Introduction

Tobacco smoking is one of the most important risk factors for cardiovascular disease [1–3]. In patients with established cardiovascular disease, smoking cessation has been associated with improved survival [4,5] and reduced numbers of new cardiovascular events [5,6], with improvement seen already within the first years following smoking cessation [4,7]. Abstinence from smoking is therefore an important goal for secondary prevention, and offering patients support to help them quit is a high priority recommendation in national [8,9] and international guidelines [10–12].

In Sweden, around 50% of patients stop smoking after stroke or myocardial infarction, with no clear improvement of this proportion seen during the last decade [13–15]. Similarly, results from the cross-sectional EUROASPIRE surveys have found no improvement in the smoking cessation rate after coronary events during the 21^st century [16,17]. This lack of improvement indicates a need for improved strategies and/or implementation to support smoking cessation in clinical practice. Several reports have also indicated a need for improved follow-up when it comes to other modifiable risk factors in patients with manifest cardiovascular disease [16,18–20]. Smoking cessation intervention as part of a multiple risk factor follow-up program may, therefore, be the most effective way to reduce recurrence of cardiovascular events in this high-risk group.

Previous studies of smoking cessation interventions indicate that counseling provided at multiple occasions during more than 1 month may increase the proportion of patients with cardiovascular disease that quit smoking [21–24], whereas less extensive counseling interventions, initiated in-hospital or in other settings and with less than 1 month of follow-up, seem to be insufficient [21,24]. Previous studies of multiple risk factor interventions indicate that high-intensity interventions, commonly with multiple contacts during the first months after a cardiovascular event [21,22] or sometimes longer [23,25], may increase the smoking cessation rate, but the design and results of available studies are heterogeneous [21–23,25–29]. In addition, common characteristics of successful interventions are difficult to distinguish because the smoking cessation component of these interventions has not always been well described [21,22,25,26,30–32], study samples have often been small [21,22,25,30,33], and there are very few examples of individual studies with significant results [21–23,25]. Few studies have evaluated less resource-demanding multiple risk factor interventions [33,34] or programs with infrequent contacts during long-term follow-up [34]. Thus, it is difficult to draw any firm conclusions regarding the effect of such interventions on smoking cessation.

The Nurse-based Age-independent Intervention to Limit Evolution of Disease (NAILED) trial was a pragmatic, secondary preventive, randomized controlled trial with a population-based recruitment approach, performed at a single center in Sweden during the years 2010–2017. The aim of the trial was to investigate if long-term, nurse-based telephone follow-up could improve secondary preventive outcome after stroke/transient ischemic (TIA) attack and acute coronary syndrome (ACS). Previously published results from the NAILED trial have shown that this relatively simple but systematic form of risk factor intervention achieved significantly lower blood pressure (BP) and low-density-lipoprotein cholesterol (LDL-C) levels as well as a reduced recurrence of major cardiovascular events compared with usual care during long-term follow-up [35–37]. The aims of the current substudy were (1) to assess if the NAILED intervention also increased the smoking cessation rate compared with usual care, and (2) to explore the possible association between smoking cessation and prognosis.

Material and methods

Results

Baseline characteristics

Out of 3228 patients with acute stroke/TIA/ACS assessed for eligibility in the NAILED trial, 1890 were included. Among those included, 332 (17.6%) reported current smoking at baseline. The corresponding proportion was similar (19.3%) among eligible patients that declined participation (n = 384). Among included smokers, 1 participant died and 10 participants withdrew during the first month after discharge, before the intervention began. Those that remained in the study at 1 month (n = 321) constitute the study population of the current study (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study flowchart.

Participants were assessed for eligibility, included, and randomized in hospital, whereas the intervention started 1 month after discharge. Smokers that remained in the study at 1 month constitute the study population of the current substudy. The maximum duration of intervention/usual care follow-up was 3 to 8 years, depending on the year of inclusion. All analyses were performed according to the intention-to-treat principle. Smoking status at the last follow-up occasion for each participant was used to analyze the primary outcome. Participants were followed for cardiovascular and mortality outcomes until Dec 31, 2017, irrespective of withdrawal from the study’s annual follow-ups at an earlier time point.

https://doi.org/10.1371/journal.pone.0311955.g001

Baseline characteristics are described in Table 1. The mean age was 63.8 years (SD 10), 33.3% (n = 107) were women, and 20.6% (n = 66) had had previous vascular events (AMI/cardiac revascularization/stroke/TIA/peripheral artery disease) prior to the index event. The types of index events were distributed as follows: 85 participants with stroke (26.5%), 41 participants with TIA (12.8%), and 195 participants with ACS (60.7%). No significant differences were found in baseline characteristics between the intervention group (n = 161) and the control group (n = 160), but there was a numerical trend toward more participants with previous vascular events in the control group (p = 0.092).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics.

https://doi.org/10.1371/journal.pone.0311955.t001

Smoking cessation

In total, 171 participants (53.3%) reported nonsmoking at their last follow-up: 81 patients (50.3%) in the intervention group and 90 patients (56.3%) in the control group, with no significant difference between groups (absolute difference 5.9%, 95% CI −5.0 to 16.7, p = 0.286). Correspondingly, we found no between-group difference in smoking status at 1 year or in the proportions that reported smoking cessation at some time point during the course of the study (Fig 2). However, at 3 years there was a significant difference in favor of the control group (absolute difference 12.8%, 95% CI 1.3–23.9, p = 0.029).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Self-reported smoking cessation.

Proportions according to intention-to-treat: The denominator includes all patients alive at the respective time points and continued smoking was assumed for participants with missing values due to study withdrawal or other reasons. The numbers of cases with missing values in the intervention and control groups, respectively, were 0 vs. 1 at 1 year, and 21 vs. 9 at 3 years. *Significant between-group difference in favor of the control group (p = 0.029).

https://doi.org/10.1371/journal.pone.0311955.g002

About one-third (n = 101, 31.5%) of the participants reported continued smoking at all their follow-ups. Among those who reported nonsmoking at their last follow-up, 80.7% (n = 138) had already reported nonsmoking at 1 month. Among patients that reported nonsmoking at 1 month (n = 179, 55.8%), 33.5% (n = 60) had at least one relapse during follow-up, and 77.1% (n = 138) reported nonsmoking at their last follow-up, again with no significant difference between groups.

Smoking cessation and recurrence of cardiovascular events and total mortality

A total of 74 participants (23.0%) reached the composite endpoint of CV events and 64 participants (19.9%) died from any cause (Table 2). The incidences of CV events and all-cause death were 4.9 and 3.8 per 100 years at risk, respectively. Smoking cessation at 1 month was associated with a lower cumulative incidence of all-cause death (Fig 3B). A similar trend was seen for CV events, but the difference was not statistically significant (Fig 3A). The sensitivity analyses, including smoking status as a time-varying variable and accounting for all-cause mortality as a potential competing risk variable, did not change any results.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3.

Association of smoking cessation at 1 month with occurrence of (A) new cardiovascular events and (B) all-cause death.

https://doi.org/10.1371/journal.pone.0311955.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Association of self-reported smoking cessation at 1 month with incidence of new cardiovascular events and all-cause death.

https://doi.org/10.1371/journal.pone.0311955.t002

Discussion

In this study, we found no effect of brief smoking cessation advice delivered by a nurse as part of telephone-based follow-up of cardiovascular risk factors in patients with stroke, TIA, or acute coronary syndrome. Compared with usual care, the intervention did not improve smoking cessation proportions in the long or short term, it did not seem to increase smoking cessation attempts overall, and we found no effect regarding sustained abstinence among those that quit smoking shortly after the index event (i.e., before the intervention began). Smoking cessation at 1 month was associated with improved survival during long-term follow-up. A similar trend was seen for CV events, but this difference was not statistically significant. Besides a positive effect of smoking cessation itself, differences in baseline characteristics between those that quit smoking early after the index event and those that did not may have contributed to the observed difference in prognosis.

Smoking cessation support is a high priority recommendation of current guidelines [9–12], but recommendations regarding specific strategies are often vague [8,11,12] and sometimes include brief counseling [9,10,12], despite little evidence for the effectiveness of such interventions among patients with manifest cardiovascular disease. In this context, our study is an important contribution and the results are in line with previous studies pointing at no short-term [31,33,34] or long-term [34] effect of counseling delivered on a single occasion [31,33] or repeatedly [34] during long-term follow-up. In a previous Swedish single-center RCT, lifestyle counseling and general risk factor assessment by a nurse at 3 months after stroke did not reduce the proportion of current smokers at 12 months compared with usual care [33]. In another RCT, performed among patients with stroke or high-risk TIA in Austria, multidisciplinary follow-up at 3 months and access to a web-based platform with educational and interactive elements had no effect on smoking cessation rates at 12 months [31]. In a single-center RCT from Norway, no effect on smoking cessation rates at any time point was found during 5 years of follow-up [34]. This Norwegian study compared a nurse-based, long-term, secondary preventive follow-up program with usual care among consecutive patients aged <80 years and hospitalized due to MI or scheduled PCI/CABG. The intervention had slightly more extensive follow-up during the first year compared with NAILED and included an offer of pharmacotherapy (nicotine replacement therapy/varenicline) as part of the intervention.

The role of smoking as a causal factor in cardiovascular disease is well established [40]. It has also been shown in several observational studies that smoking cessation is associated with reduced recurrence of cardiovascular events and death [5,6,41]. The relative risk reduction for all-cause death in our study was almost 50%, which is similar [41] or perhaps even more pronounced [5,6] than the estimates seen in previous observational studies. It should, however, be noted that the confidence interval in our study was wide and the true risk reduction could therefore be considerably larger or smaller. Similarly, the low number of participants contributing time at risk at the end of follow-up introduces uncertainty regarding the results for the cardiovascular event outcome, possibly contributing to a type 2 error. Also, as in previous studies, it is difficult to isolate the effect of smoking cessation alone, since smoking cessation was also associated with other characteristics that were of prognostic importance such as age, the type of index event, and (at a borderline significance level) previous vascular events. Age is inevitably associated with death. Previous manifestations of vascular events could indicate more severe vascular disease and thereby a worse prognosis. It can also be speculated that patients who continue to smoke despite having repeated manifestations of vascular disease are a rather select group that could have additional prognostic characteristics that are perhaps difficult to measure, such as other lifestyle-related factors, social context, or background. Few RCTs of smoking cessation interventions have evaluated their effects on cardiovascular and mortality outcomes. Mohiuddin et al. compared the addition of an intensive, postdischarge smoking cessation program based on multiple group sessions to in-hospital counseling alone in patients with ACS or decompensated congestive heart failure. They found an effect on mortality with a point estimate (RRR 77%; 95% CI 27%–93%; p = 0.014) suggesting a larger effect than in the observational studies. This was, however, a small study with few events, which limits the precision of the point estimate, and the follow-up was only 24 months [4].

Having a potentially life-threatening event such as a myocardial infarction or a stroke may be an eye-opener, increasing a smoker’s motivation to quit. Indeed, high smoking cessation rates in the early time period following a myocardial infarction or stroke has been commonly observed in many previous studies [23,28,30,31,34,42,43] and, of note, these proportions are generally larger than any additional effect achieved by smoking cessation interventions [21,22,24,44,45]. In line with previous studies [23,28,30,31,34,42,43], we found that most patients that quit smoking after a cardiovascular event did so shortly after the event. Also, the rates of additional quitting attempts and relapse during the first year were very similar to previously observed rates [34,43]. In our study, smoking cessation, especially early smoking cessation, was more common among patients with ACS as the qualifying event as compared with patients with stroke or TIA. A similar pattern was observed in a large cohort from the Netherlands [6] and may indicate that cardiac events were perceived as more serious by the patients or, perhaps, more clearly linked to smoking.

Of note, the incidence of cardiovascular events and death was numerically higher in the group that continued to smoke after the index event compared with the NAILED population as a whole [37], despite a considerably lower mean age. This emphasizes the importance of finding effective methods to support smoking cessation in this very high-risk group. Provision of smoking cessation programs is listed as one of several interventions included in the WHO framework convention on tobacco control [46] that entered into force in 2005 and has since been ratified by 183 parties including Sweden. In accordance with the convention, progressive policy and legislative changes have been implemented during the last 20 years [47,48], and many countries have seen a concurrent decline in smoking prevalence in the general population [49]. However, the proportions of smoking cessation after MI and stroke have not changed [13–17], indicating failure when it comes to implementation of effective smoking cessation interventions for this group within health care. The 1-year proportions of smoking cessation in our study are similar to the corresponding proportions reported in Swedish quality-of-health-care registers [14,15]. In addition, our study contributes data on smoking status during long-term follow-up, showing that this proportion unfortunately also remained largely the same beyond 1 year.

Within publicly funded health care, as in Sweden, it is a constant challenge to use available resources in a way that maximizes health benefits for as many as possible. In that context, the design of the NAILED intervention, a pragmatic attempt to systematically implement recommendations from secondary preventive guidelines through a relatively low-resource-intensity approach, makes a valuable contribution. Whereas communication and education about the role of smoking should be integrated into standard care, more resource-demanding interventions directed toward smoking need to be carefully worked out to be feasible and effective among those most in need of support. The epidemiological results of our study provide increased knowledge about the pattern of smoking cessation after cardiovascular events and the factors associated with continued smoking, which is important to help identify those in need of support to stop smoking. According to our results, patients that do not quit in association with a cardiovascular event are unlikely to do so during long-term follow-up. It is also apparent that repeated advice as part of an annual follow-up of other risk factors is not sufficient to make a difference in this very high-risk subgroup of patients. Instead, future studies should focus on improving the evidence for the feasibility and effectiveness of more intensive smoking cessation support among patients that present at hospital with a recurrent cardiovascular event or patients that report continued smoking at short-term follow-up after a first-time event. Based on previous studies, such an intervention should consist of repeated counseling with motivational and behavioral elements for at least 1 month and, preferably, also an offer of pharmacotherapy with systematic follow-up as part of the intervention. To be clinically relevant, it is important that studies include consecutive patients, carefully note the proportions of patients that decline participation or discontinue intervention, and assess both short-term and long-term effects on smoking cessation. To motivate both patients and the direction of resources to this type of intervention, it is important to increase knowledge of the quantitative effects of smoking cessation on new cardiovascular events and survival. Future studies should therefore also include these outcomes when evaluating the effects of smoking cessation interventions.

The current study has strengths as well as limitations. As part of the NAILED study, all consecutive patients in a geographically defined part of the country were screened for inclusion, resulting in a study sample with high resemblance of the background population in which multiple risk factor intervention for the prevention of recurrent cardiovascular events is indicated. The subgroup of current smokers in the NAILED study was not selected based on motivation to quit, and the outcome was analyzed according to the intention-to-treat principle, thereby providing realistic estimates of the effect of the intervention in clinical practice as well as epidemiological data with high external validity.

We did not use any objective method to verify the accuracy of self-reported smoking cessation, and thus it is possible that the proportions are slightly overestimated. We do, however, not find any reason to believe that the accuracy should differ between the treatment groups. In addition, the annual assessment of smoking status represented the point prevalence at those occasions; that is, we did not capture all smoking cessation attempts and relapses between contacts. Thus, the proportions of patients with registered attempts and relapses are probably underestimated. Also, the dichotomous assessment of smoking status did not capture quantitative changes in smoking habits, and therefore we do not know if there were any such changes over time or a difference between allocated treatment groups. Finally, we do not know to what extent smoking cessation support was provided in primary health care, what kind of support that was given, and to what extend pharmacotherapy was used in any of the groups. That limits our ability to analyze the overall trend toward a benefit for the control group regarding smoking cessation. Since the smoking cessation proportion in the control group was similar to corresponding data from Swedish quality registers, we do, however, believe that the support given was probably comparable to that in other regions of the country. It should also be noted that we did not adjust for multiple comparisons, increasing the risk that the between-group difference seen at 3 years could represent a type I error.

Conclusion

Patients that do not quit smoking in association with a cardiovascular event are unlikely to do so during long-term follow-up, and continued smoking is associated with an increased risk of all-cause mortality. Annual, brief information and advice regarding cessation delivered by a nurse as part of long-term, multiple risk factor follow-up is not sufficient to improve the proportion of patients that quit smoking after ACS, stroke, or TIA.

Supporting information

Acknowledgments

The authors would like to acknowledge statistician Lars Söderström for his statistical guidance and for contributing to the statistical methods and integrity of the results.