[ RadSafe ] Any comment on the latest paper in The BMJ

Tue Nov 3 12:36:48 CST 2015

If you read the paper in detail, it is clear that the uncertainty overwhelms any of the conclusions, so i think a fair reading is as follows:  

The entire significance of the linear regression (which is at 90, not 95% confidence) depends heavily upon the highest dose group at 0.6 Gy cumulative colon dose - a very small group.   
This group accounts for only 40/ 19,064 cancer deaths tabulated in this regression.    The inability to account for smoking status -  in addition to the very high uncertainty - in this cohort thus clouds the entire study.  If this cohort had a higher incidence of smokers than others, a confounding by smoking seems likely.  Indeed, and to their credit, the authors plainly calculate that the estimate of ERR encompasses 1, and is not statistically different than the lowest dose group at 90% confidence.   (The absence of smoking status is a true shame, given the quantity and scope of the data.) 

As a test of the overall regression they list regression of subsets of the dose groups, specifically ranges of 0 - 150 mSv, and 0-200 mSv.   This conveniently did not include the 250 mSv dose cohort, which is the most different from the overall regression and thus having included that in the subsets would have been a better test than excluding that dose group in the subsets and using only those ranges that are more congruent with the entire data set.   I find this omission to be a major red flag as to the statistical outcome of the overall data.  

The authors also acknowledge that a quadratic fit, more like a threshold curve, “is only slightly improved” from the linear fit, although they don’t present the conclusion that the curve observed is a little more quadratic than linear.  :}  

A fair conclusion is that this study does not establish either a linear or quadratic fit to the data as taken and treated - hence the estimate of ERR/Gy is highly uncertain even at 90% confidence; and one is pretty much as good as the other, given the imprecision of the data and limitations of the study.  Moreover,  the estimates of relative risk are sufficiently uncertain as to not refute either a linear or quadratic (i.e., threshold) relationship, although a quadratic relationship better described the data than linear, statistically speaking. Lastly, looking at the uncertainties of each dose cohort, one would need at a size of 500 or so to have a meaningfully small uncertainty as to allow examination of the dose response relationship in an adequately precise manner.     

Brad

> On Nov 2, 2015, at 12:40 PM, Brennan, Mike (DOH) <Mike.Brennan at DOH.WA.GOV> wrote:
> 
> I have several comments that I posted earlier, but upon further reflection there are three that I believe point out major issues with the paper and its conclusions:
> 
> (1)  There needs to be a control group, demographically matched to the radiation workers.  The study claims to show increased cancer with increased cumulative dose among radiation workers, but does not compare this to a population exposed only to "background" radiation.  If the cancer rate among any portion of the radiation worker population is lower than in the control, that would present a problem with the current paper, but an interesting topic for a future paper.  Also, this would go some way toward dealing with the medical radiation exposure Mohan correctly points out as an issue.
> 
> (2)  Of particular concern is the strong likelihood that cumulative dose maps very well on increasing age (it makes sense; the longer someone is a radiation worker the more radiation they've been exposed to, and the older they are).  I believe that the connection between age and the chances of developing cancer is fairly well established.  What if the driver for the graph in the paper is not cumulative radiation dose, but cumulative years?
> 
> (3)  At least for the US radiation workers, a non-trivial percentage started their careers in the Navy, and in some cases received much higher dose than they likely did as civilian rad workers.  This does not appear to be captured.  
> 
> -----Original Message-----
> From: radsafe-bounces at health.phys.iit.edu [mailto:radsafe-bounces at health.phys.iit.edu] On Behalf Of Doss, Mohan
> Sent: Sunday, November 01, 2015 6:18 PM
> To: The International Radiation Protection (Health Physics) Mailing List
> Subject: Re: [ RadSafe ] Any comment on the latest paper in The BMJ
> 
> Dear Parthasarathy,
>    You can read my rapid response criticizing the design of this study at http://www.bmj.com/content/351/bmj.h5359/rr .  There are two other rapid responses critical of the article also.  In my opinion, there is good reason to ignore this paper.
>    If there are any new publications that claim increased cancer risk from low-dose radiation, you can probably assume that they have major flaws negating their conclusions, since that has been the pathetic record of such publications todate.  Such authors have cried wolf too many times. 
>    With best regards,
>                              Mohan
> 
> -----Original Message-----
> From: radsafe-bounces at agni.phys.iit.edu [mailto:radsafe-bounces at agni.phys.iit.edu] On Behalf Of parthasarathy k s
> Sent: Sunday, November 01, 2015 8:09 PM
> To: The International Radiation Protection (Health Physics) Mailing List
> Subject: [ RadSafe ] Any comment on the latest paper in The BMJ
> 
> Our newsgroup has apparently ignored tha following paper published in The BMJ
> http://www.bmj.com/content/351/bmj.h5359
> 
> Under what this study adds are the following comments:
> "The study provides a direct estimate of the association between protracted low dose exposure to ionising radiation and solid cancer mortality. Although high dose rate exposures are thought to be more dangerous than low dose rate exposures, the risk per unit of radiation dose for cancer among radiation workers was similar to estimates derived from studies of Japanese atomic bomb survivors. Quantifying the cancer risks associated with protracted radiation exposures can help strengthen the foundation for radiation protection standards". 
> 
> Warm regardsParthasarathy
> _______________________________________________
> You are currently subscribed to the RadSafe mailing list
> 
> Before posting a message to RadSafe be sure to have read and understood the RadSafe rules. These can be found at: http://health.phys.iit.edu/radsaferules.html
> 
> For information on how to subscribe or unsubscribe and other settings visit: http://health.phys.iit.edu
> 
> 
> CONFIDENTIALITY NOTICE: This email communication may contain private, confidential, or legally privileged information intended for the sole use of the designated and/or duly authorized recipient(s). If you are not the intended recipient or have received this email in error, please notify the sender immediately by email and permanently delete all copies of this email including all attachments without reading them. If you are the intended recipient, secure the contents in a manner that conforms to all applicable state and/or federal requirements related to privacy and confidentiality of such information.
> _______________________________________________
> You are currently subscribed to the RadSafe mailing list
> 
> Before posting a message to RadSafe be sure to have read and understood the RadSafe rules. These can be found at: http://health.phys.iit.edu/radsaferules.html
> 
> For information on how to subscribe or unsubscribe and other settings visit: http://health.phys.iit.edu
> _______________________________________________
> You are currently subscribed to the RadSafe mailing list
> 
> Before posting a message to RadSafe be sure to have read and understood the RadSafe rules. These can be found at: http://health.phys.iit.edu/radsaferules.html
> 
> For information on how to subscribe or unsubscribe and other settings visit: http://health.phys.iit.edu