Article Text

Download PDFPDF

Risk and benefit: the eternal Yin and Yang of thoracic surgery

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Quality of life, as well as its length, is an essential consideration in developing and recommending therapeutic strategies

With the occasional exception, non-small cell lung cancer (NSCLC) is not curable without some form of surgical resection. The good news—ie, the benefit—is that when appropriate oncological standards are followed, cure can be obtained in a substantial majority of operated patients. However, this benefit is not gained without exposing the patient to risk. This risk/benefit relationship is the yin and yang—the two opposing considerations—of thoracic surgery. The proximate risks associated with lung resection are morbidity and perioperative mortality. The challenge, both for a given individual and for large populations, is to maximise the likelihood of benefit while minimising the risk. The paper by Strand et al1 in this issue of Thorax (see page 991) addresses the issue of operative risk with regard to 30-day postoperative mortality. This paper makes a major contribution to this issue by clearly defining specific risk factors and even developing a helpful model to estimate the risk of postoperative death.

This Norwegian study reviewed a total of 4395 patients who underwent some form of lung resection for NSCLC between the years 1993 and 2005. Considerable data were able to be reviewed because of the Norwegian law that newly diagnosed cases of cancer have to be reported to the Cancer Registry of Norway. Mining this database for patients with NSCLC revealed the following conclusions. While the overall 30-day postoperative mortality rate was 4.4%, the mortality decreased during the study period, demonstrating a continually diminishing risk of death. This rate seems appropriate as it is nearly identical to the 4.1% mortality found in a similar analysis of patients in 2001 in the USA.2

Statistically significant risk factors for patients undergoing lung resection for cancer included male gender, increasing age, operations on the right lung and the need for more extensive operations such as pneumonectomy or bilobectomy. In addition, patients with a higher Charlson co-morbidity index had a higher risk of postoperative mortality than patients with a lower index. These findings were used to generate a prediction model which can be used to define the risk of 30-day postoperative mortality for any specific patient. Knowing these risk factors and having the risk formula will be useful for thoracic surgeons and provide the opportunity for an evidence-based selection of patients for surgery. However, an important caveat which the authors make is that “… surgical treatment should not be withheld because of co-morbid conditions … alone”. In other words, there is still need for art as well as science.

Some potential risk factors were not conclusively defined. First, the relationship between surgeon specialty and outcomes could not be identified any more specifically than to say that some were general surgeons and some were cardiothoracic surgeons. This relationship, of course, would be interesting to know as previously reported studies have found better outcomes for dedicated thoracic surgeons.3 4 Two other risk and outcome considerations which have been the target of inconclusive international debate could not be definitively resolved by this study4 5—namely, the type of hospital (general vs university) and the question of hospital volume. The p values for both these analyses approached but did not reach statistical significance. The question therefore remains of the possibility of a β-type error and that, with a larger dataset, both of these differences might in fact reach statistical significance. The issue is also clouded by the fact that most of the Norwegian high-volume hospitals were also university hospitals, confounding the issue even further. The questions of the roles of these possible determinants of outcome, surgeon specialisation, hospital type and hospital volume remain unsettled.

This influential study emphatically highlights the value and importance of tracking patient outcomes in a prospective database. This report and its observations would not have been possible without the government-required Cancer Registry of Norway. In the USA the two equivalents are the Society of Thoracic Surgeons database and the American College of Surgeons’ national cancer database. Participation in these databases allows individual surgeons, surgical groups and hospitals to compare their outcomes with national outcomes which, in turn, allow them to identify areas where performance and quality can and should be improved. Further, these large databases allow determination of risk adjusted mortality and morbidity rates, which are the only legitimate and reasonable way to compare, stratify and determine appropriate outcomes.

One factor not considered by this report is the distinction between the length of life and its quality. For any comprehensive consideration of the risk-benefit of lung resection for patients with lung cancer, follow-up beyond 30 days and, in fact, for a lifetime is therefore essential. If the patient survives but experiences severe post-thoracotomy pain, becomes a respiratory cripple, or is unable to resume the desired lifestyle, the “benefit” of a curative operation is compromised. The complete picture must be kept in mind. Quality of life, as well as its length, is an essential consideration in developing and recommending therapeutic strategies.6



  • Competing interests: None.

Linked Articles