Strategies for reducing coronary risk factors in primary care: which is most cost effective?BMJ 1995; 310 doi: https://doi.org/10.1136/bmj.310.6987.1109 (Published 29 April 1995) Cite this as: BMJ 1995;310:1109
- K Field, senior lecturera,
- M Thorogood, senior lecturerb,
- C Silagyb,
- C Normand, professor of health policyb,
- C O'Neill, nurse facilitatorc,
- J Muir, senior research fellowc
- a Gwent College of Higher Education, Newport Business School, PO Box 180, Newport NP9 5XR
- b Public Health and Policy Department, London School of Hygiene and Tropical Medicine, London WC1E 7HT
- c Imperial Cancer Research Fund General Practice Research Group, University of Oxford, Radcliffe Infirmary, Oxford OX2 6HE
- Department of General Practice, School of Medicine, Flinders University of South Australia, Adelaide 5001, Australia Chris Silagy, professor. Correspondence to: Dr Thorogood.
- Accepted 19 March 1995
Objective: To examine the relative cost effectiveness of a range of screening and intervention strategies for preventing coronary heart disease in primary care.
Subjects: 7840 patients aged 35-64 years who were participants in a trial of modifying coronary heart disease risk factors in primary care.
Design: Effectiveness of interventions assumed and the potential years of life gained estimated from a risk equation calculated from Framingham study data.
Main outcome measure: The cost per year of life gained.
Results: The most cost effective strategy was minimal screening of blood pressure and personal history of vascular disease, which cost pounds sterling310-pounds sterling930 per year of life gained for men and pounds sterling1100-pounds sterling3460 for women excluding treatment of raised blood pressure. The extra cost per life year gained by adding smoking history to the screening was pounds sterling400-pounds sterling6300 in men. All strategies were more cost effective in men than in women and more cost effective in older age groups. Lipid lowering drugs accounted for at least 70% of the estimated costs of all strategies. Cost effectiveness was greatest when drug treatment was limited to those with cholesterol concentrations above 9.5 mmol/l.
Conclusions: Universal screening and intervention strategies are an inefficient approach to reducing the coronary heart disease burden. A basic strategy for screening and intervention, targeted at older men with raised blood pressure and limiting the use of cholesterol lowering drugs to those with very high cholesterol concentrations would be most cost effective.
The most cost effective strategy is to target patients with raised blood pressure or a history of coronary heart disease
Extending the strategy to include other factors increases the cost by variable amounts
All prevention strategies are more cost effective in men than in women and in older than younger age groups
Drug treatment for patients with high cholesterol concentrations is cost effective, but the incremental cost of extending treatment to lower cholesterol concentrations is high
Coronary heart disease currently costs the NHS about pounds sterling500 million annually (with an extra pounds sterling10 million for prevention). In 1988 in England and Wales coronary heart disease accounted for 153 084 deaths,1 and the government recently introduced a policy to reduce the incidence of coronary heart disease by 40% in people under 65 by the year 2000.2 The 1993 general practice contract provided financial incentives for general practitioners to screen for and treat prevalent coronary disease risk factors,3 but the effectiveness of of such programmes has been called into question.4
It is debatable whether coronary disease risk screening should be universal or targeted at high risk groups, who would gain more from the interventions. Silagy et al used data on the prevalence of coronary disease risk factors in participants in the OXCHECK trial to estimate how many people would need to be treated and how many potential coronary events would be averted with different screening strategies.5 They showed that the ratio of events averted to workload fell as the screening strategies became more comprehensive but did not address costs. Recent results of trials of health promotion clinics in primary care have shown only modest reductions in risk.6 7 To determine whether screening programmes are worth while we analysed the relative cost effectiveness of different coronary heart disease strategies in primary care.
Subjects and methods
In 1989, 11 090 men and women aged 35-64 years registered with general practices in Bedfordshire were randomly allocated to receive a health check during one of four years as part of the OXCHECK trial.6 We used data on the prevalance of coronary risk factors from 7840 consecutive subjects attending a first health check before February 1993.
We considered six prevention strategies ranging from targeting a small high risk group to an unselective strategy including the whole population aged 35-64 years (table I). The strategies represent a reasonably logical progression from minimal intervention to blanket screening of the whole population.
Patients found at screening to have coronary risk factors were assumed to have been offered lifestyle counselling together with drug treatment where appropriate. We assumed that the effectiveness of interventions was intermediate between that reported by Silagy et al5 and that in two recently published trials6 7—that is, treatment of diastolic blood pressure >/=95 mm Hg would achieve a mean reduction of 5 mm Hg (for simplicity systolic blood pressure was not considered); the 12 month sustained rate of stopping smoking among smokers would be 5%; patients with total cholesterol concentrations above 8.5 mmol/l would achieve a mean fall of 15% (20% would achieve this by dietary change alone, the remainder would be prescribed long term lipid lowering drugs); patients with a total cholesterol concentration of 7.5-8.5 mmol/l would achieve a mean fall of 5% by dietary change alone.
COST OF HEALTH CHECKS
The cost of each strategy included the costs of the initial screening, further tests for subjects at high risk, and the interventions discussed above. The cost of treating raised blood pressure has not been included since it is a fixed cost and the objective was to examine the relative cost effectiveness of the strategies. We assumed that a qualified grade G nurse conducted the health checks; the time required for collecting information on diet was 5 minutes; the time required for screening for other risk factors, such as smoking, body mass index, and family history, was two minutes each; the time required for taking a blood sample was 5 minutes; and the cost of measuring cholesterol concentration was pounds sterling4.80. These times were based on analysis of tape recorded checks from the OXCHECK trial done by one of us (CO'N).
Subjects found to be at risk were given lifestyle advice or drugs or both. We assumed that the lifestyle advice required four 15 minute sessions of nurse time plus 40 minutes of nurse organisational time (total 100 minutes); subjects with a high cholesterol (>8.5 mmol/l) would be given this lifestyle advice over three months; the 80% of subjects whose cholesterol concentration did not fall below 8 mmol/l would take long term lipid lowering drugs; the general practitioner would spend nine minutes each year advising a patient and prescribing a drug; a typical drug (synvastatin 20 mg/daily) would cost pounds sterling235 a year. Future costs have been discounted at 6% a year. (The assumed costs are given in the appendix.)
We calculated the benefit of the screening and intervention as the number of years of life gained by averting premature death from coronary heart disease. The probability of death from coronary heart disease was determined from data derived from the Framingham study. The Framingham parametric model8 was chosen because it was derived from a geographically defined population representative of similarly aged men and women. The model can provide predictions for different lengths of time and takes into account the effect of age, sex, smoking, blood pressure, the ratio of total cholesterol to high density lipoprotein cholesterol, diabetes, and left ventricular hypertrophy.
Data on risk factors in the study cohort were inserted in the parametric model and used to predict individuals' risk of coronary heart disease death over the next 10 years. Because no electrocardiographic data were available, the contribution of left ventricular hypertrophy to the regression equation was omitted, producing a conservative estimate of the risk.
The assumptions we made were used to modify the probability of coronary heart disease death and calculate avoided deaths. The expected years of survival for each age group have been taken from a survival table9 and avoided deaths have been converted to years of life saved. We calculated both the years of life saved and also years of life saved discounted at 6% a year to account for the fact that people prefer present benefits to future benefits and future costs to present costs. If someone borrows to spend now they have to repay the debt with interest. This means that the same sum of money is worth less if it is paid in the future.
The cost effectiveness of a strategy was calculated as the ratio of cost to years of life gained.
Cost effectiveness is dependent on the assumptions made about the effectiveness of the interventions. We therefore carried out a sensitivity analysis testing two further sets of assumptions—a best and worst scenario. For the best scenario we assumed that among cigarette smokers the 12 months sustained rate of stopping smoking would be 10%; all patients with a total cholesterol concentration above 8.5 mmol/l would achieve a mean fall of 20% (25% would achieve this by dietary change alone); and patients with a total cholesterol concentration of 7.5-8.5 mmol/l would achieve a mean fall of 10% by dietary change alone. For the worst scenario we assumed that the 12 month sustained rate of stopping smoking would be 3%; all patients with total cholesterol concentrations above 8.5 mmol/l would achieve a mean fall of 10% (15% by dietary change alone); and patients with a total cholesterol concentration of 7.5-8.5 mmol/l would achieve a mean fall of 3% by dietary change alone.
Table II shows the years of life gained, the costs, and the cost effectiveness of the six strategies by sex. The percentage of costs attributable to cholesterol lowering drugs was more than two thirds of the total cost for all strategies. The most basic strategy was the most cost effective, with an increasing cost effectiveness ratio as the strategies became more comprehensive. All strategies were more cost effective in men than women. Cost effectiveness improved greatly when screening and intervention were confined to the two older age groups and particularly to men over 60 years (table III).
Table IV shows the years of life gained, costs, and cost effectiveness for the best and worst scenarios for men and women. In men, the worst scenario resulted in a higher cost per year gained, and a greater difference between strategy 1 and 6. This effect was much greater in women. None of the assumptions changed the relative position of the six strategies, or the difference with sex.
INTERVENTION FOR CHOLESTEROL
Because cholesterol lowering drugs represented at least 70% of the total costs incurred we explored further the relative cost effectiveness of different protocols for the treatment of raised cholesterol concentration (table V). Providing no cholesterol screening or intervention was less cost effective in men than screening for raised cholesterol but offering drugs only if cholesterol concentrations remained above 9.5 mmol/l after dietary advice. The incremental cost effectiveness of introducing treatment at these concentrations was just pounds sterling30 per year of life gained in men in strategy 1. However, introducing drug treatment at lower cholesterol concentrations was considerably less cost effective.
Our calculation did not represent the total costs of coronary heart disease prevention in primary care as we did not include the cost of blood pressure screening and treatment, which was the same for all strategies, or cost of treating existing coronary heart disease. We also did not consider the savings generated by lower morbidity and therefore less demand for medical care. Those people surviving who would not otherwise have done so will generate other costs of care as they became at risk of other diseases. These costs are unknown.
We have shown that recording blood pressure and identifying prevalent vascular disease with appropriate interventions is more cost effective than more comprehensive screening. In men the cost was pounds sterling730 per non-discounted year of life gained for the most basic strategy and pounds sterling1270 per non-discounted year of life gained for the comprehensive strategy. The equivalent cost effectiveness ratios taking into account discounting of benefits were pounds sterling1240 and pounds sterling2180.
APPLICABILITY OF RESULTS
Evidence on the effectiveness of primary care interventions is mixed. Our assumptions about the effectiveness of interventions are more optimistic than those reported recently.6 7 The calculations may therefore overestimate the cost effectiveness of such checks. A sensitivity analysis showed that, irrespective of whether optimistic or pessimistic assumptions were used, the relative position of the strategies did not change, nor did the differential effect seen in men and women.
We studied a cohort of 7840 people screened in Bedfordshire. The prevalence of such risk factors is likely to vary in other parts of Britain, but the relative importance of different risk factors in men and women will be much the same. We used data from the United States to calculate years of life gained because no such data are available for a British population. The Framingham model provides an optimistic estimate of gain from interventions and does not distinguish between risk associated with an initial cholesterol concentration and that associated with the same concentration achieved by dietary change or drug treatment. In addition, the model gives a coefficient of risk for smokers and non-smokers but does not allow for a gradual fall in risk in those stopping smoking. The Framingham population will be genetically similar to a British population, but, nevertheless, it is impossible to determine the true predictive value of the Framingham equation in our population. However, since we were comparing relative cost effectiveness of strategies, the inaccuracies caused by using non-British data will not invalidate our conclusions.
EFFECT OF AGE AND SEX
Screening and intervention were considerably less cost effective in women. In addition, the incremental cost of extending from one strategy to the next was considerably greater in women than men. Cost effectiveness improved greatly as age increased.
Our results suggest that strategies to reduce coronary heart disease risk in primary care are most cost effective if they are limited to patients with raised blood pressure or a history of coronary heart disease. These initiatives are more cost effective in men than women, and in older rather than younger subjects. Lipid lowering drugs account for a large proportion of the cost, and protocols which limit use of these drugs to those with very high cholesterol concentrations (above 9.5 mmol/l) will increase cost effectiveness.
Our calculations are based entirely on the risk of death from coronary heart disease. A small reduction in smoking incidence would have a large effect on the incidence of other fatal diseases, and it is important, therefore, to inquire about smoking habits and encourage patients to stop.
Although these data do not define an ideal screening policy or provide exact figures for the cost of different preventive strategies, they may be useful to health planners and primary care teams. Focusing on comparative cost effectiveness allows more informed choices. Data on the relative cost effectiveness of different preventive strategies and the incremental cost per extra year of life gained when increasing the scope of a strategy should allow health planners to set a hypothetical ceiling of expenditure for each year of life gained and then choose a package tailored to age group and sex which lies below that ceiling. Further research, preferably by randomised trials, on the effectiveness of health promotion interventions would enable better comparisons between different strategies for preventing coronary heart diseases and between strategies for preventing different diseases. We need information on the long term changes that can be expected from health promotion interventions and also the effects that these changes have on the future risk of disease.
We thank the Imperial Cancer Research Fund OXCHECK Study Group for access to their data and Sue Langham, Godfrey Fowler, Martin Lawrence, and Theo Schofield for comments on the manuscript. KF and MT were supported by the Health Gain Project, funded jointly by the Health Education Authority and North West Thames Regional Health Authority. JM and CO'N were supported by the Imperial Cancer Research Fund.
Assumed costs of screening were as follows: Annual salary for a grade G nurse=pounds sterling17 725 including national insurance of 4% and superannuation of 10.6%. Syringes (20 ml)=pounds sterling3.07/50.
Needles=pounds sterling1.15/100. (10 ml tubes=pounds sterling94.50/1000.
Laboratory cholesterol test=pounds sterling4.80 (Luton and Dunstable Hospital).
General practitioner is pounds sterling44000 plus 6.5% NHS contribution toward superannuation.
Lipid lowering drug (simvastatin 20 mg)=pounds sterling18.29/28 tablets.