PURPOSE: The Philippines ranks ninth on the list of 22 high-burden tuberculosis countries (WHO Global TB Report 2009); and a larger proportion of solitary pulmonary nodules (SPN) are due to tuberculosis. We determined if two prediction models (Mayo and VA) can be used to estimate the probability of malignancy of SPN in the Philippines.
METHODS: All chest roentgenograms with SPN from October 2006 to March 2008 were reviewed by two independent radiologists. Data on variables used in the VA and Mayo Clinic models were collected. Each patient’s final diagnosis was compared with the probability of malignancy predicted by the two models (VA and Mayo Clinic). The receiver operating characteristic (ROC) curves and accuracy of each model were calculated.
RESULTS: Of a total of 17,881 chest radiographs, 81 had SPN. Five were excluded for lack of a definitive diagnosis. Of the remaining 76, thirty-nine (51%) were males. The mean age (± SD) was 55 (± 17) years old. Nineteen (25%) presented with hemoptysis; and 33 (43%) were current smokers with a mean (± SD), of 10 (±15) pack years. The mean nodule size was 2.4(±1.5) cm. Twenty-one (25%) had a spiculated nodule margin while 11 (14%) had calcifications, with 8 being speckled and eccentric. Seven percent (n=5) had history of extrathoracic malignancies. Malignancy was the pathologic diagnosis in 25 (32.9%). Both models predicted all 76(100%) as malignant, thus no ROC curves could be obtained.
CONCLUSION: The prediction models by both the VA and Mayo give a poor estimate to the probability of malignancy of SPN in the Philippines.
CLINICAL IMPLICATIONS: For individuals with SPN who come from countries with a high prevalence of tuberculosis, the VA and Mayo predicition models for SPN may not be useful.
DISCLOSURE: Albert Rafanan, No Financial Disclosure Information; No Product/Research Disclosure Information