Providing Insights that Contribute to Better Health Policy
Insurance Coverage & Costs Access to Care Quality & Care Delivery Health Care Markets Issue Briefs Data Bulletins Research Briefs Policy Analyses Community Reports Journal Articles Other Publications Surveys Site Visits Design and Methods Data Files
Counseling Patients to Quit Smoking: Missed Opportunities and the Need to Better Target Physician Counseling
Research Report No. 6
From the Kansas Health Institute, Topeka, Kan. (Dr. St. Peter); the Center for Studying Health System Change, Washington, D.C. (Mr. Swanljung and Dr. Cunningham); and the National Academy of Sciences, Washington, D.C. (Dr. McGinnis). At the time the work on this manuscript was conducted, Dr. St. Peter was at the Center for Studying Health System Change.
Context: Tobacco use is the leading cause of death in the United States. Physicians can play an important role in reducing the number of persons in this country who smoke because counseling by physicians has been shown to be effective in reducing tobacco use.
Objective: To determine the proportion of adult smokers who receive smoking cessation counseling from a physician, and to describe the characteristics associated with being counseled to quit smoking and attempting to quit smoking.
Design: Telephone interviews conducted between July 1996 and July 1997 as part of the Community Tracking Study Household Survey.
Setting: Random sample of the civilian, non-institutionalized population of the continental United States.
Participants: 23,416 persons ages 18 years or older who smoked at least 100 cigarettes during their lifetime.
Main Outcome Measures: Receipt of smoking cessation counseling by a physician and attempting to quit smoking during the last year.
Results: Twenty-seven million smokers, or more than half (51 percent) of all adult smokers in the United States, attempted to quit smoking during the last year. Fewer than half (48 percent) of the smokers who visited a physician were counseled by their physician to quit smoking. Smokers who were heavier smokers, older, in poorer health, and white were more likely to receive smoking cessation counseling (p < 0.01 for all comparisons). Having insurance and a regular source of care were also associated with a greater likelihood of being counseled (p < 0.001 for both comparisons). However, smokers who were lighter smokers, younger and non-white were more likely to try to quit smoking (p < 0.04 for all comparisons). Multivariable analysis confirmed these characteristics as independent predictors of receipt of smoking cessation counseling and attempting to quit smoking.
Conclusions: The majority of smokers do not receive counseling to quit smoking from their physicians. Furthermore, physicians are more likely to direct their counseling efforts to those smokers who are less likely to attempt to quit smoking (heavier, older smokers). This indicates that physicians are missing opportunities to counsel their patients who are younger, lighter smokers, who are more likely to attempt to quit smoking, and for whom the prevention potential may be greatest. These findings suggest that unless many physicians rethink their strategies for addressing smoking and its health consequences, much of the potential for clinical care to reduce the toll from tobacco use will go unmet.
Tobacco use is the leading cause of death in the United States, resulting in more than 400,000 deaths per year.1 Many of these deaths are avoidable since smoking cessation significantly improves the health and life span of those who quit.2 Physicians can play an important role in reducing the number of persons in this country who smoke, previously estimated to be near 50 million adults3 and 4 million children.4
Several studies have shown that counseling by physicians is effective in improving quit rates among smokers.5-10 Counseling by physicians has also been shown to be a highly cost-effective preventive intervention.11-13 Based on this evidence, smoking cessation counseling is recommended by several major physician and health organizations to be included routinely in the medical care provided to all smokers.14-18
National health objectives for the Year 2000 have been established to increase the proportion of smokers who receive effective tobacco cessation counseling from their physicians.19 Purchasers of health care also have contributed recently to the increased focus on counseling of smokers by including specific performance indicators in standardized performance measurement sets.20-22
Nevertheless, studies have shown that only between one-third and one-half of smokers receive smoking cessation advice from a physician in a given year.23,24 In response, broad initiatives have been implemented to better prepare physicians to provide effective smoking cessation counseling and to increase the proportion of smokers who are counseled to quit smoking.9,16,17,25,26
This study provides an updated estimate of the proportion and characteristics of smokers who receive counseling by a physician to quit smoking. Unlike previous studies, this analysis also identifies the characteristics of smokers who are more likely to try to quit smoking. Identifying these characteristics may improve the efficiency with which physicians reach smokers for whom the prevention potential is greatest. Reducing the number of missed opportunities to counsel those smokers who are more likely to try to quit smoking may be a particularly important component of strategies to increase the counseling of smokers by physicians.
The data for this study are from the Community Tracking Study Household Survey. The Community Tracking Study (CTS) is a major initiative of The Robert Wood Johnson Foundation to track changes in the health care system over time and to gain a better understanding of how health system changes are affecting people.27
The first stage of sample selection for the Household Survey involved the random selection of 60 sites within the continental U.S.28,29 Sites were defined on the basis of counties or groups of counties using conventionally accepted definitions of statistical and economic areas. Specifically, Metropolitan Statistical Areas (MSAs) as defined by the Office of Management and Budget were used to define metropolitan areas. Because some MSAs are too large and complex to adequately represent a single health care market (e.g. Los Angeles, Chicago, New York), the 18 larger Consolidated Metropolitan Statistical Areas (CMSAs) were divided up into smaller primary MSAs (PMSAs) for the purposes of sampling sites. The nonmetropolitan portion of Bureau of Economic Analysis Economic Areas (BEAEAs) were used to define nonmetropolitan sites.
Based on this definition of sites, 60 sites were randomly selected. Sites were selected with probability in proportion to population in order to ensure representation of the continental U.S. population (i.e., the communities where people tend to live). Sites were stratified geographically and according to medium and large metropolitan sites (200,000 persons or more), small metropolitan sites (less than 200,000 persons), and nonmetropolitan sites in order to ensure representation of these areas.
For the CTS Household Survey, households were randomly selected within each of the 60 CTS study sites. The primary sample selection method was random digit dialing (RDD). However, a sample based only on RDD methods may result in some coverage bias because families and persons who do not have telephones (about 5 percent nationally) may differ on key population characteristics, such as socioeconomic status, race/ethnicity, and insurance status. Therefore, the Household Survey also included a field sample to provide coverage of families and persons who did not have telephones or who had substantial interruptions in telephone service (2 weeks or more) during the survey year. Within these sites, the general strategy was to identify and sample Census Block Groups (excluding Census Block Groups with relatively high rates of telephone service), select housing units within these areas and screen the sample for eligible households (i.e. no working telephones), and select eligible households for interviews.
While the combined RDD and field sample from the 60 sites is nationally representative of the continental U.S. (because the 60 sites were selected randomly), the clustering of the sample into 60 sites results in considerably smaller effective sample sizes (and therefore less precise estimates) than would be the case with simple random sampling or a sample based on a much larger number of sites. To increase the precision of national estimates, a supplemental national sample of households in the continental U.S. (using RDD sampling) was added to the sample for the 60 sites.
Within sampled households, family insurance units (FIUs) were identified during the initial part of the interview. The composition of FIUs (which is hereafter referred to as families) closely approximates the concept of nuclear families and were defined to include all persons eligible for coverage under a typical private health insurance policy. These include the respondent as well as their spouse and children. Other relatives (including adult children who were not full-time students, grandparents, aunts and uncles) or nonrelatives in the household were considered to be separate families. Separate interviews were conducted with all families in the household. All interviews were conducted by telephone, including the field sample who were provided cellular telephones by field staff in order to complete the interview. Interviews were conducted in Spanish for respondents who were not fluent in English or who preferred to conduct the interview in Spanish.
During the interview, information was obtained on all adults in the family as well as one randomly selected child in the family. Information about each individuals health insurance coverage, health care use, employment status, and demographic characteristics was provided by the family informant (the person most knowledgeable about the familys health care). Information on access to care, satisfaction with care, health status, and tobacco use was self-reported by all adults.
The final sample includes 32,732 families and 60,446 individuals. This includes 635 families and 950 persons in the field sample (the non-RDD portion of the sample), and 3,276 families and 6,075 persons in the national supplement sample. All adults who smoked at least 100 cigarettes in their lifetimes (n = 23,416) and the subset of those who smoked during the 12 months prior to the survey (n = 12,849) are included in the analyses presented here.
The response rate was 64 percent for families for the RDD sample and 82 percent for the field sample, which resulted in an overall combined response rate of 65 percent for families. For the RDD sample, this response rate takes into account both families who were contacted and refused to participate, as well as households with whom contact was not made (about 5 percent of released telephone numbers) but were assumed to be eligible residences.
While there was no information about survey nonrespondents with which to assess possible nonresponse bias, our sample (weighted to correct for the complex sample design) compares favorably with published national estimates from the Current Population Survey (CPS) on key demographic characteristics, including age, gender, race/ethnicity, education, and family income.
All estimates presented in this paper were weighted to be representative of the civilian, noninstitutionalized population of the continental U.S. Weights were constructed to allow nationally representative estimates to be produced from the combined 60 site and national supplement samples, and to restore proportionality to the sample arising from the complex survey design. Weights were also poststratified by age, gender, education, and race/ethnicity in order to reduce the potential for bias resulting from survey nonresponse.
Standard errors used in tests of statistical significance take into account the complex survey design, including the clustering of the 60 site sample, the national supplement, the mixed sample frames, and selection of multiple families within a household. All estimates and standard errors were produced using SUDAAN,30 a statistical software package that allows for estimation of standard errors from a complex sample design. All point estimates reported in this paper have a relative standard error of less than 30%.
Questions on Tobacco Use
Smoking-related questions used in this survey were derived from the Behavioral Risk Factor Surveillance System.31 Persons who smoked at least 100 cigarettes in their lifetime were asked whether they now smoke every day, some days, or not at all. Persons who smoked every day were asked on average how many cigarettes they now smoke per day, and former smokers were asked how long it had been since they quit.
For this study, we examine the likelihood that smokers tried to quit smoking during the last year, and the likelihood that they were advised by physicians to quit smoking in the last year. Persons who smoked during the year prior to the survey were asked: "During the past 12 months, have you stopped smoking for one day or longer, because you were trying to quit smoking?" Persons who smoked during the year prior to the survey and who had a physician visit during that time were asked: "During the past 12 months, did any medical doctor advise you to stop smoking?"
We examine a wide range of factors that may be associated with the likelihood that individuals tried to quit smoking and the likelihood that they were advised by a physician to quit smoking. These factors include intensity of smoking, sociodemographic characteristics, health status, health system factors, and level of health care use.
Whether or not individuals tried to quit and whether they were advised by a physician to quit may be strongly influenced by how often and how much they smoke. In this study, intensity of smoking includes whether or not persons smoke every day and, if so, how many cigarettes they smoke per day. Individuals from different age, gender, and racial/ethnic groups may have different propensities for trying to quit smoking, and physicians may also be more or less likely to counsel individuals depending on these characteristics.
One would also expect the likelihood of trying to quit and being counseled to quit to be strongly influenced by an individuals health status. In the multivariate analysis, health status for adults was measured using the SF-12, a 12-item scale derived from the 36-item health status scale (SF-36) used in the Medical Outcomes Study.32 Previous studies have shown that the SF-12 scale performs well in tests of validity and reliability and is highly predictive of health care use.33 The SF-12 includes separate components for physical health and mental health, and both are used in this study as separate variables. Both the physical and mental health SF-12 component scales are transformed to have a mean of 50 and a standard deviation of 10 in the general U.S. population. As a result, all scores above and below 50, are above and below the average, respectively, in the general population, and each one point difference in scores is one-tenth of a standard deviation.34 For simplicity of interpretation, a more general measure of health status was used in the bivariate analysis. This standard, single-item measure ascertained whether an individual reported his or her health to be excellent, very good, good, fair, or poor.
The likelihood of being counseled by a physician is also likely to be influenced by an individuals health insurance coverage and their interactions with the health care system. Previous research has shown that physicians provide fewer services to uninsured persons than they do for insured persons,35-37 and this may also be the case regarding advice to quit smoking. Individuals who have a regular source of primary care-and particularly a regular physician-may be more likely to be counseled because the physician has more direct interest and knowledge of the patients health. Also, the extent to which individuals use the health care system will influence whether or not an individual is advised by a physician to quit smoking simply because more frequent contact with the health care system results in greater opportunities for this counseling to occur.
Differences in proportions were tested for statistical significance using the z test. Logistic regression analysis was used to analyze the independent effects of various factors on the likelihood that individuals tried to quit smoking and on the likelihood that they were advised by a physician to quit smoking. The logistic regression analyses included only those persons who had at least one physician visit in the previous year.
Prevalence and Intensity of Smoking
Nearly half (49 percent) of Americans ages 18 or older are either current (25 percent) or former (24 percent) smokers (Table 1). Among the 49.0 million adults who currently smoke, 38 million smoke every day; the remaining 11 million smoke only some days. Nearly 5 million Americans quit smoking just within the last year; another 42 million quit smoking more than a year ago (Table 1).
Attempting to Quit and Being Counseled to Quit
Twenty-seven million, or more than half (51 percent) of all adult smokers attempted to quit smoking during the previous year (Table 2). Lighter smokers were more likely than heavier smokers to try to quit. Sixty-six percent of some day smokers tried to quit, compared to 41 percent of every day smokers (p < 0.001). Among those smokers smoking two packs a day or more, only 25 percent attempted to quit smoking.
Thirty-seven million of the 54 million adults (69 percent) who smoked in the past year reported making at least one physician visit during the previous year. Even among those adult smokers who visited a physician during the previous year, however, fewer than half (48 percent) were counseled to quit (Table 2), leaving more than 19 million adult smokers (52 percent) who visited a physician but were not counseled by the physician to quit smoking. Assuming that those smokers who did not make a physician visit also were not counseled by a physician to quit smoking, then only one-third (33 percent) of all adult smokers were counseled by a physician to quit smoking during the previous year (Table 2).
Despite being less likely to try to quit smoking, heavier smokers were one and a half times more likely than lighter smokers to be counseled by a physician to quit. More than half (52 percent) of every day smokers, and only one-third (34 percent) of some day smokers, were counseled to quit (p < 0.001) (Table 2). Among everyday smokers, those smoking two packs a day or more (61 percent) were more likely than those smoking less than a pack a day (51 percent) to be counseled (p < 0.002).
Young adult smokers (59 percent) were more likely than middle aged (46 percent) or older smokers (52 percent) to try to quit smoking (p < 0.001 for each comparison) (Table 3). Smokers in poor health status (57 percent) were more likely than those in good (50 percent) or very good (52 percent) health status, and African American (57 percent) and Hispanic (60 percent) smokers were more likely than white smokers (49 percent), to try to quit smoking (p < 0.04 for each comparison).
In many situations, physicians are more likely to counsel those patients who are actually less likely to try to quit smoking. For example, despite being less likely to try to quit smoking, older (60 percent) and white (49 percent) smokers were more likely than younger (38 percent) and African American (43 percent) smokers to be counseled by a physician to quit smoking (p < 0.01 for each comparison) (Table 3). Smokers in poorer health status, however, were more likely than those in better health status both to attempt to quit smoking and to be counseled by a physician to quit smoking (Table 3).
Characteristics of Health Care
Smokers with insurance (52 percent) were no more likely than those without insurance (51 percent) to try to quit smoking (p = 0.646), but insured smokers (49 percent) were more likely than those without insurance (39 percent) to be counseled to quit smoking (p < 0.001) (Table 4).
Smokers with a usual source of care (53 percent) and those with both a usual source of care and a particular provider (52 percent) were slightly more likely than those with neither (48 percent) to try to quit smoking (p < 0.005 for each comparison). Among smokers with a physician visit, those with a usual source of care and a particular provider (50 percent) were more likely than those with neither (35 percent) to be counseled by a physician to quit smoking (p < 0.001) (Table 4).
Smokers with more visits to a physician were more likely both to try to quit smoking and to be advised to quit during the past year (Table 4).
Determinants of Attempting to Quit Smoking
Multivariate logistic regression was conducted to identify the factors independently associated with trying to quit smoking during the previous year among those smokers with at least one physician visit. Variables included in the model were intensity of smoking, gender, age, health status (continuous variables using SF-12 scores for physical and mental health), race, education, family income, insurance status, having a usual source of care, number of physician visits, and being counseled by a physician to quit smoking (Table 5). The analysis shows that smokers whose race is "other" (OR = 1.31, 95 percent CI 1.05-1.64) and those counseled by a physician to quit smoking (OR = 1.37, 95 percent CI 1.22-1.53) were more likely to attempt to quit smoking during the previous year. Heavier smokers (OR range from 0.11 to 0.22), older smokers (OR range from 0.66 to 0.73), those with better physical health status (for each point on the scale OR = 0.99, 95 percent CI 0.99-<1.00) or better mental health status (for each point on the scale OR = 0.99, 95 percent CI 0.99-<1.00) were less likely to try to quit. Gender, years of education and family income were not independently associated with trying to quit smoking.
Determinants of Being Counseled to Quit Smoking
Multivariate logistic regression also was conducted to identify the factors associated with being counseled by a physician to quit smoking among those smokers with a physician visit. The model was similar to the one used for trying to quit and included measures of intensity of smoking, gender, age, health status (continuous variables using SF-12 scores for physical and mental health), race, education, family income, insurance status, having a usual source of care and the number of physician visits (Table 5). The analysis shows that heavier smokers (OR range from 1.79 to 2.40), older smokers (OR range from 1.18 to 1.91), those with insurance (OR = 1.33, 95 percent CI 1.11-1.60), those with a usual source of care but no particular provider (OR = 1.37, 95 percent CI 1.05-1.78) and those with both a usual source of care and a particular provider (OR = 1.51, 1.24-1.85) were more likely to be counseled to quit smoking. Smokers who are male (OR = 0.84, 95 percent CI 0.76-0.93), those with better physical health status (for each point on the scale OR = 0.98, 95 percent CI 0.98-0.99), better mental health status (for each point on the scale OR = 0.99, 95 percent CI 0.98-0.99), and those who are African American (OR = 0.82, 95 percent CI 0.69-0.97) were less likely to be counseled by a physician to quit smoking. Years of education and family income were not independently associated with being counseled to quit smoking.
Fewer than half of adult smokers who visited a physician in the previous year were counseled by their physician to quit smoking. In one year alone, physicians missed the opportunity to counsel more than 19 million of their patients who smoke to kick the habit and significantly improve their prospects for a longer, healthier life. This is particularly distressing since fully one-half of all adult smokers demonstrated a desire to quit smoking by attempting to quit at least once during the year.
Furthermore, physicians appear to be targeting their counseling efforts on those smokers who are less likely to try to quit smoking. While older, heavier smokers were more likely to be counseled to quit smoking by physicians, younger, lighter smokers were more likely to try to quit. Because younger patients make fewer visits to physicians than older patients, physicians should take advantage of the opportunity to counsel younger patients who smoke whenever they encounter them, and not reserve counseling for preventive visits alone. Given the addictive nature of tobacco use,38,39 counseling patients who smoke fewer cigarettes and have been smoking for shorter periods of time may be a more effective approach.40 Recent debate over the pending tobacco industry settlement has emphasized the need to use some of these funds to address smoking and other forms of tobacco use through early, preventive interventions in children and young adults.41-44
Smokers in poor health, regardless of age and heaviness of smoking, were more likely both to be counseled to quit and to attempt to quit smoking. This most likely reflects the patients and physicians shared assessment of the health risks of continued smoking for these individuals. Maintained emphasis by physicians on this high-risk group of smokers clearly is indicated.
This study also found that men, non-whites and those without health insurance are less likely to receive counseling by a physician to quit smoking. The findings, again, represent important missed opportunities for physicians to counsel patients to quit smoking as non-white smokers are actually more likely than white smokers to attempt to quit; and men and patients without insurance are no less likely than women or those with insurance to attempt to quit smoking.
Having a usual source of care and a particular provider also appear to be important in determining which patients are counseled to quit smoking and which try to quit. Perhaps the rapport that exists between a physician and a patient with a relationship established over time increases the likelihood that patients will seek, and that physicians will provide, preventive counseling.45,46 Having insurance and a usual source of care previously have been associated with an increase in the number of physicians visits.47 However, the analysis presented here controlled for the effect of the number of physician visits on the likelihood of receiving counseling and trying to quit. Hence, the effect demonstrated is independent of the effect due simply to increasing the number of physician visits.
The findings from this 1996-1997 survey indicate that the proportion of smokers counseled to quit smoking by a physician has remained relatively stable since the early 1990s when it was estimated to be between 37% and 52%.23,24 Recently, efforts have been launched to better prepare physicians to identify patients who smoke and to offer them effective tobacco cessation counseling through the use of practice guidelines17 and enhanced monitoring of counseling.20-22 The effectiveness of these efforts should be tracked for evidence of progress in this important area of public health.
While it was not the objective of this analysis to examine the effectiveness of counseling by physicians to encourage smokers to attempt to quit smoking, it was demonstrated that among adult smokers with a physician visit, being counseled to quit smoking was associated with a significantly increased likelihood of attempting to quit smoking. This finding supports evidence from other studies more appropriately designed to demonstrate the effectiveness of physician counseling efforts.5-10
It is important to note that this survey asked only about being advised by a physician to quit smoking. There was no measure of the thoroughness or effectiveness of such counseling. These data, therefore, may present an overly optimistic estimate of the proportion of smokers actually receiving thorough, effective smoking cessation counseling.
Another limitation of this study is that information was collected only about counseling by physicians. Other health care providers (e.g., nurses, health educators, dentists) also play an important role in smoking cessation counseling24,48 and a more complete picture would be provided if this information was available. Finally, as in all surveys, the responses provided are subject to the recall bias of respondents. There is, however, no reason to believe that this bias systematically influenced the responses in any particular fashion.
This study provides information not available from other recent reports on physician counseling of smokers. Unlike visit-level analysis,49 the person-level analysis presented here allows an assessment of how likely individual smokers are to receive counseling over the course of a year. It may be appropriate for a physician to forgo counseling of an established patient during a particular office visit if it is known that the issue was adequately addressed during other visits. Using person-level data also allows estimates of the prevalence of counseling among important sub-populations, for instance those in poor health. The data used in this analysis are considerably more recent than other published studies, and provide analysis of characteristics associated with both being counseled to quit and attempting to quit smoking.23,24,50-52
In summary, this study shows that fewer than half of adult smokers who visited a physician were counseled by their physician to quit smoking. Physicians are missing many opportunities to counsel their patients who smoke to quit: in one year alone, more than 19 million adult smokers visited their physician, yet were not counseled to quit. In particular, physicians should remain alert for opportunities to counsel younger, lighter smokers on smoking cessation. These patients are more likely to try to quit smoking and stand to benefit the most from avoiding life-long addiction and the long-term health consequences of tobacco use. These findings suggest that unless many physicians rethink their strategies for addressing smoking and its health consequences, much of the potential for clinical care to reduce the toll from tobacco use will go unmet.
ACKNOWLEDGMENTS: The Center for Studying Health System Change is supported in full by The Robert Wood Johnson Foundation. Dr. McGinnis is supported through his position as a Scholar-in-Residence at the National Academy of Sciences. The authors would like to thank the research staff of the Center for helpful comments on an earlier draft of this manuscript. Beny Wu of Social and Scientific Systems, Inc. provided excellent programming assistance.
1 McGinnis JM, Foege WH. Actual causes of death in the United States. JAMA 1993;270:2207-12.
Table 1—Prevalence and Intensity of Smoking in the U.S., 1996-97
Table 2—Attempting to Quit and Being Counseled to Quit Smoking: Intensity of Cigarette Smoking
Table 3—Attempting to Quit and Being Counseled to Quit Smoking: Demographic Factors
Table 4—Attempting to Quit and Being Counseled to Quit Smoking: Characteristics of Health Care
Table 5—Determinants of Attempting to Quit Smoking and Being Counseled to Quit Smoking