Journal of Public Health Advance Access originally published online on August 3, 2007
Journal of Public Health 2007 29(4):379-387; doi:10.1093/pubmed/fdm045
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Reductions in cardiovascular risk in association with population screening: a 10-year longitudinal study
S. McCluskey, Research Fellow1,
D. Baker, Professor1
D. Percy, Senior Lecturer2
P. Lewis, Consultant Cardiologist3
E. Middleton, Research Fellow1
1 Centre for Public Health Research, University of Salford, C701 Allerton Building, Frederick Road Campus, Salford M6 6PU, UK
2 Centre for Operational Research and Applied Statistics, University of Salford, Salford M5 4WT, UK
3 Stockport Acute NHS Trust, Stockport, UK
Address correspondence to Serena McCluskey, E-mail: s.mccluskey{at}salford.ac.uk
Background This study was carried out in order to examine changes in cardiovascular risk associated with a population-based screening programme.
Method Cardiovascular disease (CVD) risk factor data from a representative sample of residents aged between 45 and 55 years who attended screening a total of three times over a 10-year period were chosen for analysis (n = 4113). Cohorts were defined as either ‘high risk’ or ‘normal risk’ at baseline for risk factors including blood pressure, body mass index (BMI), cholesterol, smoking and alcohol intake. Mean changes were observed for both groups over three screening episodes, and results were stratified by gender.
Results For the high-risk cohorts (after controlling for age and regression to the mean effects), there were significant decreases in all risk factors, except BMI. Conversely, the observed changes in the normal risk cohorts indicated significant increases in risk factors over the 10-year period. After adjusting for age, the pattern in the normal risk cohorts fluctuated and there were some decreases in risk, but they were not as large as the decreases in risk for the high-risk cohorts.
Conclusions Population screening for CVD is an effective strategy for identifying and reducing risk in high-risk individuals. These results have significant implications for the role of screening in preventing and controlling cardiovascular disease.
Keywords: cardiovascular disease, epidemiology, screening
 |
Introduction |
|---|
 TOP Introduction MethodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
In general, two strategies for the primary prevention of cardiovascular disease (CVD) are now widely recognized—the targeted approach, in which individuals at high risk are identified and targeted for preventive treatment, and the population approach, in which population-wide changes in risk factors are implemented.1,2 Screening for CVD risk factors is an example of the former, in that while the intervention is offered to the whole of the population, only those identified as at high risk of developing the disease are in general referred for appropriate treatment or health advice.3 It has recently been proposed that strategies targeted at the reduction of risk factors have the potential to drastically reduce the burden of CVD4 and that such initiatives may result in the current UK mortality rate being halved.5
There is a dearth of longitudinal evidence, which examines the efficacy of screening for reducing cardiovascular risk. One of the first and only major UK studies reporting on the effects of screening—the South-East London Screening Study (SELSS)6,7—was a long-term controlled trial conducted in two local group general practices. No significant differences were found between the screened and the non-screened groups in any of the outcome measures (consultation, hospital episodes, certified absence from work and mortality) after 9 years follow-up. International studies comparable to the SELSS were also conducted in the USA and Sweden, and all showed similar results in terms of total mortality and morbidity outcomes.8 Conclusions drawn were that the value of screening, when placed in the context of a cost-benefit analysis, was limited.
However, it is likely that the results reported in these studies were subject to ‘dilution bias’9 and this could have accounted for the apparent lack of impact of screening on outcomes such as CVD morbidity and mortality. ‘Dilution bias’ implies that the relationship between an intervention and an outcome measure is diluted by other intervening variables that cannot be controlled for in experimental studies based on the community and extending over a long period of time. For example, there is likely to be a significant period of time between being screened in middle age or younger and mortality due to CVD, and there are more than 200 risk markers for CVD that could have had an influence during this period.10 Medical treatment may also change the outcome of the disease, as may changes in the social lives of participants who are associated with increased risk of CVD (e.g. unemployment; stressful life events). From this basis, it has been argued that appropriate evaluation methods for interventions that are based on the community over a long-time period should be to measure effects closely related to the intervention, such as the reduction of CVD risk factors.11,12
The present study provides a 10-year longitudinal observation of changes in major CVD risk factors in association with a local population-based screening programme, and it is proposed that screening will be effective in reducing cardiovascular risk in ‘high risk’ groups, but not for reducing levels of risk in the whole population, including groups within the normal range of risk.
|
Methods |
|---|
|
TOPIntroductionMethodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
Stockport is located in the North-West of England and is a comparatively affluent metropolitan borough of Greater Manchester. In 1989, Stockport Health Authority implemented a population-wide screening programme for CVD, whereby all residents aged between 35 and 60 years were invited for screening via their general practices.
Nurse practitioners (n = 105) from all 60 of the general practices in Stockport were trained to carry out the screening procedure, whereby major risks for CVD were measured and recorded. This included blood pressure measurement and assessment, and counselling for modifiable CVD risk factors. Blood pressure was recorded using a 35 cm x 12.5 cm adult cuff, and the technique was standardized using British Hypertension Society guidelines. Body mass index (BMI), serum cholesterol concentration (obtained by blood sample), smoking status and weekly alcohol consumption were recorded on a standardized data collection card, which was also used to collect information about age, gender, occupation and employment status. All those identified as hypertensive and with high cholesterol were referred to the general practitioner for further treatment and this was noted on their screening record. Those identified as high risk in relation to one or other of the risk factors associated with lifestyle change were given health promotion advice from a standard pro forma.
High risk was defined as: systolic blood pressure >150 mmHg; diastolic blood pressure >90 mmHg; cholesterol >6.5 mmol l–1; being a smoker; alcohol consumption >21 units per week for men and >14 units per week for women; and BMI >30 (calculated using the standard measurement based on weight and height) (Although some of these cut-off points are high compared with current guidelines, it is appropriate to use the cut-off points that were employed at the time, as participants will have been identified and treated according to relevant guidelines.). On completion of screening, supporting literature about the reduction of CVD risk was given to every patient. Risk factor data were entered into a centralized database, a unique procedure, which also enabled participants to be invited for screening every 5 years. The screening process is illustrated in Fig. 1.
|
Over a 10-year period (1989–99), there were 84 646 people who had attended screening. Of these 50 788 were screened once, 25 251 were screened twice and 8607 were screened three times. The data from those individuals who had attended three screening episodes, separated by periods of 5 years, were used for a longitudinal observation of changes in cardiovascular risk. An anonymized database was made available for research purposes by Stockport Primary Care Trust. As the data were anonymized externally, ethical approval was not required. An investigation of this database reported that uptake of screening was equitable, and that the screened population was found to be representative of the Stockport population in general.13 Data on the age, gender and occupational status of the screened participants were also made available, and a ward deprivation score was calculated based on the employment domain of the DETR Index of Multiple Deprivation (2000).
In order to perform a longitudinal observation of changes in cardiovascular risk, separate cohorts for each individual risk factor were selected based on data completeness at all three screening episodes. For example, to be included in the cholesterol risk factor cohort, only those individuals who had cholesterol data recorded at all three screening episodes were included (see Fig. 2). Because screening essentially only targets those at high risk, this study was primarily concerned with examining risk changes in those people who had been classified as high risk on each of the cardiovascular risk factors. In order to establish that changes in high risk were due to screening, a control group was needed (i.e. the ‘normal risk’ population who were not targeted in the screening programme). If the same trends in risk factor changes were observed in both groups, then it would be more difficult to attribute screening to these effects as these changes were occurring in the normal population also.
|
However, preliminary analyses of the high and low-risk cohorts revealed that the high-risk groups were largely 45 years and over and the low-risk group had an average age of 37 years, and therefore any comparison between these two groups would have been confounded by age. It is widely known that cardiovascular risk does increase with age, but because the controls (i.e. the normal risk groups) had to be as similar as possible to the high-risk groups at baseline, those people aged <45 years were excluded from further analyses (n = 4494). Cohorts for each risk factor were then categorized as ‘high risk’ or ‘normal risk’ depending on risk factor status at first screen (e.g. ‘high risk’ BMI was a score of 30 and above at first screen, and ‘normal risk’ BMI is a score below 30 at first screen). Figure 2 illustrates the numbers in each risk factor cohort.
Repeated measures ANOVAs were performed using general linear model procedures. Liddell's exact test was performed to compare paired proportions (for smoking prevalence data), and significance was set at the 5% level. Using Bonferroni post hoc analyses for confidence intervals (CIs), the overall changes in risk factor levels between screening episodes 1 and 3 were examined. However, a consideration with the interpretation of studies of trends in risk factor reduction over time is regression towards the mean (RTM), whereby any variable that has a high value will be lower by chance in the next observation. Therefore, the RTM effect over a 10-year period was also calculated using the formula given in Yudkin and Stratton14 for all the high-risk factors (with the exception of smoking because this was not a continuous variable). In addition, to acknowledge the effect of RTM, the effects of ageing over a 10-year period on changes in risk factors also need to be taken into account; following the actual observed change in risk, an age-adjusted effect was also calculated using regression analysis of the population at first screen with age as the independent variable and risk as the dependent variable. Thus, the final change in the high-risk group, which could be attributed to the effectiveness of the screening programme, was the age-adjusted change minus the RTM effect, and this was assessed for significance by reporting CIs for the differences between the two means.15,16
|
Results |
|---|
|
TOPIntroductionMethodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
In order to examine the representativeness of the selected cohort (n = 4113) with the 3-screen sample from which it was selected, occupational group data were analysed as a proxy for socioeconomic status. Table 1 illustrates that the cohorts selected for analysis were largely representative of the 3-screen sample in terms of socioeconomic status.
|
Changes in cardiovascular risk were then examined for both normal and high-risk cohorts for each risk factor at each screening episode, stratified by gender. The results are described below and displayed in Table 2.
|
Blood pressure
There was a significant decrease in mean systolic blood (BPSYS) pressure in high-isk males and females [–9.49 mmHg, 95% CI (–7.20 to –11.77); –11.30 mmHg, 95% CI (–8.93 to –13.68), respectively]. Conversely, there was a significant increase in BPSYS pressure in normal risk males and females [5.43 mmHg, 95% CI (4.49–6.35); 7.33 mmHg, 95% CI (6.52–8.14), respectively]. Post hoc tests revealed that the decreases in BPSYS pressure were only significant between screening episodes 1 and 2, but not between episodes 2 and 3.
There was also a significant decrease in mean diastolic blood pressure (BPDIAS) in high-risk males and females [–5.29 mmHg, 95% CI (–4.34 to –6.24); –5.62 mmHg, 95% CI (–6.74 to –4.51), respectively]. Conversely, there was a significant increase in BPDIAS in the normal risk males and females [3.36 mmHg, 95% CI (2.70–4.71); 3.39 mmHg, 95% CI (2.87–3.90), respectively]. Again, post hoc tests revealed that the decreases in BPDIAS were only significant between screening episodes 1 and 2, but not between episodes 2 and 3.
Smoking
There was a significant smoking cessation rate for high-risk males and females (i.e. those who were smoking at screening episode 1) [–31.3%, 95% CI (–25.0 to –38.1%); –22.6%, 95% CI (–17.5 to –28.3%), respectively]. Comparing this with similar data taken from the 2003 General Household Survey,17 it was shown that the prevalence of smoking in the general population also decreased over the 10-year period, although not as substantially. Post hoc tests revealed that the reduction in smoking prevalence was significant at all three screening episodes.
Body mass index
There were no significant changes in mean BMI for high-risk males or females. However, there was a significant increase in mean BMI for normal risk males and females [0.73, 95% CI (0.56–0.90); 1.01, 95% CI (0.83–1.20), respectively].
Cholesterol
There was a significant decrease in mean cholesterol for high-risk males and females [–0.70 mmol l–1, 95% CI (–0.53 to –0.86); –0.49 mmol l–1, 95% CI (–0.29 to –0.69), respectively]. Conversely, there was a significant increase in mean cholesterol for normal risk females [0.21 mmol l–1, 95% CI (0.07–0.36)], but there was a significant decrease in mean cholesterol for normal risk males [–0.17 mmol l–1, 95% CI (–0.06 to –0.28)]. Post hoc tests revealed that for high-risk males and females, the decreases in mean cholesterol were significant at all three screening episodes.
Alcohol
There was a significant decrease in mean weekly alcohol consumption for high-risk males and females [–7.06 units, 95% CI (–4.66 to –9.46); –6.08 units, 95% CI (–4.33 to –7.73), respectively]. Conversely, there was a significant increase in mean weekly alcohol consumption for normal risk males and females [2.25 units, 95% CI (1.32–3.19); +0.76 units, 95% CI (0.20–1.32), respectively]. Post hoc tests revealed that the decrease in mean alcohol consumption was only significant between screening episodes 1 and 2, but not between episodes 2 and 3.
Age-adjusted and RTM effects
Following on from the observed change in the high-risk factors, the age-adjusted and RTM effects were then calculated. The age-adjusted change was found to be significantly greater than the RTM effect for all high-risk factors except BMI (P < 0.05), showing a reduction in high risk that was not due to chance and could be attributed to the effectiveness of the screening programme (see Table 3). Those that were not targeted in the screening programme (i.e. the normal risk cohorts) show a more fluctuating pattern in risk after controlling for age—for normal risk males there were significant increases in diastolic blood pressure and alcohol intake, significant decreases in systolic blood pressure and cholesterol and no significant changes in BMI (P < 0.05). For the normal risk females, there were significant increases in alcohol intake, and significant decreases in systolic blood pressure, diastolic blood pressure, BMI and cholesterol (P < 0.05).
|
|
Discussion |
|---|
|
TOPIntroductionMethodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
Main findings from this study
The results from the present study showed that screening was associated with a reduction of the majority of cardiovascular risks in the high-risk cohorts. After controlling for age and RTM effects, significant reductions were observed in mean systolic and diastolic blood pressure, cholesterol and weekly alcohol intake for both males and females. In addition, there was a 31% decrease in smoking prevalence for males and a 23% decrease for females in the screening programme, whereas national smoking data showed that there was only a 6% decrease in the prevalence rate for smoking in a similar age group over a similar time period. Although observed changes in the non-targeted population (i.e. those with normal levels of risk) suggested that levels of risk were increasing over time, after adjusting for age these patterns then fluctuated. Although some decreases in risk then appeared, they were not as large as the decreases in the high-risk cohorts, suggesting that screening was having a more direct influence by referring those identified for appropriate treatment.
What is already known on this topic?
The role of modifiable risk factors in the onset and progression of CVD is significant, and it has been reported that more than half the recent mortality decrease can be attributable to reductions in smoking, blood pressure and cholesterol levels.18
What this study adds?
Overall, these findings confirm the assumption that population screening identifies those at high-risk of developing CVD and referral for treatment will reduce risk for these individuals. Although the reductions in risk appear to be small, the implications on mortality from even a modest reduction in smoking, cholesterol and blood pressure were estimated in a recent study, which suggested that over 35 000 life-years could be gained over a 10-year period.19
Limitations of this study
Screening for CVD risk was not effective in reducing high levels of risk for all the risk factors examined. Levels of obesity as measured by BMI >30 were not significantly reduced for either males or females. This may reflect the fact that the management of obesity is more reliant on adherence to community-based health promotion advice about diet and levels of physical activity rather than prescribed medication or treatment regimens in the primary care setting. It was also unclear whether reductions in cholesterol for high-risk groups were as a direct consequence of screening interventions (i.e. the prescription of statins) or as part of a general downward trend, because a decrease was also observed for normal risk males. Similar results were found in a study of a UK health promotion campaign where important changes in modifiable risks for CVD were observed, but reductions in risk were also observed in the reference group.20 These findings indicate the need for more in-depth investigations into the links between levels of risk and outcome. A recent review of the evidence stated that a fundamental policy shift is required to widen responsibility for the prevention of diet, activity and weight-related ill health across the whole of Europe's population.21 It appears that further investigation may be necessary in order to establish which methods of prevention are effective for which risk factors.
One clear limitation of this study was the lack of follow-up data on treatment patterns, so that the exact mechanisms whereby risk factors were reduced could not precisely be determined. Another limitation was that only 10% of patients who attended one screening went on to attend three screenings. However, this was largely age dependent rather than being due to any systematic bias—only those people who were aged 35–49 at the beginning of the screening programme could have definitely had three screens over the screening period. A total of 8988 people between the age of 35 and 49 were screened for the first time between 1989 and 1991, meaning that there was only a small level of attrition across the 10-year period (4.2%).
The present study also highlighted discrepancies in the recording of risk factor data, reflected by the different number of participants in each risk factor cohort (selected based on data completeness). This has been examined in more detail elsewhere, whereby it was concluded that the discrepancies were most likely to have occurred because of differing procedural requirements for certain risk factors (e.g. monitoring of cholesterol involved the external analysis of a blood sample) and judgements made by the nurse practitioner on perceptions of risk in different individuals.22 However, a comparison of the selected cohort with the whole 3-screen sample indicated that the study cohort was representative in terms of socioeconomic status. It is acknowledged that a significant amount of data on occupational group were missing, but information from the available data indicates no major differences between those attending one, two or three screenings (see Table 1). It is also unfortunate that data were not collected beyond the 10-year period in the present study, as without lengthy follow-up times it is difficult to establish the true extent of changes in cardiovascular risk as a result of screening, and whether regular sustained screening is more effective than a limited number of episodes. This would make an interesting topic for future research, given the cost implications of population screening sustained over a long period of time.
|
Conclusions |
|---|
|
TOPIntroductionMethodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
Recent European guidelines suggest that overall, large benefits in terms of avoided CVD can be expected from comprehensive-risk assessment,23 and this research shows that population screening for CVD risk factors is an effective strategy for identifying and reducing risk in ‘high risk’ groups. These results have significant implications for the role of screening in preventing and controlling CVD.
|
Funding |
|---|
|
TOPIntroductionMethodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
This work was funded by the Department of Health UK.
|
Acknowledgements |
|---|
|
TOPIntroductionMethodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
The authors would like to thank Stockport Primary Care Trust for their cooperation with access to data and data analysis.
|
References |
|---|
|
TOPIntroductionMethodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
- Emberson J, Whincup P, Morris R, et al. Evaluating the impact of population and high-risk strategies for the primary prevention of cardiovascular disease. Eur Heart J (2004) 25:484–91.
[Abstract/Free Full Text] - Rose G. The Strategy of Preventive Medicine (1992) Oxford: Oxford University Press.
- NHS. National Service Framework for Coronary Heart Disease in England and Wales. (2000) London: Department of Health.
- Unal B, Critchley JA, Capewell S. Explaining the decline in coronary heart disease mortality in England and Wales between 1981 and 2000. Circulation (2004) 9:1101–7.
- Critchley JAC, Capewell S. Substantial potential for reductions in coronary heart disease mortality in the UK through changes in risk factor levels. J Epidemiol Community Health (2003) 57:243–7.
[Abstract/Free Full Text] - Trevelyan H. A study to evaluate the effects of multiphasic screening within general practice in Britain: design and method. Prev Med (1973) 2:278.[CrossRef][Medline]
- The South-East London Screening Study Group. A controlled trial of multiphasic screening in middle-age: results of the South-East London Screening Study. Int J Epidemiol (1977) 6:357–63.
[Abstract/Free Full Text] - D'Souza M. Early diagnosis and multiphasic screening. In: Recent Advances in Community Medicine—Bennett A, ed. (1978) Edinburgh: Churchill Livingstone. 195–214.
- Lindholm L, Rosén M. What is the ‘golden standard’ for assessing population-based interventions—problems of dilution bias. J Epidemiol Community Health (2000) 54:617–22.
[Abstract/Free Full Text] - Hopkins PN, Williams RR. A survey of 246 suggested coronary risk factors. Atherosclerosis (1981) 40:1–52.[CrossRef][Web of Science][Medline]
- Nutbeam D. Evaluating health promotion—progress, problems and solutions. Health Promot (1998) 13:27–44.[CrossRef][Web of Science]
- Susser M. The tribulations of trials-interventions in communities. Am J Public Health (1995) 85:156–8.
[Free Full Text] - Baker D, Hann M. Evaluation of the Stockport Cardiovascular Disease Screening Database as a Tool for Research and Evaluation Projects. (2002) Manchester: National Primary Care Research and Development Centre, University of Manchester.
- Yudkin P, Stratton IM. How to deal with regression to the mean in intervention studies. Lancet (1996) 347:241–3.[CrossRef][Web of Science][Medline]
- Gardner MJ, Altman DG. Statistics with Confidence (1989) London: BMJ Books.
- Prior JO, van Melle G, Crisinel A, Burnand B, Comuz J, Darioli R. Evaluation of a multicomponent worksite health promotion program for cardiovascular risk factors—correcting for regression towards the mean effect. Prev Med (2005) 40:259–67.[CrossRef][Web of Science][Medline]
- Office of National Statistics. General Household Survey 2003; mid-2003 Population Estimates. http://www.statistics.gov.uk/ghs (2005, date last accessed).
- Critchley JAC, Capewell S. Why model coronary heart disease? Eur Heart J (2002) 23:110–6.
[Free Full Text] - Critchley JA, Capewell S, Unal B. Life-years gained from coronary heart disease mortality reduction in Scotland. Prevention or treatment? J Clin Epidemiol (2003) 56:583–90.[CrossRef][Web of Science][Medline]
- Bennett P, Blackwell M, Clapham M, et al. South Birmingham coronary preventive project: a district approach to the prevention of heart disease. Community Med (1989) 2:90–6.
- James PT, Rigby N, Leach R. International Task Obesity Task Force. The obesity epidemic, metabolic syndrome and future prevention strategies. Eur J Cardiovasc Prev Rehabil (2004) 11:3–8.[CrossRef][Web of Science][Medline]
- Bartys S, Baker D, Lewis P, et al. Inequity in recording of risk in a local population-based screening programme for cardiovascular disease. Eur J Cardiovasc Prev Rehabil (2005) 12:63–7.[CrossRef][Web of Science][Medline]
- De Backer G, Ambrosioni E, Borch-Johnsen K, et al. European guidelines on cardiovascular disease prevention in clinical practice: third joint task force of European and other societies on cardiovascular disease prevention in clinical practice. Eur J Cardiovasc Prev Rehabil (2003) 10(Suppl.):S1–10.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||