A mixed-method systematic review and meta-analysis of mental health professionals' attitudes toward smoking and smoking cessation among people with mental illnesses

1The prevalence of smoking among people with mental illnesses and substance use disorders is much higher than in the general population. Smoking rates stand at above 70% for those with severe mental illnesses such as schizophrenia and bipolar disorder 1, 2, 3, are similarly high among individuals with substance use disorders 4, 5 and are also above average for those with common mental disorders, including depression and anxiety 1. People with mental illness or substance use disorders are more likely to be heavier and more dependent smokers 6, 7, and their life expectancy is reduced by up to 20 years 8, 9, mainly because of smoking 7. It is therefore vital that smokers from this population receive effective smoking cessation interventions.

2Mental health professionals (MHPs), broadly defined as those who received specialist training to offer services designed to improve an individual's mental health (such as clinical psychologists, psychiatrists, clinical social workers and psychiatric nurses), have a crucial role in reducing tobacco smoking among people with mental illness, as they are best placed to encourage and support smokers to quit 10. Indeed, UK clinical guidelines on smoking cessation in secondary care advise that all health and social care practitioners in in‐patient and community‐based mental health services identify smokers and offer advice and assistance to quit 11. However, smokers with mental illness are less likely to be offered advice and support to quit than those without 12, 13.

3One possible barrier to the delivery of smoking cessation intervention is MHPs' attitudes towards and beliefs about smoking and/or smoking cessation among individuals with mental illnesses and substance abuse disorders. For instance, previous research has identified that some hold concerns that patients' mental health or abstinence will suffer 14, 15, 16, 17 and that smokers are unable 18 or unmotivated to quit 19. This is in contrast to evidence that smoking cessation, if not actually benefitting mental health, does not impact mental health negatively in people with or without psychiatric disorders 20, even in life‐long, long‐term smokers 21, and that smokers with mental illnesses are just as likely to want to quit as smokers without 22. MHPs have also been reported to hold permissive attitudes towards smoking, such as the belief that smoking with patients can help build a therapeutic relationship 15, 23, 24 and that allowing patients to smoke ensures a smoother running of wards in in‐patient settings 19.

4To implement smoking cessation interventions effectively for people with mental illness and substance abuse disorders, it is necessary to understand the relevant attitudes and beliefs of MHPs who engage with these individuals. Such an understanding would lead to greater insight into the potential barriers to delivering smoking cessation support which, in turn, can then inform strategies to reduce tobacco use. However, to our knowledge, there has been no recent systematic investigation of MHPs' attitudes to smoking cessation to identify possible barriers, if any.

5The current systematic review and meta‐analysis aims to synthesize the qualitative and quantitative literature on MHPs' attitudes towards smoking and smoking cessation among people with mental illnesses and/or substance abuse disorders. Specifically, it sought to (a) identify and estimate quantitatively MHPs' attitudes towards smoking and main barriers for providing smoking cessation support such as negative attitudes towards smoking cessation and permissive attitudes towards smoking, and (b) to explore these attitudes towards smoking, smoking cessation and support in‐depth through qualitative synthesis.

6The current review followed Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guidelines 25.

7The literature was searched for studies reporting on MHPs' beliefs and attitudes relating to smoking among people with mental illnesses and substance use disorders. An initial search strategy was developed in the databases Medline and Embase, and through the identification of key terms used commonly in the literature.

8The main keywords used in the search strategy were structured around three key concepts: ‘attitudes’, ‘mental health professionals’ and ‘smoking’, and customized to each database. The final search was conducted on 17 March 2015 in the following databases: AMED (OVID platform), CINAHL (EBSCO platform), classic + Embase, Embase, HMIC Health Management Information Consortium, Medline and PsycINFO (OVID platform) (see Supporting information, Table S1 for full search strategy and number of records identified). To find further eligible papers, forward and backward citation searches of included studies were also conducted.

9Studies were eligible for inclusion if they met the following criteria: The sample was health‐care professionals working in mental health, drug or alcohol treatment (referred to as MHPs throughout).The study measured attitudes/beliefs related to patients' smoking or the delivery of smoking cessation support/advice to patients.It was a primary quantitative or qualitative study published in a peer‐reviewed journal (only baseline data from longitudinal or experimental studies were eligible for inclusion to provide unbiased, background estimates).

10The sample was health‐care professionals working in mental health, drug or alcohol treatment (referred to as MHPs throughout).

11The study measured attitudes/beliefs related to patients' smoking or the delivery of smoking cessation support/advice to patients.

12It was a primary quantitative or qualitative study published in a peer‐reviewed journal (only baseline data from longitudinal or experimental studies were eligible for inclusion to provide unbiased, background estimates).

13Studies were excluded if they were not in the English language, focused only on attitudes towards smoke‐free policies, were based on child and adolescent services (the focus of this review was adult services), if the sample was composed only of health professionals in training or if they were published prior to 2003. This date limit was chosen as this was the year in which the World Health Organization Framework Convention of Tobacco Control (WHO FCTC) was published. The WHO FCTC states that parties to the convention should adopt or implement legislation that promotes training on tobacco control for health workers, community workers and social workers 26. As the FCTC was widely ratified and embraced, with 168 signatories, we judged that the date of its publication would serve as an appropriate marker for a shift in MHPs' attitudes towards tobacco and its treatment and capture the most accurate indication of current attitudes.

14An inclusion/exclusion criteria checklist was used to screen located articles for inclusion. To establish the reliability of the checklist, two independent reviewers (K.S. and I.T.) screened a random subset (34%) of papers, yielding excellent reliability (Cohen's kappa = 0.95). Disagreements were resolved through discussion.

15Studies were evaluated with the Mixed Methods Appraisal Tool (MMAT), which has good validity and reliability 27, 28, 29. For quantitative studies, the MMAT includes three subsections distinguishing between randomized controlled trials, non‐randomized comparative studies and descriptive studies. For the current review, the ‘descriptive’ subsection was used for the appraisal of all included quantitative studies (including where baseline data had been extracted from pre‐test–post‐test or longitudinal studies), as only baseline data and non‐comparative findings were extracted for the purpose of this review.

16Data were extracted by one researcher (K.S.) using a standardized data extraction form to provide consistency, reduce bias and increase validity and reliability 30.

17For quantitative studies, all measures of MHPs' attitudes or beliefs relating to smoking among patients or the delivery of smoking cessation support to patients were extracted. Measures of attitudes towards smoke‐free policies and attitudes reported by patients were not extracted. Extracted measures deemed to be reflective of the same beliefs or attitudes were grouped into categories by K.S. and L.S. These categories were organized further into higher‐level grouping on the basis of the types of attitudes or beliefs they represented. The proportion of participants reporting either negative attitudes towards providing smoking cessation support or permissive attitudes towards patients' smoking was extracted for measures within each category. Proportions were extracted systematically using predefined criteria (see Box 1). If relevant data were not included in published reports, authors were contacted to request data.

18Where more than one measure from a single study was included in a category, the mean proportion across measures was calculated prior to pooling. Pooling of proportions within each category and within the higher‐level groups was carried out in Stata version 13 using standard methodology 31. Proportions were first transformed using the Freeman–Tukey double arcsine method, pooled with a random effects meta‐analysis with the ‘metan’ command to account for heterogeneity, and then back‐transformed to proportions. Heterogeneity was assessed with the I 2 statistic 32. Only categories for which data were available from two or more studies were included in the data synthesis to allow for confidence in the accuracy of the findings.

19Additionally, a meta‐regression was conducted to examine whether study characteristics were predictive of study outcomes using the ‘metareg’ command in Stata. Following convention 33, only outcomes with data from at least 10 studies were analysed. Study‐level covariates included in the analysis, thought to potentially influence outcomes, were year of publication (pre‐2010/2010‐onwards, based on a median split), country (US/non‐US) and service type (mental health services/drug and alcohol services). All study‐level covariates were entered into the same model concurrently, and values were adjusted for multiple comparisons.

20All text contained in the ‘Results’ or ‘Findings’ section was extracted from published reports and analysed using thematic synthesis. First, textual data were coded line‐by‐line and were then categorized by similarity to generate descriptive themes occurring across included studies 34. Line‐by‐line coding and theme generation was carried out independently by two coders (K.S. and I.T.) and agreed in discussion to establish reliability. As some qualitative studies included interview questions focused on smoke‐free policies, and/or included data gathered from interviews with patients, lines in which findings were related explicitly to smoke‐free policy or attributed to patients were not included in the coding.

21Database searches yielded a total of 327 records after the removal of duplicates. After full‐text screening, 33 papers met inclusion criteria. Forward/backward citation searches yielded an additional eight papers, resulting in a total of 41 papers reporting on 38 studies being included in this review (Fig. 1).

22Of the included studies, 31 used a quantitative methodology, six were qualitative and one employed a mixed‐methods approach. The total number of participants across quantitative studies (including the mixed‐methods study) was 16 215. One qualitative study 35 did not report a sample size; across the remainder the total sample size was 154, resulting in a total sample of 16,369 participants across all included studies. The majority of studies recruited MHPs from a range of professional roles (including nurses, psychiatrists and clinical psychologists) were set in a mental health‐care context and conducted in the United States (see Table 1 for details).

23All studies included stated clear research questions or objective, and in all but one study (which did not provide the requisite information in the published report 36), data were judged to be sufficient to address the stated research questions or objectives (however, this study provided relevant for the current review and was therefore included here).

24Twenty‐two quantitative studies were judged to have used an adequate random (stratified or non‐stratified) sampling strategy. The remainder used convenience or unclear sampling. For most studies (n = 30) the sample representativeness could not be determined, either because the inclusion/exclusion criteria were unclear or because reasons for non‐response or differences between responders and non‐responders were unknown. Two studies were judged to have unrepresentative samples: in one study the sample was attendees at a smoking cessation training programme 3, and in the other the authors stated that their sample was not representative 38. The majority of studies (n = 28) used established measures which were clearly reflective of the variable of interest. For the remainder, the source and/or wording of measures was unclear. Fewer than half (n = 15) of the studies had a response rate of ≥ 60%.

25All qualitative studies used data sources (i.e. participants/recruitment settings) that were relevant to the research question, and used relevant approaches to data collection and analysis. Four of the seven qualitative studies discussed their findings in consideration of the research context, but only one explicitly considered the researchers' influence on their findings.

26Fourteen categories of attitudes were measured in at least five of the studies included and were therefore considered to be reliable and replicable. These categories were organized into three higher‐level groups: ‘perceived barriers to providing support’ (three categories), ‘negative attitudes towards smoking cessation’ (seven categories) and ‘permissive attitudes towards smoking’ (four categories).

27The most frequently measured categories were ‘lack of knowledge/training/skills in providing smoking cessation treatment’, ‘quitting smoking might have a negative impact on symptoms/recovery’, ‘smoking cessation is not a priority’, ‘patients are not interested in quitting’ and ‘lack of time to provide smoking cessation treatment’ (see Table 2 for all categories and example measures). For some categories, suitable data could not be extracted from at least two studies, and therefore these categories were not included in the quantitative analysis.

28Because there was significant heterogeneity across measures within all categories (all I 2 ≥ 94.7, all Ps < 0.001), data were pooled using a random‐effects meta‐analysis (see Table 3).

29Around four of 10 participants reported perceived barriers to offering smoking cessation intervention, with lack of knowledge or training being the most prevalent perceived barrier, followed by lack of time and low confidence.

30A similar proportion reported negative attitudes related to smoking cessation. The most commonly held beliefs were that patients are not interested in quitting smoking and that smoking cessation interventions are not effective. The least prevalent beliefs were that delivering smoking cessation intervention is not part of MHPs' role or that of mental health/drug and alcohol services.

31Nearly half of MHPs held permissive attitudes towards patients' smoking, with more than a third reporting the belief that quitting smoking is too much for patients to take on and that smoking with patients helps to establish a therapeutic relationship or build rapport.

32Meta‐regression was conducted on only a subsection of outcomes (see Table 2) and revealed no significant associations of study characteristics with proportions for higher‐level groupings. However, in the analysis of individual categories there was a significant association between study country and the proportion of respondents who endorsed the view that quitting smoking might have a negative impact on symptoms and recovery. Controlling for other study‐level covariates, MHPs in studies from the United States were less likely than those from other countries to be worried about the negative impact of smoking cessation on mental health, with an absolute difference in proportions of 30.3% [95% confidence interval (CI) = 3.1–69.6, P < 0.05].

33Seven studies were included in the qualitative synthesis. The thematic synthesis yielded five main/recurring themes: (1) beliefs about patients quitting smoking; (2) barriers to the provision of smoking cessation treatment; (3) attitudes to the provision of smoking cessation treatment; (4) acceptance of patients' smoking; and (5) smoking as a useful tool. Table 4 shows the subthemes identified within each of these alongside illustrative quotes.

34The most apparent subtheme within this theme was negative perceptions of patients' ability and motivation to quit, which was identified in all studies. This perception appeared to arise from beliefs that patients would not be interested in quitting smoking 39, 40, 41, 42, that patients would be unable to quit smoking successfully 35, 36, 40 and the view that smoking cessation would be too much for patients to cope with alongside their other issues 36, 39, 41, 42, 43. A minority of studies reported concerns about the negative consequences of quitting smoking, such as concerns about a negative impact on psychiatric symptoms/abstinence 36, 41, the potential effects on medication 36 and a fear that smoking cessation may make patients violent 36.

35A perceived lack of opportunity to provide smoking cessation support was evident 36, 39, 40, 41, 42, 43. This generally arose from a perceived lack of treatment resources either on‐site or to refer patients to off‐site 36, 40, 41, 42 and a lack of guidance for implementing smoking cessation treatment alongside treatment for patients' primary disorders 36, 41, 42, 43. In addition, a lack of staff 36, 41, staff time 36, 39, funding 36, 40 and support from management 36, 41 were identified as further barriers to providing smoking cessation treatment.

36In a number of studies, staff did not appear to feel that they had the capability to treat nicotine dependence effectively 36, 40, 41, 43. This arose both from a lack of knowledge about nicotine dependence and pharmacological treatments 36, 40, 41, 43 and from a perceived need for specialized training in the provision of smoking cessation support to people with mental illnesses and/or drug and alcohol disorders 40, 41, 43.

37The behaviour of MHPs was identified commonly as a barrier to addressing smoking among patients 35, 40, 41, 42. This tended to relate to professionals' smoking status, as staff smoking was viewed as providing a bad example to patients 35, 36, 40, 41, 42 and could potentially undermine motivation to address patients' smoking (e.g. because it would be hypocritical) 41. In addition, professionals' role in facilitating patients' smoking (i.e. by providing lights and/or cigarettes) was a further potential barrier to providing smoking cessation treatment 35, 36.

38Smoking cessation generally was viewed to be important for patients 36, 39, 40, 41, 42, 43, primarily because of the potential to improve patients' health 39, 41, 42, 43 and reduce their financial burden 39, 42. Two studies reported that staff felt smoking cessation treatment should be integrated into patients' care 39, 43. Despite this positive attitude to the provision of smoking cessation treatment, results from the majority of studies indicated that the provision of smoking cessation treatment was not viewed as a priority in the context of treatment for patients' primary disorders [36,39–43.] For instance, some staff thought that treating tobacco dependence was less important than treating addictions to alcohol or illicit drugs, as cigarettes are not illegal or as harmful to health 36, 42. In some cases, the provision of smoking cessation treatment was seen to be dependent upon the patient; for instance, if they showed an interest in quitting 39, 41, 42, 43 or displayed any negative health effects of smoking 39.

39The provision of smoking cessation treatment was often not seen as part of MHPs' role 36, 39, 41, 42, and in many cases staff held negative beliefs about smoking cessation treatment 36, 40, 41, 42. For instance, some reported the belief that smoking cessation treatment is not effective 40, 41, while others had concerns that encouraging smoking cessation might drive patients away 36, 41.

40Overall, a general acceptance of patients' smoking by MHPs could be noted. This seemed to stem from the view that there was a long‐standing ‘culture of smoking’ in in‐patient mental health and drug and alcohol services 35, 36, 41, alongside the belief that it is a patient's right/personal choice to smoke 36, 39, 41. Further, some staff felt that smoking was a ‘core need’ for their patients—something essential that they could not be without 35, 40, 43.

41Across studies, MHPs viewed cigarettes and smoking as useful tools in managing patients 35, 36, 40, 41, 42, 43. Patients were seen to use smoking as a form of self‐medication (e.g. to cope with abstinence from other addictive drugs or with psychiatric symptoms) or stress‐relief 35, 36, 40, 42, 43, to relieve boredom 35, 40 and facilitate socializing 40. Smoking was also viewed as a useful tool to be used by staff, either through smoking with patients to build rapport 35, 36, 40, 41, by using cigarettes as a behavioural incentive/reward 35, 36, 40 or using cigarettes to keep patients calm and minimize aggression 35.

42We identified 38 relevant studies investigating MHPs' attitudes towards smoking and smoking cessation among people with mental illnesses and substance abuse disorders, most of which were quantitative rather than qualitative. A significant proportion of MHPs held negative attitudes towards smoking cessation and permissive attitudes towards smoking, and perceived a number of barriers to providing smoking cessation treatment. Meta‐regression analyses revealed no associations between attitudes and year of publication or type of service, but showed that MHPs in US‐based studies were less likely to hold concerns about the negative impact of smoking cessation on symptoms and recovery. The thematic synthesis of qualitative studies was consistent with the quantitative results, and in addition revealed a richer insight into MHPs' views on patients' smoking, particularly in the context of in‐patient services. Cigarettes were viewed as a useful tool for both patients and staff, and the smoking behaviour of staff was identified as a barrier to providing effective smoking cessation support. A general acceptance of patients' smoking was also apparent in the qualitative findings, with a culture of smoking being seen as the ‘norm’ in some services.

43This synthesis of the existing literature examining MHPs' attitudes towards smoking and smoking cessation among patients reveals that a significant minority hold attitudes and beliefs that may prevent them from delivering smoking cessation support. The relationship between attitudes/beliefs, intentions and behaviour is well established (e.g. 44, 45), and health professionals are no exception 46. This review highlights the potential for such attitudes to present a barrier to integrate smoking cessation interventions effectively into treatment for mental illnesses and substance use disorders. In addition, this review identified a number of target misconceptions that are inconsistent with the scientific literature. For instance, the belief that patients are not interested in quitting is contradicted by evidence that people with mental illnesses are just as likely to want to quit as those without 22. Similarly, concerns about the negative impact of quitting on patients' symptoms or recovery are not supported by research 20, 21.

44It is encouraging that the current review identified that most MHPs felt that it was part of their role to address patients' smoking, even though a minority of approximately one‐fifth felt that it was not part of their role. Further, the qualitative findings indicated that most felt that smoking cessation was important, suggesting that MHPs view the provision of smoking cessation treatment positively. However, despite this generally positive attitude towards providing support, MHPs' reported beliefs about patients' ability and/or motivation to quit, and perceptions of their own ability to successfully deliver interventions and intervention effectiveness which may prevent them from providing support. Most notably, more than half of MHPs in this review reported that patients are not interested in quitting smoking and almost 40% felt that quitting smoking was too much for patients to take on, suggesting that one of the main barriers to the delivery of smoking cessation support may be implicit beliefs about patients, and a tendency to shift responsibility for smoking cessation to patients alone. In agreement with previous work 18, these findings highlight a need for greater prioritization of smoking cessation treatment in mental health and substance abuse care, a need for specialist training in smoking cessation interventions and broader education to challenge misconceptions about smoking cessation in the context of mental health and drug and alcohol treatment. One notable policy change recently instigated in the United Kingdom that may prove effective is the introduction of smoke‐free policies 47; other levers which could be pursued to address existing barriers include mandatory smoking cessation training for MHPs and, in the context of private health‐care systems, governmental subsidies for cost‐effective smoking cessation pharmacotherapy. In fact, National Institute for Health and Care Excellence (NICE) guidance on smoke‐free policy indicates that this policy should only be introduced alongside training and systematic identification of smokers and treatment, as introducing such policies by themselves is unlikely to change attitudes 11. In addition, the inclusion of effective smoking cessation medications on the hospitals' formulary will ensure that smokers with and without mental illness have equal access to pharmacotherapy to treat tobacco addiction.

45It is unclear why MHPs in the United States were less likely to be concerned about the negative impact of smoking cessation than those from other countries. This may reflect differences in the training provided to MHPs in different countries. However, clinical practice guidelines on smoking cessation from the United States, Australia and the United Kingdom (the two countries that formed the majority of studies not conducted in the United States) do not provide a clear explanation for this divide. While guideline recommendations in the United Kingdom do not cover the topic of smoking cessation, psychiatric symptoms and addiction recovery explicitly, they clearly encourage smoking cessation in this setting 11, and both US 48 and Australian 3 guidelines state that smoking cessation is unlikely to have any adverse effects. The lack of evidence of any recent changes in attitudes to support cessation is also of concern in light of accumulating evidence of the positive impact of cessation on mental health 20, 21, 22, and may indicate a need to disseminate this information to MHPs more effectively.

46The main strength of this review is that is provides an up‐to‐date overview of the most prevalent beliefs and attitudes that may present a barrier to the delivery of smoking cessation intervention in mental health and drug and alcohol services. This review also has some limitations. Data for pooled proportions were not available from all studies, and there was significant heterogeneity among studies. However, quantitative results were largely confirmed by the qualitative synthesis, increasing confidence in these findings. The study selection might be biased due to the focus on English language and full‐text publications; however, every effort was made to make the search strategy as inclusive as possible. None the less, relatively little qualitative work was identified, and further studies should explore potential solutions to overcome common barriers and enable the effective integration of stop smoking service provision into mental health care.

47In summary, a significant proportion of MHPs appear to hold negative attitudes towards offering smoking cessation advice and support to patients with mental illness and substance use disorders. These present a potential barrier to the successful implementation of smoking cessation interventions in mental health/substance abuse services, and highlight the continued need for dedicated education and training amongst this group of health professionals.

48L.S. has received a research grant and honoraria for a talk and travel expenses from a Pfizer, manufacturer of smoking cessation medications.