At the long-term follow-up 5–8 years after the planned smoking cessation date, 27% of the participants were abstinent. Point prevalence was 31% in HIT and 24% in LIT, of which 26% and 19%, respectively, was 6-month continuous abstinence. Sustained abstinence was maintained by 12% in HIT and 7% in LIT (p=0.03). The 7% difference in abstinence rates between HIT and LIT remained unchanged over time and may be considered clinically significant because of the very large health gains that accrue from stopping smoking and which are attached to cost-savings . The proportion of smoke-free participants had increased by 8% from the 12-month to the long-term follow-up in both treatment arms. Abstinence at the 12-month follow-up was a strong predictor for abstinence at the long-term follow-up. A probable theoretical explanation of the observed difference between the treatment arms is that structured long-term contact may increase the possibilities for being exposed to positive reinforcement and skills training delivered by the counsellor.
Returning to our question in the introduction: how good are 1-year follow-ups as proxies for long-term and life-long abstinence? Other studies report that, among 12-month abstainers, 60-70% remains abstinent for at least eight years [21, 32]. We found that abstinence rates increased by 8% from 12-month to long-term (5-8 yrs) follow-up in both programs, and that abstinence at 12-month follow-up was the strongest predictor of long-term abstinence. However, regarding differences between the programs, we found that the group with point prevalence but <6 months at the 12-month follow-up was crucial; HIT had more abstinent participants and fewer smokers at long-term follow-up compared with LIT. Thus, a possibility could be to offer some kind of renewed support after one year to participants in this group, which needs to be studied further.
As far as we know, there is a lack of non-pharmacological clinical studies with long-term follow-up and, consequently, it is difficult to compare abstinence rates. A cessation program in a specialist clinic in Sweden, with an average of 10 treatment sessions and NRT as a recommended part of the program, shows 40% abstinence after 5–7 years, of which 18% is continuous abstinence . In the comprehensive American Lung Health Study (LHS) with long-term follow-up, smokers with mild airway obstruction are randomized to either an intensive 12-week smoking cessation intervention (SI) including nicotine gum and repeated follow-up visits/contacts or to usual care (UC). Point prevalence abstinence is high and increases in both programs over time, in SI 39% after 5 years and 52% after 11 years, in UC 22% and 43%, respectively. Participants with sustained abstinence for the first 5-year period are very likely to still be abstinent after 11 years, when validated sustained abstinence is 22% for SI and 6% for UC [34, 35]. Blondal et al. report 9% sustained abstinence after six years among smokers in Iceland receiving nicotine patches for five months and 10 individual and group meetings .
In Sweden, 85-90% of the adult inhabitants regularly visit the dental care [25, 37]. The present study, performed in a public dental setting, supports the findings of benefits by using dental professionals for cessation interventions presented in the recent Cochrane-review . The review finds a significant effect of behavioral interventions versus control for adult smokers in general dental practices with 6–12 months of follow-up (OR 2.38, 95% CI 1.70-3.35, no evidence of heterogeneity [I2=3%]). Consequently, the dental setting is a potential public health arena.
The majority of smokers quit and relapse a number of times before they achieve sustained abstinence [5, 6]. The estimated annual incidence of relapse to smoking after one year of abstinence is 2-15% in retrospective studies (with recall problems) and about 10% in prospective studies . Zhu et al. show that multiple compared with single counselling and, especially a self-help intervention, significantly lower the relapse rate during the first week, which in turn translates into a higher 12-month abstinence rate . This is in accordance with our results that significantly more participants in HIT than in LIT maintained sustained abstinence. We were not able to comment on how the participants in the present study possibly quit and relapsed, since we decided not to use the quitting attempts information. The reasons were the long follow-up period, adherent recall bias problems , a considerable internal drop-outs on the question of quitting attempts, and the study design with no measurements between baseline, 12-month and long-term follow-ups.
In the present study, abstinence at the 12-month follow-up was a very strong predictor of abstinence at the long-term follow-up, with ORs between 13.53 and 18.70 in the multivariable logistic regression analyses. Program was the only significant predictor of maintaining sustained abstinence, with an OR of 2.67 for HIT vs. LIT. Among common reported predictors for successful quitting [6, 16, 17], only other support at the 12-month follow-up was a significant predictor for point prevalence at the long-term follow-up in the present study (Table 2).
Women had higher, however not statistically significant, quitting rates than men in the present study. One explanation can be that gender interacts with other support, as women were significantly more likely than men to have access to it. These gender differences in access to social support correspond to population-based data from Sweden assessing availability of emotional support .
We did not find any strong support for the positive long-term effect of NRT use in the present study. No difference was detected in abstinence rates between those who had used NRT ≥5 weeks and those who had not used NRT at all, while those who had used NRT <5 weeks were significantly less likely to be abstinent (Additional file 2: Table S1, Additional file 3: Table S2). However, he study was not designed to assess NRT use as such, so we are not able to draw conclusions regarding the real-life effect of NRT. Abstinence rates at 12 months are significantly improved by NRT use in self-quitters without formal behavioral support , as well as among those using NRT ≥5 weeks at the Swedish National Tobacco Quitline . We did not detect such an effect at the 12-month follow-up in our original study . The contrasts concerning effectiveness of NRT in clinical trials [6, 11, 21] and population studies [12, 41] need further investigation not least in the light of policy and economic coverage decisions.
The role of snus as a substitute for smoking was low (≤8%) in the present study, and was not a significant predictor for abstinence in the multivariable analyses. Snus seems to play an insignificant role for cessation in smokers with professional support [14, 42], however, presumably plays a greater role in self-quitters .
The still smokers had reduced their number of smoked cigarettes significantly, which can be seen as a step in the process towards quitting [29, 44]. And two-thirds of this group intended to make a new quit attempt within the following six months. It is interesting what happened with intention to quit from 12-month to long-term follow-up. Among smokers at the 12-month follow-up who intended to make a new quit attempt within the following six months, 18% were smoke-free at the long-term follow-up. However, among those who did not intend to quit at the 12-month follow-up, 36% were smoke-free at the long-term follow-up (p=0.001).
The generalizability of a clinical trial is an important issue. For the present study we approached adult daily smokers fluent in the Swedish language and invited them to participate. Those interested in smoking cessation were then randomized . This means that they were more motivated to quit than smokers in the general population, but less motivated than, for example, smokers who call a quitline [14, 45], and who have themselves taken contact with a smoking cessation expert. Smoking ≤10 cigarettes/day is a criterion reported to lead to exclusion for a large percentage of individuals in clinical trials , but this was not applied in the present study since it would have reduced the generalizability.
Strengths of the present study are: i) the randomized controlled trial-design, ii) the long-term follow-up period, and iii) the high response rate. Limitations are: i) the lack of chemical validation of abstinence and ii) the risk of memory bias. However, self-reports are considered accurate in most smoking cessation studies  and as participants were free to use snus and NRT, it would be problematic to distinguish low levels of smoking and use of snus or NRT. Furthermore, we had no reason to assume a different distribution of untruthful answers in the two arms. The risk of memory bias increases with length of follow-up and we noted discrepancies between baseline, 12-month, and long-term follow-up questionnaires regarding information of last date of smoking and number of smoke-free days for 22 participants in HIT and 17 in LIT (16% of all responders).