On 2016 Jun 21, Andrew W Swartz commented:

Based upon post hoc questionnaire analysis, Shoag et al. claim that the PLCO was confounded by more prostate cancer (PCa) screening in the control group than in the intervention group. This seems highly unlikely because it is incompatible with other more objective outcomes of this trial.

First, the cumulative incidence plots for PCa [1,2] are consistent with comparison of a screened group to an unscreened (or at least less screened) group. During the first four years of screening the intervention group was diagnosed with PCa at a rate 48% greater than the control group [2]. Then, after the intervention period, the rate of prostate cancer diagnosis in the intervention group drops to that of the control group and the lines go parallel, implying similar screening exposure from that point forward. These are the expected plot shapes for screening with overdiagnosis. The Shoag et al. claim directly implies that the group with the overdiagnosis (the intervention group) was the LEAST screened. This seems implausible, as it would be contrary to all screening theory and experience.

Second, the Shoag et al. claim is also incompatible with the previously reported tumor stage-shifting. The intervention group had more favorable staging than the control group [1,3]. Though the magnitude of this effect is smaller in the PLCO than other PCa screening trials, the direction still favors the intervention group. It is highly unlikely that the intervention group would have better tumor staging if the control group had more screening.

The overdiagnosis and tumor stage-shifting are objectively measured outcomes which indicate that there was more screening in the intervention group than the control group. Therefore Shoag et al. have more likely discovered a bias in the questionnaire answers [4,5] than a flaw in the PLCO. We should also consider the possibility that this is simply the product of the “researcher degrees of freedom” which inherently accompany reanalyses[6].

Andrew W. Swartz MD, Emergency and Family Medicine, Yukon-Kuskokwim Health Corporation, Bethel, Alaska

1] Andriole G, Grubb R, Buys S, Chia D, Church T, Fouad M, et al. Mortality results from a randomized prostate-cancer screening trial. New England Journal of Medicine. 2009;360:1310–9. Andriole GL, 2009 Full Text

2] Andriole GL, Crawford DE, et al. Prostate Cancer Screening in the Randomized Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial: Mortality Results after 13 Years of Follow-up. J Natl Cancer Inst. 2012;104:125–132. Andriole GL, 2012 Full Text

3] Ilic D, Neuberger MM, Djulbegovic M, Dahm P. Screening for prostate cancer. Cochrane Database of Systematic Reviews 2013, Issue 1. Art. No.: CD004720. DOI: 10.1002/14651858.CD004720.pub3. Ilic D, 2013

4] Hebert JR, Clemow L, Pbert L, Ockene IS, Ockene JK. Social desirability bias in dietary self-report may compromise the validity of dietary intake measures. Int J Epidemiol. 1995;24(2):389-98. Hebert JR, 1995

5] Adams SA, Matthews CE, Ebbeling CB, et al. The Effect of Social Desirability and Social Approval on Self-Reports of Physical Activity. Am J Epidemiol. 2005;161(4):389-98. doi:10.1093/aje/kwi054. Adams SA, 2005 Full Text

6] Christakis DA, Zimmerman FJ. Rethinking reanalysis. JAMA. 2013;310(23):2499-500. doi: 10.1001/jama.2013.281337. Christakis DA, 2013

On 2016 Jun 21, Andrew W Swartz commented:

Based upon post hoc questionnaire analysis, Shoag et al. claim that the PLCO was confounded by more prostate cancer (PCa) screening in the control group than in the intervention group. This seems highly unlikely because it is incompatible with other more objective outcomes of this trial.

First, the cumulative incidence plots for PCa [1,2] are consistent with comparison of a screened group to an unscreened (or at least less screened) group. During the first four years of screening the intervention group was diagnosed with PCa at a rate 48% greater than the control group [2]. Then, after the intervention period, the rate of prostate cancer diagnosis in the intervention group drops to that of the control group and the lines go parallel, implying similar screening exposure from that point forward. These are the expected plot shapes for screening with overdiagnosis. The Shoag et al. claim directly implies that the group with the overdiagnosis (the intervention group) was the LEAST screened. This seems implausible, as it would be contrary to all screening theory and experience.

Second, the Shoag et al. claim is also incompatible with the previously reported tumor stage-shifting. The intervention group had more favorable staging than the control group [1,3]. Though the magnitude of this effect is smaller in the PLCO than other PCa screening trials, the direction still favors the intervention group. It is highly unlikely that the intervention group would have better tumor staging if the control group had more screening.

The overdiagnosis and tumor stage-shifting are objectively measured outcomes which indicate that there was more screening in the intervention group than the control group. Therefore Shoag et al. have more likely discovered a bias in the questionnaire answers [4,5] than a flaw in the PLCO. We should also consider the possibility that this is simply the product of the “researcher degrees of freedom” which inherently accompany reanalyses[6].

Andrew W. Swartz MD, Emergency and Family Medicine, Yukon-Kuskokwim Health Corporation, Bethel, Alaska

1] Andriole G, Grubb R, Buys S, Chia D, Church T, Fouad M, et al. Mortality results from a randomized prostate-cancer screening trial. New England Journal of Medicine. 2009;360:1310–9. Andriole GL, 2009 Full Text

2] Andriole GL, Crawford DE, et al. Prostate Cancer Screening in the Randomized Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial: Mortality Results after 13 Years of Follow-up. J Natl Cancer Inst. 2012;104:125–132. Andriole GL, 2012 Full Text

3] Ilic D, Neuberger MM, Djulbegovic M, Dahm P. Screening for prostate cancer. Cochrane Database of Systematic Reviews 2013, Issue 1. Art. No.: CD004720. DOI: 10.1002/14651858.CD004720.pub3. Ilic D, 2013

4] Hebert JR, Clemow L, Pbert L, Ockene IS, Ockene JK. Social desirability bias in dietary self-report may compromise the validity of dietary intake measures. Int J Epidemiol. 1995;24(2):389-98. Hebert JR, 1995

5] Adams SA, Matthews CE, Ebbeling CB, et al. The Effect of Social Desirability and Social Approval on Self-Reports of Physical Activity. Am J Epidemiol. 2005;161(4):389-98. doi:10.1093/aje/kwi054. Adams SA, 2005 Full Text

6] Christakis DA, Zimmerman FJ. Rethinking reanalysis. JAMA. 2013;310(23):2499-500. doi: 10.1001/jama.2013.281337. Christakis DA, 2013