Refraining from smoking shortly before lobectomy has no influence on the risk of pulmonary complications: a case-control study on a matched population†

Abstract

INTRODUCTION

Recommending smoking cessation prior to anatomical lung resection has been controversial, especially when considering the best time to do it [1–3]. Different authors have stated that smoking cessation improves prognosis in patients with early stage non-small-cell lung cancer (NSCLC) [4]. This observation should be enough to encourage every patient diagnosed with NSCLC to quit smoking, especially taking into account that those who have stopped smoking before surgery have a greater chance of maintaining more abstinence over time [5]. Although the improvement in the survival rate survival is clear [4, 6], whether or not surgery has to be postponed until the patient quits smoking remains controversial [7, 8]. Different single- and multi-centre retrospective studies have shown contradictory results regarding the postoperative risk of pulmonary resection in smokers [1–3, 2, 9]. Although some authors [1–3] and guidelines [7] recommend abstinence from tobacco for at least 2–4 weeks prior to the intervention, there is no definite evidence of the advantages of postponing surgical treatment to achieve smoking cessation [2, 8, 9].

In this study, we evaluated whether active smoking at the time of surgery increases the risk of postoperative pulmonary complications (PPCs) compared to abstention shortly before the procedure.

METHODS

Studied population

We reviewed a prospectively recorded database of 2139 consecutive patients who underwent anatomical lung resection for any indication in our institution between January 1994 and May 2015. To control for possible selection bias and to obtain a homogeneous population, we excluded 576 patients from the study because they underwent pneumonectomy (300 patients), bilobectomy (140 patients) or anatomical segmentectomy (111 patients). Of the remaining 1588, 74 were excluded because they needed reoperation in the immediate postoperative period due to technical complications. Afterwards, 9 patients were excluded because their operations included extensive chest wall resection and reconstruction. Because we previously demonstrated the effect of intensive chest physiotherapy on the rate of PPCs [10], none of the patients operated before November 2002, when our chest physiotherapy program was implemented, were included in the analysis (734 patients). Among the remaining 854 patients, 476 who were never smokers or ex-smokers of more than 16 weeks (prior to the surgical intervention) were also excluded from the study (Fig. 1). The final series comprised 378 patients who were matched according to the variables directly related to the risk of PPCs [age, body mass index, forced expiratory volume in 1 sec (FEV1%), FEV1/forced vital capacity, type of surgical approach and NSCLC diagnosis versus any other] [7, 11]. Those patients who were active smokers at the time of the operation comprised the cases; the controls were those patients who quit smoking up to 16 weeks before the operation.

Figure 1

Inclusion criteria: CONSORT flow diagram.

Open in new tabDownload slide

Perioperative management

Before the operation, the workup included physical examination, haematological and biochemical tests, electrocardiograms, chest radiographs, computed tomographic scans of the chest and abdomen and bronchoscopy when needed. Other tests were performed only if previously obtained results suggested the need.

Pulmonary function tests were routinely indicated for all patients. In this study, we analysed only FEV1% values according to age, gender and height of the patient. Although measurement of the diffusing capacity for carbon monoxide (DLCO) is currently mandatory in our department prior to any anatomical lung resection, it was not required at the beginning of the study. For this reason, we did not include DLCO values in this investigation. When needed, bronchodilator therapy was optimized, but no mucolytic treatment was prescribed perioperatively.

All of the patients with a history of relevant cardiovascular disease were referred to a cardiologist for further assessment. As a common practice, coronary stenosis was stented or by-passed before lung surgery.

Perioperative management, which has been described previously [12], was uniform for all patients throughout the study period. Briefly, all cases were anesthetized and operated on under the supervision of the same team of senior cardiothoracic anaesthetists and thoracic surgeons, respectively. The surgical approach varied among video-assisted thoracic surgery (VATS) and minimally invasive axillar or muscle-sparing mini thoracotomies, depending on the preferences of the surgeons. Systematic nodal dissection was performed according to the guidelines of the European Society of Thoracic Surgeons (ESTS) [13]. The prophylactic antibiotic regimen comprised a single dose of cefuroxime 1500 mg that was repeated 6 h later if the operation continued. All of the patients were extubated in the operating room; after 6 h in the recovery room, they were transferred to the cardiothoracic ward. Postoperative analgesia was achieved through an epidural catheter with bupivacaine and fentanyl infusion during the first 3 days and with oral paracetamol and nonsteroid anti-inflammatory drugs thereafter. Nursing care was homogeneous in all cases and included incentive spirometry and early mobilization. All of the patients participated in our specific pre- and postoperative chest physiotherapy intensive program [14]. This program is led by a dedicated chest physiotherapist who works with the patients in specific facilities located in the cardiothoracic ward. Routinely, chest physiotherapy was started the day before the surgical procedure, resumed in the recovery unit after the intervention and continued twice daily until the patients were discharged. Under the supervision of the chest physiotherapist, the patients were encouraged to exercise on an ergometric bicycle and ascend and descend hospital stairs. Also, they were taught to perform deep inspiratory manoeuvres and to produce an effective cough. Arm and shoulder exercises were indicated to prevent upper arm dysfunction.

When the abstinence syndrome developed, postoperative nicotine was administered either orally or transdermally. Other medications were given depending on the needs of the patient.

Variables and outcomes

The outcome selected as the dependent variable was the postoperative occurrence of the following PPCs: pulmonary atelectasis requiring bronchoscopy, pneumonia or both. The diagnosis was considered positive if any of the postoperative chest radiographs showed consolidation or atelectasis; when leucocytosis (white blood cells over 11000 × 10⁶/L) and fever (body temperature > 38ºC) were of pulmonary origin according to the attending surgeon; when a positive sputum culture was present or there was a change in the quality of the sputum compared to the preoperative sputum; and, finally, if respiratory insufficiency (SpO₂ < 90% on room air) was present as measured by peripheral pulsioximetry [15].

Definitions for each type of postoperative complication were agreed in advance by all team members, and all adverse events were recorded in real time when the patient was discharged. The completeness and accuracy of all records were audited by a data manager twice: at the time of the patient’s discharge and before discharge, when a report was sent to the patient or his referring physician.

The independent variable included in the analysis was smoking status. Patients who had stopped smoking up to 16 weeks before surgery were included in the control group and patients currently smoking at the time of surgery were considered cases.

Propensity score-matched analysis

To control for the influence of variables related to the occurrence of PPCs in both groups, we matched the patients according to the following variables: age, body mass index, FEV1%, FEV1/forced vital capacity, type of surgical approach (VATS or minimally invasive thoracotomies) and NSCLC diagnosis versus other [7]. In this way we obtained a quasi-experimental study using the propensity score method [16].

We used the nearest neighbour 1:1 matching method without replacement. Patients were stratified by propensity score groupings to evaluate the different PPCs rates according to their smoking status. Preoperative values of the patients grouped according to their smoking status were compared by standardized differences. After matching the cases, 134 pairs of patients were selected. The population characteristics before and after the matching are shown in Tables 1 and 2.

Estimation and comparison of postoperative pulmonary complication rates

The prevalence of PPCs was determined for the whole series and for the matched population. Finally, both groups were arranged in 2 × 2 tables and the odds ratio (OR) and its 95% confidence intervals (CIs) for PPCs were calculated. The χ² test was used to assess the probability that differences were due to smoking status.

All calculations were performed using Stata/IC 15 (StataCorp, Texas, USA). Propensity scoring matching was performed using the Leuven and Sianesi psmatch2 routine for Stata (http://fmwww.bc.edu/repec/bocode/p/psmatch2.ado).

RESULTS

Of the 378 patients included in the study, 313 (83%) patients were operated for NSCLC and 65 (17%) for other malignancies. The overall prevalence of PPCs was 4.7% (18 patients in the whole series). In the active smokers group, the prevalence of PPCs was 5.3% (13 patients out of 244); among the ex-smokers, the prevalence was 3.7% (5 patients out of 134) (Table 3). The propensity-matched score identified 134 pairs of patients (cases and controls) who differed according to smoking status. After the matching, the prevalence of PPCs in smokers and ex-smokers was 4.5% (6 out of 134) and 3.7% (5 out of 134), respectively. The OR was 1.21 (95% CI: 0.29–5.13, P = 0.76) (Table 4).

DISCUSSION

The relationship between NSCLC and smoking has been well established [6], as has the survival benefit obtained in those patients diagnosed with NSCLC and early-stage small cell lung cancer who never smoked [17–19] or who quit smoking [6] after the diagnosis. However, controversy remains regarding the benefits of smoking cessation prior to anatomical lung resection. Some studies have even shown a paradoxical increase in complications in patients who quit smoking a few weeks before surgery [20], although other authors were not able to reproduce those findings [3].

In some institutions, smoking cessation of at least 2–4 weeks is recommended prior to any anatomical lung resection, because in their series, smoking status was revealed as an independent risk factor for the development of PPCs [21–23].

Some studies do not show an increased risk of PPCs in patients who are smokers at the time of surgery [1, 2]. Even when studying the changes in the pulmonary function tests of smokers and ex-smokers 1 year after the intervention, they were not able to find any statistical differences [8]. Because there is not enough evidence to define a proper time to quit smoking prior to anatomical lung resection, we have chosen an arbitrary cut-off point based on the median time between the appearance of the first symptoms of the disease and its treatment [24] and on the cutoff points proposed by Bluman et al. [1], Mason et al. [2], Barrera et al. [3], Groth et al. [8] and Raupach et al. [9].

Despite the difference in our cutoff point compared with those reported in these studies, we observed that our complication rates in both smokers and ex-smokers were equal to or lower than those reported by these groups. We also found the following differences in our studied population: (i) We did not include in our analysis non-anatomical lung resections, as did Barrera et al. [3]. (ii) Although Mason et al. [2] considered the in-hospital mortality rate, we did not take it into account in our analysis because of the low prevalence of in-hospital mortality found in our series (two patients in total, one in each group). (iii) Indeed, because most of our long-term follow-up was done via phone calls either to the patient or to the referring physicians [12], we did not consider these data because the better prognosis in lung cancer in those who never smoked has already been proved [17–19], and differences between ex-smokers and current smokers at the time of the operation do not seem to be significant over time [8, 25]. The methodological limitations related to the heterogeneity of the studied populations and the retrospective data collection have been mitigated with the propensity score matching methods, which makes our analysis a quasi-experimental design.

Unfortunately, although DLCO is well established as an independent prognostic factor of PPCs [3, 7], we could not use it as a covariate in our propensity score matched analysis due to the lack of routine DLCO measurement in our cases before publication of the last European Respiratory Society (ERS)-ESTS guidelines [7].

In conclusion, the importance of achieving an early curative anatomical lung resection and the lack of strong evidence of an increased risk of PPCs in patients who smoke, which our study reinforces, does not seem to justify a delay in the surgical intervention while waiting for the patient to stop smoking or to be referred to a tobacco abstention program. However, it is important to remember that smoking cessation should be encouraged at any time prior to lung resection, especially in those patients diagnosed with NSCLC.

Conflict of interest: none declared.

REFERENCES

APPENDIX. CONFERENCE DISCUSSION

Dr H. Batirel(Istanbul, Turkey): This nice study addresses an important clinical issue that will affect decision-making during our daily practice. It is a case control study which spans a 20-year period. There are 134 patients in each of the propensity-matched groups, as you said, and you should be commended for a very low pulmonary complication rate, which is less than 5%.

I have four questions. Why did you include only pneumonia and atelectasis in the complications? Prolonged air leak and failure to expand is also a complication of poor pulmonary function.

Secondly. how many patients were discharged with Heimlich valves if they had prolonged air leak? Was there any difference between the two groups in terms of extubation in the OR, cardiac arrhythmias, and epidural use? In terms of pain management, was this the same in both groups?

The third question is also important. If patients are actively smoking and you operate on them, some of the patients have difficulty with abstinence after surgery. How did you manage that? I think you didn't allow patients who were active smokers to smoke, because they would want to smoke after surgery. Did you use nicotine patches and did you have any patients who had delirium after surgery because of abstinence?

Finally, you have a very low rate of pulmonary complications, 5%, almost less than half of what is reported in the literature. Do you think that your study is powered enough to show a single-digit difference between the pulmonary complications? Obviously the controversy in the literature shows that the difference in the rate of pulmonary complications after surgery between active smokers and ex-smokers is probably small: is your study powered enough to detect this small difference?

Dr Rodriguez: I will start by answering your last question on whether our study is powerful enough. We have selected a large population. We have studied more than 2000 patients. If you ask me about a patient who is 85 years old with a low pulmonary function test and unable to follow instructions or to mobilize early after the surgical procedure, I would say such patients should be considered carefully in order to schedule their surgery. But if I were asked about the vast majority of patients undergoing lobectomy, I would say the results of this study are perfectly applicable to them. That was the last question.

Then, regarding the abstinence syndrome after the procedure, sometimes we have to manage it, of course, and we manage it with nicotine patches or with other medications, to make sure the patient is okay, and doesn't feel the abstinence. Also, we try to encourage patients not to smoke during their hospital stay, and most of the time we reach our objective. That would be the second question.

Regarding the third question. I will try to answer the two questions related to prolonged air leak in one. Normally we discharge patients with prolonged air leak not with Heimlich valves, but with mini Pleur-evacs, and we follow them regularly in the outpatient clinic. For this study we didn't consider prolonged air leak as a pulmonary complication, because normally we tend to consider air leak, at least in our setting, as a technical problem. That’s the reason why we only considered pneumonia and atelectasis as pulmonary complications.

Dr D. Stavroulias(Oxford, UK): I have only one question. If you do believe in this paper, and I don't, but if you do believe in this paper, I would expect that you advise your patients not to smoke after the operation either. Is this correct?

Dr Rodriguez: No.

Dr Stavroulias: Why is that?

Dr Rodriguez: I think your question is very relevant. Before this study we recommended our patients to stop smoking, and normally we would delay the procedure up to four weeks to make sure the patients have quit smoking. Now we recommend the patients to stop smoking before the procedure, but if they don't stop smoking, we continue with the scheduled procedure.

Dr Stavroulias: So you recommend something opposite from the conclusion?

Dr Rodriguez: No. We recommend the patients to stop smoking. Maybe some patients are able to reduce their tobacco consumption before the surgery, but if they are still smoking, we won't delay the surgery.

Dr T. Laisaar(Tartu, Estonia): As I understood it, one group consisted of patients who quit smoking during the 16 weeks before surgery. Did you look at how many patients had quit smoking immediately before the surgery and how many had quit smoking some months before, because I think these patients might really do differently?

Dr Rodriguez: We established a cut-off point at 16 weeks, so unfortunately we had not looked at how many patients had stopped smoking closer to the procedure or how many patients had stopped in advance. Right now, I can't answer your question, but it is something to be taken into account for further analysis.

Dr E. Lim(London, UK): I have a comment. I actively tell my patients to continue smoking all the way through the operation if they have not stopped two months before. My own personal experience is that if you stop 1 or 2 weeks before, the cilia regenerate and you get a lot of sputum retention. But if the patients are smoking all the way through the time of operation, I actively encourage them to go on smoking on Day 1, Day 0, because it forces them to get out of the hospital, to go outside, off the oxygen. With the cold air you can cough expectorate your sputum and you can enjoy your cigarette as well without the expense of nicotine patches and without the oxygen to save the hospital money.

I have a question for you. I enjoyed your paper. I am interested in how you matched the patients, because to be able to repeat the study is quite important. Traditionally when we use propensity score matching techniques, the algorithms match for known and unknown variables in the two comparative groups. How did you actually modify this so that you only match for the known as opposed to the unknown variables as well?

Dr Rodriguez: I think this is a very relevant question. We have only matched the patients according to confounding variables, the variables we know, and that have already been demonstrated to be independent risk factors for postoperative pulmonary complications, but we didn't match them for the unknown variables.

I have been reviewing statistical literature, and in those cases some papers recommend the use of multivariate analysis. The problem with our study is that the number of outcomes was very low, less than 20, so maybe performing a multivariate analysis in that setting would have given us less accurate and less robust results than those obtained with propensity scores. That’s the reason why we decided to do propensity score matching and to match the patients only according to the known confounders.

Dr Lim: One last question. The odds ratio for your smokers to abstinence smokers was an increase in the risk of a complication by an odds of 20%. The 95% confidence interval included an odds ratio up to 5.

Would you like to comment on the uncertainty of your estimate given that the 95% confidence interval includes up to a maximum of five times the odds of developing complications if you continue smoking?

Dr Rodriguez: Yes, I know. This could be because it is a very selected population, and, although it was a very long series, we tried to control all the perioperative variables related to the postoperative pulmonary complication rate, reducing the final number of patients for the analysis. As I said to Dr Batirel, for selected patients outside the vast majority, these results should be taken carefully.

Author notes