Pulmonary Metastasectomy: Are Observational Studies Sufficient Evidence for Effectiveness?

Abstract

systematic review of pulmonary metastasectomy for Asarcoma was published in 2012 [1]. There were no control data, only multiple surgical follow-up studies. Two further uncontrolled studies of metastasectomy for sarcoma were published in The Annals after completion of the review [2, 3]. In addition, after a search of The Annals since January 2010, 10 additional follow-up reports of pulmonary metastasectomy were found for a range of cancer types. The authors all state that they operated on carefully selected patients, but none provided the characteristics or themagnitude of the patient population from which this careful selection was made. When a practice is highly selective, the apparent benefit can be a result of choosing those naturally destined to do well rather than that the surgery benefits the patients. The argument has been well formulated previously in The Annals [4]. We put it to the editor that it was time for better evidence, and he replied that “critiquing thoracic surgical diseases where we proceed on ‘clinical impression’ instead of objective data would hold the mirror to our faces.” There are many interventions and operations in surgery that were introduced and have become established practice based on what Dr Edmunds calls “clinical impression.” The phrase is in quotation marks because the very words have come to have a pejorative ring to them, but when a treatment works dramatically and consistently, and it makes a similarly favorable impression on successive observers, there may be sufficient weight of evidence from observational studies alone. Wegive twoexamples inwhichobservational studies led to an accepted clinical practice. In one, uncontrolled observational evidence appears to remain sufficient. In the other, surgeons were misled for more than 80 years, and it took a randomized controlled trial to show the error. There were important differences both in the nature of the disease and in the intervention that allowed one to be regarded as proven based on uncontrolled observation, whereas the other required a controlled trial to refute it.