Smoking cessation and the risk of cardiovascular disease outcomes predicted from established risk scores: Results of the Cardiovascular Risk Assessment among Smokers in Primary Care in Europe (CV-ASPIRE) Study

Tobacco is the only legally available consumer product which kills up to one in every two users when used as intended [1]. Smoking kills over 650,000 people in the European Union every year [2]. Smoking is a major risk factor for CVD, and can result in a seven-fold increase in the risk of peripheral arterial disease and at least a two-fold increase in the risk of coronary artery disease [3–5]. The INTERHEART study showed that if a person stopped smoking, they reduced their risk of myocardial infarction by about 65% [6].

Several multi-factorial risk models have been developed to assess the risk for development of CVD, and all of these models include tobacco use as an important predictor, along with age, blood pressure and cholesterol levels, highlighting the impact of smoking on cardiovascular risk. The Framingham Risk Score quantifies the absolute level of CVD risk based on a number of risk factors [7]. In addition to CVD risk, there have been multiple models proposed using the Framingham data to calculate various other absolute risk levels, including CVD mortality, coronary heart disease, intermittent claudication and recurring coronary heart disease [8–12].

In Europe, the Systematic COronary Risk Evaluation (SCORE) system was initiated to develop a risk scoring system for the management of cardiovascular risk in European clinical practice. The SCORE data contains approximately 3 million person-years of observation from 205,178 subjects [13–16].

General practitioners (GPs) in primary care practice play an important role in smoking cessation through measures such as counseling and pharmacotherapy [16]. Research has shown that simple advice provided by GPs can help substantially increase smoking cessation rates [17]. Data from studies on self-reports by GPs show that while the majority of GPs (62% to 98%) question all new patients as to their smoking status, the proportion of GPs offering more intensive interventions or prescribing treatments for cessation is generally lower, and also has shown to differ across and within countries [18–21]. Studies have found that the rates of GPs providing smoking cessation advice vary across European countries [18, 19, 22, 23]. Moreover, GP’s engagement in providing smoking cessation advice to patients depends on factors such as the GP’s smoking status, their attitude towards smoking, their relationship with the patient, and their knowledge of existing smoking cessation interventions [21, 24]. Researchers advocate the development of country-specific strategies for encouraging and supporting a GP’s smoking cessation activity [21].

We performed this European, multi-centre, observational, cross-sectional survey of smokers in the primary care setting to provide estimates of the risk of CVD among the smoking population attending primary care in Europe, using standard risk assessment tools. This study also reports on patient behaviors related to smoking as well as physicians practice patterns.

We performed this cross-sectional survey involving smokers in the primary care setting, selected from 3 European countries (Greece, Italy and Netherlands). These countries were chosen because they met the study data requirements and provided a geographic spread within Europe in an attempt to create a dispersed distribution. GPs were required to routinely collect lipid levels and blood pressure from smokers as standard practice based on their availabilities, their acceptance to the protocol and informed consent form and their ability to enroll the target population based on their response to their following questions:

Out of 720 possible sites, 229 met these criteria and were selected.

The final protocol and case report form, any amendments, and informed consent documentation were reviewed and approved by regulatory authorities required in each country.

Participating GPs assessed all smoker patients 40 years or older for their suitability for enrolment into the survey at the time that they attended the centre regardless of their reason for the consultation. Patients were selected consecutively from the patients that visited their doctor on a routine basis irrespective of the reason for their visit. The recruitment period for the survey was flexible to enable a maximum of 15 patients per GP. There was no set time limits for recruitment into this study. Patient recruitment continued until there were sufficient patients (approx 1400) to provide adequate precision for the study endpoints. Informed consent was obtained for those patients who met the study inclusion/exclusion criteria and who agreed to participate. Suitable patients smoked at least 10 cigarettes per day during the previous month and had blood pressure and lipids level data in their clinical records (within the last 12 months).

Once informed consent was obtained, the investigator administered the survey to patients and completed the study case report form. Demographic information was collected on age, gender, height and weight. Laboratory parameters included total cholesterol, High-density lipoprotein (HDL), Low-density lipoprotein (LDL) and systolic and diastolic blood pressure. Information on medical history was also captured (history of cardiovascular disease, diabetes, hypertension, hypercholesterolemia and chronic obstructive pulmonary disease (COPD)) including reason for consultation and family medical history.

Patients were asked whether they had persistent chronic cough, excess mucous and phlegm, shortness of breath, or wheezing, and whether they were currently being treated for hypertension, hypercholesterolemia, diabetes, or COPD. Parameters on patient behaviors related to smoking included the Fagerström Test for Nicotine Dependence (FTND) [25], years of smoking, previous attempts to quit smoking, previous use of any smoking cessation programs, longest time without smoking, interest in quitting smoking, and interest in type of smoking cessation support. Practice patterns related to smoking behavior included asking the patient whether they were provided with smoking cessation advice and whether any behavioral or pharmacotherapy support was offered.

Information about the clinic was also collected, including number of practitioners, patient volume and physician smoking status. To assess typical clinical practice patterns related to smoker behavior, we asked the clinician whether typically in their practice smoking cessation advice, behavior support or pharmacotherapy is offered to patients who have an interest in quitting smoking.

To our knowledge, CV-ASPIRE was one of the largest cross-sectional surveys of its kind to explore health risks and prevalence of comorbidities in smokers in the primary care setting and where GPs in Europe were surveyed on issues concerning patient behavior and physician attitudes toward smoking cessation. As expected, we found both patient behavior and physician attitudes toward smoking cessation to vary by country. A possible explanation for these apparent differences in practice patterns could perhaps include differences in the attitude towards the role of GPs in influencing smoking cessation. Differences in the pattern of practices and the negligence of prevention in primary care seem to be the explanation for this finding in some countries [26]. These results are consistent with previous findings which showed varying rates of GP advice on smoking cessation across Europe [18, 19, 22, 23].

A total of 49% of smokers reported at least one COPD related disorder, however only 20% of were reported as diagnosed with COPD. According to the Global Initiative for Chronic Obstructive Lung Disease COPD reference guide, “A diagnosis of COPD should be considered in any individual who has dyspnea, chronic cough or sputum production, and/or a history of exposure to risk factors for the disease, especially cigarette smoking” [27]. This finding suggests that COPD may be under diagnosed in smokers.

In Greece and Italy, the primary reason for consultation with their GP was general medical examination (57%); this was notably higher than in the Netherlands (16%), where the primary reason for consultation was cardiovascular related (32%). One of the factors that may help explain this is that more patients may be visiting their GP in Italy and Greece in order to obtain a repeat prescription. In rural Greece it has been shown that over half of patients visit their GP solely to ask for a prescription, and 70% of prescriptions are renewal of an older one [28].

Interestingly, the perception of GPs and patients differed with regards to the types of smoking advice offered to smokers in the primary care setting. For instance, a higher proportion (73%) of Italian patients reported that their GPs offered smoking advice compared to Italian GPs themselves (48%). The referral for behavioral support and pharmacotherapy was reported in the 3 participant countries consistently lower by patients than by GPs. Perhaps the difference in reporting rates could be explained in part by recall bias where GPs more readily remember the delivery of service rather than their failure to offer the service. They could also be explained through prevarication bias such that GPs could be more inclined to state that they offered the advice to their patients while patients are more inclined to deny that such advice was ever offered to them.

Demographic characteristics were for the most part distributed similarly by country (Table 3).

Various prediction models were employed in this study to estimate the absolute risk of outcomes previously established to be associated with smoking. A hypothetical population of non-smokers, derived directly from the prediction model (by setting smoking status to non-smoking), was applied in these models. This approach allowed us to mimic a counterfactual scenario where an identifiable cohort of smokers could be classed as non-smokers. The main advantage of generating a hypothetical population of non-smokers from a prediction model is that it avoids the need for sampling non-smokers into the study. Moreover, it also provides control of all confounders included as part of the prediction model since the prediction model assumes that only those variables included in the model are relevant for the prediction of events.

A limitation of this approach, however, is that the model may underperform in the real-world setting given that any known confounding factor not included in the prediction model or an unknown confounding factor may influence outcomes. For instance, the SCORE model only considers age, gender, smoking status, systolic blood pressure, and lipid levels to be important in the prediction of death due to cardiovascular disease and ignores potential important covariates such as alcohol use, family history, diet, exercise, and concomitant medications. Moreover, these prediction models do not provide complete information about what might have happened if a long-term smoker was to quit smoking, since the models themselves do not account for past exposure to smoking and its potential effects on future outcome. Another limitation, is that the simulated non-smokers have identical blood pressure and lipid levels as the smokers. It has been found that smokers tend to have lower HDL [3] and higher blood pressure [29] than non-smokers. Since the simulation is not accounting for this, the results may be considered a conservative estimate of the impact on CV risk of smoking.

Despite these limitations, the simulated population of non-smokers provided a convenient population from which to derive the relative risk attributable to smoking for cardiovascular events. The relative risk was reported in this study because it quantifies the risk in a clear and interpretable percentage that can easily be related to the impact of smoking on cardiovascular events.

The prediction models found expectedly that smokers experienced a quantitatively higher level of risk in each of the cardiovascular event categories considered when compared to a hypothetical population of non-smokers sharing the same risk factor profile. The directionality of the association was consistent (i.e., smokers were always found to be at higher risk) regardless of the prediction model. We were able to predict, for instance, through the SCORE model that this study population of smokers have a 100% increased probability of death due to cardiovascular disease in the next 10-years compared to non-smokers. The relative risk derived using the Framingham models for 10-year risk of coronary heart disease and 10-year risk for “hard” coronary heart disease showed an increase in risk of 50% (from 12% to 18%) and 65% (from 8% to 13%), respectively. The Progetto CUORE model found a similar risk increase of 68% (from 6% to 10%) for coronary or cardiovascular events.

The discrepancy in results across the prediction models just described can be explained by differences in the outcome of interest. None of the scores predicted risks are directly comparable. For example, the SCORE model predicts death due to cardiovascular disease whereas the Framingham coronary heart disease score predicts new cardiovascular events, not necessarily resulting in death. Similarly, the Progetto model predicts the risk of a major cardiovascular event, in comparison to the Framingham coronary heart disease model which predicts the risk of coronary heart disease, and one would not expect all patients diagnosed with coronary heart disease to necessarily experience a major event. Another reason why the model predictions may differ is the underlying population used to derive the models: the Framingham score was derived using a population from the United States; the SCORE model a European and the Progetto an Italian population.

The eligibility criteria for this study was restricted to individuals 40 years of age or over, which consequently may limit the generalizability of results to a younger population or those who have smoked for a shorter duration of time. Subjects were also required to have available information on blood pressure and lipids within the past 12 months of the study. Given that subjects with a history of cardiovascular disease are more likely to have available measures on the above laboratory parameters, we expect that this study population would have a higher prevalence of cardiovascular disease than a general population of smokers. Moreover, the generalizability of this study may be biased towards those at higher risk of CVD.

The findings of this study reinforce the importance of smoking as a significant predictor of long-term cardiovascular events. The consistency in the elevation of relative risk found across prediction models reinforces the need to reduce and eventually eliminate this important source of morbidity and mortality from the general population through effective smoking cessation programs. Future country-specific interventional strategies for smoking cessation should consider targeting GP practices given the GPs’ role in patient care.

It has been previously shown that smoking is the second most important modifiable risk factor associated with myocardial infarction, after LDL: HDL ratio [30], and the excess risk of experiencing a heart attack is halved within a year of quitting in comparison to continuing smokers [31]. Patients attending GP practices diagnosed with hypercholesterolemia are routinely treated for the condition. However, smokers are far less likely to receive smoking cessation treatment when visiting their GP, there may be an unfounded disparity in treatment for smoking compared to other risk factors. According to the United States Department of Health and Human Services “it is difficult to identify any other condition that presents such a mix of lethality, prevalence…and neglect …despite effective and readily available interventions” [32]. Further analyses may show that one of the best gains in health could be obtained by tackling the most important modifiable risk factors, smoking being one of them. The findings in this study may help future strategies for focusing smoking cessation programs in the primary care setting.

PM: Senior Medical Manager, Primary Care BU Europe, Pfizer, Walton Oaks, Surrey, KT20 7NS, UK, email: Pablo.Mallaina@pfizer.com. CL: Professor of General Practice and Primary Care, Head of Clinic of the Social and Family Medicine, School of Medicine, University of Crete, Greece, email: lionis@galinos.med.uoc.gr. HR: Doctor of Forensic Medicine, Bennebroek Primary Care Center, The Netherlands, email: roverz@hotmail.nl. RI: Medical Practitioner, Primary Care Varese ASL, distretto di Tradate, Italy, email:renzoimp@libero.it. AB: Statistical Scientist, Real-World and Late Phase Research, Quintiles, Reading, UK, email: Andrew.Burgess@Quintiles.com. MN: Director, Late Phase Biostatistics, Real-World and Late Phase Research, Quintiles, Reading, UK, email: Mark.Nixon@Quintiles.com. FM: Senior Medical Manager, Primary Care BU Europe, Pfizer, Primary Care BU Europe, Walton Oaks, Surrey, KT20 7NS, UK, email: Franco.Mondello@pfizer.com.