A longitudinal study of the association between the teachable moment heuristic and smoking of head and neck cancer patients

Abstract

Background

The Teachable Moment Heuristic (TMH) posits that change in its three components (ie, affective response, risk perception, and social role/self-concept) could increase motivation and confidence for health promotion, such as smoking cessation. For patients with cancer, smoking cessation can extend survival, while persistent smoking causes numerous problems (eg, cancer recurrence).

Purpose

This intensive longitudinal study with cancer patients aimed to evaluate the link between TMH constructs and smoking outcomes.

Methods

Participants included 42 newly diagnosed head/neck cancer patients who reported smoking in the past month. Participants completed a baseline questionnaire before a 30-day daily assessment. Single-item measures were used for all constructs.

Results

Participants who perceived more benefits of quitting smoking smoked fewer cigarettes, P < .05. Also, participants who reported less cancer worry smoked fewer cigarettes, P = .03. Within-persons, less cancer worry than one’s personal average predicted decreases in cigarettes the next day, P = .01. Perceiving greater benefits of quitting, and social support than one’s average predicted lower motivation but higher confidence to quit the same day, Ps < .05. Greater cancer worry than one’s average predicted day-to-day increases in cigarettes smoked, P < .01. Perceiving greater benefits of quitting and greater social support than one’s average predicted smoking fewer cigarettes that day, Ps < .05.

Conclusions

Components of the TMH are associated with favorable smoking cessation outcomes after a cancer diagnosis. Interventions to aid cancer patients with smoking cessation should focus on the benefits of quitting, bolster social support, and reduce cancer-related worry.

Introduction

Each year, approximately 800,000 patients are diagnosed with new tobacco-related cancers, including but not limited to lung, head/neck, cervical, and colorectal cancer.¹ One of the most diagnosed tobacco-related cancers is head/neck squamous cell carcinoma, which includes the disease sites of pharynx, larynx, and oral cavity, among others.² Patients with head/neck cancer are at a higher risk for symptom interference and psychological distress than other groups of cancer patients.³ In addition, smoking rates are typically highest among those with a tobacco-related cancer, with prior studies showing that 27-33% continue to smoke.⁴ There are many adverse health outcomes associated with smoking after cancer diagnosis, including a higher risk of all-cause mortality, cancer-specific mortality, new primary cancer, cancer recurrence, poor treatment response, return to the operating room, and treatment-related toxicity.⁵ Cancer patients who smoke also report worse quality of life when compared with former and never smokers.⁶ In contrast, smoking cessation is associated with multiple benefits, including increased survival.⁷ Given the clinical significance of smoking after a cancer diagnosis, it is important to understand the dynamics of health behavior change in this area and the modifiable factors that influence it.

The Teachable Moment Heuristic (TMH) is a framework for understanding the variables that underlie the motivation to adopt healthy behaviors and a launching pad for intervention development. This heuristic posits that a major health event, such as a cancer diagnosis, can serve as a catalyst for health behavior change if certain conditions are met.^8,9 These 3 conditions include a strong affective response, change in risk perception, and either change in one’s social role or self-concept.^8,10 Finally, the heuristic postulates that the changes in these components then prompt a change in motivation and self­efficacy.^8,9,11 A cancer diagnosis has the potential to change each component of the TMH, therefore also potentially increasing cancer patients’ motivation to quit smoking.^10,11 In terms of affective response, cancer patients may experience emotional distress such as fear of cancer recurrence or general worry about cancer, anxiety, and/or depression throughout the cancer care process.¹² A cancer diagnosis also poses a risk to physical health and well-being, specifically for patients who continue to smoke.⁵ Because of this, risk perceptions can change, which can result in health behavior change.^9,13 Finally, cancer patients may experience a change in their ability to carry out social roles and activities (eg, employee, active parent) and/or their self­concept (eg, healthy person, smoker, survivor) as they adopt an identity as a cancer patient.¹⁴ In sum, the TMH is a parsimonious, face valid, and widely applicable conceptual model that holds promise in its ability to identify targets in interventions to promote smoking cessation after cancer diagnosis.¹⁰

While the public may believe that cancer is a “teachable moment,” there are only a handful of empirical studies that have investigated the issue. Despite the few prior intervention studies that focus on changing the constructs within the TMH to promote smoking cessation,^15-17 little is known about how daily changes in these constructs could influence daily changes in smoking behavior after cancer diagnosis. This is largely because past studies do not focus on the acute period of cancer diagnosis when smoking cessation is urgent for clinical matters (eg, smoking cessation prior to surgery optimizes clinical outcomes). In addition, many past studies involve only cross­sectional designs or longitudinal surveys spaced too far apart in time to fully capture changes in the TMH constructs and key events in the smoking cessation process.^15-18 Additionally, this is the first known study to extend the TMH to predict actual smoking behavior change (eg, cigarettes per day) of cancer patients. Reported here is a hypothesis-generating, intensive longitudinal study with new head/neck cancer patients who participated in a 30-day daily assessment protocol. This study aims to evaluate the association between day-to-day changes in the TMH and 3 smoking cessation outcomes (namely, motivation and confidence to quit smoking and cigarettes per day).

Method

Participants

The sample consisted of 42 adults (ie, ≥18 years old), recently diagnosed, first primary head/neck cancer patients who reported smoking in the past month at the time of study enrollment. Exclusion criteria included: (1) prior cancer diagnosis except nonmelanoma skin cancer; (2) unreliable phone access; (3) inability to speak, read, or write in English; (4) serious cognitive or psychiatric impairment; and (5) pregnant or planning to become pregnant in the next 6 months.

Procedure

Study procedures are detailed elsewhere.^19,20 Briefly, participants were recruited prospectively from an academic cancer center. At the time of the study, the tobacco treatment program at this cancer center was relatively new and therefore only available to 2 participants. Individuals who were screened, confirmed to be eligible, and provided written informed consent completed a baseline assessment and were then scheduled to start a 30-day daily assessment, with 1 time per day data collection primarily via computer-assisted, phone-based, interactive voice response (IVR). Participants completed 22.5 ± 7.9 days of the daily assessments. Data were collected from October 2015 to January 2019. Participants received $20 for the baseline and up to $80 for the daily assessments.

Measures

Demographics

At baseline, participants reported their demographic background (eg, age, race, gender, marital status, education, income) based on standard items from the Behavioral Risk Factor ­Surveillance System.²¹

Smoking-related outcomes

During the baseline survey, participants answered questions about their smoking behavior and quit attempt history, with questions commonly used in clinical and population research among cancer patients and survivors.²² The daily assessment evaluated the number of cigarettes smoked the prior day plus intention (here, motivation and confidence) to abstain from smoking the following day via Contemplation Ladders.²³ ­Participants separately rated their motivation and confidence on a scale from 0 = very definitely no/not at all confident to 10 = very definitely yes/very confident, respectively. Evidence of validity and reliability of these questions comes from past studies that use IVR and ecological momentary assessment technology, as well as other single-item indicators of behavior for daily assessment of smoking behavior.^24-26

TMH variables

The baseline survey and daily assessment evaluated the TMH using 4 items that asked about their thoughts and feelings that day. These items tapped into affective response (cancer worry; “How much did you worry about your cancer today?”), risk perception (perceived harms of smoking; “How harmful is smoking to you?” and perceived benefits of abstinence; “How beneficial is not smoking to you?”), and one social role/self­concept (social support; “How much support did you receive from others today?”). All items were measured on a scale from 0 to 9, for example, 0 = never and 9 = almost all the time. The single-item measures for cancer worry and social support were created from corresponding multi-item measures on the baseline assessment (results reported elsewhere¹⁹). The single-item cancer worry measure was created to be consistent with the Impact of Event Revised Scale and the single-item social support measure with the Social Support ­Questionnaire.^27,28 Both measures have demonstrated adequate reliability and ­validity.^29,30 Social support was chosen as a proxy for the social role/self-concept component to determine the relationship between support from others (eg, family, friends, healthcare providers) and key smoking cessation outcomes over time.¹¹ Pearson’s r values were conducted to evaluate how well the single-item measures for cancer worry and social support related to their corresponding multi-item measures. Those results can also be found elsewhere.¹⁹ Finally, there is also research evidence for the reliability and validity of single-item measures to assess smoking-related risk perceptions.²⁶

Data analysis plan

The data analysis plan was determined post hoc. Descriptive statistics were conducted to obtain sample characteristics of participants’ smoking-related outcomes and TMH variables at baseline, and between- and within-person correlations among all variables of interest. Two generalized linear mixed models (GLMMs) using proc glimmix in SAS and 1 hierarchical linear model (HLM) using proc mixed in SAS were used to test the daily associations among TMH components, cigarettes per day, and motivation and confidence to quit. ­Participants’ scores were disaggregated on each TMH construct into between- and within-person variability. Between-person scores were created by (1) calculating each participant’s mean value on each construct across all observations, (2) calculating a grand mean of each construct from all participants’ means, and (3) subtracting the grand mean from each participant’s mean. Within-person scores were created by subtracting each participant’s mean score on each construct from their raw score at each observation. To improve the fit of our second GLMM, motivation to quit was reverse-scored so that it closely aligned with a negative binomial distribution. Alpha was set to P < .05.

In the first GLMM, the number of cigarettes smoked per day was regressed on between- and within-person scores on all TMH variables. A log link was used with a Poisson distribution, given this aligned with the count nature and distribution of the number of cigarettes smoked variable (median = 2.5, mode = 0, skewness = 1.27, kurtosis = 0.79; Figure S1). In the second GLMM, daily motivation to quit was regressed on between- and within-person scores on all TMH variables. A log link with a negative binomial distribution was used, given the distribution of the motivation to quit variable (median = 8, mode = 9, skewness = –1.27, kurtosis = 0.26; Figure S2). In the HLM, daily confidence to quit was regressed on between- and within-person scores on all TMH variables with no link function given the continuous nature and distribution of the ­confidence variable (median = 7, mode = 9, skewness = –0.57, kurtosis = –1.12; Figure S3). A lag-1 autoregressive residual covariance structure was applied, and the Kenward-Roger method was used to calculate denominator degrees of freedom.

To test the associations regarding day-to-day changes in smoking cessation outcomes, 2 GLMMs and one HLM were conducted using the same modeling specifications as above. In each model, we regressed 1 outcome at time t + 1 on all between- and within-person TMH variables at time t, controlling for the respective outcome variable at time t. In total, we tested 6 models (4 GLMMs and 2 HLMs).

Results

Sample characteristics

The sample was White, non-Hispanic (97.6%, n = 41) and Black/African American (2.4%, n = 1). The sample was identified as 71.4% male (n = 30) and 28.6% female (n = 12), with an average age of 57.6 ± 7.3 years (range = 43 to 77). About half of the participants were married or partnered (45.2%, n = 19) and reported that they were disabled (50.0%, n = 21). In terms of income, 64.3% (n = 27) reported annual income of $20,000 or less, 28.6% (n = 12) $20,000 to $50,000, and 12.5% (n = 3) $50,000 or more (range = <$10,000 to $75,000 or more). The most common level of education was high school (42.9%, n = 18). In terms of clinical characteristics, the sample was predominantly diagnosed with stage IV or metastatic cancer (52.4%, n = 22), and the most common cancer site was the larynx (42.9%, n = 18). Most participants received both radiation and chemotherapy (33.3%, n = 14). Participants were diagnosed with cancer an average of 1.2 ± 1.6 months before enrollment.

Smoking behavior at baseline

Participants reported smoking an average of 12.4 ± 11.3 cigarettes per day. Over half of participants (54.8%, n = 23) reported at least 1 24-hour quit attempt in the past year, with half (n = 21) having tried since diagnosis. Motivation and confidence to quit smoking for the next month were in the moderate-to-high range (7.2 ± 3.8 and 5.8 ± 3.8, respectively). People who had greater motivation to quit also tended to express more confidence in their ability to quit, r = 0.45, P < .01.

Teachable moment heuristic variables at baseline

Participants reported experiencing a moderate degree of cancer-related worry (5.7 ± 3.5). Both the perceived benefits of abstaining from smoking (7.2 ± 3.0) and the perceived harms of smoking were rated in the high range (8.1 ± 2.1). Moderate levels of social support were also reported (5.7 ± 3.3).

Correlations (Table 1)

Table 1.

Between- and within-person correlations among variables of interest.

Variable	1	2	3	4	5	6	7
Outcomes
1. Cigarettes per day		–.66**	–.83**	.10	–.25	–.13	.16
2. Motivation to quit	–.28**		.75**	.06	.49**	.24	–.01
3. Confidence to quit	–.33**	.41**		–.23	.25	.06	.09
Teachable Moment Heuristics
4. Cancer worry	.04	–.09**	–.10**		.13	.33*	–.04
5. Benefits	–.03	.13**	.14**	.18**		.43**	.02
6. Harms	.04	.11**	.16**	.19**	.32**		–.21
7. Social support	–.09**	.09**	.09*	–.05	.09**	.04

Within-person correlations below the diagonal. Between-person correlations above the diagonal.

^* P < .05,

^** P < .01.

Between-persons, the only significant relation between smoking-related outcomes and TMH constructs was a positive correlation between motivation to quit and perceived benefits of abstaining from smoking, r = 0.49, P < .01 (Table 1). ­Within-persons, TMH constructs were significantly related to most smoking-related outcomes; although all these associations were small-sized, rs: –0.10 to 0.16.

Cancer-related worry

Between-persons

People who reported more cancer-related worry on average across the study period tended to report smoking more cigarettes on average, B = 0.41, P = .03, than those lower in ­cancer-related worry (Tables 2 and 3). By contrast, average cancer-related worry was not significantly related to average intentions to or confidence in quitting, Ps > .10.

Table 2.

Between and within-person daily associations between teachable moment components and smoking outcomes.

Daily (Same-Day)
Cigarettes per day					motivation to quita					Confidence to quitb
Variable	B	SE	P	95% CI	Variable	B	SE	P	95% CI	Variable	B	SE	P		95% CI
Intercept	–0.04	0.46	.94	[–0.96 to 0.89]	Intercept	–0.15	0.26	.58	[–0.67 to 0.38]	Intercept	5.87	0.43	<.01		[5.00 to 6.74]
Between persons					Between persons					Between persons
Cancer worryc	0.41	0.18	.03	[0.04 to 0.79]	Cancer worry	0.11	0.26	.58	[–0.10 to 0.32]	Cancer worry	–0.29	0.17		.10	[–0.63 to 0.06]
Benefitsd	–0.89	0.42	.04	[–1.74 to –0.04]	Benefits	–0.76	0.24	<.01	[–1.25, to –0.26]	Benefits	0.57	0.38		.14	[–0.19 to 1.35]
Harmse	.18	0.36	.63	[–0.55 to 0.91]	Harms	0.27	0.21	.21	[–0.16 to 0.71]	Harms	0.09	0.29		.76	[–0.50 to 0.68]
Social supportf	0.01	0.17	.97	[–0.35 to 0.36]	Social support	0.08	0.10	.42	[–0.12 to 0.28]	Social support	0.08	0.16		.61	[–0.24 to 0.41]
Within persons					Within persons					Within persons
Cancer worry	0.01	0.01	.07	[–0.001 to 0.03]	Cancer worry	0.09	0.02	<.01	[0.05 to 0.13]	Cancer worry	–0.06	0.03		.05	[–0.12 to 0.0003]
Benefits	–0.01	0.01	.02	[–0.03 to –0.003]	Benefits	–0.05	0.02	<.01	[–0.08 to –0.02]	Benefits	0.07	0.03		.01	[–0.02 to 0.12]
Harms	0.00	0.01	.86	[–0.02 to 0.02]	Harms	–0.07	0.03	.02	[–0.13 to –0.01]	Harms	0.22	0.07		<.01	[0.09 to 0.36]
Social support	–0.03	0.01	<.01	[-0.04 to –0.01]	Social support	–0.05	0.02	.01	[-.09 to –0.01]	Social support	0.07	0.03		.04	[0.004 to 0.13]
Random intercepts σ = 7.67, SE = 2.06. Ps < .05 are deemed statistically significant.					Random intercepts σ = 2.46, SE = 0.70. Ps < .05 are deemed statistically significant.					Random intercepts σ = 6.89, SE = 1.74, p < .01. AR(1) ρ = .049, SE = 0.03, P < .01. Residual σ = 2.93, SE = 0.19, P < .01. AIC = 3525.90. Ps < .05 are deemed statistically significant.

^aMotivation: 0 = very definitely no and 9 = very definitely yes.

^bConfidence: 0 = not at all confident and 9 = extremely confident.

^cCancer worry: 0 = none and 9 = a great deal.

^dBenefits: 0 = not at all beneficial and 9 = extremely beneficial

^eHarms: 0 = not at all harmful and 9 = extremely harmful.

^fSocial support: 0 = none and 9 = a great deal.

Abbreviations: CI, confidence interval; SE, standard error.

Table 3.

Between and within-person associations of daily changes in teachable moment components and smoking outcomes.

Next day (day-to-day)
Cigarettes per day					Motivation to quit					Confidence to quit
Variable	B	SE	P	95% CI	Variable	B	SE	P	95% CI	Variable	B	SE	P	95% CI
Intercept	–0.23	0.41	.58	[–1.05 to 0.60]	Intercept	–0.45	0.25	.08	[–0.95 to 0.05]	Intercept	2.22	0.23	<.01	[1.75 to 2.69]
Cigaretteslagged	0.04	0.003	<.01	[0.03 to 0.04]	Motivationlagged	0.13	0.01	<.01	[0.10 to 0.16]	Confidencelagged	0.63	0.03	<.01	[0.57 to 0.68]
Between persons					Between persons					Between persons
Cancer worry	0.38	0.16	.03	[0.05 to 0.71]	Cancer worry	0.09	0.10	.38	[–0.11 to 0.28]	Cancer worry	–0.11	0.07	.13	[–0.25 to 0.04]
Benefits	–0.77	0.38	.05	[–1.54 to 0.002]	Benefits	–0.54	0.23	0.03	[–1.01 to –0.07]	Benefits	0.25	0.15	.12	[–0.07 to 0.57]
Harms	0.13	0.33	.71	[–0.55 to 0.80]	Harms	0.19	0.21	.37	[–0.23 to 0.61]	Harms	0.04	0.12	.71	[–0.20 to 0.29]
Social support	–0.02	0.15	.90	[–0.33 to 0.29]	Social support	0.05	0.09	.56	[–0.13 to 0.24]	Social support	0.05	0.06	.46	[–0.08 to 0.18]
Within persons					Within persons					Within persons
Cancer worry	0.02	0.01	.01	[0.004 to 0.04]	Cancer worry	0.06	0.02	<.01	[0.02 to .10]	Cancer worry	–0.04	0.03	.14	[–0.10 to 0.01]
Benefits	–0.01	0.01	.02	[–0.03 to –0.003]	Benefits	<0.01	0.02	>.99	[–0.30 to 0.30]	Benefits	0.01	0.02	.58	[–0.04 to 0.06]
Harms	–0.01	0.01	.20	[–0.04 to 0.008]	Harms	–0.07	0.03	.01	[–0.12 to –0.02]	Harms	0.07	0.05	.15	[–0.03 to 0.17]
Social support	–0.01	0.01	.08	[–0.03 to 0.001]	Social support	0.00	0.02	.84	[–0.04 to 0.03]	Social support	–0.02	0.03	.47	[–0.08 to 0.04]
Random intercepts σ = 5.98, SE = 1.64. Ps < .05 are deemed statistically significant.					Random intercepts σ = 2.12, SE = 0.64. Ps < .05 are deemed statistically significant.					Random intercepts σ = 1.01, SE = 0.36, P < .01. Lag-1 ρ = –.20, SE = 0.05, P < .01. Residual σ = 2.28, SE = 0.13, P < .01. AIC = 3155.30. Ps < .05 are deemed statistically significant.

Within-persons

On days participants experienced greater cancer-related worry than normal, they demonstrated significantly greater increases in the number of cigarettes smoked than their personal average the next day, B = 0.02, P = .01 (Tables 2 and 3). On days participants experienced greater cancer-related worry than normal, they reported greater daily motivation to quit on that day, B = 0.09, P < .01, and greater increases the next day, B = 0.06, P < .01. However, within-person changes in cancer-related worry were not significantly related to same-day confidence to quit, B = –0.06, P = .05, or day-to-day changes in confidence, B = –0.04, P = .14.

Perceived benefits of abstaining from smoking

Between-persons

People who perceived greater benefits of abstaining from smoking on average across the study period smoked significantly fewer cigarettes on average than those who perceived fewer benefits in abstaining, B = –0.89, p = .04 (Tables 2 and 3). ­Furthermore, people who perceived greater benefits of abstaining from smoking reported greater motivation to quit smoking than those who perceived fewer benefits from abstaining, B = –0.76, P < .01. By contrast, there was not a significant association between perceived benefits of abstaining from smoking and confidence in quitting, P = .10.

Within-persons

On days participants perceived greater benefits of abstaining than normal, they smoked significantly fewer cigarettes that day, B = –0.01, P = .02, and reported significantly greater reductions in the number of cigarettes smoked from one day to the next, B = –0.01, P = .02 (Tables 2 and 3). Although participants also reported lower motivation to quit on days they perceived greater benefits of abstaining than normal, B = –0.05, P < .01, the perceived benefits of abstaining did not predict day-to-day changes in intention to quit, B < 0.01, P > .99. Similarly, on days participants perceived greater benefits of abstaining than normal, they reported greater confidence to quit, B = 0.07, P < .01, but these perceptions of benefits did not predict day-to-day changes in confidence to quit, B = 0.01, P = .58.

Perceived harms of continued smoking

Between-persons

There were no significant between-person associations between perceived harms of continued smoking and the number of cigarettes smoked, motivation to quit, or confidence to quit, Ps > .10 (Tables 2 and 3).

Within-persons

Within-person changes in perceived harms of smoking did not significantly predict same-day or day-to-day changes in the number of cigarettes smoked, Ps > .15 (Tables 2 and 3). On days participants perceived greater harms of smoking than normal, they reported significantly lower motivation to quit, B = –0.07, P = .02, and greater reductions than normal in motivation to quit the next day B = –0.07, P < .01. Although on days participants perceived greater harms of smoking than normal, they reported significantly greater confidence to quit, B = 0.22, P = .01, these perceptions of harm did not significantly predict changes in confidence to quit the next day, B = 0.07, P = .15.

Social support

Between-persons

There were no significant between-person associations between average social support and the average number of cigarettes smoked, intentions to, or confidence in quitting, Ps > .40 (Tables 2 and 3).

Within-persons

On days participants experienced greater social support than normal, they smoked significantly fewer cigarettes than their personal average the same day, B = –0.03, P < .01, but not the next day, B = –0.01, P = .08 (Tables 2 and 3). Similarly, on days participants experienced greater social support than normal, they reported significantly lower motivation to quit that day, B = –0.05, P = .01, but no significant day-to-day changes in motivation to quit, B = –0.004, P = .84. Finally, on days participants experienced greater social support than normal, they reported significantly greater confidence to quit on the same day, B = 0.07, P = .04, but no significant day-to-day changes in confidence to quit, B = –0.02, P = .47.

Discussion

In this sample of newly diagnosed head/neck cancer patients, variables meant to represent components of the TMH and key smoking cessation variables (cigarettes per day and both motivation and confidence to quit) were examined via 30 days of intensive longitudinal daily assessment. In terms of the TMH conceptual framework, intention was measured via motivation to quit and self-efficacy was measured via confidence to quit. These items were measured separately, but in prior studies, these items are often highly correlated and thus combined. We chose to keep them separate because some research shows motivation and confidence have different relations with other key variables, including smoking cessation.³¹ Furthermore, previous studies evaluating the TMH primarily utilize cross-sectional or longitudinal designs spaced too far apart, which prohibit the discernment of how changes in its components influence the overall smoking cessation process.^15-18 Finally, this study took the TMH a step further than motivation and confidence to quit smoking, and predicted actual smoking behavior change (ie, number of cigarettes per day). This study is the first to evaluate daily changes in the TMH components and their association with motivation and confidence to quit smoking, as well as cigarettes smoked per day, which holds potential for the development of intervention targets for cancer patients who smoke.

Affective response

In terms of affective response, when participants had more cancer-related worry than their personal average, they reported greater motivation to quit smoking. This finding supports the TMH, in that a strong affective response after a health event can increase motivation to quit smoking.⁸ In fact, there is evidence in the general population that worry is associated with increases in motivation to quit smoking.³² However, there was no association found between cancer worry and confidence to quit smoking, but more intense cancer-related worry on a given day did predict day-to-day increases in the number of cigarettes smoked. Studies have demonstrated the relationship between smoking status and cancer-related anxiety/worry and call for the integration of treatment for smoking and psychological distress.^33,34 Unfortunately, it may be that increases in motivation to quit smoking due to cancer worry do not translate to lasting changes in smoking behavior. This finding could be explained in a cyclical pattern, such that higher cancer worry leads to smoking to cope with worry, which, in turn, again increases cancer worry, regardless of an increased motivation to quit smoking. It is important to note that only one component of psychological functioning was measured, so other components, such as depression, could have produced different results.

Perceived benefits of abstinence and harms of smoking

When participants perceived greater benefits of abstinence than their personal average, they smoked fewer cigarettes that day and the next day. In contrast, perceived harms of persistent smoking were not associated with the number of cigarettes participants smoked per day. Overall, these findings demonstrate the importance of positive messaging by focusing on the benefits of abstinence when discussing smoking cessation with cancer patients. It also underscores the need to draw upon the 2020 Surgeon General’s Report on Smoking Cessation³⁵ instead of solely focusing on the 2014 report on the negative consequences or risks of persistent smoking.⁵ These results also indicate that focusing on perceived risks of smoking does not change actual smoking behavior or motivate cancer patients to quit. This could be driven by fatalism or beliefs about helplessness and pessimism regarding cancer diagnosis and prognosis.³⁶

When participants perceived greater benefits of abstaining and harms of smoking, they had lower same-day motivation to quit smoking, but higher confidence to quit. This is in line with prior research, which demonstrates that low motivation, but high confidence, may mean that a patient recognizes the benefits of quitting, and believes a quit attempt would be successful, but currently lacks a strong desire to take action toward quitting. Finally, as the question regarding perceived harms of smoking pertained to general health-related risks, not ­cancer-specific risks, construct measurement might have impacted the results. Participants may have experienced a weaker relation between general health-related harms and their smoking behavior than with the more specific cancer-related harms they may have been receiving messaging about from their oncologists or other cancer care professionals.

Social support

Partly due to measurement issues and lack of prior research on the social role/self-concept component of the TMH, social support was used as a proxy in this study.¹⁰ Results show that within-persons changes in social support predict same day, but not next day, decreases in cigarettes per day and increases in motivation and confidence to quit smoking. Importantly, social support is implicated in both healthy adjustment to cancer diagnosis and successful smoking cessation.^37,38 Prior research demonstrates that social support may serve as a buffer against stress, thus aiding in the smoking cessation process.³⁹ Prior studies also show that cancer patients with lower perceived social support are more likely to continue to smoke,³⁸ and that quitting smoking with a friend or spouse may increase quit success.⁴⁰ Social support may then be important for cancer patients throughout the smoking cessation process, specifically from family members or peers who are also trying to quit smoking or do not currently smoke. However, social support tends to decrease as cancer treatment comes to an end,⁴¹ so patients in this study may have been experiencing the highest levels of support they will receive after diagnosis. Interventions to increase social support for smoking cessation should leverage the increase in social support that occurs immediately after diagnosis. More research is needed to determine consistent and adequate measurement of the social role/self-concept component of the TMH.

Implications

This study presents a few important clinical implications. First, this study provides initial evidence for the ability of the TMH to predict not only daily changes in motivation to quit smoking but also smoking behavior. These findings indicate that treatments that focus on reducing cancer worry, amplifying perceptions of the benefits of smoking abstinence, and increasing social support might also produce increases in motivation to quit and reductions in cigarettes per day for patients recently diagnosed with head/neck cancer. Finally, these results show some day-to-day variability in the TMH variables, which supports assessing these constructs repeatedly in the context of intervention and patient care to explore the potential for change. Second, those providing smoking cessation services and advising patients to quit should talk about the benefits of quitting instead of the risks of smoking, since patients higher in perceived benefits in abstaining smoked fewer cigarettes. Third, interventions to bolster patients’ social support and include supporters/caregivers in cessation treatment plans may help to improve cancer patients’ confidence to quit as well as their tobacco use outcomes. These interventions should capitalize on the immediate post-diagnosis increase in social ­support. Finally, cancer patients in this sample seem to use smoking to cope with cancer worry. Tobacco counseling should focus on introducing other coping strategies and treatments to reduce cancer worry (eg, cognitive behavioral therapy, acceptance and commitment therapy). Although access to ­evidence-based tobacco treatment has improved for cancer patients who are seen at NCI-designated cancer centers,⁴² there are still financial and other practical barriers to receiving these services for patients with low financial resources. In terms of future research directions, studies should focus on other high-risk cancer sites to improve generalizability of findings. ­Additionally, interventions for the 3 TMH components should be developed and evaluated to determine their effect on improving smoking cessation outcomes for cancer patients.

Limitations

Results of this study must be viewed considering its limitations. First, these analyses were determined post hoc, making them neither a definitive nor exhaustive examination of the TMH components in the context of cancer diagnosis. Second, while the measurement of the affective response and risk perception components aligned with those of past studies, social support was used as a proxy for social support/self-concept. There remains a lack of consensus about how to best measure social role/self-concept¹⁰ given that only 2 prior studies have focused on this component of the TMH. Third, this sample consisted only of head/neck cancer patients, which limits generalizability. However, focus on this particularly high-risk group with clear smoking related outcomes is important, and it is unlikely that results would not generalize to other disease sites, at least those that are generally thought to be tobacco-related. Fourth, our sample contained low racial and ethnic diversity, which might further limit generalizability, although research on racial and ethnic differences in the TMH components results in mixed findings.⁴³ More research is needed to determine if racial and ethnic differences exist in the TMH. Finally, this study was hypothesis-generating and observational in design; therefore, all conclusions and implications should be interpreted with caution and viewed in the context of this limitation. ­Additionally, although our lagged analyses provide some information on Granger causality, we encourage future researchers to experimentally manipulate the TMH in intervention to provide stronger tests of causality.

Conclusion

This is the first known study to evaluate the association between key smoking cessation outcomes (eg, motivation and confidence to quit) as well as smoking behavior (eg, cigarettes per day) and the TMH components. Key targets should include cancer-related worry, social support, and risk perceptions, specifically focusing on the benefits of quitting, rather than the harms of persistent smoking.³⁵ More research is needed to further explore the relationship between these TMH components and confidence to quit smoking. In sum, the TMH holds promise for intervention development to help cancer patients quit smoking.

Author contributions

Tia Borger (Data curation [equal], Writing—original draft [lead]), Matthew W. Southward (Data curation [equal], Formal analysis [lead]; Writing—original draft [supporting]), Jessica Maras (Writing—original draft [supporting], Writing—review & editing [supporting]), Abigayle Feather (Writing—original draft [supporting], Writing—review & editing [supporting]), and Jessica L. Burris (Conceptualization [lead], Funding acquisition [lead], Investigation [lead], Methodology [lead], Supervision [lead], Writing—review & editing [lead])

Supplementary material

Supplementary material is available at Annals of Behavioral Medicine online.

Funding

Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under award number K07 CA181351, the Patient Oriented and Population Sciences Shared Resource Facility and Biostatistics and Bioinformatics Shared Resource Facility of Markey Cancer Center under award number P30 CA177558, and the National Center for Advancing Translational Sciences of the National Institutes of Health under award number UL1 TR001998. M.W.S.’s efforts on this project were partially supported by the National Institute of Mental Health under award number K23 MH126211. The content of this publication is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Conflicts of interest

The authors have no conflicts of interest to declare.

Transparency statements

Study Registration: This study was not formally registered. Analytic Plan Pre-Registration: The analysis plan was not formally pre-registered. Analytic Code Availability: Analytic code used to conduct the analyses presented in this study are not available in a public archive. Materials Availability: Materials used to conduct the study are not publicly available.

Data availability

De-identified data from this study are not available in a public archive.

Statement of human rights

All procedures were approved by the University of Kentucky Institutional Review Board and have been performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.