Comparative risk analysis of dioxins in fish and fine particles from heavy-duty vehicles.

Abstract

Dioxins and airborne fine particles are both environmental health problems that have been the subject of active public debate. Knowledge on fine particles has increased substantially during the last 10 years, and even the current, lowered levels in the Europe and in the United States appear to be a major public health problem. On the other hand, dioxins are ubiquitous persistent contaminants, some being carcinogens at high doses, and therefore of great concern. Our aim was to (a) quantitatively analyze the two pollutant health risks and (b) study the changes in risk in view of the current and forthcoming EU legislations on pollutants. We performed a comparative risk assessment for both pollutants in the Helsinki metropolitan area (Finland) and estimated the health effects with several scenarios. For primary fine particles: a comparison between the present emission situation for heavy-duty vehicles and the new fine particle emission standards set by the EU. For dioxins: an EU directive that regulates commercial fishing of Baltic salmon and herring that exceed the dioxin concentration limit set for fish meat, and a derogation (= exemption) from the directive for these two species. Both of these two decisions are very topical issues and this study estimates the expected changes in health effects due to these regulations. It was found that the estimated fine particle risk clearly outweighed the estimated dioxin risk. A substantial improvement to public health could be achieved by initiating reductions in emission standards; about 30 avoided premature deaths annually in the study area. In addition, the benefits of fish consumption due to omega-3 exposure were notably higher than the potential dioxin cancer risk. Both regulations were instigated as ways of promoting public health.

Full text

Page 1

Risk Analysis, Vol. 28, No. 1, 2008
DOI: 10.1111/j.1539-6924.2008.01005.x
Comparative Risk Analysis of Dioxins in Fish and Fine
Particles from Heavy-Duty Vehicles
Olli Leino,1∗Marko Tainio,1and Jouni T. Tuomisto1
Dioxins and airborne fine particles are both environmental health problems that have been
the subject of active public debate. Knowledge on fine particles has increased substantially
during the last 10 years, and even the current, lowered levels in the Europe and in the United
States appear to be a major public health problem. On the other hand, dioxins are ubiquitous
persistent contaminants, some being carcinogens at high doses, and therefore of great con-
cern. Our aim was to (a) quantitatively analyze the two pollutant health risks and (b) study
the changes in risk in view of the current and forthcoming EU legislations on pollutants. We
performed a comparative risk assessment for both pollutants in the Helsinki metropolitan
area (Finland) and estimated the health effects with several scenarios. For primary fine par-
ticles: a comparison between the present emission situation for heavy-duty vehicles and the
newfineparticleemissionstandardssetbytheEU.Fordioxins:anEUdirectivethatregulates
commercial fishing of Baltic salmon and herring that exceed the dioxin concentration limit set
for fish meat, and a derogation ( = exemption) from the directive for these two species. Both
of these two decisions are very topical issues and this study estimates the expected changes
in health effects due to these regulations. It was found that the estimated fine particle risk
clearly outweighed the estimated dioxin risk. A substantial improvement to public health
could be achieved by initiating reductions in emission standards; about 30 avoided premature
deaths annually in the study area. In addition, the benefits of fish consumption due to omega-
3 exposure were notably higher than the potential dioxin cancer risk. Both regulations were
instigated as ways of promoting public health.
KEYWORDS: Dioxin;EuropeanUnionlegislation;fineparticles;fish;riskassessment;riskcomparison
1. INTRODUCTION
Exposures to dioxins and ambient fine particles
are both ranked high as health hazards, but these pol-
lutants display many important differences. Data for
fine particle risk come mainly from epidemiological
studies whereas most of the information on dioxin
comes from toxicology. There are also differences in
their biological half-lives. Furthermore, exposure to
fine particles is rather uniform within a given area
1National Public Health Institute of Finland.
∗Address correspondence to Olli Leino, National Public Health
Institute of Finland, PO Box 95, FI-70701 Kuopio, Finland.
while exposure to dioxins varies according to food
consumption habits. This leads to another difference
between these two risks. Fine particle exposure is
perceived as an unavoidable risk, whereas the risk
from dioxin can be individually controlled, at least
to some extent.
Dioxins are a group of highly toxic chemi-
cals. The most potent dioxin congener is 2,3,7,8-
tetrachlorodibenzo-p-dioxin (TCDD). Due to their
lipophilicity, dioxins are very slowly metabolized and
excreted, thus they bioaccumulate and become bio-
magnified in wildlife and humans. We use the term
“dioxin” in this study to refer to polychlorinated
dibenzodioxins and dibenzofurans (PCDF) and
127
0272-4332/08/0100-0127$22.00/1C ?2008 Society for Risk Analysis

Page 2

128Leino, Tainio, and Tuomisto
polychlorinated biphenyls with dioxin-like toxicity
(DL-PCB). Dioxins have been demonstrated to be
animal carcinogens at high doses. The International
AgencyforResearchofCancer(IARC)oftheWorld
Health Organization (WHO) has classified TCDD
as a group 1 human carcinogen.(1)They have been
linked to many serious health effects, especially in
animals and also in humans, including cancer, repro-
ductive and developmental effects, altered immune
function, and disruption of the endocrine system.
Dioxins are believed to be a powerful cancer pro-
moter, rather than an initiator.(2)
The ecosystem of the Baltic Sea has been badly
polluted by dioxins. The EU has set the maximum
dioxin concentration of 8 pg/g (WHO-TEQ in fresh
weight) for fish products.(3)However, the dioxin
concentrations of wild salmon and herring from
the Baltic Sea frequently exceed 10 pg/g (WHO-
PCDD/F-PCB-TEQ in fresh weight).(4)In com-
parison, wild salmon from north-east Europe dis-
play dioxin concentrations of approximately 2 to
3 pg/g (WHO-PCDD/F-PCB-TEQ in fresh weight)
and salmon from South and North America have
less than 2 pg/g (WHO-PCDD/F-PCB-TEQ in fresh
weight).(5)In Finnish-farmed salmon, the concentra-
tions of dioxins are lower since these fish are fed
cleaner fish feed compared with the diet of wild
salmon in the Baltic Sea.(4)In Finland, the principal
human exposure from dioxins comes from fish, with
fish from the Baltic Sea being the main source.(6)
In 2001, EU authorized a five-year transitional
period for Finland and Sweden to allow Baltic her-
ring and salmon to be sold on their domestic mar-
kets. During this five-year period, countries were
obligated to study the health effects due to the con-
sumption of these fish species. In the year 2006,
Finland and Sweden were granted another transi-
tional period, up till the end of the year 2011 (EC
199/2006).(3)Again, studies about health risks and
benefits due to consumption of these fish will play an
important role in the decision making concerning fu-
ture regulation due in 2011.
Airborne ambient fine particles with aerody-
namic diameter less than 2.5 μm (PM2.5) are one of
the major environmental health problems in modern
Western societies. Fine particles have been linked
to several adverse health effects. The adverse health
effects have been seen in both short-term (daily
variations)(7)and long-term (chronic)(8)studies. The
strongest association has been found between ambi-
ent particulate matter (PM) and elevated cardiopul-
monary mortality, lung cancer mortality, and re-
duced lung function.(9)The Clean Air for Europe
(CAFE) program, funded by the European Commis-
sion, claimed that fine particles are responsible for
over 300,000 premature deaths annually in Europe
(EU25) and lower the average life-expectancy by
8.6 months.(10)
In Finland, traffic and domestic wood combus-
tion are the main sources of primary fine particles.(11)
Emissions of particles due to traffic were highest
in the 1980s.(12)Changes to fuel composition, espe-
cially the decline in the levels of sulfur compounds,
have lowered the particle emissions. A major de-
crease took place in 1994, when reformulated fuels
entered general use.(12)At present, heavy-duty vehi-
cles are responsible for 60% of the total fine particle
emission of road traffic in the Helsinki metropolitan
area, although the number of heavy-duty vehicles ac-
counts for only 13% of total number of vehicles on
the roads.(12)That is, heavy-duty vehicles emit more
fine particle emissions than the automobiles powered
by gasoline engines. For this reason heavy-duty vehi-
cles are of particular interestin any attempt toreduce
health effects of traffic-generated fine particles.
The aim of the study was to carry out a compar-
ative risk assessment of these two pollutants and to
compare health effects of the two regulations being
initiated by the European Union.
2. MATERIALS AND METHODS
We chose the Helsinki metropolitan area as the
geographical area. In this way, we could gain full ac-
cess to the actual road traffic data measurements per-
formed in the Helsinki metropolitan area and define
the estimated risk of fine particles more accurately
than elsewhere in Finland. To estimate the dioxin
risks due to fish consumption, we calculated the risk
for the Finnish population and scaled it down to the
population of the Helsinki metropolitan area. We as-
sumed that the citizens of the Helsinki metropolitan
area would have similar fish consumption patterns as
the rest of the Finnish population.
We had to use toxicological information to esti-
mate the dioxin risk and epidemiological information
to estimate the fine particle risk. When there was a
discrepancy, we preferred to utilize assumptions ex-
aggerating rather than understating the risk due to
dioxins. This was because our prior hypothesis was
that the estimated dioxin risk would be smaller and
we wished to minimize the probability of encounter-
ing a false negative result for the dioxin risk.

Page 3

Comparative Risk Analysis of Dioxins and Fine Particles129
Table I. Mortalities and Population
Characteristics of the Helsinki
Metropolitan Area in 2004
Helsinki Metropolitan Area Mortality StatisticsValueICD-10 codes
Population size
Mortality rate
Total cancer mortality
Total mortality
Lung cancer mortality
Nonaccidental deaths
Cardiopulmonary mortality
Traffic-related fatalities in Helsinki
CHD mortality
980,412
0.007454
1,727
7,308
C00-D48
A-Q
C34
Total mortality-V01-Y98
I11-I70 and J15-J47
V01-V99
I21,I22 and I20, I24, I25
313
6,560
2,888
62
1,488
For demographics statistics, we used the database
from Statistics Finland(13)and for mortality data,
data from Statistics Finland(13)combined with the
WHO database.(14)The coronary heart disease mor-
tality estimate consisted of acute myocardial infarc-
tion and other ischemic heart diseases.(15)Mortality
statistics are summarized in Table I.
2.1. Scenarios
We estimated the health effects for the alter-
native scenarios. EU has set emission standards for
the fine particle emissions of new heavy-duty vehi-
cles. The fine particle emission standards scenarios
are called EURO IV and EURO V, which have the
same emission limit of 0.02 g/kWh(16)for particles.
Therefore, we combined these two scenarios into
one scenario, EURO IV and V. We compared this
EURO IV and V scenario to the present situation
“business as usual” (CURRENT PRACTICE PM).
EURO standards represent total suspended parti-
cles, but we assumed that virtually all of the particles
are < 2.5 μm.
The two decision alternatives concerning diox-
ins were based on the commission regulation (EC)
N:O 1881/2006,(3)setting maximum levels for certain
contaminants in foodstuffs (see Table II). EU has
set the directive for dioxins (scenario NO DERO-
GATION), which regulates the consumption of fish
products exceeding dioxin concentration of 8 pg/g
WHO-PCDD/F-PCB-TEQ. However, Finland and
Sweden have been granted an exemption (scenario
DEROGATION) for Baltic salmon and herring.
These scenarios, based on the EU directives, are
used in the model and are described in Table II.
In the case of dioxins, we used premature cancer
deaths as the endpoint; and for fine particles, we
used cardiopulmonary, lung cancer, and other nonac-
cidental causes of death.(17)
2.2. Fish Consumption and Dioxins
The major part of Finnish dioxin exposure comes
from fish. This is because the Baltic Sea is heav-
ily contaminated with persistent organic pollutants
suchasdioxinandpolychlorinatedbiphenyls(PCBs),
while the environment is otherwise relatively clean
of dioxins. Typical sources in other countries, such as
dairy products or meat, make only a small contribu-
tion to the total dioxin exposure in Finland.(6)There-
fore,itisverydifficulttoreduceFinnishdioxinintake
without affecting fish intake. It is therefore necessary
to study the collateral effects, that is, the detrimental
effects on health, of reduced fish consumption when
evaluating the overall risks of dioxin.
We selected the most common species available
for consumers in Finland, including farmed salmon
(Salmo gairdneri), wild salmon (includes wild salmon
(Salmo salar), wild rainbow trout (Salmo gairdneri)
andwildtrout(Salmotrutta)),herring(Clupeaharen-
gusmembras),whitefish(Coregonuslavaretus),sprat
(Sprattus sprattus), perch (Perca fluviatilis), flounder
(Platiochthys flesus), pike-perch (Stizostedion luciop-
erca), bream (Abramis brama), pike (Esox lucius),
vendace (Coregonus albula), and burbot (Lota lota).
The fishery catch data were obtained from
the Finnish Game and Fisheries Research Insti-
tute (RKTL). Recreational and commercial fishery
catches were 40,952 metric tons and 109,025 metric
tons, respectively, in 2002.(18)These values include
both sea areas and fresh waters. Units were reported
in fresh weight, that is, uncleaned, so filleting factors
for the different species were used in order to ob-
tain gutted weight. The filleting factor is a ratio of
the gutted fish weight and whole fresh fish weight
and this variable includes uncertainty estimated by
the experts of the RKTL and varies between species.
The herring species exhibits a strong correlation be-
tween size and dioxin concentration. Therefore, we
included size distribution of the fishery catch for her-

Page 4

130Leino, Tainio, and Tuomisto
Table II. The Fine Particle Emission
Scenarios by the EU for New
Heavy-Duty Diesel Engines and the Fish
Consumption Scenarios
PollutantEndpointScenario Description
ParticlesCardiopulmonary and lung CURRENT PRACTICE
cancer mortality due to
heavy-duty vehicles
Business as usual
0.077 g/kWh
EURO IVand V
Commission regulation
98/69/EC and 99/96/EG
0.02 g/kWh
No derogation (salmon and
herring must meet 8 pg/g)
Derogation (salmon and
herring exempted)
DioxinTotal cancerCommission regulation
EC 1881/2006
Commission regulation
EC 199/2006
ring. Finally, we estimated the proportion of fish (by
species) that will actually be consumed by humans.
The reminder of the catch is used as animal feed,
waste, and for other purposes.(18)In this way, we ob-
tained an estimate of consumption of Finnish fish.
The pollutant concentrations of fish were ob-
tained from the National Food Agency of Fin-
land.(4)Dioxin concentrations of the different species
ranged from 0.2 to 14 pg/g (WHO-TEQ in fresh
weight), large herring and wild salmon being the
species with the highest concentrations and fresh wa-
ter fish in general exhibiting the lowest concentra-
tions. Samples included skin and ventral fat. This
approach overestimates the concentration of diox-
ins in the edible part, as not everyone consumes
these parts as food. In addition, we assumed a lin-
ear exposure-response relationship for excess can-
cers associated with dioxin intake as reported in the
IRIS database(19)of the U.S. Environmental Protec-
tion Agency (EPA). The cancer slope factor (CSF)
for TCDD is 156,000 per mg/kg/day. The estimated
pollutant health risk was calculated assuming addi-
tivity between the pollutants. All cancer cases were
assumed to be lethal.
Estimated risks from consuming Finnish fish
were calculated in commensurable units, premature
deaths, because this is readily comparable with both
fineparticlesandconsumptionoffish.Nonlethalend-
points, for example, developmental effects, were not
quantitatively taken into consideration in this study.
The exposure was calculated as the product of
the pollutant concentration of fish and the fish con-
sumption and the estimate of risk was the product of
exposure-response, exposure, and background mor-
tality.
A number of studies have shown the beneficial
effects of omega-3 fatty acids in the reduction of
coronary heart diseases (CHD).(20–24)CHD includes
acute myocardial infarction and other ischemic heart
diseases. In particular, fatty fish species, like salmon
and herring, are rich in omega-3 fatty acids. For
evaluating the concentrations of omega-3 fatty acids
of fish species, we used the nutritional database
Fineli,(25)maintained by the National Public Health
Institute, Finland, and scientific articles as reference
values.(22,23)Omega-3 fatty acids are also associated
with some other beneficial endpoints, for example,
risk reduction of stroke, improved cognitive devel-
opment, prevention of depression, and decrease in
hypertension.(26,27,28)These results are less definitive
and the effects of these endpoints on public health
would be smaller than that of CHD, so they were not
taken into account in this study.
We were careful not to overestimate the ben-
eficial effects of omega-3 fatty acids. A large pro-
portion of the omega-3 benefit literature is based
on studies of cardiac patients. We included a factor
that reflected the uncertainty whether there was car-
diac health benefit for everyone or only for CHD pa-
tients. According to Mozaffarian and Rimm,(29)mod-
est consumption (250–500 mg/day) of omega-3 could
reduce CHD deaths by 14.6% per each 100 mg/day
of omega-3 exposure. After this limit, no extra bene-
fit was assumed from omega-3 fatty acids in terms of
reducing CHD incidence.
The estimate of the health effects was calculated
as the product of omega-3 concentrations in the dif-
ferent fish species, consumption of fish by species,
and background mortality.
2.3. Fine Particles Emitted by Heavy-Duty Vehicles
The estimated risks due to primary fine particle
emissions were based on a recent study, which es-
timated emissions, exposure, and associated health

Page 5

Comparative Risk Analysis of Dioxins and Fine Particles 131
effectsofprimaryfineparticlesduetolocalbustraffic
in the Helsinki metropolitan area.(30)Brief overviews
for the exposure and health effect submodels are de-
scribed in the following two paragraphs. The emis-
sion submodel was totally renewed for this study and
itis described after the exposure and the health effect
submodels.
Annual average population exposure to traffic-
emitted primary PM2.5in the Helsinki metropolitan
area was estimated using two alternative exposure
models. The first model was based on the EXPOLIS-
Helsinki study,(31)in which the observed average ex-
posure to total PM2.5 in this area was 10.7 μgm−3
in 1996–1997.(32)The average exposure was appor-
tioned to source categories using elemental composi-
tions. The exposure fraction attributable to the local
traffic emissions was separated from the source-
categorized results by comparing the emission
rates of different emission sectors. In an alterna-
tive approach, exposure was calculated based on
ULTRA study, in which the contribution of local
traffic emissions was analyzed by using an absolute
principal component analysis and multivariate linear
regression, based on both particle and gaseous air
pollutant concentrations.(33)
An exposure-response submodel described the
slope of the exposure-response function and the
plausibility of the PM2.5 health effect. Only mor-
tality due to long-term PM2.5exposure was consid-
ered. The exposure-response coefficient for three
mortality outcomes (cardiopulmonary, lung cancer,
and other nonaccidental) were estimated by using
values with equal probability from the result dis-
tributions reported in Dockery et al.(34)and Pope
et al.(17)They assumed that the exposure-response
function was linear with no threshold. The plausibil-
ity of the estimated health effects was included in
the exposure-response submodel using author judg-
ment. Plausibility was defined as the probability
that the observed exposure-response relationship ac-
tually represents a causal association. Background
cardiopulmonary (International Classification of
Disease (ICD-10) codes: I11-I70 and J15-J47), lung
cancer (C34), and total mortality (A-Q) were 2,888,
313, and 7,308 deaths per year, respectively, in the
Helsinki metropolitan area in 1996.(13)(see Table I).
Anemissionsubmodelwascreatedforthisstudy.
Data for the emission submodel were received from
the LIISA emission model maintained by the Techni-
cal Research Centre of Finland (VTT).(12)The emis-
sion model included annual fine particle emissions of
all heavy-duty vehicles in the cities of Helsinki, Van-
taa, Espoo, and Kauniainen. Emissions were calcu-
latedbydataofroadandstreettrafficvolumeincases
of the cities of Helsinki, Vantaa, and Espoo. Emis-
sions of the municipality of Kauniainen were calcu-
lated by average Finnish road and street traffic data
in proportion to the population of the city of Kau-
niainen. To calculate the present situation (CUR-
RENT PRACTICE PM), we used also the data of
VTT.(12)
2.4. Simulation
The variables and the uncertainty distributions
included in the model are summarized in Table III.
The whole model was implemented using the Ana-
lytica TM version 3.1.1 (Lumina Decision Systems,
Inc., CA, USA) Monte Carlo simulation program.
We used Latin hypercube sampling and the model
was run with 20,000 iterations. An illustrative de-
piction of the graphical layout of the model is pre-
sented in Fig. 1. A more detailed description of this
type of illustration can be found from an article
and the model by Tuomisto and Tainio.(35,36)The
complete model of this study is published on the
HEANDE webpage. For a more detailed descrip-
tion of the variables and calculation, please see that
model (URN:NBN:fi-fe20071159).(37)
Uncertainty analysis was performed by calculat-
ing absolute rank-order correlations between the un-
certain input variables and the model outputs.
3. RESULTS
TheestimatedhealthriskduetodioxinsfromFinnish
fish was 1.2 cancer deaths (90% confidence interval
1.1–1.4) per year in the Helsinki metropolitan area
population (980,412 inhabitants, year 2004). Most of
the estimated total cancer risk was due to PCDD/F;
PCBs were responsible for only 13% (0.16 cancer
death) of the total pollutant risk. Over 50% of the to-
tal risks of dioxins was attributable to large (size over
17 cm) Baltic herring. The extent of Finnish herring
consumption has been declining in recent years. Ac-
cording to RKTL, in 2005, it was approximately 20%
of the total fish consumption.(38)
In the NO DEROGATION scenario, the can-
cer deaths would be decreased by 0.7 per year due
to reduced dioxin exposure. At the same time, there
would be almost 40 more CHD deaths due to di-
minished omega-3 intake (see Table IV and Fig. 2).
The net health effect, annual avoided CHD deaths,
of consuming Finnish fish are 170 (90% CI 50–350)
and 140 (90% CI 40–270) in scenarios DEROGA-