Peter B. Bach, MD, MAPP
Acknowledgment: The author thanks Geoffrey Schnorr, BS, from the Memorial Sloan-Kettering Cancer Center, who provided research, editorial, and administrative assistance.
Potential Conflicts of Interest: None disclosed. The form can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M13-2926.
Requests for Single Reprints: Peter B. Bach, MD, MAPP, 1275 York Avenue, New York, NY 10065.
Harry J. de Koning, MD, PhD
February 7, 2014
Reply to Bach’s editorial
The recent U.S. Preventive Services Task Force recommendation on lung cancer screening represents a major synthesis of trial evidence, model-based outcomes, and expert judgment to quantify the trade-offs of CT screening for the millions of people at high risk for lung cancer (1, 2). Dr. Bach, however, states in his editorial that one could have been more cautious about relying on modeling for extrapolation well beyond the empirical data to fill in gaps in the evidence (3). He questions the net benefit of screening annually over many years. Does he mean there is only evidence to screen 3 times, as was done in the NLST? Does he mean women should only receive 5 breast cancer screens, since this was the average number of screens in the breast-screening trials? Randomized trials are set up to prove efficacy of an intervention. In translating that evidence to public health, much more is needed, especially to estimate the long-term benefits and harms for the target population. In fact many would argue that modeling is essential to make that translation from trials to population guidelines (4), particularly as we face an ever-increasing pace of technology where questions far outpace our ability to conduct multiple trials. Our model-based analyses (2) required a joint consideration of numerous factors, including smoking-dose response, and age-specific incidence and other cause mortality by smoking behavior and birth cohort. These factors were superimposed onto over 1,000 schedules of screening examinations using the NLST as a guide, something too complex to evaluate without aid of a model. Dr. Bach, however, states that we were unable to generate models that parallel the natural history of lung cancer and that our models produced inconsistent mortality benefits in reproducing the early years of NLST. It should not be surprising that model variability would produce results that differ in the early years when event rates are low and variability large. Even Data Monitoring Committees place low value on the early years. Dr. Bach also points out that the models differed in their predictions of the absolute number of cases and deaths prevented. Absolute counts have considerable natural variability, and are more difficult to estimate accurately, but in the ranking of competing scenarios, all five models rank the 27 scenarios consistently. Moreover, the models reproduce the outcomes observed in the trials (5), and we showed the range of absolute effects in the table on harms and benefits of the advantageous scenario.At the end Dr. Bach mentions the term sharpshooter. Although it is a dismaying example, the analogue with the sharpshooter is striking. With the models, we indeed draw the target around the greatest cluster of data: based on 200,000 persons enrolled in the screening trials, we can give the best estimate about the screen-detectable preclinical period, test sensitivity, and improvements in prognosis by screening and early treatment, by gender, age and histology. We therefore hope that clinical researchers will engage more closely with modelers and contribute to deliberations about the best use of models, with a deeper understanding of the model development and validation process. Harry J. de Koning, R. Meza & S.K. PlevritisReferences1. Moyer VA, on behalf of the U.S. Preventive Services Task Force. Screening for Lung Cancer: U.S. Preventive Services Task Force Recommendation Statement. Ann Intern Med. Published online 31 December 2013 doi:10.7326/M13-27712. Koning HJ de, Meza R, Plevritis SK, ten Haaf K, Munshi VN, Jeon J, Erdogan SA, Kong CY, Han SS, van Rosmalen J, Choi SE, Pinsky PF, Berrington de Gonzalez A, Berg ChD, Black WC, Tammemägi MC, Hazelton WD, Feuer EJ, McMahon P. Benefits and Harms of Computed Tomography Lung Cancer Screening Strategies: A Comparative Modeling Study for the U.S. Preventive Services Task Force. Ann Intern Med. Published online 31 December 2013 doi:10.7326/M13-23163. Bach PB. Editorial. Raising the Bar for the U.S. Preventive Services Task Force. Ann Intern Med. Published online 31 December 2013 doi:10.7326/M13-2926 4. Heijnsdijk EA, Wever EM, Auvinen A, Hugosson J, Ciatto S, Nelen V, Kwiatkowski M, Villers A, Páez A, Moss SM, Zappa M, Tammela TL, Mäkinen T, Carlsson S, Korfage IJ, Essink-Bot ML, Otto SJ, Draisma G, Bangma CH, Roobol MJ, Schröder FH, de Koning HJ. Quality-of-life effects of prostate-specific antigen screening. N Engl J Med. 2012 Aug 16;367(7):595-605. doi: 10.1056/NEJMoa1201637.5. Meza R, ten Haaf K, Kong CY, Erdogan A, Hazelton WD, Black W, Tammemagi M, Choi S, Jeon J, Han S, Munshi V, van Rosmalen J, Pinsky P, McMahon PM, de Koning H, Feuer EJ, Hazelton WD, Plevritis SK (In Press). Comparative analysis of five lung cancer natural history and screening models that reproduce outcomes of the NLST and PLCO trial. Cancer, in press.
Bach PB. Raising the Bar for the U.S. Preventive Services Task Force. Ann Intern Med. 2014;160:365–366. doi: https://doi.org/10.7326/M13-2926
Download citation file:
Published: Ann Intern Med. 2014;160(5):365-366.
Healthcare Delivery and Policy, Prevention/Screening.
Results provided by:
Copyright © 2020 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use