Interpersonal Communication in Response to an Intervention and Its Impact on Smoking Cessation Within a Low-Income Population

Abstract

Background. It is a priority to develop population-based strategies for reducing barriers to smoking cessation among low-income populations. Harnessing secondary transmission such as interpersonal communication (IC) has helped to reduce tobacco use, but there is a dearth of quasi-experimental research that examines IC and the full spectrum of smoking cessation behaviors, particularly in the context of population-level programs. Aims. Using quasi-experimental methods, we examined IC in response to a population-level intervention and its impact on the full spectrum of smoking cessation outcomes among low-income smokers. Method. We used propensity score matching; three different propensity score matching procedures were used to estimate and approximate experimental effects. We assessed four cessation outcomes: utilization of a free tobacco quitline (QL), making a quit attempt, and being smoke-free for 7 and 30 days at follow-up. We also examined predictors of IC. Results. IC was significantly related to QL utilization (effect sizes ranging from 0.135 to 0.166), making a quit attempt (effect sizes ranging from 0.115 to 0.147), being smoke-free for 7 days (effect sizes ranging from 0.080 to 0.121), and being smoke-free for 30 days at follow-up (effect sizes ranging from 0.058 to 0.082). Program-related and participant characteristics predicted IC, such as receiving emotional direct mail materials and living with a fellow smoker. Discussion. IC in response to a population-based program affected the cessation process, and IC had a marked impact on sustained cessation. Conclusion. Population-based programs should aim to harness psychosocial dynamics such as IC to promote sustained cessation among low-income populations.

Keywords

health promotion programs interpersonal communication population-based practice tobacco control tobacco-related disparities

Tobacco use is the leading preventable cause of morbidity and mortality in the United States (Lim et al., 2013; McAfee, Davis, Alexander, Pechacek, & Bunnell, 2013). In addition, smoking accounts for half of the disparity in mortality associated with socioeconomic status (SES) among men in the United States, and a similar effect is emerging among women (Gregoraci et al., 2017; Jha et al., 2006). The adult smoking rate has decreased by 44% since 1985 (Centers for Disease Control and Prevention, 2014), but high rates of smoking persist among low-income women and men (Slater, Nelson, Parks, & Ebbert, 2016).

A range of initiatives have helped decrease smoking rates (Gielen & Green, 2015), but health communication interventions have been especially effective (Livingood, Allegrante, & Green, 2016; McAfee et al., 2013). One of the strongest paths of influence stemming from health communications interventions on individuals’ behavior is a two-step flow process (or secondary transmission) that results in communication among interpersonal networks (i.e., interpersonal communication [IC]) of the target population (Katz & Lazarsfeld, 1955; Moran et al., 2016; Southwell & Yzer, 2009). Health communication research underscores the importance of IC within social networks in response to population-based health interventions in general (Hornik & McAnany, 2001; Southwell & Yzer, 2009), but interventions that encourage IC are potentially the most penetrating type of population-based intervention for smoking cessation (Livingood et al., 2016; McAfee et al., 2013; Southwell & Yzer, 2009).

Evidence shows that campaign-generated IC is positively associated with smoking cessation (Popova, 2016; Southwell, 2013; van den Putte, Yzer, Southwell, de Bruijn, & Willemsen, 2011), as IC can amplify a campaign’s direct effects and encourage secondary effects on behavior through multifaceted processes. For instance, IC can influence the salience of a campaign’s message, activate behavioral intentions, deepen the processing of information related to a campaign, increase personal relevance of a message, increase norm awareness related to a health behavior, or generate social support for behavioral change, among others (see Jeong & Bae, 2017; Southwell, 2013; Southwell & Yzer, 2009). These psychosocial dynamics are critical for sustained smoking cessation (e.g., Christakis & Fowler, 2008; Montano & Kasprzyk, 2008). As more evidence accumulates regarding IC and smoking cessation, IC’s potential impact has become a critical area of research (Dunlop, 2011; Jeong & Bae, 2017; Ramanadhan, Nagler, McCloud, Kohler, & Viswanath, 2017).

Yet there is a dearth of experimental or quasi-experimental research on IC in response to population-based tobacco treatment programs (Jeong, Tan, Brennan, Gibson, & Hornik, 2015; Livingood et al., 2016; Parks, Slater, Rothman, & Nelson, 2016; Popova, 2016). In addition, it is a public health priority to develop and disseminate health communication and population-based interventions that promote cessation and reduce barriers to sustained smoking cessation among low-SES populations, particularly because health communication campaigns are often less effective in socioeconomically disadvantaged populations (Niederdeppe, Kuang, Crock, & Skelton, 2008; Parks et al., 2016; Slater et al., 2016). Understanding how campaign-generated IC relates to cessation behavior could shed light on the potential role of IC as bridge to long-term cessation among low-SES populations.

Current Study

The current article examines IC in response to a population-based tobacco treatment program implemented within a low-income population. The program used health communication strategies paired with free cessation services as well as a financial incentive. Health communication interventions paired with financial incentives or free cessation services, such as those provided by state tobacco quitlines (QLs), are most effective, particularly among socioeconomically disadvantaged populations (Niederdeppe et al., 2008; Robinson et al., 2014; Slater et al., 2016).

Research Questions

Individuals vary in their willingness to share health information and to engage in interpersonal talk after being exposed to campaigns (Southwell, 2013). Differential propensities to engage in IC influence “selection effects” that potentially affect IC and cessation outcomes. Put differently, there are potential spurious factors that we must account for when examining the relationship between IC and smoking cessation because phenomena such as the willingness to engage in IC could explain both the presence of the explanatory variable (IC) and the outcome variable (smoking cessation). Outside of a randomized controlled trial, these selection effects or spurious relationships complicate evaluations of IC’s direct impact on smoking cessation. Consequently, methods such as quasi-experimental analyses should be used that can approximate experimental evidence in practice settings (when randomized controlled trials are infeasible) in order to assess IC’s impact on tobacco cessation (Livingood et al., 2016; Popova, 2016). To our knowledge, no quantitative research has evaluated a population-based intervention using methods that specifically predict and adjust for the propensity to engage in IC in order to approximate the experimental effect size of IC on smoking cessation. Subsequently, our first research question is the following:

Research Question 1: Does IC relate to smoking cessation within a low-income population after accounting for the propensity to engage in IC?

Tobacco cessation can include a range of behaviors from steps toward quitting to complete tobacco cessation (Prochaska, Redding, & Evers, 2008). Subsequently, the full spectrum of cessation behaviors (i.e., initial, intermediate, and sustained behavioral steps toward cessation) is important to consider. Our second research question asks the following:

Research Question 2: Is IC related to the full spectrum of smoking cessation behaviors (i.e., short- to long-term cessation) within a low-income population after accounting for the propensity to engage in IC?

Finally, we also consider how participant- and program-level characteristics are related to IC. Individuals have various reasons they communicate (or not) with others in response to a health communication intervention. Despite the increasing importance of documenting IC and its impact, there is limited research on the predictors of interpersonal discussions in response to health campaigns (Lee & Kim, 2015; Southwell, 2013), and in particular, predictors of IC in response to population-based smoking cessation programs have not been thoroughly documented (Brennan, Durkin, Wakefield, & Kashima, 2017). Consequently, our third research question asks the following:

Research Question 3: What predicts IC in response to a smoking cessation intervention within a low-income population?

Method

Overview and Participants

From September 2010 to September 2012, a smoking cessation intervention was implemented through Minnesota’s cancer early detection program, called “Sage” (see Slater et al., 2016). Sage provides cancer-related services to inadequately insured women 40 years of age or older with household incomes at or below 250% of the U.S. federal poverty level. The intervention was implemented within a population that was disproportionately female and entirely low-income (Sage also provides colorectal cancer services to men, and therefore a limited number of participants were male).

Intervention

Two recruitment strategies were designed to connect individuals to Minnesota’s free QL via three-way calls conducted through Sage’s central call center. The Sage call center is staffed by patient navigators (see Freund et al., 2008) who guide individuals through preventive services by coordinating appointments, transportation, and follow-ups.

The two recruitment strategies were direct mail (DM) and a phone referral system coined “opportunistic referral with connection” (ORC). For DM, participants were sent mailers that consisted of a folded card with an emotionally evocative message and graphic that were rooted in health communication theory (i.e., loss-frame and high-efficacy messaging; see Slater et al., 2016; Witte & Allen, 2000). The mailers gave directions to call Sage’s call center and included a small insert card advertising a financial incentive offer. For ORC, the Sage call center received phone calls regarding cancer-related services in accordance with its ongoing programmatic work and obtained smoking statuses of callers (i.e., ORC callers did not receive tobacco-related mailers); these self-identified smokers were subsequently offered to be connected to Minnesota’s free QL (see Slater et al., 2016).

Sage patient navigators connected all willing participants (for both recruitment routes) to the QL via three-way phone calls among participants, QL operators, and Sage patient navigators. Sage patient navigators were required to confirm a connection was made with the QL. The intervention offered a $20 incentive for having a confirmed connection to Minnesota’s free QL. Connected participants were offered cessation services, including free telephone counseling with a maximum of five sessions within a 6-month period. The QL also provided self-help materials and free nicotine replacement therapy if requested.

Survey Data

Survey data were gathered from participants who completed QL connections. Interviews were conducted by trained staff at least 7 months after participants’ QL connection, following evidence-based practices (see North American Quitline Consortium [NAQC], 2011). Since most relapse events occur within 6 months of an intervention, it is recommended that follow-up assessments are at least 6 months (and at most 3 years) after an intervention (Fiore et al., 2008). Approximately 77% of participants were interviewed between 7 and 8 months, and 13% between 8 and 12 months, with the maximum amount of time being between 12 and 13 months. A total of 844 smokers recruited via DM were connected to the QL, and 1,612 were connected via the ORC recruitment route. All participants who completed QL connections were targeted for the survey; respondents were unresponsive if there was no response after 10 call attempts—1,218 participants completed the survey. Not all participants who made QL connections were offered QL services for multiple reasons, and the current analyses included only individuals who were offered QL services (n = 995). Participants reported miscellaneous reasons for not receiving services, including accidental disconnections, not receiving call backs by QL, and reasons attributable to insurance coverage (see Slater et al., 2016). After accounting for missing data, the analytic sample consisted of 970 participants.

Measures

Table 1 provides descriptive statistics on all measures. Following past research (see DiClemente et al., 1991; NAQC, 2011), outcome measures consisted of four smoking cessation measures: (a) QL utilization, (b) quit attempt subsequent to intervention, (c) being smoke free for 7 days at follow-up, and (b) being smoke free for 30 consecutive days at follow-up. QL utilization was measured by asking whether respondents had used the tools and services that were offered from the QL (1 = yes, 0 = no). For making a quit attempt, participants were asked, “Since we connected you with the tobacco quitline, have you stopped smoking tobacco for 24 hours or longer because you were trying to quit” (1 = yes, 0 = no). For smoking abstinence, participants were first asked whether they had not smoked for 30 consecutive days at the time of their follow-up interview after QL connections (30-day abstinence rate; 1 = yes; 0 = no); a subsequent question was asked to participants who reported they were not smoke-free for 30 consecutive days in order to assess whether they had not smoked for 7 consecutive days at follow-up after QL connections (7-day abstinence rate; 1 = yes, 0 = no).

Table 1.

Outcomes, Interpersonal Communication, and Other Participant Characteristics (N = 970).

Characteristics	% (n)
Outcomes
QL utilization	66.29% (643)
Quit attempt^a	75.15% (729)
Smoke free—7 days^b	23.81% (231)
Smoke free—30 days^c	18.97% (184)
Predictors and demographics
Interpersonal communication about program	50.41% (489)
Incentive important for QL connection	68.66% (666)
Sage patient^d	61.86% (600)
Recruitment group^e	35.15% (341)
Sex	2.06% (20)
Race/ethnicity
White	74.74% (725)
Non-White	25.26% (245)
Smoked every day	94.02% (912)
Quit attempt with medications in past	62.68% (608)
Lives with smoker	37.63% (365)
Unlikely to contact QL without intervention	71.96% (698)
	Mean (SD)	Range
Age (in years)	52.24 (7.71)	18-75
Education^f	3.50 (0.90)	1-6
Years smoked	30.92 (10.93)	1-62

Note. QL = quitline.

Quit attempt longer than 24 hours made within 7 months from beginning of intervention. ^bThis is a measure of 7-day point prevalence abstinence 7 months from beginning of intervention, which is the recommended measure of abstinence by the North American Quitline Consortium. ^cThis is a measure of 30-day point prevalence abstinence 7 months from beginning of intervention, which is the recommended measure of abstinence by the North American Quitline Consortium. ^dWhether respondent was in Sage database as previous client (1 = yes, 0 = no). ^eMeasure of two different recruitment methods (1 = direct mail, 0 = transfer via phone). ^fEducation ranges from 1 (completed 8th grade or less) to 6 (graduate school).

IC about the incentive-based program was measured by asking the following question: “Did you tell others about the Sage offer that rewards smokers $20 for being connected with a tobacco quitline through the Sage call center” (1 = yes, 0 = no). Influence of financial incentives was measured as whether participants found the incentive to be important for their QL connection (1 = important, 0 = not important). Other program-related covariates included whether participants were ever previously involved with the Sage program (1 = Sage, 0 = non-Sage) and a dichotomous measure of the two recruitment routes (1 = DM, 0 = ORC).

Demographic variables included age (continuous years), sex (1 = male, 0 = female), race/ethnicity (1 = White, 0 = non-White), and education (ordinal scale of 1 = completed eighth grade or less to 6 = graduate school). Smoking characteristics included a continuous measure of years smoked, daily smoker (1 = yes, 0 = no), whether participants lived with a smoker (1 = yes, 0 = no), past quit attempts with medication prior to the program (1 = yes, 0 = no), and whether participants were unlikely to contact the QL without the current intervention (1 = yes, 0 = no).

Analytic Strategy

The primary analysis consisted of propensity score matching techniques. As previously noted, certain individuals may have been more likely to engage in IC, which potentially influences the direct effect of IC on smoking cessation because there are potential underlying variables that could explain both the presence/absence of IC and smoking cessation. Propensity score matching techniques are designed to account for these potential spurious relationships. In other words, propensity score matching procedures were used to estimate and approximate an experimental effect of IC on smoking cessation by accounting for individuals’ differential propensities to engage in IC.

The usefulness of accounting for the propensity to engage in IC depends on how well we quantitatively capture propensity scores. We relied on previous research and theory (i.e., Katz & Lazarsfeld, 1955; Southwell, 2013) to select predictors of IC. More specifically, we aimed to account for individual-level factors (e.g., education, likeliness to call quitline without intervention, addiction level), social network/community factors (e.g., living with a smoker), and content-related factors (e.g., receiving emotionally evocative mailing or not, incentive importance, previous Sage participation; see Southwell, 2013). Factors that influence IC are not easily distinguishable from predictors of the health behavior itself (e.g., being smoking free). Research shows communication behavior and health behavior change can have common predictors (e.g., Kim, Southwell, & Slater, 2011); consequently, predictors were also selected based on previous literature on smoking cessation behavior.

The propensity score matching techniques used here provided estimates of average treatment effects of the treated (ATT) for the relationship between IC and the four smoking cessation outcomes. Results were based on the ATT using a sample matched in accordance with the propensity to engage in IC (all covariates described in the Measure section were included in the analysis). First, the matching process consisted of a logit equation predicting the propensity to engage in IC, which required a “balancing of the sample”; a balancing property was satisfied, and individuals who met the balancing requirement were retained. Second, individuals’ propensity scores were used to match individuals according to the likelihood of engaging in IC—individuals who engaged in IC were matched with individuals who did not engage in IC based on propensity scores.

The second part of the analysis generated “treatment effects” for IC on smoking cessation outcomes by using three different matching techniques: (a) nearest neighbor matching, (b) kernel matching, and (3) stratification matching. Each matching procedure had strengths and weaknesses and therefore all three were used. Information on these statistical methods are detailed elsewhere (see Heckman, Ichimura, & Todd, 1998; Rosenbaum & Rubin, 1983; Smith, 1997). All analyses were conducted in Stata v.13. The Stata command is “pscore,” which is free to download. More details on the estimation of average treatment effects based on propensity scores within Stata are available (including example syntax and outputs), which offer materials for applied researchers on how to use the Stata command, how to interpret the output, and how to use the different matching techniques (see Becker & Ichino, 2002).

Results

Displayed in Table 2, IC was significantly related to all smoking cessation outcomes in all three matching techniques. There were minor differences in the effect sizes for the different matching techniques and the four outcomes; however, effect sizes were generally similar. Accounting for the propensity to engage in IC, IC was significantly related to QL utilization with effect sizes ranging from 0.135 to 0.166. Because of the binary outcome, this effect size can be interpreted as an absolute difference—that is, utilization rates for individuals who engaged in IC were higher than individuals who did not engage in IC by 13.5 percentage points (nearest neighbor), 16.1 percentage points (kernel), and 16.6 percentage points (stratification). IC was significantly related to making a quit attempt with effect sizes ranging from 0.115 to 0.147, indicating absolute differences in quit attempt rates were between 11.5 and 14.7 percentage points. Finally, IC was significantly related to being smoke-free in the past 7 and 30 days at follow-up. Effect sizes ranged from 0.121 to 0.080 for 7-day tobacco abstinence and 0.058 to 0.082 for 30-day tobacco abstinence. To highlight one set of results on tobacco abstinence rates, we focus on the nearest neighbor matching technique: abstinence rates for the IC group were higher than the non-IC group by 12.1 percentage points for 7-day abstinence and 8.2 percentage points for 30-day abstinence.

Table 2.

Propensity Score Matching Results for Treatment Effect of Interpersonal Communication on Smoking Cessation Outcomes.

Matching estimation procedure	Outcome measures
	Quitline utilization	Quit attempt^a	Smoke-free 7 days^b	Smoke-free 30 days^c
	IC^d effect	IC effect	IC effect	IC effect
Nearest neighbor (489)	.135*	.147*	.121*	.082*
	(.052, .218)	(.062, .233)	(.049, .192)	(.007, .157)
	(.061, .213)	(.061, .227)	(.075, .171)	(.006, .155)
Kernel (489)	.161*	.115*	.080*	.058*
	(.089, .234)	(.057, .174)	(.010, .149)	(.009, .107)
	(.086, .218)	(.039, .164)	(.006, .156)	(.020, .108)
Stratification (482)	.166*	.118*	.083*	.063*
	(.102, .230)	(.057, .179)	(.016, .149)	(.015, .111)
	(.104, .226)	(.056, .168)	(.024, .189)	(.017, .106)

Note. The first set of parentheses under effect sizes are normal theory 95% confidence intervals; the second set of parentheses under each effect size are bias-corrected 95% confidence intervals. Parentheses next to matching estimation procedures are the respective numbers of treated individuals (who engaged in interpersonal communication) who were matched with control individuals.

Quit attempt made within 7 months from beginning of intervention. ^bSeven-day point prevalence abstinence at follow-up. ^cThirty-day point prevalence abstinence at follow-up. ^dInterpersonal communication about the program.

p < .05.

Logit coefficient results for the propensity to engage in IC are presented in Table 3. In terms of program-level covariates, perceived incentive importance was positively related to IC (adjusted odds ratio [AOR] = 1.60, 95% confidence interval [CI] = 1.20, 2.12), and individuals with past Sage involvement were less likely to engage in IC (AOR = 0.73, 95% CI = 0.54, 0.99). The DM recruitment group was more likely to engage in IC compared with the ORC recruitment group (AOR = 1.54, 95% CI = 1.14, 2.09), adjusting for covariates. Living with a smoker was positively related to IC (AOR = 1.66, 95% CI = 1.27, 2.18). Age was negatively related to IC (AOR = 0.97, 95% CI = 0.95, 0.99).

Table 3.

Predictors of Propensity to Engage in Interpersonal Communication in Response to Intervention (N = 970).

Predictors	Coefficient
Incentive important for QL connection	0.47*** (0.15)
Sage patient^a	−0.31* (0.15)
Recruitment group^b	0.43** (0.15)
Sex	0.56 (0.50)
Race/ethnicity (1 = White)	−0.20 (0.16)
Smoked every day	−0.13 (0.29)
Quit attempt with meds in past	−0.10 (0.14)
Lives with smoker	0.51*** (0.14)
Unlikely to contact QL without intervention	−0.26 (0.15)
Age (in years)	−0.03* (0.01)
Education^c	−0.05 (0.08)
Years smoked	0.00 (0.01)

Note. QL = quitline. Standard errors are in parentheses.

Education ranges from 1 (completed 8th grade or less) to 6 (graduate school). ^bWhether respondent was in Sage database as previous client (1 = yes, 0 = no). ^cMeasure of two different recruitment methods (1 = direct mail, 0 = transfer via phone).

***

p < .001. **p < .01. *p < .05.

Discussion

IC in response to a population-level health communication intervention was associated with smoking cessation within a low-income population after accounting for the propensity to engage in IC (Research Question 1). We approximated the experimental impact of IC on smoking cessation, heeding the call for applied public health research to use methods that disentangle IC’s impact on smoking cessation in response to population-based interventions (see Jeong et al., 2015; Livingood et al., 2016; Popova, 2016). We also found that IC was positively related to initial, intermediate, and sustained behavioral steps toward smoking cessation (Research Question 2).

There is a dearth of information on experimental effect sizes associated with IC’s impact on smoking cessation (see, e.g., Jeong et al., 2015). According to one matching technique, individuals who engaged in IC had a 30-day abstinence rate at follow-up that was approximately eight percentage points higher than individuals who did not engage in IC. For a comparison, we considered an analogous scenario: according to a meta-analysis, the absolute difference of adding cessation medication to counseling interventions relative to interventions that included only counseling increased the abstinence rate (measured as both 7-day and continuous abstinence rates) by 7.5 percentage points (Fiore et al., 2008). In other words, the presence of IC in addition to receiving counseling had an effect size similar to adding cessation medication to counseling (using only 30-day quit rates rather than 7-day and 30-day quit rates combined).

In terms of predictors of IC, we found that both program-related components and participant characteristics influenced IC (Research Question 3). Supporting previous research (Jeong & Bae, 2017; Southwell, 2013), recruitment methods were related to IC. Specifically, individuals who received emotionally evocative mailers were more likely to engage in IC relative to individuals who were recruited through the ORC route. This indicates message factors can inspire IC, potentially through harnessing emotions or other tendencies related to salience and appraisal (Southwell, 2013). Recruitment route was related to IC adjusting for individuals’ self-reported likeliness to contact a QL without the intervention (i.e., a potential proxy for readiness to quit), indicating message factors or recruitment route predicted IC net of readiness to quit.

Other mechanisms underlying the influence of IC include social network characteristics (Jeong & Bae, 2017; Southwell, 2013; Southwell & Yzer, 2009). Even though we did not include direct measures of social network characteristics and activity, we found living with a fellow smoker was related to IC. This potentially supports research and theory demonstrating that social networks and communities within which an individual is embedded can influence IC (e.g., availability and type of conversation partner; Jeong & Bae, 2017; Southwell, 2013). Since our measure of living with a smoker is limited, future research should examine how social network characteristics affect IC and health behavior change in response to interventions. We found younger participants were less likely to engage in IC, potentially because they have fewer motivations (or health concerns resulting from fewer years smoked) to change behaviors; alternatively, younger participants have fewer opportunities to chat about the smoking cessation program given potential different characteristics of social networks.

Past Sage participation also had a negative relationship with IC, potentially resulting from the novelty of the program for non-Sage participants. The role of previous Sage participation also could have been contingent on perceived satisfaction with the program—that is, satisfactory experiences with health programs can influence future program engagement (e.g., Peipins, Shapiro, Bobo, & Berkowitz, 2006).

We found that the salience of the financial incentive also played a role in activating IC. Research on incentives and smoking cessation has not devoted specific attention to the potential role of IC (Parks et al., 2016). Financial incentives influence smoking cessation, particularly among low-income populations (Blumenthal et al., 2013; Sigmon & Patrick, 2012; Volpp et al., 2009). Yet incentives may only influence short-term cessation (Sigmon & Patrick, 2012; Volpp et al., 2009), and therefore a central concern within this area of research is how to improve long-term cessation outcomes within low-income populations through the use of financial incentives (Blumenthal et al., 2013). Our results indicate that there is a link between social interactions (IC) and financial incentive offers that is important for both short- and long-term smoking cessation.

Implications for Practice

Ecological health communication deals not only with individual-level and macro-level contexts but also microsystems (i.e., contexts between individual and macro contexts) that affect health outcomes (Moran et al., 2016). As Livingood et al. (2016) and Popova (2016) highlighted, macro- and individual-level processes are important, but the processes activated within microsystems (i.e., IC, group social norms) by health communication interventions have been critical for tobacco control efforts over the past 20-plus years within the United States. Population-based practice should continue to target microsystem dynamics by encouraging IC and harnessing IC’s influence on smoking cessation.

In addition, pairing health communication interventions with free tobacco cessation services can increase effectiveness, particularly within socioeconomically disadvantaged populations (Niederdeppe et al., 2008; Robinson et al., 2014). Future programs should continue to couple evidence-based health communication interventions with free cessation services, but programs should also consider pairing those interventions with other programmatic components that promote IC (i.e., health communication materials that appeal to emotions, financial incentives), primarily because IC can serve as a bridge to sustained cessation within low-SES populations.

Limitations

The data used in the current study were not derived from a randomized trial. The quasi-experimental methods provided an estimate for IC’s impact on smoking cessation that adjusted for the propensity to engage in IC. However, we were not able to assess the direct experimental effect of IC. Randomized controlled trials are ideal for identifying causal mechanisms, and future research should continue to explore the possible causal relationships among interpersonal talk, smoking cessation, and health communication programs. It was important to examine how a health communication intervention might affect tobacco-related disparities (Niederdeppe et al., 2008), but because we focused on a low-SES population, generalizing the current findings beyond low-SES populations should be done with caution. Our measure of IC captured whether individuals engaged in IC in response to the program, but we did not capture the content of conversations. Future research should examine how the content of conversations is related to tobacco cessation (Brennan et al., 2017).

Conclusion

This article evaluated a population-based program geared toward reducing barriers to sustained smoking cessation within a population of low-income smokers. The intervention used health communication strategies, free tobacco cessation services, and financial incentives. Our analyses focused specifically on IC in response to the program and how this form of communication related to smoking cessation. IC has a marked impact on the full spectrum of smoking cessation behaviors, and these relationships emerge even after accounting for the propensity to engage in IC. Program-related components (i.e., message type, financial incentives) in addition to individual-level and social network characteristics (e.g., age, living with a fellow smoker) influence IC in response to a health communication intervention. As researchers and practitioners continue to address the public health priority of establishing population-based programs geared toward promoting smoking cessation and reducing tobacco-related disparities, they should focus on activating and harnessing the influence of IC.

Footnotes

Acknowledgements

The authors thank Jonathan Slater, Christina Nelson, QUITPLAN® Helpline staff, Shelly Madigan, Sage patient navigators, Janis Taramelli, and Michelle Waste for their efforts.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was funded through the Centers for Disease Control and Prevention (American Recovery and Reinvestment Act, and Patient Protection and Affordable Care Act).

References

Becker

S. O.

Ichino

(2002). Estimation of average treatment effects based on propensity scores. Stata Journal, 2, 358-377.

Blumenthal

K. J.

Saulsgiver

K. A.

Norton

Toxel

A. B.

Anarell

J. P.

Gesten

F. C.

. . . Volpp

K. G.

(2013). Medicaid incentive programs to encourage healthy behavior show mixed results to date and should be studied and improved. Health Affairs, 32, 497-507.

Brennan

Durkin

S. J.

Wakefield

Kashima

(2019). Why do smokers talk about antismoking campaigns? Predictors of the occurrence and content of campaign-generated conversations. Health Communication, 32, 1539-1556.

Centers for Disease Control and Prevention. (2014). Trends in current cigarette smoking among high school students and adults, United States, 1965-2014. Retrieved from www.cdc.gov/tobacco/data_statistics/tables/trends/cig_smoking/

Christakis

N. A.

Fowler

J. H.

(2008). The collective dynamics of smoking in a large social network. New England Journal of Medicine, 358, 2249-2258.

DiClemente

C. C.

Prochaska

J. O.

Fairhurst

S. K.

Velicer

W. F.

Velasquez

M. M.

Rossi

J. S.

(1991). The process of smoking cessation: An analysis of precontemplation, contemplation, and preparation of stages of change. Journal of Consulting and Clinical Psychology, 59, 295-304.

Dunlop

S. M.

(2011). Talk “truth”: Predictors and consequences of conversations about a youth antismoking campaign for smokers and nonsmokers. Journal of Health Communication, 16, 708-725.

Fiore

M. C.

Jaen

C. R.

Baker

T. B.

Bailey

W. C.

Benowitz

N. L.

Curry

S. J.

. . . Wewers

M. E.

(2008). Treating tobacco use and dependence: Clinical practice guideline 2008 update. Rockville, Washington, DC: U.S. Department of Health and Human Services.

Freund

K. M.

Battaglia

T. A.

Calhoun

Dudley

D. J.

Fiscella

Paskett

, . . .The Patient Navigation Research Program Group. (2008). National Cancer Institute patient navigation research program: Methods, protocol, and measures. Cancer, 113, 3391-3399.

10.

Gielen

A. C.

Green

L. W.

(2015). The impact of policy, environmental, and educational interventions: A synthesis of evidence from two public health success stories. Health Education & Behavior, 42(1 Suppl.), 20S-34S.

11.

Gregoraci

van Lenthe

F. J.

Artnik

Bopp

Deboosere

Kovacs

, . . . DEMETRIQ Consortium. (2017). Contribution of smoking to socioeconomic inequalities in mortality: A study of 14 European countries, 1990-2004. Tobacco Control, 26, 260-268.

12.

Heckman

J. J.

Ichimura

Todd

(1998). Matching as an econometric evaluation estimator. Review of Economic Studies, 65, 261-294.

13.

Hornik

R. C.

McAnany

(2001). Theories and evidence: Mass media effects and fertility change. Communication Theory, 11, 454-472.

14.

Jeong

Bae

R. E.

(2017). The effect of campaign-generated interpersonal communication on campaign-targeted health outcomes: A meta-analysis. Health Communication. Advance online publication. doi:10.1080/10410236.2017.1331184.

15.

Jeong

Tan

A. S. L.

Brennan

Gibson

Hornik

R. C.

(2015). Talking about quitting: Interpersonal communication as a mediator of campaign effects on smokers’ quit behaviors. Journal of Health Communication, 20, 1196-1205.

16.

Jha

Peto

Zatonski

Boreham

Jarvis

M. J.

Lopez

A. D.

(2006). Social inequalities in male mortality, and in male mortality from smoking: Indirect estimation from national death rates in England and Wales, Poland, and North America. Lancet, 368, 367-370.

17.

Katz

Lazarsfeld

P. F.

(1955). Personal influence: The part played by people in the flow of mass communication. Glencoe, IL: Free Press.

18.

Kim

Southwell

B. G.

Slater

J. S.

(2011). Socioeconomic disparities in peer referral and information sharing about mammography. Paper presented at the 2011 annual convention of the International Communication Association (ICA), Boston, MA.

19.

Lee

H. O.

Kim

(2015). Linking health information seeking to behavioral outcomes: Examination of the antecedents and the outcome of childhood vaccination information seeking. Journal of Health Communication, 20, 285-296.

20.

Lim

S. S.

Vos

Flaxman

A. D.

Danaei

Shibuya

Adair-Rohani

. . . Ezzati

(2013). A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990-2010: A systematic analysis for the Global Burden of Disease Study 2010. Lancet, 380, 2224-2260.

21.

Livingood

W. C.

Allegrante

J. P.

Green

L. W.

(2016). Culture change from tobacco accommodation to intolerance: Time to connect the dots. Health Education & Behavior, 43, 133-138.

22.

McAfee

Davis

K. C.

Alexander

R. L.

Pechacek

T. F.

Bunnell

(2013). Effect of the first federally funded U.S. antismoking national media campaign. Lancet, 382, 14-20.

23.

Montano

D. E.

Kasprzyk

(2008). Theory of reasoned action, theory of planned behavior, and the integrated behavioral model. In Glanz

Rimer

B. K.

Viswanath

(Eds.), Health behavior and health education: Theory, research, and practice (4th ed., pp. 67-96). San Francisco, CA: Jossey-Bass.

24.

Moran

M. B.

Frank

L. B.

Zhao

Gonzalez

Thainiyom

Murphy

S. T.

Ball-Rokeach

S. J.

(2016). An argument for ecological research and intervention in health communication. Journal of Health Communication, 21, 135-138.

25.

North American Quitline Consortium. (2011). Guides on quitlines and research. Retrieved from http://c.ymcdn.com/sites/www.naquitline.org/resource/resmgr/research/researchguideapril2013.pdf

26.

Niederdeppe

Kuang

Crock

Skelton

(2008). Media campaigns to promote smoking cessation among socioeconomically disadvantaged populations: What do we know, what do we need to learn, and what should we do now? Social Science & Medicine, 67, 1343-1355.

27.

North American Quitline Consortium. (2011). Guides on quitlines and research. Retrieved from http://c.ymcdn.com/sites/www.naquitline.org/resource/resmgr/research/researchguideapril2013.pdf

28.

Parks

M. J.

Slater

J. S.

Rothman

A. J.

Nelson

C. L.

(2016). Interpersonal communication and smoking cessation in the context of an incentive-based program: Survey evidence from a telehealth intervention in a low-income population. Journal of Health Communication, 21, 125-133.

29.

Peipins

L. A.

Shapiro

J. A.

Bobo

J. K.

Berkowitz

(2006). Impact of women’s experiences during mammography on adherence to rescreening (United States). Cancer Causes & Control, 17, 439-447.

30.

Popova

(2016). Can we resolve the disconnect between how communication interventions work and how we evaluate them? Health Education & Behavior, 43, 121-124.

31.

Prochaska

J. O.

Redding

C. A.

Evers

K. E.

(2008). The transtheoretical model and stages of change. In Glanz

Rimer

B. K.

Viswanath

(Eds.), Health behavior and health education: Theory, research, and practice (4th ed., pp. 97-121). San Francisco, CA: Jossey-Bass.

32.

Ramanadhan

Nagler

R. H.

McCloud

Kohler

Viswanath

(2017). Graphic health warnings as activators of social networks: A field experiment among individuals of low socioeconomic position. Social Science & Medicine, 175, 219-227.

33.

Robinson

M. N.

Tansil

K. A.

Elder

R. W.

Soler

R. E.

Labre

M. P.

Mercer

S. L.

, . . . Community Preventive Services Task Force. (2014). Mass media health communication campaigns combined with health-related product distribution: A community guide systematic review. American Journal of Preventive Medicine, 47, 360-371.

34.

Rosenbaum

P. R.

Rubin

D. B.

(1983). The central role of propensity score in observational studies for causal effects. Biometrika, 70, 41-55.

35.

Sigmon

S. C.

Patrick

M. E.

(2012). The use of financial incentives in promoting smoking cessation. Preventive Medicine, 55, S24-S32.

36.

Slater

J. S.

Nelson

C. L.

Parks

M. J.

Ebbert

J. O.

(2016). Connecting low-income smokers to tobacco treatment services. Addictive Behaviors, 52, 108-114.

37.

Smith

H. L.

(1997). Matching with multiple controls to estimate treatment effects in observational studies. Sociological Methodology, 27, 325-353.

38.

Southwell

B. G.

(2013). Social networks and popular understanding of science and health: Sharing disparities. Baltimore, MD: Johns Hopkins University Press.

39.

Southwell

B. G.

Yzer

M. C.

(2009). When (and why) interpersonal talk matters for campaigns. Communication Theory, 19, 1-8.

40.

van den Putte

Yzer

Southwell

B. G.

de Bruijn

G.-J.

Willemsen

M. C.

(2011). Interpersonal communication as an indirect pathway for the effect of antismoking media content on smoking cessation. Journal of Health Communication, 16, 470-485.

41.

Volpp

K. G.

Troxel

A. B.

Pauly

M. V.

Glick

H. A.

Puig

Asch

D. A.

. . . Audrain-McGovern

(2009). A randomized, controlled trial of financial incentives for smoking cessation. New England Journal of Medicine, 360, 699-709.

42.

Witte

Allen

(2000). A meta-analysis of fear appeals: Implications for effective public health campaigns. Health Education & Behavior, 27, 591-615.