Plausibility visvis Attributability in considering Health Effects due to Low Radiation Dose Exposure

1
Plausibility vis-à-vis Attributability in
considering Health Effects due to Low Radiation
Dose Exposures
Abel J. González Representative to the United
Nations Scientific Committee on the Effects of
Atomic Radiation (UNSCEAR) Vice-President of the
International Commission on Radiological
Protection (ICRP) Member of the Commission of
Safety Standards of the IAEA Autoridad
Regulatoria Nuclear?Av. Del Libertador 8250
(1429)Buenos Aires,Argentina?54
1163231306?agonzale_at_sede.arn.gov.ar
2
The conundrum Effect or no EffectIs that the
question?
3
Apparently contradictory premises

• It is highly plausible that lowrad-induced
  mutations in the DNA may eventually evolve into
  detrimental health effects.
• Detrimental health effects cannot not be
  attributed to lowrad.

4
Thesis

• The two premises are not contradictory but in
  fact are complementary!
• ?
• therefore

5
(i)  It is plausible that exposure to lowrad
might induce detrimental health effects.
consequently radiation protection ought to be
required! (ii) Individual health effects cannot
be attributed to lowrad situations.
(theoretical enumerations of these effects are
technically wrong!)
6
The link between the two premises/theses is the
intrinsic uncertainty of current knowledge on
radiation health effects at lowrad. This implies
limitation in the epistemology, namely
restrictions to the theory of knowledge on the
phenomena linking lowrad exposure and health
effects, especially with regard to its methods,
validity, and scope.
7
The Facts
8
Likelihood of effects
Limit of pathological knowledge
Limit of epidemiological knowledge
Burns, sickness death
100 (certainty)
Clinically observable (pathology of individuals)
Statistically estimated (epidemiology of
populations)
? Decreasing risk of cancer
5
?
Dose (mSv)
100
1000
8
17 November, 2009
9
Clinical Diagnosis
Epidemiological Estimations
Epistemological limitations
Knowledge
Animal Experimentation
Human Biology
10
Timing of events leading to radiation effects.
11
Timing of events leading to radiation effects.
12
Timing of events leading to radiation effects.
13
Timing of events leading to radiation effects.
14
Timing of events leading to radiation effects.
15
Physiology ?
Biology
Physics and chemistry
Epidemio- logy
Time
10-15s.
10-9s.
100 years
10-3s.
10 2 m.
Manifestation of effects
Exposure
The time scale of the phenomena limits knowledge.
16
A simple example of epistemological limitation
17
Deterministic effects Do we know all?
What is the threshold for protracted exposures?
Likelihood 100
? ?
gt1000mSv ?
Dose
Time
18
Misunderstandings
19
Likelihood
Certainty (100)

• SLOPE
• Is 0.005/mSv the probability of cancer at
  lowrad?

Doses
19
17 November, 2009
20
Modeling effects
21
UNSCEAR estimates
22
UNSCEAR modeling for estimating effects

• (estimating because it is an approximate
  calculation, appraisal or judgment)
• Risk of exposure-induced cancer incidence (REIC)
• Risk of exposure-induced death (REID)
• Excess cancer deaths (ECD)
• Years of life lost per unit dose (YLL)
• Years of life lost/radiation-induced cancer
  death(YLLRIC)

23
China
Puerto Rico
U.S.A.
Japan
Distribution of the REID from solid cancer for
various current populations, assuming a test dose
of 0.1 Sv, and using generalized linearquadratic
excess relative risk models fitted by Bayesian
Markov Chain Monte Carlo methods, calculated for
a population in equilibrium.
24
Nominal statistical uncertainty distribution
for REID
approximately log-normal
25
UNSCEAR Summary

• The estimated average REID (lifetime risk of
  death) following an acute dose of 1 sievert (Sv)
  is
• between 4.3 and 7.2 per cent for all solid
  cancers, and
• between 0.6 and 1.0 per cent for leukaemia.
• (namely, about 5 per Sv for all malignancies)

26
Modeling of protection
Modeling of effects
27
Radiation Protection
Epistemology
Paradigm
P r a t i c e
27
28
The radiation protection modelor radiation
protection paradigm
29
(No Transcript)
30
ICRP ought to introduce Detriment-adjusted
Nominal Risk Coefficients
31
Detriment-adjusted Nominal Risk Coefficients

• Risk Coefficient A numeral, expressed in Sv-1,
  which multiplied by dose quantifies the
  plausibility of harm.
• Nominal The stated numeral does not necessarily
  correspond to its real value it relates to
  hypothetical, rather than real, people averaged
  over age and sex.
• Detriment-adjusted The numeral is
  multidimensional, expressing plausible
  expectation of harm, and including inter alia
  the weighted plausibility of fatal and non-fatal
  harm, and life-lost should the harm actually
  occur.

32
Rounded value used in RP standards?5Sv-1
33
Plausible Probability of Stochastic Effects, p
increment of p
Background incidence
In this zone the relationship is irrelevant for
radiation protection
increment of D
Dose, D
backgroundannual dose
BARC
33
17 November, 2009
34
An unpractical alternative
35
Plausible Probability of Stochastic Effects, p
?D1 ?D2 ?p1 ? ?p2 (?p1ltlt?p2)
increment of p2
Background incidence
increment of p1
increment of D1
increment of D2
Dose, D
backgroundannual dose
BARC
35
17 November, 2009
36
Likelihood
But uncertainty remains! (not known, not
completely confident or sure)
Certainty (100)
Doses
36
17 November, 2009
37
The Fight
38
The Fight

• LOWRAD IS BAD
• Mutagenic properties
• Genetic instability
• Bystandards effects clastogenic plasma factors
• Epidemiological evidence

• LOWRAD IS GOOD
• Minute mutagen
• Adaptive response
• Apoptosis gtgt cancerogenesis then..
• hormesis!
• Epidemiological evidence

39
The challenge
Plausibility versus Attributability
17 November, 2009
39
40
Plausible(from L. plausibilis, from plaus-,
plaudere applaud)

• Detriment-adjusted nominal risk coefficients
  
  are plausible
• because they are
• apparently reasonable, likely, and probable,
• butwithout compulsorily being so!

40
41
The detriment-adjusted nominal risk coefficients
are presumed to be seemingly or apparently
likely, fair, reasonable, valid and valuable and
therefore acceptable.

• However, they should not give a deceptive
  impression of reliability, although this
  limitation should not be construed to mean that
  they are specious,
• namely superficially plausible, but actually
  wrong, misleadingly attractive
  only in appearance!

42
Formalization of plausibility

• The product of nominal risk coefficients times
  dose is a probability but it is a
  probability conditional to the LNT model
  assumption.
• Whether the assumption is valid or not is
  unknown, but its validity may be assigned a
  weight indicating the expert's degree of belief
  in it.
• If the weighting factor were a true probability,
  the weighted probability would be the
  unconditional probability of the effect actually
  occurring.
• However, the weighting factor is not a true
  stochastic quantity (even though in
  Bayesian statistics it may be treated as a
  probability).
• It follows that the weighted probability is not
  an unconditional probability in the formal
  Bernoullian definition of probability.
• Thus, Beninson and Lindell suggested that it be
  termed plausibility.
• J.Radiol.Prot. 21(2001)39-44

43
Nominal statistical uncertainty distribution
for REID
approximately log-normal
Cumulative plausibility
confidence limits 1.28.8 Sv-1
44
95 upper limit
1.0-
Cumulative plausibility
Assuming a 20 degree of disbelief
0.8-
0.6-
8.8/Sv
0.4-
0.2-
Risk ()/Sv
5
2
4
6
8
14
10
12
45
95 upper limit
1.0-
Cumulative plausibility
Assuming a 50 degree of disbelief
0.8-
0.6-
8.8/Sv
0.4-
7/Sv
0.2-
Risk ()/Sv
5
2
4
6
8
14
10
12
46
95 upper limit
1.0-
Cumulative plausibility
Assuming a 80 degree of disbelief
0.8-
0.6-
8.8/Sv
0.4-
5/Sv
0.2-
Risk ()/Sv
5
2
4
6
8
14
10
12
47
(No Transcript)
48
There are other subjective qualifiers of
plausibility
49
Plausibility(also require)
Verisimilitude
Believability
Logicalness
Admissibility
Acceptability
Fidelity
Integrity
50
Namely given the epistemological limitations

• we should assume that lowrad may plausibly be
  detrimental to health and, therefore,
• we ought to protect people against lowrad
  although protection must be commensurate
  (optimization).

51
but plausibility ? attributability
52
Attributable

• Regarding a health effect as being caused by
  lowrad.
• (from L. attribut- allotted both from
  attribuere, from ad- to tribuere assign).

17 November, 2009
52
53
Attributability
Demonstrability
Provability
Counterfactuality
Attestability
54
Provability

• Demonstrability by evidence of the actual
  existence of radiation effects
• from L. probare test.

55
In radio-epidemiology

• Radio-epidemiological studies encompass inherent
  constraints for the
  provability of lowrad effects

BARC
55
17 November, 2009
56
Exposed group N people E cancers n
probability of naturalcancer pD probability
of radiation cancer
Control group N people C cancers n
probability of natural cancer
BARC
56
17 November, 2009
57
E number of total cancers (natural
radiation-induced cancers)
C number of natural cancers
(E-C) number of radiation cancers
58
Limit in knowledge!
E-C
BARC
58
17 November, 2009
59
Limitation of knowledge in epidemiological studies

• The standard deviation is
• ? ? 2 n N pd D N
• If the excess cancers are to be proved with a
  statistical confidence of 95
• E C gt 2 ?

BARC
59
17 November, 2009
60
Epidemiological Limit of Knowledge

• Operating algebraically and as n gtgt pd D,
• N gt constant / D2
• which is the equation giving the number of
  people, N, needed for proving excess cancers at
  dose D.

(Constant 8 n / pd2)
BARC
60
17 November, 2009
61
Dose (mSv)
SOLID CANCERS
provable (knowledge)
unprovable
1 mSv
10 9 p.
People
61
17 November, 2009
62
Dose (mSv)
HEREDITABLE EFFECTS
provable (very limited knowledge)
unprovable
1 mSv
1012 people!
People
BARC
62
17 November, 2009
63
Counterfactuality
64
Counterfactuality

• Can the premise
• a given low-dose radiation exposure caused
  health effects in a given individual
• be explained in terms of a counterfactual
  conditional premise of the form
• if the radiation exposure had not occurred, the
  health effects would not have occurred?

65
Yes, it can, for deterministic effects

• No, it cannot, for stochastic effects!

66
Other subjective qualifiers of attributability
67
Attributability(also require)
I n c o n t r o v e r t i b i l i t y
Indubitability
Indisputability
Undeniability
Unquestionability
68
Namely given the epistemological limitations

• Individual health effects cannot be attributed to
  lowrad and, therefore,
• Lowrad cannot be accused to kill people!
  (Death bodies shall not be
  invented, e.g., by multiplying lowrad collective
  doses and people!)

69
Conclusions
70
Likelihood of Health Effects
Certainty (100)
epidemiology
pathology
?
Dose (mSv)
Provability limit of epidemiologists
Provability limit of pathologist
70
17 November, 2009
71
Likelihood of Health Effects
Certainty (100)
epidemiology
pathology
?
Dose (mSv)
Plausible
Individual diagnosis
Collective estimate
72
Likelihood of Health Effects
Certainty (100)
epidemiology
pathology
?
Dose (mSv)
No attribution
Collective attribution
Individual attribution
73

• Deterministic effects (which occur at high doses)
  are demonstrable and
• provable on the individual exposed, and
• attestable by a qualified radio-pathologist.
• They are therefore individually attributable.
• DETERMINISTIC EFFECTS MUST BE PREVENTED!

17 November, 2009
74

• 2. Stochastic effects can be
• collectively (no individually) attributable,
• but only if doses and number of people are high!

17 November, 2009
75

• 3. It is impossible and therefore incorrect
• to attribute individual health effects
• to low-dose radiation exposure situations.

17 November, 2009
76

• 4. However, it is plausible that lowrad may
  cause stochastic effects and, therefore
• people shall be protected against lowrad!
• THE PLAUSIBLE PROBABILITY OF STOCHASTIC EFFECTS
• MUST BE LIMITED!

17 November, 2009
77
Epilogue

• The good news
• the recent UNSCEAR report to the United Nations
  General Assembly and its resolution

78
UNSCEAR intents..to clarify further the
assessment of potential harm owing to chronic
low-level exposures among large populations and
also the attributability of health effects..
Report on the fifty-fourth session of UNSCEAR
(29 May-2 June 2006 General Assembly Sixty-first
session Supplement No. 46 (A/61/46)
79
Resolution adopted by the UN General Assembly (on
10 January 2008 Resolution A/RES/62/100)

• The UN General AssemblyRecalls the intention
  of UNSCEAR.to clarify further the assessment
  of potential harm owing to chronic low-level
  exposures among large populations and also the
  attributability of health effects, and
• encourages the Committee to submit a report on
  that issue at its earliest convenience

80

Av. del Libertador 8250 Buenos Aires, Argentina
541163231758
Thank you!
agonzale_at_sede.arn.gov.ar
BARC
80
17 November, 2009