Comparative Environmental Health

Assessments
A Brief Introduction and Application in China
Kirk R. Smith
Professor of Global Environmental Health, University of California,
School of Public Health, Berkeley, California, USA

Conceptual and methodological issues in calculating and comparing the health impacts
from environmental risk factors in ways that are not only compatible across environ-
mental hazards but also can be fairly compared to burdens from nonenvironmental risk
factors, such as poor nutrition, unsafe sex, and smoking, are discussed. It is emphasized
that a focus on environmental health burden does not always produce priorities that
correspond to those related to environmental quality alone. The methods when applied
to China’s environmental and other risks using the Chinese burden of disease in terms
of lost healthy life years as the metric are illustrated. Household environmental risks
are still quite important in China, because of rural poverty, but have been exceeded
by community environmental risks nationally. Global risks from climate are small at
present, but have the potential to rise. Although not a major greenhouse gas emitter
on a per capita basis compared to rich countries, China has already passed the thresh-
old of imposing more global risk than it receives. The study ends with the suggestion
that environmental risk assessment should use as a baseline estimates that are based
on methods developed in international collaborative assessments, such as those in the
WHO Comparative Risk Assessment, in order to foster comparability and policy and
public confidence in the methods.

Key words: burden of disease; environmental risk transition; household environmental

health risks; climate-change health risks; community environmental health risks

Introduction fuzziness about the boundaries of what consti-

tute “environmental” risk factors. The old “na-
It is well recognized that environmental risk ture versus nurture” debate might lead some to
factors are important determinants of human conclude that anything that is not genetic (na-
health, sometimes being listed along with ge- ture) is left to the environment (nurture). Even
netic/biological and behavioral/social deter- genes, however, are the result of environmental
minants as the triad of primary influences on processes (mutation and natural selection), and
health. As these three groups of factors cover thus environment might be thought to include
nearly all the known universe, however, and “nature” as well, that is, everything. Logical as
in detail are sometimes difficult to disentan- this might be, it is not useful. For clarity, there-
gle from one another, the operational implica- fore, I take a narrow version of what constitute
tions of such a triad are not immediately clear. environmental health risk factors; those involv-
Partly this is due to a high degree of remaining ing identifiable physical/biological agents me-
diated through environmental pathways. Thus,
it includes passive but not active smoking, food
Address for correspondence: Kirk R. Smith, Professor of Global contamination not nutrition, among others. It
Environmental Health, University of California, School of Public Health,
Berkeley, CA 94720-7360. Voice: 510-643-0793; fax: 510-642-5815.
does include non-human-engendered agents,
[email protected] however, such as adverse weather, natural
Ann. N.Y. Acad. Sci. 1140: 31–39 (2008). 
C 2008 New York Academy of Sciences.
doi: 10.1196/annals.1454.032 31
32 Annals of the New York Academy of Sciences

contaminants, such as arsenic, and the many does not directly threaten human health,
other impacts that are not created by human as tragic as it is for other reasons. In-
actions per se, but rather exist because of hu- direct pathways certainly exist, but they
manity’s attempt to keep healthy in an uncer- tend to be weaker in effect and far more
tain and sometimes hostile natural world.1 uncertain.
Importantly in addition, here I give priority 2. We have grown to decry the outrage of
to the second word of the term “environmen- much of environmental contamination,
tal health” by examining how understanding for example, the factory dumping waste
environmental factors can affect efforts to im- into the river, but often forget that just be-
prove overall health rather than the reverse, cause something is outrageous does not
that is, how understanding of environmental mean it is particularly unhealthy. The two
health risks can affect efforts to improve envi- are just not on the same axis. U.S. na-
ronmental quality (EQ). This may seem a trivial tional investments for cleaning up aban-
change in framing, but actually produces a pro- doned toxic-waste dumps illustrate how
found shift in understanding and, potentially, an outrageous situation has led to ma-
priorities. jor improvements in EQ with little actual
Protecting and improving EQ is an im- improvement in health in comparison to
portant goal for society, but one that is not the expenditure. People are not exposed
equivalent and can sometimes actually be much to toxic material even when living
counterproductive to protecting and improv- nearby, except in a few cases, although
ing environmentally mediated ill-health. This many hundreds of thousands have had
diversion is due to several factors, inter alia: their peace of mind and property values
protected by cleanups.
1. The basic indicators of progress in the 3. Scientific ability to measure small amou-
two arenas are not equivalent and in nts of contaminants has improved by fac-
some cases not highly correlated. EQ tors of millions in recent decades. Thus,
relies on ambient and widespread mea- a toxin that could once only be found at
sures, such as outdoor air pollution, for- the ppm level, might now be measured
est destruction, and river contamination, easily at the ppt level. It is only a slight
but environmental health relies on mea- stretch to say that, with persistent pollu-
sures of human exposure, which are of- tants, science is trending toward a time
ten heavily influenced by local factors when everything can be measured every-
that do not significantly affect EQ , such where if the time and money are spent. In
as the proximity of people to the pollu- such a circumstance, what is the meaning
tion source. The same amount of an air of EQ or environmental health—certainly
pollutant released indoors, for example, not just the presence of the contaminant?
produces roughly a thousand times more This quandry is not new, after all the pri-
exposure than it would outdoors. Thus, mary directive of environmental health
environmental tobacco smoke in many that the dose (exposure) makes the poi-
parts of the world dominates air-pollution son, was proposed some 500 years ago.
exposures even, with essentially no impact Modern measurement science, however,
on EQ. The relationship can be inverse, is forcing us to focus more fully on its
that is, even heavy contamination of rivers implications.
does not lead to much human exposure 4. Scientific ability to measure relative risk
in many circumstances because people do (increase in ill-health per unit exposure)
not drink river water directly in most parts has also improved dramatically in recent
of the world. Destruction of primary forest decades, due to great strides in toxicology
Smith: China’s Burden of Environmental Disease 33

and epidemiology, the primary environ- 7. For many pollutants, the lack of atten-
mental health risk sciences. In epidemiol- tion also seems to be part of the still-
ogy, it is now possible to estimate relative common feeling that “natural” things are
risks for some indicator of ill-health, even somehow benign (or even beneficial), in
if not actual disease, for what used to be spite, for example, of toxicologic evidence
considered very small exposures. In the that natural pesticides are about as toxic
case of some end points, cancer in partic- as artificial ones, woodsmoke as toxic as
ular, there is now general acceptance of fossil-fuel smoke, and natural arsenic in
using animal experiments at high expo- drinking water a large health burden.
sures to do so and then extrapolating to 8. A final sharp dichotomy exists between so-
human exposures at low levels, thus pro- ciety’s evident concerns over public health
viding risk estimates with no human data. risks compared to those affecting workers.
5. A consequence of the scientific advances In spite of clear evidence that workers in
in measurement and risk assessment has all societies, rich and poor, experience ex-
been that fewer and fewer important con- posures to nearly any toxic agent at much
taminants have identifiable thresholds of higher levels than the public at large,
effect. Even if possibly the case for an in- all societies seem content with protecting
dividual, the wide variety of vulnerability and in general worrying about them less.
in the population makes the concept of There are a number of factors that play
threshold dubious—someone somewhere a role, for example, the impression that
is likely to be substantially more sensitive workers may somehow choose to take on
than the mean. As a result, pollutant lev- the risk voluntarily in compensation for
els that might once be considered accept- the job, but the fact remains that mea-
able because they were below the identi- sured either by total burden of disease or
fied threshold at the time, that is, “safe” by risk per person, workers’ environmen-
are no longer so and, indeed, might thus tal health risks are greatly underappreci-
be considered “outrageous” in some cir- ated and undercontrolled.
cumstances.
Perhaps “outrage” might be considered the
6. Without someone to blame, it is difficult to
main thread running through these various is-
generate interest on the part of many sci-
sues. We can all probably agree that a human
entists, policymakers, and environmental
activity found to measurably expose the public
nongovernmental organizations about an
to a pollutant that has had a risk study done
environmental health problem, even one
showing a health-related effect is outrageous
with a large health impact. Indoor air pol-
and ought to be stopped. What is not clear at
lution in poor households is an example—
all from this statement, however, is when, that
no big industry to blame. At most, such
is, what priority it ought to have among all
environmental health problems are acts
the other demands to improve health, which
of omission rather than ones of commis-
are legion just within environmental health, let
sion, that is, the only ones to blame are
alone across all health risk factors. There are
governments (or you and I) that do not
several additional steps needed to answer this
take action even though they (we) did not
question
cause the problem. Perhaps also it is ra-
tional understanding of the need to make I. If everything is measured everywhere (3
the situation seem “outrageous” in order in the preceding list), everything is toxic
to successfully find public funding or me- (4 in the preceding list), and thresh-
dia sympathy. Nothing like an identifiable olds no longer exist (5 in the preced-
villain to generate real outrage. ing list), then qualitative distinctions fail
34 Annals of the New York Academy of Sciences

and quantitative distinctions must prevail. list is the Global Comparative Risk Assessment
No longer can we rely on establishing managed by the World Health Organization’s
“safe” levels that are not therefore “out- Global Burden of Disease Project.2 Published
rageous,” but must instead conduct quan- in 2004, it responds to each of the needs out-
titative assessments of effect. Only in this lined:
way can we know how to apply our efforts
I. Quantitatively compares risks based on
effectively.
standard population attributable-burden
II. If health improvement is our goal, then
calculations applied to detailed estimates
factors, such as outrage, villainy, volun-
of exposure distributions and systematic
tariness, naturalness, commission, or lo-
review of epidemiological evidence by
cation in a workplace, are irrelevant, at
age, sex, and 14 world regions for 26
least at the level of the initial compara-
major risk factors, including 6 mediated
tive analysis. We may choose to not spend
through environmental pathways.
resources on controlling worker risks for
II. Is based on carefully developed proce-
political or public perception reasons, for
dures and concepts that treat all popu-
example, but these factors should not play
lations, risk factors, and disease outcomes
a role in our primary comparison.
equally. Impacts are weighted differently
III. If comparison across possible ways to de-
only according to the age and sex of
vote our resources is essential to deciding
those affected, universal human experi-
on how to proceed, then we need to ac-
ences, and not by any nonuniversal cri-
cept common metrics in doing so. It is
teria, such as income group affected, ge-
just not possible to fairly compare risks
ographical location, or whether the risk
using deaths alone without considering
factor is undertaken voluntary or is part
age patterns. In addition, the large variety
of natural risks.
of nonfatal conditions (disease and injury)
III. Uses a lost healthy life-year metric (the
that affect health are not easily combined
disability-adjusted life year (DALY)) as the
together or with mortality. Think just of
metric of ill-health, thus providing a way
those end points often found in studies of
to compare disparate health outcomes, fa-
outdoor air pollution alone (hospital ad-
tal and nonfatal. Although other metrics
missions, asthma attacks, reported respi-
might be considered to have better char-
ratory symptoms, physician visits, use of
acteristics by some observers (e.g., Ref. 3),
medication, school absences, cancer diag-
there is no other systematic and coher-
noses, premature births, etc.). How would
ent global database available for any but
these be compiled into a common metric
the DALY, which essentially constrains in-
that could even compare the impacts in
ternational activities, such as the Com-
different populations or different emission
parative Risk Assessment (CRA), to its
sources, let alone compare with impacts
use.
from water pollution, high blood pressure,
and unsafe sex to know where our efforts I refer readers to the basic references for the
might be most effectively applied to pro- CRA and its parent project, the Global Bur-
tect health? den of Disease, for further detail on its history,
conceptual basis, methods, procedures, and ev-
The only international effort to compare im- idence bases.2 Here, however, to illustrate its
pacts across risk factors, environmental and utility in gleaning insights into environmental
other, in a way that systematically and trans- health risks, I discuss some of its results in the
parently address tasks I–III in the preceding context of China.
Smith: China’s Burden of Environmental Disease 35

Figure 1. Distribution of major causes of death in

China for 2002.12

Figure 2. Distribution of major causes of lost

Chinese Burden of Environmental disability-adjusted life years (DALYs) lost in China for
2002.12
Disease

Figure 1 shows the pattern of mortality in

China, showing the importance of a fairly small
number of causes—cancer, stroke, and chronic
obstructive pulmonary disease (COPD), which
together account for more than half of all
deaths. China has the pattern of other middle-
income countries, except perhaps for unusually
high COPD and stroke mortality. As deaths
do not account for age or for nonfatal diseases
and injuries, however, Figure 2 based on lost
Figure 3. Burden of disease (disability-adjusted
healthy life years (DALYs) is a better indication life years (DALYs)) by major risk factors in China for
of the true pattern of ill-health—the target list 2000. Note that unlike Figures 1 and 2, these at-
for public health interventions. Note first that tributable burdens cannot be directly added as there
no diseases dominate the total as much as for are overlaps in some cases. (From Reference 13
deaths and that 50% of the burden is reached based on data in Reference 2.)
only by including the top 11 diseases. Note also
the mix of the diseases of affluence, for exam- from which one can make valid comparison
ple , cancer, stroke, and ischemic heart disease, across ages, sexes, parts of the world, and risk
and of poverty, for example, diarrhea, TB, and factors. The result is shown in Figure 3, a which
perinatal conditions. Finally, as with nearly all shows the result by risk factor, most of which
parts of the world, depression is quite impor- contribute to several diseases in Figure 2.
tant, even though it does not feature as a direct Note first the importance of substance abuse,
cause of death. alcohol and tobacco, in the country. The bur-
The CRA for China is based on the 2000 den of tobacco is also growing rapidly as the
burden of disease, as in Figure 2, that is, it delayed impact of China’s smoking patterns
is constrained by the total ill-health found in becomes manifest. With two-thirds of men
the country. This characteristic is vital for pro-
ducing a believable coherent result, one within a
Percentages based on the WHO region, WPRO-B, which is, by pop-
which all risk factors are treated equally and ulation, ∼85% China.
36 Annals of the New York Academy of Sciences

smoking, tobacco in China is causing one of the way toward explaining the high rates of COPD
largest health tragedies in history as epidemi- among nonsmoking women in the country.
ology indicates that at least half of them will Although the smoke from coal, which is
die prematurely from the habit.4 Female smok- widely used for cooking and heating in rural
ing rates, although traditionally low, are also Chinese households, is a well-established hu-
starting to grow. Within a few years, tobacco man carcinogen6 and the exposures to women
will exceed all other risk factors as a cause of cooks are high, the total burden of lung can-
ill-health in the country. cer is small compared to that from ALRI and
That both malnutrition (underweight) and COPD—perhaps only 5% of the total.7 This
overweight (obesity) made the top 10 risk fac- relatively small impact from what might seem
tors is a stark indication of the growing inequity to be a large exposure to a known carcinogen
in the country between rural and urban areas illustrates a discrepancy in how the environ-
and, to some extent, between east and west. mental health community has treated cancer
Kids are still malnourished in western villages compared to other health outcomes. Risks of
while adults in Shanghai and Guangzhou are one per million or even 100 per million, which
becoming fat. This fits with other evidence, are commonly considered levels of concern for
for example, that, by income, China is now carcinogens, translate into rather small relative
the most inequitable nation in Asia east of risks compared to other risk factors, such as
Afghanistan with an inequity ratio (income of particle air pollution. One hundred per million
the richest 10% divided by income of the poor- lifetime cancer risk is equivalent to a relative
est 10%) of 18.5 risk of only about 1.0001; miniscule compared
The lack of development in rural areas is to typical relative risks for Chinese particle pol-
also evident in the pattern of ill-health from lution indoors or outdoors of >1.5. It may be
environmental risk factors. Chief among the that cancer is more to be feared than COPD,
public environmental risks are those associated but it seems unlikely that most people would
with rural poverty—indoor air pollution from think that the difference is orders of magni-
solid fuels and poor water/sanitation. The bur- tude. In any case, in the CRA all diseases are
den from either exceeds that from either urban treated with the same metric of lost DALYs—
outdoor particle pollution or lead pollution, al- no allowance is given for fear of certain dis-
though these are still associated with hundreds eases in some cultures. This means that can-
of thousands of premature deaths. Indeed, each cer risks from pollutants, even those with what
of these risk factors exceeds 1% of national lost might be considered high exposures and high
DALYs, levels that can be considered to be ex- potencies, do not produce much impact. This
tremely high. is why exposures to carcinogenic chemicals do
Compared to other poor rural areas in the not rate a separate appearance in Figure 3, al-
world, China has a similar burden of disease though they do play a role in the impact of some
from indoor air pollution, but somewhat less major risk factors, for example, smoking and
from poor water/sanitation, perhaps because occupation.
of the long tradition of using boiled water for As noted earlier, the environmental health
cooking and somewhat better access to life- burden falls heavily on workers in China, as it
saving health care for acutely ill children, the does in many countries. Indeed, occupational
main cause of lost DALYs from this risk factor. risks are ranked fifth among risks of all kinds.
Similarly, in contrast to rural India, the bur- On a per capita basis, the risk is higher still,
den of indoor air pollution is much more borne since these risks tend to be concentrated in cer-
by women in the form of COPD than by chil- tain industries employing a small fraction of
dren in the form of acute lower respiratory dis- the population. The insert shows the distribu-
ease (ALRI). Indeed, this risk factor goes a long tion of the burden among different types of
Smith: China’s Burden of Environmental Disease 37

ation, in general the burden of this class of

environmental risks first rises with develop-
ment, peaks in middle income countries, and
again falls in rich countries, an inverted U-
shaped curve sometimes referred to as the En-
vironmental Kuznets Curve. China lies close
to the maximum for these risks at present
driven by traffic accidents and occupational
hazards.
The only global environmental risk quanti-
fied in the CRA was that due to climate change.
It is important to understand that for all risk
Figure 4. Global distribution of burden of dis- factors the values shown in the figure are at-
ease (disability-adjusted life years (DALYs)) for major tributable risks, i.e., the degree of ill-health that
environmental risk factors by income in 2000.8 would not exist today if exposure to the risk fac-
tor had been at some feasible lower level in the
occupational hazards. Note the importance of past. This is distinct from what might actually
injuries and particle pollution. As might be ex- be considered the more policy-relevant value,
pected, 85% of the burden falls on men as the avoidable risk, i.e., the amount of ill-health in
compared, for example, to indoor air pollution the future that could be averted if we took action
where two-thirds of the adult burden falls on today. For those many risk factors that are not
women. changing rapidly one way or another on their
own, the difference between avoidable and at-
tributable may not be large. For climate change,
Chinese Burden Compared however, the fact that attributable risk by 2000
to Other Countries (base year for the CRA) was relatively modest
(150,000 premature deaths; 0.15% of global
Figure 4 shows the distribution of the burden DALYs) does not provide comfort, because the
of major classes of environmental risk factors avoidable risk is likely many times larger if noth-
by development status, here defined as pur- ing is done. Thus the scale of global risk in
chasing power per capita in international dol- Figure 4 is likely to increase steadily, making
lars (United Nations Development Program). A avoidable much larger than attributable risk.
quite similar pattern is found if using the Hu- On the other hand, there is reason to think that
man Development Index (this section is closely the pattern of the burden shown in the figure
based on Reference 8). will not change drastically, i.e., it will continue
Household environmental risks quantified to fall mostly on poor countries and, in such
here include indoor air pollution from solid fuel countries, on children.9
use, poor water/sanitation, and lack of malaria Although China has had rapidly growing an-
control. As can be seen in Figure 4, these risks nual CO 2 emissions in recent years, its Natu-
decline steadily with economic development. ral Debt (cumulative historical emissions de-
China’s household burden lies well below the pleted by natural processes) is still relatively low
level in the poorest countries, but has some dis- by comparison to that of rich countries (e.g.,
tance further to improve. Ref. 10). Using Natural Debt as the indicator
Community environmental risks quantified of the portion of the global burden of disease
here include urban ambient air pollution, lead from climate change that is due to each coun-
pollution, occupational hazards, and road- try, Figure 5 shows the distribution of global
traffic accidents. Although not without vari- risk imposed and global risk experienced, the
38 Annals of the New York Academy of Sciences

Conclusion

Although there is much left to learn and

develop, there is a growing set of interna-
tional assessments, exemplified by the Global
Burden of Disease and Comparative Risk As-
sessment efforts, that provide a common basis
for comparison across countries, risk factors,
potential interventions, and so forth, based
on common methods agreed by international
Figure 5. Distribution of disability-adjusted life
years (DALYs) lost from global climate change by per
groups of experts managed in a way designed
capita income. The indicates the loss by those regions to be consistent, coherent, peer-reviewed, and
who receive the burden, that is, experienced burden. nonpolitical. Other such efforts are the In-
The upper curve indicates the distribution according tergovernmental Panel on Climate Change,
to how much the regions have contributed to climate and the Millennium Development Goals, the
change, that is, the imposed burden.9
Commission on Macroeconomics and Health,
and the Millennium Ecosystem Assessment. Al-
though we as individual specialists may differ
in details about the methods proposed by these
latter being the same as shown in Figure 4. groups, there is value in starting from the ap-
Note that rich countries impose some 500 times proaches established by them as internationally
more risk than they experience and poor coun- agreed assessment baselines. Elaboration or lo-
tries experience some 15 times more risk than cal adjustment can then be made, but at least
they impose, a very large degree of global in- there is a common basis on which to start, re-
equity.9 Interestingly, however, China, although ducing the considerable amount of confusion
by no means rich, has already passed the point that can result among policymakers and the
at which it has imposed more risk than it re- public by different results from different meth-
ceives, with a ratio of about 3.4 between the ods. The results discussed in this chapter are
two. By this measure, China might wish to con- examples.
sider shouldering some of the cost of global As a middle-income nation, China still has
greenhouse gas mitigation, even though not a significant national environmental burden of
have nearly the accountability of rich countries ill-health due to rural household poverty. Sev-
yet. eral hundred thousand premature deaths still
Of course, there are other environmental occur annually. Community-level burden from
risks that have not yet been quantified. Mostly, occupational and environmental risks have now
these are due to lack of good exposure data exceeded these, however, and are growing
across wide populations rather than a paucity faster. It thus has the classic management chal-
of epidemiological risk information. Among lenge posed by the environmental risk transi-
these, for example is that from environmental tion, that is, how to maintain efforts to reduce
tobacco smoke, which seems to be significant in traditional household environmental risks as
China.11 Efforts to quantify additional factors fast as possible, while also blunting the rise and
contributing to the burden from other environ- soon bringing under control community-level
mental health risks, such as those from toxic risks.14 All of this, of course, must now be done
chemical releases, would improve the evidence in the context of dealing with global impact of
base for regulatory policy in the country. climate change.
Smith: China’s Burden of Environmental Disease 39

Acknowledgments 7. Zhang, J. & K.R. Smith. 2007. Household air pol-

lution from coal and biomass fuels in China: mea-
I appreciate the comments and assistance of surements, health impacts, and interventions. Environ.
Dr. Duan Xiaoli and John Minardi. Health Perspect. 115: 848–855.
8. Smith, K.R. & M. Ezzati. 2005. How environmental
health risks change with development: the epidemi-
Conflicts of Interest ologic and environmental risk transitions revisited.
Annu. Rev. Environ. Resour. 30: 291–333.
The author declares no conflicts of interest. 9. Patz, J.A. et al. 2007. Climate change and global
health: quantifying a growing ethical crisis. EcoHealth
4: 397–404.
References 10. Smith, K.R. 1996. The natural debt: north and
south. In Climate Change: Developing Southern Hemisphere
1. Smith, K.R., C.F. Corvalan & T. Kjellstrom. 1999. Perspectives, Ch. 16. A.H.-S.T. Giambelluca, Ed.: 423–
How much global ill health is attributable to environ- 448. Chichester, England: John Wiley & Sons.
mental factors? Epidemiology 10: 573–584. 11. Gan, Q. et al. 2007. Disease burden of adult lung
2. Ezzati, M. et al., Eds. 2004. Comparative Quantification cancer and ischemic heart disease from passive to-
of Health Risks: Global and Regional Burden of Disease due bacco smoking in China. Tob. Control 16: 417–
to Selected Major Risk Factors. Geneva: World Health 422.
Organization. 12. Lopez, A.D. et al., Eds. 2006. Global Burden of Disease
3. Hofstetter, P. & J.K. Hammitt. 2002. Selecting hu- and Risk Factors. Disease Control Priorities Project.
man health metrics for environmental decision- New York: Oxford University Press and the World
support tools. Risk Anal. 222: 965–983. Bank, P. 475.
4. Hu, T.W., Ed. 2008. Tobacco Control Policy Analysis in 13. Smith, K.R., J. Rogers & S.C. Cowlin. 2005. House-
China. Shanghai: World Scientific Press. hold Fuels and Ill-Health in Developing Countries: What
5. United Nations Development Programme. 2005. Hu- improvements can be brought by LP Gas? World LP Gas
man Development Report. New York: UNDP. Association and Intermediate Technology Group.
6. Strait, K.O. et al. 2006. Carcinogenicity of household Paris, p. 59.
solid-fuel use and high-temperature frying. Lancet- 14. Smith, K.R. 1990. The risk transition. International
Oncology 7: 977–978. Environmental Affairs 2(3): 227–251.