P1.03-052 The Effect of Rounding on Rate of Positive Results on CT Screening for Lung Cancer

Abstract

Effective management of small pulmonary nodules to reduce frequency of false positives has been one of the most challenging issues to implementation of screening. Measurement of size is important as it determines whether a nodule is positive result and also whether growth has occurred. Lung-RADS v.1 guideline requires nodule measurement to be rounded to the nearest whole number, it is not specified whether individual length and width measures should also be rounded prior to rounding the diameter. An alternative approach is the one used in I-ELCAP where measurements were recorded to one decimal place. This study explored how rounding would affect the frequency of positive results both for baseline and annual rounds. Using data collected from CT screenings of 21,136 I-ELCAP participants, we evaluated four different approaches for calculating the nodule diameter (D) based on measurements of the length (L) and width (W) listed below: 1) Measurement of L and W to one decimal place (x.x) and calculation of D without rounding; 2) rounding D to the nearest integer; 3) rounding the L and W measurements to the nearest integer before calculating D with no further rounding; and 4) rounding the calculated D determined by method 3 to the nearest integer. Threshold of positive results was 6.0 mm for baseline round and 4.0 mm for annual repeat rounds of screening. Frequency of positive results in the baseline and annual repeat rounds were compared. For baseline screening using the current I-ELCAP definition (Method 1), the rate of positive results was 10.2%. Using method 2, 3 and 4, positive rates were 12.8%, 10.5% and 13.2%, respectively. Use of rounding would have increased the frequency of positive results by 25.7%, 3.0%, and 28.9%, respectively. Of 85,877 repeat screenings, the rate of positive results was 8.0% using method 1. Using method 2, 3 and 4, positive rates were 9.7%, 8.3% and 9.8%, respectively. Use of rounding would have increased the frequency of positive results on repeat screenings by 20.5%, 3.2%, and 22.3%, respectively. Regardless of where the rounding occurred, it results in more nodules designated as positive. This effect is most pronounced when the rounding occurs in average diameter, and since frequency of nodules increases as size decreases, small nodules are therefore the most frequent cause for positive results and rounding can lead to large increases in positive rates.