Abstract
Background Many programmes to detect and prevent cardiovascular disease (CVD) have been performed, but the optimal strategy is not yet clear.
Aim To present a systematic review of cardiometabolic screening programmes performed among apparently healthy people (not yet known to have CVD, diabetes, or cardiometabolic risk factors) and mixed populations (apparently healthy people and people diagnosed with risk factor or disease) to define the optimal screening strategy.
Design and setting Systematic review of studies performed in primary care in Western countries.
Method MEDLINE, Embase, and CINAHL databases were searched for studies screening for increased cardiometabolic risk. Exclusion criteria were studies designed to assess prevalence of risk factors without follow-up or treatment; without involving a GP; when fewer than two risk factors were considered as the primary outcome; and studies constrained to ethnic minorities.
Results The search strategy yielded 11 445 hits; 26 met the inclusion criteria. Five studies (1995–2012) were conducted in apparently healthy populations: three used a stepwise method. Response rates varied from 24% to 79%. Twenty-one studies (1967–2012) were performed in mixed populations; one used a stepwise method. Response rates varied from 50% to 75%. Prevalence rates could not be compared because of heterogeneity of used thresholds and eligible populations. Observed time trends were a shift from mixed to apparently healthy populations, increasing use of risk scores, and increasing use of stepwise screening methods.
Conclusion The optimal screening strategy in primary care is likely stepwise, in apparently healthy people, with the use of risk scores. Increasing public awareness and actively involving GPs might facilitate screening efficiency and uptake.
INTRODUCTION
Cardiovascular diseases (CVD) are the leading causes of death and disability worldwide.1 Early detection and treatment of cardiometabolic risk factors can prevent development of CVD. GPs could play a central role in primary prevention as they are easily accessible to patients. Performing screening in a familiar setting — inside the general practice — is likely to enhance participation.2 In addition, counselling and treatment could easily follow screening, as these are already part of usual care.
Screening programmes can be performed among apparently healthy individuals (not yet known to have CVD, diabetes, or cardiometabolic risk factors) and in mixed populations (including apparently healthy people and people already diagnosed with a risk factor or established disease). Screening entire populations could lead to considerable unnecessary testing, with a fairly low yield, as most people screened will be healthy. Other ways of screening have been advocated, for example, targeted or stepwise screening methods.3 In targeted screening, a presumed high-risk group is considered, for example, people with central (abdominal) obesity. Stepwise methods can be used to identify high-risk groups, thereby limiting the number of people qualifying for further examinations.
Even though many screening programmes have been conducted in primary care over recent decades, no attempt had been made to compare the various approaches and define lessons. In the authors’ opinion, the optimal screening strategy should identify all people with an increased cardiometabolic risk, with a minimum effort to detect these people. Therefore, this study was a systematic review of cardiometabolic screening programmes in a primary care setting. The aim was to define the optimal screening strategy, by focusing on the population in which the screening was performed, the approach taken with patients, the different screening steps, the uptake, and the yield of screening.
METHOD
Criteria for considering studies for this review
In this study, the focus was on screening studies that were aimed at detecting an increased cardiometabolic risk, performed in primary care. Studies were excluded when they assessed only the prevalence of risk factors without further follow-up or treatment; those without involving a GP in the screening process; when fewer than two cardiometabolic risk factors (hypertension, dyslipidaemia, impaired glucose tolerance, diabetes, overweight/obesity) were considered as the primary outcome; and those with study populations constrained to an ethnic minority.
It was assumed that people with established disease or risk factors already receive extensive care through existing pathways, although this probably only applies to countries with a well-established primary care system. Therefore, only studies performed in Western countries were included: European countries, Australia, New Zealand, Canada, and the US.
How this fits in
Screening programmes can identify cardiometabolic risk factors that, when treated, can prevent development of cardiovascular disease. Several strategies, settings, and populations can be eligible. This study defines that the optimal screening strategy in primary care could be a stepwise approach, using, for example, risk scores to select people qualifying for further screening examinations. Increasing public awareness and actively involving GPs may facilitate screening efficiency and uptake.
Search methods
The MEDLINE, Embase, and CINAHL databases were searched for synonyms for primary care, screening, and cardiometabolic risk factors, on 27 January 2013. For the detailed search strategy see Appendix 1 (available from the authors on request). Language was restricted to English.
Data collection and analysis
Two reviewers independently checked all titles and abstracts. Potentially relevant articles were retrieved full-text, and subsequently assessed for inclusion by two reviewers independently. When a full-text article was not available, the author and/or editor was contacted. In cases of disagreement between two reviewers, these were discussed and resolved by the third reviewer.
For included studies, relevant data were extracted using a standardised template. Data were extracted on the screened population, screening method, patient approach, response rates, and yield of screening in terms of detected cardiometabolic risk factors. When more than one article reported on the same study population, the article with the most information on the method was included.
RESULTS
Included studies
Figure 1 represents the study flow diagram. The search strategy yielded 11 445 hits, of which 26 met the study inclusion criteria. Data on population characteristics, screening method, patient approach, and response rates are listed in Tables 1 (apparently healthy populations) and 2 (mixed populations).
The first four studies concerning screening for cardiovascular risk were published between 1967 and 1972.4–7 These were all so-called multiple screening studies: besides cardiovascular risk factors the screening programmes also screened for other diseases like glaucoma, anaemia, or cervical cancer. The first study that focused solely on screening for an increased cardiovascular risk was published in 1978.8
Screening population
Apparently healthy populations
Five studies focused on apparently healthy patients only (Table 1),9–13 including the four most recently published. The number of people eligible for screening ranged from 361 to 24 166. Four studies defined a specific age-category, the lower threshold varying between 20 and 40 years and the upper threshold varying between 69 and 75 years. Only the study by Lambert et al set a minimum age, excluding males <40 years.10
Mixed populations
Twenty-one screening programmes were performed in mixed populations (Table 2).4–7,14–27 The British Family Heart Study reported which part of their study population was not previously diagnosed with coronary heart disease, high blood pressure, high cholesterol, or diabetes. None of the other studies provided any information about the composition of their study population. Most studies included males and females. The number of people eligible for screening ranged from 120 to 40 000. Sixteen studies defined a specific age-category, with lower thresholds from 15 to 45 years and upper thresholds from 50 to 65 years. The remaining studies excluded school children,4,7 or specifically considered older patients.5,26,28
Screening method
Apparently healthy populations
Three of the five studies performed in an apparently healthy population used a stepwise approach.9,11,13 Calculation of a risk score based on a questionnaire completed by patients,9 data available in electronic medical records,11 or self-measuring waist circumference,13 were used as first screening steps; subsequently people with scores above a threshold were invited for additional examinations.
Mixed populations
Hellénius et al were the only group to use a stepwise approach in a mixed population. People were asked to fill in a questionnaire about the presence of cardiometabolic and lifestyle risk factors. Those with at least one risk factor were invited for further examinations.22
Patient approach
Apparently healthy populations
Four studies in apparently healthy populations actively invited people to participate in the screening, by a written invitation from their GP,9,11,13 or by either their own GP or an alternative provider.10 One study used an opportunistic approach: eligible males were asked during a regular GP visit to participate in screening.12
Mixed populations
Most studies performed in mixed populations actively invited people. Five studies used a different approach. Two studies asked people to participate in screening during a regular GP visit;22 in one of them people could also attend on their own request.16 Cope et al informed people about the screening through public advertising inside and outside the general practice.4 Two studies combined the active and opportunistic approaches.17,25
Response rate
Apparently healthy populations
Four studies performed among apparently healthy people had response rates between 50% and 79%; if necessary, one or more reminders were sent. One study had a lower response rate of 24%; sending reminders was not reported.10 The age group eligible for screening did not seem to influence the response rate. The study with the highest rate was the only one with an opportunistic approach and lasted for 3 years.12
Mixed populations
Response rates in mixed populations ranged from 6% to 93%. The age group eligible for screening did not seem to influence the response rate. The lowest response rate (6%) was observed by Hellénius et al using a stepwise method and an opportunistic approach.22 Devroey et al reported a rate of 7%; in this study the local authorities invited all inhabitants of three Belgian towns to visit their GP for a health check.20 The highest response rate was found in the Swedish Västerbotten Intervention Programme, in which all people of a specific age were annually invited.27
Yield in terms of cardiometabolic risk
Apparently healthy populations
An overview of the yield of the studies in apparently healthy populations is given in Table 3. Four studies calculated a CVD risk score for participants. The results varied from a 10-year cardiovascular mortality risk ≥10% in 6% of the study population,9 to a 10-year cardiovascular mortality and morbidity risk ≥20% in 20% of the study population.10
Mixed populations
The yield of the studies performed in mixed populations is presented in Table 4. All but one of the studies measuring obesity (BMI ≥30 kg/m2) report prevalence rates between 10% and 21%. This percentage does not clearly increase in the more recent studies. The yield of diabetes ranged from 1% to 3%. Prevalence of hypercholesterolaemia (total cholesterol ≥6.5 mmol/l) ranged between 24% and 48% for females, and 26% and 46% for males.14,17,21,22,25–27 Seven studies calculated a cardiovascular risk score for participants:15,16,19–21,24,28 five studies reported results.15,19,20,24,28
Apparently healthy people versus a mixed population
The British Family Heart Study Group is the only study performed in a mixed population that separately reports results for their apparently healthy subpopulation.15 Of the 2246 males, 1716 were apparently healthy, and of the 1604 females, 1321 were apparently healthy. The prevalence rates were comparable or slightly lower in the apparently healthy population. A diastolic blood pressure ≥90 mmHg was present in 38% and 23% of all males and females, and in 33% and 18% of the apparently healthy males and females, respectively. For a total cholesterol level ≥6.5 mmol/l, the prevalence rates were 22% and 18% for all males and females, and 19% and 16% for the apparently healthy males and females.
DISCUSSION
Summary
To the best of the authors’ knowledge, this is the first systematic review, having systematically searched and assessed the literature, of screening programmes to identify individuals with an increased cardiometabolic risk in primary care. The screening studies were published between 1967 and 2012. Over the past 50 years the attitude towards screening for an increased cardiometabolic risk has changed. Earlier studies focus on the benefits of screening, whereas later publications search for the best way to screen. With increasing knowledge of benefits of treating cardiometabolic risk factors, the cut-off values for the separate risk factors have become stricter.
A time trend in eligible populations was also observed. Not until 1995 was the first study focusing on an apparently healthy population published.12 In the last 8 years, four of the five existing studies on apparently healthy populations were performed and only two studies in mixed populations were published.19,20 This finding supports the authors’ assumption that people with established CVD, diabetes, hypertension, or dyslipidaemia already receive further risk assessment and treatment through existing pathways; therefore screening among them should not be necessary.
Most published screening programmes were performed in England; no studies were found performed in the US. This might be because of the requirement for actively involving GPs, as the GP has a less strong gatekeeper function in the US.
The first stepwise screening was published in 1993;22 the other three studies using a stepwise method were published in or after 2008.9,11,13 The latter studies all concerned apparently healthy populations. Particularly in these populations a stepwise method can be useful, as a substantial part of them will be healthy. A stepwise method will limit the number of people qualifying for elaborate examinations. A first step that requires action from the invited participants does not seem to influence the response rate compared with being invited for a total risk assessment at once. As only one study also invited a sample of people who did not qualify for further examinations after the first step, it was not possible to compare the discriminative ability of the different stepwise methods.
Sending reminders resulted in a higher screening uptake in apparently healthy populations. As expected, with an opportunistic approach the highest response rate was found with the longest study period.
Public awareness seems to be an important determinant of screening uptake. The highest response rate (93%) was found by Weinehall et al who describe the results of the first 8 years of the Västerbotten Intervention Programme in Sweden.27 The individual screening strategy was combined with a population strategy by creating a local health promotion collaboration to raise public awareness. The opportunistic screening by Cope et al lasted 1 week, but the screening uptake was 39%.4 The investigators advertised their health week in advance, within the general practice and by giving talks.
Devroey et al had an active approach but also the lowest response rate; people were invited by the local authorities to visit their GP for a health check.20 In most studies, people were invited by their GP; in no other study were the authorities responsible for inviting people. Six of the seven studies published in or after 2004 calculated a risk score; this reflects the increased use of risk scores in clinical practice. One might expect lower prevalence rates in an apparently healthy population than in a population with participants already diagnosed with risk factors. When comparing prevalence rates of hypertension, diabetes, or dyslipidaemia between healthy and mixed populations, no substantial differences were observed. From this, it can be concluded that a substantial part of the apparently healthy people have an increased cardiometabolic risk, which makes screening among this population worthwhile.
Strengths and limitations
Before discussing the yield of the screening programmes, some limitations of this systematic literature review need to be considered. As mentioned above, the cut-off values for risk factors have become stricter over the years, hindering the comparison of the yield over time. As a result of heterogeneity of used thresholds and risk scores, it was not possible to compare prevalence rates of cardiometabolic risk factors, nor relate the yield of the different studies to programme characteristics. Because most of the studies performed in mixed populations did not provide information about the composition of their study population, it was not possible to compare the yield of screening between healthy and mixed populations.
A systematic review often includes a risk of bias assessment of quality items like randomisation, selection bias, blinding, and loss-to-follow-up. The main interest in this study were cross-sectional aspects of screening; the focus was on the eligible population, screening method used, patient approach, response rate, and yield of screening. Because the focus was on ‘how’ rather than ‘how well’, no critical appraisal was performed on the above-mentioned items.
Another limitation of this study is that five articles were missed in the analyses as it was not possible to retrieve them full-text.
Comparison with existing literature
Ultimately, it is not the yield of the screening that makes sense, but the reduction of cardiovascular morbidity and mortality by early treatment of risk factors. Si et al examined the effect of screening versus no screening in general practice on surrogate endpoints, and found significant improvement in several cardiovascular risk factors, especially in high-risk populations.29 A Cochrane review, however, found no effect of health checks on total and cardiovascular mortality.30 The studies included in these reviews all started in the 20th century. As shown here, much has changed over the years, not only with regard to the characteristics of the screening programmes, but also with regard to new treatment insights.
Before implementing a screening programme it is first necessary to identify optimal programme efficiency, with regard to method and targeted population. This may influence the effects of screening on morbidity and mortality. In the authors’ opinion, the optimal screening strategy should identify all people with an increased cardiometabolic risk, with a minimum effort to detect these people. Defining ‘optimal’ in such a way, the authors realise that not all the Wilson and Jungner criteria are taken into account,31 such as cost-effectiveness, or psychological harm. Screening might reduce the costs for treating diseases, but screening programmes are also expensive. Therefore, it is important to use the available resources as efficiently as possible. A stepwise approach is likely to be the best way to reduce these costs through selection of a high-risk population, reducing the number of further examinations. Future research will have to prove whether stepwise screening methods are indeed more cost-effective.
Implications for research and practice
As people with established disease or risk factors are likely to receive extensive care through existing chronic care pathways, screening programmes should focus on apparently healthy people.
Stepwise methods can limit the number of people qualifying for further examinations; a stepwise method, in which the first step requires action from the invited participants, does not seem to influence response rate.
An invitation, and if necessary reminders, sent by the GP, appears to increase screening uptake.
Increased public awareness of the opportunity and relevance of screening seem to lead to a higher screening uptake.
In this review, 26 screening programmes are described for detecting people with an increased cardiometabolic risk, performed in primary care. Observed time trends were the shift in focus from a mixed population to an apparently healthy population, an increased use of risk scores, and an increasing use of stepwise methods, especially in apparently healthy populations.
In apparently healthy populations a substantial number of people were detected with an increased cardiometabolic risk, stressing the need for ongoing detection. Stepwise methods, increasing public awareness, and actively involving GPs could improve screening efficiency and uptake.
Notes
Funding
Not applicable.
Provenance
Freely submitted; externally peer reviewed.
Competing interests
The authors have declared no competing interests.
Discuss this article
Contribute and read comments about this article: bjgp.org.uk/letters
- Received February 1, 2014.
- Revision requested March 25, 2014.
- Accepted May 20, 2014.
- © British Journal of General Practice 2014