Int. J. Cancer: 15, 954-961 (1975)

LUNG CANCER AND SMOKING IN DANISH WOMEN

Ole M. JENSEN The Itpernational Agency for Research on Cancer, Unit of Epidemiology and Biostatistics, 150, cours Albert Thomas, 69008-Lyons, France

In most developed countries, including Denmark, cancer of the lung is the most frequent malignant disease among men, whereas the problem is less among women. Examination of Danish mortality data for the period 1931 to 1972 reveals a 5-fold increase in female rates as opposed to a 16-fold increase in male rates. Since about 1960 female rates have, however, increased faster than male rates with a consequent decline in the male-female ratio. It is demonstrated that more recent female birth-cohorts have higher mortality rates than the older ones. The increasing mortality from lung cancer among more recent female cohorts is shown to parallel increasing proportions of sniokers in these cohorts. Further data are presented to indicate that not only is smoking more widespread among young than among old women, but the proportion o f cigarette smokers among all smokers falls from 98 in the youngest to 33 in the oldest agegroup. No causal relationship between smoking and lung cancer can be claimed from the evidence presented. The present findings are, however, what would be expected if cigarette-smoking were an aetiological factor in female lung cancer.

During this century mortality from cancer of thelung, ICD 162-163 (WHO, 1967), has increased rapidly in Denmark as in other developed countries, and among Danish men lung cancer is now the most frequent malignant disease. The problem is less important, but is increasing, for Danish women, where lung cancer ranks sixth among deaths caused by cancer from 1966 to 1970 (National Health Service of Denmark, 1963-1974). I n Europe the highest morbidity and mortality rates for both sexes are found in the British Isles (Doll etal., 1970; WHO, 1974). On the European continent Danish women exhibit the highest mortality rate, having a level about 2-3 times higher than that observed in other Scandinavian countries (Fig. I ) . Cigarette-smok i ng is a well-esta blished aet iological factor in lung cancer. The apparent inconsistency between amount smoked by women and level of lung cancer as compared to men has, Received: February 7, 1975.

954

however, cast doubt on the role of cigarettes in female lung cancer. As cigarette-smoking in Denmark is believed to have spread among w3men particularly in the years immediately after the Second World War (Clemmesen, 1953) and reckoning an induction period of about 20 years between onset of smoking and the development of lung cancer (Clemmesen and Nielsen, 1955 and 1973), it is pertinent to review the time trends in lung cancer in Denmark from 1931 to 1972 and the relationship between lung cancer mortality and cigarette-smoking in Danish women. MATERIAL AND METHODS

Mortality data

For a rapidly fatal disease such as lung cancer, mortality is an adequate measure of morbidity (Larsson, 1971a). The present review has therefore been based entirely on the mortality data available in Denmark since 1931. The data from

SMOKING AND FEMALE LUNG CANCER

S c o t land England & Wales Iialand

Dsnnark lcalanl Swadan Caimanr F.R.

Bsliiun [inland Nalhsrlands Swi tiailand Norway

FIGURE 1

Mortality from lung cancer in selected European countries, 1970-1972, ranked according to the female mortality rates. Age-standardized to world standard population.

1931 t o 1968 (Clemmesen, 1965; National Health Service of Denmark, 1963-1974), have been coded according to the 7th Revision of the ICD (WHO, 1957) whereas the 8th Revision of the ICD (WHO, 1967) has been used for the years 1969-1972 (National Health Service of Denmark, 1963-1974). In contrast to the United States (Percy et al., 1974), the change from the 7th to the 8th Revision of the ICD in 1969 has not influenced the time trends observed in Denmark. The probable explanation is that code number 163 in the 7th Revision of the ICD (Lung, unspecified as to primary or secondary) has never been adopted in Denmark and that a smooth decline in the number of cases described as secondary in the thoracic organs has taken place since 1960. Thus no change in the practices of coding lung tumours has been introduced with the 8th Revision of the ICD. To avoid the bias introduced by the changing age-structure of the population with time all rates have been age-adjusted by direct standardization to a world population (Doll et al., 1970). Smoking among Danish women

As annual routine statistics on the consumption of cigarettes by sex or age are not available for Denmark, information on smokingamong women has been adapted from publications on the Danish Morbidity Survey 1952-54 (Hamtoft and Lindhardt, 1957a, b ; Nlzrrgaard, 1957). Questions

on smoking habits were included during two periods of the Morbidity Survey in 1952-1953 and 1954 covering 11,500 (1952-1953) and 5,500 (1954) persons of either sex. Using a multi-stage sampling technique the sample for the morbidity survey was drawn from the adult Danish general population. The first stage was a stratified random sample with the probability proportional to size of well-defined geographical areas (communities, towns), the second stage a random-start systematic sample of all persons 15 years and above. The resulting 3 % sample was found to be representative of the general population (Lindhardt, 1960). TABLE I AGE-STANDARDIZED ' MORTALITY RATES A N D SEX-RATIO FOR CANCER OF THE LUNG IN DENMARK, 1931-1972

Year

Males

Females

Male/female ratio

1931-1935 1936-1940 1941-1945 1946-1950 1951-1955 1956-1960 1961-1965 1966-1970 1971-1972

2.96 5.02 8.40 12.70 18.58 26.08 34.58 40.30 47.19

1.83 2.06 3.17 3.08 4.20 4.78 6.12 7.83 9.85

1.62 2.44 2.65 4.12 4.42 5.46 5.65 5.15 4.79

1

World standard population (Doll et al., 1970).

955

JENSEN

parallel lines representing the birth cohorts with increased mortality rates for more recent cohorts.

RESULTS

Mortality trends

Age-standardized mortality rates and sex-ratios since 1931 are shown in Table I and Figure 2. During the years 1931-1960, mortality among men rose much faster than among women, for whom the rates did not start to go up till around 1940. From around 1960 female mortality has continued to rise at the same or perhaps an even

Smoking trends among Danish women

Using published material from the Danish Morbidity Survey on the age when smokers and ex-smokers started smoking, classified by birthcohorts, it has been possible to calculate the proportion of persons in the different cohorts

-

looM o r t o l i t y per 100.000

M/F r o t i o

Men

. - - - Wo me n Mole

.........

- female rotio

iao

- 2

- 1

I

1 .o

1931-31

1

I

1936-40 1941-45

1

1

1%-M

1951-55

I

1

I

1956-60 1961-65 1966-70

I

1971-72

Year

FIGURE

2

Mortality for men and women and male-female ratio for lung cancer, Denmark 1931-1 972. Age-standardized to world standard population.

faster rate than previously, whereas among men the rate of increase has been levelling off. As a result of these changing rates of increase the sex-ratio has declined during the 1960s. The mortality data for women plotted by birth-cohorts on a log-log scale are shown in Figure 3. Although there is some crossing-over, especially at lower ages, probably due to the small number of cases, the pattern is one of 956

who at a given age were classifiable as current smokers or ex-smokers. A smoker in the Morbidity Survey was defined as “ a person smoking regularly ”. Figure 4 shows the proportion of current female smokers and ex-smokers at different ages by birth-cohort. More recent cohorts contain at each age a larger proportion of persons who have smoked. Data are not available for cohorts

SMOKING AND FEMALE LUNG CANCER

I

M w l o l i t y per 100.000

1891

/

born prior to 1900-1904 as numbers of smokers were small and information on age when smoking was started is supposedly increasingly unreliable. Furthermore, by restricting Figure 4 to ages below 55 at the time of interview the bias due to excess mortality in smokers is minimized. The proportion of cigarette smokers among all smokers interviewed in the Morbidity Survey is also given for women in Figure 4. A decline from 98% to 33% is seen from the youngest to the oldest age-group. It is not possible to know whether this represents a cohort phenomenon. DISCUSSION

As in most countries, lung cancer in Denmark has increased rapidly during this century. Among men the increase started around 1930(Clemmesen, 1965). In both sexes part of the increase in rates

around 1940 is thought to be due to the improved diagnosis following the introduction of bronchoscopy (Clemmesen, 1965) and it thus appears that the increase among women did not start till between 1940-1945. During the past 20 years, lung cancer mortality in Denmark has more than doubled in both sexes, and now increases faster among women than among men. This has led to a decline in the male-female ratio of lung cancer mortality since around 1960 after a steady increase from 1931 (Fig. 2). The different rates of increasing mortality for the two sexes in Denmark are similar to the patterns observed in the United Kingdom and the United States (Lancet, 1971 ; Schneiderman and Levin, 1972), where the male-female ratios started to decline around 1960. In Norway and Sweden a levelling-off in the sex ratio started around 1965 due to an abrupt increase in female lung-cancer mortality, whereas in Finland no increase in lung cancer among women has been noted during the period 1950 to 1971 (Segi and Kurihara, 1960-1972; WHO, 1974). In Denmark, as elsewhere, it is difficult if not impossible to say how much of the increase in mortality is real and what proportion is due to improved diagnosis prior to death. It has been argued that up to 50% of the increase is spurious and due to increased interest in the disease among the medical profession (Rosenblatt et al., 1969). However, studies on the validity of the diagnosis (Bonser and Thomas, 1959, Heasman and Lipworth, 1966; Larsson, 19716), the parallel increase in mortality, morbidity and the number of cases diagnosed at X-ray mass screening for tuberculosis (Clemmesen, 1965), and the different behaviour of the rates in the two sexes strongly support the view that the increases observed are real. A cohort phenomenon similar to that demonstrated in Figure 3 for lung cancer mortality in Danish women has previously been described in males in a number of countries including Denmark (Clemmesen and Nielsen, 1955) and in females in the United Kingdom (Case, 1956) and the United States (Schneiderman and Levin, 1972). In contrast to what is seen among Danish male birth-cohorts shown in Figure 5 , where, in cohorts born after 1906, increased risks are only minor, or non-existant, each female birth-cohort had a greater risk than the preceding one (Fig. 3).

957

JENSEN

The presence of a cohort phenomenon may be taken as an indication of the introduction of a new aetiological factor in the changing environment, and there is overwhelming evidence that cigarette smoking is this new factor in lung cancer.

with older cohorts. The differences in the populations at risk i.e. the differences between the younger and older cohorts of cigarette smokers would be further widened should the age-specific differences in type of tobacco consumed (Fig. 6 ) constitute a cohort phenomenon.

/o Smokers + e x s m o k e r s

0

8o 70

t

60 50 40

30 20

10

13

16

21

26

31

36

41

46

51

56

M e a n a g e o f b i r t h cohorts

FIGURE4

Proportion of female smokers and ex-smokers among all women at different ages by year of birth, Denmark. Insert: proportion of cigarette smokers among all smokers by age (women, Denmark, 1953).

As n o annual routine statistics for Denmark are available on cigarette consumption by sex, it is impossible to compare the increasing lung cancer mortality in women to the number of cigarettes smoked by them. The availability of both mortality and smoking data by birth cohorts shows, however, that more recent female cohorts both contain an increased proportion of smokers at any given age and exhibit an increased risk of dying from lung cancer in comparison

958

For comparison with the female data the proportion of current smokers plus ex-smokers in Danish male birth cohorts, obtained in the same way as the female data, is given in Figure 6. The variation in the proportion of smokers in different cohorts at any given age is much smaller than that seen among women, but as for women smaller proportions in the old age-groups smoke cigarettes (Fig. 6 ) . As for women, also, it is impossible to know whether this is a true

SMOKING AND FEMALE LUNG CANCER

cohort phenomenon. It is, however, most likely t o be so. The small variation in the proportion of smokers by year of birth makes the smoking data for males unsuitable for a comparison with the changes in cohort mortality, as the unknown but probably increasing proportion of cigarette smokers in more recent cohorts may compensate for the sometimes lower proportion of smokers in these. In line with the observations of increases in both total cigarette consumption and male lung cancer mortality in a number of countries including Denmark (Doll, 1953; Ringertz, 1955), the present study shows that also among women in Denmark an increased risk of lung cancer is paralleled by increased cigarette consumption. No causal relationship between cigarette smoking and lung cancer can be postulated on this

500.0

lOQ0

50.0 1881

iao

5.0

4

1.0 32.5

37.3

42.3

47.5

w.5

62.3

023

(17.5

ns

77.3 m.3

AGE

FIGURE 5 Male lung-cancer mortality by birth-cohort (log-log scale) in Denmark, 1931-1970.

loo L

12

is

n

ze

31 Meon as.

30 4$ of b i r t h .ohor#*

si

Y)

61

66

FIGURE 6

Proportion of male smokers and ex-smokers among all men at different ages by year of birth, Denmark. Insert: proportion of cigarette smokers among all smokers by age (men, Denmark, 1953).

959

JENSEN

evidence alone, but overwhelming evidence as to this relationship has been established from both case-control and longitudinal studies (Doll and Hill, 1952 and 1964). The present investigation thus removes some of the doubt that cigarette smoking is an impartant aetiological factor in lung cancer in women and it is most likely, as shown by Burbank (1972), that the present sex difference in lung cancer rates may be explained simply by a dose-response effect. The rise in female lung cancer mortality in Denmark is expected to continue in the years to come due to a further increase in the proportion of female smokers in the population. Since 1964 this proportion seems to have become stabilized around 50% compared to the 7540% smokers in the male population, which proportion has remained largely unchanged during the past 20 years (Hamtoft and Lindhardt, 19576; Bonnevie, 1964; Egsmose, 1974). The Danish male-female ratio for lung cancer is expected to continue its present decline, as the

sex differences in the proportions of smokers at a given age diminish (Fig. 4 and 6). I t has been suggested (Wynder, 1972) that the low sex ratios prevailing before the dramatic increase in lung cancer started might again be reached although based on much higher rates than previously. Whether the sex-ratio of 1.6 observed in Denmark from 1931-1935 will again be reached is unknown and would be determined by the amount smoked, sex-specific smoking habits and exposure to other environmental factors. ACKNOWLEDGEMENTS

The author wants to thank Dr. C. S. Muir for critical comments in the preparation of the manuscript. Miss M. J. Picard helped with the typing and Mr. J. Lamy drew the figures. The work reported in this paper was undertaken during the tenure of a Research Training Fellowship awarded by the International Agency for Research on Cancer.

CANCER PULMONAIRE ET TABAGISME CHEZ LES DANOISES Dans la plupart des pays dtveloppds, et en particulier au Danemark, le cancer pulmonaire est le nkoplasrne le plus friqquent chez I’homme, alors que le problPme est moins grave chez la femme. L’examen des taux de mortalit6 enregistrts au Danemark entre 1931 et 1972 rPv2le un accroissement par un facteur 5 chez les femmes et 16 chez les hommes. Cependant, depuis 1960, le taux a augmentt plus rapidement chez les femmes que chez les honmes, d’ou une diminution du rapport entre les deux taux. I1 a ttt dtmontrt que dans les cohortes f h i n i n e s jeunes, le taux de mortalitt est suptrieur tr celui que I’on enregistre dans les cohortes plus dgtes. L’augmentation de la mortalitt par cancer pulmonaire dans les cohortes feminines jeunes est parallPle a l’accroissement du pourcentage de femmes qui fument dans ces mimes cohortes. L’auteur prtsente d’autres donntes qui indiquent non seulement que I’usage du tabac est plus rtpandu chez les jemmes jeunes que chez les femmes dge‘es, mais aussi que la proportion de fumeuses de cigarettes par rapport a I’ensemble des fumeuses passe de 98 % dans les groupes les plus jeunes a 33 % dans les plus dgts. Aucune relation de cause b effet entre le tabagisme et le cancer pulmonaire ne peut &re ttablie d’aprks ces observations. Cependant, c’est aux mimes constatations que I’on aboutirait si l’usage de la cigarette Ptait un facteur Ptiologique dans le cancer pulmonaire chez la femme. REFERENCES

BONNEVIE,P., Befolkningens rygevaner. Ugeskr. f

Laeger, 126, 1433-1435(1964). BONSER,G. M., and THOMAS, G. M., An investigation of the validity of death certification of cancer of the lung in Leeds. Brit. J. Cancer, 18, 1-12 (1959). F., US lung cancer death rates begin to BURBANK, rise proportionately more rapidly for females than

960

for males: a dose-response effect? J. chron. Dis., 25, 473-479 (1972). CASE,R. A . M., Cohort analysis of mortality as a historical and narrative technique. Brit. J . prev. soc. Med., 10, 159-171 (1956). CLEMMESEN, J., Aetiological studies in malignant neoplasms, Vol. I , MunksgBrd, Copenhagen (1965).

SMOKING AND FEMALE LUNG CANCER

CLEMMESEN, J., and NIELSEN, A., The geographicaland racial distribution of cancer of the lung. Schweiz. Z. Path. Bakt., 18, 803-819 (1955). CLEMMESEN, J., and NIELSEN,A., Age-adjusted mortality rates for lung cancer plotted against national cigarette consumption. Acta path. microbiol. Scand., (A) 81, 95-96 (1973). CLEMMESEN, J., NIELSEN, A., and JENSEN, E., Mortality and incidence of cancer of the lung in Denmark and some other countries. Acta Unio int. Cancr., 9, 603-636 (1953).

DOLL, R., Bronchial carcinoma: incidence and aetiology. Lancet, ii, 585-590 (1953). DOLL,R.,and HILL,A. B., A study of the aetiology of carcinoma of the lung. Brit. med. J., 2,1271-1286 (1952).

DOLL,R., and HILL,A. B., Mortality in relation to smoking: ten years’ observations of British doctors. Brit. med. J., 1, 1399-1410, 1460-1467 (1964). DOLL,R., MUIR,C. S., and WATERHOUSE, J. A. H., Cancer incidence in five continents, Vol. 11, UICC, Geneva (1970). EGSMOSE, T., Personal communication of Observa POIIS 1968-1972 (1974). HAMTOFT, H., and LINDHARDT, M., Tobaksforbruget i Danmark. I. Tidsscrift f. danske sygehuse. 33, 149-157 (1957a).

HAMTOFT, H., and LINDHARDT, M., Tobaksforbruget i Danmark. 11. Tidsskryt f. danske sygehuse. 33, 167-172 (1957b).

HEASMAN, M. A., and LIPWORTH,L., Accuracy of certification of cause of death. General Registry Ofice Studies, No. 20, HMSO, London (1966). LANCET EDITORIAL, Mortality trends. Lancet 1, 1138 (1971).

LARSSON, S., Comparison of lung cancer morbidity and mortality in Sweden 1959-1966. Acta path. microbiol. scand. (A), 79, 524-528 (1971a). LARSSON, S., Completeness and reliability of lung cancer registration in the Swedish Cancer Registry. Acra path. microbiol. scand. (A), 79, 389-398 (19716).

LINDHARDT, M., (ed.), Sygdomsundersogelsen i Danmark 1951-1954, Munksgkrd, Copenhagen (1960). NATIONAL HEALTHSERVICE OF DENMARK, Causes of death in the Kingdom of Denmark 1961-1971, Copenhagen (1963-1974). NATIONAL HEALTHSERVICE OF DENMARK, Mortality from lung cancer, 1972. Personal communication (1974).

A., Tobaksforbruget i Danmark 111. Tidsskrqtf. danske sygehuse, 33, 167-172 (1957). PERCY,C., GARFINKEL, L., KRUEGER, D. E., and DOLMAN,A. B., Apparent changes in cancer mortality in 1968 as a result of the introduction of the Eighth Revision, International Classification of Diseases. Publ. Hlth. Rep., 89, 418-428 (1974). RINGERTZ, N., Environmental factors and smoking in the causation of cancer of the lung. Schweiz. Z. Path. Bukt., 18, 866-884 (1955). ROSENBLATT, M. B., LISA, J. R., TENG, P., and BECK,J., Validity of lung cancer mortality data. Bull. N. Y. Acad. Med., 45, 519-527 (1969). SCHNEIDERMAN, M. A., and LEVIN,D. L., Trends in lung cancer mortality, incidence, diagnosis, treatment, smoking and urbanisation. Cancer, 30, 1320-1325 (1972). SEGI,M., and KURIHARA, M., Cancer mortality for selected sites in 24 countries, No. 1-6, Department of Public Health, Tokohu University School of Medicine, Sendai, Japan (1960-1972). WORLD HEALTHORGANIZATION, Manual of the international classification of diseases, injuries and causes of death, 7th Revision, WHO, Geneva (1957). WORLD HEALTHORGANIZATION, Manual of the international statisticaI classifcation of diseases, injuries and causes of death, 8th Revision, WHO, Geneva (1967). WORLD HEALTHORGANIZATION, Mortality tables, 1956-1971, WHO, Geneva (1974). WYNDER, E. L., Etiology of lung cancer. Reflections on two decades of research. Cancer, 30, 1332-1339 (1972). NBRGAARD,

96 1