RIBs & BEIR P2

(P1 can be found here)

We know we can't run the "perfect" epidemiological experiment on humans. 

But we have options:

1.  We can run the type of study described in the last post on non-human animals, which we've done.  But the trade off is, we don't know how to map those results to humans.  We have lifespan differences, preferential cancer differences, diet differences, etc.  BEIR mentions these, and we shouldn't totally ignore them.  But they have limited general inferential power.

2.  We can infer from observed specific, non-cancer dose response relationships at low doses.  This is covered in BEIR (if you don't want to read that level of detail, there is a summary chart on the page following the linked page).

These are primarily human cell studies as well as animal cell studies in low dose ranges.  The dose response curves are typically LNT.  This is an improvement in knowledge, but still not cancer.

3.  Cohort studies.  These are lesser quality studies than the type described in P1 of this series.  These are frequently called "follow up" studies.  You have an exposed group (that you didn't choose) and you have an unexposed group (that you need to choose that best represents the exposed group).  You have an exposure (or dose) situation you didn't design.  These are real people with complicated lives.  In a prospective study, you observe in real forward-going time as effects are manifested over time.  This means a lengthy study (the LSS or A-bomb study is mostly prospective, we learn more with time.  LNT-deniers seem to think an "old" study is a bad thing, when the opposite is true for a prospective study.  An old study is usually a bad thing for a retrospective cohort study, where all the data is in the past and the researcher has to find  or reconstruct it).

Obviously, such a study is going to have a large detection gap between 0  and whatever its lowest +dose, + effect ends up being.

4. Other studies.  There are other types of studies that have been done, but these usually suffer from problems with low numbers or averaging or very specific or unique circumstances that make them inappropriate for general application.  Let's briefly discuss averaging which was employed by Cohen:

Imagine two groups of people, 4 in each group, each dosed to an average of 50 (let's avoid units).  In one group we find 1 death, and the other we find 2 deaths.  This is very troubling from an epidemiological point of view...you exposed each group to the same average dose, but there is a 100% difference in effect.  With this sort of situation, it is highly unlikely you will find a 95% confidence level correlation between dose and effect. 

This is why we need to look at individuals not averages.  We might look at the 4 folks in one group and find doses like this (bold indicates the person died):  100, 100, 0, 0.  And we might look at the second group and find this:  200, 0, 0, 0.

If we had looked at individuals, we would see a clear linear relationship between dose and death, totally masked by averaging.

Back to cohort studies.  Every cohort study has certain pro's and con's....after all, the important parameters are not in the control of the researcher.  LNT denialists are quick to point out the con's of the LSS study:

1.  Pulse burst dose rate versus chronic irradiation at low dose rates.
2.  Confounding due to the survivors having been in war conditions.

But they usually ignore the pro's:

1.  A single, easy to identify dose rather than people getting different doses over long periods of time.  Dose estimation errors are therefore minimized.
2.  Large population compared to most others.
3.  Whole body external exposure versus internal exposures typical of power plant accidents.
4.  Good dose distribution...from very high to very low doses.
5.  Since the dose occurred long in the past and many years have passed...we've accumulated knowledge with time!!!
6.  The effect we're studying isn't a single cancer...we're studying many kinds of cancers.  This is different than radon (lung cancer) or iodine studies (thyroid cancer), which show an LNT response.

Overall, the LSS study is the best we have for whole body, low LET radiation.  That's why we want to see it through until the last individual in the study dies.

It's not some grand conspiracy by the "establishment"...it's good science.

Next:  The LSS study.