Coronary heart disease (CHD) is the most common cause of death in the UK, accounting for almost one in five deaths in 2003.¹ However, death rates from CHD in the UK have halved in the past two decades. Most of this decline (58%) has been attributed to reductions in population risk factors.²

In March 2000 the UK government introduced the National Service Framework (NSF) for CHD³ which recommended, among other things, reducing the prevalence of coronary risk factors in the population, preventing CHD in high risk patients, and improving treatment for people with established CHD. Although the NSF recommends health promotion activities, most interventions—for example, the Quality and Outcomes Framework of the 2003 GP contract⁴—focus on secondary prevention.

Population impact measures (PIMs) can be used to estimate the population impact of reductions in the prevalence of risk factors and improved uptake of treatment.^5,6 Building on our previous work that assessed the population impact of secondary prevention on mortality in patients following acute myocardial infarction (AMI) and heart failure,⁷ here we use PIMs to assess the impact of primary and secondary intervention strategies on the prevention of CHD events over a 1 year period. We present data for England, but the method is applicable to any country or population to apportion the potential benefit of alternative interventions. The method is designed to assist policy making through the simple and understandable description of benefits to the population.

METHODS

Two PIMs were used to assess the impact of strategies to reduce the burden of CHD in England. The PIN-ER-t can be defined as the “the potential number of disease events prevented in a population over the next t years by eliminating a risk factor”,⁵ which we have modified here to be “the potential number of disease events prevented in your population over the next t years by eliminating all or a specified proportion of the risk factor, or (for continuous variables) reducing the risk factor to below a specified threshold”. This is used to estimate the impact of a reduction in the prevalence of a risk factor. It is derived from the population attributable risk (PAR), and the calculation requires the size of the population at risk, the incidence of the disease event in t years in the population, the proportion of the population exposed to the risk factor (or, for continuous variables, the proportion with levels above the threshold), and the relative risk of the disease event among the exposed compared with those not exposed (see formula in Appendix 1).

The number of events prevented in your population (NEPP) can be used to estimate the impact of an intervention in populations with a specified condition. It is defined as “the number of events prevented by the intervention in your population”.⁶ In this paper we have used the NEPP to estimate the population impact of moving from current to best practice treatment levels among patients with or at high risk of CHD. The calculation requires the size of the population at risk, the proportion of the population with the disease, the proportion of those with the disease exposed to the intervention currently, the proportion who adhere to the intervention, and the reduction in the risk of death in those receiving the intervention compared with those not receiving the intervention (relative risk reduction, RRR).

We obtained information from a variety of sources to calculate PIMs. For primary prevention we used the incidence of AMI from the Health Survey for England (HSE)⁸ combined with the proportion of the population exposed to each risk factor (table 1) and the relative risk of AMI associated with each risk factor (table 2). To calculate the NEPP we used data from the HSE to estimate the prevalence of AMI and the prevalence of those at >30% risk of developing CHD over 10 years based on the Framingham equation. The baseline risk of readmission among new hospital admissions for AMI was obtained from Scottish data.⁹ The RRR associated with each intervention was obtained from meta-analyses or systematic reviews where available (tables 3 and 4). These data were then used in the modelling process to estimate the population impact of the combined pharmacological interventions and combined lifestyle interventions. The combination of treatments and lifestyle interventions was based on the formula of Mant and Hicks.¹⁰ The formula used to calculate the NEPP for a combination of incremental changes in treatment and lifestyle interventions is given in Appendix 1. All calculations were for populations aged 55+ years. Due to restrictions imposed by the availability of data, we have often used AMI as a surrogate measure for CHD. Confidence intervals were derived using probabilistic sensitivity analysis.¹¹

Primary prevention

The PIN-ER-t was used to estimate the impact of a change in the levels of CHD risk factors in the general population assuming UK government targets were achieved. Thus we estimated the impact of a reduction in adult smoking from 28% to 24%, an increase in the numbers doing moderate exercise by 25%, a reduction in the level of obesity by 25% in men and 33% in women, and an increase in the consumption of fruit and vegetables by 50%. The UK government’s target to reduce mean systolic blood pressure by 5 mm Hg can be reasonably expressed as a 50% reduction in the proportion of the population with high blood pressure.¹² The target for everyone with a cholesterol level >6.5 mmol/l to achieve a level of <5 mmol/l has been described as unrealistic,¹² and we have estimated the impact of a cholesterol level of 6.5 mmol/l in these subjects which we consider to be a more achievable target. The relative risk associated with each of the above risk factors is shown in table 2.

Those at high risk

To estimate the impact of interventions among those at high risk of developing CHD, we used the NEPP. Table 3 shows the RRR for a reduction in CHD events over 1 year associated with interventions among people at high risk of developing CHD.

The proportion of those people at high risk of developing CHD who were currently treated was estimated as 57% for antihypertensives,^13,14 62% for men treated with statins,^13,14 and 36% for women treated with statins.^13,14 Adherence to treatment was estimated as 85% from the WOSCOPS study for statins¹⁵ and as 69% for antihypertensives.¹⁶ The proportion of patients at high risk of developing CHD who were eligible for treatment with one or more lifestyle intervention was derived from the HSE. We estimated the proportion of people who had a CHD risk score of over 30% based on the Framingham equations and found that 96% of those classed as high risk had at least one of the risk factors used in the Framingham equation (current smoker, overweight, poor diet, or did little or no physical activity). A study of 40 primary care providers in the UK between 2000 and 2001 suggested that around 25% of obese patients were recorded as being advised to lose weight.¹⁷ We used this as a surrogate for the percentage of at risk patients who currently receive lifestyle advice.

Secondary prevention

Table 4 shows the RRR for reduction in CHD events over 1 year associated with interventions among people with established CHD/AMI. In a previous publication focusing on mortality among people with established CHD we used the Quality and Incentives in Primary Care (QUIP) database to describe the proportion prescribed aspirin, beta-blockers, statins and ACE inhibitors.⁷ The proportion of people with established CHD who were eligible for treatment with these drugs, who adhered to treatment, and who were eligible for, received and adhered to lifestyle based interventions were also described in the earlier publication. These data are summarised in table 5 and are used here to estimate the population impact of these interventions on preventing repeat events among people with established CHD.

RESULTS

Table 1 shows the prevalence and baseline risk estimates used in the calculation of the PIN-ER-t and NEPP. We used the 1 year period prevalence of AMI as our estimate of incidence for the PIN-ER-t and as our estimate of prevalence for the NEPP. Table 1 also provides the prevalence of the population exposed to CHD risk factors for use in calculation of the PIN-ER-t. Tables 2–4 show the RR and RRR associated with risk factors and interventions.

Using the information provided in tables 1–4, we can calculate the population impact of different strategies (table 6). The results in table 6 are based on reductions in the prevalence of risk factors from current levels to those recommended by the UK government, and on moving from current treatment levels to 90% of all those eligible receiving the appropriate interventions among those at high risk of developing CHD and among those with established CHD. Table 6 shows that, if government targets were met, the greatest impact would be achieved by meeting targets for primary prevention. In particular, reducing cholesterol levels so that no-one has a cholesterol level >6.5 mmol/l would prevent 59 680 CHD events, and reducing the proportion of the population with high systolic blood pressure (140 mm Hg) by 50% would prevent 18 105 CHD events. Our findings suggest that improving levels of drug treatment among people at high risk of CHD would prevent 2008 (95% CI 790 to 3627) events within 1 year, and improving the levels of lifestyle advice among this group would prevent 4410 (95% CI 1993 to 8014) events. A strategy aimed at intervening among those with established CHD would prevent 3067 (95% CI 1572 to 5878) repeat events, and improving the levels of lifestyle advice among this group would prevent 1103 (95% CI 179 to 2097) events. It should be noted that the 95% confidence intervals for each of these intervention strategies overlap.

DISCUSSION

The results presented here show that the greatest impact on preventing CHD in England would be achieved if the UK government targets for reductions in the prevalence of risk factors were met. This is consistent with recent evidence suggesting that the reductions in CHD that have taken place in England and Wales over the past 20 years have been more as a result of primary prevention strategies than from secondary prevention.³⁵ It also emphasises the point made by Geoffrey Rose of the importance of the population approach to a condition that has population level causes.³⁶

However, the government’s targets are challenging and are unlikely to be met by simply giving advice on lifestyle.³⁷ A wide range of strategies across health, education, and other aspects of public policy will be needed to address targets relating to weight and exercise. In fact, it is likely that only the smoking target is actually achievable and may even be met by the continuation of current trends.³⁸ In our analysis we have adjusted the blood pressure and cholesterol targets to represent potentially achievable targets and, while these do not simply reflect a continuation of current trends, they do support the current downward trend in population wide levels of blood pressure and cholesterol.³⁸ Current trends suggest an increase in the proportion of the population who are sedentary and overweight or obese. Data on dietary trends over time are difficult to obtain because of the inconsistency of methods of recording and reporting diet. The UK government targets for these risk factors are thus likely to require a reversal of current trends and are therefore much more ambitious than those for blood pressure, cholesterol and, in particular, smoking.

Our estimates of differential benefit to the population from different strategies (aimed at the general population, high risk groups, and secondary prevention) could guide a prioritisation of alternative intervention strategies. The advantage of the NEPP is that it uses estimates of change from intervention trials in relation to current levels of achievement. The estimates of benefit are thus realistic. The problem with the use of the PIN-ER-t is that the estimates are theoretical only. It converts PAR into a measure of absolute risk for the population as a whole, but it assumes that this risk is reversible. Such criticism can be extended to most of the commonly used measures of burden of disease, and reinforces the need for measures which truly estimate the benefit of interventions. The estimates derived from the NEPP are therefore likely to be more clearly useful for prioritisation. We see that it is possible to rank the benefit to the population of drugs and lifestyle interventions for high risk and secondary prevention interventions. Additional information on costs can be added to these measures.³⁹

The measures used in this paper are subject to the limitations associated with using published data. RR and RRR vary from different studies and, where possible, our estimates are based on data from meta-analyses. However, this was not always possible due to limitations in the availability of data. Furthermore, many published sources use different definitions of disease status and outcome and present their results over different time frames, age groups, and may or may not split by sex. However, despite the limitations of using published literature, PIMs such as the PIN-ER-t and the NEPP can provide a method of estimating the potential impact of different strategies for reducing the burden of CHD in a population. Application of these methods suggests that, in order to have maximum benefit, UK public policy should be oriented towards primary prevention of CHD. However, meeting the government targets for primary prevention will present a considerably greater challenge than meeting the targets for secondary prevention.

APPENDIX 1

The number of events prevented in your population (NEPP) to estimate the incremental change in moving from current to best practice is calculated as:

NEPP  =  n × P_d × (P_b − P_t)× P_a × r_u × RRR

where n  =  population size, P_p  =  prevalence of disease in the population, P_b  =  proportion treated if best practice was achieved, P_t  =  proportion currently treated, P_a  =  proportion who adhere to treatment, and RRR  =  relative risk reduction associated with the treatment.

In our analyses we have adjusted the original formula for NEPP to estimate the incremental change in moving from current to best practice for a combination of n treatments. Thus, NEPP can be calculated as:

NEPP  =  n × P_d × r_u × [1 − (1 − P_einc1 × RRR) × (1 − P_einc2 × RRR₂) × ……. × (1 − P_eincn × RRR_n)]

where P_einc1 × RRR₁  =  P_b1 × P_a1 × RRR₁ − P_t1 × P_a1 × RRR₁,

P_einc2 × RRR₂  =  P_b2 × P_a2 × RRR₂ − P_t2 × P_a2 × RRR₂,

………

P_eincn × RRR_n  =  P_bn × P_an × RRR_n − P_tn × P_an × RRR_n

The PIN-ER-t for estimating the number of disease events prevented by eliminating the proportion of the risk factor above a certain threshold was calculated by subtracting the PIN-ER-t obtained for the new exposure prevalence from the PIN-ER-t for the previous exposure prevalence (for example, 24% of population smoking PIN-ER-t subtracted from 28% of population smoking). The PIN-ER-t is calculated as:

where n  =  population size, P_exp  =  prevalence of the exposure in the population, I_p  =  incidence of the outcome in the population, and RR  =  relative risk of the outcome if the risk factor is present.