Multivariable analysis for smoking cessation at 6 months after myocardial infarction. CI indicates confidence interval; OR, odds ratio; PHQ, Patient Health Questionnaire.21
Customize your JAMA Network experience by selecting one or more topics from the list below.
Dawood N, Vaccarino V, Reid KJ, et al. Predictors of Smoking Cessation After a Myocardial Infarction: The Role of Institutional Smoking Cessation Programs in Improving Success. Arch Intern Med. 2008;168(18):1961–1967. doi:10.1001/archinte.168.18.1961
Smoking cessation after myocardial infarction (MI) is an important goal for secondary prevention of mortality. Whether new initiatives to promote cessation improve patients' quit rates after MI is unknown.
The Prospective Registry Evaluating Outcomes After Myocardial Infarction Events and Recovery (PREMIER) enrolled 2498 patients with MI from 19 US centers between January 2003 and June 2004. Smoking behavior was assessed by self-report during hospitalization and 6 months after an MI. Extensive sociodemographic, comorbidity, psychosocial, disease severity, and treatment data were collected by interview and medical record abstraction. Hierarchical multivariable logistic regression models with random site effects were constructed to predict smoking cessation 6 months after admission, with a focus on the presence of an inpatient smoking cessation program as a hospital-level covariate.
Among 834 patients who smoked at the time of MI hospitalization, 639 were interviewed and reported their smoking habits 6 months post-MI (77%). Of these, 297 were not smoking at 6 months (46%). The odds of smoking cessation were greater among those receiving discharge recommendations for cardiac rehabilitation (odds ratio [OR], 1.80; 95% confidence interval [CI], 1.17-2.75) and being treated at a facility that offered an inpatient smoking cessation program (OR, 1.71; 95% CI, 1.03-2.83). However, medical chart–based individual smoking cessation counseling did not predict smoking cessation rates (OR, 0.80; 95% CI, 0.51-1.25). Patients with depressive symptoms during the MI hospitalization were less likely to quit smoking (OR, 0.57; 95% CI, 0.36-0.90).
While individual smoking cessation counseling was not associated with smoking cessation post-MI, hospital-based smoking cessation programs, as well as referral to cardiac rehabilitation, were strongly associated with increased smoking cessation rates. Such programs appear to be underutilized in current clinical practice and may be a valuable structural measure of health care quality. Moreover, smoking cessation programs should likely incorporate screening for and treating depressive disorders.
Smoking has long been recognized as an important risk factor for acute myocardial infarction (MI) and coronary heart disease (CHD).1,2 Smoking cessation is associated with a reduction in readmission rates and up to a 50% lower mortality in patients with established cardiovascular disease.3-5 In fact, smoking cessation may be more effective in reducing mortality in patients with MI than therapy with aspirin, β-blockers, or angiotensin-converting enzyme inhibitors.6 However, despite the clear benefits of smoking cessation after MI, only one-third to one-half of patients who smoke at the time of MI subsequently quit.7
In light of the significant benefits of smoking cessation, the American Heart Association and American College of Cardiology have deemed counseling at the time of MI discharge to be a class I indicated behavior.8,9 Moreover, the Centers for Medicare & Medicaid Services and The Joint Commission (formerly the Joint Commission on the Accreditation of Hospital Organizations) have embraced smoking cessation counseling as a performance measure of health care quality.10,11 Despite these efforts, the effectiveness of smoking cessation interventions such as these have been inconclusive in patients with CHD.
Effective inpatient smoking programs have been developed and validated for medical, surgical, and cardiac patients, including post-MI patients. Among cardiac patients, smoking quit rates in patients enrolled in smoking cessation programs range from 28% to 54% (median, 31%), and improvements in absolute quit rates range from 7% to 36% (median, 15%).12 In randomized controlled trials, smoking cessation programs have been shown to be a cost-effective method to significantly increase quit rates in patients with MI.13-16 Yet many hospitals do not provide such programs as part of routine MI care, possibly because providing such a program is not considered in quality performance measures. Instead the focus of hospitals is to improve documentation of smoking cessation counseling at discharge after MI. While efforts to systematically document smoking cessation counseling may improve such documentation, these efforts may not achieve the intended goal of getting smokers to quit smoking. We hypothesized that the presence of smoking cessation programs and referral to cardiac rehabilitation programs might be associated with higher smoking cessation rates after MI because both programs deliver a greater intensity of smoking cessation advice and guidance than mere documentation of smoking cessation counseling. Accordingly, we examined the smoking cessation rates among smokers recovering from an MI in the multicenter Prospective Registry Evaluating Outcomes After Myocardial Infarction Events and Recovery (PREMIER) registry.17
The methods of the PREMIER study have previously been described.17 Briefly, patients were prospectively enrolled between January 1, 2003, and June 28, 2004, from 19 US hospitals to examine processes of care and patients' health outcomes as a function of sociodemographic, clinical and health status characteristics, with the purpose of improving MI care. All patients with positive troponin findings or elevated creatine kinase myocardial band fraction in the initial 24 hours of admission were screened for eligibility. Subjects were included if they were 18 years or older, had elevated levels of cardiac enzymes or biomarkers within 24 hours of arrival to the study hospital, and had additional clinical evidence of MI (eg, prolonged ischemic symptoms or ST changes in the admitting electrocardiogram). Subjects were excluded from PREMIER if they were transferred to the study institution from another facility more than 24 hours after their original presentation; were unable to provide or refused consent; did not speak English or Spanish; or were discharged or died before being contacted by the site coordinator. For the purpose of the present study, we restricted the analysis to those patients who were self-reported smokers at the time of MI. Patients gave informed consent, and the research protocol was approved by the institutional review board at each participating center.
Each patient enrolled in PREMIER underwent a detailed interview at the time of hospitalization to collect information on sociodemographic, behavioral, and psychosocial status as well as smoking behaviors. In addition, a comprehensive medical chart abstraction was performed at each site during the MI admission to further ascertain patients' demographic characteristics, medical history, clinical status, hospital treatments, and discharge recommendations. Centralized telephone follow-up interviews were conducted 6 months after MI admission to reassess relevant clinical and behavioral characteristics and smoking habits.
Smoking behavior was assessed by asking patients how many cigarettes per day they smoked, on average, and at what age they started smoking. The smoking behavior questionnaire was based on questions from national surveys, including the Behavioral Risk Factor Surveillance System, Society for Research on Nicotine and Tobacco, and the Question Inventory on Tobacco.18,19 Patients were classified as smokers if they had had even a puff of tobacco smoke in the past 30 days. Information on availability of a smoking cessation program at the admitting hospital was obtained through a site survey in which all participating hospitals were asked whether they offered an inpatient smoking cessation program to smokers. Information on individual smoking cessation counseling during the hospital admission was obtained through querying the documentation of such advice by the health care professionals using medical chart abstraction performed after discharge. Referral to cardiac rehabilitation is defined as an order in the medical chart referring a patient to cardiac rehabilitation at discharge.
Economic burden was defined by patient interview questions in which patients reported avoiding health care owing to cost over the past year. Social support was assessed by the Enhancing Recovery in Coronary Heart Disease Social Support Inventory (ESSI),20 a 7-item scale developed for, and validated in, patients with MI. The range of ESSI scores is 0 to 27, higher scores indicating greater social support. Depressive symptoms were assessed by means of the 9-question Primary Care Evaluation of Mental Disorders Brief Patient Health Questionnaire (PHQ).21 Patients were classified as depressed if they scored 10 or higher on the PHQ, which corresponds to a level of at least moderate depression and represents the minimum number of symptoms required for the diagnosis of major depression. Data were entered at each participating institution and transmitted to the national data coordinating center using a Web-based data entry interface (Outcome Inc, Cambridge, Massachusetts) that allowed front-end range and logic checks to ensure accuracy of collected data.
The primary outcome of this study was smoking cessation at 6 months. Smoking behavior at 6 months was assessed by telephone interview using the same questionnaire used at baseline. Patients were classified as having quit if they had not smoked, even a puff, within the past 30 days.
Because current assessments of health care quality use only information available at the time of hospital discharge, only those variables known at the time of hospital discharge were analyzed. Baseline patient characteristics were compared between patients who quit smoking and those who continued smoking 6 months after their MI. We used t tests to compare continuous variables and χ2 or Fisher exact tests for categorical variables, as appropriate. To test whether or not availability of an inpatient smoking cessation program in the admitting facility was effective in achieving better quit rates, we compared the smoking cessation rates of smokers admitted to hospitals with smoking cessation programs with those admitted to hospitals without such programs. Multivariable, hierarchical logistic regression modeling, using site as a random effect to account for clustering of patients by site, was used. Variables that had a statistically significant bivariate association, or were considered a priori to be clinically meaningful, were included in the final model. The final model included sociodemographic variables (age, race, sex, marital status, and economic burden), medical history (alcohol use, cocaine use, diabetes mellitus, lung disease, prior MI, and prior coronary revascularization procedures), clinical status on admission (ST elevation MI and congestive heart failure), psychosocial variables (depression and social support), and referral of patients to cardiac rehabilitation, smoking cessation counseling, and availability of a smoking cessation program at the admitting hospital as a site-level variable. Logistic regression model fit and discrimination were assessed using the Hosmer and Lemeshow22 goodness of fit test and the model C statistic, respectively.
In secondary analysis, to evaluate whether there were any other factors common to hospitals with smoking cessation programs that may have affected likelihood to quit smoking, we compared baseline patient characteristics between patients admitted to hospitals with smoking cessation programs and those without smoking cessation programs. In addition, because 21% of the patients did not participate in the 6-month interview, we sought to evaluate any potential biases arising from the exclusion of missing 6-month outcome data. In the overall sample, including patients without 6-month interviews, but excluding patients who had died by the time of the 6-month follow-up, we calculated a propensity score for missing 6-month outcome data using nonparsimonious logistic regression analyses. Variables used in the propensity analysis included site of enrollment, all demographics, socioeconomic and lifestyle factors, clinical characteristics, laboratory findings, disease severity, baseline health status, medications, and acute and nonacute treatments received during the patients' initial MI hospitalization (Table). From these models, a probability of failure to complete an interview was calculated. To assess the bias from patients lost to follow-up, the reciprocal of this probability was assigned to each patient to more heavily weigh the behaviors of patients most likely to be similar to those missing follow-up.23 All tests for statistical significance were 2-tailed with an alpha level of .05. All analyses were conducted using SAS software, release 9.1 (SAS Institute, Cary, North Carolina) and R, version 2.1.0.
A total of 2498 patients with MI were enrolled in the PREMIER study between January 1, 2003, and June 28, 2004. Thirty-three percent smoked at the time of hospitalization and were discharged alive, thus forming the population for this study (n = 834). From this sample, we excluded patients who died within 6 months (n = 26). Of the remaining 808 patients, 169 did not have follow-up smoking data available (21%), leaving 639 patients available for the final regression analysis. Smokers without follow-up data were similar in age to smokers with follow-up data; however, they were less likely to be white (62% vs 72%) or married (35% vs 55%) and were more likely to have less than a high school education (69% vs 55%). Smokers who were lost to follow-up were more likely to be male (79% vs 67%), have a history of prior MI (28% vs 16%), and have a history of chronic lung disease (21% vs 12%) (P < .05 for all comparisons).
The mean (SD) age of our study population was 54 (10) years with 69% being male (n = 576) and 69% white (n = 576). Forty-six percent of patients were not smoking 6 months after their MI (n = 297). Patients who quit smoking were more likely to be married, have a higher income, and enjoy a higher level of social support (Table). Patients who quit smoking were less likely to have a history of alcohol or cocaine abuse, depression, a prior MI, prior percutaneous coronary interventions, or congestive heart failure. Age, sex, education, average number of cigarettes smoked per day, and the duration of smoking history were not associated with smoking cessation.
Several important features about patients' exposure to smoking cessation advice were observed. Ten of 19 hospitals offered smoking cessation programs. Interestingly, while those who did and did not quit were equally likely to have had medical chart–based individual level counseling to quit smoking (75% vs 72%) (P = .36), patients who quit smoking were significantly more likely to be admitted to a hospital offering smoking cessation programs (69% vs 56%) (P < .001) (Table). Patients who quit smoking were also more likely to have been referred to cardiac rehabilitation at discharge (63% vs 47%) (P < .001).
After adjusting for demographic factors, medical history, and clinical status on admission, several characteristics were independently associated with quit rates at 6 months. Patients with a history of cocaine use (odds ratio [OR], 0.26; 95% confidence interval [CI], 0.09-0.75) and those with significant depressive symptoms as indicated by a PHQ score of 10 or higher (OR, 0.57; 95% CI, 0.36-0.90) were significantly less likely to quit smoking (P <.05). Patients with a documented referral to cardiac rehabilitation were more likely to quit (OR, 1.80; 95% CI, 1.17-2.75). Importantly, documented individual counseling to stop smoking was not associated with quitting (OR, 0.80; 95% CI, 0.51-1.25), while admission to a hospital with an inpatient smoking cessation program was associated with quitting smoking after discharge (OR, 1.71; 95% CI, 1.03-2.83). The model C statistic was 0.71. These results are graphically presented in the Figure. In the multivariable models, the absolute risk reduction for smoking cessation was 17% for referral to cardiac rehabilitation (mean predicted probabilities for quitting for those with and without referrals to cardiac rehabilitation, 54% and 37%, respectively) and 15% for smoking cessation program (mean predicted probability for quitting for admission to a hospital with and without a smoking cessation program, 52% and 37%, respectively).
In secondary analysis, hospitals with inpatient smoking cessation programs were more likely to admit patients with a lower economic burden than were hospitals without such programs (30% vs 28%) (P = .04). Hospitals with and without such programs did not differ in patient comorbidity including depression, documented smoking cessation counseling, or referral to cardiac rehabilitation.
To examine whether patients without follow-up might have influenced these observations, the analyses were repeated with a propensity-based weighting score based on patients' likelihood not to participate in follow-up. No significant changes in the identified factors or their association with quitting were observed. This suggests that no observable bias was present in our analyses.
Given the importance of smoking cessation after MI, multiple organizations have established performance measures that include documentation that patients with MI have been counseled to stop smoking.8,11 Yet the effectiveness of individual counseling to quit smoking, compared with other strategies (eg, smoking cessation programs), has not been well described. In this multicenter, prospective MI registry, we found that 1 in 3 patients were self-reported smokers at the time of admission and that more than half continued to smoke 6 months later. While several patient characteristics, including cocaine use and depressive symptoms, were associated with a lower likelihood of quitting smoking, several interventions during the hospital stay were associated with successful quitting. Both a referral to cardiac rehabilitation and the presence of an inpatient smoking cessation program were associated with successful smoking cessation among smokers. These findings extend current understandings of smoking habits after an MI and have important implications for current quality assessment efforts.
Consistent with previous smoking cessation trials in patients with MI, we found that the self-reported smoking cessation rate at 6 months post-MI remains low, thus making it an important target for quality improvement.24,25 Second, consistent with some,26 but not other studies,27 sex, age, marital status, education, and social support were not significant predictors of smoking cessation post-MI. Third, our study demonstrates that comorbidities and severity of MI were not independently associated with smoking cessation after MI. Our study findings are discordant with previous small studies that suggest a greater likelihood of quitting with increased severity of CHD.28,29 These studies enrolled patients undergoing cardiac procedures from single hospitals, while we enrolled MI patients from a national registry. In addition, most of these previous studies only adjusted for cardiac procedure, age, and packs per day smoked, while we adjusted for a comprehensive set of clinical, psychosocial, and quality of care variables. Thus, it should not be assumed that the presence of comorbidities alone is a sufficient motivation to quit smoking.
Importantly, we found that smoking cessation counseling, as documented in the patients' medical record, was not associated with smoking cessation, while the presence of a formal inpatient smoking cessation program was associated with successful quitting. These findings corroborate previous studies suggesting the limited effectiveness of smoking cessation counseling to quit smoking at the time of an MI.14,16 We believe that there are several potential limitations in the current patient-level documentation of smoking cessation counseling. It is possible that patients did not effectively note or understand the advice when smoking counseling was documented. On the contrary, it is also possible that effective advice was given without documentation. In contrast to these apparent counterintuitive findings, there were characteristics of care, namely a hospital-based program and referral to cardiac rehabilitation, that were associated with smoking cessation. We hypothesize that the presence of a formal inpatient hospital smoking cessation program was able to ensure more consistent and effective delivery of smoking cessation advice than individual provider advice. Given these findings, it appears that the current performance measure of documentation of smoking cessation counseling to quit smoking is not a good surrogate for actual quitting, while the presence of an inpatient hospital smoking cessation program that targets smokers with an MI may be a better marker of the desired goals of this performance measure. Rather than be a performance measure in the usual sense of the term, the presence of an inpatient smoking cessation program could be a valuable structural measure of high-quality care.30,31
Depression was highly prevalent in our study population, and its presence was inversely associated with smoking cessation. Thus, screening and treatment of depression should likely be incorporated into smoking cessation programs to potentially improve the rate of smoking cessation after MI.
In our study, hospitals with smoking cessation programs admitted patients with higher socioeconomic status than did hospitals without these programs. However, in multivariable analysis, the socioeconomic status of the patients did not predict whether they quit smoking. Although it is reassuring that patients quit smoking regardless of their socioeconomic status, efforts should be made to eliminate socioeconomic inequities in the availability of smoking cessation programs because such programs are found to be very cost-effective compared with other treatment methods in patients with MI.13
Consistent with previous studies,32 our study demonstrates that referral to cardiac rehabilitation is associated with a greater probability of quitting. While referral to rehabilitation does not guarantee subsequent participation, it is a class I indication for patients with MI and is now an American Heart Association/American College of Cardiology performance measure of quality.9 Thus, in conjunction with our proposal that smoking cessation programs be considered a structural measure for quality, both of these efforts could achieve the desired goal of the current, ineffective performance measure of documented smoking cessation counseling.
An important limitation of this study is the limited insights available about the types of inpatient smoking cessation programs available. Important potentially mediating benefits of smoking cessation programs may include the type or intensity of smoking cessation interventions offered, the training of counselors, the use of pharmacologic aids in smoking cessation,14,16 education materials, and follow-up plans. Nevertheless, in our study the presence of any formal, systematic program to encourage inpatients to stop smoking was found to be associated with the desired outcome of getting smokers to quit smoking. Further research will be needed to define the important components of an inpatient hospital smoking cessation program so that these can be incorporated into the design of a formal structural measure of MI treatment quality.
An additional potential limitation of this study includes the loss to follow-up. However, there was no difference in our findings when adjusting for patient characteristics associated with loss to follow-up or when we used propensity methods to more heavily weight the outcomes of patients most like those who did not participate in the follow-up interviews.
Third, smoking status was self-reported using interviewer-administered questionnaires, and we did not use biochemical assessments to validate patients' reported smoking behavior. Although biochemical validation may be more important than self-report of smoking cessation in intervention studies and specific clinical populations where there has been an intense pressure to quit smoking, biochemical validation is expensive, obtrusive, requires more contact with respondents, and may result in increased refusals.33 Moreover, self-reported smoking behavior is shown to be a valid and practical method with 88% sensitivity and 89% specificity for smoking in the general population.33 In addition, using interviewer-administered questionnaires, such as those in our study, yields higher estimates of sensitivity and specificity than self-administered questionnaires.33
Finally, although in multivariable analysis after controlling for multiple patient and clinical characteristics, referral to cardiac rehabilitation was significantly associated with smoking cessation, unmeasured confounding is always a potential limitation in an observational study.
In conclusion, despite significant benefits in post-MI outcomes, smoking cessation rates remain low after MI. Individual smoking cessation counseling during the MI hospitalization, as documented in the chart, is not associated with smoking cessation after MI, although other modifiable processes of care, such as referral to cardiac rehabilitation, are. Most importantly, availability of hospital-based smoking cessation programs in the admitting facility is associated with increased smoking cessation rates and should be potentially considered as a structural measure of health care quality. Such programs should likely incorporate screening for and treating depressive disorders because depression is highly prevalent in this population and is strongly associated with persistent smoking.
Correspondence: Susmita Parashar, MD, MPH, MS, Department of Medicine, Division of Cardiology, Emory University School of Medicine, EPICORE, 1256 Briarcliff Rd, Building A, Ste 1 N, Atlanta, GA 30307 (firstname.lastname@example.org).
Accepted for Publication: January 21, 2008.
Group Information: The following is a complete list of the PREMIER Registry Investigators: John A. Spertus, MD, MPH; Carole Decker, RN, PhD; Phillip Jones, MS; Kimberly J. Reid, MS; Gary Collins, MD; Richard Bach, MD; David Cohen, MD, MSc; Frederick Masoudi, MD, MSPH; Edward Havranek, MD; John Rumsfeld, MD, PhD; Eric Peterson, MD, MPH; Susmita Parashar, MD, MPH, MS; Viola Vaccarino, MD, PhD; William S. Weintraub, MD; Sanjaya Khanal, MD; Jane Jie Cao, MD, MPH; David Magid, MD, MPH; Wallace Radtke, MD; Mohamed Rahman, MD; John E. Brush Jr, MD; Paul Heidenreich, MD; Timothy Dewhurst, MD; Annette Quick, MD; John Canto, MD; Vijay Misra, MD; John Messenger, MD; Harlan Krumholz, MD, SM.
Author Contributions: Ms Reid and Dr Spertus had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Dawood, Spertus, and Parashar. Acquisition of data: Dawood, Reid, and Hamid. Analysis and interpretation of data: Vaccarino, Reid, and Spertus. Drafting of the manuscript: Dawood, Reid, Spertus, Hamid, and Parashar. Critical revision of the manuscript for important intellectual content: Dawood, Vaccarino, Spertus, and Parashar. Statistical analysis: Vaccarino and Reid. Obtained funding: Vaccarino, Spertus, and Parashar. Administrative, technical, and material support: Dawood, Vaccarino, Spertus, Hamid, and Parashar. Study supervision: Dawood, Vaccarino, Spertus, and Parashar.
Financial Disclosure: Dr Spertus has received consulting fees and honoraria from CV Therapeutics and United Healthcare as well as research grants from CV Therapeutics.
Funding/Support: This study was supported by a grant from CV Therapeutics, American Heart Association Scientist Development Award 0630084N (Dr Parashar), National Institutes of Health (NIH) National Center for Research Resources grant K12RR17643 and 1K23RR023171 (Dr Parashar), NIH grants K24HL077506, R01-HL68630, and R01 AG026255 from the NIH (Dr Vaccarino), and funding from the Emory University General Clinical Research Center.
Create a personal account or sign in to: