
- Rodney T Jackson, professor of epidemiologyd
- aHealth Promotion Research Unit, Department of Public Health and Policy, London School of Hygiene and Tropical Medicine, London WC1E 7HT
- bGlaxoSmithKline, New Frontiers Science Park, Third Avenue, Harlow CM19 5AW
- cClinical Research Division II, University Medical Unit, Manchester Royal Infirmary, Manchester M13 9WL
- dFaculty of Medical and Health Sciences, University of Auckland, Grafton Mews, 52-54 Grafton Road, Auckland, New Zealand
- Correspondence to: Kiran Nanchahal
- Accepted 13 April 2002
In 2000 the UK government launched the national service framework for coronary heart disease, setting national standards for improving prevention, diagnosis, and treatment. In agreement with recent recommendations on preventing coronary heart disease1 and managing hypertension,2 this programme includes use of coronary risk appraisal models from the Framingham study published in 19913 to help identify patients eligible for drug treatment. These models were updated in 2000,4 incorporating further follow up and additional risk factors. We compare the predicted risks calculated using the two models and assess the implications for preventing heart disease.
Methods and results
The health survey for England is an annual, nationwide, household based, cross sectional surveyof a representative sample of the population. We used the 1998 survey data for 5518 (62.3% of 8852) participants aged 35-74 with complete information on factors needed for assessment of coronary disease risk, after exclusion of 738 (7.7% of 9590) participants reporting angina, heart attack, orstroke diagnosed by a doctor.5 The 2000 models allow calculation of risk over a period of four years,4 whereas the 1991 models permit estimation of risk over 4-12 years.3 We estimated the 10 year and four year probabilities of developing heart disease predicted using the 1991 equations and the four year risk predicted using the 2000 equations.
Summary statistics for four year coronary disease risk per 100 population based on the 1991 and2000 models within a range of risk categories show that both models generally produce similar distributions (table). Although substantial statistical agreement exists between classification of participants into risk categories based on the two models, participants within each category based onthe 1991 models were distributed across a wide range of risk categories based on the 2000
Although population distributions of coronary risk calculated with the two models are generallysimilar, a significant number of people meeting criteria for drug treatment on the basis of the 1991 models would not meet the equivalent criteria on the basis of the 2000 models. Current UK guidelines generally recommend offering drug treatment for hypertension or hypercholesterolaemia to patients with a 10 year risk 15%. 1 2 We used a 5% risk of a coronary event in four years as being equivalent to a 10 year risk of 15%, rather than 6% over four years, because risk increases exponentially rather thanlinearly with age. Had we used 6%, the discrepancy between the 1991 and 2000 models would have been even greater.
Our study confirms that risk of coronary disease in Britain is high. On the basis of the 1991 risk appraisal models, approximately 32% of men and 7% of women aged 35-74 in England are at 15% risk of developing heart disease in the next 10 years. The 2000 models give figures for a four year risk 5% of 29% for men and 6% for women. Although only 1-2% of men and women ineligible for drug treatment under current criteria would be eligible if the 2000 models were used, 20% of men and 43% of women currently recommended drug treatment would not be eligible if their four year risk based on the updated models was used. Sensitivity and specificity for the 1991 risk appraisal models would be 97.6% and 90.0% for men and 79.7% and 96.0% for women, considering the updated models to provide the most up to date assessment of coronary disease risk for asymptomatic men and women. Although thresholds for drug treatment are somewhat arbitrary and depend to a large degree on the resources available, we recommend that these findings are taken into account when guidelines for coronary heart disease prevention are updated in accordance with emerging scientific evidence for statin treatment and management of mild hypertension.
We thank J N Morris for comments on an earlier draft of the manuscript.
Contributors: KN devised this study and drafted the manuscript of the paper, JD undertook the statistical analyses, and all authors contributed to writing the paper. KN will act as guarantor.
JD received a SmithKline Beecham scholarship while an MSc student at the London School of Hygiene and Tropical Medicine when some of this work was done.
Competing interests None declared