Primary care physicians are potentially important sources of interventions aimed at preventing youth smoking. Yet recent surveys suggest that physician smoking prevention practices are less than optimal.
To document prevention counseling practices and to identify correlates of these activities in a random sample of general practitioners in Montreal, Quebec.
A cross-sectional mail survey.
Of 440 eligible general practitioners (GPs), 337 (77%) completed the questionnaire. General practitioners were more likely to ascertain the smoking status of adolescents (70.9%) than preadolescents (35.7%). Although about half of the GPs offered advice to prevent smoking onset in young adults (48.6%) and adolescents (48.3%), fewer did so for preadolescents (34.4%); only 12.1% advised parents to discuss smoking onset with their children. Correlates of ascertaining smoking status included female sex (odds ratio [OR], 1.90; 95% confidence interval [CI], 1.07-3.41), lower proportion of walk-in patients (OR, 2.73; 95% CI, 1.31-5.80), awareness of the "stage of behavior change'' model (OR, 2.17; 95% CI, 1.18-4.04), and higher self-efficacy (OR, 4.12, 95% CI, 2.00-8.69). Correlates of provision of prevention advice included more hours spent in direct patient care (OR, 1.93; 95% CI, 1.13-3.34), favorable beliefs and attitudes (OR, 1.73; 95% CI, 1.06-2.83), and higher self-efficacy (OR, 4.32; 95% CI, 2.25-8.44).
Our results point to the need for renewed efforts to enhance preventive efforts in primary care settings. Intervention programs for GPs should emphasize overcoming unfavorable beliefs and attitudes and low self-efficacy. Future research should evaluate the effect of brief prevention counseling adapted to increasingly busy practices.
PREVENTION OF smoking initiation is becoming even more crucial because of recent evidence suggesting that children and adolescents may become nicotine dependent very early in the smoking process.1 Pediatricians and family practitioners are potentially important sources of counseling aimed at preventing youth from smoking,2- 6 and current recommendations advocate that physicians include tobacco prevention in routine medical visits with parents and children beginning with prenatal visits and continuing throughout childhood and adolescence.7- 10 The US National Cancer Institute and the American Academy of Pediatrics recommend that physicians undertake the 5 A's practice activities (ie, anticipate, ask, advise, assist, and arrange).11- 13
Despite these recommendations, recent surveys suggest that few physicians provide smoking prevention counseling, particularly prior to and in early adolescence when counseling might be most effective at preventing the onset of smoking.14- 19 Few studies to date have identified predictors of physician prevention practices that could help identify targets for interventions to improve physician prevention counseling behavior.16,19
We conducted a cross-sectional mail survey20 as part of the evaluation of an intervention program to improve smoking cessation and prevention counseling practices of general practitioners (GPs) in Montreal, Quebec.21 This article describes the prevention counseling practices of GPs in Montreal, as well as the sociodemographic, clinical, and psychosocial correlates of these practices.
The survey was conducted between April 1, 1998, and July 31, 1998, in a simple random sample of 670 physicians selected from the 1997 Quebec College of Physicians database for Montreal. Eligible physicians included those holding a GP license from the Quebec College of Physicians and working in any clinical setting in Montreal. Three mailings of a self-administered French or English questionnaire with follow-up telephone calls to all nonrespondents were undertaken to maximize participation. More details on the mailing protocol and the survey instrument are available in a previous article.20
We collected data on GPs' sociodemographic characteristics (age, sex, spoken-language, number of years since graduation from medical school, completion of a family medicine residency, and training in smoking counseling), practice profile (type and number of practice settings, number of hours per week spent in direct patient care, number of nonhospitalized patient visits per week, proportion of patients seen by appointment and without an appointment, proportion of patients aged ≤18 and ≥65 years, and provision of obstetric care), psychosocial characteristics related to smoking counseling, and tobacco prevention counseling behaviors.
Psychosocial characteristics of GPs included 1 item in each of the 4 categories to measure the following: awareness of the "stage of behavior change" model,22 knowledge of community resources to help patients quit smoking, self-efficacy ("I have the skills required to help younger patients not start smoking."), and beliefs and attitudes ("Family physicians should advise parents of their pediatric patients to quit smoking.''). Three items measured physicians' perceptions of patient-related barriers: lack of influence on patients, lack of patients' interest, and lack of compliance with physicians' advice. Five items measured physician-related barriers: complexity of smoking cessation guidelines, and the lack of time, reimbursement, training, and patient education materials.
Awareness and knowledge variables were dichotomized (yes/no). Although these 2 variables pertain mostly to cessation and not to prevention practices, they were included in these analyses because they might predict an important aspect of physicians' smoking prevention practices, namely, screening for parental smoking and counseling those who smoke to quit. Also, these 2 variables were considered indicators of the physicians' heightened awareness and level of interest regarding smoking-related issues. For the self-efficacy and beliefs and attitudes items, subjects indicated their level of agreement with the statement on a 6-point Likert-type scale that ranged from "strongly agree'' (1) to "strongly disagree'' (6). Response categories and descriptive analyses of the barrier items, as well as the methods for creation of the barrier scores and their psychometric properties, were previously reported.20
We collected data on the following 2 aspects of smoking prevention practices: (1) ascertainment of smoking status in preadolescents (aged 9-12 years) and adolescents (aged 13-19 years) (2 items) and (2) provision of advice to prevent smoking onset in preadolescents, adolescents, and young adults (aged 20-24 years) (3 items) and encouragement of parents to discuss smoking with their children (1 item). Subjects indicated the proportion of their patients to whom they provided these aspects of counseling on a 6-point Likert-type scale that ranged from "all patients'' (1) to "few/no patients'' (6). For descriptive analyses, responses were dichotomized into "more than half of patients'' and "less than half of patients.'' For multivariate analyses, we summed the 2 ascertainment items to create an "ascertainment'' score (range, 2-12; mean [SD], 6.7 [3.5]), which was categorized by tertile into "most patients" (range, 2-3), "some patients" (range, 4-7), and "few patients" (range, 8-13). Similarly, we summed the 4 advice-related items to create a "prevention advice'' score (Cronbach α = 0.86; range, 4-24; mean [SD], 16.3 [6.4]), which was categorized by tertile into "most patients" (range, 4-11), "some patients" (range, 12-20), and "few patients" (range, 21-34).
Independent correlates of the ascertainment and prevention advice scores were identified in multiple polychotomous logistic regression analyses. Potential correlates included sociodemographic characteristics, practice profile, and psychosocial characteristics. Continuous independent variables were categorized by tertile for analysis except the beliefs and attitudes item for which distribution only allowed dichotomization. Correlates significant at P≤.20 in univariate analyses were entered into the multivariate models concurrently. Those identified in stepwise procedures as significant at P≤.05 were retained in the final models. To examine if the proportion of patients younger than 18 years seen in practice modified the results, we tested interaction terms between each of the correlates identified as statistically significant and the proportion of patients younger than 18 years.
Of the 670 individuals selected from the physician database, 440 were eligible and 337 (77%) completed the questionnaire; 211 (62.6%) were males. Table 1 summarizes the sociodemographic characteristics and practice profiles of GPs by sex. The reported proportion of patients 18 years or younger seen in practice was low overall and did not exceed 20% for 76.1% of the physicians. Female physicians reported seeing significantly more younger patients than their male counterparts (Table 1).
Only 26.1% of GPs had heard of the stages of behavior change model and 45.7% knew of community resources to help patients quit. Although 84.6% of the respondents strongly agreed/agreed that family physicians should advise parents of their pediatric patients to quit smoking, only 32.7% (strongly agree/agree) felt they had the skills required to help their younger patients not to start smoking.
More GPs ascertained the smoking status of more than half of the adolescents (70.9%) than of the preadolescents (35.7%). Almost half of the GPs offered advice to prevent smoking onset for more than half of the young adults (48.6%) and adolescents (48.3%). Fewer did so for preadolescents (34.4%), and only 12.1% suggested to more than half of the parents who have children aged 5 through 19 years that they discuss smoking prevention with their children.
Female sex, lower proportions of walk-in patients, and awareness of the stage of behavior change model were independent correlates of ascertainment practices; more hours spent in direct patient care and favorable beliefs and attitudes were strongly correlated to provision of prevention advice (Table 2). Higher self-efficacy was a strong correlate of both ascertainment and prevention advice. None of the interaction terms that included the proportion of patients aged 18 years in practice were statistically significant (data not shown).
Our results concur with those of previous studies that indicate that physicians are missing opportunities during their interactions with youth and their parents to prevent tobacco use.14- 19,23 Specifically, physicians are more likely to ascertain adolescents' smoking status (36%-71% in our study and 56%-66% in previous surveys14,15) than to provide advice to prevent smoking onset (34%-49% in our study and 17%-35% in a previous survey14), and these practices are more often undertaken with older than younger adolescents.14,16 Low physician engagement has been described for other recommended prevention practices, including immunization; measurement of blood pressure and cholesterol level23,24; counseling related to nutrition; developmental conditions23,24; car, bike, and gun safety issues24,25; use of alcohol and other substances24,26; and sexual behavior.23,24,27 These findings suggest that physicians do not have access to published recommendations and guidelines for preventive practices or that these recommendations might not be sufficient to change physician behaviors. Specific strategies including reimbursement of physicians for prevention practices and evaluation of these practices as a qualitative indicator of the care physicians provide to adolescents have been suggested as mechanisms to promote physician prevention counseling behaviors.18 Also, new models of prevention might be necessary to help physicians integrate the growing body of guidelines and practice recommendations.
Our data suggest that several sociodemographic and clinical characteristics are related to physician prevention behaviors. Similar to prevention practices related to alcohol abuse and sexual behaviors,24,26,27 female physicians were more likely to ascertain smoking status among adolescents and preadolescents. This might reflect that female GPs see a higher proportion of younger patients and, therefore, are possibly more familiar with and have a higher level of comfort working with children and adolescents. Clinical predictors included lower proportions of walk-in patients for ascertainment practices and more time spent in patient care for provision of advice. Physicians seeing fewer walk-in patients might represent those doing more routine care visits during which prevention messages are more likely to be provided, in contrast to urgent or walk-in appointments. Those spending more time in patient care might see more adolescent patients; it has been reported that physicians who see a higher proportion of older adolescents in particular, are more likely to provide preventive messages.24,26 Halpern-Felsher et al24 and Marcell et al26 argue that a physician might think that older adolescents are at higher risk of engaging in risky behaviors and are more eligible for prevention advice. This hypothesis could not be verified in our study because we did not differentiate the proportion of adolescents seen by study subjects by age group (ie, we asked only about the proportion of patients <18 years seen in practice).
Previous surveys suggest that physicians with favorable beliefs and attitudes about adolescent interest in their counseling and compliance with their advice are more likely to screen for adolescent smoking.16 High confidence in the effectiveness of prevention counseling and the perception that colleagues are highly committed to prevention practices have also been identified as predictors of the provision of advice to deter the onset of smoking19 as have beliefs and attitudes and self-efficacy.25,26 Our data support these findings and underscore the need for interventions among physicians who target these psychosocial factors. Interestingly, our results suggest that self-efficacy levels, although not optimal, might be improving. Compared with a Massachusetts survey of 350 pediatricians conducted between 1995 and 1996, in which only 16% of 211 respondents were "quite confident" in their role to prevent smoking,19 33% of GPs in our survey strongly agreed/agreed that they have the skills required to help younger patients not start smoking.20 This might reflect the success of recent efforts to promote physicians' commitment to preventive practices including smoking prevention and cessation counseling.8 Finally, the association between awareness of the stage of behavior change model and ascertainment practices suggests that providing physicians with information about this model might increase their likelihood to screen for smoking among their younger patients. Alternatively, this variable could reflect physicians' heightened awareness and level of interest regarding smoking-related issues.
Although several barriers to counseling (including lack of training, time, and reimbursement; perceived lack of influence on patients; and lack of patients' interest in and compliance with physician advice) have been reported to limit counseling practices,14,17,19,20,23 none of the barriers studied herein were retained as independent correlates of preventive practices, concurring with results from previous work.19,25 Nevertheless, some authors have argued that intervention programs aimed at promoting physician prevention practices should consider time restraints and lack of reimbursement given the high proportion of physicians perceiving these issues as important barriers, and that the current health care delivery context exposes physicians to the high pressure of increasingly busy practices.
Our results concur with those of previous research in that physician preventive practices are well below current recommendations. These findings suggest that, in addition to dissemination of recommendations and guidelines, research investigating factors facilitating inclusion of preventive messages in routine medical care of adolescents is imperative. The importance of beliefs and attitudes, self-efficacy, and lack of confidence in approaching adolescents identified in this and other studies suggest that physician-training programs should target these factors. Finally, randomized trials investigating the effectiveness of physicians' delivery of prevention messages, particularly those relating to tobacco, are essential. In the meantime, the continued high prevalence of smoking among youth, the emerging data regarding the development of nicotine dependence at low exposures to cigarettes, and the high morbidity and mortality related to tobacco use justify active physician involvement in tobacco prevention messages.28
Because our data are based on self-reports, physicians' reports of counseling behaviors could be overestimated. Our response proportion was relatively high but generalizability of results might be limited by nonresponse. Given the low proportion of subjects reporting patients aged 18 years or younger, the failure to identify any effect of this variable on ascertainment and prevention advice practices could be because of a lack of power. Finally, our study did not collect detailed data on patient characteristics that could be correlated to GPs' practices. Data on the actual kinds of assistance physicians provide to youth to prevent smoking were also not collected. Future surveys may need to consider collecting these data to better evaluate physicians' smoking prevention counseling practices and their correlates.
Accepted for publication July 26, 2002.
This project was funded by the Direction de la Santé Publique de Montréal-Centre.
Dr O'Loughlin was a National Health Research Scholar during the study and is an Investigator of the Canadian Institutes of Health Research.
We acknowledge contributions by Garbis Meshefedjian, MSc, and thank the physicians who responded to the questionnaire.
Corresponding author and reprints: Héla Makni, MD, MSc, Direction de la Santé Publique de Montréal-Centre, Pavillon Lafontaine, 1301 Sherbrooke E, Montréal, Québec H2L 1M3 (e-mail: firstname.lastname@example.org).
Despite current recommendations for tobacco prevention counseling by physicians, these practices remain far from optimal. Research to identify predictors of physicians' smoking prevention practices, although crucial to better-tailored training and intervention programs aimed at promoting these practices, is as yet underdeveloped.
Our results suggest that female sex, lower proportion of walk-in patients, awareness of the stage of behavior change model, and higher self-efficacy to provide smoking counseling are independent correlates of ascertaining the smoking status of preadolescents and adolescents. Correlates of provision of prevention advice included more hours spent in direct patient care, favorable beliefs and attitudes about counseling, and higher self-efficacy to provide counseling.
Makni H, O'Loughlin JL, Tremblay M, Gervais A, Lacroix C, Déry V, Paradis G. Smoking Prevention Counseling Practices of Montreal General Practitioners. Arch Pediatr Adolesc Med. 2002;156(12):1263-1267. doi:10.1001/archpedi.156.12.1263