Understanding the Association Between Spontaneous Quit Attempts and Improved Smoking Cessation Success Rates: A Population Survey in England With 6-Month Follow-up

Abstract

Introduction

Almost half of smoking quit attempts are “spontaneous” (initiated as soon as the decision to quit has been made) and are associated with increased success rates. This study aimed to assess to what extent other factors may account for this association.

Methods

Data were used from respondents to a survey representative of the adult population in England from 2006 to 2016. We included 2018 respondents who were current smokers at baseline and had attempted to quit between baseline and 6-month follow-up. Logistic regression models assessed the association between quit success and spontaneous quit attempts while adjusting for smoking, sociodemographic, and quit attempt characteristics.

Results

Spontaneous quit attempts were associated with greater odds of quit success (OR = 1.31, 95% CI = 1.07 to 1.60) but the association was not significant in the fully adjusted model (OR_adj = 1.19, 95% CI = 0.95 to 1.49). In this adjusted model, those who attempted to quit without cutting down first (OR_adj = 3.08, 95% CI = 2.46 to 3.88) and were male (OR_adj = 1.44, 95% CI = 1.16 to 1.80) had greater odds of success; although a greater number of attempts in the past 6 months, stronger urges to smoke (strong vs. none), higher daily cigarette consumption, and lower social grade (E vs. AB) were associated with lower odds of success (OR_adj range = 0.32–0.98, p < .030). Quit attempts made without cutting down first were correlated with spontaneous quit attempts (r = .150, p < .001) and appeared to account for the diminished association between spontaneous quitting and success (OR_adj = 1.18, 95% CI = 0.96 to 1.46).

Conclusions

The increased success rate of spontaneous quit attempts appears to be because spontaneous quit attempts are more likely to be made without cutting down first.

Implications

The apparent benefit of spontaneous over planned quit attempts may be attributable to the former being more likely to involve quitting without cutting down first (ie, abrupt cessation) than cutting down first (ie, gradual cessation) and so this may be a more useful target for advice to improve the chances of successful quitting.

Introduction

Smoking is a major public health problem and causes a wide range of diseases.¹ Almost half (45.8%) of quit attempts are spontaneous, meaning that a smoker tries to quit as soon as they make the decision.² Spontaneous quit attempts have been found to be associated cross-sectionally with a greater chance of success compared with quit attempts that have been planned to take place in the future, even after adjusting for sociodemographic factors.² It remains possible that the difference is due to other unmeasured confounding factors, particularly other attributes of the quit attempt and level of cigarette dependence. This study aimed to assess whether spontaneous quit attempts are associated with increased likelihood of success after controlling for a wider range of potential confounders than has been assessed previously.

It is crucial to understand the factors associated with successful quit attempts as this has important implications for recommendations to individuals trying to quit smoking. Making a spontaneous quit attempt is one such factor associated with improved quit success.^2,3 Spontaneous quit attempts among smokers in England are between two and three times more likely to succeed than planned attempts, after adjusting for age, sex, and socioeconomic status.² This has also been found in Canada, where 51% of quit attempts are spontaneous and successful quit attempts were more likely to be spontaneous³ and a separate retrospective study in Canada found that planned quit attempts did not increase the likelihood of quit success.⁴ However, other important predictors of quit success were not adjusted for in either of these studies. The International Tobacco Control Four Country study did adjust for potential confounders such as cigarette dependence and use of aids during a quit attempt alongside sociodemographic factors and subsequently found no clear effects of spontaneous quit attempts on quit success rates across different waves of the survey.^5,6 However, the International Tobacco Control study found that only 19% of quit attempts were spontaneous,⁵ compared with other studies that have found around half of quit attempts were spontaneous.^2,3 It is unclear why their results differ so markedly.⁵ In the United States, it has been found that delaying a quit attempt prospectively predicted lower quit success.⁷ In addition to this, among moderate and heavy smokers in the United States, the impact of spontaneous quit attempts on quit success interacted with race: among white smokers, spontaneous quit attempts were associated with a longer period of abstinence whereas among black smokers, spontaneous quit attempts were associated with a shorter period of abstinence.⁸ However, no significant differences were detected in the impact of spontaneous quit attempts on abstinence among light or nondaily smokers.⁸ Therefore, more evidence is required on whether spontaneous quit attempts are independently associated with improved success rates after adjusting for potential confounders.

There are a number of known factors that predict quit success, including lower levels of cigarette dependence^9–11 and the strength of urges to smoke.¹² It is important to assess cigarette dependence and strength of urges to smoke prospectively as these can be underestimated if assessed retrospectively among ex-smokers. Quit success is also positively associated with general cessation behaviors such as having made fewer quit attempts in the past year, the time since the attempt was initiated, and quit attempts made without cutting down first.^5,13–15 Quit attempts made without cutting down first are sometimes termed “abrupt,” whereas quit attempts in which the smoker cuts down first are sometimes called “gradual.” Quit attempts made with or without cutting down first can both be part of either a spontaneous or planned quit attempt (see Table 1 for definitions). Use of behavioral or specialist National Health Service support, including pharmacotherapy, are known to improve the likelihood of quit success, compared with unaided quit attempts,^13,16 despite requiring planning. In the United Kingdom, there is evidence that higher social grade is associated with quit success.^10,13 No association between quit success and sex was found in a systematic review,¹⁰ though the effects were heterogeneous across studies. The evidence of an association between quit success and age was inconsistent in a review¹⁰ with some studies finding that quit success is associated with older age.¹³

Table 1.

Definitions of Spontaneous Versus Planned and With Cutting Down First (Abrupt) Versus Not (Gradual) Quit Attempts

Quit attempts	Without cutting down first (abrupt)	With cutting down first (gradual)
Spontaneous (quit at the moment the decision was made)	Attempted to cease all smoking the moment the decision to stop was made	Attempted to begin the process of stopping smoking the moment the decision to stop was made by reducing consumption, or after a period of reduction had already taken place
Planned (for later the same day or in the future)	Attempted to cease smoking in one step but delayed initiation until a later point in time	Attempted to stop smoking by reducing cigarette consumption first, delaying initiation until a later point in time

As there is mixed evidence about whether spontaneous quit attempts are associated with improved success rates after appropriately adjusting for potential confounders, it is important to understand which, if any, of these confounders account for this association. This study used data from a large, nationally representative population survey in England to assess the association between spontaneous quit attempts and quit success, after adjusting for a range of potential confounders.

Research Questions

1. Are spontaneous quit attempts associated with improved success rates among adult smokers in England?

2. Are spontaneous quit attempts independently associated with improved success rates among adult smokers in England, adjusting for other known predictors of success rates?

3. If spontaneous quit attempts are not independently associated with improved success rates, which of the potential confounders account(s) for the association?

Methods

Design and Study Population

The Smoking Toolkit Study is an ongoing, monthly, population survey in England. The Smoking Toolkit Study consists of cross-sectional household surveys of nationally representative samples of 1700–1800 adults (aged ≥16 years) in England.¹⁷ The study sampling is a hybrid of random probability and simple quota—England is split into more than 170 000 areas (consisting of approximately 300 households each) stratified according to a geodemographic analysis of the population. Areas are then randomly allocated to interviewers who conduct interviews within that area until the quota is fulfilled.

The study used data from November 2006 until September 2016 from responders who were current smokers at baseline and had attempted to quit between baseline and 6-month follow-up. There was missing data between April 2012 and August 2014 when 6-month follow-up data were not collected.

The required sample size was 495, based on a power calculation with the following parameters: power equal to .80, α equal to .05, odds ratio (OR) equal to 2.0 (based on results from West and Sohal²).

Measures

Explanatory variable: Spontaneous quit attempt was measured at 6-month follow-up with the question “Which of the following applies to your most recent serious quit attempt?” Respondents had two options: “I started the quit attempt the moment I made the decision I was going to stop” (spontaneous) or “I planned the quit for later the same day or for a date in the future” (planned).

Outcome variable: Quit success was measured at 6-month follow-up by self-reported smoking status with two options: successful or unsuccessful.

Covariates: Quit attempt characteristics were measured at 6-month follow-up and were related to quit attempts between baseline and follow-up. Quit attempt made without cutting down first was assessed with the question “Did you cut down the amount you smoked before trying to stop completely at your most recent serious quit attempt?” and two options: “Stopped without cutting down” or “Cut down first.” Time since start of most recent quit attempt (last week, between a week and a month, 1–2 months, 2–3 months, or 3–6 months) and number of quit attempts in past 6 months (1, 2, or 3 or more) were both categorical variables. Use of evidence-based aids during most recent quit attempt was dichotomous (no or yes); evidence-based aids included prescription nicotine replacement therapy (NRT), NRT over-the-counter, varenicline, bupropion, electronic cigarettes, and face-to-face support.

Smoking and sociodemographic characteristics were measured at baseline. Smoking characteristics measured were strength of urges (categorical: slight, moderate, strong, very strong, or extremely strong) and daily cigarette consumption (continuous; indicating cigarette dependence). Sociodemographic characteristics measured were age (16–24, 25–34, 35–44, 45–54, 55–64, or ≥65), sex, social grade (AB: higher and intermediate managerial, administrative, or professional; C1: supervisory, clerical and junior managerial, administrative, or professional; C2: skilled manual workers; D: semi and unskilled manual workers; or E: state pensioners, lowest-grade workers or unemployed), year (continuous, 2006–2016), and month of survey (continuous, 1: January through to 12: December).

Analysis

All analyses were conducted in R with complete cases for all variables of interest. The protocol and analysis plan were preregistered on Open Science Framework (https://osf.io/g9h3b/). A series of unadjusted binary logistic generalized linear models (GLMs) were conducted to assess the univariate association between the outcome variable (quit success) and each of the explanatory variable (spontaneous quit attempt) and covariates. An adjusted GLM was conducted with the outcome variable (quit success) regressed on to the explanatory variable (spontaneous quit attempt) and all covariates. Continuous variables were transformed by dividing the original variable by the SD to express the variable as a proportion of the SD.

If the adjusted GLM accounted for any unadjusted association between spontaneous quit attempt and quit success, then a series of additional GLMs were planned to identify the key confounder variable. A correlation matrix between the dichotomous explanatory variable (spontaneous quit attempt) and all covariates was inspected to assess the collinearity between variables. The correlation matrix reported tetrachoric correlations for two dichotomous variables (quit attempt made without cutting down first; use of evidence-based aids during quit attempt; sex) and point-biserial for one dichotomous variable and one continuous (daily cigarette consumption; survey year; survey month) or categorical (time since start of most recent quit attempt; number of quit attempts in past 6 months; strength of urges; age; social grade) variable.

The correlation matrix and adjusted GLM informed which covariates were tested for confounding (ie, significantly correlated with the explanatory variable and associated with the outcome variable). Quit success was then regressed on to two variables: (1) spontaneous quit attempt and (2) any covariates identified as potential confounders.

Bayes factors (BFs) were calculated to examine for nonsignificant results whether the associations indicated evidence of no effect or data being insensitive to detect an effect, and for significant results, the strength of evidence.¹⁸ Alternative hypotheses were represented by half normal distributions and the absolute expected effect size for the quit success outcome was set to OR is equal to 2.0 in the observed direction (ie, OR = 2.0 for observed ORs > 1 and OR = 0.5 for observed ORs < 1) based on previous research.^2,3 BFs at least 3 provide evidence for the alternative hypothesis, BFs less than or equal to 0.33 provide evidence for the null hypothesis, and 0.33<BFs<3 indicate the data are insensitive to detect an effect.^19,20

Results

Of 33 646 current smokers with complete data at baseline during the study period, 7302 were followed up (a comparison of the sociodemographic and smoking characteristics of the baseline and follow-up samples are summarized in Supplementary Table 1). Of the 7302 current smokers followed up, a total of 2018 adults aged 16 years and older were included in the analytic sample. Table 2 reports descriptive statistics on the sociodemographic, smoking, and quitting behavior characteristics. Respondents had a mean age of 46 years, over half (54.5%) were female and the most common social grade was E (25.1%). About half of quit attempts were spontaneous (49.0%) and were made without cutting down first (48.9%). The most common time since the start of their most recent quit attempt was 3–6 months (31.4%). The majority of respondents had only made one quit attempt in the past 6 months (69.5%) and almost half reported moderate strength of urges (46.8%). The mean daily cigarette consumption was 13.1 and over half had used evidence-based aids during their quit attempt (58.4%).

Table 2.

Descriptive Statistics for Sociodemographic and Smoking and Quitting Behavior Characteristics

Sociodemographic characteristics
Age, mean (SD)	46.0 (15.96)
Age, % (n)
16–24	11.1 (225)
25–34	15.6 (315)
35–44	20.6 (415)
45–54	20.6 (415)
55–64	17.8 (360)
≥65	14.3 (288)
Sex, % female (n)	54.5 (1100)
Social grade, % (n)
AB	12.9 (260)
C1	23.0 (465)
C2	21.7 (438)
D	17.3 (349)
E	25.1 (506)
Smoking and quitting behavior characteristics
Spontaneous quit attempt, % (n)	49.0 (988)
Quit attempt made without cutting down first, % (n)	48.9 (986)
Time since start of most recent quit attempt, % (n)
Last week	8.6 (174)
Between a week and a month	20.4 (412)
1–2 months	20.5 (413)
2–3 months	19.1 (386)
3–6 months	31.4 (633)
Number of quit attempts in past 6 months, % (n)
One	69.5 (1403)
Two	21.3 (429)
Three or more	9.2 (186)
Strength of urges, % (n)
None	7.6 (154)
Slight	13.5 (272)
Moderate	46.8 (944)
Strong	23.2 (468)
Very strong	6.7 (135)
Extremely strong	2.2 (45)
Daily cigarette consumption, mean (SD)	13.1 (8.89)
Use of evidence-based aids during quit attempt, % yes (n)	58.4 (1178)

Are Spontaneous Quit Attempts Associated With Improved Success Rates Among Adult Smokers in England?

Spontaneous quit attempts were associated with a greater likelihood of quit success (see Table 3). Greater likelihood of quit success was also associated with quit attempts made without cutting down first and time since start of most recent quit attempt (3–6 months compared with last week). Number of quit attempts in past 6 months, strength of urges (strong compared with none) and daily cigarette consumption were all negatively associated with likelihood of quit success. Males were also more likely to have a successful quit attempt and those of social grade E were less likely to have a successful quit attempt compared with those of social grade AB. Survey year and month were both positively associated with the likelihood of quit success. An unplanned sensitivity analysis (conducted following external peer review) in which a log binomial GLM was conducted (see Supplementary Table 2) found no difference in the pattern of associations in the unadjusted models between relative risk and odds ratios.

Table 3.

Unadjusted and Adjusted Association Between Quit Success and Spontaneous Quit Attempts and Other Potential Confounders

		Unadjusted		Adjusted
	Quit succewss, % (n)	OR (95% CI)	p	OR (95% CI)	p	BFHN
Spontaneous quit attempt
Noa (n = 1030)	22.7 (234)
Yes (n = 988)	27.7 (274)	1.31 (1.07 to 1.60)	.010	1.19 (0.95 to 1.49)	.121	0.94c
Quit attempt made without cutting down first
Noa (n = 1032)	14.9 (154)
Yes (n = 986)	35.9 (354)	3.19 (2.58 to 3.97)	<.001	3.08 (2.46 to 3.88)	<.001	>10 000d
Time since start of most recent quit attempt
Last weeka (n = 174)	21.8 (38)
Between a week and a month (n = 412)	22.3 (92)	1.03 (0.67 to 1.59)	.896	0.80 (0.51 to 1.28)	.352	0.78c
1–2 months (n = 413)	21.1 (87)	0.96 (0.62 to 1.48)	.834	0.68 (0.43 to 1.08)	.100	2.18c
2–3 months (n = 386)	23.1 (89)	1.07 (0.70 to 1.66)	.750	0.68 (0.43 to 1.09)	.105	2.00c
3–6 months (n = 633)	31.9 (202)	1.68 (1.14 to 2.52)	.011	0.93 (0.60 to 1.44)	.727	0.39c
Number of quit attempts in past 6 months
Onea (n = 1403)	29.9 (420)
Two (n = 429)	15.6 (67)	0.43 (0.32 to 0.57)	<.001	0.45 (0.33 to 0.62)	<.001	>10 000d
Three or more (n = 186)	11.3 (21)	0.30 (0.18 to 0.47)	<.001	0.32 (0.19 to 0.51)	<.001	6678d
Strength of urges
Nonea (n = 154)	30.5 (47)
Slight (n = 272)	36.8 (100)	1.32 (0.87 to 2.03)	.193	1.27 (0.81 to 2.01)	.306	0.86c
Moderate (n = 944)	24.6 (232)	0.74 (0.51 to 1.09)	.117	0.80 (0.54 to 1.21)	.290	0.81c
Strong (n = 468)	19.4 (91)	0.55 (0.36 to 0.83)	.004	0.59 (0.37 to 0.93)	.022	5.68d
Very strong (n = 135)	20.7 (28)	0.60 (0.34 to 1.02)	.060	0.71 (0.39 to 1.28)	.257	1.16c
Extremely strong (n = 45)	22.2 (10)	0.65 (0.28 to 1.38)	.281	0.88 (0.36 to 2.04)	.777	0.66c
Daily cigarette consumption	n/a	0.79 (0.70 to 0.88)	<.001	0.82 (0.71 to 0.93)	.003	11.44d
Use of evidence-based aids during quit attempt
Yesa (n = 1178)	25.1 (296)
No (n = 840)	25.2 (212)	1.01 (0.82 to 1.23)	.955	0.82 (0.65 to 1.03)	.085	1.25c
Age
16–24a (n = 225)	23.6 (53)
25–34 (n = 315)	28.6 (90)	1.30 (0.88 to 1.93)	.193	1.37 (0.90 to 2.10)	.151	1.36c
35–44 (n = 415)	24.3 (101)	1.04 (0.72 to 1.53)	.825	1.17 (0.77 to 1.78)	.460	0.58c
45–54 (n = 415)	27.5 (114)	1.23 (0.85 to 1.80)	.282	1.35 (0.90 to 2.06)	.153	1.35c
55–64 (n = 360)	25.0 (90)	1.08 (0.73 to 1.60)	.693	1.37 (0.90 to 2.11)	.147	1.36c
≥65 (n = 288)	20.8 (60)	0.85 (0.56 to 1.30)	.461	0.98 (0.62 to 1.55)	.923	0.34c
Sex
Femalea (n = 1100)	23.0 (253)
Male (n = 918)	27.8 (255)	1.29 (1.05 to 1.58)	.014	1.44 (1.16 to 1.80)	.001	58.35d
Social grade
ABa (n = 260)	30.0 (78)
C1 (n = 465)	31.8 (148)	1.09 (0.79 to 1.52)	.610	1.10 (0.77 to 1.57)	.615	0.41c
C2 (n = 438)	25.8 (113)	0.81 (0.58 to 1.14)	.229	0.92 (0.64 to 1.33)	.650	0.38c
D (n = 349)	23.5 (82)	0.72 (0.50 to 1.03)	.072	0.84 (0.57 to 1.24)	.384	0.61c
E (n = 506)	17.2 (87)	0.48 (0.34 to 0.69)	<.001	0.65 (0.45 to 0.96)	.030	5.65d
Survey year		1.14 (1.04 to 1.26)	.008	1.08 (0.96 to 1.20)	.191	0.38c
Survey month (continuous)		1.10 (0.99 to 1.22)	.068	1.14 (1.02 to 1.27)	.020	1.75c

BF = Bayes factor; CI = confidence interval; OR = odds ratio.

^aReference level.

^bEvidence for the null hypothesis.

^cData are insensitive to detect an effect.

^dEvidence for the alternative hypothesis.

Are Spontaneous Quit Attempts Independently Associated With Improved Success Rates Among Adult Smokers in England, Adjusting for Other Known Predictors of Success Rates? If Not, Which of the Potential Confounders Accounts for the Association?

No independent association between spontaneous quit attempts and improved success rates was detected (see Table 3) and the BF calculated indicated that the data were insensitive to detect an effect (BF_HN = 0.94). Quit success was independently positively associated with an attempt made without cutting down first and being male, and the BFs calculated provided extremely strong evidence for these positive associations (BFs_HN > 58.35). Quit success was independently positively associated with survey month though the BF indicated that the data were insensitive to detect an effect (BF_HN = 1.75). A sensitivity analysis (conducted following internal peer review) in which survey month was transformed from a continuous into a categorical (year quarters) variable found no difference in the pattern of associations in the adjusted model.

Quit success was negatively associated with number of quit attempts in past 6 months, strength of urges (strong vs. none), daily cigarette consumption, and social grade (E vs. AB). The BFs calculated provide extremely strong evidence for the association between quit success and number of quit attempts in the past 6 months (BFs_HN > 6678) and provide moderate evidence for the association between quit success and strength of urges (BF_HN = 5.68), social grade (BF_HN = 5.65), and daily cigarette consumption (BF_HN = 11.44).

Spontaneous quit attempts was correlated with quit attempt made without cutting down first (r = .150, p < .001), strength of urges (r = .052, p = .020), daily cigarette consumption (r = .094, p < .001), use of aids (r = .230, p < .001), and social grade (r = −.053, p = .017) (see Supplementary Table 3). Of these variables significantly correlated with spontaneous quit attempts, the following four variables were also independently associated with quit success: (1) quit attempt made without cutting down first, (2) strength of urges, (3) daily cigarette consumption, and (4) social grade. These variables were regressed onto quit success with spontaneous quit attempt in four separate models (see Table 4).

Table 4.

Adjusted Models Between Quit Success and Spontaneous Quit attempts and Potential Confounder

	Adjusted
	OR (95% CI)	p	BF(HN)
Model 1
Spontaneous quit attempt (not spontaneousa)	1.18 (0.96 to 1.46)	.113	0.94c
Quit attempt made without cutting down first (with cutting down firsta)	3.15 (2.54 to 3.91)	<.001	>10 000d
Model 2
Spontaneous quit attempt (not spontaneousa)	1.28 (1.04 to 1.57)	.017	3.86d
Strength of urges
Nonea (n = 154)
Slight (n = 272)	1.32 (0.87 to 2.02)	.199	1.18c
Moderate (n = 944)	0.75 (0.52 to 1.09)	.127	1.46c
Strong (n = 468)	0.55 (0.37 to 0.84)	.005	24.33d
Very strong (n = 135)	0.62 (0.36 to 1.05)	.079	2.73c
Extremely strong (n = 45)	0.66 (0.29 to 1.41)	.302	1.24c
Model 3
Spontaneous quit attempt (not spontaneousa)	1.25 (1.02 to 1.54)	.029	2.16c
Daily cigarette consumption	0.80 (0.71 to 0.89)	<.001	136.82d
Model 4
Spontaneous quit attempt (not spontaneousa)	1.36 (1.11 to 1.67)	.003	31.70d
Social grade
ABa (n = 260)
C1 (n = 465)	1.10 (0.79 to 1.54)	.561	0.40c
C2 (n = 438)	0.81 (0.57 to 1.14)	.216	0.83c
D (n = 349)	0.71 (0.49 to 1.02)	.064	2.25c
E (n = 506)	0.48 (0.33 to 0.68)	<.001	1114.57d

BF = Bayes factor; CI = confidence interval; OR = odds ratio.

^aReference level.

^bEvidence for the null hypothesis.

^cData are insensitive to detect an effect.

^dEvidence for the alternative hypothesis.

Spontaneous quit attempts was not significantly associated with quit success in model one only, where spontaneous quit attempts and quit attempt made without cutting down first were regressed onto quit success, though the BF calculated (BF_HN = 0.94) indicated the data were insensitive to detect an effect. The positive association between quit success and quit attempt made without cutting down first remained in model one and the BF provided extremely strong evidence for the association (BF_HN > 10 000). The significant association between spontaneous quit attempts and quit success remained for models two, three, and four, though the BF for model three indicated that the data were insensitive to detect an effect. An unplanned sensitivity analysis (conducted following external peer review) in which a log binomial GLM was conducted (see Supplementary Table 4) found no difference in the pattern of associations in these adjusted models between relative risk and odds ratios.

Exploratory Analysis

An exploratory analysis was conducted to assess whether there was an association between quit success and spontaneous quit attempts when stratified by whether the respondents quit without or with cutting down first (see Supplementary Table 5 and Supplementary Figure 1). Of the respondents who had quit without cutting down first (n = 986), 54.0% (n = 532) had quit spontaneously and 37.4% (n = 199) of these had quit successfully compared with 34.1% (n = 155) of those who had not quit spontaneously. Of those who had quit with cutting down first (n = 1032), 44.2% (n = 456) had quit spontaneously and 16.4% (n = 75) of these had quit successfully compared with 13.7% (n = 79) of those who had not quit spontaneously. No significant association was detected between spontaneous quit attempts and quit success for respondents who quit without cutting down first (OR = 1.15, 95% CI = 0.89 to 1.50, p = .287) or for those who quit with cutting down first (OR = 1.24, 95% CI = 0.88 to 1.75, p = .222). The BF indicated that the data were insensitive to detect an effect for the association between those respondents who quit either with or without cutting down first (BF_HN range = 0.55–0.89). However, no significant interaction effect was detected in a fully adjusted logistic regression model including an interaction between spontaneous quit attempts and whether the respondent quit without cutting down first (OR = 0.85, 95% CI = 0.54 to 1.33, p = .465).

Discussion

Spontaneous quit attempts were associated with the likelihood of quit success among adult smokers in England, but no association independent of other known predictors of quit success was detected between spontaneous quit attempts and improved success rates. Attempting to quit without cutting down first and being male were both independently, positively associated with quit success. The number of quit attempts in the past 6 months, strength of urges (strong compared with none), daily cigarette consumption, and social grade (E compared with AB) were all independently, negatively associated with quit success. Quit attempts made without cutting down was correlated with spontaneous quit attempts and when both were regressed onto quit success, quit attempts made without cutting down accounted for the univariate association between spontaneous quit attempts and quit success.

About half of quit attempts were spontaneous (49.0%) in this study, which reflects what has been found in other studies.^2,3 Spontaneous quit attempts were associated with quit success in a univariate model, similar to previous research.^2,3 However, no association was detected between spontaneous quit attempts and quit success among adult smokers in England where other known predictors were adjusted for. An exploratory analysis stratified by those respondents who had made a quit attempt without or with cutting down first found no association between quit success and spontaneous quit attempts. This provides further support for the conclusion that making a quit attempt without cutting down is more important for quit success than making a spontaneous quit attempt. However, the BFs for the exploratory analyses indicated that the data were insensitive to detect an effect and so further research in this area is warranted.

This study replicated previous findings that quit success is positively associated with attempting to quit without cutting down first,^5,13,14 and negatively associated with number of quit attempts in the past 6 months,^5,13,14 strength of urges,¹² cigarette dependence,^9–11 and social grade (E vs. AB).^10,13 There were also contradictions between the findings from this study and previous research. This study found a significant positive association between being male and quit success that was supported by the BF calculated, which is not in line with previous research.¹⁰ Therefore, the evidence on whether there is an association between quit success and sex remains inconclusive. This study also found no association between quit success and age, time since quit attempt was initiated and use of evidence-based aids, and the BFs indicated that the data were insensitive to detect an effect. This is contrary to some previous research that found quit success is associated with older age,¹³ longer time since quit attempt was initiated,^5,13,14 and use of evidence-based aids.^13,16 These findings highlight that the evidence is still inconclusive and further research is needed to better understand these associations.

A major strength of this study is that the sample was from a large, nationally representative population survey in England and, therefore, these findings are likely to be generalizable to the population of smokers in England. The sample consisted of individuals who were current smokers at baseline and had made a quit attempt between baseline and the 6-month follow-up survey. Therefore, at the follow-up survey individuals could be current smokers (unsuccessful quit attempt) or ex-smokers (successful quit attempt). To avoid potential underestimation among ex-smokers, cigarette dependence (indicated by daily cigarette consumption) and strength of urges to smoke were assessed prospectively at baseline, rather than retrospectively at follow-up.

A limitation of this study was that there was a reliance on recall data for quitting behavior characteristics, which involved recall of the past 6 months, introducing scope for bias. Another limitation is that it was not possible to compare the analytic sample—respondents at follow-up who had made a quit attempt in the past 6 months—with those who were not followed up and who had made an attempt to stop in the 6 months between baseline and follow-up. Attempts during that period were unknown because respondents were not followed up. The overall process of following up only recontacts 21.7% and results in a sample that differs from baseline (see Supplementary Table 1). However, the pattern of differences between those who had and had not attempted to quit in the 6 months previously were broadly similar at both baseline and 6-month follow-up (see Supplementary Table 4). Another limitation is that smokers may have interpreted the questions relating to spontaneous quit attempts and quit attempts without cutting down differently. Spontaneous quit attempts are those that started the moment the decision to stop was made while planned quit attempts refer to any quit attempt that did not start immediately, either starting later that same day or for a date in the future. It may be that some respondents interpreted the term “planned” as requiring some form of formal preparation for the quit attempt (eg, buying NRT and choosing a quit date), whereas simply waiting, potentially for more favorable conditions, is also defined as a planned quit attempt. Similarly, confusion may arise from spontaneous quit attempts in which the respondent cut down first; these are defined as “attempted to begin the process of stopping smoking the moment the decision to stop was made by reducing consumption, or after a period of reduction had already taken place” (see Table 1). However, some respondents may have classed their spontaneous quit attempt as without cutting down first if a period of reduction had already taken place.

This study has important practical implications in that smokers should be encouraged to quit without cutting down first, that is, “abruptly” as this is associated with greater likelihood of quit success. This recommendation is in line with the National Centre for Smoking Cessation and Training Standard Treatment Programme²¹ in the United Kingdom and United States guidelines, which encourage total abstinence after the quit date.²²

In addition, smokers of social grade E, those who have a higher daily cigarette consumption and those who have made more quit attempts in the past 6 months are likely to require additional support to achieve a successful quit attempt. The adjusted models included the use of evidence-based aids, which covered prescription NRT, NRT over-the-counter, varenicline, bupropion, electronic cigarettes, and face-to-face support. It is possible that the use of different evidence-based aids has different consequences for the relationship between a spontaneous quit attempt and success. A future study with larger numbers could address this issue. Insofar that it were true, changes in the popularity of different aids—such as the rapid rise in use of electronic cigarettes in England around 2011/2012—would likely affect the overall association observed between spontaneous quit attempts and success.²³

About half of quit attempts made among adult smokers in England between 2006 and 2016 were spontaneous. Spontaneous quit attempts were associated with improved rates of quit success though this association was not detected when adjusting for other known predictors of quit success. Attempting to quit without cutting down first was independently associated with increased likelihood of quit success and this association appeared to account for the positive univariate association between spontaneous quit attempts and quit success.

Ethics Approval

Ethical approval for the Smoking Toolkit Study was granted by the University College London Ethics Committee (ID 2808/005). The data are not collected by University College London and are anonymized when received by University College London.

Funding

CG is funded by Cancer Research UK (CRUK: C1417/A22962) and National Institute for Health Research (NIHR) School for Public Health Research. The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care. LS, RW, and JB are funded by Cancer Research UK (CRUK: C1417/A22962). TR has not received any funding in relation to this study. The Smoking Toolkit Study has been primarily funded by Cancer Research UK (C1417/A14135, C36048/A11654, C44576/A19501) but has previously been funded by Pfizer, GlaxoSmithKline, and the Department of Health. The funders played no role in the design, conduct, or analysis of the study, nor in the interpretation or reporting of study findings.

Declaration of Interests

CG has no competing interests. JB has received unrestricted research grants from Pfizer related to smoking cessation. RW has received research funding and undertaken consultancy for companies that manufacture smoking cessation medications. LS is a HEFCE-funded member of staff at University College London. He has received honoraria for talks, an unrestricted research grant, and travel expenses to attend meetings and workshops from Pfizer and an honorarium to sit on advisory panel from Johnson & Johnson, both pharmaceutical companies that make smoking cessation products. He has acted as paid reviewer for grant awarding bodies and as a paid consultant for health care companies. Other research has been funded by the government, a community-interested company (National Centre for Smoking Cessation) and charitable sources. He has never received personal fees or research funding of any kind from alcohol, electronic cigarette, or tobacco companies. TR has received honoraria from Pfizer, Novartis, Glaxo Smith Kline, Astra Zeneca, and Roche as a speaker in activities related to continuing medical education. He has also received financial support for investigator-initiated trials from Pfizer and Johnson & Johnson.