Objective: To validate the influence of the Charlson comorbidity index (CCI) in patients with operated primary non-small cell lung cancer. Methods: From January 1996 to December 2001, 205 consecutive resections for non-small cell lung cancer were performed at the Erasmus Medical Center Rotterdam. The patients ranged in age from 29 to 82 years, with a mean age of 64 years. In a retrospective study, each patient was scaled according to the CCI and the complications of surgery were determined. Results: The hospital mortality was 2.4% (5/205). Of the 205 patients 167 (32.7%) experienced minor complications and 32 (15.6%) major complications. In univariate analysis, gender, grades 3–4 of the CCI, any prior tumor treated in the last 5 years and chronic pulmonary disease were significant predictors of adverse outcome. Multivariate analysis showed that only grades 3–4 of the CCI was predictive (odds ratio=9.8; 95% confidence interval=2.1–45.9). Although only comorbidity grades 3–4 was a significant predictor, for every increase of the comorbidity grade the relative risk of adverse outcome showed a slight increase. Conclusion: The CCI is strongly correlated with higher risk of surgery in primary non-small cell lung cancer patients and is a better predictor than individual risk factors.
Mortality due to cancer in the Netherlands is the highest for lung cancer in men and the second in place for women . Although pulmonary resection is recognized as the best treatment for lung cancer, there are still questions about the relationship between comorbidity and the 30-day mortality and morbidity. Accurate identification of comorbidity is essential in assessing patient's health status and quantifying risk of mortality and morbidity.
The Charlson comorbidity index (CCI), developed by Charlson and colleagues  in 1987, was developed based on a longitudinal study of 559 patients admitted to a medical service during a 1-month period. Nineteen conditions were found to significantly influence survival in the study population and were given a weighted score based on the relative mortality risk (Table 1) . The sum of the weighted scores of all of the comorbid conditions present in cancer patients was then scaled to establish the CCI. The score can be divided into four comorbidity grades: 0, 1–2, 3–4, ≥5.
The weighted index was tested for its ability to predict mortality in a cohort of women with histologically proven primary cancer of the breast. With each increased level of the comorbidity index, there was a stepwise increase in the cumulative mortality attributable to comorbid disease .
The purpose of this retrospective study was to evaluate the usefulness of the CCI in patients with operated primary non-small cell lung cancer.
Patients and methods
We reviewed the medical records of 205 consecutive patients (148 men, 57 women) who underwent resection for primary non-small cell lung cancer at the Department of Cardio-Thoracic Surgery of the Erasmus Medical Center Rotterdam between January 1, 1996, and December 31, 2001. The patients ranged in age from 29 to 82 years, with a mean age of 64. The patient demographics are listed in Table 4.
Each patient was scaled on the CCI. Patients were considered to have a comorbid condition if a listed disorder was mentioned in the records or if the patient was treated for it. Cardiac disease is associated with a higher risk of operation in patients with lung cancer [3–7]. Therefore, we modified the CCI by scoring all forms of coronary artery disease (myocardial infarction, angina, coronary artery bypass graft, percutaneous transluminal coronary angioplasty) with a value of 1. The comorbid conditions of the patients are summarized in Table 1.
For all cases, preoperative workup included chest radiography, computed tomography scan of the chest and upper abdomen, bronchoscopy, electrocardiography, basic biochemical tests, liver and kidney function tests and pulmonary function studies. Additional staging procedures, i.e. mediastinoscopy, liver, bone and brain scans were selectively performed to aid in treatment planning.
The types of procedures performed consisted of pneumonectomy (55), bilobectomy (19), lobectomy (126) and wedge resection (4). The histological typing occurred according to The World Health Organization Histological Typing of Lung Tumours . All tumours were staged according to the international tumour-node-metastasis (TNM) classification . Staging was based on pathological assessment of the primary tumour and surgical sampling of bronchopulmonary, hilar and mediastinal lymph nodes. Histological subtypes and stage of disease are presented in Table 2 .
The length of hospital stay (LOS) was calculated as the difference, in days, between date of discharge and date of surgery. Complications were classified as minor (non-life-threatening) or major (potentially life-threatening), occurring within 30 days of the operation or during a longer period in the same postoperative hospital stay . When a given patient had both minor and major postoperative complications, he was coded as having major complications only, although the nature of the minor complication was also recorded. Hospital mortality was defined as death occurring within 30 days of surgery or any death later in the same postoperative hospital stay.
The χ2 or Fisher exact test was used to analyze the categorical data. Continuous variables were analyzed using the Student's t-test. Univariate and multivariate logistic regression analysis was used to discriminate independent risk factors for major complications after surgical resection. One way analysis was used to determine the influence of comorbidity on length of hospital stay. All data analysis was performed with SPSS for Windows (release 10.1; SPSS Inc, Chicago, IL). A P value <0.05 was considered significant.
When comparing gender, the prevalence of comorbidity was not significantly (P=0.179) higher in males compared to females (79 versus 70%). The prevalence of comorbidity for lung cancer in patients >70 years was significantly (P=0.009) higher than for younger patients (89 versus 72%).
Of the 205 patients, 67 (32.7%) experienced minor complications and 32 (15.6%) major complications (Table 3) . The most common complications were supraventricular arrhythmia and air leak lasting more than 5 days. Five patients (2.4%) died postoperatively. The causes of death were inferior myocardial infarction (n=1), cardiac failure (n=1), multiple pulmonary embolism (n=1) and empyema (n=2). These patients had a CCI of respectively 2, 3, 3, 1 and 0.
Univariate analysis showed that significant predictors of major complications were gender, grades 3–4 of the CCI, any prior tumor treated in the last 5 years and chronic pulmonary disease (Table 4) . Although only comorbidity grades 3–4 was a significant predictor, for every increase of the comorbidity grade the relative risk of major complications showed a slight increase.
Age, pneumonectomy, squamous cell carcinoma, smoking, diabetes, congestive heart failure, coronary artery disease and respiratory function were not significant predictors of major complications.
In the multivariate model only grades 3–4 of the CCI was associated with an increased risk of major complications (odds ratio, OR 9.8; 95% confidence interval, CI 2.1–45.9). Given these results, the CCI is a better predictive factor than individual risk factors.
The mean length of hospital stay was 14.4 days, ranging from 2 to 116. An increase of comorbidity grade showed a slight increase of the LOS, although this was not significant (P=0.107) (Table 5) .
Comorbidity in general has been considered to be an important prognostic factor in patients operated for cancer . The presence of clinical data (symptoms) and comorbidity have been repeatedly evaluated as prognostic factors [12–14]. Battafarano et al.  concluded in a prospective study of 451 patients who underwent surgical resection for pathological stage I non-small cell lung cancer that comorbidity has a significant impact on survival. Even in the presence of another tumor with a theoretical good prognosis (for example squamous cell cancer of the skin), survival decreases in patients with lung cancer .
In general, the reported morbidity after operative treatment of lung cancer is high, because the majority of patients are elderly and most have chronic obstructive pulmonary disease. Diffusing capacity of the lung for carbon monoxide, predictive postoperative FEV1% and VO2max are respiratory function test which can be used to assess these patients and which have been proven to be a predictive value for postoperative outcome . A few reports [19–23] have shown, however, that advanced age is not necessarily associated with a higher morbidity. They reported an overall mortality rate from 1.2 to 7.4% in patients older than 70 years and concluded that no patients should be denied thoracotomy because of age alone. In the present study age was also not a significant risk factor for major complications. Pneumonectomy is considered to be a predictor of postoperative complications in particular mortality . However in this series only two patients died after pneumonectomy. In the logistic regression analysis the odds ratio of pneumonectomy for major complications was 1.8 which points to a predictive effect of type of resection on major complications. However, probably due to the small number of patients, the odds ratio was not significant. This finding underscores our conclusion that in many instances it is better to use a morbidity index score than a single variable to predict postoperative outcome.
The CCI has been found to be useful in some reports. Beddhu et al.  used the CCI in a retrospective study of peritoneal dialysis and hemodialysis patients and found that the CCI was a strong predictor. Fried et al.  also found that the CCI was a strong predictor of mortality in peritoneal dialysis patients. Singh and colleagues  reported in a multi-institutional study of patients with head and neck cancer that the CCI was a valid prognostic indicator. In this study we also found that the CCI is the best predictor of major complications of surgery. Most of the factors necessary for the CCI are standard clinical variables. These factors are easy to find in clinical records. In our retrospective study only data on FEV1% were incomplete in 5% of the cases.
In summary, we conclude that the CCI is a strong predictor of major complications of surgery in non-small cell lung cancer patients and is a better predictor than individual risk factors. The index is easy to use and could have widespread applicability.