The study was approved by uploading the raw data onto the research data deposit (RDD) public platform, with the approval RDD number as RDDA2020001408.
Clinicopathological characteristics of all patients who had lung tumor resection were collected from electronic medical records at Sun Yat-sen University Cancer Center between January 2008 and December 2013. The exclusion criteria were as follows: benign lung tumor or small cell lung cancer, metastatic lung cancer, other malignancy, history of lung surgery, bilateral lung cancer, American Society of Anesthesiologists (ASA) physical status equal to or greater than IV, receiving corticosteroids due to chronic obstructive pulmonary disease (COPD)/asthma exacerbation/ inflammatory bowel disease, TNM stage Ⅲb or Ⅳ, severe perioperative complications and those who lacked histologic conﬁrmation or clinical details[17, 18]. All enrolled patients were divided into two groups on the basis of whether they received dexamethasone during lung tumor resection. Whether patients receive dexamethasone depends on the preference of anesthesiologists. Patients in the dexamethasone group (DEX group) received intraoperatively dexamethasone, and those in the non-dexamethasone group (non-DEX group) didn’t receive intraoperatively dexamethasone.
Patient characteristics and Outcome
The following data were extracted from electronic medical records: age, gender, height, weight, smoking status, smoking index, co-morbid illness, Karnofsky performance score (KPS), TNM stage, histology type, tumor size, American Society of Anesthesiologists (ASA) score, type of anesthesia, anesthetic time, operation methods, dexamethasone, NSAIDS, transfusion, type of postoperative pain control, date of surgery, postoperative complications, chemotherapy and radiotherapy. Co-morbid illnesses consisted of coronary heart disease, heart failure, cerebrovascular disease, hemiplegia, diabetes mellitus, hypertension, hepatitis B, hyperthyroidism or hypothyroidism, nephropathy, peptic ulcer, asthma, chronic obstructive pulmonary disease and connective tissue diseases. Smoking index is the number of cigarettes per day times years of smoking. Tumor size was the largest diameter of tumor mass. Charlson comorbidity index was a tool to calculate the cumulative prognostic burden of co-morbid illnesses in an objective method. Type of postoperative pain control consisted of intravenous analgesia and epidural analgesia; Core component of intravenous analgesia was fentanyl or sufentanil; While core component of epidural analgesia was morphine. The Clavien-Dindo classification was based on the therapy which was used to rectify a specific postoperative complication.
The primary endpoint of our study was disease-free survival (DFS) and overall survival (OS). DFS was the interval between the date of surgery and the date of relapse of lung cancer, metastasis or death. Relapse of lung cancer was defined as locoregional new occurrences of tumor mass and confirmed by imaging or histopathological examination. Metastasis of lung cancer was defined as the dissemination of lung cancer from lung to another part of the body, and confirmed by imaging or histopathological examination. OS was the interval between the date of surgery and the date of death. We recorded the dates of death from the hospital information system (HIS) of Sun Yat-sen University cancer Center. The final follow-up was December 31, 2018. The follow-up period was at least 5 years. Patients lost to follow-up during the study period were censored.
Propensity score matching was conducted to balance baseline characteristics between two groups by reducing the potential confounding factors, which was performed using the IBM SPSS Statistics 22.0 (SPSS Inc, Armonk, NY, USA). Propensity score was the possibility of covariates and calculated by logistic regression analysis. The propensity score for each individual was measured by incorporating the covariates of age, sex, BMI, smoking index, Charlson comorbidity index, KPS, TNM stage, histological type, tumor size, American Society of Anesthesiologists (ASA) score, anesthetic time, type of anesthesia, NSAIDS, operation method, Clavien-Dindo classification, blood transfusion, postoperative pain control, chemotherapy and radiotherapy using logistic regression analysis. A propensity score-matched cohort of patients receiving vs. not receiving dexamethasone was generated at a ratio of 1:2. For each patient in the DEX group, two patients not receiving dexamethasone were matched using the nearest neighbor method. We tested multiple caliper in 0.1, 0.01 or 0.001 of standard deviation (SD) of the logit of the estimated propensity score, which meant the maximum distance of two units was 0.1, 0.01 or 0.001. Ultimately, the caliper of 0.1 met the appropriateness of matching with preferable homogeneity and minimum loss of sample size. We evaluated the efficiency of matching to balance baseline characteristics by using standard differences [19, 20]. Standardized differences less than 10% was regarded as homogeneity of baseline characteristics between two groups.
Categorical variables were presented as the number. Continuous variables were presented as the mean or median. The Student’s t test or the Mann-Whitney U test were used for comparisons of continuous variables. The chi-square test or the Fisher’s exact test were used to compare categorical variables. In the propensity score-matched cohort, disease-free survival (DFS) and overall survival (OS) were compared between the non-DEX and DEX groups, and calculated using the Kaplan–Meier method. Multivariable cox proportional hazards models were conducted to identify other potential confounding factors associated with intraoperatively administration of dexamethasone. All variables were entered into multivariable cox proportional hazards regression analysis to compare the hazards ratio between the two groups by using the “enter” method. Multivariable cox proportional hazards regression model was then to assess the association between clinicopathological variables and survival after lung cancer resection for NSCLC patients. Associations between intraoperatively administration of dexamethasone and disease-free survival and overall survival in the high-risk factor subgroups of postoperative nausea and vomiting (PONV) were calculated by cox proportional hazard regression analysis and presented as forest plot. Univariable, multivariable analyses and propensity score matching analysis were carried out using the IBM SPSS Statistics 22.0 (SPSS Inc, Armonk, NY, USA). A two-tailed P < 0.05 was considered statistically significant.