Smoking Behaviour and Beliefs About Smoking Cessation After Bariatric Surgery

A cross-sectional study was performed in a high volume bariatric surgery centre where bariatric surgery has been performed since 2007, from July 2015 until September 2016. The hospital has been accredited as Bariatric and Metabolic Surgery Centre of Excellence by the European Accreditation Council for Bariatric Surgery in 2013 [26]. At least 18 months after surgery, patients who underwent primary or revisional bariatric surgery between the 1st of July 2012 and 31st of December 2013 were eligible for the current study. Preferably, patients were approached when attending the clinic for routine follow-up visit. If patients did not show up, then they were contacted by mail, e-mail or telephone if details were available. All patients in this cohort were advised to stop smoking at least 2 weeks prior to the surgery. They received regular follow-up visits after surgery, after the visits of the first year at least one visit every year.

This study was performed in accordance with the ethical standards of the Helsinki Declaration. All patients provided written consent. The Institutional Review Board of the hospital confirmed that formal ethical review was not required.

After enrolment, patients were asked to complete a written questionnaire. The compiled questionnaire composed of the modified Fagerström test and other questions were derived from the questionnaire based on the ASE-model of STIVORO, the Michigan Alcohol Screening Test, Jellinek’s self-test on addictions, the DSM V criteria and Compulsive Behavior Questionnaire [27‐32]. The questionnaire was tested with cognitive interviewing [33]. Subsequently, we made minor changes that simplified the completion of the questionnaire. The compiled questionnaire is available upon request (English, Dutch). It consisted of 51 questions on general information, current and past smoking, alcohol and addictive drugs use. In this study, we do not report the results on alcohol, drugs and other addictions.

In addition, preoperative and follow-up data were retrieved from medical records. Preoperative characteristics that were extracted from medical records included sex, preoperative weight, preoperative body mass index (weight in kg/(height in meter)²), diabetes mellitus type 2 (DM 2), hypertension, previous abdominal surgery, type and date of bariatric surgery, haemoglobin A1c (HbA1c) and C-reactive protein (CRP). Postoperative characteristics that were collected at the time of questionnaire included remission of hypertension and DM 2, symptomatic gallstones, (history of) reported physical or mental problems due to excess skin (including already performed plastic surgery), reported hypoglycaemia, readmissions related to bariatric surgery or due to abdominal complaints, any reoperation change of HbA1c, change of CRP and % total weight loss (%TWL). Reoperations included laparoscopic cholecystectomy, diagnostic laparoscopy, closure of mesenteric defects (with or without active internal hernia), redo surgery, undo surgery, revisional surgery (for instance, because of bleeding, leakage, stenosis, abscess), surgical neurectomy and incisional hernia repair.

The questionnaires and other data were collected and managed by two researchers (PW and OA). The data were collected and recorded systematically, in accordance with instructions that were formulated in advance. Deviations were discussed until consensus was reached.

With this study, firstly, we aim to estimate the frequency of current smoking, former smoking and never smoking in patients who underwent bariatric surgery at least 18 months ago.

Secondly, in patients with a history of smoking, we aim to compare questionnaire findings on history and thoughts about smoking cessation between smoking behaviour categories.

We also aim to establish the association between long-term outcomes of bariatric surgery and smoking behaviour categories.

Characteristics of patients before surgery were summarized for all patients and by postoperative smoking behaviour group. Characteristics of patients before surgery, questionnaire findings on smoking history and thoughts about smoking and long-term outcomes of bariatric surgery were compared between the postoperative smoking behaviour groups.

Values are expressed as mean and standard deviation (SD) in case of continuous variables that are normally distributed. Median and interquartile range (IQR) are presented in case of non-normal distribution. Categorical variables are expressed as frequencies and percentages.

Categorical variables were compared using chi-square tests. The independent t test (two groups) or one-way ANOVA (more than two groups) were used for normal distributed continuous variables. In case of non-normal distributed data the Mann-Whitney U test (two groups) or Kruskal-Wallis test (more than two groups) were used. Finally, in case of p value < 0.2, group differences were tested separately using chi-square test, independent t test or Mann-Whitney U test as appropriate.

The associations between postoperative smoking behaviour categories and long-term outcomes of surgery were explored using univariable logistic or linear regression. Odds ratios (ORs) with 95% confidence interval (95 CI) or betas with standard errors (SE) were presented. Since this evaluation was not the primary aim of the study, we proceeded only to a multivariable model if there was an indication for a relation for the specific outcome (p < 0.2). In a multivariable model, variables with a p value < 0.4 at univariable analysis were then further explored using multivariable logistic or linear regression (full model). By means of stepwise backward selection, variables with p > 0.05 were eliminated from the model (final model).

A total of 609 (52.0%) patients responded and were included in the study, whereas 557 (47.5%) patients did not respond and 6 (0.5%) patients had died before the inclusion period. Respondents were slightly older than non-respondents (mean (SD) age was 45.9 (9.7) versus 42.2 (10.4) years; p < 0.001, respectively), but the remaining characteristics were not significantly different (Table 1).

Overall, 522 (85.7%) of the patients were female, the mean (SD) age was 45.9 (SD 9.7) years and the median (interquartile range [IQR]) body mass index (BMI) was 42.5 (40.1–46.2) kg/m². Medical history revealed that 242 (39.7%) patients had hypertension, 119 (19.5%) had DM 2 and 328 (53.9%) patients had a previous abdominal surgery, including 67 (11.0% of total population) patients with a previous bariatric surgery. The educational level was reported to be ‘low’ by 54 (9.0%) patients, ‘middle’ by 435 (72.5%) patients and ‘high’ by 111 (18.5%) of all patients.

Table 2 describes and compares the preoperative characteristics of respondents for each smoking behaviour as reported at the time of questionnaire after surgery, separately.

Median (IQR) time after surgery was 3.0 (2.3–3.3) years, when patients completed the questionnaire. After surgery, 101 (16.6%) patients were current smokers, 239 (39.2%) were former smokers (226 (94.6%) of them were already former smoker before surgery), and 269 (44.2%) patients were lifetime never smokers. Figure 1 demonstrates the postoperative smoking status, with information on the smoking behaviour category on the day of surgery.

In Fig. 1 and Table 3, the smoking, smoking cessation, resuming and starting rates after surgery are demonstrated.

Of the current smokers after surgery, 69 (68.3%) had smoked continuously after surgery (by definition > 90%, median (IQR) 100% (97.8–100); not continuously smokers had smoked median (IQR) 48.0% (15.4–78.3) of the time after surgery.

Postoperative current smokers had a median (IQR) smoking history of 13.4 (5.0–22.3) pack-years; continuous smokers 15.0 (6.0–22.5) years and incomplete continued smokers 9.1 (1.5–20.1) years.

Median (IQR) time of not smoking before surgery was 0.5 (0–2.3) months in postoperative current smokers; 0.25 (0–1) months in those who smoked continuously after surgery; and 2.3 (0.6–24) months in not continuous smokers.

The 54 patients (recent or former smokers before surgery) who resumed smoking after surgery had a median (IQR) history of 10.0 (4.4–20.0) pack-years. Median (IQR) time of not smoking before surgery was 1.8 (0.8–12.0) months. The median (IQR) percentage of time that they had smoked after surgery was 90.6% (47.0–99.4).

Thirteen (5.4%) of all former smokers had smoked more than 10% of the time after surgery (median (IQR) 73.7% (51.0–92.3)) but were stopped already when they filled in the questionnaire.

Postoperative former smokers had a median (IQR) history of 15.0 (5.0–30.0) pack-years; this was 4.6 (2.4–12.0) pack-years for those who had a temporary relapse after surgery and 15.0 (5.5–30.0) pack-years in those who continued not to smoke anymore. Median (IQR) time of preoperative cessation was 99.0 (31.5–204.0) months; 1.8 (0.5–11.5) months in those who had a temporary relapse and 108.0 (36.0–204.0) months in those who continued not to smoke anymore.

The 12 (5.0%) former smokers who were recent or current smokers before surgery had a median (IQR) history of 10.0 (3.8–17.3) pack-years. Median (IQR) time of cessation before surgery was 0.3 (0.0–0.5) months. They had smoked a median (IQR) of 0.0% (0.0–80.9) of the time after surgery.

Table 4 demonstrates the ideas about the health consequences of smoking cessation in postoperative current and former smoking patients (based on combination of pre- and postoperative smoking behaviour). Former smokers were more aware of the beneficial effects of smoking cessation on general health.

Also, the result of bariatric surgery after smoking cessation was thought to be ‘much better’ by 67% of former smokers versus 17% of current smokers. Characteristics of current and/or previous quit attempts of all current and former smokers at the time of questionnaire after surgery (based on lifetime history of smoking) are described in Table 5.

Table 6 shows the long-term bariatric outcomes per smoking behaviour category after surgery (based on lifetime smoking history). Mean %TWL in current smokers was 33.6 (SD 8.9). %TWL was 3.4% (SE 1.0) higher in current smokers compared with former smokers and 2.1% (SE 1.0) higher compared with lifetime never smokers (overall p < 0.001). In a multivariable model with adjustments for sex, preoperative BMI, HbA1c before surgery and time after surgery, %TWL in current smokers was 2.8% (SE 0.9) higher compared with former smokers and 2.1% (SE 0.9) higher compared with never smokers (overall p = 0.011).

Remission of DM 2, remission of hypertension, reported hypoglycaemia, readmissions because of abdominal complaints, reoperations related to bariatric surgery, symptomatic gallstones and reported physical or mental hindrance of excess skin were not statistically different between the groups. Univariable analysis of smoking subgroups and CRP change showed an indication to proceed to multivariable analysis. However, in the multivariable model, the association was not statistically significant.

This is the first study exploring the attitudes of smokers and ex-smokers toward benefits of smoking cessation and risks of smoking in bariatric surgery patients. The attitudes and beliefs have been investigated in other populations [12‐19]. We could imagine that the success and willingness to quit are affected by several aspects of which the concern of weight gain after smoking cessation is one, both before and after bariatric surgery. Other factors that could give impact on the willingness to quit are of a social and cultural nature; as in our study we found differences in the experienced support and amount of smokers in social environment between former and current smokers. Future (qualitative) research could deepen the knowledge about (former) smokers, in respect of the contributing or interfering beliefs and factors associated with successful cessation in bariatric surgery populations. Psychosocial theories are used to investigate behavioural change, and smoking cessation in particular [14‐19, 40, 41]. In general, smokers are more likely to deny or rationalize their risk of developing smoking-related diseases (self-exempting beliefs). When smokers experience enough cognitive dissonance, this could discomfit the self-exempting beliefs, thereby inducing smoking cessation attempt. [17, 41]. This could explain the differences in beliefs between smokers and former smokers about the effects of smoking cessation on general health, bariatric surgery and weight loss. The proportion of responses of ‘do not know’ can be explained in several ways; respondents truly did not know, or they were not able to figure it out due to cognitive dissonance. Some factors are known to be associated with poor self-reported health rate, including smoking and, among other things, BMI, gender, age and educational level [42, 43]. It is possible that these factors also influence the attitudes toward health effects of smoking cessation, but we did not adjust for them.

Our study has several limitations. First of all, self-reported smoking can be subject to recall and reporting bias [12, 44]. It is known that reporting bias is greater in situations where quitting expectations on part of the healthcare team are higher [44]. In our experience, these biases are highest before surgery, because continuation of smoking could result in postponement of the surgery. In this study, we used written self-report without interference of the attending doctor. Thus, the (retrospective) design could have been a positive aspect, because patients did not have to fear rejection when they retrospectively reported that they were a current smoker at the day of surgery. Additionally, we previously demonstrated that the sensitivity and specificity of written self-reported smoking behaviour compared with serum cotinine in patients more than 18 months after bariatric surgery were 93.5% and 96.4% [34]. Thus, we expect that the contamination of the ‘former smoker category’ with smokers that reported their smoking status incorrectly was limited. However, associations with smoking could actually be more distinct when we could have prevented any misclassification. Second, the response rate was only 52%. Apparently, the bariatric surgery population is not easily willing to participate in studies concerning smoking behaviour, considering most of the response rates in other questionnaire studies concerning bariatric surgery and smoking were even lower or unclear [21, 24, 31]. Additionally, the percentages of loss-to-follow-up after more than 2 years in bariatric surgery patients are high (26–95.7%), even in studies involving only medical record review, often above 40% [7, 23, 45, 46]. We compared preoperative characteristics of responders and non-responders to check for selection bias, which seems limited. Still, the response rate could influence the representativeness of the data. Patients with specific characteristics may have systematically refused to participate, i.e., based on smoking, the most (un)satisfied, (un)healthy or optimistic patients could have ignored our request to participate. Therefore, we interpret the results with caution, because attrition bias could affect the magnitude and even direction of the associations that we found. Because of the possibility of Berkson’s bias, current findings cannot be externalized to other populations and should first be examined in other bariatric surgery populations.

Third, the questionnaire was developed specifically for this study and not validated. However, we carefully selected questions used in the previous publications and in clinical practice. Also, we used cognitive interviewing to improve the formulating of the questions. Therefore, we are convinced that we can draw reliable and clinically relevant conclusions. Lastly, we included a heterogeneous group in terms of type of bariatric procedures and primary and revisional surgery, which could have influenced our results. On the other hand, in terms of preoperative assessment and postoperative follow-up duration and method, this group was uniform.

In conclusion, 3 years after bariatric surgery, 16.6% of the patients was smoking, and 44.2% had never smoked in their entire life. Former smokers were more aware of the beneficial effects of smoking cessation on the general health and the result of the bariatric surgery compared with current smokers. Actual mean weight loss was somewhat more in smokers compared with former and never smokers. We could not confirm that other long-term bariatric outcomes were significantly different between current, former and never smokers after surgery. The present study emphasizes the need for larger cohort studies with long-term follow-up, investigating better pre- and postoperative strategies to convince smokers to quit and prevent former smokers to relapse. Future research should focus on these strategies, possibly by addressing concerns about weight gain and other reasons interfering with quitting.