The Marshfield Clinic is a multispecialty group practice located in rural central Wisconsin, providing both primary and referral type care with a full range of oncology-related services. The Marshfield tumor registry averages 260 new cases of lung cancer per year. The tumor registry provided identifying information on consecutive cases of lung cancer with established cell types according to the 1981 World Health Organization criteria. Radiographs were considered acceptable if taken within 30 days of diagnosis, and posteroanterior (PA) and lateral radiographs were available in most cases. Each radiograph was submitted to one of two participating radiologists who did not have knowledge of the cell type. Specific radiographic findings on the chest x-ray film were evaluated and recorded on a standard form. The radiograph was then reread by the second radiologist and discrepancies were settled prior to analyzing data. Table 1 is a listing of principal findings recorded on the data forms.
As a basis for historic comparison, we used a widely referenced series from Mayo Clinic that details radiographic findings according to cell type for 600 cases of lung cancer in the 1950s and 1960s. We examined differences between Marshfield Clinic and Mayo Clinic for the following abnormalities: parenchymal mass (tumor within lung substance), apical mass, pleural effusion, obstructive findings, and mediastinal masses or adenopathy. We also formed a “peripheral primary tumor” group by including parenchymal and apical masses.
To compare with Mayo Clinic’s “hilar or perihilar mass or prominence” group, we combined hilar mass, perihilar mass, and hilar adenopathy. This group was compared in the following two ways: single abnormality on chest radiographs referred to as “central mass only,” and central mass either with or without any other abnormalities (such as peripheral mass, effusion, volume loss) is referred to as “central mass plus.”
After completion, the radiographic data forms were entered on the computer and analyzed using software (SPSS/PC+; SPSS Inc; Chicago; and SAS; SAS Institute; Cary, NC). Statistical comparison of proportions of cases with various findings between the Marshfield Clinic and Mayo Clinic series was based on the likelihood-ratio x2 test. In cases in which at least 1 expected value was less than 5, Fisher’s Exact Test (two-tailed) was used; p values <0.05 were considered statistically significant.