Document Sample
                                                    N T FOR QUOTATION
                                                    WITHOUT THE PERMISSION
                                                    OF THE AUTHORS

~ ~ C A s E r n D Y

Martin Rusnak
A n a t o l i Yashin
Ihge M e r i n s k a

March 1986

C o l l a b o r a t i v e Papers r e p o r t work which has not been performed solely
at t h e International Institute f o r Applied Systems Analysis and which
has received only limited review. Views o r opinions expressed herein
do not necessarily represent those of the Institute, its National Member
Organizations, o r o t h e r organizations supporting the work.

2361 Laxenburg, Austria
      The authors would like to thank Professor Plesko and his colleagues from t h e
Slovakian Cancer Registry Office in Bratislava f o r supplying the data and f o r giv-
ing t h e i r comments. Thanks also g o to Susanne Stock f o r h e r brilliant work in typ-
ing this document, and to Shaba Venkataramn f o r h e r preparation of t h e data.

    I n i t i d Lung C a n c e r P r e v a l e n c e
    P r o p o r t i o n s o f S m o k e r s i n the Population
    Risk o f Lung C a n c e r
    Assumption One
    Assumption Two
    Assumption T h r e e
    Mathematical Model D e s c r i p t i o n
    Computer Realization
                            SLOVAKUN CASE STUDY

                Martin Rusnuk*, dnutolt Yashtn**, Inge Mertnska***

     In t h e f i r s t decade of this century lung c a n c e r was an uncommon tumor. This
is ir! s h a r p contrast to t h e late nineteenseventies and early eighties (Efron 1984).

     In 1977 the World Health Organization reported t h a t in many countries death
     rates were e i t h e r stationary or declining in both m a l e s and females, f o r canc-
     ers o t h e r than lung. The USA, Australia, Austria, Canada, Japan, Mexico,
     Sweden, Switzerland, and o t h e r s were among t h e affected countries.

     In 1979 t h e American Cancer Society reported t h a t t h e overall incidence of
     c a n c e r had decreased slightly in t h e past 25 y e a r s and t h a t t h e r e w a s a n in-
     creased death rate in men, which w a s mainly t h e result of lung c a n c e r (Figure

     In 1982 the American Cancer Society reported, "Lung c a n c e r rates are indeed
      t h e monster of c a n c e r statistics, causing t h e overall c a n c e r death rate to in-
     crease over 18 y e a r s f r o m 157.0 to 169.0 p e r 100,000persons".

     Most industrialized countries have recorded similar increases of o v e r 100%
incidence in neoplasms of t h e lung between 1950 and 1964 (Liebow 1975). As a
result o intensive epidemiological r e s e a r c h c a r r i e d out in this field during t h e
last 20 years, i t i s now generally accepted t h a t c a n c e r of the lung i s a disease of
modern civilization and, in large p a r t , preventable. The incidence of lung neo-
p h m s correlates directly with population density, urbanization, industrialization,
tobacco smoking, and even with t h e registration of automobiles (Hoffman and Gilli-
a m 1954). All these facts suggest t h a t w e are facing a real epidemic of lung can-
*Martin Ruenak, Reeearch Inetitute lor Medical Blonice, Jedlova 6, 883 46 Bratielava, Caechoelo-
**Anatoll Yaehin, IIASA, A-2361 Laxenburg, Auetria
***Inge Merineka, Reeearch Inetitute lor Medlcal Bionice, Jedlova 6, 883 46 Bratielava, Czechoelo-
vaki a
o e r . The oounteraotions of health o a r e eystems are well known but w e are in-
terested in t h e future development d this proosss and how I t uould affeot t h e p
pulation in forthaoraing years. How effective oould preventive aampaigns be, as-
suming different approaohes? Where to oonoentrate preventive efforts-in                the
younger or in t h e older part of t h e population? Many soientists are lodring f o r t h e
answers to suah questions. To develop a mathematical description of processes in
t h e population suffering from the s p r e a d of lung canuer may help answer s o m e of
these questions and forecast future development. The descriptive model, being
realized on a digital oomputer, oould b e of substantial help to health care
managers, specialists in epidemiology , o t h e r physicians, and even to nonphysicians
with interests in this field.

     The etiologiaal factors in lung a a n a e r are divided into: personal a i r pollu-
tants (e.g. smoking) and nonpersonal air pollutants (e-g. atmospheric contaminants
and industrial exposure). Recent new evidence suggests some dher personal and
llonpersonal hazards f o r mankind.
     Tobaoco smoking i s enoountered as t h e most uonunon etiologiual f a c t o r in
bronohogenic oarcinoma. The suggestion t h a t smoking, and in partioular oigarette
smoking, znay b e important in t h e production of lung cancer has been made by many
writers on the subject, even though well-controlled and large-saale oliniual studies
are lacking. Adler (1912) was one of t h e f i r s t to think that tobaoco might play
stme role in this respect. Miiller (1939), from a careful but limited clinical shtist-
iual study, offered good evidence t h a t heavy smoking is a n important etiological
factor. In 1941 Ochsner and De Bakey (1941) d e d attention to t h e similarity of
t h e curve of increased sales of oigarettes to t h e g r e a t e r prevalence of primary
c a n c e r of the lung.   They emphasized t h e possible etiological relationship of
cigarette smoking to this condition. Based on a study of 684 cmses of proved lung
a a n o e r cases wing special interviews (634 personal interviews, and f o r 33 cases
t h e information was obtained by mailing a questionnaire), Wynder and Graham
(1950) ooncluded t h a t excessive smoking, and in particular oigarette moking, o v e r
a long period i s at least one important faotor in t h e striking increase of bronoho-
genic carcinoma.
     Reaently, strong evidenae f o r t h e aomeation between smoking and lung aana-
er has appeared. The lung oanaer epidemiology i s under extensive etndy a l over
t h e world. Let us have a look at data f o r several countries.
     The 40-year incidence trends o bronchogenic aarcinoma in O l m s t e d County,
Minnesota, show t h a t in men lung aanaer has risen rapidly with each decade. Dur-
ing t h e l s decade this increase was due ta an increase o t h e rate among men
           at                                             f
o v e r 85, rates in men under 65 appearing to have plateaued. The incidence in wom-
en increased f o r t h e f i r s t t i m e in the decade 1863-1974. For all t h e cases together
t h e five-year survivorships were l l Z (Seidman et al. 1976). While in t h e USA both
t h e inaidence and mortality of o t h e r neoplasmas have leveled off or decreased in
t h e last deaade, t h e death rate f o r lung a a n c e r in-men has increased exponentially
and is today 18 t i m e s higher than 40 years ago (Figure 1 ) (Seidman et al. 1976).
These changes were acaompanled with changes in the smoking population. Two im-
portant phenomena are disaernible:

     The rate of self-reported smoking has been declining significantly.
     The predominance of males in t h e smoking population has been receding. This
     reflects t h a t t h e rate of smoking among adult m a l e s has decreased persistent-
     ly and significantly since 1964, while t h e rate of smoking among women actual-
     ly rose through muah of t h e lseOs, falling slowly in t h e 1970s (Warner 1983).

     Mortalfty f o r lung cancer in England, 1968-1980. expressed as standardized
mortality ratios, is stable in males, while in females i t i s rising (Figure 2) (Prey et
al. 1984). There seems ta b e a n overall decline in t h e number of cigarette smok-
ers. Between 1972 and 1980 t h e proportion of smokers in all groups fell, but espe-
cially among professionals. The average weekly cigarette consumption in smokers
in 1980 was 124 (18 per day) f o r men and 102 (15 p e r day) f o r women. The de-
a r e a s e of smoking was remarkable mainly in certain social groups, notably physi-
cians. The decrease was acaompanied with a decline in lung aancer death among
medical doctors, aontrasting with a significant increase in t h e overall population
(Table 1 ) (Prey et al. 19434).
     Also. in Japan t h e p a t t e r n of lung cancer has been changing rapidly. Until
several years ago, t h e number of deaths in Japan from pulmonary tuberculosis was
f a r higher than t h a t from lung cancer. As recorded in 1947, t h e death toll due ta
pulmonary tuberculosis was 121,912, 159 times t h a t f o r cancer cases (768). By
1972 this figure had decreased to 11,983 deaths from tuberaulosis, but t h e number
of lung canaer deaths approached 12,290 (Hirayama 1976). The data f o r 1977 were
Table 1 Trend in lung aancer mortality of English dootors, 1953-1985.

             Mortality/Smoking habits                                     Trend
             Lung c a n c e r death in doctors                       2SZ decrease
             Lung a a n c e r death in t h e general population      26% inarease
             Ex-smokers in doctors                                   12Z increase
             Filter cigarettes in England                                1-6Z

SOURCE:   Fray et a . ( 8 1 .
                   l 18)

8,803 f o r tuberculosis, aompared with 17,235f o r lung aancer. If t h a t pace aontin-
ues, the death rate f o r lung c a n a e r is expeated to equal t h a t f o r stomach aancer
shortly (Hirayama 1 7 ) Per capita increase in cigarette consumption in recent
y e a r s in Japan c o r r e l a t e s with recent increases in lung c a n c e r morbidity and mor-
tality rates (Hirayama 1 7 )
     A u s W a n mortality statistics show that, in 1977, lung c a n c e r was t h e m o s t
commonly reported cause of death from cancer in men. Looking at standard mor-
tality ratios plotted o v e r t i m e (Figure 3) one aan see an accelerating rate of mor-
tality from lung a a n c e r in t h e aase of women. The situation f o r men is f a r more op-
timistic, with a definite slowing down in the rate at which lung c a n c e r mortality is
inareasing (Rohan and Christie 1 8 )
     Data from Czechaslovakia a p p e a r to have similar features to thoee from o t h e r
developed aountries. Based on routine uancer statistics, t h e incidence of lung
c a n a e r in men has generally increased, with a n average yearly increase of 1 2 . A
similar trend is found in male mortality, but with a smaller yearly increment. Fe-
male incidence and mortality shows lower values, with a n average yearly increase
of incidence of 0.7% P l e s k o et al. 1 8 ) The dependence between smoking and
lung aancer incidence has also been fully proved in Czechoslovakia (Trefny 1978;
Kubik 1 8 )
     The general incidenae of c a n a e r in India is lower than in European countries
or the USA. Canaers of t h e upper alimentary and respiratory tracts (oral aavity,
pharynx, Larynx,oesophagus, and lung) account f o r m o r e than half of t h e aancers
in men and about a q u a r t e r in women (Steinfeld 1 8 ) Smoking habits are quite
different in India, smoking being synergistic with tobacco chewing. Cigarette
smoking is of comparatively r e c e n t origin and its effect on lung a a n c e r prevalence
is rising.
           (a) Falling mortality

                                                        'Cancers - Stomach
                                                                 -  C e ~ x
                             head dlsease
                  'Peptic ulcers (males)
              1   'Ischaernic heart disease (females)

           @) No change
                  'Cancer of lung (males)
                  'Diabetes (males)

                  'Ischaemic hearl d~sease

            (c) R~sing

                  'Cancer of lung (females)
                  'Cancer breast

                                                        'Peptic ulcers (females)
                                                        'Chronic renal farlure

P g r 2.
 L ue       h
           T e mortality features of different diseases in England. [Source: Prey
           et al. (1984).]

Figure 3.   Standardized mortality ratios f o r lung cancer In Australla, 1950-1977.
            [Source: Rohan and Christie ( 9 0 .
     The poesibility of tobaooo as a major etiolagiaal faotor in hurnm aanaer is n o w
acoepted by all reputable major medioal and soientific organizations (Steinfeld
1985). In addition to epidemiological studies, many experimental animal studies
have been undertaken, including t h a t of Auerbach et al. (1WO). who oonducted
long-term experiments on aaroinogenesis in animals. The number of research re-
p o r t s on the problem i s ourrently about 40,000 (Steinfeld 1985); i t is not possible
to mention all of these here! However, w e have described s o m e of them to illus-
trate the problem itself.
     Several findings from epidemiologioal studies were not potssible to explafn as
due to smoking alone. Stocks and Campbell (1955) found urban a n c e r mortality
rates in ad-1950 England to be twice as high as rural cancer mortality rates.
They attributed 502 of t h e lung uanoer rate in Liverpool to smoking and 402 to air
pollution. Similar findings w e r e soon reported elsewhere and an exhaustive litera-
t u r e overview is given in Greenberg (1983). A list o chemicals with proved associ-
ation to lung c a n c e r i s given in Table 2. They range f r o m the very potent radioac-
tive emitters to some with no activity (Beamio et al. 1975).

Table 2. Caroinogens associated with lung cancer.

                       X-irradiation     beuropyrem
                       uranium           iron oxide
                       oobalt            tar
                       chromium          coal distillates
                       niokel            petroleum distillates
                       asbestos          beryllium
                       molybdenum        arsenic
                       vanadium          bis(chloromethy1) e t h e r

SOURCE: Ioaohim (isre).

     The debate on air pollution and lung cancer is still ongoing. The arguments
f o r a relationship are as follows:

     Urban air contains substances with proven carcinogenic effects (cigarette
     smoke, industrial pollutants, motor vehicle exhaust gases, and construction
     Urban excess of lung cancer oannot always be attributed solely to cigarette
     smoking and occupational exposure (Greenberg 1983).
The arguments against are that cities with the w o r s t quality air d o not neoessarily
have t h e highest lung c a n a e r rates and that muah higher lung oanaer rates in men
ampared to women seem to be associated with t h e i r smoking h a b i b and oaaupa-
tions .
     There seems to be a close m r r e l a t i o n between lung oanaer risk in different
oaaupations. Data from t h e Rnnish Cancer Registry and R n n b h National Census of
Deaember 31, 1970, reveal that t h e standardized incidence ratio (SIR) was highest
f o r m a l e s in mining and quarrying (2.08). A higher than expected r a t i o was also
observed in manufacturing (1.29).         Among females the only SIR significantly dif-
f e r e n t from unity w a s found in agriculture, forestry, and fishing (0.44) (Pukkala et
al. 1983).
     The f a c t that t h e prevalence of lung diseases is high in individuals with a his-
tory of alcohol abuse has led to t h e suggestion of a correlation between alcohol in-
take and lung oancer. A study of t h e consumption of alcohol and tobacco in rela-
tion to oanaer in t h e USA found a positive, simple correlation between wine and
spirits and lung cancer, but a negative relationship between bser aonsumption and
lung cancer in women a f t e r mntrolllng f o r cigarette mnsumption (Breslow and En-
s t r o m 1974). An exhaustive discussion on the results of different studies i s given
in P o t t e r and McMiahaelDs (1984) overview. However, the situation is still not
a l e a r enough to enable quantification of the risk of lung cancer in men, due to al-
aohol consumption.

     The introduction of computers to enable our understanding of cancer epi-
demiology has led toward t h e creation of National Cancer Registries in many aoun-
hies all o v e r the world. That is why reasons f o r the prevalence of c a n c e r cases
are much better understood than those f o r the prevalence of other chronic
diseases. It reminds one of t h e history of tuberculosis between World W a r s I and I1
and immediately a f t e r World W a r 11. However, the future development of cancer
diseases in t h e population i s still being disaussed. The association of different
types of aancer with different risk factors is complex and must be solved if w e are
to forecast t h e development of c a n c e r prevalence.
     As is a l e a r from t h e previous section, t h e main risk f a c t o r in lung m c e r i s
undoubtedly smoking. Smoking can be understood as one of those diseases mused
by the individual's own actions. Studies of the future impacts of ahanges in smok-
ing and smoking habits, oould be of much help in establishing different anti-
smoking polioies. Because of the oomplex c h a r a c t e r of these ohanges and t h e i r im-
pacts, a model oould forecast what ohangea in lung aamer prevalenoe muld be an-
ticipated and how a ohange in prevalenoe will effect the health aare system and so-
ciety. The model might also be meful for international oomparisons. The educa-
tional process oould benefit in teaching postgraduate medical doctars to quantify
their knowledge, as well as to interpret static epidemiological results.

     W e made use o data from the Slovak Socialist Republic, stratified according
to age (18 age categories) and sex. Some coefficients were not available f o r Slo-
vakia; w e used data from studles in Europe and, in one oase, from the USA.

Initial        Curer Prevalence
     Data on lung cancer prevalence f o r Slovakia a r e currently under preparation
by the National Canoer Registry; w e made use of incidence data to roughly estimate
lung cancer prevalence.
     The prevalenoe has been estimated according to the following formula:

where P ( t ) stands for lung cancer prevalence at t i m e t , f o r sex t and age
category j ; Itljstands for lung cancer incidence f o r sex t and age j , and, finally,
ptSj(t) is the lung cancer mortality rate f o r sex t and age j a t t i m e t . W e did not
use a m o r e sophisticated approach because, a f t e r receiving the original pre-
valence data, w e intend ta skip the procedure of prevalence estimation. However,
if t h e r e are no available data on lung aancer prevalence, the use of model DYMOD
(Kitsul1980) or a similar one is highly recommended.
     W e had aocess to lung cancer incidenoe data from 1971 to 1983 f o r the Slovak
Sociallst Republlc according to the Slovak National Cancer Registry. The model
uses only the prevalence data from y e a r 1983 (Table 3) stratified by sex and age.
Yortality hr
     The data on lung m o e r mortality by sex and age oome from t h e Slovak Na-
tional Canoer Registry (Table 1, l!'igure 1). The data on general mortality oome
from t h e offioial demographic statistIos yearbook of 1983.

P m p o r t i o m o f Smoke-   in the Popalation

     The principal source of data on smokers in Czechoslovakia is Katriak (1983).
Unfortunately he did not olassify smokers into as many age categories as w e would
like to have, which i s why w e use his data as a basis f o r our e x p e r t estimation. In
Table 5 w e show the proportion of nonsmokers, c u r r e n t smokers, and quitters by
s e x and age. Figure 5 displays proportions smoothed by Q-spline.
     The study on sociologioal aspeots of tobaocoism (Katriak 1983) in Czechoslo-
vakia was a source f o r data on transition coefficients between nonsmokers and
smokers. Coefficients f o r transition from smokers to quitters were estimated from
the data of t h e Hammond study of ex-smokers (Hanunond and Percy 1958). The
numbers roughly correlate with findings of Russell (1976) and Olejnikov et al.
(1983). Table 6 and Figure 6 summarize these transition ooefficients.

     Because of lack of data from Czechoslovakia w e use the results of a case-
control interview study of lung c a n c e r carried out in five European oountries (Lu-
bin et al. 1984). The results of these are given in the form of relative risks associ-
ated with smoking and stopping smoking oompared with nonsmokers. The data of
lung cancer risk f o r nonsmokers were found in Enstrom (1979). Table 7 and Fig-
ure 7 display the risk of lung canoer in terms of t h e number of cases per 100,000
persons by age, sex. and smoking habits.

     The previous t w o sections indicate what t h e possible m u s e s of lung cancer
are and what data are available. Prom this information the model s t r u c t u r e was
easily developed a f t e r making some preliminary assumptions.
Table 3. The lung aanaer prevalenae data for the Slovak Socialist Republic es-
         timated for 1983.


                             Ae     Male    Female
                             0-4       0           0
                             5-9       1           0
                            10-14      2           0
                            15-19      0           0
                            20-24      1           0
                            25-29      5           1
                            30-34      2           2
                            35-39     m            3
                            40-44    Q9            1
                            45-49   259            9
                            50-54   434           21
                            55-59   628           39
                            60434   635           51
                            65439   Q64           59
                            70-74   848          105
                            75-79   831          118
                            80-84   354          121
                             85+    351          197

Table 4. Lung aancer deaths in the Slovak Scmialist Republic, 1983, by   we   and

                             Ae     Male    Female
                             0-4      0           0
                             5-9      0           0
                            10-14     0           0
                            15-19     0           0
                            20-24     2           0
                            25-29     0           0
                            30-34     2           1
                            35-39    14           2
                            40-44    36           6
                            45-49    77           8
                            50-54   165          20
                            55-59   288          21
                            60434   255          28
                            65-69   212          m
                            70-74   279          36
                            75-79   189          24
                            8044     72          14
                             85+      9           8
                 kge Di stri buti on of Lung Cancer
             C X 1E-3     Mortality Rate
             5    I
                      I , , , 1 , , I , 1                    ,   ,   ,   I

                  1    .   .   . . ; . . . . .:. .
       0          -

       F     3:       . . . . i . . . . .i . . . . .

                                                       .. i n . . . .

             2 1 . ..
               .                   .i.     0   .   .

       Y          -
             j,   : . . . . ; . . . . .:. . . . .
                  I            t   .


             O ; I I ' I I I ' ~ ' I '
                  0                    3                 6                   9   12   15   18
             0                                          age groups

Figure 4.        Lung c a n c e r mortaltty rates f o r Slovakia, 1983.

Il.rarption One
     Smoking cigarettes i s generally recognized to be t h e principal cause of lung
c a n c e r , which led us to discount other. not yet definitely proved, possible etiologi-
aal factors, m h as air poUution and alcohol. This allows us to underline t h e im-
p o r t a n t role anti-smoking actions have in diminishing t h e prevalence of this

-ption            Two
     According to t h e results of s e v e r a l epidemiology studies, stopping smoking
leads to a decrease in t h e risk of developing lung cancer, a decrease t h a t is
directly related to t h e t i m e elapsed since stopping. However, many people used to
quit smoking s e v e r a l times during t h e i r adulthood, so t h e duration of nonsmoking i s
v e r y different f o r each individual. I t is difficult to find reliable data on this pro-
a e s s , which is why w e assume that:
     Nonsmokers never             during t h e i r U f e rrpan.
     Smokers smoke during t h e i r whole life.
     Quitters stop and never start again.

Table 5. Population dlvided according to smoking habits.

                                           Smoking history
                      Never smoked         Current smokers          Quitters
              Age     Male     Female      Male       Female      Male   Female
              0-4     1         1          0            0         0       0
              5-9     1         1          0            0         0       0
             10-14    1         1          0            0         0       0
             15-19    0.49      0.62       0.31         0.18      0.20    0.20
             20-24    0.49      0.62       0.31         0.08      0.20    0.30
             25-29    0.26      0.46       0.24         0.16      0.50    0.38
             30-34    0.26      0.46       0.24         0.16      0.50    0.38
             35-39    0.26      0.46       0.24         0.16      0.50    0.38
             40-44    0.26      0.46       0.24         0.16      0.50    0.38
             45-49    0.23      0.54       0.32         0.13      0.45    0.33
             50-54    0.23      0.54       0.32         0.13      0.45    0.33
             55-59    0.23      0.54       0.32         0.13      0.45    0.33
             6044     0.23      0.54       0.32         0.13      0.45    0.33
             65-69    0.37      0.81       0.39         0.08      0.32    0.11
             70-74    0.37      0.81       0.39         0.08      0.32    0.11
             75-79    0.37      0.81       0.39         0.08      0.32    0.11
             80-84    0.37      0.81       0.39         0.08      0.32    0.11
              85+     0.37      0.81       0.39         0.08      0.32    0.11

lLmrption Three
     There are plenty of differences in smoking habits: c i g a r e t t e versus o t h e r
types of smoking; low, medium and high tar cigarettes; depth of inhalation; etc.
The assumption t h a t t h e r e i s no difference between smoking habits has been adopt-
ed in this model.
     Based on these assumptions, t h e model s t r u c t u r e (Figure 8) can be derived. I t
consists o two main blocks:

     Population forecast.
     Lung c a n c e r prevalence forecast.
                                     age g r o u p s
                           f ' r o ~ o r i on of Smoke;-s

                                  aye groups
                           Prop01-ti on   o f Quitters

Figure 5 .   Population divided according to smoking habite, curves smoothed by Q-
Table 6. Coefficient of transftfons from nonsmokers to smokers and from 'smokers t o quitters.

                                                          Ae groups
 Risk groups     Sex      0-4     5-9    10-14    15-18   2044    25-29   30-31     36-39    40-44
   Smokers      Male     0        0       0.29    0.69     0.35   0.24    0.13      0.06     0.07
               Female    0        0       0.28    0.67     0.20   0.16    0.09      0.03     0.0
   Qultters     Male     0       0        0.11    0.11     0.11   0.11    0.14      0.14     0.19
               Female    0       0        0.11    0.11    20.11   0.11    0.14      0.14     0.19

                                                          A e groups
 Rlsk groups     Sex     9       50-51   55-58   60-64    65-89   70-74   '7!j-79   8081     mi+
   Smokers      Male     0.07     0.m     0.09    0.m      0.08   0.06    0.04      0.03     0.01
               Female    0.02     0.02    0.02    0.01     0.M    0.005   0.003     0.0001   0
   Quitters     Male     0.19     0.26    0.26    0.28     0.29   0.40    0.40      0.40     0.40
               Female    0.19     0.28    0.28    0.29     0.29   0.40    0.40      0.40     0.40
                      Transi ti on Rate from
                      Nonsmokers t o Smokers

                                 a       groups

                      Transition Rate from
                      Smokers to Quitters

Figure 6.   Transition rates between smoking categories by sex and age.
Table 7. Risk of lung oanoer acoording to smoking habits (cases/100,000persons).

                                                            Age Emups
 Smoking habit    Sex      0-4     !5-Q   10-14    is19     20-24    -29        30-34    -9       4044
 Never amoked     Male     0       0         0       0        0          0        2.3      2.3      2.3
                 Female    0       0         0       0        0          0        0.5      0.5      0.5
 Smokers          Male     0       0         0       7.6      7.6        7.6      7.6      7.6      7.6
                 Female    0       0         0       0.85     0.06       0.85     0.85     0.85     0.66
 Quitters         Male     0       0         0       1.38     1.38       1.98     1.38     1.38     1.38
                 Female    0       0         0       0.3      0.3        0.3      0.3      0.3      0.3

                                                            A e groups
 smoking habit    sex     B       so61    m-m      6661     ~XWB     '10-74     n m      80-81     m+
 Never smoked     Male     3.5     3.5     32.2    32.2      85.6    85.6       89.9      89.9     89.9
                 Female    2.7     2.7     11.4    11.4      19.6    19.6       38.8      38.8     30.8
 Smokers          Male    25.5    25.5    289.5    299.5    6!
                                                             i       658        899      899      899
                 Female    9.15    9.45    51.3     51.3    170.5    170.5      337.6    337.6    337.6
 Quitters         Male     5.25    5.25     66.9    86.9    203.4    203.4      278.7    278.7    278.7
                 Female    4.6     4.6      20.5    20.5     56.8     36.8      112.5    ll2.6    112.6
            Risk of Lung Cancer O n s e t f o r Nonsmokers

                     Risk o f L u n g Cancer O n s e t for
                 (     1~-5)  Curyen-t Smokers
                     1   1   1   1   ,   ,   (   1   1   ,   1   1   1   1   1   ,   (   1   )

                                                         age g r o u p s

                                                         asre moups                              +
Figure 7.    Risk o f lung cancer o n s e t by smoking habits, sex, and age.
                        0  START

                     QINPUT OF DATA

                     NUMBER OF N
                     FORECASTED YEARS

                    DEFINITION OF


            I     RISK GROUPS AND
                  PREVALENCE FORECAST        I I

                                         OF RESULTS

                    RESULTS I N TABLES

Figure 6.   The lung cancer model realization in computer program.
The f i r s t bloak i s t h e same as t h e one w e used in o u r previous study on smoking and
chronic obstrvctive pulmonary disease (COPD) (Rusnak st al. 1885). The second
aomputes t h e amount of people In different mnoking statuses, by s e x and age. A s
f a r as w e know how m y people in eaoh risk a t e g o r y are becoming ill with lung
eanoer, t h e prevalence ean be estimated. By ohanging t h e model's parameters, t h e
user oan test different hypotheses about smoking distribution o v e r population, or
about t h e effects of anti-smoking campaigns on lung m o e r prevalence.

Y athaatical Y ode1 Demmiption
       The population forecast i s based on a simplified idea of population dynamics.
Denoting pi, (t ) f o r population at t i m e t , sex i , and age group j , t h e equation used
is :

        if j   =1 , a = b      (b   - number of births)

The death rate i s t h e total death rate f o r t h e population, denoted p, t h e mortality
rate f o r nonlung oancer cases i s c , and lung c a n o e r mortality rate i s       &   One can

In o r d e r to describe t h e dynamics of t h e populations at different rfsk, w e have to
introduce some m o r e variables:

       nt , (t ) = number of nonsmokers with sex i , a g e j , at t i m e t .
       st,j(t ) = number of smokers with sex i , a g e j , at t i m e t .
       qtj(t)   = number of quitters with sex i , a g e j , at time t .

Coefficient p t j describes t h e risk of lung c a n c e r onset f o r nonsmokers with sex 1 ,
age j (per 100.000 persons). Coefficients p f j and           p
                                                              :    stand f o r t h e s a m e type of
risk, but for smokers and quitters, respectively. Transitions between groups are
marked by t h e coefficients     T:
     rimj tmwition from nonsmokers to smokers
     r f m j transition imn smokers to quitters.

One oan derive the following equations for the foreaast of lung cancer development
in nonsmokers, smokers, and quitters:

      n i , j ( f ) =nfDj(t-1) + a -($I    + c + p t j +Pf,j)ni,j(t-l)

                                               i   = 1,2 ,   j   = 1,...,18

      if j = 1 , a = b        (b   - number of births)

The dynamics of lung cancer prevalence, I f ,j ( t ), can be expressed as:
      The model was implemented on an IBM PC micro-computer and the program was
written in IBM Compiler Basic. The user oan define t h e range of forecast; maximum
is 40 years.
      The model allows t h e user to test several hypotheses, wlth t h e quantification
of each hypothesis being expressed by t h e coefficient change (in peroent). Com-
muniuation between t h e user and t h e model is done interactively, with only graphio
routines invoked separately using the STATGRAPH program s y s t e m on a n IBM PC XT
or NEWPLOT graphic system on the VAX.
      The model oommunicates with o t h e r programs and systems. The results of
model runs can be stored in database system dBase I11 1n t h e database LCA. I t
allows t h e user to r e t r i e v e necessary information in t h e interactive way.   A
detailed description of the LCA database exploitation is to be found in Joestl-
Segalla et al. (1986). Data f o r graphic routines as STATGRAPH and NEWPLOT may
be retrieved from database LCA. Piles used f o r input into t h e model and f o r com-
munication purposes are listed in Table 8.

      As w a s already stated, the main t a r g e t of this model is to project future
development in lung cancer prevalence. For this purpose t h e projection of risk
f a c t o r s was also done.
      Basic projection on data from Slovakia with no scenario are shown in R g u r e 9.
The steady increase in the number of cases is more significant in t h e female popu-
lation, as compared to t h e males. The growth of t h e number of female cases wl
Table 8. Files ueed by the lung aanaer model.

File name        Structure       Description           File name      Structure       Description
POPFORC.SSR      SEX    1,2      population forecast           .TAB   table           results of model run
                 AGE    1,18     for Slovakia                                         in a form of a table
                 TIME   1,20

CADEATHR.SSR     SEX    1,2      lung cancer death     .BSF
                                                       I see dBase I11                file with model's
                 AGE    1,18     rates and general          database LCA              results to be entered
                                 mortality rates                                      into database LCA

SICKRISK.SSR     TYP    1,2,3    TYP 1 nonsmokers      I   I   .KOM   see Keyfitz's   communication file for
                 SEX     1,2     TYP 2 smokers                        model           Keyfitz's model
                 AGE     1,18    TYP 3 quitters
                                 risk of lung cancer   .SUM
                                                       I sums                         sums of the model's
                                 onset according to                                   results
                                 smoking habits

        .        TYP    1,2      transitions                   .LCS                   output file from data-
                 SEX    1,2      TYP 1 nonsmokers-                                    base LCA
                 AGE    1,18           smokers
                                 TYP 2 smokers-

       .         TYP    1,2      proportions of        U       sNWP                   file with data for
                 SEX    1,2      TYP 1 smokers                                        NEWP LOT
                 AGE    1,18     TYP 2 quitters

LUNGCASE.SSR     SEX    1,2      number of lung
                 AGE    1,12     cancer cases
               L UNG     CRNCER        PREVALENCE              FORECRST


                  eoa                                                          I

         S                                                                     I
'       C                                                                       1
!        =-;
         L L                                                                   iI
'       r ,
t        CL
!                                                                              I
!                                               "                              I

/              L U N G   CRNCER        PREVRLENCE
I                                                                              I

Figure 9.          Lung cancer prevalence projection under basic aonditions.
inarease up to four times in s o m e age oatsgories, u o m p w d with two time8 in the
sanre aategory f o r nraTes. While some age aatsgorlee -9)          show only a moderate
t r e n d of inorease in the male population, the same female population m t i n u e e to
grow at a mwh steeper rate. Three male age groups (50-54, 15-19, and W )
have t h e opposite hndency, while t h e female population keeps on growing. The
reasons f o r these patterns are more understable if the profeation of risk f a c t o r s
is taken into account. Projections of populations in risk are summarized in Plqure
10. The steady increase in t h e numbers of smokers is shifted towards the older age
categories, which indicates t h e reoent t r e n d in many countries that more and more
women begin smoking and they retain this habit during a significant part of their
     The model was run on several scenarios. W e tried to highlight t h e impacts of
preventive measurea in t e r m s of diminution of population in risk by changing t m n -
sition coefficients. For example, a change in t h e transition between nonsmokers to
smokers would imply a change in the smoking population. A similar effect is as-
sumed a f t e r changing the transition from smokers to quitters. The o t h e r type of
scenario concerns t h e change in rlsk of disease onset. I t is not likely that d i r e a t
influence could be done to the lung cancer onset. A majority of t h e screening ac-
tivities are usually uneffective and too late to prevent the disease onset and as a
l a w expensive and invasive. Nevertheless, introduction of lower tar c i g a r e t t s s or
more effective filters could be concerned as effective preventive m e a s u r e s as well
as diminution of t h e air pollution in towns or in working places.
     W e have tested several different scenarios. The list of scenarios aan be found
in Table 9. The original assumption of model sensitivity to minor changes was not
proven, as can be seen in Figure 1 . The f i r s t t w o scenarios displayed (7 and 8) d o
not show any significant decrease in the three-dimensional graph. However, in the
tabular form of results, the change is discernible. The overall tendency shows
that a 20-year forecast span is too s h o r t f o r a m o r e significant improvement. Can
w e call the prevention of up to 100 lung aancer cases insignifiaant? Scenario 9
shows retardation in steady increase in lung c a n c e r prevalence, especially in t h e
male population. Several o t h e r scenarios t r y to highlight the effects of more sig-
nificant changes. Scenario 13 (F'igure 12) assumes a reduction in t h e transition
from nonsmokers to smokers to zero, introduced in 1986. While the general pat-
tern from the f i r s t sight is t h e same compared to the basic projection, a closer
view will reveal t h e reduction of cases in all age categories. Even this reduction
w a s not powerful enough to completely stop the steady increase. A similar reduc-
       Forecast of Nonsrnokors                                 Forecast of Honsrnokers
         f emal e , scenari o 0                                   male, 0

   Forecast of'                      Foreca~tof C i yrtrette Smokers
          femal e , scenari 00                                  amle. scenario 0

Forecast of Cigarette Smoking Q u i t t e r s
           f emal e, scenari o 0

        Flgure 10. Projections o risk group development under basic conditions.
Table 9. Saensrfo desaription.

Scenario                           Ywrof      Age            Percentage
 munber       Cbangod factor       change   cmtegory   Sex    of change             Comment
    1      trawltlon                                                      &crew4 in ntmber of
           aonarookera-aaokers      1966     15-29     WF       M)        people nho d a r t smoking
    2      trawltion
           emkersquitters           1966     40-69     WF       130       increase in quitters
    3      trassltions
           no~mokera-wokers         1966     15-29     WF       M)        comMnatlon of scenarios
           mmkere-qaltters                   40-59              130       land2
    4      risk i n nonsmokers      1986     40-69     WF       75        reduction i n risk
    5      risk i n smokere         1906     40-59     WF        75       risk reduction i n mmkere
    6      risk i n quitters        1906     40-!W     WF        75       risk reduction Ln quttters
    7      trawltlon
           aonanokera-mnokers       1966     15-29     WF       M)        combination of scenarios
           risk i n smokers                  40-59              130       1and 5

    8      traasttlon
           mmkers-quitters          1966     40-59     WF       130       combination of scenarios
           riek i n qoitters                 40-59              75        2 and 6

   9       trawltion
           wlallettiquitters        1968     15-59     WF        10       docrewe i n smokers
   10      risk i n mnokers         1966     20-59     WF       SO        rirrk redoction i n &ere
   11      trawitlon
           wlallerbqut t t e r e    1906     20-59     WF       200       incream i n quitters
  12,13    transition
           aommmkemwtmokers         1966      All      WF        0        decrawe i n mokere
   14      transition
           nonanokera-wokers        1986      All      WF       SO        decrease i n d   e   n
   15      trawltion
           nonmmokera-smokers       1986      All      WF        !
                                                                 2J       4ec1-0~4 mookern
   16      risk i n mnokere                                               ridt reduction i n
           and quitters             1968      All      WF        !0
                                                                 5        aaakere and quitters
   17      ridt i n smokers                                               ridt rdoction i n
           and quitters             1986      Nl       WF        25       amkern and quitters
   18      ridt i n aaokers                                               risk reduction i n
           and qui t t e r s        1986      All      M/F       0        mmkers and quitters
     LUNG      CRNCER       PREVR ENCE    FORECRST                      LUNG                 CANCER        PREVRLCNCE      FORECRST

       -                     M A L ~


       ....                                                                         .
       ....                                                                         .
       ....                                                       Y
                                                                                   ".                                                          1
g        .                                                        m

                             - * " -
                                                                  b     .          .

             mGE   GROUPS        -c€--=

     L U W G   CRNCER       PREVALENCE    FORECRST                      LUNG                 CPNCER        PREVALENCE      FORECRS-
                             MALE                                                                          FEMALE



       ..                                                                           .
       .-                                                         W
                                                                  w                ".
L        .                                                                          r

                                                                                       mGE   C R O U P ~
                                                                                                            -   % -
                                                                                                                 -c emm5

                                              OCCYIInIO   a
                                                                        ,               ,.
                                                                                        --                                     OCCUCI*II   a

                                                              I         LUNG

                                                                                             CRNCER        PREVRLENCE


                                                                  L                .

                   riawe 11. Cornparison o several scenarios.
        LUNG      CRNCER            PREVRLENCE            FORECAST

              GSE    GROUPS                   -(E--~

        LUNG        CANCER          PREVALENCE             FORECRST

Plgure 12. Results of scenario 13   - no transition from nonsmokers to smokers i s
Uon to 502 is shown in R g u r e 13 and t h e dearease is not as outstanding as aom-
pared to soenario 15 in R g u r e 14 (reduction to 252). SaeMI-los 16, 17, and 18 an-
Uaipate ahanges in the rlsk d the onset of lung aanaer in smokers and quitters.
The introduction of a 302 ahange in risk will =use a dramatic ahange in lung aanc-
er prevalenae trends (Figure 15), as compared to saenario 14 (Figure 1 3 ) or basic
projection (P'igure      9).   Even more optilnlstia projections were achieved by
scenarios 17 and 18 (Figures 1 6 and 17).

      The results of t h e lung cancer model c d i r m e d the general notion on complex-
ity of t h e relationship between smoking and lung cancer. The quantifiaation of this
relationship i still a task which remains to be solved. A continuous increase in
lung aancer prevalence (as aan be seen from the projections made upon t h e
c u r r e n t situation) is similar to the one done by r e s e a r c h e r s from t h e Finnish
Cancer Registry (Teppo et al. 1985). A simflar trend can be imagined by looking at
the t i m e s e r i e s data on lung aancer all o v e r t h e world. In o r d e r to change the
a u r r e n t situation and praspective development, one lnlght ask several questions of
t h e type: 'What will happen when         ...?".   The assumed change i s in t h e form of a n
effective, antismoking aampaign or in t h e more effwtive lung c a n c e r prevention.
even treatment. I t i s not rational to e x p e c t dmmatic changes in smoking behavior
of people in developed aountries. Besides public education, several therapeutic
methods are applied to cure those who wish to gtop smoking. The results from both
of these methods are still not encouraging enough. Lebeau in Lehrl et al. (1985)
has clearly demonstrated both on t h e basis of personal experience and from data
taken from literature on the subject, t h a t t h e percentage sucaess rate at t h e end
of treatment, which is always of s h o r t duration, is relatively good and is approxi-
mately B O X whatever the method used, except in the case of treatment with drugs,
this being the least effective method with a success rate of only 452. The results
of t h e f i r s t saenarios testing w e r e not too encouraging to us. But a f t e r t h e t i m e of
delay between starting to smoke and t h e onset of disease aras taken into account,
t h e situation became clear. Now w e understand what potential is hidden above t h e
antismoking aampaigns. The aessation of smoking would definitely lead toward
diminution of the incidenae of lung cancer, and the o t h e r ahronic diseases as well.
W e have to be patient enough and wait f o r the results. Many scientists try to
understand how the life expectancy of people can be increased. But t h e regular
a i g a r e t t e smoker sacrifiaes seven y e a r s of life f o r his habit and addiction. This

          LUNG            CANCER   PREVRLENCE           FORECAST


    W           *ma


                                                                SCENRRIO    14
          CP,         '   8-

          LUNG            CANCER   PREVRLENCE          FORECAST

Flgure 13. Results of suenario 14 - reduction in transition rates from nonsmokers
           to smokers to one-hall ofthe orlginal value for a l ages.
              LUNG                             CANCER      PREVALENCE   FORECAST

                     I   mom




                                                                            SCENARIO   15
          4    SDS                 I       .
                                           >    ,.lH

          LUNG                                 CANCER      PREVALENCE   FORECAST



     w               lee0

     e                    l

                                                                            SCENRR30   15
      r       err,             I       ,
                                       >       w,x.,xmm.

Figure 14. Results of scenario 15 - transitions from nonsmokers to smokers
           reduced to one-quarter of the origlnal value for all ages.
          LUNG    CANCER        PREVRLENCE            FORECAST



              QGE   GROUPS
                                        .                                  I


          LUNG    CRNCER       PREVRLENCE            FORECAST

Flgure 15. Results of scenario 16 - risk of lung cancer onset in smokers and
           quitters reduced to half of the original value for all ages
         LUNG                 CANCER     PREVALENCE       FORECRST



    W         a mm
    e           D

                 RGE            GROUPS

                                                                  SCCNRRIO    17


         LUNG                 CRNCER     PREVALENCE       FORECAST
                                          M A L E




    - 2

                 QGE            GROUPS

                                                                  SCCNRRIO    1 7
          m          ,   .I   .,,.,

Figure 16. Results of scenario 17       - risk of lung canoer onset in smokers and
               quitters reduced to one quarter of the original value.
       L U N G    CANCER       PREVALENCE             FORECRST
                               F E M A L E


            QGE    GROUPS

Figure 17. Results of scenario 18 - risk of lung cancer onset in smokers and
           quitters approaches zero in all ages.
~ranslat.8~ about 5.5 minutes of life f o r each olgarette smoked (Swann 1978).
     The prevalenae m u l d d l y be cmnverted into eoonomio terms as sick-leave
Qys. hmpital Qys,       or o t h e r health o r social-related expenditure. The inorease
in prevalence w l l l be tailed by the inarease in aonstmption o health anre
resouraes, whiah are wudly limited. Early identifiaation programs f w lung oanc-
er (annual chest roentgenograms and sputum cytology) effective in identifying
squamous aell oarcinoma at a t i m e when e a r l y deteation aan improve sumtval
(Plehinger 1984) are very expensive and mass screening i s not recommended as
a o s t a f f e c t i v e (Early Lung Canaer Study Group 1984). That is why sinoking cessa-
tion probably seems to be t h e only way to stop t h e increase in lung canoer in-
cidence and death. That is why Petty (1985) proclaims t h a t every physician's of-
fice o r clinic can become a smoking cessation center.
     The model itself is still under development. The authors plan to use i t f o r
international oomparisons within developed countries. A more detailed stratifica-
tion of smoking habits i desirable as well. Such refinement wlll be possible only if
a detailed study on smoking would be available. The possible merge of this model
with t h e o t h e r s on chronic diseases will be a worthwhile task f o r the future.
A d l e r . I. (1912) Prfnubry Malignant Orowths     u        the Lungs and Bronchi. N e w
         York: L o n g ~ ~ a r r Green and Co.
Annegers. J.F., D.T. Carr, L.B. Woolner, and L.T. Kurland (1878) Incidenae, Trend,
         and Outoome o Bmmhiogenici Caraimma in OLPnsted County, Minnesota, 1 3 j      9 !-
         1874. Mayo CLinCc P r o d i ngs 53:432-436.
Auerbach, O., E.L. Haaraond, D. Kinnan, et al. (1970) Effects o Cigarette Smoking
         on Dogs: II. Pulmonary Neoplasms. A r c h i w s     u      Environmental Hnalth
Beamio, J.F., A. Stein, and J.L. Andreus (1975) Changing Epidemiology of Lung
         Cancer. Inareasing Incidence in Women. Medicart Clinics       u      North America
Breslow. N.E. and J.E. Enstrom (1974) Geographic Correlation Between Cancer
         Morhlity Ratss and Alcohol-Tobacco Consumption in the United States. Jour-
         nal of National C a n m r Institute 53:631639.
Early Lung Cancer Study Group, Early Lung Cancer Detection (1984) Summary and
         Conclusions. American Review U R e s p i r a t o r y Diseases 130:565-570.
Efron, E. (1984) T%a A p d y p t i c s . Cancer and the Big Hi. New York: Simon and
         Schuster. 588 pp.
Enstrom, J.E. (1979) Rising Lung Cancer Mortality b o n g Nonsmokers. Journal of
         National Cancer Institute 62755-760.
P e y , J.. D. Brooks, and I. McColl (1984) AEBDrrta Book. MTP Press Limited. 271
Greenberg, M.R. (1983) U r k n i z a t i o n and Cancer Mortality. Oxford University
         Press. 276 pp.
Hammond, E.C. and C. Percy (1958) Ex-Smokers. N w York S a t e Joutncrl          u    Medi-
         cine pp. 2!356-2959 (Sept. 15).
Hirayama, T. (1976) Epidemiology of Lung Cancer Based on Population Studies. In
         CLinicaL Implications 4f Air FbUutwn Research, edited by A.J. Finkel and
         W .C. Duel. American Medical Association, Publishing Sciences Group, Acton,
Hirayama, T. (1977) Smoking and Cancer: A Prospective Study on Cancer Epi-
     demiology Based on Census Population in Japan. In h d f h Consequences,
     Education, Cessation A d i v i t t e s and Oovernmental Action. Smoking and
     Health II. D E No. 77-1413 (NIH), pp. 65-72. Washington, D.C.
Hirayama, T. (1979) Cancer Epidemiology in Japan. Environmental Health Per-
     spectives 32:ll-15.
Hoffman, E.F. and A.G. Gilliarn (1954) Lung Cancer Mortality. Geographic Dtstri-
     butwn in the United SYates jbr 1848-iB48. USPHS Report 69, pp. 1033-
Ioachim, H.L. (1978) Present Trends in Lung Cancer. In 7 7 u Lung, SYructure,
     finction and DCsecrss, edited by W.M. Thurlbeck and R.A. Murray. The Willi-
     ams and Wilkins Co. 322 pp.
JoestlSegaLla, M., M. Rusnak, L. Smolak, A. Yashin. and L. Zeold (1986) The Popu-
     Lution Data Base at IUS4. Forthcoming Working Paper. Laxenburg. Austria:
     International Institute f o r Applied Systems Analysis.
Katriak, M. (1983) SxAoLoqtcb A m k t y Z M a k b u u . U s t a v Z & a u r a v o t n i c ~
        Wchovy. Bratislava. 97 pp.
Kitsul, P. (1980) A Dpnamic Approach to thr Estimation # Morbidity. WP-80-71.
        Laxenburg, A w t r i a : Intsrnatlonal Institute f a r Applied Systems Analysis.
Kubik, A. (1981) ProepekUve Studien d e r Atiopathagenetisohen Faktoren des Bron-
        ohialkrebes.             Z.ftschrVY f i r Erkrankungen dar Atmungs Organa
Lehrl, S., E. Trell, B. Luban-Plozza, B. Lebeau, Th. Abelfn, T. Salvador-Llivina, and
        R. Masironi (1985) Smoking Control Programmers. European JournuL 4J
        R e s p i r a t o r y Diseases 66(139):35-38 (Supplement).
Liebow, A.A. (1975) Tbr&                   a n d CardwvascuLar 2krgm-y w i t h ReLated h f h o l -
        og y . New York : Appleton-CenturyXrafts .
Lubin, J.H., W.J. Bloh, F. Berrfno, R. -ant,             C.R. Gillfs, M. Kunze, D. Schmihl, and
        G. Visoo (1956) Modifying Risks of Developing Lung Cancer by Changing Habits
        of Cigarette Smoking. B r i t i s h Medical JournuL 288:1953-1956.
Miiller, F.H. (1939) Tabakmissbranch und Lungencartinom. Z e i t s c h r w f i r
        K r e b m r s c h u n g 49357-85.
Ochsner, A. and M. De Bakey (1941) Carcinoma of t h e Lung. Archives o J 2 k r g a r y
Olejnlkov, S.P. et al. (1983) Smoking and Some Social-Demographic Characteris-
        tics. T s r a p e v t i c e s w Archiv 55(1):5741.
Petty, T.L. (1985) Pulmonary Medicine. JournuL # the American Madfret Associa-
        t i o n 254(16):2271-2273.
Plesko, I., J. Mrnitrova, J. Somogyi, A. Sestaz, and V. Vlasak (1985) Contemporary
        Trends of t h e Incidenae of Malignant Tumors of t h e Lungs in Czechoslovakia.
        SYudia ALeumoLogica et Phtiseologica CechosLovacu 45(3):147-155.
P o t t e r , J.D. and A.J. McMichael (1984) Alaohol, Beer and Lung Cancer - A Meaning-
        ful Relationship? InternationaL JournaL o / ~ i d e m i o L o g y   13(2):240-242.
Pukkala, E., L. Teppo, T. Hakullnen and M. Rimpelii (1983) Occupation and Smoking
        as Risk Determinants o Lung Cancer. International Journal of Epidemiol-
        o g y 12(3):290-295.
Rohan, T. and D. Christie (1980) Australian Lung Cancer Mortality. MedicaL Jour-
        naL 4JAustraLia 1:489-490.
Rusnak, M., A. Yashin, and P. Kristufek (1985) 7!he F b t u r e #Lung Diseases: COPD
        Model f i r SLovakia. CP-85-49. Laxenburg, Austria: International Institute
        f o r Applfed Systems Analysis.
Russell, M.A.H. (1976) Prevention of Smoking-Related Disease. B r i t i s h Medical
        Journal 1:1430.
Seidman, H., E. Silverberg, and A.I. Holleb (1976) Cancer Statistics, 1976. A Com-
        parison of White and Black Populations. Ca-A Cancer Journal 4 J C l i n W m
Steinfeld, J.L. (1985) Smoking and Lung Cancer. A Milestone in Awareness. Jour-
        na.L 4J tha American MedicaL Association 24/31,20,253:2995-2997.
Stocks, P. and J. Campbell (1955) Lung Cancer Death Rates Among NonSmokers and
        Pipe and Cigarette Smokers. B r i t i s h M e d M Journal 2923-928.
Swann, P.G. (1978) Smoking or Health: A Report of t h e Royal College of Physi-
        cians. M e d W Bumtin 38:40-65.
Teppo, L., M. Hakamd, T., Hakulinen. E. Pukkala, and E. Saxen (1985) Phnning
    and EuuLuati ng Awmti w Ahasure~.Lyon, h a n o e : Intezmtional Agenay
    f o r Researoh on Canoer OARC). Pages 27-44.
Trefny, J. (1978) Erfaaeung dm3 Bronchialoareinolns in d e r CSSR. Z.ftsch+Vr f i r
    Erkrankungen dn Atnsungs Urgcrnu 150(2):250-254.
Vaupel, J.W., B.A. Gambfll and A.I. Yashin (1985) Contour Y a p s @PopuhtionSbr-
    jbcrs. WP-85-47. Laxenburg, A u s t r i a : International 1netitut.e f o r Applied
    Systems Analysis.
Warner, K.E. (1983) Reactions to Perceived Risk: Changes in the Behavior o           f
    Cigarette Smokers. In l h e Analysis of Actual k s u s Psrceived Risks. edited
    by Conello et al. New York: Plenum Press.
Wynder, E.L. and E.A. Graham (1950) Tobacco Smoking as a Possible Etiologic Fac-
    tor in Bronchogenic Carcinoma. JournaZ of the Americcm Medical Assodation

Shared By: