Cunningham P. Patient Engagement During Medical Visits and Smoking Cessation Counseling. JAMA Intern Med. 2014;174(8):1291-1298. doi:10.1001/jamainternmed.2014.2170
Increased patient engagement with health and health care is considered crucial to increasing the quality of health care and patient self-management of health.
To examine whether patients with high levels of engagement during medical encounters are more likely to receive advice and counseling about smoking compared with less engaged patients.
Design, Setting, and Participants
Cross-sectional survey using multivariate regression analysis of 8656 current and retired autoworkers and their spouses younger than 65 years who are or were employed by the 3 major US auto companies.
Main Outcomes and Measures
Clinician advice and counseling about smoking; patients who tried to quit smoking.
Among 1904 current smokers, 58.5% of those who were more highly engaged during medical encounters were counseled by clinicians about specific strategies and methods to stop smoking, compared with 45.4% of patients who were less engaged. Patient engagement and being advised by clinicians to stop smoking had independent effects on smoking cessation efforts by patients. Accounting for differences in other patient characteristics, patients with high engagement levels were more likely to try to stop smoking compared with patients with lower engagement (odds ratio, 1.62; P < .01). Patients who were both highly engaged and had received counseling from clinicians were the most likely to try to stop smoking (74.6%) while patients with low engagement who did not receive counseling were the least likely (46.0%). Nevertheless, counseling is still effective among even less engaged patients; 60.4% of smokers with low engagement who received counseling tried to quit smoking in the past year compared with 46.0% who did not receive counseling.
Conclusions and Relevance
The study results provide evidence that clinicians respond differently to patients who are highly engaged during medical encounters than they do to less engaged patients in terms of smoking cessation advice. Clinicians should not assume that low patient engagement and greater passivity during medical encounters is evidence of unwillingness to quit. The results show that smoking cessation counseling is associated with a higher likelihood of quit attempts even for patients who are less engaged during medical encounters.
The importance of patients being engaged in their own health and health care is increasingly recognized among both clinicians and policymakers. Patient engagement is defined as “actions individuals must take to obtain the greatest benefit from the health care services available to them.”1(p2) Evidence is accumulating about the benefits of increased patient engagement, including higher quality of care, greater adherence to treatment regimens, better care experiences on the part of patients, and lower costs.2- 5 More engaged patients also tend to use healthy behaviors (eg, diet, exercise), preventive health care use, and self-management of chronic diseases to a greater extent than less engaged patients.6
Although patient engagement is strongly determined by many aspects of a patient’s background,7 there is also evidence that patient engagement can be changed, for example, through interventions in the community, at the workplace, and by clinicians.1,8 Another strategy is for clinicians to assess initial patient engagement at intake and use the information to match patients with appropriate medical staff within a practice and to guide treatment decisions.1 The latter is based on 2 key assumptions: (1) that clinicians respond differently to patients based on their level of engagement and (2) the effects of clinical interventions to address specific health behaviors will differ depending on the level of engagement that patients bring into a medical encounter. One study found that physicians were more likely to refer highly engaged patients to a Web-based patient portal that had information about their care compared with less engaged patients.9 Otherwise, there is little empirical evidence on how treatment and advice provided by clinicians differs depending on patient engagement.
The objective of this study is to examine how 1 specific patient health behavior—smoking cessation—and physicians’ efforts to encourage and advise patients on smoking cessation strategies differ depending on how engaged patients are during medical encounters. The US Preventive Services Task Force recommends that clinicians ask all adults about tobacco use and provide tobacco cessation interventions for patients who use tobacco products.10 Nevertheless, less than half of adult smokers nationally report that physicians provide any kind of advice or counseling about smoking cessation, and about two-thirds of physicians cite lack of patient motivation to quit smoking as significant barriers to interventions.11,12 Therefore, clinicians’ efforts to counsel patients about smoking cessation—and potentially the effectiveness of these interventions—are likely to vary depending on patient engagement levels.
The data for this study are based on the 2012 Autoworker Health Care Survey (AHCS),13 a survey of active and retired hourly wage workers from Chrysler, Ford, and General Motors. The survey was sponsored by the National Institute for Health Care Reform (NIHCR), a nonprofit, nonpartisan organization established by the International Union, United Automobile, Aerospace, and Agricultural Implement Workers of America (UAW); Chrysler Group LLC; Ford Motor Co; and General Motors. The study received approval from the New England Institutional Review Board. The total survey sample includes 8656 hourly wage workers, retirees younger than 65 years (ie, not eligible for Medicare), and their spouses (all ages). Retired autoworkers 65 years or older and their spouses were excluded from the sample. All survey respondents received $40 for completing the survey.
The sample was randomly selected, with some oversampling of active workers so that the proportion of active and retired workers in the sample was about evenly split. Survey weights were designed to produce representative estimates of active and retired autoworkers and their spouses and to account for survey nonresponse based on information from health plan eligibility files, such as age, sex, and whether the spouse was enrolled in a company-sponsored plan. Standard errors reflect the complex sample design, primarily owing to the oversampling of active workers and the clustering of the sample within families (ie, the worker and his or her spouse). The survey was administered by mail, and a final response rate of 64% was obtained.
Sample persons have roughly similar health benefits as a result of contract negotiations between the UAW, the union representing autoworkers, and each auto company, and, therefore, differences in health benefits among the sample can be largely ruled out as affecting the results. Similarly, all sample persons have a similar choice of physicians because most are enrolled in a preferred provider organization type of health plan, which includes a very broad choice of physicians in the plan networks.
Four questions on patient engagement during medical encounters were included in the survey. The questions did not ask about a specific medical encounter or a specific type of clinician (eg, primary care physicians, specialists, or nonphysicians could be considered). These questions were derived from a survey sponsored by the National Business Group on Health (NBGH) in 2007.13 The following questions were asked of respondents about things they might have done before or during a medical visit (Table 1):
Have you ever:
Brought information you found on an Internet website to a medical visit and talked about it with your physician?
Taken notes during a medical visit to help you remember what the physician or nurse said?
Brought along a friend or family member to your medical visit as your advocate or to give you support?
Brought along a list of questions to ask during a medical visit?
Responses to the questions include never (coded as 0); once (1); and more than once (2). An index of patient engagement was constructed by summing the responses to the 4 measures. Scores range from a high of 8 (answered “more than once” on all 4 questions) to a low of 0 (answered “never” to all 4 questions).
The 13-item Patient Activation Measure (PAM) is more commonly used in research on patient engagement and focuses on how much control people believe they have over their health and how their health is affected by health care.14 Conceptually, the PAM provides an assessment of the potential or capacity for patients to be engaged in their health care, while the NBGH questions used in this analysis relate more specifically to how patients are actually engaged during their medical encounters with physicians. In practice, both measures are likely to be correlated because prior research has shown that patients with high scores on the PAM were much more likely to bring a list of questions to the physician visit compared with the patients with the lowest scores on the PAM.15
Questions on smoking behavior and cessation are based on similar questions asked in the Behavioral Risk Factor Surveillance System conducted by the Centers for Disease Control and Prevention.16 These include questions on (1) whether the individual has smoked a total of at least 100 cigarettes; (2) whether he or she currently smokes and frequency of smoking; (3) whether a clinician advised the individual to stop smoking in the past year; (4) whether a clinician discussed specific methods or strategies to stop smoking; and (5) whether the patient actually stopped smoking for 1 day or longer in the past year because he or she was trying to quit.
Logistic regression analysis is used to examine the association between patient engagement, clinicians’ counseling to stop smoking, and patient smoking cessation efforts. The initial analysis of data included all autoworkers and spouses and examined the association between patient engagement and whether the person was a current smoker. The next series of regressions restricted the sample to autoworkers and spouses who were current smokers (n = 1904) in order to examine the effects of patient engagement on smoking cessation activities. These regressions included the association between patient engagement and whether a clinician advised them to stop smoking in the past year, and whether a clinician had discussed specific methods or strategies to stop smoking. A final regression was estimated that examined the effect of patient engagement and clinician counseling on the likelihood of a patient trying to stop smoking in the past year.
Also, separate equivalent logistic regressions for the likelihood of trying to stop smoking were estimated for low, moderate, and highly engaged patients who were current smokers. Statistically significant differences in the coefficients associated with a clinician’s counseling would indicate that the effects of counseling on smoking cessation efforts differ depending on patient engagement levels.
All logistic regression analyses control for other variables that may also affect health behaviors and smoking cessation efforts by clinicians and patients. These include worker status (retirees, recent hires or longer-term employees [hired after or before November 2007]), age, sex, race/ethnicity, educational attainment (based on years of education and attainment of degree), family income (before taxes), and measures of health status. The latter include chronic conditions that survey respondents reported had been diagnosed by a physician, and measures of perceived physical and mental health. Only coefficients with P < .05 were considered to be statistically significant.
Regression-adjusted means are computed for key variables (patient engagement, clinician counseling efforts) based on the results of the model. The adjusted means are computed by taking the average of the individual predictions assuming different values for patient engagement and/or a clinician’s counseling.
Given the cross-sectional nature of the survey data, the causal direction between patient engagement and clinician counseling about smoking cannot be conclusively established. It is possible that highly engaged patients are more likely to select clinicians with whom they have more positive communication and interpersonal exchanges rather than clinicians being more responsive to highly engaged patients.
The analysis tests for this by examining whether there are differences by level of engagement in how long patients have been with their personal physicians. Evidence of greater switching of personal physicians (who have likely performed much of the smoking cessation counseling) by highly engaged patients may suggest greater selection behavior on the part of patients.
In addition, the association between patient engagement and the quality of communication with a patient’s personal physician is examined, including how well the physician explains things, how carefully the physician listens, and whether the physician spends enough time with the patient. If highly engaged patients switch physicians more frequently, then more favorable assessments of their current physicians might suggest that these patients are selecting physicians with whom they are better able to communicate and therefore are more likely to receive counseling about smoking.
Among the 4 patient engagement questions, 44.1% of 8656 individuals responded that they did not bring a list of questions to ask their physician or nurse, 66.2% did not take notes during a visit, 60.7% did not bring a friend or relative to ask questions, and 74.7% did not bring information they obtained from an Internet site (Table 1). Overall, only 9.7% said that they engaged in all 4 activities (once or more than once), while 28.9% did not engage in any of the activities.
When the patient engagement questions are summed to create an index of patient engagement, about one-fifth of patients (21.1%) are classified as being highly engaged (scores of 5-8), while 40.5% are classified as having low engagement (scores of 0 or 1) (Table 2). Patient engagement is most strongly correlated with educational attainment and health status. Persons with a college degree or higher have higher engagement levels compared with people with less education. Higher engagement is also observed among those with multiple chronic conditions and those who describe their physical and/or mental health as fair or poor. Retirees and women also tend to have higher levels of patient engagement.
About 23% of autoworkers and their spouses currently smoke (1904 individuals), which is somewhat higher than the national average of 19% for people aged 18 to 64 years (Table 3, national estimates not shown).17 Prevalence of smoking is slightly higher for people with low engagement compared with more highly engaged patients, although the differences are not statistically significant.
Among people who smoke, people with high or moderate levels of engagement were more likely to be counseled about specific methods or strategies to stop smoking (58.5% and 60.1% of smokers, respectively) compared with smokers with low engagement (45.4%). Smokers with high levels of engagement were much more likely to have attempted to quit smoking in the previous year (70.5%) compared with smokers with low levels of engagement (54.3%).
The results of the logistic regression analysis were largely consistent with the descriptive results. There were no statistically significant differences in the likelihood of being a current smoker between highly engaged patients and less engaged patients (Table 4). The likelihood of being a current smoker is most strongly associated with education, income, and perceived physical and mental health.
Among people who smoke, highly engaged patients were more likely to be advised by a clinician to quit smoking compared with people with low engagement (odds ratio [OR], 1.51; P < .05) (Table 4). Women, those with higher family income, and people with chronic conditions (especially multiple chronic conditions) were much more likely to be advised to stop smoking by a clinician. More highly engaged patients were also more likely to report that a clinician discussed specific methods or strategies for smoking cessation, although the effect is less clearly linear than for whether physicians advised them (Table 4).
Patients with higher engagement were also more likely to try to quit smoking in the past year compared with people with lower engagement (OR, 1.62; P < .01) (Table 4). In addition, patients were also more likely to try to quit smoking when a clinician discussed specific methods or strategies, although just advising a patient to stop smoking without discussing specific strategies had no statistically significant association with cessation attempts. Smoking cessation efforts were greater among people who did not smoke every day, among African Americans, and college graduates. People with 3 or more chronic conditions were also more likely to try to stop smoking, although the difference with people who had no chronic conditions was not as large as for whether people were counseled by clinicians to stop smoking.
Separate logistic regression models were estimated for patients with high, moderate, and low engagement to test whether the association between physician counseling and smoking cessation efforts differed depending on engagement levels (findings not shown). Tests revealed no statistically significant differences in the coefficients for clinician advice and counseling variables across the 3 models, suggesting that people are just as likely to try to quit smoking if they were counseled by a clinician regardless of their overall level of engagement.
Table 5 shows the regression-adjusted means computed from these models. The results show that the percentage who tried to quit smoking is highest among highly engaged patients who received counseling on cessation strategies (74.6%) and lowest among patients with low engagement who did not receive such counseling (46.0%). However, even among patients with low engagement, smoking cessation efforts were higher among those whose clinicians discussed specific methods or strategies for smoking cessation (60.4%) compared with those who did not receive any advice or counseling (46.0%).
There were no differences between highly engaged patients and less engaged patients in the length of time that they had been with their personal physician (Table 6). In addition, there were no statistically significant differences between highly engaged patients and less engaged patients in their assessment of how well their physician explained things, how well the physician listened to them, and the amount of time their personal physician spent with them. These results suggest that highly engaged patients do not change clinicians more frequently based on the quality of their interactions.
Although the study did not examine the extent to which patient engagement levels could be increased among this particular sample of autoworkers, the results suggest that increasing overall engagement levels could potentially assist smoking cessation efforts among workers. Increasing smoking cessation can also be facilitated by physicians and other clinicians regardless of the level of patient engagement.
Interventions to provide smoking cessation counseling, such as those based on the Transtheoretical Model (TTM) and the 5-A framework (Ask, Advise, Assess, Assist, and Arrange), have been developed to promote health behavior change at all stages of a patient’s readiness for such change and have proven successful for smoking cessation.10,18,19 Specific guidelines have been developed for physicians to discuss cessation strategies and methods for patients who are willing to try and quit.20 For patients who are unwilling to quit, guidelines advise physicians to attempt “motivational intervention,” such as discussing the “5 R’s” (Relevance, Risks, Rewards, Roadblocks, Repetition), although discussion of specific smoking cessation strategies is unlikely to be productive with these patients. The key distinction in these guidelines is physicians’ and other clinicians’ assessment of willingness to quit on the part of patients. Clinicians should not misinterpret lack of patient engagement during medical encounters as unwillingness to quit because the results of this study suggest that counseling of even less engaged patients is effective in getting them to attempt quitting.
Information on patient engagement can be used to develop treatment plans and to match patients with medical staff within a practice or clinic who may be especially effective with highly engaged or less engaged patients in providing counseling on smoking cessation and other health behaviors. Some innovative delivery systems, such as Fairview Health Services in Minnesota, are increasingly using instruments such as the PAM or TTM to identify a patient’s level of engagement or receptivity to clinician’s advice and counseling about health and health behaviors. The extent to which physician practices collect and use such information is unknown, although at this point it is likely to be quite limited. For many practices, limitations on the types and numbers of medical staff, inadequate electronic health records and information systems may severely limit their ability to obtain and use patient engagement information.
Several limitations with the study should be noted. First the survey questions used to ascertain patient engagement were general in nature and not designed to capture all possible ways in which an individual may be engaged with their health care. Although the questions in the NBGH were designed to identify the most salient ways that a patient could be engaged during their encounters with clinicians, it is possible that actual patient engagement is underreported in this study.
The fact that patient engagement, smoking cessation counseling, and smoking cessation effort are all self-reported by patients should also be noted, especially given the cross-sectional nature of the survey data. It could be argued that highly engaged patients are more likely to recall being counseled by clinicians as well as other details of their clinician visits compared with less engaged patients. The questions on clinician counseling have been used extensively in other major surveys, such as the BRFSS and the National Health Interview Survey, and therefore are the most credible and validated measures available.
Also, the causal sequence between patient engagement and smoking cessation cannot be determined with certainty. A desire to quit smoking may prompt greater patient engagement during their encounters with clinicians, or that they occur more or less simultaneously (ie, are collinear). Experimental research on patient engagement has found that efforts to increase patient engagement affected subsequent health behaviors.2,5
In addition, the survey did not ascertain whether recent attempts to stop smoking were successful. It is possible that the success of cessation efforts differs by level of patient engagement, which cannot be determined by the data used for this study.
Despite these limitations, the study results strongly suggest that clinicians respond differently to patients who are highly engaged during medical encounters than they do to less engaged patients in terms of advising patients to stop smoking. Nevertheless, even patients with low levels of engagement can benefit from this counseling. While lack of health plan reimbursement for smoking cessation activities has previously been a barrier for both clinicians and patients, the Affordable Care Act now requires that health plans cover such services. Counseling activities can also be increased through delivery system reforms that use financial incentives and bonuses to encourage clinicians to provide smoking cessation counseling.
Corresponding Author: Peter Cunningham, PhD, Department of Health Care Policy and Research, Virginia Commonwealth University School of Medicine, 830 E Main St, Fourth Floor, Richmond, VA 23298-0430 (firstname.lastname@example.org).
Accepted for Publication: March 19, 2014.
Published Online: June 9, 2014. doi:10.1001/jamainternmed.2014.2170.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by the National Institute for Health Care Reform (NIHCR).
Role of the Sponsor: The NIHCR provided input on the sample and questionnaire design and also facilitated access to names and contact information of autoworkers for the purposes of sample selection. The NIHCR also provided input on the research questions and objectives for this article but had no role in the analysis and interpretation of data. The NIHCR had no role in the preparation of the manuscript other than to provide comments on an earlier draft.
Additional Contributions: The following individuals reviewed an earlier version of this draft and provided comments. None of these individuals received any compensation for reviewing the manuscript: Ann O’Malley, MD, MPH, Mathematica Policy Research, Washington, DC; Paul Ginsburg, PhD, formerly President of the Center for Studying Health System Change; and Alwyn Cassil, formerly Director of Public Affairs at the Center for Studying Health System Change. In addition, Cynthia Saiontz-Martinez of Social Scientific Systems Inc, Silver Spring, Maryland, was paid to provide the programming for statistical analysis.