Institut für Küstenforschung

Download Report

Transcript Institut für Küstenforschung

Miljøbelastning med
anthropogene stoffer
Eksempel „bly“:
konsekvenser, iagttagelser og helbred
Hans von Storch
Institut für Küstenforschung, GKSS Forschungszentrum
Kemisk Institut, København, 13 Dezember 2006
IfK
Institut für Küstenforschung
Institut für Küstenforschung
The case of Germany
IfK
Questions
• Is it doable to estimate lead concentrations in the
atmosphere and in human blood with a simple
regression-type model?
• What were the levels of lead concentration in
human blood in Germany in the 196s and 1970s (a
time for which no measurements are available)?
• How can we design scenarios of lead levels in
human blood conditional upon alternative political
regulations?
Institut für Küstenforschung
Annual gasoline sales and lead emissions in Germany. Volume of
gasoline sold (millions of liters per year; solid) and of leaded gas
(after 1985; red crosses); amount of lead added to gasoline (in tons;
yellow).
IfK
Data set “G” (“Germany”; Heinzow, 1998) is
unsystematically collected, with samples at different
locations, different methods, different age and gender
groups.
The second data set “M” (“Münster”; Human-Probenbank
Münster 2002;) is better, as it is derived from a controlled
sampling strategy – for groups of students living in the
industrial town of Münster (51.5oN, 7.4oE) in NordrhineWestfalia, close to the Ruhr area with heavy industry and
intensive road traffic.
Institut für Küstenforschung
We have two data sets with time series of lead
concentrations in human blood (LHB), beginning in the
late 1970s. No reliable earlier data about lead in the human
blood in Germany is available.
IfK
lead in human blood
according to G and M data sets
G children
G adults
M
120
g/l
80
40
0
1980
1983
1986
1989
1992
1995
1998
2001
Data set G (“Heinzow”) and M (“Münster”) with lead concentration
in human blood (in g/l). The G data are split into adults and
children.
25
120
G samples (g/l)
20
100
15
80
Similarity of
lead
concentration in
human blood
sampled in the
same year.
Based on G and
M data.
10
60
405
200
0
20
5
40
10
60
15
80
M samples (g/l)
20
100
25
120
Lead concentration in
gasoline and in blood
in Germany. Heinzow
(G) and Münster (M)
data sets. The lead
concentration in 1985
to 1995 is an
"effective"
concentration for
West Germany, by
proportionally
weighting the
concentrations in of
leaded (0.015 g/l) and
unleaded (0.013 g/l)
gasoline
lead concentration
in gas (g/l) and in human blood (g/l)
120
lead concentration in blood (g/l)
M
G adults
80
40
0
0
0.04
0.08
lead concentration in gasoline (g/l)
0.12
0.16
Data set G: There is a linear relationship between the mean
concentration and the dispersion of the sample distributions.
standard deviation = 0.43 mean – 3.9,
(1)
95% quantile = 1.63  mean + 10.9
(2)
80
Scatter diagrams of sample
means (horizontal axis) and
sample standard deviations
(dots, red) and 95% quantiles
(diamonds, green) in lead
concentrations in human
blood in data set G. In g/l.
40
Institut für Küstenforschung
standard deviation and 95%ile
120
0
0
40
80
mean
120
160
IfK
800
frequency distribution of LHB (g/l)
in adults in 1991/1992 in the M dats set
600
.
Thus, it is more appropriate to use percentiles:
If the mean concentration is 150g/l,
then according to (2)
5% of the population may
have a concentration of
more than 250 g/l.
400
200
0
0
100
200
Institut für Küstenforschung
Data set G: According to (1), if the mean blood
concentration is 150 g/l, then on average one sixth of the
population will have more than 200 g/l or less than
100g/l. This rough estimate is based on the assumption of
a normal distribution, which is not really valid as the
distribution of lead levels is skewed, with a long tail
towards large values
IfK
Critical Values of Lead Concentration in Human Blood
Defined by the `Human-Biomonitoring-Kommission´ in
Germany
category
value (μg//l)
children (6-14 years)
women* (25-45 years)
other adults
* women in child-bearing
age
1
2
3
1
2
3
≤ 100
100 – 150
> 150
≤150
150 – 250
> 250
Category 1: unobstrusive value
Category 2: no health risks are expected but monitoring is recommended
Category 3: health hazards are possible, clarification and mitigation is needed
Source: Krause et al. 1996
Sample statistics
mean vs. 90% quantile
mean vs. 95% quantile
160
120
80
Scatter diagrams of sample means
(horizontal axis) and sample 95%
(diamonds) and 90% (dots) quantiles
(vertical axis) in lead concentrations in
human blood in data set M. In g/l.
40
0
0
20
40
60
80
100
90% quantile = 1.46  mean – 2.2
95% quantile = 1.75  mean – 4.6.
(3)
(4)
According to (3) and (4), a mean concentration of
150g/l is associated with 5% (10%) of the population
having more than 258 g/l (217 g/l) lead in their
blood.
Institut für Küstenforschung
In case of data set M, the 90% and 95% quantiles are
given. Also in this case, a clear linear relationship
between the mean and the quantiles is emerging with
IfK
LCt+1 =  LCt +  LEt+1
LHBt+1 =  LHBt +  LCt+1 + 
(5)
(6)
Equation (6) is equivalent to
(LHBt+1 - ) =  (LHBt - ) +  LCt+1
(7)
with  = /(1-). Formulation (7) describes the dynamics
of “anomalies” LHBt- relative to a “normal” state 
towards which the system converges as soon as the forcing
LCt ceases if <1. For 0 <  <1 the air concentration
approaches  asymptotically with a time scale of 1/(1-) if
LC =0.
Institut für Küstenforschung
A simple dynamical relationship between three variables,
namely the emission of lead LEt in an area AE in the year t,
the atmospheric concentration LCt in an area AC in the
year t and the mean concentration of lead in human blood
LHBt in the year t in the area AC.
IfK
12000
estimates of lead emission
related to use in gasoline (g/l)
MWV estimate
Pacyna estimate
10000
6000
4000
2000
0
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
Estimates of emitted lead in Germany in tons/year.
Dots: Linearly interpolated estimates by Pacyna and Pacyna (2000).
Crosses: Estimates based on volume of gasoline sold in West Germany,
according to MWV (1998, 2002)
Institut für Küstenforschung
8000
IfK
Best guess of emissions in AE
Muenster case
600
Pacyna linearly interpolated
best guess according to (8)
difference
200
0
1960
-200
1970
1980
1990
Best guess of lead emissions in the 6-grid box AE surrounding
Münster in tons/year. The times with an estimate from Pacyna
is given by a big dot. The blue time series is used as input in
the reconstruction 1955-1995 of lead concentrations in the air
(5) and in the blood (7).
Institut für Küstenforschung
400
IfK
Equation 5/6
LCt+1 = α LCt + β LEt+1
LHBt+1 = γ LBHt + δ LCt+1 + ε
LC = lead concentration in the atmosphere
LE = total lead emission
LHB = lead concentration in human blood
Lead concentration LCt in the
air, in ng/m3, in the 50  50
km2 grid cell containing the
town of Münster as simulated
in the reconstruction, and as
estimated using the linear
model (5) (red) Münster
emissions LEt.
600
400
LCt+1 =  LCt +  LEt+1
200
(5)
modelling lead concentration
in air using (5)
6 cell AE
linearly interpolated LC
annual series LC
0
1960
1970
1980
1990
95% quantile = 1.75  mean – 4.6.
(7)
(4)
Fit of equation (7) and test of equation (4) for data set M.. The upper
two curves refer to the 95%-iles, and the lower two to the means. The
estimated mean curve is derived by using simulated air concentrations
in Münster and the 1981 observed blood level as initial value; the curve
for the estimated 95%-iles is obtained by using the estimated means and
applying formula (4).
Institut für Küstenforschung
(LHBt+1 - ) =  (LHBt - ) +  LCt+1
IfK
lead concentration in blood (g/l)
400
300
95%-iles
90%-iles
200
means
100
0
1955
1960
1965
1970
1975
1980
1985
1990
1995
Estimated mean, 90%ile and 95%ile lead concentrations in human blood in Münster,
according to (6/7) and (3/4 (red solid lines, M). Additionally the mean level estimated with
the Germany model is given as dotted blue curve (G). A level of more than 150 g/l are
considered in Germany as potentially harmful for children and women in child-bearing age
(HBM 3). For other adults the limit for serious concern is set to 250 g/l.
Institut für Küstenforschung
Three scenarios of reduction of lead used as anti-knock in
gasoline in Germany. The big black symbols describe the
actual concentrations.
IfK
lead concentrations in human blood
adults in Münster
scenario 1: means
scenario 1: 90%iles
scenario 1: 95%iles
scenario 2: 90%iles
scenario 3: 90%iles
300
200
100
0
1950
1960
1970
1980
1990
2000
Scenarios for mean lead concentrations (g/l) in human blood, as derived by the Münster
model. Scenario 1 describes an evolution without regulation (i.e., ongoing use of 0.6 g/l
lead in gasoline in Germany, upper curves). In scenario 2 no unleaded gasoline has been
introduced in Germany in 1985 (middle curves), and in scenario 3, regulation was
instituted in Germany already in 1961 (lower curves).
Institut für Küstenforschung
lead concentration in blood, g/l
400
IfK
Conclusions
• It is possible to reconstruct atmospheric lead
concentrations and blood lead levels using a
simple regression model
• It is possible to estimate lead concentrations in
human blood using only lead emissions
• Reducing the lead content in gasoline was a
successful environmental policy to limit human
health risks
Institut für Küstenforschung
Rest of the world
IfK
Institut für Küstenforschung
USA
After Kitman, Nation 270, March 20, 2000
IfK
May 8, 2001
New Warnings on
Lead and Children
By ERIC NAGOURNEY
new study raises questions about whether the current measure used to define lead poisoning is stringent
enough.
Lead poisoning has been redefined over the years, as doctors have decided that smaller and smaller amounts are
acceptable. It is now considered to occur at 10 micrograms of lead per deciliter of blood.
But researchers told a recent gathering of the Pediatric Academic Societies of evidence that even at levels lower
than that, lead in the bloodstream appears dangerous.
The primary researcher, Dr. Bruce Lanphear of the Children's Hospital Medical Center of Cincinnati, said that in
the children he and his colleagues had studied, I.Q. declined in those with less than 10 micrograms of lead per
deciliter of blood. The findings, he said, suggest that millions more children in the United States than previously
suspected may be at risk.
"This is clearly a major public health crisis, and there really is too much complacency about this as a public
health issue," he said.
The researchers studied 276 children born in five hospitals in Rochester, N.Y., and then kept track of them for
five years, measuring their lead levels and then their I.Q.'s at age 5.
The researchers reported finding an inverse relationship between I.Q. and lead levels. Among all the children,
they said, there was an average 5.5 percent reduction in I.Q. for every 10-microgram increase in lead.
Lead Use in Gasoline in 1996
Country Western
Europe
Lead Content in
Gasoline (g/l)
Market Share of
Leaded Gasoline (%)
Austria
0
0 (since 1993)
Belgium
0.013
26
Denmark
0
0
Finland
0
France
Germany
Lead Content in
Gasoline (g/l)
Market Share of
Leaded Gasoline (%)
Bulgaria
0.15
95
0
Croatia
0.6
70
0.013
38
Czech Republic
0.15
45
0.013
3
Hungary
0.15
36
Greece
0.4 (0.15 in Athens)
67 (since 1995)
Moldova
0.4
100
Iceland
0.013
15
Poland
0.15
30
Ireland
0.013
35
Romania
0.6
94
Italy
0.013
56
Russian Federation
0.6
50
Luxembourg
0.013
18
Slovak Republic
0
0 (since 1995)
Netherlands
0.013
14
Norway
0.013
2
Portugal
0.4
61
Spain
0.4
77
0
0 (since 1994)
0.013
13
0.4
82
0.013
33
Sweden
Switzerland
Turkey
United Kingdom
Country Central and
Eastern Europe
Source: modified from World Bank 1997
Thomas and Kwong, 2001
blood lead level (g/l)
Blood Lead Levels in Different Cities in 1980s and 1990s
years of sampling
US- scientists expect health dangers for children above a blood lead level of 100g/l.
German experts are convinced there can be health dangers above 150gPb/l.
Source: World Bank (1997), Heinzow et al. (1998)
http://w3g.gkss.de/staff/blei/index.html
Institut für Küstenforschung
Web-page
IfK