Impact of smoking habits on postoperative outcomes following lung surgery for cancer: results from the Epithor database

Abstract

OBJECTIVES

Smoking is a modifiable risk factor for lung resections but to what extent preoperative smoking cessation reduces that risk remains unclear. The study hypothesis was that the potential benefit of smoking cessation can be assessed by measuring the risk difference between active and former smokers in a large cohort of patients.

METHODS

Data were extracted from the French Society of Thoracic and Cardiovascular Surgeons (Société Française de Chirurgie Thoracique et Cardiovasculaire) database. The study cohort comprised patients who underwent lung resection for cancer from January 2002 to December 2020 and for which information on smoking status was available. The risk of overall and specific postoperative complications according to smoking status was defined by logistic regression models, and results were presented in terms of odds ratios (ORs) and relative 95% confidence intervals (CIs) adjusted for confounding factors.

RESULTS

Of the 7204 analysed patients at the time of their operations, 20.2% were active smokers, 60.7% were ex-smokers and 19.1% were never smokers. Compared to former smokers, active smokers experienced a higher rate of respiratory complications (OR 1.5, CI 1.2–1.7) and infections (OR 1.6, CI 1.3–1.9). Postoperative atelectasis was significantly reduced in former smokers (3%) compared to active smokers (6.9%, P < 0.01). In active smokers, the risk was related to the level of exposure, being higher for smokers of more than 40 pack-year.

CONCLUSIONS

After lung surgery, active smokers experience a higher risk of respiratory complications, infections and prolonged air leak compared to former smokers. This risk seems to be related to the level of exposure.

Open in new tabDownload slide

INTRODUCTION

Smoking is a well-recognized risk factor for postoperative morbidity affecting exercise capacity [1], risk of pulmonary adverse events, infections, delayed wound cicatrization [2, 3], prolonged air leak and intensive care unit admission [4] even when a minimally invasive approach is adopted [5, 6]. Given the fact that smoking is a modifiable factor [7, 8], it would be logical to assume that smoking cessation before an operation eliminates that risk. In fact, current guidelines suggest that cigarettes should be stopped at least 4 weeks before an operation [9] because the benefits of shorter periods of abstinence remain controversial [10, 11].

Despite expected benefits, available data showed that quitting smoking before lung surgery reduces the risk only moderately. In a large cohort study, smoking cessation reduced only respiratory complications, and the ideal duration for cessation was not identified [12]. Probably due to the lack of strong evidence, in real life many surgeons do not require patients to quit smoking before surgery because they consider the clinical advantages inferior to the risk of tumour progression while time elapses from smoking cessation, risk of alienating patients or referring physicians and difficulties in referring patients to a tobacco control unit [13]. Consequently, a certain proportion of active smokers can be detected even among high-risk patients undergoing operations [14, 15].

If smoking cessation is beneficial but does not reduce the level of risk to that of never smokers, it probably means that surgical risk linked to smoke is not fully reversible. Based on this assumption being correct, postoperative risk is expected to be low in never smokers, intermediate in ex-smokers and high in active smokers. Additionally, the irreversible risk is represented by the difference in outcome between never smokers and ex-smokers, and the reversible part is measured by the difference between former and active smokers.

The goal of the study was to estimate the potential benefit of preoperative smoking cessation by analysing a large cohort of patients from Epithor, the database of the French Society of Thoracic and Cardiovascular Surgery [Société Française de Chirurgie Thoracique et Cardiovasculaire (SFCTCV)].

METHODS

End points and the database

The hypothesis of this study was that surgical risk increases from never smokers to former smokers, reaching its highest level in active smokers. Therefore, the main end point of the study was to measure differences in postoperative outcome according to smoking habits. Secondary end points were (i) to test the impact of the level of exposure on surgical outcome in active smokers and (ii) to measure persistency of active smoking postoperatively.

The data were extracted from Epithor, the national database established in 2002 by the SFCTCV, endorsed by the High Authority for Healthcare and by the French National Cancer Institute. The database was designed to collect variables regarding comorbidity, respiratory function, staging, surgery and postoperative outcomes, with an external regular auditing process monitoring the quality of the data. Details about Epithor design were published previously [16, 17]. The study cohort was composed of patients who underwent lung resection for primary lung cancer from January 2002 to December 2020 and for which information on smoking status was available.

In Epithor, a smoking habit is described by the combination of 3 different fields: smoke (yes/no), smoking status (controlled/uncontrolled) and exposition (pack-year). The cohort was divided into 3 groups: active smokers (variable smoke: yes, variable status: uncontrolled), ex-smokers (variable smoke: yes, variable status: controlled) and never smokers (variable smoke: no). Cases with incomplete information on smoking history were excluded from the analysis.

Clinical variables available in the database were compared among groups (age, sex, body mass index, American Society of Anesthesiologists score, comorbidity, forced expiratory volume in 1 s, diffusing capacity of the lungs for carbon monoxide, neoadjuvant treatments, type of lung resection, surgical access, duration of the procedure, histological result and stage). Postoperative outcome was analysed in the 3 groups. Postoperative complications were defined by the occurrence of any adverse event classed grade II or higher according to the Clavien-Dindo classification. These events were classified into 4 different categories: respiratory complications (atelectasis, pneumonia, respiratory failure, pulmonary embolism), cardiovascular complications (coronary syndrome, cardiac failure, arrhythmia), infections (pleural, surgical site, other bacterial or viral infection) and surgical complications (prolonged air leak, haemothorax, bronchial fistula, reoperation).

Statistical analysis

In order to analyse the relevance of smoking to postoperative outcome, the 3 groups were compared in terms of 30-day postoperative mortality and complication rates. Classic descriptive statistical indicators (mean or median for continuous variables and percentages for categorical variables) were compared using one-way analysis of variance with the related Bonferroni test for multiple comparisons in case of continuous variables and the χ² or Fisher’s exact test when appropriate for the comparison of categorical variables. Given that the never smoker, past smoker and current smoker have a certain gradation, the significance of any highlighted trends was also assessed. The trend analysis was conducted by estimating a linear regression model for quantitative variables (the significance of the trend was deduced from the significance of the estimated regression coefficient) and by the χ² test for trend for categorical variables.

The risk represented by smoke to postoperative outcomes was adjusted for other clinical variables by logistic regression models. Dependent variables considered in the analysis were postoperative mortality, overall morbidity and any specific complication outlined by the analysis as previously described. Results were presented in terms of odds ratio (OR) and relative 95% confidence intervals (95% CIs) adjusted for confounding factors identified. The analysis was performed using Stata 18 (StataCorp 2023, College Station, TX, USA) and SPSS (IBM SPSS Corp. 2023, Armonk, NY, USA) statistical software. Data are available on reasonable request, pending SFCTCV ethical committee approval.

Ethical committee approval

The study was endorsed by the SFCTCV and approved by their ethical committee (n° 35085-24). Patient consent was waived.

RESULTS

At the time of data extraction, 11989 patients were recorded in the database; 7204 of them had complete information on smoking habit (60.5%) and represent the cohort of the study. The clinical characteristics of the population are listed in Table 1. At the time of the operations, 20.2% (n = 1455) of patients were active smokers; 61% (n = 4375) were ex-smokers and 18.9% (n = 1374) were never smokers. Compared to ex-smokers, active smokers were significantly younger (median age 63 vs 67, P < 0.05) and had a higher exposition history (median 40 pack/years vs 35, P < 0.05).

Postoperative complications according to smoking status are shown in Table 2. Trend analysis showed that smoking history specifically affected respiratory complications, infections and prolonged air leak. The main difference in terms of respiratory complications was the occurrence of postoperative atelectasis, which occurred in 4% of cases in smokers (n = 232) and in 1% of cases in non-smokers (n = 15, P 0.01). The overall hospital stay was shorter in ex-smokers compared to active smokers (7.2 and 7.8 days, respectively, P < 0.05), and the difference increased when complications occurred (11.6 vs 13 days, respectively, P < 0.05).

Smoke and surgical outcome

Smoking status did not influence postoperative mortality. In terms of overall morbidity, a past history of smoking was confirmed as an independent predictor of adverse events (OR 1.7, CI 1.5–1.9, Table 3), respiratory complications (OR 2.9, CI 2.1–3.9), infections (OR 2.8, CI 2–3.9) and prolonged air leak (OR 1.6, CI 1.2–2.1). When the analysis focused on the comparison of active and former smokers, considering respiratory complications, infections and prolonged air leak as independent variables, active smoking was an independent factor for the first 2 (respiratory complication OR 1.47, CI 1.2–1.7; infections OR 1.6, CI 1.3–1.9, Table 4).

Level of exposure

In active smokers, a correlation was found between the level of exposure and the level of risk. Compared to active smokers with mild exposure (30–40 pack-years), patients with high exposure (> 40 pack-years) exhibited double the overall risk (OR 2.1, CI 1.1–3), due to an increased rate of respiratory complications (OR 2.48, CI 1.44–4.25) and infections (OR 2.1, CI 1.24–3.5) (Table 5). Smoke is also an independent risk factor for prolonged air leak in case of mild exposure compared to low exposure (<30 pack-years, OR 1.7, CI 1.2–2.5).

Active smoking after surgery

After surgery, 456 patients from the initial cohort were re-hospitalized because they needed further surgery. The rate of active smokers at the time of the second surgical procedure was 16.2% (74 patients). All of them were active smokers at the time of the first operation.

DISCUSSION

The expected benefits of preoperative smoking cessation in thoracic surgery are difficult to measure because available evidence remains limited. A proxy may be represented by a retrospective comparison of data from large cohorts, in which the difference in outcome between active and former smokers can be easily assessed.

Results from this SFCTCV cohort indicated 3 different aspects of smoking cessation before lung surgery. The first is that the main expected benefits from smoking cessation are a significant reduction in postoperative infections and respiratory complications, with a dramatic decrease in postoperative atelectasis. The second aspect is that in active smokers the risk is related to the level of exposure, because patients who smoked more than 40 pack-years are at higher risk than patients with a lower exposure (<30 pack-years). This information suggests that reducing the number of smoked cigarettes may eventually reduce the risk in patients unable to quit. Finally, a small residual risk seems to remain in former smokers and is not completely set aside by the time, probably due to a permanent alteration at the level of the lung and the immune system.

How accurate is this information? The question is legitimate because of the long period of time covered by the study, during which surgical techniques evolved from standard to minimally invasive surgical approaches. In Epithor, minimally invasive operations increased over time and reached 63% in 2022, 25% of which were robotic procedures. Nevertheless, the rate of active smokers remained similar, both in patients undergoing minimally invasive surgery (19.9%, 823/4129) and in those undergoing an open approach (20.5%, 632/3075).

Two aspects limit the accuracy of published evidence on preoperative smoking cessation, and our study is not exempt from them: (i) obtained information is not fully representative of the entire database, because the ratio between the cohort population and the total number of patients in the database may be low, and (ii) smoking habits are self-reported by the patient. Regarding the first point, general databases are rarely designed to collect exhaustive information on smoking, and information may sometimes be scattered, considering it more as a risk factor for lung cancer than a predictor of postoperative outcome. Nevertheless, comparing this study with the only available large cohort published in 2009 [12], the rate of available information was significantly higher (60.5% vs <50%), suggesting that results from the study may be considered relevant. Concerning the second point, self-reporting is accepted mainly for 2 reasons: (i) the first is that serial testing in smokers is not practical. (ii) The second, even more important, reason is that tests for smokers are of little help because they become quickly negative after cessation (4–6 h for carbon monoxyde testers and 72 h for blood or salivary cotinine levels) and do not allow clinicians to confirm that cessation lasted for several weeks.

Even if risk reduction by preoperative smoking cessation is a time-dependent event, this study was not designed to define the ideal interval between quitting smoking and surgery because no information on the time of cigarette cessation was available in Epithor. This problem has been overcome because additional fields specifying the time of cessation were recently inserted in the Eras section of Epithor.

One additional point of interest related to the study is that the number of active smokers decreased slightly postoperatively, despite the link between level of exposure and long-term survival pointed out in the Red Book of the SFCTCV 2016–2022. Surveillance during the follow-up period should probably include active referral to a tobacco unit for those patients unable to quit after their operation.

CONCLUSION

The study confirmed that active smoking negatively affects postoperative outcome and that the risk of pulmonary complications, infections and prolonged air leak is lower in patients who quit smoking. Given these potential benefits, surgeons are best placed to motivate their patients and should probably acquire basic competence in discussing smoking cessation in order to reduce their complication rates.

FUNDING

Epithor was financially supported by the French Society of Thoracic and Cardio-vascular Surgery (SFCTCV). No addition funding was allocated to this project.

Conflict of interest: none declared.

DATA AVAILABILITY

Data are available on reasonable request, pending SFCTCV Ethical Committee approval.

Author contributions

Francesco Leo: Conceptualization; Methodology; Project administration; Writing—original draft. Giuseppe Migliaretti: Formal analysis; Methodology. Simona Sobrero: Conceptualization; Project administration; Validation; Writing—review & editing. Dan Angelescu: Resources; Supervision; Validation; Writing—review & editing. Tarun Mc Bride: Resources; Validation; Writing—review & editing. Marcel Dahan: Conceptualization; Data curation; Supervision; Writing—review & editing. Jacques Jougon: Conceptualization; Investigation; Methodology; Supervision; Writing—review & editing.

Reviewer information

The European Journal of Cardio-Thoracic Surgery thanks Frank A. Baciewicz Jr, Meinoshin Okumura and the other anonymous reviewers for their contributions to the peer review process of this article.

REFERENCES

ABBREVIATIONS

ABBREVIATIONS
 
• CI

  Confidence interval

 
• OR

  Odds ratio

 
• SFCTCV

  French Society of Thoracic and Cardiovascular Surgery (Société Française de Chirurgie Thoracique et Cardiovasculaire)