bs_bs_banner

REVIEW

doi:10.1111/add.12680

Do doctors’ smoking habits influence their smoking cessation practices? A systematic review and meta-analysis Maria J. Duaso1, Máirtín S. McDermott2, Agurtzane Mujika3, Edward Purssell4 & Alison While5 Department of Postgraduate Research, Florence Nightingale School of Nursing and Midwifery, King’s College London, London, UK,1 School of Information Systems and Technology, University of Wollongong, Wollongong, Australia,2 Faculty of Nursing, University of Navarra, Pamplona, Spain,3 Florence Nightingale School of Nursing and Midwifery, King’s College London, London, UK4 and Department of Postgraduate Research, Florence Nightingale School of Nursing and Midwifery, King’s College London, London, UK5

ABSTRACT Aims To assess the association between doctors’ smoking status and the use of the ‘5As’ of smoking cessation. Methods A systematic search of 11 databases covering English and Spanish language publications since 1996 was undertaken. Studies were included if they reported doctors’ smoking status (current, former or never smoker) and rates of practising any of the 5As of smoking cessation (Ask; Advise; Assess; Assist; and Arrange). Frequencies and proportions were extracted from individual papers and risk ratios (RR) were calculated. A random-effects meta-analysis model was used to assess the effect of the doctor’s personal smoking history. Covariate effects were explored using metaregression for three pre-specified study characteristics: doctors’ role, smoking prevalence of the sample and study quality. Results Twenty studies were included in this systematic review. The RR of always asking patients about their smoking was not associated significantly with doctors’ smoking status [RR = 0.98; 95% confidence interval (CI) = 0.94–1.02; P = 0.378; I2 = 0.00%; 10 studies]. Meta-analysis suggested that doctors who were current smokers had a 17% increased risk of not advising their patients to quit compared with never-smokers (RR = 0.83; 95% CI = 0.77–0.90; P < 0.000; I2 = 82.14%; 14 studies). However, high levels of heterogeneity were found that were not explained by the meta-regression. Regarding assisting patients to quit, never smokers were more likely to counsel than current smokers (RR = 0.92; 95% CI = 0.85–0.99; P = 0.036; I2 = 0.00%; three studies) but less likely to make a referral (RR = 1.40; 95% CI = 1.09–1.79; P = 0.009; I2 = 0.00%; five studies). No statistically significant differences were found in arranging future contact by smoking status (RR = 0.80; 95% CI = 0.52–1.23; P = 0.315; I2 = 47.03%; four studies). Conclusions Smoking status of doctors may affect their delivery of smoking cessation treatments to patients, with smokers being less likely than non-smokers or ex-smokers to advise and counsel their patients to quit but more likely to refer them to smoking cessation programmes. Keywords

5As, doctor, meta-analysis, smoker, smoking cessation, systematic review, tobacco.

Correspondence to: Maria J. Duaso, Department of Postgraduate Research, Florence Nightingale School of Nursing and Midwifery, King’s College London, London SE1 8WA, UK. E-mail: [email protected] Submitted 8 November 2013; initial review completed 18 February 2014; final version accepted 30 June 2014

INTRODUCTION Doctors play a leading role in reducing tobacco use. Given the frequency with which they encounter members of the public, they are ideally placed to make a significant impact on rates of smoking, which remains the single most preventable cause of death and illness in the world today [1]. Physician’s interventions can be effective in helping patients to stop smoking. The long-term success rate of those quitting unaided is 2–3%, with advice given © 2014 Society for the Study of Addiction

by doctors increasing this by a further 1–3% [2]. While the effect of such advice may seem small, if implemented routinely it could have a major impact at the population level. Since 1996 the US guidelines have recommended the ‘Five As approach’ (5As) for treating tobacco use and dependence [3]: ask about tobacco use; advise to quit; assess willingness to make a quit attempt; assist in quit attempt; and arrange follow-up. This approach is endorsed by the World Health Organization (WHO) [4], Addiction, 109, 1811–1823

1812

Maria J. Duaso et al.

and similar guidelines without the mnemonic exist in the United States [5] and elsewhere [6,7]. Despite these recommendations, the current implementation of the 5As is suboptimal and varies widely among doctors [8]. One major source of the varied implementation of stop smoking advice appears to be doctors’ own beliefs and attitudes towards smoking cessation. A systematic review of 13 studies [9] found that a sizable minority of primary care doctors held negative attitudes towards broaching smoking cessation with their patients. The doctors’ primary concern was that it was too timeconsuming, followed by beliefs that such intervention would be ineffective, and they had no confidence in their smoking cessation promotion ability. An earlier UK study [10] found that the majority (57%) of general practitioners (GPs) surveyed did not feel that it was appropriate to check the smoking status of their patients at each visit, a sizeable minority (39%) felt that it was inappropriate to advise all smokers to stop smoking at each visit and 23% felt that it was inappropriate for GPs to provide assistance to all smokers who wanted to stop. As well as beliefs about smoking cessation, health professionals also share the prevailing societal beliefs about smoking; for example, that smoking relieves stress and that smoking cessation will have a negative impact on mood [11,12], despite evidence to the contrary [13], which can also lead to a reduced likelihood of providing advice to quit [14]. One potential source both of these negative beliefs and of variations in practice are the doctors’ own health behaviours. Several studies have suggested that doctors who smoke have less positive attitudes towards smoking cessation [15,16] and are less likely to both take patient’s smoking history and provide advice than non-smoking colleagues [17–20]. However, the extent to which personal smoking behaviour affects the physician’s clinical practice is an important public health issue that has not been comprehensively investigated. The aim of the current review, therefore, is to determine if doctors’ smoking status (current, ex- or never smoker) is associated with engagement in smoking cessation, defined as practising any of the ‘5 As’: asking, advising, assessing, assisting or arranging follow-up.

METHOD Eligibility criteria Studies were selected for this review if they: 1 Reported and categorized doctors’ current smoking status, including whether they had previously smoked. 2 Reported doctors’ current smoking cessation practices comparable to the 5As of smoking cessation. 3 Reported statistical relationships between (1) and (2). 4 Were published in English or Spanish. © 2014 Society for the Study of Addiction

5 Were published from 1996 onwards. Studies were excluded if: 1 Past smoking history was not reported. Previous studies have suggested that health professionals with a past smoking history may have different preventive practices to those who have never smoked [21]. 2 Studies only reported doctors’ attitudes towards smoking cessation, confidence in their ability to help patients to quit or intention to counsel their patients. 3 Reported smoking cessation practices that were not comparable to the 5As of smoking cessation. 4 Doctors’ data were not reported independently, but mixed with data from other health professionals (e.g. nurses, dentists) or unqualified staff (e.g. medical students). Study identification Five electronic English databases (MEDLINE, EMBASE, PsycINFO, CINAHL and ERIC) and six Spanish databases (IBECS, Scielo, CUIDEN, ENFISPO, LILACS and MEDES) were searched using a combination of free text search terms [including doctor$, GP$ (quit$ or stop$ or ceas$ or giv$) adj smoking]. An example of MEDLINE full search strategy can be found in the Supporting information, Table S1. Database searches were run up to 15 February 2013. No attempt was made to access unpublished studies or other ‘grey’ literature. Two reviewers working independently screened all abstracts and titles (M.J.D. for English databases and A.M. for Spanish ones) selecting potentially eligible studies for full text evaluation. Reviewers working independently and in duplicate (M.J.D., A.M. and M.McD.) determined the eligibility of full text reports. A subsample of screened decisions was reviewed by a second author (M.McD.). A kappa coefficient agreement of 0.91 was reached. Data extraction from all finally included papers was doublechecked by an independent reviewer (A.W. or M.J.D.). Disagreements were resolved by consensus. Data synthesis Smoking status data were extracted. Doctors were grouped into current smokers, former smokers or never smokers. All smoking cessation practices reported in the included studies were grouped into five categories (5As). A coding checklist was developed and variables were included in the analysis if the authors reported comparable measurements (see examples in Table 1). For the first variable (Ask), studies were included if they reported ‘always’ identifying tobacco users. The other four smoking cessation practices (Advise, Assess, Assist and Arrange) were converted into a dichotomous variable if several frequencies were reported. Always or more than Addiction, 109, 1811–1823

Doctors’ smoking and their smoking cessation practice

1813

Table 1 Five As for treating tobacco use and dependence. The five As

Examples of interventions Ask always about or otherwise identify a patient’s smoking status

Ask—systematically identify all tobacco users at every visit Advise—strongly urge all tobacco users to quit Assess—determine willingness to make a quit attempt Assist—aid the patient in quitting attempt

Arrange—ensure follow-up contact

In a clear, strong and personalized manner, urge every tobacco user to quit Ask smokers about their interest in quitting smoking Provide brief counselling about how to quit smoking Give out written stop-smoking materials Discuss use of medications Discuss a quit date Refer to a smoking cessation class or programme Suggest a follow-up visit or telephone call about quitting smoking Schedule follow-up contact, in person or via telephone

Source: Fiore 2000 [22].

occasionally was considered as practising and occasionally or less considered as not practising. Moderator variables Studies have suggested that professionals who perceive smoking cessation as ‘highly’ relevant to their practice are more likely to advise their patients to stop [23,24]. Doctors’ specialities were coded independently by two researchers according to whether smoking cessation was considered an essential part of their role (e.g. primary care practitioners, pulmonologists, cardiologists, oncologists, infertility specialists) or not (hospital doctors, nonvascular surgeons, etc.). Therefore, samples were classified into two groups, ‘specialist’ or ‘non-specialist’. Studies with mixed samples (specialist + non-specialist) were classified as ‘non-specialist’ (Table 2). Another potential factor that could affect between study heterogeneity is the general tobacco control activity at national level at the time that the study took place. Doctors’ smoking status might have less impact on smoking cessation efforts in countries with high legislative and preventive regulations. The smoking prevalence of the sample was extracted as proxy indicator of national levels of tobacco control activity, as these have been linked to smoking rates among health professionals, particularly doctors [40–42]. The quality of the studies was assessed using an adaptation of the Centre for Evidence-Based Management Survey Scoring System [43]. The adapted tool used a 0–6 scoring system to appraise the methodological quality of cross-sectional surveys, including representativeness of the sample, response rate, validity of the tool and assessment of statistical significance. Rather than defining a minimum value below which a study would be excluded, a study’s quality score was used as a covariate to assess heterogeneity. © 2014 Society for the Study of Addiction

Data analysis Statistical analysis was carried out using Comprehensive Meta-analysis Software version 2 and Metafor for R. Frequencies and proportions were extracted from individual studies and three risk ratios (RR) (i.e. current versus never smoker; current versus ex-smoker; ex- versus never smoker) were calculated per each smoking cessation practice (5As). Random-effects meta-analysis models were used to assess the effect of the doctor’s personal smoking history on smoking cessation practices. Lack of homogeneity was tested using the χ2 Q statistic (significance level P < 0.1) and the descriptive percentage of variance due to heterogeneity among studies was assessed using the I2 statistic [44]. To assess the effect of the three potential moderator variables, univariable meta-regressions were carried out for each primary comparison (e.g. Ask/current versus never smokers). Interaction terms were calculated to test if the effect varied by covariate subgroup (speciality, smoking prevalence of the sample and quality score). Funnel plots were used to inspect visually for potential publication bias. Egger tests were conducted to test the symmetry of the plot [45]. Where there was evidence of asymmetry suggesting a possible publication bias, a sensitivity analysis and the ‘trim-and-fill’ method was carried out to assess the impact of each study on the combined effect [46].

RESULTS Description of studies After removing duplicates, a total of 3213 abstracts were screened. A total of 295 full texts were assessed for eligibility, from which 250 papers were excluded [see Preferred Reporting Items for Systematic Reviews and Addiction, 109, 1811–1823

© 2014 Society for the Study of Addiction

2007 2010 1994 1994 2009 2002 2004 2001 2000/4 2003 2000 Not stated 2000 Not stated 2000 Not stated 2003/5 2002 2003 2004

Aboyans 2009 [18] Araya 2012 [25] Easton 2001 [26] Easton 2001 [19] Freour 2011 [27] Jacot Sadowski 2009 [28] Jiang 2007 [20] Kossler 2002 [29] Meshefedjian 2010 [21] Ng 2007 [30] Ohida 2001 [15] Ozturk 2012 [31] Rico Lezama 2001 [32] Samuels 1997 [33] Sanchez 2003 [34] Schnoll 2006 [35] Sotiropoulos 2007 [36] Steinberg 2007 [37] Thankappan 2009 [38] Zylbersztejn 2007 [39]

*Non-specialist.

Survey year

First author (publication year)

Table 2 Details of included studies.

France Chile US US France Switzerland China Austria Canada Indonesia Japan Turkey Uruguay Israel Ecuador Russia Greece US India Argentina

Location 371 235 1452 1644 341 1856 3652 1395 618 447 3771 80 152 260 679 63 1284 336 339 6497

Sample size Cardiologists Hospital physicians* Non-primary care women physicians* Primary care women physicians Doctors specialized in infertility Primary care physicians Physicians* General practitioners General practitioners Physicians with primary clinical responsibilities* Physicians* Non-vascular surgeons* Hospital physicians* Hospital doctors* Physicians not specified* Oncologists Physicians from all specialities* General practitioners and general internists* Physicians (faculty and health service)* Hospital physicians*

Physician role National National National National National National Regional National Regional Regional National Institutional Institutional Institutional Regional Institutional National sample Regional Regional National

Sample origin Paper Paper Paper Paper Authors Authors Paper Paper Author Author Paper Paper Paper Author Paper Author Paper Paper Paper Paper

Source of data

8.1 16.2 4.7 3.5 12.5 17.7 22.9 7.1 7.4 11.9 20.3 17.5 30.9 15.7 32.3 27.0 38.6 3.3 10.8 30.0

Smoking prevalence

4 2 4 4 4 4 6 4 5 5 5 2 0 2 4 2 5 5 5 2

Qualityscore

1814 Maria J. Duaso et al.

Addiction, 109, 1811–1823

Records idenƟfied through database searching (n =5245)

Full-text arƟcles assessed for eligibility (n =295)

Figure 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram

Included

Studies included in qualitaƟve synthesis (n = 45 )

Records excluded based on Ɵtle, abstract and duplicity (n =2516)

Full-text arƟcles excluded, with reasons (n = 250) (a) physicians’ smoking status not reported (b) No assessment of past smoking (c) no reporƟng of 5As smoking cessaƟon pracƟces (d) No associaƟons tested (e) Mixed sample Authors did not provide requested informaƟon for meta-analysis (n = 25)

Studies included in Meta-analysis (n=20)

Meta-Analyses (PRISMA) Fig. 1]. Of 45 studies that were deemed potentially relevant, the authors of 31 studies were contacted to obtain further information to determine their eligibility. Some of the authors no longer held the original data, while others could not be contacted. Authors were contacted on at least two occasions, and we were able to obtain additional data relating to six studies. Finally, 20 studies were included in this systematic review. The study characteristics are presented in Table 2. The year of publication ranged from 1997 [33] to 2012 [25,31]. The studies included were conducted in 17 countries, eight of them in Europe, four in North America, four in South America and four in Asia. Eight of the studies included samples of doctors for whom smoking cessation was an important part of their role as GPs [19,21,28,29,37], cardiologists [18], oncologists [35] and infertility specialists [27]. The study sample sizes ranged from 63 to 6497 participants. Half the studies included national samples, six were regional and four were conducted in single study sites. The prevalence of smoking among the doctors in the included studies ranged from a minimum of 3% in two studies from the United States [19,37] to 38.6% in the Greek study [36]. The quality scores of the included studies ranged from 0 to 6, with a median of 4 (interquartile range = 2–5) (Table 2). Survey response rates ranged from 18.2% [29] to 97.0% [36]. Regarding the use of the five smoking cessation practices investigated, 10 studies [18,21,25,27,28,30,32,34, © 2014 Society for the Study of Addiction

1815

Spanish databases (n =484)

Records screened (n = 3213)

Eligibility

Screening

IdenƟficaƟon

Doctors’ smoking and their smoking cessation practice

35,38] included data on whether doctors always enquired about the smoking status of their patients (Table 3) and 14 studies [19–21,25–27,29,31,33– 36,38,39] on advising smokers to quit (Table 4). Assessing smokers’ motivation was reported in only one of the studies [35]. Assisting with a quit attempt can take many forms, and this was reflected by the different measurements reported (Table 5): counselling patients [21,27,28], providing written materials [15,21,35], setting a quit date [15,20,21,35], providing medications [15,20,21,28,31,35,37] and referring to a smoking cessation programme [15,21,28,31,35]. The fifth A, arranging a follow-up contact, was included in four studies [15,18,21,35] (Table 6).

Asking patients about smoking status A wide range of frequencies of reported practices were found, with 96.2% of French cardiologists routinely assessing smoking [18] compared only to 7.1% of Indonesian physicians reporting doing so [30]. Only two of the 10 studies [35,38] found statistically significant differences between doctors’ smoking status (Table 3). The meta-analysis confirmed the results of these individual studies (Table 7), as the pooled RR of always asking patients about their smoking was not associated significantly with the doctors’ smoking status. Heterogeneity in results was not significant across any of the three comparisons. Addiction, 109, 1811–1823

1816

Maria J. Duaso et al.

Table 3 Reported smoking cessation practices (ASK) by doctor’s smoking status in individual studies. Smoking status Study Aboyans 2009 [18] Current Former Never Araya 2012 [25] Current Former Never Freour 2011 [27] Current Former Never Jacot Sadowski 2009 [28] Current Former Never Meshefedjian 2010 [21] Current Former Never Ng 2007 [30] Current Former Never Rico Lezama 2001 [32] Current Former Never Sanchez 2003 [34] Current Former Never Schnoll 2006 [35] Current Former Never Thankappan 2009 [38] Current Former Never

ASK

n

%

n

%

P

30 120 220

8.1 32.4 59.3

27 113 216

90 94.2 98.2

0.110

38 86 11

8.1 63.7 28.1

30 71 91

78.9 82.4 82

0.886

42 103 192

12.5 30.6 57.0

33 88 159

79 85 83

0.599

325 537 969

17.7 29.3 52.9

264 455 791

81.2 84.7 81.6

0.257

44 192 358

7.4 32.3 60.3

9 41 85

20.5 21.4 23.7

0.761

52 117 268

11.9 26.8 61.3

3 13 19

5.8 11.1 7.1

0.335

47 36 69

30.9 23.7 45.4

45 34 64

95.7 94.4 92.8

0.931

218 176 280

32.3 26.1 41.5

97 84 136

44.5 47.7 48.6

0.649

17 32 14

27.0 50.8 22.2

0 3 5

0 9.4 35.7

0.042

36 87 210

10.8 26.1 62.1

8 36 75

22.2 41.2 35.6

0.001

Overall P-value corresponds to χ2 or Yates’ χ2 correction if any expected frequency was below 1.

Advising patients to quit smoking Six of the 14 studies included reported statistically significant lower rates of advising their patients to quit among doctors who smoked [19,20,25,29,34,36]. Primary random-effect meta-analysis suggested that, compared to never-smokers, current smokers had a 17% increased risk of not advising their patients to quit [RR = 0.83; 95% confidence interval (CI) = 0.77–0.90; P < 0.001]. They © 2014 Society for the Study of Addiction

were also less likely to advise when compared to former smokers (RR = 0.82; 95% CI = 0.74–0.90; P < 0.000). There was evidence of heterogeneity in both comparisons (Q = 72.79, d.f. = 13, P < 0.000, I2 = 82.14) and (Q = 81.02 d.f. = 13, P = 0.000, I2 = 83.95), respectively. However, a mixed-effect meta-regression model indicated that heterogeneity in the RR was not significantly affected by doctors’ speciality, overall prevalence of the sample or study quality (Supporting information, Table S2). Addiction, 109, 1811–1823

Doctors’ smoking and their smoking cessation practice

1817

Table 4 Reported smoking cessation practices (ADVISE) by doctor’s smoking status in individual studies. Smoking status Study Araya 2012 [25] Current Former Never Easton 2001a [19] Current Former Never Easton 2001b [26] Current Former Never Freour 2011 [27] Current Former Never Jiang 2007 [20] Current Former Never Kossler 2002 [29] Current Former Never Meshefedjian 2010 [21] Current Former Never Ozturk 2012 [31] Current Former Never Samules 1997 [33] Current Former Never Sanchez 2003 [34] Current Former Never Schnoll 2006 [35] Current Former Never Sotiropoulos 2007 [36] Current Former Never Thankappan 2009 [38] Current Former Never Zylbersztejn 2007 [39] Current Former Never

n

ADVISE %

n

%

38 86 11

8.1 63.7 28.1

24 72 88

66 301 1030

4.7 21.5 73.7

55 316 1219

P

63.2 83.7 79.1

0.035

24 123 465

32 43 46

0.193

3.5 19.9 76.7

29 223 858

32 43 46

0.019

42 103 192

12.5 30.6 57.0

40 95 96

95 92 96

0.599

813 97 2642

22.9 2.7 74.4

432 65 1789

53.1 67 67.7

0.043

150 710 1245

7.1 33.7 59.1

104 504 1096

69.0 71.0 88.0

0.001

44 192 358

7.4 32.3 60.3

25 129 231

56.8 67.9 63.1

0.308

14 8 58

17.5 10.0 72.5

12 8 52

85.7 100.0 89.7

0.928

40 53 162

15.7 20.8 63.5

34 52 150

85.0 98.1 92.6

0.138

218 176 280

32.3 26.1 41.5

117 135 202

53.2 76.3 71.6

0.001

17 32 14

27.0 50.8 22.2

4 16 8

23.5 50.0 57.1

0.115

496 177 611

38.6 13.8 47.6

369 150 521

74.4 84.7 85.3

0.001

36 87 210

10.8 26.1 62.1

25 63 168

69.4 72.4 80.0

0.071

1916 1429 3039

30.0 22.4 47.6

1110 1140 2368

58.9 81.0 79.0

0.000

Overall P-value corresponds to χ2 or Yates’ χ2 correction if any expected frequency was below 1.

© 2014 Society for the Study of Addiction

Addiction, 109, 1811–1823

© 2014 Society for the Study of Addiction

12.5 30.6 57.0

17.7 29.3 52.9

22.9 2.7 74.4

7.4 32.3 60.3

20.3 28.0 51.8

17.5 10.0 72.5

27.0 50.8 22.2

3.3 22.9 73.8

325 537 969

813 97 2642

44 192 358

765 1054 1952

14 8 58

17 32 14

11 77 248

%

42 103 192

n

– – –

– – –

– – –

– – –

25 110 215

– – –

209 401 682

29 87 143

n

– – –

– – –

– – –

– – –

55.6 57.9 58.9

– – –

63.9 74.4 70.1

69.0 84.4 74.5

%

0.258

0.005

0.071

P

ASSIST counselling

– – –

– – – 1 2 1

33 62 83

6 18 38

– – –

– – –

– – –

n

– – –

– – – 5.9 6.3 7.1

4.3 5.9 4.3

14.3 9.6 10.7

– – –

– – –

– – –

%

0.989

0.111

0.673

P

ASSIST written matter

Overall P-value corresponds to χ2 or Yates; χ2 correction if any expected frequency was below 1.

Freour 2011 [27] Current Former Never Jacot Sadowski 2009 [28] Current Former Never Jiang 2007 [20] Current Former Never Meshefedjian 2010 [21] Current Former Never Ohida 2001 [15] Current Former Never Ozturk 2012 [31] Current Former Never Schnoll 2006 [35] Current Former Never Steinberg 2007 [37] Current Former Never

Study

Smoking status

– – –

– – – 1 3 3

29 53 90

20 90 151

37 6 172

– – –

– – –

n

4.6 6.2 6.5

– – –

5.9 9.4 21.4

– – –

3.8 5.0 4.6

47.6 48.1 42.4

– – –

– – –

%

0.354

0.453

0.410

0.123

P

ASSIST quit date

Table 5 Reported smoking cessation practices (ASSIST) by doctor’s smoking status in individual studies.

8 55 185

1 2 0

0 0 0

108 142 268

22 115 207

48 13 175

273 447 829

– – –

n

72.7 71.4 74.6

5.9 6.3 0.0

0.0 0.0 0.0

14.1 13.5 13.7

50.0 61.5 58.0

5.9 13.4 6.6

83.5 82.9 85.2

– – –

%

ASSIST NRT

0.929

0.637

0.933

0.360

0.020

0.470

P

– – –

– – –

– – –

– – –

– – –

40 5 116

224 350 631

– – –

n

– – –

– – –

– – –

– – –

– – –

4.9 5.2 4.4

68.5 64.9 64.9

– – –

%

0.782

0.458

P

ASSIST bupropion

38 42 95

– – –

1 3 1

7 4 24

12 8 15

16 63 105

– – –

– – –

n

– – –

5.9 9.4 7.1

50.0 50.0 41.4

1.6 0.8 0.8

37.2 33.5 29.5

– – –

11.6 7.8 9.8

– – –

%

ASSIST referral

0.904

0.786

0.118

0.435

0.164

P

1818 Maria J. Duaso et al.

Addiction, 109, 1811–1823

Doctors’ smoking and their smoking cessation practice

Table 6 Reported smoking cessation practices (ARRANGE) by doctor’s smoking status in individual studies.

Study

Smoking status

ARRANGE

n

%

n

8.1 32.4 59.3

23 76.7 107 89.2 0.018 204 92.7

7.4 32.3 60.3

13 31.0 43 23.0 0.309 102 28.7

Aboyans 2009 [18] Current 30 Former 120 Never 220 Meshefedjian 2010 [21] Current 44 Former 192 Never 358 Ohida 2001 [15] Current 765 Former 1054 Never 1952 Schnoll 2006 [35] Current 17 Former 32 Never 14

20.3 28.0 51.8 27.0 50.8 22.2

%

2 9 29

P

0.3 0.9 0.015 1.5

2 11.8 5 15.6 0.935 2 14.3

1819

Two studies reported the use of bupropion when helping smokers to quit [20,28]. Data provided by the authors from a survey of 3385 Swiss primary care doctors suggests that a third prescribed bupropion [28], while fewer than 5% of hospital-based Chinese physicians used it when helping smokers to quit [20]. Overall, no statistically significant differences were found across the doctors’ smoking status (Table 7). Finally, the rates of referring to cessation services also varied. For example, 41% of non-vascular surgeons in a Turkish hospital referred their patients to specialist services [31], while fewer than 1% of a national sample of 3771 Japanese physicians reported doing so [15]. None of the five studies including data on referral found statistically significant differences by smoking status of the doctors. However, the pooled RR suggested that smokers are 1.4 times more likely to refer than never smokers (RR = 1.40; 95% CI = 1.09–1.79, P = 0.009) (Supporting information, Fig. S9). Arranging follow-up

Overall P-value corresponds to χ2 or Yates’ χ2 correction if any expected frequency was below 1.

Assisting in quit attempt Meta-analysis of three studies reporting on counselling of patients (Table 7) suggests that smokers were 8% less likely to counsel than never smokers and 14% less likely than former smokers (RR = 0.92; 95% CI = 0.85– 0.99, P = 0.036 and RR = 0.86; 95% CI = 0.79–0.94; P < 0.001, respectively) with no significant heterogeneity. None of the three studies reporting on assistance in the form of written materials found significant differences by doctors’ smoking status [15,21,35]. The pooled RRs were also non-significant (Table 7). Despite the well-known effectiveness of setting a quit date, overall rates of reported assistance to enable smokers to do so were low. Doctors’ smoking status did not appear to have any impact in individual studies [15,20,21,35] or pooled RRs (Table 7). Six studies reported doctors’ provision of nicotine replacement therapy (NRT) when assisting smokers to quit (Table 5). There were marked differences across the studies in terms of rates of provision, with a study of Russian oncologists providing none at all [35] compared to more than two-thirds of a national sample of US primary care doctors [19]. Of the seven studies, only one [20] found doctors’ smoking status to be a statistically significant factor in assisting smokers to quit using NRT, with 5.9% of current smokers prescribing NRT compared to 13.4% of former and 6.6% of never smokers (P = 0.02). The pooled RRs in the meta-analysis did not suggest an overall effect (Table 7). © 2014 Society for the Study of Addiction

The last step of the 5As is to schedule a follow-up contact either in person or over the telephone. Once more, reported rates were variable (Table 6). Higher levels of follow-up were found among French cardiologists (92%) [18] compared to 24% of GPs in Montreal [21], and fewer than 1% of Japanese hospital-based doctors [15]. Two of the four studies found statistically significant differences, with smokers reporting lower levels of follow-up [15,18]. The pooled RRs in the meta-analysis did not suggest an overall effect (Table 7). However meta-regression suggests that the effect varies by speciality subgroup, with specialist doctors being more likely to arrange follow-up (RR = 4.89; 95% CI = 1.16–20.67; P = 0.031) (Supporting information, Table S2). Publication bias There was evidence of asymmetry in two of the metaanalyses: Asking current versus never smokers and Asking current versus ex-smokers (Supporting information, Table S3, Fig. S1), with fewer smaller studies showing no relationship than might be expected. Sensitivity analysis using the ‘leave-one-out’ method [47] revealed that the association of asking and smoking status remained statistically non-significant when any one of the studies was excluded.

DISCUSSION We hypothesized that doctors’ own smoking status or smoking history could be a source of variation in the current variable and suboptimal implementation of recommended smoking cessation advice in the form of the Addiction, 109, 1811–1823

1820

Maria J. Duaso et al.

Table 7 Meta-analysis doctor’s reported practice of 5As smoking cessation by smoking status. Pooled effect size

ASK Current versus never-smokers Current versus ex-smokers Ex- versus never-smokers ADVISE Current versus never-smokers Current versus ex-smokers Ex- versus never-smokers ASSIST counselling Current versus never-smokers Current versus ex-smokers Ex- versus never-smokers ASSIST written materials Current versus never-smokers Current versus ex-smokers Ex- versus never-smokers ASSIST quit date Current versus never-smokers Current versus ex-smokers Ex- versus never-smokers ASSIST NRT Current versus never-smokers Current versus ex-smokers Ex- versus never-smokers ASSIST bupropion Current versus never-smokers Current versus ex-smokers Ex- versus never-smokers ASSIST referral Current versus never-smokers Current versus ex-smokers Ex- versus never-smokers ARRANGE Current versus never-smokers Current versus ex-smokers Ex- versus never-smokers

Heterogeneity

Relative risk

95% CI

P-value

No of studies

0.98 0.96 1.01

0.94–1.02 0.92–1.00 0.96–1.05

0.378 0.087 0.774

10 10 10

0.83 0.82 1.02

0.77–0.90 0.74–0.90 0.99–1.05

0.000 0.000 0.112

0.92 0.86 1.07

0.85–0.99 0.79–0.94 1.00–1.14

1.07 0.85 1.23

d.f.

P-value

I2

8.77 6.18 12.95

9 9 9

0.459 0.722 0.165

0.00 0.00 30.51

14 14 14

72.79 81.02 22.99

13 13 13

0.000 0.000 0.042

82.14 83.95 43.46

0.036 0.000 0.046

3 3 3

0.08 0.76 2.48

2 2 2

0.961 0.683 0.290

0.00 0.00 19.30

0.75–1.51 0.57–1.28 0.94–1.61

0.720 0.443 0.139

3 3 3

0.41 2.10 1.91

2 2 2

0.816 0.349 0.385

0.00 4.97 0.00

0.85 0.87 1.10

0.64–1.14 0.67–1.13 0.94–1.30

0.272 0.295 0.234

4 4 4

4.87 1.16 1.74

3 3 3

0.181 0.763 0.628

38.42 0.00 0.00

0.98 0.93 0.98

0.93–1.03 0.79–1.10 0.93–1.03

0.412 0.418 0.412

6 6 6

1.84 9.78 1.84

5 5 5

0.871 0.082 0.871

0.00 48.87 0.00

1.06 1.05 1.00

0.97–1.15 0.96–1.16 0.93–1.08

0.176 0.283 0.949

2 2 2

0.10 0.05 0.13

1 1 1

0.749 0.830 0.722

0.00 0.00 0.00

1.40 1.19 1.17

1.09–1.79 0.92–1.53 0.96–1.42

0.009 0.177 0.114

5 5 5

1.66 0.37 0.37

4 4 4

0.798 0.985 0.985

0.00 0.00 0.00

0.80 0.92 0.95

0.52–1.23 0.64–1.33 0.87–1.02

0.315 0.667 0.162

4 4 4

5.66 4.43 3.03

3 3 3

0.129 0.219 0.387

47.03 32.25 1.10

Q

CI = confidence interval.

5As. Overall, there was limited evidence in our metaanalysis to suggest this. Of the smoking cessation practices investigated, only advising smokers to quit, assisting with counselling and referring were found to vary by smoking status. With almost 60 years since the first research linking smoking and ill health [48], it may be that the negative impact of smoking on health is accepted to the point that doctors’ own smoking cessation practices are unaffected by their current or former smoking status. Also, the gradual increase of teaching on tobacco in medical schools [49] may have reduced any potential differences in practice. A recent systematic review of randomized trials concluded that participation in formal training in © 2014 Society for the Study of Addiction

smoking cessation significantly increases the rate of quit advice provided to patients [50]. It would have been valuable to include training as a moderator variable, but unfortunately this was not reported in most of the studies. Former smokers advised and assisted with counselling more than current smokers, suggesting that stopping smoking might have a positive effect on smoking cessation practices. It is important to note that most of the studies included here were not designed to evaluate the association between smoking status and smoking cessation practices, and therefore may be underpowered to detect this. Interestingly, doctors who smoke overall were more likely to refer to smoking cessation services or equivalent Addiction, 109, 1811–1823

Doctors’ smoking and their smoking cessation practice

than current and former smokers. A comparative study of 11 European Union (EU) countries concluded that GPs who smoke tended to feel less effective in helping patients to reduce tobacco consumption than non-smoking GPs (39.34 versus 48.18%, P < 0.01) [16]. It could be hypothesized that doctors do not personally feel able to assist with quitting if they themselves are smokers, and so refer out to other agencies. Our findings suggest that overall the delivery of the 5As is suboptimal. These results should be viewed with caution, as only a subset of studies addressing 5As delivery is included here. However, our results are consistent with other studies that have also found that, while patients are often asked about their smoking habits and advised to stop smoking, assistance and follow-up rates are lower [51–53]. Key findings from the Smoking Toolkit Study (STS) in England suggest that only a minority of smokers have discussed smoking with their GP in the past year, and just 25.9% have been offered a prescription or referred to a smoking cessation service [54]. The strengths of this review include a comprehensive search strategy of both English and Spanish databases including a wide range of countries, rigorous and reproducible extraction of data and the contacting of authors for further information. The use of a random-effect model and meta-regression analyses to explain heterogeneity also strengthens confidence in the estimation results presented here. A key limitation of this work is the between-study heterogeneity. This systematic review has brought together studies that are diverse in terms of setting, methodological quality and speciality of participants. Studies reported over a period of almost 20 years and both smoking prevalence and practices may have changed over this time. Reported smoking cessation practices varied widely across studies. We used doctors’ specialities as a proxy indicator of how integral smoking cessation was to their role. However, this has some limitations, as smoking role perceptions may vary according to national guidance or local policies. Table 7 shows high levels of heterogeneity across the studies that assess Advise practices that are not explained in the meta-regression; consequently, the overall effect must be viewed with caution. Conversely, a visual inspection of the forest plot (Supporting information, Fig. S3) suggests that, despite the different RR estimates of the individual studies, there is no inconsistency in the direction of the effect, with smoking doctors seeming to be less likely to advise their smoking patients to quit. Another limitation is the omission of grey literature in the search; however, due to the importance of the subject and the relative robustness of our findings in the sensitivity analysis there is little reason to think these are significant. © 2014 Society for the Study of Addiction

1821

Several recommendations can be made for future research based on our findings. This review confirms that provision of the 5As of smoking cessation among doctors remains low and variable. It appears that the impact of doctors’ own smoking behaviour is limited. Future research should investigate other reasons for the shortfall and variability in adherence to this recommended practice, and could also investigate whether interventions to reduce smoking among doctors increase the delivery of smoking cessation interventions. Finally, it would be interesting to ascertain whether the findings reported in this review are consistent in other professional groups providing valuable smoking cessation support, such as nurses, midwives, dentists or pharmacists.

CONCLUSIONS Meta-analyses of the currently available studies suggest that, while the smoking status of doctors does not affect whether or not they monitor patient’s smoking status, it may have an impact on them advising their smoking patients to quit. Smoking doctors are at higher risk of not assisting their patients with counselling when compared to former smokers and non-smokers. Conversely, they seem more likely to refer their patients to a smoking cessation programme. These findings must be interpreted within the context of the limitations of the data, but suggest that smoking cessation among doctors might extend beyond their personal health and benefit their patients. Declaration of interests None. References 1. World Health Organization. WHO Report on the Global Tobacco Epidemic, 2008: The MPOWER Package. Geneva: World Health Organization; 2008. Available at http:// www.who.int/tobacco/mpower/mpower_report_full_2008 .pdf (accessed 28 July 2014) (Archived at http:// www.webcitation.org/6RPZouBsu on 28 July 2014). 2. Stead L. F., Bergson G., Lancaster T. Physician advice for smoking cessation. Cochrane Database Syst Rev 2008; (2): CD000165. 3. Tobacco Use and Dependence Guideline Panel. Treating Tobacco Use and Dependence: 2008 Update. Washington, DC: US Department of Health and Human Services; 2008. 4. Silva D. Tools for Advancing Tobacco Control in the 21 St Century: Policy Recommendations for Smoking Cessation and Treatment of Tobacco Dependence. Geneva: World Health Organization; 2003. Available at: http://www.who.int/ tobacco/resources/publications/en/intro_chapter3.pdf (accessed 15 May 2014) (Archived at http://www .webcitation.org/6PaTgjxbu). 5. National Institute for Health and Clinical Excellence (NICE). Brief Interventions and Referral for Smoking Cessation in Addiction, 109, 1811–1823

1822

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

Maria J. Duaso et al.

Primary Care and Other Settings. London: National Institute for Health and Clinical Excellence; 2006. Zwar N., Richmond R., Borland R., Peters M., Litt J., Bell J. et al. Supporting Smoking Cessation: A Guide for Health Professionals. Melbourne: The Royal Australian College of General Practitioners; 2011. Available at: http://www.racgp.org .au/download/documents/Guidelines/smoking-cessation .pdf (accessed 28 July 2014) (Archived at http://www .webcitation.org/6PaTl3YI6 on 15 May 2014). CAN-ADAPTT (Canadian Action Network for the Advancement, Dissemination and Adoption of Practice-informed Tobacco Treatment). Canadian Smoking Cessation Clinical Practice Guideline for Smoking Cessation 2011. Available at: https://www.nicotinedependenceclinic.com/English/ CANADAPTT/Documents/CAN-ADAPTT%20Canadian %20Smoking%20Cessation%20Guideline_website.pdf (accessed 15 May 2014) (Archived at http://www .webcitation.org/6PaTZVYKW). Aveyard P., Begh R., Parsons A., West R. Brief opportunistic smoking cessation interventions: a systematic review and meta-analysis to compare advice to quit and offer of assistance. Addiction 2012; 107: 1066–73. Vogt F., Hall S., Marteau T. M. General practitioners’ and family physicians’ negative beliefs and attitudes towards discussing smoking cessation with patients: a systematic review. Addiction 2005; 100: 1423–31. McEwen A., Akotia N., West R. General practitioners’ views on the English national smoking cessation guidelines. Addiction 2001; 96: 997–1000. Praveen K. T., Kudlur S. N. C., Hanabe R. P., Egbewunmi A. T. Staff attitudes to smoking and the smoking ban. Psychiatr Bull 2009; 33: 84–8. Dickens G. L., Stubbs J. H., Haw C. M. Smoking and mental health nurses: a survey of clinical staff in a psychiatric hospital. J Psychiatr Ment Health Nurs 2004; 11: 445– 51. McDermott M. S., Marteau T. M., Hollands G. J., Hankins M., Aveyard P. Change in anxiety following successful and unsuccessful attempts at smoking cessation: cohort study. Br J Psychiatry 2013; 202: 62–7. Campion J., Checinski K., Nurse J., McNeill A. Smoking by people with mental illness and benefits of smoke-free mental health services. Adv Psychiatr Treat 2008; 14: 217– 28. Ohida T., Sakurai H., Mochizuki Y., Kamal A. M. M., Takemura S., Minowa M. et al. Smoking prevalence and attitudes toward smoking among Japanese physicians. JAMA 2001; 285: 2643–8. Brotons C., Bjorkelund C., Bulc M., Ciurana R., Godycki-Cwirko M., Jurgova E. et al. Prevention and health promotion in clinical practice: the views of general practitioners in Europe. Prev Med 2005; 40: 595– 601. Frank E., Segura C., Shen H., Oberg E. Predictors of Canadian physicians’ prevention counseling practices. Can J Public Health 2010; 101: 390–5. Aboyans V., Pinet P., Lacroix P., Laskar M. Knowledge and management of smoking-cessation strategies among cardiologists in France: a nationwide survey. Arch Cardiovasc Dis 2009; 102: 193–9. Easton A., Husten C., Malarcher A., Elon L., Caraballo R., Ahluwalia I. et al. Smoking cessation counseling by primary care women physicians: women physicians’ health study. Women Health 2001; 32: 77–91.

© 2014 Society for the Study of Addiction

20. Jiang Y., Ong M. K., Tong E. K., Yang Y., Nan Y., Gan Q. et al. Chinese physicians and their smoking knowledge, attitudes, and practices. Am J Prev Med 2007; 33: 15–22. 21. Meshefedjian G. A., Gervais A., Tremblay M., Villeneuve D., O’Loughlin J. Physician smoking status may influence cessation counseling practices. Can J Public Health 2010; 101: 290–3. 22. Fiore M. C. Treating Tobacco Use and Dependence: An EvidenceBased Clinical Practice Guideline for Tobacco Cessation. Rockville, MD: US Department of Health and Human Services; 2000. 23. Tremblay M., Cournoyer D., O’Loughlin J. Do the correlates of smoking cessation counseling differ across health professional groups? Nicotine Tob Res 2009; 11: 1330–8. 24. Tong E. K., Strouse R., Hall J., Kovac M., Schroeder S. A. National survey of U.S. health professionals’ smoking prevalence, cessation practices, and beliefs. Nicotine Tob Res 2010; 12: 724–33. 25. Araya M. V., Leal F., Huerta P., Fernandez N., Fernandez G., Millones J. P. The influence of smoking habits of Chilean physicians on the use of the structured medical advice about smoking. Rev Med Chile 2012; 140: 347–52. 26. Easton A., Husten C., Elon L., Pederson L. L., Frank E. Nonprimary care physicians and smoking cessation counseling: women physicians’ health study. Women Health 2001; 34: 15–29. 27. Freour T., Dessolle L., Jean M., Barriere P. Smoking among French infertility specialists: habits, opinions and patients’ management. Eur J Obstet Gynecol Reprod Biol 2011; 155: 44–8. 28. Jacot Sadowski I., Ruffieux C., Cornuz J. Self-reported smoking cessation activities among Swiss primary care physicians. BMC Fam Pract 2009; 10: 22. 29. Kossler W., Lanzenberger M., Zwick H. Smoking habits of office-based general practitioners and internists in Austria and their smoking cessation efforts. Wien Klin Wochenschr 2002; 114: 762–5. 30. Ng N., Prabandari Y. S., Padmawati R. S., Okah F., Haddock C. K., Nichter M. et al. Physician assessment of patient smoking in Indonesia: a public health priority. Tob Control 2007; 16: 190–6. 31. Ozturk O., Yilmazer I., Akkaya A. The attitudes of surgeons concerning preoperative smoking cessation: a questionnaire study. Hippokratia 2012; 16: 124–9. 32. Rico Lezama M. N., Cataldi Belotti S., Grolero de Cat M. L. [Survey to medical teaching and non-teaching staff attending the XV Uruguayan Congress of Pneumology in September 2000.]. Arch Med Intern 2001; 23: 177–82. 33. Samuels N. Smoking among hospital doctors in Israel and their attitudes regarding anti-smoking legislation. Public Health 1997; 111: 285–8. 34. Sanchez P., Lisanti N. Prevalencia de tabaquismo y actitud hacia ese hábito entre médicos del Azuay, Ecuador [Smoking prevalence and attitudes toward smoking among physicians in Azuay, Ecuador]. Rev Panam Salud Publica 2003; 14: 25–30. 35. Schnoll R. A., Engstrom P. F., Subramanian S., Demidov L., Wielt D. B. Smoking cessation counseling by Russian oncologists: opportunities for intervention in the Russian Federation. Int J Behav Med 2006; 13: 8–15. 36. Sotiropoulos A., Gikas A., Spanou E., Dimitrelos D., Karakostas F., Skliros E. et al. Smoking habits and associated factors among Greek physicians. Public Health 2007; 121: 333–40. Addiction, 109, 1811–1823

Doctors’ smoking and their smoking cessation practice

37. Steinberg M. B., Nanavati K., Delnevo C. D., Abatemarco D. J. Predictors of self-reported discussion of cessation medications by physicians in New Jersey. Addict Behav 2007; 32: 3045–53. 38. Thankappan K. R., Pradeepkumar A. S., Nichter M. Doctors’ behaviour and skills for tobacco cessation in Kerala. Ind J Med Res 2009; 129: 249–55. 39. Zylbersztejn H. M., Cardone A., Vainstein N., Mulassi A., Calderón J. G., Blanco P. et al. [Smoking among Argentinian doctors: TAMARA study]. Rev Argentina Cardiol 2007; 75: 109–16. 40. World Health Organization. Tobacco Control Country Profiles. Atlanta: American Cancer Society; 2003. Available at: http://www.who.int/tobacco/surveillance/policy/ country_profile/en/ (accessed 15 May 2014) (Archived at http://www.webcitation.org/6PaTqyzMh). 41. John U., Hanke M. Tobacco-smoking prevalence among physicians and nurses in countries with different tobaccocontrol activities. Eur J Cancer Prev 2003; 12: 235–7. 42. Lopez A. D., Collishaw N. E., Piha T. A descriptive model of the cigarette epidemic in developed countries. Tob Control 1994; 3: 242–7. 43. Centre for Evidence Based Management. CEBMa Critical questions for a survey. Available at: http://www.cebma .org/wp-content/uploads/Critical-Appraisal-Questions-for -a-Survey.pdf (accessed 15 May 2014) (Archived at http:// www.webcitation.org/6PaVNvrim). 44. Higgins J. P., Thompson S. G., Deeks J. J., Altman D. G. Measuring inconsistency in meta-analyses. BMJ 2003; 327: 557–60. 45. Egger M., Davey Smith G., Schneider M., Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: 629–34. 46. Duval S., Tweedie R. Trim and fill: a simple funnel-plotbased method of testing and adjusting for publication bias in meta-analysis. Biometrics 2000; 56: 455–63. 47. Borenstein M., Hedges L., Higgins J., Rothstein H. Comprehensive Meta-Analysis. Englewood, NJ: Biostat; 2005. 48. Doll R., Hill A. B. Smoking and carcinoma of the lung; preliminary report. BMJ 1950; 2: 739–48. 49. Richmond R., Zwar N., Taylor R., Hunnisett J., Hyslop F. Teaching about tobacco in medical schools: a worldwide study. Drug Alcohol Rev 2009; 28: 484–97. 50. Carson K. V., Verbiest M. E., Crone M. R., Brinn M. P., Esterman A. J., Assendelft W. J. J. et al. Training health professionals in smoking cessation. Cochrane Database Syst Rev 2012; 16: (5): CD000214. 51. Glasgow R. E., Emont S., Miller D. C. Assessing delivery of the five ‘As’ for patient-centered counseling. Health Promot Int 2006; 21: 245–55.

© 2014 Society for the Study of Addiction

1823

52. Hazlehurst B., Sittig D. F., Stevens V. J., Smith K. S., Hollis J. F., Vogt T. M. et al. Natural language processing in the electronic medical record: assessing clinician adherence to tobacco treatment guidelines. Am J Prev Med 2005; 29: 434–9. 53. Hollis J. F., Bills R., Whitlock E., Stevens V. J., Mullooly J., Lichtenstein E. Implementing tobacco interventions in the real world of managed care. Tob Control 2000; 9: i18–24. 54. Fidler J. A., Shahab L., West O., Jarvis M. J., McEwen A., Stapleton J. A. et al. ‘The Smoking Toolkit Study’: a national study of smoking and smoking cessation in England. BMC Public Health 2011; 11: 1471–2458.

Supporting Information Additional Supporting information may be found in the online version of this article at the publisher’s web-site: Figure S1 Funnel plots showing relation between the association of smoking cessation practices/smoking status of physicians (log risk ratio) and the standard error of the log risk ratio Figure S2 Forest plot ASK. Current smokers versus never smokers Figure S3 Forest plot ADVISE. Current smokers versus never smokers Figure S4 Forest plot ASSIST counseling current smokers versus never smokers Figure S5 Forest plot ASSIST providing written material. Current smokers versus never smokers Figure S6 Forest plot ASSIST setting up a quit date. Current smokers versus never smokers Figure S7 Forest plot ASSIST with nicotine replacement therapy (NRT). Current smokers versus never smokers Figure S8 Forest plot ASSIST with bupropion. Current smokers versus never smokers Figure S9 Forest plot ASSIST making a referral. Current smokers versus never smokers Figure S10 Forest plot ARRANGE. Current smokers versus never smokers Table S1 Medline search strategy. Table S2 Meta-regression. Table S3 Publication bias.

Addiction, 109, 1811–1823