SEARCH WITHIN CONTENT
VOLUME 2018 , ISSUE 1 (March 2018) > List of articles
Citation Information : Evidence Base. VOLUME 2018 , ISSUE 1 , ISSN (Online) 1838-9422, DOI: 10.21307/eb-2018-001, March 2018 © 2018.© Australia and New Zealand School of Government and the authors
License : (CC-BY-NC-ND 4.0)
Published Online: 03-April-2018
Cigarette smoking costs Australia and New Zealand billions of dollars per year and is the single most preventable risk to health. Though governments have initiated numerous public health policies which have reduced the incidence of smoking, current usage remains around 15 percent. Making further inroads is likely to require augmenting these interventions with action at the individual level. The hospital setting provides a unique opportunity to assess the efficacy of individual attention. The aim of this study is to make an initial assessment of this efficacy by collating the existing evidence of outcomes achieved by health professionals working with individuals in hospital settings. The systematic literature search resulted in 69 studies (72 citations) for evaluation. Results indicated that a multicomponent intervention comprised of high intensity counselling with a minimum of one month of post-discharge follow-up in addition to either nicotine replacement therapy or varenicline tartrate is the most effective combination of individual treatments for improving smoking abstinence, particularly for general inpatients. Further, there was an indication that patients admitted to specialist wards (e.g. cardiovascular) would benefit most from high intensity interventions, regardless of the use of adjunct pharmacotherapy. The evidence for a positive effect on sustained quit smoking rates for peri-operative patients is not definite, but as smoking adversely affects surgical success, implementing multicomponent interventions should still be considered. This review found no clear evidence to support implementation of smoking cessation interventions in the emergency department setting. Overall, interventions throughout the review were heterogeneous, making the estimate of a true effect difficult. Furthermore, there were only low numbers of local studies, with the findings of this review relying mostly upon extrapolation from overseas studies. Given the severity of the burden placed on the health system by smoking, there is a need for continuing endeavours by researchers with the support of the government to identify innovative and effective interventions for smokers that can be delivered by health professionals caring for smokers in the hospital setting.
Cigarette smoke is composed of more than four thousand known chemical species with toxic and carcinogenic properties (Colombo et al. 2014). Smokers breathe this aerosolised gas and particulate cocktail into their bodies, even though it has been identified as the cause of a long list of diseases and cancers affecting the circulatory, respiratory, reproductive, nervous, urinary and digestive systems (Colombo et al. 2014). Cigarettes’ main psychoactive component, nicotine, has strong reinforcing and rewarding properties, and together with the psychological and environmental cues associated with smoking, makes cigarette smoking an extremely addictive behaviour (Balfour 2009; Laviolette and van der Kooy 2004).
Recent estimates indicate a decrease in the prevalence of smoking, from 29.1 percent of Australians and 25 percent of New Zealanders in the 1990s compared to the most recent figures of 12.8 percent and 16.6 percent respectively (Australian Institute of Health and Welfare 2008; Australian Institute of Health and Welfare 2014; Ministry of Health 2014; Ministry of Health 2015). Though these figures appear to be heading in the right direction there are still many who engage in a behaviour which is the greatest preventable cause of morbidity and mortality, accounting for approximately 15,500 and 5,000 Australian and New Zealander deaths per year respectively (Begg et al. 2007; New Zealand Government 2016).
In common with most other countries, Australia and New Zealand have implemented a range of public policy initiatives targeting cigarette use over the past decades. Key examples include prohibiting the advertisement and promotion of cigarette products, followed by mass media public education campaigns and labelling of cigarette packaging with health warnings, and most recently implementing plain packaging requirements (Cotter 2011; Grace 2016; Miller and Scollo 2016). Financial interventions and methods aimed at increasing quit rates have also been introduced in the form of subsidisation of pharmacotherapy, and price and taxation increases, with the cost of purchasing cigarettes in Australia now almost 10 times more expensive than in the 1990s (Purcell et al. 2012). Perhaps the strongest legislation is the ban on smoking in public places, including (but not limited to) workplaces, bars and restaurants, schools, hospitals and other healthcare buildings (New Zealand Drug Foundation 2011; Greenhalgh et al. 2016). These broad public policies and legislature initiatives to date have contributed to a drop in smoking rates (Woodruff et al. 1993; Glasgow et al. 1997; Farkas et al. 1999; Farrelly et al. 1999; Albers et al. 2007; Azagba and Sharaf 2013), and as such have demonstrated the viability of continuing to target smoking cessation on a societal level.
Policy interventions have enjoyed considerable success in reducing smoking. However, smoking reduction is not as great in absolute terms as the percentage decline might imply. That is, the decline in absolute number of smokers is now less, for instance due to population growth. Additionally, smoking rates are influenced by fewer young people taking up the habit, as opposed to reducing current smoker rates. These recalcitrant smokers are less receptive to changing their behaviour, even with the concerted effort of enforcing the tobacco related policies outlined above (Borland et al. 2012; Australian Institute of Health and Welfare 2014). Hence, additional interventions aimed at the level of individual smokers are required to support these larger scale initiatives. The need to engage the healthcare system in this effort has been outlined in a number of policy documents, including Australia’s National Tobacco Strategy 2012-18, Framework Convention on Tobacco and the National Preventative Health Strategy (World Health Organization 2005; National Preventative Health Taskforce 2008, Intergovernmental Committee on Drugs 2012). In broad terms these documents outline the need to improve uptake of existing infrastructure (e.g. Quitline), develop systems where health professionals engage with patients around this issue, and provide policy guidelines on brief interventions for health professionals to implement with smokers in their care (World Health Organization 2005; Greenhalgh et al. 2016). The National Tobacco Strategy 2012-18 (Intergovernmental Committee on Drugs 2012) specifies that as an industry we must:
Improve management of smoking cessation for all patients in health care facilities, particularly for patients on admission to hospital.
There are a number of evidence-based approaches that have been proven to work in the general smoking population, when attempting to address smoking cessation with patients. One can provide face-to-face counselling (Lancaster and Stead 2005) or can refer to a telephone counselling service (Stead et al. 2013), hereafter referred to as a Quitline. Alternatively, there are a number of smoking cessation medications that can assist during a quit attempt, including varenicline tartrate (varenicline) (Ebbert et al. 2010; Cahill et al. 2013), bupropion hydrochloride (bupropion) (Raupach and van Schayck 2011; Cahill et al. 2013) or nicotine replacement therapy (NRT) (Ferguson et al. 2011; Cahill et al. 2013). Combining pharmacotherapy and counselling, hereafter referred to as a multicomponent strategy, is an additional option, which has been shown to increase quit rates from 10 to 25 percent compared to pharmacotherapy alone (Stead and Lancaster 2012). There are also some alternative methods that often get mentioned when discussing smoking cessation, most notably hypnotherapy, for which conclusive evidence is currently lacking (Barnes et al. 2010).
Healthcare institutes’ practices and specific organisational policies in relation to smoking cessation are often underutilised, ad hoc, outdated and variable between institutions, departments, and even individual healthcare practitioners (Freund et al. 2008; Freund et al. 2009; Bartels et al. 2012; George et al. 2012; Ohakim et al. 2015). In the hospital setting, anecdotal and published evidence suggests that clinical practice guidelines for smoking cessation are sub-optimally translated into practice (Fiore et al. 2012; Freund et al. 2009; Regan et al. 2012; Slattery et al. 2016; Smith et al. 2012), despite a growing evidence base underpinning the recommendations. This is a substantial missed opportunity for effective intervention, with almost 300,000 hospitalisations per year in Australia attributable to cigarette smoking (Hurley 2006). Furthermore, hospitalisation presents a unique opportunity to apply individualised approaches as it:
provides a reflection period during an inpatient stay for smokers to reconsider lifestyle factors contributing to their illness/admission, a so-called ‘teachable moment’ (McBride et al. 2003);
makes it difficult (although not impossible) to smoke through enforced initiation of abstinence while in a hospital bed, as a result of widespread smoking bans in hospitals (Rigotti et al. 2000; Purcell et al. 2012), allowing patients to focus on the achievement that they have already quit by the time of discharge; and
facilitates the initiation of smoking cessation medication under supervision, which allows monitoring of nausea, craving and titration of medication to avoid adverse events.
Hence, in order to provide a current snapshot of effective quit smoking interventions in the hospital setting, the aim of this review is to evaluate the existing evidence for interventions delivered to smokers admitted or presenting to hospital, who are receiving treatment and advice from healthcare professionals.
The compiled evidence will be discussed with a view to underpin practical and achievable recommendations for policy improvement in this area. Given that smoking is a national and international priority area, there is an opportunity to make a significant impact on patient services for smoking cessation by making a change to current practice approaches.
A systematic literature search of the Medline, EMBASE, PsycINFO and The Cochrane Library databases was conducted in November 2016. Publications investigating smoking cessation policy within the hospital setting (for admitted patients and those attending the emergency department) were examined. We used the following free text search terms to identify relevant records: (smoking cessation) AND (pharmacotherapy OR drug therapy OR counselling OR hypnotherapy OR hypnosis OR aversive therapy OR psychotherapy OR smoke-free policy OR bans OR fines OR penalties OR motivation OR goals).
Grey literature were also searched to identify potentially relevant articles through the International Clinical Trials Registry Platform and ClinicalTrials.gov (searched November 2016), using the key words (smoking cessation) AND hospital. Reference lists of publications meeting all the inclusion criteria were also screened for potentially eligible studies.
In order to determine the most effective smoking cessation interventions delivered by healthcare professionals in the hospital setting, we reviewed evidence from randomised controlled trials with a minimum three-month follow-up. In the case of perioperative interventions, there was no restriction set at the follow-up period. Trial participants were current smokers admitted to a hospital ward or presenting to a hospital emergency department. Healthcare professionals delivering the intervention were defined as doctors, nurses, pharmacists, physiotherapists, and other clinical professionals providing care within the hospital setting. The intervention could also be delivered by researchers in instances where the procedure would, in practice, be provided by a healthcare professional.
Interventions included were:
Behavioural interventions: counselling, support groups, self-help, seminars, motivational lectures, web- and mobile phone-based interventions
Pharmacological interventions adjunct to counselling: NRT, bupropion and varenicline. These are generally accompanied by at least minimal counselling by health professionals. When minimal counselling was standardised across each study group, this was considered as isolation of the pharmacological intervention effect
Multicomponent interventions: pharmacological and non-pharmacological approaches combined as a package intervention
Comparison groups consisted of usual care, minimal intervention (such as a pamphlet or referral back to regular General Practitioner for follow-up) or co-intervention (where the control group receives some parts of the intervention but not all).
From the title, abstract or descriptors, two reviewers screened the retrieved citations to identify potentially relevant trials. Data for included studies were then extracted into standardised templates for trial characteristics and outcome variables. Studies not meeting all the inclusion criteria reported above for study design, subject description, and intervention/comparison group characteristics, were excluded from the narrative synthesis of results. In addition, a risk of bias assessment was conducted to assess the quality of the evidence base. This was undertaken using the report by Tooth et al. (2005) and standard Cochrane risk of bias grading criteria. Bias was categorised as ‘high risk of bias’ when a particular quality procedure did not occur, ‘unclear risk of bias’ when data were not reported in the article or when criteria were not relevant and ‘low risk of bias’ when data for a criterion were reported and adequately addressed in the study design. Review Manager Version 5.3 software was used to generate the risk of bias graphs. Outcome variables of interest were:
Seven-day point-prevalence smoking abstinence: this refers to whether, at a pre-defined time point, an individual has smoked cigarettes/tobacco in the previous seven days.
Continuous smoking abstinence: relates to cessation of smoking from the initial quit date to a pre-defined time point.
These outcomes are often measured by the participant’s self-report, and may be validated through biochemical measures, e.g. cotinine samples obtained via blood, urine or saliva, or carbon monoxide assessed through expired breath or blood specimen (Benowitz et al. 2002; Hughes et al. 2010).
For this review we will report seven-day point-prevalence in the first instance, and where this is not measured we will report 30-day continuous smoking abstinence. These self-reported measures are highly correlated and will lead to the same conclusion (Velicer and Prochaska 2004). Other outcomes which will be discussed in the review are: number and type of adverse events in case of pharmacological interventions; number and type of postoperative complications in case of interventions focusing on surgical patients; and costs associated with each intervention. While not direct smoking cessation measures, these outcomes are clinically relevant and should be taken under advisement when making recommendations for change of practice.
After duplicates were removed, 2768 citations from the electronic search were screened for eligibility. Of these, 62 citations were eligible for evaluation and a further 10 were identified through hand searching included articles and relevant systematic reviews. Tables 1, 2 and 3 are a summary of the characteristics and findings of the 69 studies (72 citations) included for evaluation, separated into the three main types of intervention.
Initially this document aimed to focus on Australia and New Zealand studies, with reference to international studies. However, the comprehensive literature search identified only seven studies undertaken in Australia and none from New Zealand. As such, all relevant studies have been included together for narrative synthesis; studies from North America and Europe being the main contributors to the current evidence base.
Methodological assessment was undertaken for all 72 citations and a visual summary is provided in Figure 1, while individual study quality is presented in Figure 2. Overall, study quality was determined to be average. There were many studies where a low or high risk of bias judgement was precluded due to poor reporting of study methods in the publication; this resulted in a high volume of unclear bias assessments. High risk of bias assessment was predominantly noted in the performance bias domain; this is likely attributable to difficulty blinding participants with behavioural interventions. For pharmacological interventions this was generally well done.
The results below are split for interventions targeting the general inpatient population and patients on cardiac wards. This decision was made because pharmacological interventions are often perceived to be associated with increased risk and caution for cardiac patients; furthermore, there is a substantial body of research targeting this specific patient group. Upon review of the evidence base, cancer patients were not placed in a dedicated subgroup, as it was hypothesised that they predominantly fall within the pre-operative or outpatient group depending on disease status.
In addition, it is important to note that the ‘intensity’ of behavioural interventions will be discussed below. This refers not necessarily to the quality of the intervention but to the composition of the intervention itself; for example, duration and number of contacts by health care professionals (longer and more equates to a higher intensity). Quality of these interventions is an important factor when considering the results, and may have had a bearing on the relative success or otherwise of the studies included for review. Aspects of the interventions which would facilitate a quality judgement (e.g. delivery of intervention by a specialist or a generalist; training of study personnel to ensure consistent intervention delivery) are reported variably, making it difficult for the authors to rate the intervention quality.
For reasons explained previously, high intensity behavioural interventions are recommended for inpatient smokers to successfully quit their habit. These recommendations are influenced by the results of a Cochrane systematic review and meta-analysis that concluded that high intensity behavioural interventions, being those that consist of a hospital counselling session and a follow-up for at least one month, lead to a significant improvement in quit rates (Rigotti et al. 2012). These recommendations are accurate when behavioural and multicomponent (counselling and access to free pharmacotherapy) interventions are put together, which was the case in the Rigotti et al. (2012) Cochrane review. However, when solely looking at behavioural counselling interventions, specifically split for admitting ward, results point to different conclusions: no clear evidence in favour of either low or high intensity behavioural interventions.
Three studies used lower intensity interventions, and unsurprisingly for such a small pool of evidence, conclusive evidence of effectiveness was lacking. While Meysman et al. (2010) found that a brief nurse-delivered stage-based intervention using trained nurses was better than a booklet in getting patients to quit, a study by Rigotti et al. (1997) failed to find a significant advantage with a 15-minute bedside counselling session combined with post-discharge counselling over usual care at 6-month follow-up. Similarly, Pederson et al. (1991) did not find any differences when comparing brief quit advice with more intensive counselling consisting of up to eight 15–20 minute sessions while participants were still hospitalised.
Interestingly, studies using higher intensity interventions also showed no significant differences in favour of intervention. De Azevedo et al. (2010) used tailored counselling and up to seven follow-up telephone calls, while Hennrikus et al. (2005) used counselling and up to six follow-up telephone calls, with neither trials showing higher smoking cessation rates compared to brief advice or usual care. Similarly, the study by Smith et al. (2011) did not find significant benefits from a high intensity intervention (bedside counselling, seven telephone follow-ups and minimal intervention package) over minimal intervention. These trials, with a combined sample size of 2228 participants, are only opposed by a small study of 77 participants, which found that an intensive 12-week nurse-delivered relapse management intervention, including eight telephone follow-up calls, showed a 42 percent quit rate compared to 15 percent in the usual care group (Caruthers et al. 2006).
All studies above used either telephone or face-to-face follow-up. Only one study (Harrington et al. 2016) used at a different mode of delivery – a web-based intervention. The intervention focused on quit smoking education and allowed for asynchronous communication with quit smoking counsellors. The authors found that at end of follow-up, results significantly favoured control, 18.5 percent versus 13.8 percent. These results need to be considered in the context of problematic treatment adherence, which may have negatively impacted the findings. Treatment adherence problems for web-based and mobile-based interventions in the general population are widely recognised, and can be attributed (among other reasons) to the inexperience of intervention designers in designing engaging solutions (Kelders et al. 2012). This problem will hopefully be solved as the eHealth and mHealth research field matures and interventions become more sophisticated.
Evidence for the effectiveness of bupropion prescribed in the general hospital setting is currently lacking; three trials used bupropion in a cardiac population (see the cardiac section) and two trials targeted pre-operative patients (see pre-operative section). One trial assessed its use in the general population (Simon et al. 2009). The study found no significant improvement in quit rates with bupropion. Interestingly, however, it reported non-significant higher rates of abstinence among the placebo group (31 percent) compared with the bupropion group (15 percent) at six-month follow-up.
The sole study looking at the effect of NRT versus placebo in the hospital setting did not yield significantly higher quit rates when both groups were provided as an adjunct to intensive counselling with five face-to-face follow-up sessions (Campbell et al. 1991). All other NRT studies were either performed in the cardiac or pre-operative setting, or were provided as a multicomponent intervention.
The two studies that tested the use of varenicline in the hospital setting found conflicting results. A well-powered Australian study found that varenicline adjunct to counselling by Quitline significantly increased quit rates to 31.1 percent compared to the 21.4 percent of smoke-free patients treated by Quitline alone (Smith et al. 2013). Alternatively, a much smaller pilot study by Steinberg et al. (2011) found no significant difference between varenicline and placebo adjunct to low intensity counselling. In this trial, only just over half of the participants treated with varenicline were compliant, but those patients who were compliant with the medication showed higher quit rates when treated with varenicline (80 percent versus 56 percent); unfortunately, the study was not sufficiently powered to detect this kind of difference.
Multicomponent interventions for inpatient smoking cessation are widely encouraged by best-practice guidelines (Fiore et al. 2000; West et al. 2000; Global Initiative for Chronic Obstructive Lung Disease 2015; Yang et al. 2016). Most studies found in this review evaluated a bedside counselling program combined with NRT and post-discharge telephone and/or outpatient cessation support. Similar to behavioural interventions, counselling can be broken up into high and low intensity. For low intensity interventions, three studies were found that tested the effectiveness of low intensity counselling in addition to provision of NRT, none of which found a significant effect on quit rates (Molyneux et al. 2003; Nagle et al. 2005; Thomas et al. 2016).
The results of eight studies investigating high intensity multicomponent interventions, on the other hand, showed favourable smoking cessation rates. The largest of these studies (Miller et al. 1997), comprising 2024 patients, found that an intervention consisting of counselling, access to an educational video, NRT and one follow-up telephone call was not demonstrably superior to usual care for smoking cessation. However, after increasing the behavioural counselling intensity, by adding three extra follow-up telephone calls and provision of extra face-to-face counselling in the event of relapse, smoking cessation significantly improved by 7 percent over usual care. Similarly, Simon et al. (1997; 2003) conducted two studies evaluating a multicomponent intervention comprised of individual counselling, educational video, provision of three months of NRT, printed resources and five follow-up telephone calls. In the first instance it was compared to a brief pre-discharge counselling session and printed resources and though not significant at six months, abstinence favoured the intervention group by 8 percent, and at 12 months this further improved to 14 percent and became a statistically significant difference (Simon et al. 1997). In the second evaluation, the multicomponent intervention was compared to two months of NRT and a brief counselling session (Simon et al. 2003). Once again results favoured intervention at six months (35 percent vs 21 percent) and 12 months (33 percent vs 20 percent); both differences were statistically significant. This is the first time an identical intervention of this type has been evaluated with reproducible results.
High intensity individually oriented multicomponent interventions may be a resource-intensive exercise, and as such several studies have researched different intervention formats, specifically referring to outpatient services, the use of group-formats, automated telephone counselling, and the use of computers to deliver the interventions.
Simply referring patients to outpatient counselling is not sufficient to maintain high quit rates. Sherman et al. (2016) found that providing patients post-discharge (up to 42 days) counselling via telephone in addition to eight weeks of NRT out-performed a simple referral to Quitline for proactive counselling by almost 10 percent at two months and 5 percent at six months for 30-day point prevalence abstinence.
Delivering inpatient counselling within a group setting is a promising alternative to individual bedside counselling. Borglykke et al. (2008) tested group counselling combined with standard cessation information and provision of NRT compared to standard cessation information alone. Results favoured the multicomponent strategy, where cessation at one-year follow-up was 17 percent superior to control.
Two recent studies investigated the use of innovative technology aimed at automating follow-up counselling and triaging smokers who needed additional human-delivered counselling. Rigotti et al (2014), found significant results when using an automated interactive voice response system to provide follow-up counselling, in addition to free NRT, with superior quit rates of 11 percent higher in the intervention group. They also performed a cost analysis that determined the costs per patient for the intervention were US$354 for the first 12 months and US$108 for subsequent years (Rigotti et al. 2014). Fellows et al. (2016) similarly tested an innovative voice recognition intervention in addition to assisted outpatient referrals and a multicomponent inpatient intervention, however, they reported no significant difference in quit smoking rates. Uptake of NRT in the Fellows et al. (2016) trial was considerably lower, which might explain the difference in results. These results indicate that simple voice recognition counselling on its own is not sufficient, but that it proves a promising future research area when tested as a complement to a multicomponent intervention and free NRT.
Only one study assessed the effectiveness of a computer delivered intervention in addition to other familiar components (NRT, print resource, individual counselling and follow-up telephone support) (Prochaska et al. 2014). This multicomponent intervention proved successful compared to usual care, improving smoking abstinence by approximately 10 percent at three-month follow-up. This was sustained at 18-month follow-up with 20 percent of participants reporting smoking abstinence in the intervention group compared to just 7.7 percent in the control. Interestingly, the intervention group also demonstrated decreased risk of hospital readmission. When viewed in combination with the promising results of another web/computer-based intervention, discussed above (Harrington et al. 2016), this is likely a priority area for further cessation research given its potential for the easy implementation of a standardised program with long-term results.
There is insufficient evidence to suggest that the general inpatient population benefits from brief counselling, referral to professional counselling services after discharge or even high intensity behavioural interventions that include telephone follow-up counselling. Furthermore, evidence for the use of pharmacotherapy on its own is inconsistent. The only intervention type that was significantly more effective in increasing quit rates was the use of multicomponent interventions. Specifically, the biggest evidence based interventions involve combining provision of (free) NRT with intensive counselling. Using innovative approaches (automated telephone counselling and computer-delivered interventions) to complement or substitute human-delivered counselling looks promising, but the current evidence base is not strong enough to recommend widespread adoption.
Three studies evaluated the effectiveness of low intensity behavioural interventions in patients with a cardiac admission, failing to find sufficient evidence for effectiveness. Neither Bolman et al. (2002) nor Hajek et al. (2002) found superior quit rates for low intensity interventions consisting of brief bedside counselling and other behavioural components (e.g. declaration to commit to quitting). Adding a single one-week follow-up call to an extensive cognitive behavioural intervention was not sufficient in inducing higher quit rates compared to usual care (Rigotti et al. 1994). Quit rates for cardiac patients were, however, higher as compared to the general inpatient setting, despite not being higher than usual care, which does indicate a higher susceptibility to quit smoking for this population.
This notion is further supported by results from studies using higher intensity behavioural interventions. In contrast to the general inpatient population, cardiac patients show significantly higher quit rates of between 39–70 percent as found by five studies. Bedside counselling and education complemented with telephone follow-ups were effective in improving quit rates, with one study finding these results were sustained up to five years later (Chouinard and Robichaud-Ekstrand, 2005; Dornelas et al. 2000; Feeney et al. 2001; Ockene et al. 1992; Smith and Burgess 2009).
Using a group format as opposed to individual counselling, or targeting smoking cessation as part of a larger cardiac care improvement interventions, are also promising. A bi-weekly group smoking cessation session, combined with telephone follow-up after discharge, resulted in abstinence up to 12 months in 50 percent of patients treated in the group sessions versus 37 percent of patients given no further instruction on how to quit (Quist-Paulsen and Gallefoss 2003). Quit rates of up to 70 percent were found when evaluating a multicomponent intervention focusing on improving case-management of cardiac patients (DeBusk et al. 1994). The intervention assessed smoking and nutritional counselling, the use of lipid-lowering drug therapy and exercise training, and included 12 nurse-initiated telephone follow-ups and 12 patient visits to either the nurse case manager or the laboratory for blood testing.
Four trials investigated the differences between pharmacotherapy targeting smoking cessation, and placebo as a complement to low intensity counselling. Review of the existing literature demonstrates that there is no definitive evidence to support adding bupropion to counselling for cardiac patients, as two trials failed to find a significant benefit in this population (Eisenberg et al. 2013; Planer et al. 2011). A third trial by Rigotti et al. (2006) only found a significant short-term difference favouring bupropion when looking at a subset of treatment compliant participants; this result was not maintained at one-year follow-up. A further study investigated the use of varenicline as a complement to low intensity counselling, which saw significantly higher rates of non-smokers at the end of the trial – 47.3 percent versus 32.5 percent (Eisenberg et al. 2016).
Multicomponent interventions for cardiac inpatients were largely led by nurses and included a higher intensity behavioural component. Interventions overall point to increased cessation rates of at least 10 percent up to a period of 3–6 months, but results were in general not sustained at 12 months. Froelicher et al. (2004) found higher point prevalent smoking cessation rates of 10.7 percent at 6 months and overall higher continuous smoking abstinence (p=0.04) for their multicomponent intervention versus brief counselling. Reid et al. (2003) similarly found 11 percent higher quit smoking rates at 6 months, for an intervention that provided extra follow-up care for those patients who had relapsed. These results were replicated when they tested a similar intervention but used voice response technology to take over the follow-up counselling (Reid et al. 2007).
These results were countered on the one hand by a study with significantly higher quit smoking rates, and on the other hand a study that failed to find results. Taylor et al. (1990) found that a multicomponent intervention over usual care almost doubled smoking cessation rates (61.6 percent vs 32 percent). Conversely, a recent three-arm study (Berndt et al. 2017) comparing usual care to follow-up counselling performed by telephone or follow-up counselling done face-to-face did not find significant between-group differences. However, when the latter split the groups into low socioeconomic status and high socioeconomic status, telephone counselling and face-to-face counselling outperformed usual care.
High intensity behavioural as well as multicomponent interventions are effective in increasing quit smoking rates for patients with cardiac conditions. While behavioural interventions were heterogeneous in their intervention designs, making it difficult to nominate the exact program which would be most effective, it can be said that any inpatient smoking cessation intervention for cardiac patients should include a behavioural component extending into the post-discharge period. Where possible and appropriate, the provision of free quit smoking medication could be considered as results show overall favourable short to medium term results.
There is no clear evidence to support the use of behavioural smoking cessation interventions in the emergency department. The only trial available failed to find a difference between two-minute generic advice and a more intensive intervention, including a self-help workbook and three follow-up telephone calls, when targeting smoking youth aged 14–19 years old. Quit rates were 2.5 percent for the intervention and 2.9 percent for control (Horn et al. 2007).
There appeared to be no difference between provision of a multicomponent intervention with either low or high intensity behavioural counselling in the emergency department setting. Three trials studied multicomponent interventions including low intensity behavioural support. Bernstein et al. (2013) and Bernstein et al. (2015) tested brief counselling, six weeks of NRT and referral to Quitline (for general smokers) or a single follow-up telephone call (for substance abusers). Though these results were promising at the three-month follow-up, they were not sustained at 12 months. Richman et al. (2000) failed to find even short-term positive results for a similar program compared to an educational pamphlet, with quit rates of 10.9 percent versus 10.4 percent.
Only one emergency department initiated a multi-faceted intervention including a higher-intensity behavioural component, consisting of four follow-up telephone calls (Neuner et al. 2009). Contrary to results of higher-intensity interventions in the general and cardiac inpatient settings, this study demonstrated similar results between groups when compared to usual care: 14.2 percent versus 11.3 percent respectively.
Lack of successful outcomes in the emergency department, even for multi-faceted interventions including a higher-intensity behavioural component, is inconsistent with results seen in other hospital settings. Given the inundation of patients in the emergency department and the need to triage patients according to urgency, as well as costs associated with intervention delivery, this may not be the most suitable opportunity for dedicated smoking cessation interventions.
Providing a personalised letter from a consultant in combination with nurse quit smoking advice and a referral to a stop smoking service improved quit rates pre-operation (18 percent), as opposed to providing a general quit smoking booklet and nurse advice (8 percent) (Andrews et al. 2006). The only other peri-operative behavioural intervention did not detect a difference between brief advice and brief advice plus a carbon monoxide check on the day of surgery (Shi et al. 2013).
Only two pharmacological studies were identified for the peri-operative setting. Myles et al. (2004) did not find bupropion to be more effective than placebo prior to elective surgery in terms of overall abstinence rates at hospital admission and six-month follow-up (this trial started two months before surgery was scheduled and encompassed two face-to-face counselling sessions, as well as weekly telephone follow-up). Alternatively, a 12-week course of varenicline as part of a perioperative intervention including standardised counselling identified a significant 10 percent improvement in smoking abstinence with varenicline compared to placebo which was sustained up to 12 months (Wong et al. 2012).
Multicomponent interventions that included higher-intensity behavioural support found favourable quit rates for the peri-operative setting overall. Lindström et al. (2008) and Lee et al. (2013) used similar interventions consisting of weekly long counselling sessions, referral to Quitline and free NRT, with both trials finding significantly higher quit rates. The Lindstrom trial found that 33 percent versus 15 percent of smokers had quit at 12-month follow-up, and the Lee trial found 30 percent versus 11 percent had quit at three-month follow-up, for intervention compared to control. A study comparing peri-operative nurse-led telephone, computer and in-person support in conjunction with NRT, found that in the two weeks pre-surgery, abstinence was significantly higher than control (89 percent vs 13 percent), which was maintained in the post-surgery period (92 percent vs 50 percent) (Sørensen and Jørgensen 2003). Furthermore, a multi-modal intervention delivered in person, by telephone and computer as well as free NRT resulted in 73 percent pre-surgical abstinence compared to 56 percent in the usual care group. Though these results diminished after the peri-operative period, the intervention group still out-performed the control group (18 percent vs 5 percent) (Wolfenden et al. 2005).
These favourable results were opposed by two studies that failed to find a significant difference between intervention and control. Ratner et al. (2004) found 20 percent more abstinence within 24 hours of commencing the intervention, but failed to find significant differences beyond the post-operative period. Referral to Quitline in combination with NRT and clinician-driven motivation to use Quitline did not lead to significant differences in abstinence rates at one or three-month follow-up (Warner et al. 2011).
Finally, two studies employed a weaning/scheduled quit program to encourage peri-operative smoking cessation. Adding a scheduled quit date and cigarette weaning program to best practice care (five nurse-led bedside and telephone counselling and NRT) was ineffective; the control group who received best practice care alone demonstrated similar three and six-month abstinence prevalence (approximately 30 percent) (Ostroff et al. 2014). Moller et al. (2002) provided NRT and a consultation session with a nurse, where the participant received a personalised nicotine substitution schedule and was encouraged to quit smoking entirely, or at least reduce consumption by 50 percent. Though this study did not record smoking abstinence outcomes, it demonstrated a significant improvement in post-surgical complications, particularly wound healing, as well as, a non-significant improvement in the need for second surgery and cardiovascular complications, compared to the control group (Møller et al. 2002).
The evidence for long-term abstinence when targeting peri-operative patients is not clear-cut. Overall, high intensity multicomponent interventions seem to point to sustained quit smoking rates compared to usual care, but more research is still needed. Referring to Quitline and placing patients on a weaning schedule, however, did not lead to significant longer-term quit rates. Overall, higher quit rates were demonstrated in this population as compared to the general inpatient population. While control often comprised basic cessation advice, it also resulted in an increase in abstinence from baseline. This effect is likely attributable to the ‘hard deadline’ and gravity of impending surgery followed by enforced smoke-free environments during post-operative hospital stays. Although results are not definitive, implementing multicomponent interventions should be considered, due to the adverse effects smoking has on the success of surgical procedures and related potential post-operative complications.
Out of the 13 pharmacological studies, 11 reported information pertaining to adverse events as part of their investigation. Five studies used bupropion, four used varenicline and four used NRT in forms of gum, lozenges and/or patches. Among the five studies reporting on bupropion, only one reported a significant increase in adverse events between the intervention and comparator population, finding dizziness to be more common among bupropion participants compared to placebo (14 percent vs 1.4 percent; p = 0.005) (Planer et al. 2011). Other common adverse events reported among all five studies and across both arms (intervention and placebo) include insomnia, dry mouth, vomiting, sleep disturbance and re-hospitalisation. An increased risk of suicidal thoughts, changes in thinking and behaviour and worsening of depression are known to occur with antidepressants including bupropion. However, none of the five studies evaluated found this to be significantly worse among the bupropion participants.
Among the four studies reporting on varenicline, three found a statistically significant increase in nausea and two reported an increase in abnormal dreams, both among varenicline users. Nausea occurred in 13.9–25 percent of participants on varenicline compared to 1.5–8.6 percent of comparator participants. Other common side effects include insomnia, headache and irritability. A black box warning did exist for varenicline related to possible development of serious neuropsychiatric symptoms including suicidal thoughts, hostility and agitation. However, in December 2016 the US Food and Drug Administration approved dropping the warning due to recent evidence from the EAGLES publication (Evaluating Adverse Events in a Global Smoking Cessation Study) (Anthenelli et al. 2016). This study compared the safety of varenicline, bupropion, NRT and placebo in approximately 8000 smokers with and without psychiatric disorders, finding that varenicline did not increase neuropsychiatric events among those with a history of psychiatric disorders. Although the black box warning has been lifted, the labelling still states that post-marketing studies have reported serious or clinically significant neuropsychiatric adverse events. None of the four studies evaluating varenicline in this review found any significant increase in neuropsychiatric events, cardiovascular events or serious skin reactions between groups.
Among the four NRT studies only two reported information about adverse events and none of these were found to be significantly different between groups. The most common side effects were gastrointestinal such as nausea and vomiting. Of the 35 multicomponent interventions, only six studies reported that an adverse event had occurred. The most common side effects with use of transdermal nicotine patch were skin reactions such as site irritation and erythema, which resolved with discontinuation of the patch. Other common side effects included nausea, headache, sleep disturbances/nightmares and dizziness. However, among studies using a placebo patch as a comparator there was no significant difference between groups in relation to these outcomes (Lewis et al. 1998; Molyneux et al. 2003; Sørensen and Jørgensen 2003). No adverse events were mentioned specifically for other types of nicotine replacement therapy such as gums, lozenges, inhalers, and mists.
When considering smoking cessation interventions for the inpatient setting, health institutes need to consider intervention and patient type, as patients admitting to hospital for different reasons may have variable responses to different types of interventions. Patients admitted to a cardiac ward, as opposed to a general ward, showed higher quit rates when they were offered face-to-face counselling, which needs to be followed by either face-to-face or telephone follow-up for at least one month. A simple one-off counselling session, even if it is extensive, is not sufficient to improve cessation rates. The behavioural counselling program can be supplemented with NRT, but it is not necessary as these patients respond to the intervention with or without NRT. This is welcome news, as treatment compliance to smoking cessation pharmacotherapy and enthusiasm for its use in general is often sub-optimal (Ferguson et al. 2011).
When looking at the general inpatient setting, participants do not show higher quit rates when treated solely with a behavioural intervention, regardless of whether there is intensive follow-up or not. This patient group requires the provision of (free) NRT in order for smoking cessation interventions to become effective. Alternatively, patients can be prescribed varenicline, as this medication has been shown to be effective and safe in the hospital population, as well as the general population (Anthenelli et al. 2016; Sterling et al. 2016).
Smoking cessation counselling interventions can successfully be embedded within already existing models of care, specifically by training ward nurses to counsel and provide follow-up. This is a similar conclusion as that found by the Cochrane review on nursing interventions for smoking cessation (Rice et al. 2013). Health institutes should consider making smoking cessation, and counselling on other health related factors, such as alcohol use and exercise, an official part of the health professional’s role. They should ensure that the health professional has enough resources, time and training to ensure they are capable to perform these important duties.
There was no evidence to support quit smoking interventions in the emergency department. While smoking cessation should be discussed as part of the ‘teachable moment’ that is hospitalisation (McBride et al. 2003), health professionals should carefully consider whether the patient is likely to benefit from a quit smoking intervention (based on admitting reason, access to pharmacotherapy, and feasibility of running a high intensity behavioural intervention), or whether it would be more fruitful to address other health problems or behaviours that would be more likely to change. Similarly, healthcare institutions and individual professionals should weigh the costs and available resources of focusing on smoking cessation in this setting in light of the current lack of evidence.
Though there has been substantial research in the field of interventions for inpatient smoking cessation, a definitive conclusion on the exact elements of the most effective approach is lacking. This is largely attributed to heterogeneous interventions and a large proportion of neutral results. Overall, we may infer that interventions including both high intensity behavioural counselling, and either NRT or varenicline have been most effective.
Previously, diagnosis of a chronic condition has been associated with increased smoking cessation. More specifically, patients with cardiac and cancerous disease demonstrate a greater desire and aptitude for quitting (Hermanson et al. 1988; McWhorter et al. 1990; Novotny et al. 1990; Salive et al. 1992). This is consistent with the findings of this review, where interventions were most successful when implemented in cardiac as opposed to general medical, surgical and emergency units. While it may be that health professionals in specialist units (e.g. cardiovascular) place greater emphasis on the benefits of smoking cessation than general medical units (Emmons and Goldstein 1992), future research exploring patients’ attitudes toward the importance of quitting in relation to their type of disease may provide the required insight to close the gap.
Surprisingly, there was only limited evidence for more innovative interventions, with only four studies testing the effectiveness of a computer or web-based solution (Wolfenden et al. 2005; Ostroff et al. 2014; Prochaska et al. 2014; Harrington et al. 2016), and no studies testing the use of smartphone apps. While these studies did not favour an electronic intervention as such, issues with adherence were identified. As this field of research evolves, further understanding of effective implementation strategies could unlock the potential of these interventions. As the burden on our health professionals and the demand for cost-effective care increases, developing, testing and implementing innovative tools to track patient progress on vital health behaviour such as smoking cessation becomes increasingly important. Preliminary results from the general population are promising in terms of feasibility, acceptability and smoking abstinence (Bricker et al. 2014; Ubhi et al. 2015). As this technology becomes ingrained in society’s daily life, it presents the greatest opportunity for innovative research.
The current evidence base lacks data for the cost-effectiveness of these hospital-based interventions. A high intensity behavioural intervention in combination with provision of NRT was estimated to cost US$354 in the first year of implementation and US$108 for subsequent years (Rigotti et al. 2014). However, when in-person counselling was replaced with a batch of telephone sessions in combination with NRT, there was an estimated cost of approximately US$75 per patient (Sherman et al. 2016). While this is by no means sufficient evidence of cost-effectiveness to change policy and practice, it does not exclude the hospital as a viable setting for intervention, suggesting that using existing models of care and infrastructure can keep costs low.
The socioeconomic burden of smoking on the health system, a great deal of which can be attributed to hospital use, is disproportionate to the research effort from Australia and New Zealand. Though smoking cessation is a national priority area and policies recommend that hospital and health care professional intervention is necessary, just seven of the included studies were conducted in Australia and none in New Zealand. Of these, only three demonstrated beneficial effects in favour of intervention; a multicomponent computer based intervention (Wolfenden et al. 2005), inpatient prescription of varenicline and bedside facilitation of Quitline contact (Smith et al. 2013), and high intensity behavioural counselling with an extended period of follow-up telephone calls (Feeney et al. 2001). Translational clinical research can be a lengthy and costly process, and within the hospital system this is often investigator-led with costs coming from already restricted operating budgets. Redirecting some governmental funds into a research grant scheme could and should target two important areas for bridging the knowledge-translation gap that currently exists. The first initiative would be a commitment to the development of an evidence-based hospital smoking cessation program, which also takes into account strategies that support the ongoing provision of this intervention beyond the study period. This should be accompanied by a methodologically rigorous and reproducible evaluation, with a commitment to regular review for currency. Secondly, a grant scheme could invite more interest in continued innovation in this area. Despite the existence of a large body of evidence, hospital cessation research needs to continue to evolve and grow to incorporate such emerging approaches as eHealth and mHealth interventions, which could prove invaluable to the cessation effort.
Overall, this review has provided an evidence base predominantly from overseas, which indicates the merits of ward-based intervention for smoking cessation and provision of post-discharge follow-up by health professionals. Though results are not overwhelmingly favourable, given the current state of socioeconomic burden caused by smoking, perhaps something is better than nothing. However, it is apparent from this review that local and innovative (web and smartphone-based) research is lacking, and would benefit from a more concerted effort by health researchers with support from the government.