The authors provide an excellent analysis of the public data, essentially confirming Van Pelt's findings of 12 years ago, but the question arises as to whether the two authors and six editors and reviewers involved in the genesis of this paper, 8 people in all, were concerned with the statistical details yet overlooked a fundamental fact of lung cancer. People with lung cancer commonly suffer shortness of breath. People with shortness of breath commonly move to lower altitudes where their respiratory difficulties are alleviated by the higher oxygen partial pressure. Some of these people will initially be diagnosed with lung cancer after moving to a county at a lower elevation. Others will be diagnosed while living at a higher elevation, but move to a lower elevation for their own comfort, and subsequently be recorded in the statistics of the county of their new home. In both cases, these movements of people may account for the authors' finding of a negative association between elevation and county lung cancer rates, making less compelling the authors' suggestion of carcinogenic activity of reactive oxygen species.

There may be a rationale for oxygen as carcinogenic, but the mouse study cited has an unusual context, is the first of its kind, and is of uncertain interpretation. The cited association of neonatal oxygen supplementation and cancer may result from the reverse causality, or from other illnesses acting as common-causal variables.

The hypothesis that the reported association is an artefact of patient movement could be tested by interviewing patients with lung, breast, colorectal and prostate cancer at clinics at low elevation, and determining whether the proportion of people reporting having previously lived at high elevation locations is different among these four groups. My expectation is that a higher proportion of such people will be found among the lung cancer patients, having been motivated by their shortness of breath to move to a lower elevation. The breast, colorectal and prostate patients will tend to remain in their home counties regardless of elevation. The authors' thorough analysis beneficially reveals the need for this follow-up research to be done. -Maxwell G. Anderson, PhD