INTRODUCTION
Current US smokers overwhelmingly want to quit (68.8 %), and most make at least one quit attempt each year (52.4 %), yet they rarely achieve sustained abstinence (6.2 % per year).
1 As a result, the prevalence of smoking in the US has plateaued at approximately 18 % of adults.
2 Evidence-based smoking cessation therapies such as medication and counseling significantly increase the success of quit attempts,
3,
4 but these therapies are underutilized.
1,
5 Current models of care for tobacco cessation treatment rely on highly motivated smokers to initiate therapy (e.g., by calling state telephone quitlines) or on clinical providers to offer therapy. The US Clinical Practice Guideline
6 instructs providers to offer active therapy only to smokers who are “willing” to quit in the next 30 days. The guideline recommends that smokers who are not ready to quit receive brief motivational interventions (e.g., motivational interviewing) to enhance readiness to quit.
The theoretical justification for evaluating smokers’ readiness to quit prior to offering therapy is rooted in the transtheoretical model (TTM).
7 This model describes progression through five stages of change (SOC) (precontemplation, contemplation, and preparation for current smokers; action and maintenance for those who have quit) that correlate with ten processes of behavior change. Hundreds of published validation, population, and intervention studies have evaluated the TTM in the context of tobacco use.
8,
9 According to the TTM, “action-oriented” interventions such as cessation pharmacotherapy are most effective in the advanced stages.
10‐
12 Unfortunately, 80 % of U.S. smokers have historically fallen into the precontemplation and contemplation stages.
13,
14 As motivated smokers have quit in response to public health campaigns and policy initiatives,
15 the proportion of smokers in preparation has dropped even further, with levels now at only 9–12 %.
16,
17 Efforts to help early-stage smokers transition to higher stages through motivational interviewing have produced mixed results.
18 As the percentage of smokers in preparation shrinks, the practice of offering active therapy only to those preparing to quit will have diminishing returns.
While the TTM has served as a useful framework for understanding behavior change, it has limitations as the basis for clinical practice guidelines. The TTM systematically underestimates smokers’ motivation to quit,
19‐
21 as many, if not most, precontemplators and contemplators both want to and try to quit.
22 In fact, several interventions have documented successful abstinence among precontemplators and contemplators.
23‐
25 These results substantiate critiques of the construct validity
21,
26‐
28 and inherent instability of the TTM stages.
21,
29 Interventions that proactively offer evidence-based smoking cessation therapies to all smokers, regardless of SOC, may provide an opportunity to reduce the prevalence of smoking.
30
The Veterans Victory Over Tobacco Study randomized smokers to usual care or to a proactive, population-based tobacco cessation intervention that offered telephone or in-person counseling, as well as access to cessation medications, to smokers regardless of SOC. The primary results revealed a statistically significant higher population-level 6-month prolonged smoking abstinence rate at 1 year for proactive care (13.5 %) compared with usual care (10.9 %, p = 0.02).
31 In this secondary analysis, we evaluate the effectiveness of proactive care among smokers at different baseline SOC. Our primary question is whether a proactive outreach intervention will increase prolonged abstinence even among those who say they are not ready to quit. Secondary outcomes include the uptake of cessation therapies and quit attempts by smokers at each SOC.
RESULTS
Of the 5123 eligible, randomized participants, 3006 provided complete baseline survey data, including the RQL, and thus constitute the sample for this secondary analysis (58.5 % [1473/2519] of those randomized to the proactive care intervention and 58.9 % [1533/2604] of those in usual care). At baseline, 781 smokers were in precontemplation (26.0 %), 1148 were in contemplation (38.2 %), and 1077 were in preparation (35.8 %) (Table
1).
Table 1
Participant Characteristics by Baseline Stage of Change
All participants | 3006 | 781 | 1148 | 1077 | – |
Treatment group | |
Usual care | 1533 (51.3) | 396 (50.6) | 585 (50.7) | 552 (52.5) | 0.64 |
Proactive care | 1473 (48.7) | 385 (49.5) | 563 (49.2) | 525 (47.5) |
Demographic characteristics: |
Age (years) | 57.7 (10.6) | 59.4 (10.3) | 57.0 (10.7) | 57.3 (10.5) | <0.001 |
Race | |
White | 1852 (67.2) | 543 (74.4) | 737 (70.0) | 572 (58.4) | <0.001 |
Black | 774 (22.2) | 154 (16.4) | 280 (20.2) | 340 (29.0) |
Hispanic | 85 (4.5) | 41 (4.2) | 59 (3.6) | 85 (5.8) |
Other | 95 (6.1) | 43 (5.1) | 72 (6.2) | 80 (6.8) |
Gender | |
Male | 2838 (94.4) | 741 (94.9) | 1080 (94.2) | 1017 (94.1) | 0.71 |
Marital status | |
Married | 1460 (50.0) | 337 (44.8) | 602 (53.6) | 521 (50.0) | 0.001 |
Socioeconomic status: |
Income ($) | |
< 10,000 | 511 (17.0) | 123 (16.1) | 185 (16.0) | 203 (18.9) | 0.001 |
10,000–20,000 | 879 (30.8) | 238 (32.8) | 307 (28.1) | 334 (32.4) |
20,001–40,000 | 841 (30.5) | 841 (29.7) | 207 (30.4) | 323 (31.3) |
≥ 40,001 | 593 (21.6) | 155 (21.4) | 266 (25.5) | 172 (17.4) |
Social and environmental pressures: |
Home smoking rules | |
Not allowed anywhere | 1116 (41.1) | 237 (33.3) | 406 (39.8) | 473 (48.6) | <0.001 |
Allowed some places/times | 602 (20.7) | 142 (19.2) | 219 (20.0) | 241 (22.3) |
Allowed anywhere | 1087 (38.2) | 354 (47.5) | 438 (40.2) | 295 (28.7) |
Friends who smoke | |
None | 454 (16.2) | 116 (15.8) | 171 (16.2) | 167 (16.5) | 0.001 |
< half | 809 (28.7) | 185 (24.6) | 296 (27.9) | 328 (32.7) |
About half | 614 (22.2) | 163 (22.3) | 238 (22.7) | 213 (21.7) |
> half | 530 (19.4) | 134 (18.8) | 217 (21.0) | 179 (18.0) |
All | 384 (13.6) | 130 (18.6) | 135 (12.2) | 119 (11.2) |
People important to me want me to quit smoking |
Strongly disagree to neutral | 580 (21.6) | 267 (37.8) | 188 (17.9) | 125 (13.2) | <0.001 |
Somewhat agree | 628 (22.8) | 196 (26.7) | 234 (23.1) | 198 (19.3) |
Strongly agree | 1569 (55.6) | 263 (35.5) | 635 (59.0) | 671 (67.5) |
Smoking behaviors: |
Cigarettes per day | |
≤ 10 | 1018 (31.3) | 211 (24.7) | 290 (22.3) | 517 (46.7) | <0.001 |
11–20 | 1286 (45.0) | 341 (46.3) | 547 (49.5) | 398 (38.9) |
≥ 21 | 652 (23.7) | 213 (29.0) | 301 (28.3) | 138 (14.4) |
Time to first cigarette, min | |
≥ 31 | 809 (25.8) | 185 (23.2) | 249 (20.7) | 375 (33.6) | <0.001 |
6–30 | 1535 (52.6) | 402 (52.2) | 615 (55.1) | 518 (50.1) |
< 5 | 638 (21.6) | 188 (24.6) | 278 (24.2) | 172 (16.3) |
Quit in past 12 months | |
Yes | 1673 (54.4) | 181 (22.4) | 626 (53.2) | 866 (80.6) | <0.001 |
Longest quit length | |
Never quit | 255 (8.3) | 128 (16.3) | 72 (6.1) | 55 (4.7) | <0.001 |
< 1 month | 824 (27.1) | 230 (29.5) | 334 (27.9) | 260 (24.2) |
1–6 months | 809 (27.2) | 187 (24.4) | 309 (27.5) | 313 (28.9) |
> 6 months | 1092 (37.5) | 225 (29.8) | 426 (38.5) | 441 (42.2) |
Within precontemplation and contemplation, observed baseline characteristics were balanced across treatment groups (Tables
3 and
4, available online). However, for the preparation group, male gender and level of agreement with the statement “People important to me want me to quit smoking” were not balanced across treatment groups (Table
5, available online). These imbalanced variables were included in the complete case and the NMAR models as adjusting covariates.
Primary Outcome: 6-Month Prolonged Abstinence (Table 2)
Table 2
Primary Outcome: 6-Month Prolonged Abstinence by Baseline Stage of Change and Treatment Arm
Precontemplation | 18 (5.6 %) | 15 (5.3 %) | 0.9 (0.5, 1.9) | 1.2 (0.8, 1.8) |
Contemplation | 31 (6.5 %) | 48 (11.0 %) |
1.8 (1.1, 2.8) |
1.8 (1.3, 2.4) |
Preparation | 60 (13.1 %) | 82 (21.1 %) |
1.8 (1.2, 2.6) |
1.6 (1.4, 2.0) |
Six-month prolonged abstinence at 1 year varied by baseline SOC (5.4 % for precontemplators, 8.6 % for contemplators, and 17.1 % for preparers [p < 0.001]). The overall interaction between SOC and treatment arm was not statistically significant (p = 0.30). Among smokers in preparation, those randomized to the proactive intervention were more likely to quit than those in usual care (21.1 % vs 13.1 %, respectively,
p = 0.003). Logistic regression mixed modeling analysis, taking into account treatment arm and facility as well as the adjusting covariates described above, found a significant effect of proactive care compared with usual care among preparers (OR, 1.8 [95 % CI, 1.2–2.6]). Similarly, smokers in contemplation who were randomized to the proactive intervention were more likely to quit than those in usual care (11.0 % vs 6.5 %, p = 0.018; OR, 1.8, [95 % CI, 1.1–2.8]).). Smokers in precontemplation quit at similar rates in the two treatment arms (5.3 % vs 5.6 %, p = 0.85; OR, 0.9, [95 % CI, 0.5–1.9]). Analyses accounting for nonresponse using likelihood-based NMAR models showed a similar effect of the proactive care intervention on prolonged abstinence at each SOC (Table
2). Of the 254 study participants who achieved 6-month prolonged abstinence, 55.6 % began the study in preparation, while 44.4 % were not ready to quit at baseline (12.5 % began in precontemplation and 31.9 % in contemplation).
DISCUSSION
This proactive population-based smoking cessation intervention included smokers at all stages of change (SOC), and increased long-term prolonged abstinence for smokers in both preparation and contemplation. Smokers in precontemplation did not quit at higher rates in response to proactive outreach, but they were more likely to try evidence-based cessation interventions, including telephone counseling and combined counseling and pharmacotherapy. Proactively offering evidence-based cessation therapies to all smokers led to increased therapeutic engagement and higher long-term population-level quit rates.
Our assessment of an overall interaction between SOC and treatment group revealed no statistically significant difference among SOC subgroups with respect to the intervention. Interaction tests are often underpowered and may be biased when the study is not designed to detect these subgroup effects and randomization is not stratified by subgroup. Thus, we presented additional stratified analyses demonstrating the effect of proactive outreach on smokers at each baseline stage. Among highly motivated smokers (preparation), 73.5 % of whom made at least one quit attempt during follow-up, proactive outreach increased the likelihood of quitting successfully by 50 % (21.1 % vs 13.1 %). Though proactive outreach targets less motivated smokers,
37 we found that even highly motivated smokers benefited from proactive outreach through increased uptake of cessation therapies.
Proactive outreach to contemplators appeared to increase cessation by helping smokers overcome high nicotine dependence through the use of evidence-based therapies. Contemplators had attempted to quit in the past at rates similar to preparers, but had much higher levels of nicotine dependence (Table
1). A history of past quit attempts is a predictor of future attempts, but nicotine dependence levels predict the success of those attempts.
38,
39 Increased therapeutic engagement by contemplators resulted in more successful quit attempts. Our finding that treatment engagement was associated with cessation (OR 1.55 [95 % CI 1.06–2.28] for use of combined therapy) supports this proposed mechanism.
Although proactive outreach increased treatment engagement among precontemplators, we did not observe a difference in smoking cessation rates. It may be that proactive outreach that offers standard cessation therapies is ineffective for precontemplators, who face greater barriers to quitting
40 and may require tailored or high-intensity therapy.
23 Alternatively, our analysis may have been underpowered to show a difference in cessation among precontemplators, given the smaller size of the subgroup and lower baseline quit rates. Further research is needed for a definitive answer to this question. We found no evidence that the proactive intervention increased unsuccessful quit attempts among precontemplators (31.4 % of precontemplators in usual care and 31.3 % in proactive care made at least one quit attempt during follow-up [
p = 0.98]), and thus have no reason to believe that proactive care poses harm to precontemplators.
Our results replicate and extend prior research that has evaluated population-based smoking cessation interventions for smokers at all SOC. Since 1995, when Curry and colleagues first demonstrated the potential effectiveness of telephone-based interventions with non-volunteer smokers at all SOC,
25 dozens of telephone-based trials in various populations have supported this proactive approach. However, most have measured only point-prevalent abstinence and/or found short-term (6–9-month follow-up) effects.
37,
41,
42 In contrast, our proactive care intervention achieved prolonged abstinence at long-term (12-month) follow-up among both preparers and contemplators. One reason for this robust effect may be that we combined population-based outreach (using electronic health technology to identify current smokers and to offer telephone-based care) with individual care management (linking care to VA providers to facilitate pharmacotherapy). Smokers in proactive care were much more likely to use combined counseling and pharmacotherapy (Fig.
1), which has been shown to be highly effective.
4,
6
While earlier telephone counseling studies have included smokers at all SOC, those studies did not provide information as to whether smokers at lower SOC benefitted from the interventions or merely diluted the population-level treatment effect. In 2015, Haas and colleagues
43 addressed this deficiency, reporting that their telephone-based intervention using interactive voice response increased abstinence both among those planning to quit within the next 30 days and among those with no plans to quit. Our report provides additional information on the magnitude of the treatment effect on less motivated smokers and divides less motivated smokers into subgroups of precontemplators and contemplators.
Interest in treating smokers at all SOC has grown over the past decade. In 2005, Pisinger and colleagues first reported the success of the Inter99 trial, which found that a high-intensity intervention could engage less motivated smokers and increase rates of abstinence.
23,
44 Several editorials, citing the success of Inter99, have questioned current guideline recommendations to assess motivation to quit prior to offering cessation therapy.
26,
45 Aveyard and colleagues conducted a systematic review and meta-analysis of brief physician interventions among smokers at all motivation levels and concluded that combining physician advice to quit with the offer of cessation support motivated an additional 40–60 % of smokers to attempt to quit, compared with advice alone.
30 The authors suggested that offering assistance in quitting to smokers at all SOC may be effective because the offer itself increases confidence in success. To explain why less motivated smokers often respond to offers of cessation therapy, others have advanced an alternative theory of smoking cessation based in “catastrophe theory,” which posits that motivational tension fluctuates, and small triggers can induce apparently spontaneous quit attempts.
46,
47
Carpenter and colleagues subsequently reported that among unmotivated smokers, NRT sampling was more successful for inducing quit attempts and short-term cessation than practice quit attempts alone.
48 Two additional small studies compared the offer of NRT to usual care and found promising short-term results regardless of motivation.
49,
50 After more than a decade of small studies, meta-analyses, and editorials, we now report on the largest low-intensity, pragmatic, proactive care intervention with long-term outcomes among smokers who were not planning to quit at baseline.
Limitations include a study population of mostly older male US veteran smokers, which may limit generalizability to other populations of smokers. Our study, a non-pre-specified subgroup analysis, is restricted to those participants who completed the baseline survey questions that established SOC, and thus this sample may be more engaged than the overall population. Follow-up data availability is limited to an even smaller group. We address the potential for response bias in follow-up data by including a likelihood-based NMAR model analysis. In the primary analysis,
31 we also addressed possible differential non-response between treatment groups, and found that taking into account non-response bias did not substantially alter the results. Additional limitations result from inconsistencies in operationalizing the TTM across the literature, making it difficult to compare our SOC with those in other studies.
9
This large, pragmatic randomized trial of a telephone-based intervention demonstrated increased uptake of smoking cessation therapies and prolonged abstinence at 1 year both for smokers who were already planning to quit and for those who were not. Similar to the results of prior studies,
23,
25 we found that among participants who quit successfully, nearly half began the study stating that they were not ready to quit. Restricting therapy to only those in the preparation stage would exclude 64 % of the smokers in our sample, and 44 % of those who quit. Our results add to the growing body of evidence that smoking cessation therapy should be proactively offered to all smokers, regardless of stated plans to quit.