Articles by Geyman, J. In calculating the sensitivity, specificity, positive predictive value, and negative predictive value of the different nodule size thresholds, positive screens having no nodule 4mm or larger or nodules of unknown size were counted as positive for all thresholds.
Participants indicated a strong preference for expectant treatment, but gave physician recommendation a significant role in the final decision. Physicians need to offer both options to patients and consider individual patient preferences when making recommendations regarding management of first-trimester incomplete spontaneous abortion.
Molnar , Lynn M. Oliver , MD and John P. Address reprint requests to Alexandra M. Services Email this article to a colleague Alert me when this article is cited Alert me if a correction is posted Alert me when eletters are published Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Download to citation manager Get Permissions.
Responses Submit a response No responses published. Google Scholar Articles by Molnar, A. Articles by Geyman, J. Search for related content. Positive predictive value using various nodule size thresholds to define a positive screen. Approximately one follow-up CT scan was performed per positive T0 screening examination, regardless of nodule size Table 3. Follow-up CT rates were substantially lower in the T period compared with T0, reflecting the practice of recommending follow-up with the next annual screen for nodules that were stable for one year.
False-positive screen is defined as a positive screen in which no lung cancer is diagnosed within one year of the screen. Percentages are calculated using the number of false positive screens as the denominator, obtained from Table 1 by subtracting the number of lung cancers from the number of screens with each maximum nodule diameter.
The projected reductions in follow-up CT scans and invasive diagnostic procedures increased with increasing nodule size threshold Table 4. There were no statistically significant differences between size categories in mortality or five-year survival Table 5. P values comparing each feature for differences by nodule size category. NLST investigators chose a nodule size threshold of 4mm in greatest transverse diameter to define a screening examination as positive with recommendation for further evaluation.
Other screening programs have used no threshold 3 , 12 , 13 to 5mm average of length and width 14 , The detailed screening and clinical outcome data of the NLST now allow for an informed analysis of the effect of nodule size positivity threshold on CT screening performance and projected patient outcomes. From these NLST data, we project that each millimeter increase in the nodule size threshold for screen positivity would result in a small but steadily increasing proportion of lung cancer diagnoses being delayed or missed, and a marked decrease in false-positive rates.
Specifically, the percentage of screen-detected cancers that would have been delayed or missed, over all three screening years, was 1. False-positive rates would have been reduced by nearly one-eighth, one-third, one-half, and two-thirds, respectively, at these thresholds. Sensitivity was greater than specificity for thresholds below 7mm, and specificity was greater than sensitivity above 7mm.
Positive predictive value steadily increased, and there was minimal effect on negative predictive value. Our findings also suggest that an increase in the size threshold for a positive screen would lead to a considerable reduction in the number of follow-up tests, paralleling the reductions in false-positive rates. In the NLST, diagnostic follow-up procedures were recommended, but not mandated by the trial protocol. Therefore, the follow-up evaluations performed during the NLST and estimated reductions should be a reasonable reflection of patient and physician behavior in the real-world clinical setting.
Invasive procedures for lesions subsequently determined to be benign represent a major potential source of harm related to screening. Since invasive testing was performed infrequently when the largest nodule was smaller than 10mm, an increased nodule size threshold would have had less effect on the number of invasive procedures than on the number of follow-up CT scans.
Because some of the invasive procedures may have been falsely negative if lung cancer was subsequently diagnosed , not all of these invasive procedures may have been unnecessary.
A surprisingly high proportion of participants with cancer and largest nodule of 4 to 7mm had small cell histology, and an even higher proportion with largest nodule of this size had stage IV cancer. However, these observations are relevant to the expected clinical impact of raising the nodule size threshold for a positive screen.
For example, a threshold of 8mm would have resulted in However, those with 4 to 7mm nodules and stage IV cancer would likely have obtained little benefit from a threshold under 8mm. In addition, more than one third of cancer cases with 4 to 7mm nodules had other suspicious findings, and many of these may have been called positive even with a higher nodule size threshold, potentially preventing a delay in diagnosis.
Furthermore, some indolent small nodule lung cancers may never become clinically apparent in the absence of screening 16 , and for these there would be no harm in delaying the diagnosis until growth is demonstrated on a subsequent annual screen.
Thus, the proportion of cases in which raising the threshold would be detrimental is likely smaller than the proportion of cancer diagnoses that would be delayed or missed because of the higher threshold. There was no clear relationship between nodule size and survival or lung cancer mortality.
Using the December 31, cutoff date for counting lung cancer deaths, the risk ratio for death relative to the chest X-ray CXR arm with the 4mm size threshold was 0. At thresholds of 6mm, 8mm, and 10mm, the projected number of extra CT arm lung cancer deaths and the projected risk ratios would be 7 and 0.
The number needed to screen to prevent one lung cancer death would increase from to , , and , respectively, at these same thresholds. The relationship between nodule size threshold, delayed or missed diagnosis, and false-positive rates is generally similar to an analysis of baseline screening data from the Early Lung Cancer Action Project ELCAP 4.
Notable differences from our study are that nodule size in ELCAP was defined by the average of length and width, and in the ELCAP study the effects of nodule size threshold on the number of invasive procedures, sensitivity and specificity, positive and negative predictive value, stage, mortality, or beyond the first round of screening were not assessed To our knowledge, no other screening studies have examined the effects of different nodule size thresholds.
However, the National Comprehensive Cancer Network guidelines on lung cancer screening recently raised the recommended nodule size threshold for a positive screen to 6mm mean diameter One limitation of this study is that the screening occurred predominantly in academic medical centers within a supervised clinical trial, and performance with different screen positivity thresholds in the general medical community could be different than estimated here.
This may be unlikely, though, since nodule detection and measurement are basic tasks for radiologists who regularly interpret chest CT scans, and substantial variability was seen even among the NLST radiologists 18 , Screening efficacy with higher thresholds also could be different than estimated here if screening eligibility criteria or compliance with annual screening are different than in the NLST.
Because some cancers were not diagnosed until more than one year after the first positive screen, and screening outcomes here were based on annual intervals, the false-positive rates are likely overestimated. Diagnosed cancers could not be correlated to specific nodules, so the degree of false-positive overestimation is unknown. However, this should not affect the estimated increases in the annual numbers of cancers missed or diagnostic procedures avoided with increased nodule size thresholds.
In conclusion, this analysis suggests that false-positive CT screenings and medical resource utilization would be substantially reduced by raising the nodule size threshold for a positive screen, with a delay in the diagnosis of lung cancer and an impact on outcomes in a small percentage of all lung cancer cases. A reduction in the number of false-positive screens and the ensuing work-ups could substantially affect the cost-effectiveness of low-dose CT screening for lung cancer.
The specific tradeoffs demonstrated here may help to inform decisions regarding the optimal threshold to use in clinical practice, or the thresholds to evaluate in a clinical trial setting. In the future, risk models that take into account patient factors and nodule characteristics other than size 20 may help to further refine the criteria for screening CT interpretation. The authors acknowledge the screening center investigators and staff of the National Lung Screening Trial and the staff from Information Management Services and Westat.
Most important, we thank the study participants, whose contributions made this study possible. Pinsky is employed by the National Cancer Institute and collaborated in the design, implementation, interpretation, and writing of study results.
Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide. Sign In or Create an Account. Close mobile search navigation Article navigation. View large Download slide. Diagnostic procedures following false-positive screens by maximum size of noncalcified nodule. Reduced lung-cancer mortality with low-dose computed tomographic screening. Definition of a positive test result in computed tomography screening for lung cancer: Guidelines for management of small pulmonary nodules detected on ct scans: A statement from the fleischner society.
Baseline characteristics of participants in the randomized national lung screening trial. Determining the cause of death. Results of the two incidence screenings in the national lung screening trial.
Results of initial low-dose computed tomographic screening for lung cancer. The national lung screening trial: Results stratified by demographics, smoking history, and lung cancer histology. Screening for early lung cancer with low-dose spiral computed tomography: Results of annual follow-up examinations in asymptomatic smokers. Ct screening for lung cancer: Suspiciousness of nodules according to size on baseline scans.
The pittsburgh lung screening study pluss: Outcomes within 3 years of a first computed tomography scan. Lung cancer screening version 1. Interobserver agreement on interpretation of pulmonary findings at low-dose ct screening.