Systematic Review of Psychosocial Smoking Cessation Interventions for People with Serious Mental Illness

Abstract

Objective

Tobacco smoking is a major driver of premature mortality in people with serious mental illness (SMI; e.g., schizophrenia, bipolar disorder). This systematic literature review described randomized control trials of psychosocial smoking cessation interventions for people with SMI, rated their methodological rigor, evaluated the inclusion of racial/ethnic and sexual/gender minorities, and examined smoking cessation outcomes.

Methods

Eligible studies included peer-reviewed articles published between 2009 and 2020 that examined psychosocial smoking cessation interventions in people with SMI. We used the Preferred Reporting Items for Systematic Review and Meta-analysis guidelines to conduct our review and the Methodological Quality Rating Scale to evaluate methodological rigor.

Results

Eighteen studies were included. Ten were categorized as high methodological rigor given their study characteristics (e.g., longer follow-up) and eight as lower methodological rigor based on their characteristics (e.g., not intent-to-treat). Racial/ethnic and sexual/gender minorities were underrepresented in these studies. A range of psychosocial interventions were examined including motivational enhancements, smoking cessation education, cognitive behavioral strategies, and contingency management. Most studies also provided smoking cessation medications (e.g., NRT, bupropion), although provision was not always uniform across treatment conditions. Nine studies found the intervention condition achieved significantly higher smoking abstinence or reduction compared to the comparison group.

Conclusions

The most promising psychosocial smoking cessation interventions seemed to be initiated in inpatient psychiatric units and employed either a transtheoretical approach (i.e., stage-tailored, decisional balance) or individually-tailored telephone behavioral smoking counseling using a motivational-interviewing framework. Clinical trial designs (e.g., SMART, factorial) that control for the provision of psychosocial medications and allow for the identification of optimal psychosocial treatments are needed. Future studies should also ensure greater inclusion of racial/ethnic and sexual/gender minorities and should be culturally/linguistically adapted to improve treatment engagement and study outcomes.

1. Introduction

Tobacco smoking is a major driver of premature mortality and chronic disease in people with serious mental illness (SMI; e.g., schizophrenia, bipolar disorder). The prevalence of current smoking among people with SMI is two to four times higher than in the general population with estimates ranging from 40% to 80% depending on the sample and psychiatric diagnosis (Callaghan et al., 2014; Dickerson et al., 2018). Smokers with SMI tend to be heavy smokers (i.e., smoke ≥25 cigarettes per day) and to extract more nicotine with each cigarette than those without SMI, further exacerbating the negative health effects (Evins et al., 2015). Smoking among people with SMI is associated with lower quality of life, poorer mental health, worse cognitive functioning, and worse functional outcomes (e.g., financial and personal care skills) compared to nonsmokers with SMI (Depp et al., 2015; Dickerson et al., 2013). Additionally, studies of the general population and of people with SMI indicate that tobacco smoking is also associated with increased risk of suicide, although quitting seems to mitigate this risk (Das and Prochaska, 2017).

Many factors contribute to the high prevalence of smoking among people with SMI. Historically, smoking cessation treatments have not been offered to people with SMI because behavioral health staff viewed smoking as a secondary concern compared to more immediate problems, believed that smoking provides therapeutic benefits, and that quitting might cause psychiatric symptoms to worsen (Das and Prochaska, 2017; Schroeder and Morris, 2010). Further compounding this problem, the tobacco industry targeted this population by providing free cigarettes in behavioral health settings, funding research designed to push the self-medication hypothesis (i.e., cigarettes reduce psychiatric symptoms), and lobbying to block hospital smoking bans (Das and Prochaska, 2017; Prochaska et al., 2008; Schroeder and Morris, 2010). Additionally, there is evidence that tobacco use screening by psychiatrists declined from 1993 to 2010 (Rogers and Sherman, 2014). This is concerning because quitting without appropriate supports may be more difficult for people with SMI, since evidence suggests that tobacco smoke normalizes cognitive deficits and temporarily lowers blood levels of antipsychotic medications, thus reducing medication side-effects (Das and Prochaska, 2017; Schroeder and Morris, 2010; Sharma et al., 2016).

Despite these challenges and high rates of smoking, people with SMI who smoke want to reduce or quit smoking at similar rates as smokers without SMI (Aschbrenner et al., 2015; Ashton et al., 2013; Hall and Prochaska, 2009; SAMHSA, 2019). Current smoking cessation guidelines for people with SMI advise that both pharmacological and psychosocial smoking cessation treatments should be provided, and a combination of these approaches is more effective than either of these treatments alone (SAMHSA, 2019). Pharmacological treatments include varenicline, bupropion, and nicotine replacement therapy (NRT, e.g., patch, gum) (Das and Prochaska, 2017). Psychosocial treatment approaches include cognitive behavioral strategies, motivational interviewing (MI), acceptance and commitment therapy (ACT), contingency management/reinforcement, the 5A’s (i.e., Ask, Advise, Assess, Assist, Arrange), brief motivational and educational interventions, quitlines, and technology driven interventions (e.g., web-based, smart phone apps) (Das and Prochaska, 2017). Importantly, many of these psychosocial intervention approaches are not discrete; for instance, quitlines may incorporate cognitive behavioral strategies and motivational interviewing.

Several reviews of randomized clinical trials of smoking cessation interventions for people with SMI indicate that smoking cessation mediations (e.g., varenicline, bupropion, NRT) can help people with SMI achieve smoking abstinence (Evins et al., 2015; Pearsall et al., 2019; Peckham et al., 2017; Roberts et al., 2016), however, the evidence regarding psychosocial interventions is less clear. For instance, the 2009 Schizophrenia Patient Outcomes Research Team (PORT) reviewed pharmacological and psychosocial cessation interventions and found evidence to support the use of psychosocial interventions in conjunction with pharmacological treatments, but there was not sufficient data to determine the most effective type or key components of behavioral interventions (Buchanan et al., 2010). Similarly, Evins et al. (2015) found support for the use of psychosocial interventions in combination with cessation medications, but there was not enough evidence to determine the optimal duration and format of these psychosocial interventions. Bennett et al. (2013) reported that psychosocial interventions could help people with schizophrenia reduce or quit smoking in the short term (i.e., end of treatment), but evidence of long-term change (i.e., abstinence at post intervention follow-up) was lacking.

Two reviews published more recently also evaluated psychosocial interventions along with pharmacological interventions, and both reviews found inconsistent findings regarding the beneficial effects of psychosocial smoking cessation interventions on smoking abstinence outcomes (Pearsall et al., 2019; Peckham et al., 2017). Both reviews had mental health diagnosis inclusion criteria that was more stringent (e.g., 100% of participants required to have psychotic disorder or bipolar disorder) than the current review, and both reviews primarily evaluated abstinence from smoking without focusing on other cessation outcomes (e.g., reductions in cigarettes per day [CPD], nicotine dependence, quit attempts) (Pearsall et al., 2019; Peckham et al., 2017). How smoking cessation interventions effect these other outcomes could prove significant since some evidence suggests that reductions in smoking may lead to quit attempts and predict future abstinence (Begh et al., 2015; Evins et al., 2004; Hughes and Carpenter, 2006), and in some studies a reduction in smoking has accompanied a reduction in psychopathology (Gelkopf et al., 2012). Although reducing cigarette consumption may convey benefits to future quit attempts and cessation, it should be noted that there is no conclusive evidence that smoking reduction has a direct health benefit. To date, most reviews of psychosocial smoking cessation interventions for people with SMI have each reviewed a smaller number of studies (≤11) than the current review and have not systematically evaluated the inclusion of racial/ethnic minorities and sexual/gender minorities, which is a critical area for addressing health disparities and for ensuring these interventions work for the diversity of people served in community mental health systems.

Understanding the type of psychosocial interventions available and determining which show the most promise for helping people with SMI quit or reduce their smoking is important for several reasons. First, some cessation medications (e.g. varenicline) may be contraindicated due to comorbid medical conditions and some may not want to take an additional medications to quit smoking (Bennett et al., 2015). Second, identifying the most promising interventions gives community mental health agencies the information they need to provide evidence-based cessation treatments (Bennett et al., 2013). Lastly, while pharmacotherapy is an important component of smoking session treatment, current guidelines state that pharmacotherapy should be offered in conjunction with psychosocial interventions, but the most effective types of psychosocial interventions have not been determined.

1.1. Objectives

To help fill these gaps, we reviewed randomized clinical trials published since the publication of the 2009 Schizophrenia PORT recommendation, which did not find enough evidence to recommend specific psychosocial interventions or to delineate the specific treatment components that were most effective (Buchanan et al., 2010). The aims of this systematic review are to: 1) describe the components of psychosocial smoking cessation interventions for people with SMI, 2) rate the methodological rigor of these intervention studies, 3) evaluate the inclusion of racial/ethnic minorities and sexual/gender minorities in these studies, and 4) examine four smoking cessation intervention outcomes, including abstinence from smoking, reductions in smoking, decreases in nicotine dependence, and quit attempts.

2. Methods

2.1. Data Sources and Search Methods

This review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher, 2009). To locate articles written in English and published in peer-reviewed journals from January 2009 to February 2020, we searched the following electronic bibliographic databases: PsychINFO, Scopus, Medline, and CINAHL. The search strategy included terms for SMI (e.g., schizophrenia, bipolar disorder), smoking cessation (e.g., quit smoking, smoking abstinence), psychosocial interventions (e.g., cognitive behavioral therapy, motivational interviewing) and intervention design (e.g., randomized controlled trials). Manual searches were also conducted of the reference sections of existing reviews and eligible articles to check for additional relevant studies.

2.2. Study selection

Articles were included if they met the following criteria: 1) enrolled people with schizophrenia or bipolar disorder, or if the author(s) described participants as having SMI; 2) were randomized trials testing the impact of a psychosocial smoking cessation intervention among people with SMI who smoke cigarettes; and 3) reported measures of smoking abstinence or reduction as a primary outcome. Any measure of smoking abstinence or reduction was accepted. Measures of smoking abstinence included: 7-day point prevalence abstinence (PPA, i.e., self-report no smoking in the past seven days), biologically confirmed (e.g., expired carbon monoxide [CO], cotinine levels) 7-day PPA, and CO-confirmed recent abstinence. Measures of smoking reduction included: number of cigarettes per day (CPD), CO levels, and cotinine levels. While we were primarily interested in smoking abstinence and reduction outcomes, we also evaluated quit attempts and nicotine dependence outcomes if they were included. To evaluate articles’ eligibility for inclusion, two of the authors reviewed the identified articles’ abstracts and full-text and any disagreements on eligibility were resolved via the third author. Figure 1 presents our study flow chart. The initial search identified 1,817 articles. After removing duplicates, we screened 1,235 articles for eligibility and excluded 1,214. Twenty-one articles from 18 unique studies were included in our review.

Figure 1.

Study Flow Chart

2.3. Analytical strategy

A standardized data abstraction form was used to systematically code study characteristics, including study aims, sites, intervention characteristics (e.g., format, duration), sample characteristics (e.g., race/ethnicity, psychiatric diagnosis), smoking cessation outcome measures, findings, and limitations. Data were abstracted by one reviewer and then verified by a second reviewer. We rated the methodological quality of the studies across thirteen dimensions (e.g., study design, length of follow-up, dosage, multi-site) using an adapted version of the Methodological Quality Rating Scale (MQRS) (Miller et al., 1995). The adapted MQRS has been successfully used in prior systematic reviews (Cabassa et al., 2017, 2010; Cabassa and Hansen, 2007). Cumulative MQRS ratings for each study could range from 0 (low) to 17 (high). Similar to prior studies, the mean MQRS score (11.8) of studies included in our review was used to categorize studies into relative lower and higher methodological rigor (Vaughn and Howard, 2004). Studies with MQRS scores < 11.8 were considered to have relatively low methodological rigor and studies with a MQRS score ≥ 11.8 were considered to have higher methodological rigor. Two authors worked independently to rate each study with the adapted MQRS. The percentage of agreement between the two coders across all MQRS dimensions was 93%. Differences in ratings were resolved by consulting with the third author.

3. Results

3.1. Study characteristics

The characteristics of the 18 studies included in our review are displayed in Table 1. Study samples ranged from 20 to 754 participants (Median 150). The most common psychiatric disorders were schizophrenia spectrum disorders and bipolar disorder. The average age of participants ranged from 37.6 to 56.7 (median 45.1) and the percentage of female participants included in studies ranged from 0% to 64%. The average number of cigarettes smoked per day (CPD) at baseline ranged from 15.2 to 28.6 (median 21.0). Thirteen studies were conducted in the United States (U.S.), two in Australia, two in the United Kingdom (U.K.), and one in Israel. Studies took place in a variety of settings, including community mental health centers (CMHC), primary care clinics, outpatient mental health clinics, and research centers. Four studies were initiated in inpatient psychiatry units and continued in outpatient settings when participants were discharged, and one study was conducted in an inpatient psychiatric hospital for the entirety of the intervention among participants who were hospitalized for at least one year.

Table 1.

Characteristics of Included Psychosocial Smoking Cessation Intervention Studies

Author(s) and Year	SS	Age (M)	Female (%)	Race / Ethnicity (%)b	Psychiatric Diagnosis	CPD (M)	Setting / Location	Smoking Cessation Outcomes	SE	Total MQRS Score
Baker et al., 2015; 2018	235	41.6	41.0	84.0% Australian born	59.0% Schizophrenia spectrum 22.0% Bipolar disorder 19.0% Nonorganic psychotic syndrome	28.6	Three sites: Local research center and community clinics / Australia	CO confirmed 7-day PPA CPD 50% reduction in CPD Expired CO Nicotine dependence	156	13
Bennett et al., 2015	178	54.8	10.7	70.8% Black 22.5% White 6.7% Other	43.3 % Schizophrenia spectrum 27.5 % Bipolar disorder 21.9% Posttraumatic stress disorder 5.1 % Depression with psychotic features 2.2% Psychosis not otherwise specified	15.2	3 outpatient VA medical centers / USA	CO confirmed 7-day PPA CPD Quit attempts Nicotine dependence	12	12
Brody et al., 2017	34	56.7	0	53.0% Black 35.0% White 12.0% Asian	100.0% Schizophrenia	19.0	VA health center / USA	CO confirmed 7-day PPA* CPD* Expired CO Nicotine dependence	26	9
Brunette, Pratt et al., 2018	661	45.0	64.0	93.0% White	22.0% Schizophrenia spectrum 23.0% Bipolar disorder 24.0% Major depression 31.0% Anxiety and other disorders	17.3	Ten community mental health centers / USA	CO and urine cotinine confirmed 7-day PPA	52	12
Christiansen et al., 2018	222	44.0	48.3	26.4% Black 7.2% Hispanic	51.2% Schizophrenia 54.6% Bipolar disorder 45.4% Major affective disorder 14.2% Alcoholism	21.5	Twelve community support programs / USA	CO confirmed 7-day PPA* CPD Quit attempts*	13	10
Dixon et al., 2009	304	44.3	47.7	50.3% White 45.1% African American 4.6% Other	75.7% Schizophrenia 24.3% Affective and other psychoses	19.2	Six outpatient mental health clinics / USA	CO confirmed recent abstinence Cigarettes smoked per week Self-reported past 7-day smoking status	52	13
Gelkopf et al., 2012	53	46.3	28.6	Not reported	100.0% Schizophrenia	21.0	Inpatient psychiatric hospital / Israel	CPD*	13	10
Gilbody et al., 2015 Peckham et al., 2015	97	46.8	40.2	90.6% White 6.2% Black 2.0% Multiracial 1.0% Asian	59.0% Schizophrenia and other psychotic illness 10.0% Schizoaffective disorder 31.0% Bipolar disorder	24.8	Three sites: either primary care or specialist mental health services / England	CO confirmed 7-day PPA CPD Nicotine dependence	52	13
Gilbody et al., 2019 Peckham et al., 2019	526	46.0	41.1	89.7% White 3.6% Multiracial 3.4% Asian 3.0% Black	50.0% Schizophrenia 22.0 % Bipolar disorder 13.0% Schizoaffective disorder 15.0% Other psychotic disorder	24.0	Twenty-two sites: either primary care or specialist mental health services / England	CO confirmed 7-day PPA CPD Nicotine dependence	52	13
Heffner et al., 2020	51	49.0	45.0	24.0% Racial or ethnic minority	88.0% Bipolar I disorder 12.0% Bipolar II disorder	19.1	4 outpatient clinics / USA	CO confirmed 7-day PPA 50% reduction in CO level	14	12
Hickman III et al., 2015	100	39.5	35.0	44.0% Caucasian 27.0% Black 9.0% Hispanic 11.0% Asian American / Pacific Islander 9.0% Multiracial / Native American / Other	52.0 % Unipolar depression 29.0% Psychotic disorder 15.0% Bipolar disorder 4.0% No diagnosis reported	19.3	Initiated on inpatient psychiatric unit; continued as outpatient / USA	CO confirmed 7-day PPA	52	16
Japuntich et al., 2020	20	52.1	55.0	60.0% Caucasian	Specific breakdown not reported	22.6	Health home in a community mental health center / USA	Self-report 7-day PPA CPD* Quit attempts	13	10
Metse et al., 2017	754	38.7	38.8	13.6% Aboriginal or Torres Strait Islander	26.7% Mood disorders 23.0% Substance-related disorders 19.5% Schizophrenia and related psychosis 17.8% Personality and other disorders 13.0% Anxiety and stress-related disorders	NR	Initiated in four inpatient psychiatric facilities; continued as outpatient / Australia	CO confirmed 7-day PPA Prolonged abstinence (30 day or greater) CPD* 50% reduction in CPD* Quit attempts* Nicotine dependence	52	13
Morris et al., 2011	123	43.0	58.5	81.0% White 8% Black 7.9% Latino/a or Hispanic or Latin American 2.5% Native American 2.0% Asian 1.0% Other	35.0% Other 31.5% Bipolar disorder 27.0% Schizophrenia 18.5% Depression	20.4	Four community mental health clinics in rural and urban areas / USA	CO verified cessation 50% reduction in CPD* Nicotine dependence	26	11
Prochaska et al., 2014	224	40.0	37.5	65.2% White 11.2% Multiracial/other 8.9% African American 8.5% Asian American/Pacific Islander 6.3% Hispanic / Latino	46.9% Unipolar depression 25% Bipolar depression 15.2% Schizophrenia spectrum disorders 12.9% Other	19.0	Initiated on inpatient psychiatric unit; continued as outpatient / USA	CO confirmed 7-day PPA*	78	14
Stockings et al., 2014	205	37.6	46.3	Not reported	33.7% Schizophrenia and related psychosis 21.5% Substance-related disorders 18.5% Unipolar depressive disorders 14.1% Bipolar disorders 4.9% Anxiety and stress-related disorders 2.9% Personality disorders 4.4% Other	23.0	Initiated on inpatient psychiatric unit; continued as outpatient / USA	CO confirmed 7-day PPA Continuous abstinence 50% reduction in CPD* Quit attempts* Nicotine dependence*	26	11
Tidey et al., 2011	52	45.1	29.0	75.2% White 15.3% African American 9.6% Other 7.6% Hispanic	100.0% Schizophrenia or schizoaffective disorder	27.1	Outpatient research center / USA	CPD CO level* Cotinine level*	4	11
Williams et al., 2010	87	45.2	36.8	65.5% White 21.9% African American 3.4% Hispanic 5.8% Other 3.5% Asian	56.3% Schizophrenia 43.7% Schizoaffective disorder	22.5	Outpatient mental health facilities across two behavioral health care systems / USA	CO confirmed 7-day PPA CO confirmed continuous abstinence CPD CO levels Quit attempt	52	10

Note. SS = Sample Size; M = Mean

^aRacial/ethnic categories were taken from the original manuscript

CPD = Cigarettes per day; SE = study endpoint (weeks); MQRS = Methodological Quality Rating Scale; CO = Carbon monoxide; PPA = point prevalence abstinence; NR = Not reported; Nicotine dependence was measured with the Fagerstrom Test for Nicotine Dependence

^*= Indicates significant between group differences favoring the primary intervention condition.

The most common smoking cessation outcomes measured in the included studies were abstinence (88.9%), reduction in smoking (83.3%), nicotine dependence (44.4%), and quit attempts (33.3%). Of the 16 studies that measured abstinence from smoking, 15 used expired carbon monoxide (CO) to confirm abstinence (primarily CO-confirmed 7-day PPA) and one used self-report only. One study measured urinary cotinine levels along with expired CO to confirm 7-day PPA. Of the 15 studies measuring reductions in smoking, 14 measured CPD, five measured expired CO, and one used urinary cotinine levels to measure reduction. Four studies measured smoking reduction as the percentage of participants who achieved a 50% reduction in CPD, and one study measured 50% reduction in expired CO. Eight studies measured changes in nicotine dependence, all using the Fagerström Test for Nicotine Dependence (FTND), and six studies measured quit attempts using a variety of self-report measures (e.g., The Smoking History Form, Timeline Follow-Back method). Study endpoints ranged from 4 to 156 weeks (median 26 weeks).

3.2. Methodological Quality Assessment

Study MQRS scores (see Table 1) ranged from 9 to 16 with a mean score of 11.8 (sd = 1.7) and a median score of 12.0. All studies employed randomized controlled designs, contained sufficient details for replicability, reported baseline scores, and had standardized intervention manuals or procedures. Using the mean split (11.8) of the MQRS total score to categorize studies into relative low or high methodological rigor, eight studies were categorized as having low methodological rigor and ten as high methodological rigor.

In the low methodological rigor studies, follow-up lengths were as follows: 50% less than six months, 37.5% 6-to-11 months, and 12.5% 12-months or longer. Fifty percent had participant follow-up rates of less than 70%, 12.5% of studies had follow-up rates of 70%−84.9%, and 37.5% of studies had follow-up rates of 85%−100%. None of the low methodological rigor studies used collateral verification of participants’ self-reported smoking status, 25% did not use objective verification (e.g., CO feedback) of participants’ self-reported smoking status, 37.5% did not enumerate study dropouts, 62.5% did not conducted follow-up assessments with a person blinded to participants’ treatment condition, 25% of studies did not conduct intent-to-treat analysis, and 62.5% recruited from a single site.

In the high methodological rigor studies, 20% had follow-up lengths less than six months and 80% were 12-months or longer. Thirty percent had participant follow-up rates of less than 70%, 60% had follow-up rates of 70%−84.9%, and 10% had follow-up rates of 85%−100%. Twenty-percent of the high methodological rigor studies used collateral verification of participants’ self-reported smoking status, all studies used objective verification of participants’ self-reported smoking status and enumerated study dropout, 40% did not conduct follow-up assessments with a person blinded to participants’ treatment condition, all studies conducted intent-to-treat analysis, and 20% conducted recruitment from a single site with the remaining studies recruiting from multiple sites.

3.3. Inclusion of racial/ethnic minorities, sexual/gender minorities, and cultural and linguistic adaptations

Fifteen studies (83.3%) reported the percentage of racial/ethnic minorities included in their sample. The total sample size of these 15 studies was 3,433, with 24.12% (N=828) of participants being racial/ethnic minorities. Of these 15 studies, 12 (nine conducted in the U.S., two in the U.K., one in Australia) disaggregated and reported the racial/ethnic characteristics of their sample. Among the nine U.S.-based studies (N=1,324 participants), 738 (55.74%) participants were white, 424 (32.02%) African American, 57 (4.31%) Hispanics, 39 (2.95%) Asian, 3 (0.23%) Native American, and 70 (5.29%) other race/ethnicity. Among the two U.K. studies (N=623), 559 (89.73%) participants were white, 22 (3.53%) Black, 19 (3.05%) Asian, and 21 (3.37%) were other racial/ethnic minorities. The Australian study (N=754) reported that 13.6% of participants were Aboriginal or Torres Strait Islanders. Only three studies reported the percentage of participants who identified as sexual or gender minorities. Heffner et al. (2020) and Hickman et al. (2015) reported that twenty percent of participants were sexual or gender minorities, and 2.7% of participants in Prochaska et al. (2014) were transgender.

Only one study reported making cultural adaptations to their intervention. Hickman et al. (2015) adapted the intervention computer program and manual to include more ethnically diverse and urban images and references the tobacco industry’s targeting of Black smokers. All studies required participants to speak English and offered no linguistic adaptations.

3.4. Description of Psychosocial Interventions

Table 2 summarizes the interventions reviewed in this study. Studies ranged from 22-day contingency management interventions to multifaceted healthy lifestyle interventions lasting nine months. Thirteen (72.22%) interventions were delivered via individual sessions only, one (5.56%) used only group sessions, and four (22.22%) used a combination of individual and group sessions. Ten (55.56%) interventions were delivered in-person, one (5.56%) via telephone, one (5.6%) was technology-based (e.g., computer-based), and six (33.33%) used a combination of in-person and telephone/technology-based delivery approaches. Interventions were provided by an array of interventionists, including mental health clinicians (e.g., psychologist, counselors), advanced practice nurses, telephone counselors, and research staff. Christiansen et al. (2018) was the only study to use certified peer specialists (i.e., state certified individuals in recovery from mental illness). Ten interventions (55.56%) provided NRT to all participants, while five studies (27.78%) provided education on smoking cessation medications and then offered to help participants obtain medications from prescribers if the participants desired. Two studies (11.11%) provided bupropion as part of the intervention condition.

Table 2.

Descriptions of the Included Psychosocial Smoking Cessation Interventions

Author(s)	Intervention Length	Format	Intervention facilitator	NOS	Smoking Cessation Medications	Intervention Description	Key outcomes
							QR	CPD	FTND	QA
							%	Mean	Mean	%
Baker et al., 2015; 2018	9 months	In-person / Individual	Psychologists	17	NRT	Condition 1: Healthy lifestyle intervention using MI to increase readiness to change tobacco use and other CVD risk factors, cognitive behavioral strategies to build smoking cessation skills, and contingency reinforcement to encourage initiation and maintenance of change. Education provided information on NRT and interactions between nicotine and symptomatology. Consisted of an initial 90-minute session followed by 16 one-hour sessions: sessions 2–8 weekly, sessions 9–11 fortnightly, sessions 12–17 monthly. Received 24-week supply of NRT.	11	9.7	N/A	N/A
	9 months	Telephone / Individual	Psychologists	17	NRT	Condition 2: Compared to the in-person healthy lifestyle intervention, the telephone condition was provided in an abbreviated format and without cognitive behavioral therapy and contingency reinforcement. It consisted of an initial 90-minute face-to-face session, 14 10-minute telephone sessions, and two additional 30-minutes face-to-face sessions at weeks 4 and 8 which allowed for the provision of NRT and biomedical measures to be taken.	8	7.9	N/A	N/A
Bennett et al., 2015	12 weeks	In-person / Individual + group	Master’s level clinicians	25	Not provided	Condition 1: The behavioral condition offer two weeks of engagement -focused outreach (i.e., reminder calls, transportation assistance) and provided education on smoking cessation medications and assistance/facilitation with obtaining medications from their prescriber. An initial individual session discussed how smoking negatively affected the participants and their motivations for wanting to quit. The 24 twice-weekly group meetings began with a CO breathe test. Contingency rewards (maximum of $3.50) were offered for negative tests. Group sessions discussed goal setting and taught smoking cessation skills.	14.1	8	2.3	73.4
	12 weeks	In-person / Group	Master’s level clinicians	24	Not provided	Condition 2: The supportive condition also offered two weeks of engagement-focused outreach and basic education on smoking cessation medications. The extensive medication support provided in the behavioral condition was not provided in the supportive condition. The 24 twice-weekly meetings provided support for quitting and education on a range of smoking cessation topics. Meetings started with participants taking a CO breath test, but contingency rewards were not offered.	9.2	7.4	2	71.9
Brody et al., 2017	6 months	In-person / Individual + groups	Psychologists	26	Bupropion and NRT	Condition 1: Combination extended treatment plus home visits (COMB- EXT+HV) provided rapid initiation of a combination of three smoking cessation medications (bupropion, nicotine patch, nicotine lozenge), CBT, and HV. 15-minute medication management visits and one-hour CBT groups took place weekly for 6-months. CBT sessions focused on education, skill building (e.g., coping skills, avoiding triggers). Expelled CO was measured at each session. HV were designed to reduced secondhand smoke (SHS) exposure by looking for visible signs of smoking activities and providing brief behavior counseling to minimize and avoid SHS exposure and smoking triggers.	45	16.1	4.2	N/A
	6 months	In-person / Individual + groups	Psychologists	26	Bupropion and NRT	Condition 2: COMB-EXT condition provided the same treatment as the COMB- EXT+HV, but without the HV.	20	14	3	N/A
	6 months	In-person / Individual + groups	Psychologists	26	NRT, bupropion, or varenicline	Condition 3: Treatment as usual provided weekly CBT group sessions and medication management visits, and a single first line smoking cessation medication (nicotine patch, bupropion, or varenicline) and did not provide HV.	8	7.4	2.2	N/A
Brunette, Pratt et al., 2018	NR	In-person + Telephone / Individual	Psychiatrist Advanced practice nurses Bachelor’s level staff CBT therapist	13	Received prescriber visit to discuss medication;obtained medications if desired.	All conditions first received the motivational Let’s Talk about Smoking intervention described in Brunette et al. 2018. Condition 1: Participants received a prescriber (PV) with their CMHC prescriber who was trained in the 5A’s and cessation pharmacotherapy for people with SMI. Twelve telephone counseling sessions were provided with a trained CBT therapist. Telephone sessions were an adaptation of the Freedom from Smoking Program for people with SMI. Contingency rewards: Within each condition, half of the participants were randomized to receive contingency rewards (up to $450) for verified abstinence during a four-week cessation attempt.	NRa	N/A	N/A	N/A
	NR	In-person + Telephone / Individual	Psychiatrist Advanced practice nurses Bachelor’s level staff	4	Received prescriber visit and obtained medications if desired.	Condition 2: Participants received a PV as describe in condition 1, plus facilitated Quitline counseling through the New Hampshire Tobacco Helpline. The Helpline provided up to three manualized counseling sessions designed to help people quit smoking. Contingency rewards as described in condition 1.	NRa	N/A	N/A	N/A
	NR	In-person / Individual	Psychiatrist, Advanced practice nurse, Bachelor’s level staff	1	Received prescriber visit and obtained medications if desired.	Condition 3: Participants received a PV as described in condition 1. Contingency rewards as described in condition 1.	NRa	N/A	N/A	N/A
Christiansen et al., 2018	4 weeks	In-person / Individual	Certified peer specialist & RA’s	4 + 4 OS	NRT	Condition 1: Participants received four weekly 20–30 minute sessions, which consisted of a motivational activity followed by a behavioral component designed to prepare the participant for quitting. NRT was provided. Session 1: reviewed quit attempts, lessons learned, setting smoking reduction goals. Session 2: Decisional balance worksheet, planning a practice quit attempt. Session 3: imagined themselves smoke-free at future timepoints and combined activities from previous 2 sessions. Session 4: Reviewed previous three sessions. At the end of session four peer researchers explained the Wisconsin Telephone Quitline (WTQL) and offered to assist participants in making the call. At the end of session 4, participants in all conditions were offered the option of four additional sessions.	8.5	8.5	N/A	50
	4 weeks	In-person / Individual	Certified peer specialist & RA’s	4 + 4 OS	NRT	Condition 2: Attention control participants received 4 weekly 20–30 minute sessions where peer specialist reviewed booklets on the harms of smoking. NRT was provided for the four weeks during the intervention. Participants were also provided a referral to the WTQL.	1	5	N/A	21.4
Dixon et al., 2009	NR	In-person / Individual	Psychiatrists	NR	Medication not provided but could be prescribed during 5A’s.	Condition 1: 5A’s: The intervention implemented the 5A’s in six outpatient mental health clinics and aimed to have physicians implement the 5A’s at each clinic visit. The intervention promoted the 5A’s in the clinics and created a standardized set of chart notes to document implementation.	3.9	0.27	N/A	N/A
					See condition 1	Condition 2: Delayed intervention condition	1.6	1.2	N/A	N/A
Gelkopf et al., 2012	5 weeks	In-person / Group	Hospital staff trained by clinical psychologist	5	Not provided	Condition 1: Based on the transtheoretical model and consisted of five weekly one-hour smoking cessation groups. Education was provided regarding the risks and harms of smoking; behavioral change and maintenance techniques were taught. Groups provided a mechanism to discuss coping with anxiety and urges to smoke. Contingency rewards (fruit, stars on the board) were awarded to participants who significantly reduced their smoking. Each department had group facilitators devoted to reducing smoking by supporting participants throughout the week.	N/A	7.2	N/A	N/A
					Not provided	Condition 2: Delayed intervention condition	N/A	1.99	N/A	N/A
Gilbody et al., 2015 Peckham et al., 2015	NR	In-person / Individual + group	Mental health professional trained in smoking cessation interventions (MHSCP)	NR	Advised of options and provided by general practitioner when deemed necessary.	Condition 1: The bespoke smoking cessation intervention was individually tailored to participants’ needs and was delivered by a mental health practitioner in conjunction with the participant’s general practitioner (GP). The participant’s GP provided smoking cessation medications when deemed necessary. The intervention provided evidence-based behavior change techniques (e.g., measuring CO, relapse prevention) and home visits. Participants also received usual care.	26	6.4	0.7	N/A
	NR	N/A	NR	NR	Could be provided as part of usual care services.	Condition 2: Usual care participants received advice on accessing local NHS and family doctor smoking cessation services. They also received information on NHS Quitline services and continued to receive their usual services from their mental health provider.	16	4.9	1.2	N/A
Gilbody et al., 2019 Peckham et al., 2019	NR	In-person / Individual	Mental health professional trained in smoking cessation interventions (MHSCP)	12	Advised of options and provided by general practitioner when deemed necessary.	Condition 1: The bespoke smoking cessation intervention was individually tailored to participants’ needs and was delivered by a mental health practitioner trained in smoking cessation interventions (MHSCP) in conjunction with the participant’s general practitioner (GP). The MHSCP worked with the GP to facilitate the provision and monitoring of smoking cessation medications if the GP deemed medication necessary. The initial session was one-hour and remaining sessions were 30-minutes. Sessions began weekly but shifted to fortnightly if the participant preferred. Sessions took place in the participant’s home or another mutually agreed upon location and provided evidence-based behavior change techniques (e.g., measuring CO, relapse prevention). Participants also received usual care.	13	4.5	0.9	N/A
	NR	N/A	NR	NR	Could be provided as part of usual care services.	Condition 2: Usual care participants received advice on accessing local National Institute of Health (NHS) and family doctor smoking cessation services. They also received information on NHS Quitline services and continued to receive their usual services from their mental health provider.	8	4.5	1.1	N/A
Heffner et al., 2020	10 weeks	Web-based / Individual	N/A	NR	NRT	Condition 1: WebQuit Plus is a web-based Acceptance and Commitment Therapy (ACT) intervention for smokers with bipolar disorder. Participants received 8 weeks of NRT. The program has four parts related to making a quit plan, smoking triggers, coping skills, and engaging personal values to support long term abstinence. Participants receive text messages promoting the use of NRT and can use keyword text messaging to receive on demand assistance. Users are prompted to track smoking, use of cessation medications, and to practice ACT skills. The web program tracks the amount of money and minutes of life saved based on user reports of reduced or quitting smoking.	8	N/A	N/A	N/A
	10 weeks	Web-based / Individual	N/A	NR	NRT	Condition 2: The comparison condition used the National Cancer Institute’s website, Smokefree.gov. Smokefree.gov provided a multimedia format and offered guidance on quitting, interactive content, and information on the health effects of smoking. Participants received 8 weeks of NRT and email and text messages promoting engagement.	8	N/A	N/A	N/A
Hickman III et al., 2015	6 months	In-person + computer / Individual	Study Staff	3	NRT	Condition 1: The Transtheoretical Model (TTM)-tailored computer-assisted intervention was initiated on inpatient psychiatric units and provided one 15–30 minute on unit counseling session covering decisional balance, managing temptations, and motivational enhancement. The computer-assisted intervention provided a printed report that directed users to relevant sections in a staged-matched treatment manual (6th grade reading level). The computer intervention was repeated at the 3- and 6-month post-hospitalization follow-ups. Ten weeks of NRT was provided post-hospitalization.	26.2	N/A	N/A	N/A
	NR	In-person / individual	Study staff	NR	NRT	Condition 2: Enhanced usual care participants received a tobacco use assessment, NRT to manage withdrawals while hospitalized, advice to quit smoking, a quit smoking pamphlet, and referrals to smoking cessation services upon discharge.	16.7	N/A	N/A	N/A
Japuntich et al., 2020	3 months	Telephone-based / Individual	Clinical psychologist	9	Advised and helped facilitate receipt of medication if participant desired.	Condition 1: The proactive care intervention provided an initial outreach call which provided the 5A’s and offered the option to participant in telephone counseling and helped facilitate the provision of smoking cessation medications if the participant was interested. Eight telephone counseling sessions were provided. The initial session was 60-minutes and the remaining calls were 30-minutes. Telephone sessions used MI and engaged in problem solving related to smoking cessation.	11	17	N/A	78
	N/A	N/A	N/A	N/A	Could be provided as part of usual care services.	Condition 2: Usual care participants received whatever care was normally provided in their behavioral health home. At the time of the study the health home had no formal cessation program. Usual care could entail the case manager contacting the participant’s medical provider about cessation treatment.	0	6.45	N/A	36
Metse et al., 2017	4 months	Telephone + in-person / Individual	Research staff / Telephone counselor	13	NRT	Condition 1: The intervention provided self-help material and a 10–15 minute MI session to smokers in an inpatient psychiatric hospital. At discharge participants received NRT and at least 11 individually tailored telephone behavioral counseling sessions. The 15-minute telephone sessions were weekly for 5 weeks, weekly or fortnightly (participant preference) during the next 7 weeks, and fortnightly during the final month. Sessions used a MI framework. Counselors encouraged participants to use a free government telephone Quitline. Participants’ general practitioner or outpatient mental health clinician were notified of their involvement in the study so they could encourage smoking cessation intervention.	13.4	6.45	NR	77.5
	N/A	N/A	N/A	N/A	NRT	Condition 2: The control group received routine hospital smoking cessation services which could include assessment of smoking status, brief advice to quit, NRT while inpatient, a 3-day supply of NRT upon discharge, referral to a Quitline, and a post-discharge smoking cessation care plan.	10	4.16	NR	58.3
Morris et al., 2011	NR	In-person + Telephone / individual + group	CMHC clinicians / Quitline counselors	15	NRT	Condition 1: The Smoking Cessation for Persons with Schizophrenia (SANE) group provided ten weekly one-hour sessions. Sessions were based on a MI and CBT approach and provided smoking cessation education, positive reinforcement, coping and anxiety reduction skills training, learning and practicing alternative behaviors, and strategies for building supportive relationships. Quitline services were provided by quitline counselors trained by the research team to work with people with psychiatric disorders. The Quitline provided five proactive calls designed to assist with quit attempts, promote healthier lifestyles, and prevent relapses. The quitline also provided 12-weeks of NRT.	NRb	4.93	1.38	N/A
	NR	Telephone / Individual	Quitline counselors	5	NRT	Condition 2: Participants in this condition received the same quitline services provided to the SANE condition.	NRb	4.91	1.63	N/A
Prochaska et al., 2014	6 months	Computer-based + in-person / Individual	Study counselor	4	NRT	Condition 1: The Transtheoretical Model (TTM)-tailored computer-assisted intervention was initiated on inpatient psychiatric units and provided one 15–30 minute on unit counseling session covering decisional balance, managing temptations, and motivational enhancement. The computer-assisted intervention provided a printed report that directed users to relevant sections in a staged-matched treatment manual (6th grade reading level). A letter was mailed to the participant’s outpatient provider to elicit support for smoking cessation. The computer intervention was repeated at the 3- and 6-month post-hospitalization follow-ups. Ten weeks of NRT was provided post-hospitalization.	20	N/A	N/A	N/A
	N/A	N/A	N/A	N/A	NRT during hospitalizat-ion only.	Condition 2: Usual care services provided NRT to participants only while they were hospitalized but did not provide any other smoking cessations services during hospitalization or upon discharge.	7.7	N/A	N/A	N/A
Stockings et al., 2014	4 months	Telephone + in-person / Individual	Project officer / Telephone counselor	NR	NRT	Condition 1: The intervention provided self-help material and a 10–15 minute motivational interviewing session to smokers in an inpatient psychiatric hospital. Participants received a two-week supply of NRT at discharge. Patients received fortnightly telephone smoking cessation support and an additional 12 weeks of NRT. The telephone sessions offered psychological support, NRT monitoring and advice, and techniques to improve smoking outcomes. Counselors encouraged participants to use a free government telephone Quitline and offered referrals to community smoking cessation programs.	7.7	10.3	1.5	31.7
	NR	NR	NR	NR	NRT during hospitalization; 3-day supply post discharge.	Condition 2: The control group received routine hospital smoking cessation services which could have included assessment of smoking status, brief advice to quit, NRT while inpatient and a 3-day supply upon discharge, referral to a Quitline, and a post-discharge smoking cessation care plan.	5.9	4.7	0.1	13.9
Tidey et al., 2011	22 days	In-person / Individual	Not reported	9	See intervention description.	Condition 1: Contingency management (CM) + bupropion (BUP) Condition 2: CM + placebo Condition 3: Non-contingent reinforcement (NR) + BUP Condition 4: NR + placebo Participants went to a laboratory three times per week to provide urine and CO samples and to complete additional study measures. NR participants received a $25 gift card for attending the study and providing biomedical samples regardless of cotinine levels. CM participants received a $25 gift for attendance and could receive cash bonuses for reduced cotinine levels. Bonuses increased for every reduced or abstinent sample. BUP participants received a medication organizer filled with bupropion according to study protocols.	N/A	N/Ac	N/A	N/A
Williams et al., 2010	26 weeks	In-person / Individual	Mental health clinicians	24	NRT	Condition 1: Treatment of Addiction to Nicotine in Schizophrenia (TANS) treatment sessions were 45-minutes each and incorporated motivational interviewing, skills training, and relapse prevention techniques. TANS used role plays to practice concrete smoking cessation skills. Both conditions received 16 weeks of NRT beginning on the target quit date (week 5) and both conditions had three stages of treatment (engagement, achieving abstinence, relapse prevention).	7d	NRe	N/A	44.4
	26 weeks	In-person / Individual	Mental health clinicians	9	NRT	Condition 2: The Medication management (MM) intervention was of moderate intensity and provided nine 20-minute individual sessions. MM sessions provided education about NRT, monitoring psychiatric symptoms, and education on interactions between medications and tobacco. Received NRT as described in condition 1.	7d	NRe	N/A	47.6

Note. NOS = Number of sessions; MI = motivational interviewing; CVD = cardiovascular disease risk factors; NRT = nicotine replacement therapy; CO = Carbon monoxide; CBT = cognitive behavioral therapy; OS = optional sessions; CMHC = community mental health center; 5A’s = ask, advise, assess, assist, arrange; SMI = serious mental illness; NR = not reported; N/A = not applicable or did not include this outcome; QR = Percentage of participants who achieved 7-day point prevalence abstinence at the study endpoint; CPD = Mean reduction in cigarettes per day at the study endpoint; FTND = Mean reduction in Fagerstrom Test for Nicotine Dependence scores at the study endpoint; QA = Percentage of participants who made a quit attempt during the study.

^aBrunette et al. (2018) used figures to illustrate quite rates, rather than providing the numeric value for each condition.

^bMorris et al. (2011) did not report quit rates for each treatment condition; only a combined quit rate across both conditions was reported.

^cTidey et al. (2011) measured reductions in cotinine and carbon monoxide levels, not cigarettes per day

^dQR was operationalized as continuous abstinence at one year instead of 7-day PPA

^eWilliams et al. (2010) did not report mean cigarettes per day by treatment condition; only a rate that combined both treatment conditions was reported for this outcome.

Several modes of therapy were used in the included studies. Sixteen studies provided smoking cessation education and taught behavioral skills strategies (e.g., managing cravings, avoiding triggers, coping skills) and 13 studies included motivational interviewing or enhancement sessions. Three studies were based on the Transtheoretical Model (TTM), one on Acceptance and Commitment Therapy (ACT), and three employed the 5A’s (i.e., Ask, Advise, Assess, Assist, Arrange). Six studies used CO feedback at each intervention session as an educational and motivation tool and five studies used continency reinforcement (i.e., rewards for reduced smoking) to encourage initiation and maintenance of change. Three of these studies combined CO feedback with contingency rewards (i.e., rewards for negative CO test). Most studies (N=13) used a combination of therapeutic approaches. Two interventions provided home visits, with Brody et al. (2017) providing visits for the purpose of reducing secondhand smoke exposure and Gilbody et al. (2015, 2019) providing intervention sessions in the home as an adaptation for people with SMI.

3.5. Study outcomes

In this section we summarize the findings for each of the smoking cessation outcomes examined across the 18 studies. For each outcome, we highlight the studies that reported statistically significant differences between the intervention and comparison groups.

3.5.1. Smoking abstinence outcomes

Sixteen studies conducted statistical analysis to test whether the smoking cessation intervention resulted in statistically significant differences in abstinence rates between treatment conditions. Twelve (median MQRS 12.0, range 10–13) of the 16 studies did not find any statistically significant differences between or within groups at the study endpoint, and Hickman et al. (2015) (MQRS = 16) found that both the intervention and comparison conditions achieved statistically significant abstinence rates from baseline, although there were no between group differences.

The remaining three studies reported statistically significant differences favoring the intervention condition compared to the comparison condition at the study endpoint. Two were categorized as having low methodological rigor (Brody et al., 2017; Christiansen et al., 2018), and one as having high methodological rigor (Prochaska et al., 2014). Brody et al. (2017) found CO-confirmed 7-day PPA rates were significantly different between the combination extended treatment plus home visits (COMB-EXT+HV) and treatment as usual (TAU) conditions (45% versus 8%, respectively) at the study endpoint (26 weeks). Differences were not significant between the COMB-EXT+HV and COMB-EXT without HV conditions or between the COMB-EXT and TAU conditions. The authors found that HV targeting secondhand smoke exposure provided an added benefit over COMB-EXT without HV. Christiansen et al. (2018) tested a motivational and preparatory intervention delivered by peer specialists and found that CO-confirmed 7-day PPA rates were statistically higher in the motivational intervention condition (8.5%) versus the attention control condition (1%) at the study endpoint (13 weeks). Prochaska et al. (2014) tested a Transtheoretical Model (TTM) computer-assisted intervention initiated on an inpatient psychiatry unit and found that CO-confirmed 7-day PPA rates were significantly higher in the TTM condition compared to usual care at 3 months (13.9% vs 3.2%, respectively), 6 months (14.4% vs 6.5%), 12 months (19.4% vs 10.9%), and 18 months (20.0% vs 7.7%).

3.5.2. Smoking reduction outcomes

Of the 15 studies that conducted statistical analysis to test whether the smoking cessation intervention achieved statistically significant reductions in smoking (e.g., CPD, expired CO) at the study endpoint, five (median MQRS 13.0 range 10–13) did not find any statistically significant differences between or within groups, and three studies (median MQRS 12.0 range 10–13) reported significant smoking reductions in both the intervention and comparison conditions, although there were no between group differences (Baker et al., 2018; Bennett et al., 2015; Williams et al., 2010).

The remaining seven studies found significant differences favoring the intervention versus the comparison condition at the study endpoint with six categorized as having low methodological rigor (Brody et al., 2017; Gelkopf et al., 2012; Japuntich et al., 2020; Morris et al., 2011; Stockings et al., 2014; Tidey et al., 2011) and one categorized as having high methodological rigor (Metse et al., 2017). Brody et al. (2017) reported that the COMB-EXT+HV and COMB-EXT without HV achieved significantly greater reductions in CPD over time compared to TAU (mean reduction, 16.1, 14.0, 7.4, respectively). Gelkopf et al. (2012) found the transtheoretical model intervention group significantly reduced CPD compared to the waitlist condition (mean reduction, 7.2 vs. 2.0, respectively) at the study endpoint (13 weeks). Similarly, Japuntich et al. (2020) found the proactive care group reported significantly greater reductions in CPD compared to the usual care group (mean reduction, 17.0 vs. 6.5, respectively) at the study endpoint (13 weeks). Morris et al. (2011) found that participants in the quitline plus community group were more likely than the quitline only group to achieve 50% or greater reduction in cigarette consumption at the study endpoint (21% vs. 8% respectively), and the number of CPD decreased significantly in both groups but did not differ across conditions. Stockings et al. (2014) found a significant time-by-condition interaction for CPD, favoring the telephone cessation support condition relative to the control condition at each timepoint (mean reduction, 10.3 vs. 4.7, respectively), and the telephone cessation support condition was significantly more likely to achieve 50% or greater reduction in CPD at each time point. Tidey et al. (2011) reported that cotinine and CO levels significantly decreased over the study period for patients in the contingency management condition compared to the non-contingent reinforcement condition and bupropion did not increase the efficacy of contingency management. Finally, Metse et al. (2017) reported that the telephone behavioral counseling condition relative to the control condition was significantly more likely to reduce CPD (mean reduction 6.45 vs. 4.16, respectively) and to reduce cigarette consumption by 50% or greater (53.9% vs. 44.1%, respectively) at the study end point (52 weeks).

3.5.3. Nicotine dependence (FTND) outcomes

Of the eight studies that conducted statistical analysis to test whether the smoking cessation intervention resulted in statistically significant reductions in nicotine dependence (FTND score), four (median MQRS 13.0, range 9–13) did not find any statistically significant differences between or within groups at the study endpoint, and three studies (median MQRS 12.0, range 11–13) found that FTND scores significantly decreased in both the intervention and comparison conditions, although there were no between group differences (Baker et al., 2015; Bennett et al., 2015; Morris et al., 2011). Only Stockings et al. (2014) (MQRS = 11) reported a significant time-by-condition interaction in which the telephone smoking cessation support condition had greater reductions in FTND scores over the course of the study relative to the control condition (mean reduction, 1.5 vs. 0.1 respectively).

3.5.4. Quit attempt outcomes

Six studies conducted statistical analysis to test whether smoking cessation interventions resulted in a statistically significant greater percentage of participants making a quit attempt relative to comparison conditions. Three studies (median MQRS 10.0, range 10–12) did not find any statistically significant differences between or within groups at the study endpoint. The other three studies found statistically significant differences between groups at the study endpoint, two were categorized as having low methodological rigor (Christiansen et al., 2018; Stockings et al., 2014) and one as high methodological rigor (Metse et al., 2017). Christiansen et al. (2018) found that the motivational and behavioral intervention condition was significantly more likely to have made a quit attempt by the study endpoint (13 weeks) relative to the attention control condition (50.0% vs. 21.4%, respectively). Stockings et al. (2014) did not find a significant condition-by-time interaction, however, significantly more participants in the telephone support condition made a quit attempt at each time point relative to the control condition. Metse et al. (2017) found that the telephone behavioral counseling condition was significantly more likely to have made at least one quit attempt compared to the control condition (77.5% vs. 58.3%, respectively) at the study endpoint (52 weeks).

4. Discussion

Findings from our review do not allow us to state which type or specific components of psychosocial interventions are most effective at achieving smoking abstinence or reduction due to methodological limitations and because no one type of intervention consistently rose to the top. However, studies finding significant differences between the intervention and comparison groups shared common evidenced-based components, including providing smoking cessation medications (e.g., NRT, bupropion), motivational enhancement techniques, and cessation education and skills training, but differed in intensity (e.g., number and frequency of sessions), duration, and modality (e.g., group, individual, technology). The most promising interventions included in this systematic review seem to be those that found significant between group differences in smoking cessation outcomes and were rated as high methodological rigor on the MQRS (Metse et al., 2017; Prochaska et al., 2014). These interventions had common elements, including being initiated while participants were admitted to inpatient psychiatric facilities, individualized to the participant, provided NRT, used technology (i.e., telephone or computer), contacted participants’ primary care providers to illicit support for smoking cessation, and had follow-up periods of at least 12 months. The intervention conducted by Prochaska et al. (2014) used a transtheoretical model computer-assisted intervention which provided an individually tailored stage-matched treatment manual, and Metse et al. (2017) provided a minimum of 11 individually tailored telephone behavioral counseling sessions using a motivational interviewing framework.

While it is encouraging that most studies included in this review provided interventions which align with current smoking cessation guidelines regarding counseling, motivational enhancement, and smoking cessation medications (Barua et al., 2018; Prochaska et al., 2017), it is still an open question as to the optimal intensity, duration, and modality of treatment. Understanding the necessary behavioral intervention components is further complicated by the fact that many interventions either facilitated the provision of smoking cessation medications or provided them to participants, but taking the medications was not required so the use of medications among participants was not uniform. Further, while some studies had the same medication protocol for all study conditions, other studies had different medication protocols for each study condition making interpreting the effect of psychosocial intervention components on study outcomes challenging. Based on these study designs, it is difficult to isolate the specific effect that behavioral interventions had on smoking cessation since whether or not participants were taking medications was not accounted for in most outcome analyses.

Understanding optimal treatment components is important so that treatment providers can understand the investment in resources needed to improve the health of people with SMI. Trying to determine the optimal components necessary for psychosocial smoking cessation interventions has been elusive as other reviews of smoking cessation interventions have found (Buchanan et al., 2010; Evins et al., 2015). Given this consistent finding, future intervention studies should move towards using sequential multiple assignment randomized trials (SMART), factorial designs, and N-of-1 trial designs (Davidson et al., 2014; Ng and Weisz, 2016; Parker et al., 2011). Factorial designs allow for multiple interventions to be assessed in a single trial and provides the possibility of assessing the effects of delivering the interventions separately or in combination with each other (Parker et al., 2011). One study in the current review, Brunette et al. (2018) used a factorial design in which they tested three intervention conditions and each of these conditions was further randomized to receive contingent reinforcement. They found that participants who received monetary incentives were significantly more likely to stop smoking over the course of the intervention, although this effect did not differ among the three intervention conditions. Further analysis of each intervention condition by monetary incentive group at each timepoint revealed that participants in the provider visit plus quitline condition who received monetary incentives were significantly more likely to achieve abstinence at 12-months compared to those in the provider visit plus quitline condition without incentives.

Only half (N=9) of the studies in this review found that the intervention condition achieved significantly higher smoking abstinence or reduction rates compared to the comparison group. This may be partially explained by the fact that most of the studies included active comparison conditions that received evidenced-based smoking cessation interventions including, NRT, quitlines, telephone counseling, and in-person peer counseling sessions. Another reason for the high number of nonsignificant findings could be the chronic nature of tobacco dependence. Smokers with SMI may need a treatment structure that allows them to enter and exit treatment as needed, and past studies show that people may need to make multiple quit attempts before they are successful (Prochaska et al., 2017), indicating that time limited interventions may not be successful in helping people with SMI quit smoking. This further highlights the utility of SMART trials and staged-matched interventions in which participants enter the intervention or trial at the stage of treatment personalized to their needs, and then based on their treatment response, can be further randomized to the next stage of treatment. It is also possible that factors outside of the smoking cessation intervention are creating barriers to smoking abstinence and reduction. People with SMI are overrepresented in neighborhoods of lower socioeconomic status which are less likely to have resources that promote health and have an easy availability of cigarettes and alcohol (Henwood et al., 2013; Weinstein et al., 2019). There are studies that suggest that tobacco retailer density is associated with greater tobacco dependence, and policies limiting tobacco retailer density and aggressive marketing may benefit smokers with SMI (Prochaska et al., 2017). Another barrier to cessation that may have confounded intervention findings could be the community mental health settings in which the smoking cessation interventions were provided. Whether or not the behavioral health setting promotes a culture of smoking cessation through smoke-free policies and attitudes can influence smoking behavior (Das and Prochaska, 2017). Studies included in this review did not report the smoking policies and culture of the intervention sites, which could be an important area for future intervention studies.

Equally as important as understanding the optimal intervention components is understanding possible differential effects of these interventions on specific subpopulations of people with SMI. In the current systematic review, overall, we found that trials were not representative of racial/ethnic minorities with SMI in the U.S., particularly Hispanics, Asians, and Native Americans were vastly underrepresented. This is concerning because some studies of people with SMI indicate that racial and ethnic minorities have poorer cardiovascular health than non-Hispanic whites (Carliner et al., 2014; Hawes et al., 2020) and quitting smoking can yield some of the greatest improvements in cardiovascular health (Creamer et al., 2019). Studies of smokers in the general population indicate that Native Americans smoke at higher rates than any other racial/ethnic minority group, followed by people of multiple races (CDC, 2017). These two racial groups had the lowest representation in the included studies. Future psychosocial smoking cessation trials need to increase the enrollment of racial and ethnic minority groups in these studies and carefully examine the differential outcomes of these interventions in these historically marginalized and underserved populations.

Another concerning finding was that 94% of studies provided no cultural or linguistic adaptations. Only Hickman et al. (2015) made cultural adaptations, but no intervention was offered in a language other than English. This is highly problematic given that over 60 million U.S. residents speak a language other than English at home and over 20% of this group indicates they either do not speak English well or not at all (Ryan, 2013). Research indicates that limited English language proficiency among Latino immigrants with psychiatric disorders is a barrier to mental health service use (Kim et al., 2011). Also, reviews of emergency medicine research provide evidence that the exclusion of non-English speakers from research studies has increased over time (Brodeur et al., 2017). It is imperative that interventions be tested and available in multiple languages since the number of people who speak languages other than English at home is increasing in the U.S. (Ryan, 2013). In addition, rather than adapting existing interventions for non-English speakers and racial/ethnic minority populations with SMI, researchers should consider creating these interventions specifically for non-English speakers and for racial/ethnic minority groups to address their unique needs and circumstances.

Another finding of this review is the low representation of sexual and gender minorities in psychosocial smoking cessation studies for people with SMI. Non-SMI sexual and gender minority studies report that smoking rates are significantly higher in this population compared to non-sexual and gender minority populations (Vogel et al., 2019). Given that people with SMI also face a heightened risk of smoking, it is possible that sexual and gender minorities with SMI have a compounded risk of smoking. For this reason, it will be important to include sexual and gender minorities in future smoking cessation trials for people with SMI.

Findings from the MQRS revealed that there are aspects of psychosocial smoking cessation intervention studies that could be strengthened. Follow-up length was less than 12-months in half of the studies and six of these studies had follow-up lengths of less than six months. It will be important for smoking cessation interventions to provide follow-up lengths of 12-months or more since other reviews of smoking cessation interventions for people with SMI have found that smoking abstinence and reduction declines substantially after the end of treatment (Bennett et al., 2013). Our findings also indicate that low intervention retention is an issue that needs to be addressed. One strategy for increasing retention may be designing or adapting trials for multiple racial/ethnic minority groups and multiple languages. Hickman et al. (2015) had one of the most diverse racial/ethnic minority group samples and provided cultural adaptations while having a follow-up rate of 90% at the study endpoint. The MQRS also highlighted the need for more studies to use assessors that are blinded to treatment conditions to reduce the risk of bias.

Our review has several limitations. The heterogeneity of both intervention and comparison groups prevented us from conducting a meta-analysis. Also, no systematic review is free of biases (e.g., publication bias) and there is always the possibility of missing studies that meet the inclusion criteria. We minimized these biases by following the PRISMA guidelines, searching a variety of databases and using manual searches, and by having two authors double screen all titles, abstracts, and full-text for inclusion. Rating the methodological rigor of studies with the MQRS is subjective. To minimize errors, we used an established measure, two independent raters, and resolved rating differences through consensus with a third rater. Our study excluded grey literature which could result in increased effects of publication bias in this review. Also, due to the study designs we were not able to isolate the effect of smoking cessation medications from the effect of psychosocial intervention components. Finally, since many of the psychosocial interventions employed were not distinct approaches, one approach may include a combination of different psychosocial approaches making it challenging to determine which components had the most impact on intervention outcomes.

5. Conclusion

A small number of interventions with large heterogeneity achieved significantly greater abstinence and reductions in smoking compared to comparison conditions. Findings from this review indicate that there is not yet enough evidence to support specific intensities, durations, or modalities of psychosocial smoking cessation treatments for smokers with SMI. Answering this question may require the use of factorial and SMART trial designs that can help elucidate the most effective treatment components. Future trials may also need to control for the provision of smoking cessation medications to understand the unique effect of psychosocial intervention components versus medication effects. Further, studies should describe the culture and policies related to smoking cessation at the interventions sites, particularly when the intervention takes place in community mental health agencies or outpatient psychiatric rehabilitation programs. Also, given the chronic nature of tobacco dependence, interventions may need to be structured to where participants can enter and exit treatment as needed according to their stage of change, rather than being of shorter duration (i.e., four weeks) and time limited. Finally, psychosocial smoking cessation trials need to make greater progress in developing and delivering interventions in multiple languages, incorporating cultural adaptations, and including and engaging more racial/ethnic minorities and sexual/gender minorities. These changes are needed to ensure that psychosocial smoking cessation interventions are available and effective for all people with SMI.