Progress in understanding the multifactorial nature of cardiovascular disease and the cumulative effect of combinations of risk factors has evolved rapidly in recent years since the pioneering research of Anderson and colleagues [
Full details of our study design have been given in our previous publications [
Absolute CVD risk was based on the Framingham 10-year CVD risk as recommended in the Joint British Societies’ Guidelines [
In the risk estimator used in this study, the component of risk due to the nonmodifiable combination of gender and age,
The modifiable components,
We have used the above formulae to calculate the component of absolute CVD risk attributable to the combined nonmodifiable risk factors and each of the three modifiable risk factors individually. For the risk factors systolic blood pressure and total to HDL cholesterol ratio, where a threshold needs to be exceeded before the factor is considered a risk, we have further divided the risk into an unmodifiable and a modifiable component. For example, a patient is only considered to have a modifiable blood pressure component of CVD risk if systolic blood pressure exceeds 140 mmHg. Thus systolic blood pressure up to 140 mmHg was taken as the unmodifiable part and anything above this threshold was taken as the modifiable component. Similarly, a patient is only considered to have a modifiable cholesterol component of CVD risk if total to HDL cholesterol ratio exceeds 4.5.
For each modifiable risk factor, the change in risk was calculated as a proportion of the modifiable risk attributable to that factor so that an estimate of the proportion of modifiable risk reduced or added to could be derived for each patient. The mean population proportions were then compared with zero (i.e., no change in risk factor), using the one-sample
The baseline characteristics of the sample population are summarised in Table
Baseline characteristics of sample population.
Characteristic | Category |
|
---|---|---|
Gender | Female | 68 (11.3%) |
Male | 533 (88.7%) | |
Ethnic group | White | 581 (96.7%) |
Other | 20 (3.3%) | |
Socioeconomic statusa | Deprived | 288 (48%) |
Intermediate | 195 (32%) | |
More affluent | 118 (20%) | |
|
||
Measure | Mean (SD) | |
|
||
Age (years) | 63.7 (6.4) | |
Systolic blood pressure (mmHg) | 145 (16.7) | |
Total/HDL cholesterol ratio | 4.8 (1.1) | |
Weight (kg) | 83.6 (14.1) | |
Body mass index (kgm−2) | 28 (4.5) |
As can be seen from the table, this population was predominantly of older age, male, White, and of low or intermediate socioeconomic status. The risk factor profile of this high CVD risk population is shown in Table
Risk factor profile of sample population.
Risk factor | Definition |
|
---|---|---|
High blood pressure | SBP ≥ 140 and DBP ≥ 90 | 158 (26%) |
High cholesterol | TC/HDL ratio ≥ 4.5 | 344 (57%) |
Current smoker | Yes | 319 (53%) |
Overweight | BMI ≥ 25 and BMI < 30 | 291 (48%) |
Obese | BMI ≥ 30 | 166 (28%) |
SBP: systolic blood pressure (mmHg); DBP: diastolic blood pressure; TC/HDL: total to high density lipoprotein cholesterol ratio; BMI: body mass index (kgm−2).
The most prevalent of the established CVD risk factors was high cholesterol (57%) followed by smoking (53%) and high blood pressure (26%). It is also worth noting that overweight or obesity was highly prevalent in this sample (76%), though this risk factor is not included explicitly in the currently (at the time of conducting this research) recommended CVD population risk estimator.
Figure
Components of absolute CVD risk. SBP: systolic blood pressure; TC/HDL: total to HDL cholesterol ratio; (u): unmodifiable; (m): modifiable.
The proportion of the modifiable risk at baseline that was reduced for each of the primary risk factors is summarised in Table
Proportion of modifiable risk reduced for each risk factor.
Proportion of |
95% CI |
| |
---|---|---|---|
Smoking | 56 | 51–62 | 319*** |
Systolic blood pressure | 68 | 58–77 | 322*** |
Total to HDL cholesterol ratio | 53 | 42–64 | 309*** |
Cumulative effects of multiple risk factors. RF1: gender/age; RF2: + SBP; RF3: + TC/HDL; RF4: + smoking.
The first bar shows the negative effect of age increasing by 1 year (age and gender component of risk) (RF1). This was more than compensated for by the beneficial change in the systolic blood pressure component (RF2), with additional benefit when reduction in TC/HDL ratio (RF3) and smoking components (RF4) were added.
There was wide variation in individual patient reduction in % CVD risk, ranging from −16.4 to +26.2 but no real pattern characterising those who showed no change or increased their risk and those who reduced their risk. We performed binary logistic regression analysis on the dichotomous outcome variable (0 = no change or worse, 1 = reduction in CVD risk) on predictors of age at baseline (expressed as decade = age/10 in order to obtain an odds ratio for a meaningful change in age) and categories of gender and smoking status, high blood pressure, high cholesterol, and obesity at baseline. Just two factors made a significant explanatory contribution in the model. Older patients and those who had a high blood pressure at baseline were more likely to show a reduction in CVD risk. However, the overall predictive power of the model was low, Nagelkerke’s
The mean population 10-year CVD risk for females decreased from 26% at baseline to 21.2% after one year of intervention and for males mean population CVD risk decreased from 33.4% to 30.3%. Furthermore, the proportion of the male population estimated to be at high risk of experiencing a CVD related event was 13.9% (95% CI, 13.6–14.2), while that for females was 3.4% (95% CI, 3.3–3.6). Extrapolating these data to the whole of the at risk population of Stoke on Trent (~150,000, 77,500 males, 72,500 females), estimated ~450 serious CVD events could be prevented over 10 years.
The multifactorial risk factor approach to population CVD risk reduction offers significant advantage over the single risk factor approach and has the potential to reduce the incidence of major vascular disease related events, though this would need to be confirmed in prospective longer term studies. Moreover, the success of routine screening of electronic medical records allows for a more proactive and more precisely targeted approach to the management of population CVD risk than has been available until now.
Fifty-seven percent of the modifiable CVD risk in our sample population was attributable to smoking, 26% to high blood pressure, and 17% to a high TC/HDL cholesterol ratio. The INTERHEART study [
More importantly, perhaps, in the context of this research, approximately 59% of the modifiable CVD risk in this high risk population was removed after just one year of intervention. If replicated nationally this would represent a significant public health gain and should make a valuable contribution to reducing the burden of chronic diseases, of which the vascular disease cluster remains dominant [
Another key finding from this research relates to the importance of unmodifiable factors (in particular gender and age) included in the risk estimator. For example, risk is unavoidably inflated for males and people of older age. This may explain why our sample population (identified on the basis of having a CVD risk of ≥20%) was predominantly male (~8 : 1 ratio of males to females) and of older mean age than would be expected from the profile of CVD events in the whole population. This may be a more important consideration than has been recognised hitherto and represents a serious limitation of the Framingham risk estimator approach to population CVD risk reduction. It is possible that our study participants included older people who, although they had a confirmed CVD risk ≥20% as per national guidelines, actually had a
These latter issues raise questions about the continuing suitability of the Framingham-based approach as the primary tool for use in the prevention of cardiovascular disease. The Framingham equations were developed from the research carried out in the town of Framingham, Massachusetts, in the United States, beginning in 1948 and still continuing today. Whilst there is no denying the importance of the factors included in the original equations, they form a limited subset of the factors now known to predispose individuals to cardiovascular disease risk and do not include factors such as obesity [
Although not without some challenges, the roll-out of software (Oberoi Clinical Observations, Oberoi Consulting, Derby, UK) for the management of CVD risk across the majority of general practices in Stoke on Trent was achieved relatively smoothly and at reasonable cost (approximately ~
Multifactorial risk factor interventions for the prevention of cardiovascular disease have been shown to be effective in well-controlled, well-funded randomised controlled trials.
Greater emphasis on prevention has been advocated to address the growing burden of chronic disorders as the world’s population both grows in number and ages.
Routine screening of electronic medical records in general practice to estimate population CVD risk is feasible and affordable.
Coordinated high risk screening and additional support to reduce multiple risk factors simultaneously resulted in a reduction in mean population CVD risk by about 10% of baseline level. This should translate to a proportional reduction in the incidence of cardiovascular disease events.
The inclusion of nonmodifiable factors in the CVD risk estimator has the potential to bias patient selection towards older males in particular who may have lower
It was not feasible to record all constituent treatments (and compliance with these treatments) offered to the patients included in this research. Thus we were unable to attribute changes observed in blood pressure, total to HDL cholesterol ratio, and smoking habits directly to specific components of the intervention. In addition to this point, the current guideline CVD risk estimator is based predominantly on clinical measures and has limited sensitivity to other aspects of lifestyle modification that might influence CVD risk such as weight loss, increased physical activity, reduced psychosocial stress, avoidance of excessive intake of alcohol, and increased self-esteem that might result from enhanced support from within the community.
The sample population was predominantly male and of older age than would be expected from the profile of CVD events in the local population as a whole. We believe this to be due to an inadvertent sampling bias caused by the inclusion of the nonmodifiable risk factors of gender and age in the CVD risk estimator used. This makes it difficult to generalise beyond the specific population included in this evaluation of the NHS Health Checks programme as implemented in Stoke on Trent.
In assessing the components of absolute CVD risk attributable to each risk factor, we have partitioned the original parametric model into its various components and examined the contribution of each component to the overall risk included in the model. This is an intermediate stage in the actual estimation of absolute CVD risk. Thus, we were unable to attribute changes in risk factors directly to changes in absolute CVD risk estimated.
The NHS Health Checks programme as implemented in Stoke on Trent was successful in reducing estimated mean population cardiovascular disease risk. Around 59% of the modifiable risk attributable to the combination of high blood pressure, high cholesterol, and current smoking was removed after one year of intervention.
The authors declare that there is no conflict of interests regarding the publication of this paper.
The authors are indebted to Linda Picariello and the team of project support workers, Karen Hales, Tracy Pepper, Joanne Fynn, Dianne Machin, and Claire Whitehead, who were responsible for overseeing the recruitment process, contacting patients, and recording outcomes. The role of the clinical champions in supporting and leading implementation within each participating general practice is also acknowledged. Neil Ryder was responsible for coordinating the download of the “at risk” patient lists from electronic medical records in each practice and for coordinating data transfer associated with the project. Marion Beloe and her team of Lifestyle Coaches, Cath Dale (Leisure, Stoke on Trent City Council), and Voluntary Action Stoke on Trent (VAST) were responsible for coordinating enhanced support for lifestyle change within the community. Chris Leese managed the project on behalf of NHS Stoke on Trent and Donna Bailey provided administrative support. This work was supported by Stoke-on-Trent Primary Care Trust as part of the local implementation of the NHS Health Checks programme. The sponsor had input to the design of the study in that the evaluation was to be embedded in primary care and was to take at least the medium term (one year) view. Collection, analysis and interpretation of data, writing of the report, and the decision to submit the paper for publication were entirely the responsibility of the authors.