Hysterectomy and Subsequent Risk of Cancer

Transcription

1 International Journal of Epidemiology International Epidemiological Association 1997 Vol. 26, No. 3 Printed in Great Britain Hysterectomy and Subsequent Risk of Cancer RIITTA LUOTO,* ANSSI AUVINEN,**, EERO PUKKALA** AND MATTI HAKAMA**, Luoto R (National Public Health Institute, Department of Epidemiology and Health Promotion, Mannerheimintie 166, FIN Helsinki, Finland), Auvinen A, Pukkala E and Hakama M. Hysterectomy and subsequent risk of cancer. International Journal of Epidemiology 1997; 26: Background. The objective of this retrospective cohort study was to assess the effect of hysterectomy on subsequent risk of cancer among hysterectomized and a similar number of non-hysterectomized control women, registered in in the Mass Screening Registry (MSR). Methods. Cancer cases were obtained from the Finnish Cancer Registry (FCR) and standardized incidence ratio (SIR); the expected number of cases based on cancer incidence rates of the Finnish female population in , was used. Relative risk (RR) was calculated as SIR among the hysterectomized relative to non-hysterectomized women, adjusted for follow-up, education and parity. Results. The RR estimates of non-genital cancers among women with any hysterectomy were approximately 5% higher than in the non-hysterectomized cohort. Relative risks of rectal cancer (RR = 1.4, 95% confidence interval [CI] : ) and thyroid cancer (RR = 2.1, 95% CI : ) were significant and largest among women who had undergone total hysterectomy pre- or perimenopausally. Relative risk estimates of breast cancer were close to unity. Conclusions. Hysterectomy is not associated with any substantial protective or promoting effect on cancers in general. Elevated risk of papillary thyroid cancer following hysterectomy is biologically plausible, as there are reproductive and endocrinological causes of thyroid cancer. Keywords: neoplasms, cohort, hysterectomy Hysterectomy is one of the most frequent surgical procedures for women. At the national level, the highest prevalence rates of hysterectomy have been reported from the US, 550 per for all ages in There the prevalence rates increased slightly until the year 1990, and have begun to decrease recently. 2 In the US, one-third of women aged 45 years have undergone this operation, 2 whereas rates reported in Europe have been lower. In the Nordic countries, the prevalence of hysterectomy is highest in Finland (390/ women of any age) 3 and Denmark (360/ women of any age). 4 In 1989, one-fifth of Finnish women aged years reported having had a hysterectomy. 5 Earlier studies have reported that hysterectomy including bilateral oophorectomy, not hysterectomy alone, reduces risk of breast cancer 6 13 and is associated with lower risk of ovarian cancer However, in most studies conducted before the 1970s some confusion remains concerning the extent of hysterectomy, i.e. whether both ovaries were removed or only one. The incidence of breast cancer has decreased by a third subsequent to hysterectomy. 17,18 Artificial menopause has been associated with an increased risk of thyroid cancer No other studies assessing the risk of cancers other than gynaecological or overall cancer have been published to our knowledge. According to a Danish study, supravaginal amputation did not cause increased risk of overall cancer or cervical cancer. 25 The aim of this study was to assess whether hysterectomy is associated with risk of cancer other than in the reproductive organs, with special emphasis on hormone-related cancers. * National Public Health Institute, Department of Epidemiology and Health Promotion, Mannerheimintie 166, FIN Helsinki, Finland. ** Finnish Cancer Registry, Liisankatu 21 B, FIN Helsinki, Finland. Finnish Center for Radiation and Nuclear Safety, PO Box 14, FIN Helsinki, Finland. University of Tampere, Tampere School of Public Health, PO Box 607, FIN Tampere, Finland. 476 MATERIAL AND METHODS In Finland, a countrywide organized mass screening for cervical cancer has been operating since the mid1960s and it covers predominantly age groups from 35 to 50 years. 26 The method for screening has been the Papanicolau test and the re-screening interval 5 years. At the time of the screening, each woman fills in a

2 HYSTERECTOMY AND SUBSEQUENT RISK OF CANCER 477 questionnaire including, among other items, a question whether or not she has had a hysterectomy. The response options include a total hysterectomy and a partial hysterectomy, in which the cervical stump is left intact and the body of the uterus removed. Ovarian status is not assessed in the questionnaire nor the indication for hysterectomy. All women reporting a hysterectomy for the first time on the screening questionnaire in were identified from the Mass Screening Registry (MSR). As control cohorts, a similar number of non-hysterectomized women, separately for totally and partially hysterectomized, were randomly selected from the MSR with a similar distribution of birth years and municipalities of residence. Women with a gynaecological cancer (ICD- 7 codes ) prior to the reporting date of hysterectomy (85 cases among women with total hysterectomy and 28 among women with partial hysterectomy) were excluded from the analyses. Among the control cohort for total hysterectomy there were only 14 cases of gynaecological cancer prior to the beginning of followup, and no cases among the control cohort for partial hysterectomy. A computerized record-linkage with the Central Population Registry (CPR) using the personal identifier (PID) was performed to obtain follow-up for death and emigration. The coverage of follow-up was practically complete: only six women were not found in CPR. The final cohort consisted of women reporting a total hysterectomy and 6727 with a partial hysterectomy. The non-hysterectomized cohorts included and 6714 women, respectively. Follow-up for cancer incidence through the files of the population-based nationwide Finnish Cancer Registry (FCR) was done automatically with the PID as the key. The follow-up for cancer for the hysterectomized women started at the screening round when she reported a hysterectomy for the first time. The actual age at hysterectomy can be estimated to be about 2 3 years lower, between two Pap screenings. For the non-hysterectomized cohorts, the follow-up started at the same date or if she had not attended mass screening prior to that time, at the date of the first screening. The follow-up ended at emigration, at death, or on 31 December 1993, whichever was first. The number of person-years for totally hysterectomized women was and for their control cohort ; for partially hysterectomized women it was , and for their control cohort The mean length of followup was 20.5 years in both hysterectomized groups and 19.9 years among the non-hysterectomized women. The observed numbers of cancer cases and personyears at risk were calculated separately for four calendar periods ( , , and ) by 5-year age groups. The expected numbers of cases for overall cancer and for primary cancer sites were calculated by multiplying the number of person-years in each stratum by the corresponding cancer incidence in the whole Finnish female population. The cancer sites were a priori selected for the analysis and included hormonally dependent cancers and other common cancer sites to yield the whole picture of cancer among hysterectomized women in Finland. The observed and expected numbers of cases were calculated separately for subcohorts defined by the following variables: (i) age at entry to the cohort: 45, 45 54, 55 years, (ii) parity (live births): 0, 1 2, 3, (iii) education: basic, medium, high, unknown, (iv) follow-up: 0 4, 5 9, 10 14, 15 19, 20 years since the entry to the cohort. To calculate the standardized incidence ratio (SIR), the observed number of cases was divided by the expected number. The statistical significance of the difference in SIR between exposed and unexposed women was assessed using the Mantel-Haenszel χ 2 test, on the presumption that the observed number of cases followed a Poisson distribution. Relative risk (RR) estimates were calculated as the SIR among the hysterectomized relative to the SIR among the non-hysterectomized control cohorts. The adjusted estimates were obtained from the model including hysterectomy and the variables to be adjusted for (education, parity and follow-up). Due to missing information in the covariates, added numbers of cancer cases of specific cancer sites in Table 2 are not equal to the number of all sites together. RESULTS Women with total or partial hysterectomy were more often nulliparous and less often had more than three children than women in control cohorts (Table 1). In addition, women with total hysterectomy less often had high education compared to non-hysterectomized women. Information on education was missing for a large proportion of subjects already in the source file from Statistics Finland, among both hysterectomized and non-hysterectomized women. However, the distribution of available educational information was equal among both hysterectomized and non-hysterectomized women, in both partial and total hysterectomy groups (Table 1). During the follow-up, 1964 cancers were observed versus 2191 expected in the total hysterectomy group and 763 versus 794 in the partial hysterectomy group (Table 2). As expected, the incidence of genital cancers was much below the national level in both groups. The

3 478 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY TABLE 1 Person-years (in thousands) at risk by education, parity, follow-up (years since the entry to the cohort), and period among totally and partially hysterectomized women and among age- and municipality-balanced non-hysterectomized women from Mass Screening Registry Partial hysterectomy Total hysterectomy Yes No Yes No Education Low Medium High Unknown Parity Follow-up (years) Period observed numbers of non-genital cancer among both totally and partially hysterectomized women was 3 4% above the expected value (Table 2). The SIR of rectal and thyroid cancer were significantly increased among women with total hysterectomy. In particular the risk of papillary thyroid cancer (SIR = 2.15, 95% CI : ) was increased among totally hysterectomized women, whereas the risk of follicular thyroid cancer (SIR = 0.87, 95% CI : ) was not. In all, among the control cohort for totally hysterectomized women, 2030 cancer cases were observed versus 2133 expected, and among control women for partial hysterectomy, 762 cases were observed versus 774 expected. The only site-specific SIR among the control cohort for totally hysterectomized women which differed significantly from unity were those for cervix (SIR = 0.63, 95% CI : ) and lung (SIR = 0.80, 95% CI : ). Within the control cohort for partially hysterectomized women, incidence of cancers of the gallbladder and biliary tract (SIR = 0.47, 95% CI : ) and the papillary type of thyroid cancer (SIR = 0.23, 95% CI : ) were significantly decreased. In the multivariate analysis with adjustment for education, parity and follow-up, the RR of all nongenital cancers (RR = 1.07), rectal cancer (RR = 1.40) and thyroid cancer (RR = 2.10) were significantly increased among women who had undergone total hysterectomy in comparison to their control cohort (Table 2). None of the RR for partially hysterectomized women were significantly different from unity. The adjusted RR for overall cancer and cancers of the rectum, lung and thyroid were increased only among women with total hysterectomy below the age of 55 (Table 3). The risks for lung cancer and rectal cancer were highest years after entry to the cohort whereas the risk of thyroid cancer increased constantly throughout the follow-up (Table 4). The risk of melanoma was statistically significantly increased after 15 years of follow-up, but not before that. DISCUSSION The main result of our study was that hysterectomy is not associated with a substantial protective or promoting effect for cancers in general. There was a slight increase in the overall risk due to an excess of papillary thyroid cancer, rectal cancer, lung cancer and skin melanoma. These findings were clearest among women with hysterectomy performed pre- or perimenopausally and 10 years or more after hysterectomy. An exception was thyroid cancer, the risk of which increased throughout the follow-up time. Increased risks were not due to treatment of cancers leading to the hysterectomy, because hysterectomized women as well as non-hysterectomized control women with any gynaecological cancer prior to the beginning of follow-up were excluded from the analyses. However, the risk of papillary thyroid cancer may reflect a similar aetiology for hysterectomy and papillary thyroid cancer or differences in the probability of diagnosis, given disease status, between hysterectomized and non-hysterectomized women. The cancer registration system in Finland has almost complete coverage, 27 and the computerized record linkage procedure is precise. 28 Therefore, technical incompleteness is not likely to bias the results. The follow-up for deaths and emigration for study cohorts was complete for the period of this study. The quality of hysterectomy information in the MSR has not been evaluated, and there is a possibility of non-differential misclassification error, which may decrease the observed RR. In a British study comparing a postal survey and hospital records with regard to recall of surgical history, there was 90% agreement for the operations undergone, and 82% agreement for the year of surgery. 29

4 HYSTERECTOMY AND SUBSEQUENT RISK OF CANCER 479 TABLE 2 Numbers of cancer cases (N), standardized incidence ratios (SIR, reference group all Finnish women of the same age and period) and relative risks (RR, reference group non-hysterectomized women with same distribution of age and municipality from Mass Screening Registry) of cancer with 95% confidence intervals (CI), by primary site and extent of hysterectomy Primary site Partial hysterectomy Total hysterectomy N SIR RR a N SIR RR a 95% CI 95% CI 95% CI 95% CI All sites Genitals Cervix Corpus uteri Ovary Other All non-genitals Stomach Colon Rectum Gallbladder Pancreas Lung, bronchus Breast Kidney Bladder Skin melanoma Thyroid All haematological b a Adjusted for education, parity, and follow-up. b Includes non-hodgkin lymphoma, Hodgkin s disease, myeloma and leukaemia.

5 480 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY TABLE 3 Relative risks (RR) and 95% confidence intervals (CI), adjusted for follow-up years, education and parity of non-genital cancer by age at reporting total hysterectomy. Reference group: age- and municipality-matched non-hysterectomized women from Mass Screening Registry Primary site Age at reporting total hysterectomy 45 years years 55 years RR 95% CI RR 95% CI RR 95% CI All non-genitals Stomach Colon Rectum Bladder Gallbladder Pancreas Lung, bronchus Breast Kidney Melanoma Thyroid All haematological a a Includes non-hodgkin lymphoma, Hodgkin s disease, myeloma and leukaemia. These results can most probably be generalized for Finland as well, i.e. self-reported hysterectomy information in our study may be considered valid. However, the major weakness of the study is the lack of information concerning ovariectomy. Therefore, we cannot clearly discern the effects of oophorectomy from those due to hysterectomy alone. There is a strong geographical variation within Finland in the occurrence of several cancer types including thyroid cancer 30 and also for hysterectomy. 31,32 We were able to control for the effect of geographical variation by choosing control cohorts with the same distribution of municipalities of residence as for the women reporting a hysterectomy. Even if educational information was missing for a large proportion of women in the source file from Statistics Finland, it is not likely to affect strongly the main results of the study. We did not observe a clear association between hysterectomy and breast cancer. In several earlier studies 6 7,9 12,17 women who have undergone hysterectomy have had decreased risk of breast cancer. These findings are likely to be due to ovarian ablation accompanying hysterectomy. Irwin et al. 17 found a 30% decrease in breast cancer risk after hysterectomy with bilateral oophorectomy and a 20% decrease after hysterectomy with preservation of one or both ovaries. These risk estimates were adjusted for use of oestrogen replacement therapy. At the end of the 1980s, approximately a quarter of all hysterectomies in Finland were accompanied by a bilateral oophorectomy. 4 Information on the frequency of bilateral oophorectomy in the 1970s in Finland is not available. Our result that the RR of ovarian cancer was 40% lower among women who had undergone total hysterectomy and close to expected among women who had undergone a partial operation, suggests a high frequency of bilateral oophorectomy with total hysterectomy, although it has also been suggested that hysterectomy without oophorectomy may reduce the risk of ovarian cancer In Finland hormone replacement therapy (HRT) has been routinely used after oophorectomy which may have resulted in the disappearance of the protective effect despite frequent oophorectomy. Modern HRT contains lower levels of hormones than the very first products used at the beginning of the 1960s. Therefore it may be possible that modern HRT does not increase the risk of breast cancer, even if older HRT regimens did. Our results represent the effect of oophorectomy combined with the older HRT treatments used two decades ago. Even if the specific biological mechanisms underlying the associations reported in Table 3 are not clear, results from previous studies concerning both risk factors and pathogenesis of these cancers provide useful background for interpretation of our results. The risk of thyroid cancer has been associated with reproductive factors and weight, e.g. increased weight in late

6 HYSTERECTOMY AND SUBSEQUENT RISK OF CANCER 481 TABLE 4 Relative risks (RR) and 95% confidence intervals (CI), adjusted for education and parity, of non-genital cancer by year of follow-up since total hysterectomy. Reference group: age- and municipality-matched non-hysterectomized women from Mass Screening Registry Primary site Years of follow-up RR RR RR R 95% CI 95% CI 95% CI 95% CI All non-genitals Stomach Colon Rectum Bladder Gallbladder Pancreas Lung, bronchus Breast Kidney Melanoma Thyroid All haematological a a Includes non-hodgkin lymphoma, Hodgkin s disease, myeloma and leukaemia. adulthood, use of fertility drugs, and miscarriage or stillbirth at first pregnancy, pregnancy soon after puberty, 36 parity, 37 late last birth and long reproductive period. 38 All these reproductive features including obesity describe a high oestrogen level in adolescence and premenopause. Other suggested risk factors for the papillary type of thyroid cancer are family history of goitre, heart disease, biliary disorder and female genital cancer. 36 It is unlikely that any of these factors would be associated with hysterectomy so that there would be confounding. The most common benign indication for hysterectomy is uterine leiomyosis, which is not a recognized risk factor of thyroid cancer. Thus, there is no evidence for counfounding by indication. According to Levi et al. 19 women of all ages who underwent artificial menopause had an increased risk of epithelial thyroid cancer (odds ratio = 6.3, 95% CI : ) compared to all premenopausal women. This is in line with our results, according to which an excess of papillary type of thyroid cancer was found among hysterectomized women. Although the risk of thyroid cancer among both control cohorts was slightly lower than among the Finnish general female population, this did not explain our result. However, the lack of latency between hysterectomy and appearance of increased risk of thyroid cancer suggests a possibility of bias. Finally, a number of previous studies both from Europe 19,20 as well as the US have concluded that hysterectomy increases significantly the risk of thyroid cancer. Oestrogen receptors also occur in thyroid cancer tissue, which provides further biological plausibility for the association between female sex hormones and thyroid cancer In bilateral oophorectomy, levels of oestrogen decrease, which affects not only pituitary gonadotropins, but also other pituitary hormones directly or indirectly involved in the reproductive function. A decrease in thyrotropin response to thyroid stimulating hormone (TSH) following ovariectomy has been reported, 42 which may enhance the growth of thyroid tumours. The risk of melanoma has also been associated with reproductive factors and oral contraceptive use. Multiparity 43 and hyperpigmentation during a prior pregnancy 44 have been associated with increased risk of melanoma. High socioeconomic status has been associated with a high frequency of hysterectomy 32 and an increased risk of papillary thyroid cancer. 45 However, in our study, women reporting a total hysterectomy had less education than those in the control cohort. Furthermore, we adjusted for education in the analysis to control for possible confounding. In spite of our main result, that hysterectomy does not have a large effect on the risk of total cancer, our study raises some concern about excess risk of thyroid cancer and possibly an association with rectal cancer and skin melanoma. REFERENCES 1 Pokras R, Hufnagel V G. Hysterectomy in the United States Vital and Health Statistics: Hysterectomies in the United States Series 13, Data from the National Health Survey. Publ. no. 92. USA, 1987.

7 482 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY 2 Wilcox L S, Koonin L M, Pokras R et al. Hysterectomy in the United States, Obstet Gynecol 1994; 83: Luoto R, Kaprio J, Keskimäki I et al. Hysterectomy in Finland ; incidence, indications and surgical methods. Int J Epidemiol 1994; 23: Andersen T F, Madsen M, Loft A. (Regional variations in use of hysterectomy). In Danish, with English summary. Ugeskrift Laeger 1987; 149: Luoto R, Hemminki E, Topo P et al. Hysterectomy in Finland prevalence and women s own opinions. Scand J Soc Med 1992; 20: Lilienfeld A M. The relationship of cancer of the female breast to artificial menopause and marital status. Cancer 1956; 9: MacMahon B, Feinleib M. Breast cancer in relation to nursing and menopausal history. J Natl Cancer Inst 1960; 24: Wynder E L, Bross I J, Hirayama T. A study of the epidemiology of cancer of the breast. Cancer 1960; 3: Feinleib M. Breast cancer and artificial menopause. A cohort study. J Natl Cancer Inst 1968; 41: Salber E, Trichopoulos D, MacMahon B. Lactation and reproductive histories of breast cancer patients in Boston, J Natl Cancer Inst 1969; 43: Trichopoulos D, MacMahon B, Cole P. Menopause and breast cancer risk. J Natl Cancer Inst 1972: 48: Sartwell P E, Arthes F G, Tonascia J A. Exogenous hormones, reproductive history and breast cancer. J Natl Cancer Inst 1977; 59: Vorherr H, Vorherr U F, Argubright K F. Tubal sterilization in relation to fibrocystic disease and breast cancer (Abstract). Arch Gynecol 1985; 237(Suppl.): Booth M, Beral V, Smith P. Risk factors for ovarian cancer: a case-control study. Br J Cancer 1989; 60: Hankinson S E, Hunter D J, Colditz G A et al. Tubal ligation, hysterectomy, and risk of ovarian cancer: a prospective study. J Am Med Assoc 1993; 270: Parazzini F, Negri E, LaVecchia C et al. Hysterectomy, oophorectomy and subsequent ovarian cancer risk. Obstet Gynecol 1993; 81: Irwin K L, Lee N C, Peterson H et al. Hysterectomy, tubal sterilization, and the risk of breast cancer. Am J Epidemiol 1988; 127: Hirayama T, Wynder E L. A study of the epidemiology of cancer of the breast. Cancer 1962; 15: Levi F, Franceschi S, Gulie C et al. Female thyroid cancer: the role of reproductive and hormonal factors in Switzerland. Oncology 1993; 50: Franseschi S, Fassina A, Talanini R et al. Risk factors for thyroid cancer in North Italy. Int J Epidemiol 1989; 18: Ron E, Kleinerman R, Boice J E, Livolsi V A, Flannery J T. A population-based case-control study of thyroid cancer. J Natl Cancer Inst 1987; 79: Preston-Martin S, Bernstein L, Maldonado A S, Henderson B E. Thyroid cancer among young women related to prior thyroid disease and pregnancy history. Br J Cancer 1987; 55: McTiernan A M, Weiss N S, Daling J R. Incidence of thyroid cancer in women in relation to reproductive and hormonal factors. Am J Epidemiol 1984; 120: Kolonel L H, Hankin J H, Wilkins L R, Fukunaga F H, Hinds M W. An epidemiologic study of thyroid cancer in Hawaii. Cancer Causes Control 1990; 1: Storm H H, Clemmesen I H, Manders T et al. Supravaginal uterine amputation in Denmark and risk of cancer. Gynecol Oncol 1992; 45: Hakama M, Louhivuori K. A screening programme that worked. Cancer Surv 1988; 7: Teppo L, Pukkala E, Lehtonen M. Data quality and quality control of a population-based cancer registry. Experience in Finland. Acta Oncol 1994; 33: Pukkala E. Use of record linkage in small-area studies. In: Elliot P, Cuzick J, English D, Stern R (eds). Geographical and Environmental Epidemiology. Oxford: Oxford University Press, 1992, pp Coulter A, McPherson K, Elliott S et al. Accuracy of recall of surgical histories: a comparison of postal survey data and general practice records. Community Med 1985; 7: Pukkala E, Gustavsson N, Teppo L. Atlas of Cancer Incidence in Finland Finnish Cancer Registry, Cancer Society of Finland. Publication no. 37. Helsinki, Keskimäki I, Teperi J, Aro S et al. Regional Variation in the Rates of Surgical Procedures in Finland. National Center for Health Services and Development. Reports 83. Gummerus, Jyväskylä, Luoto R, Keskimäki I, Reunanen A. Socioeconomic variations in hysterectomy evidence from a linkage study of the Finnish Hospital Discharge Register and Population Census. J Epidemiol Community Health 1996 (in press). 33 Goodman M T, Kolonel L N, Wilkens L R. The association of body size, reproductive factors and thyroid cancer. Br J Cancer 1992; 66: Preston-Martin S, Jin F, Duda M J et al. A case-control study of thyroid cancer in women under age 55 in Shanghai. Cancer Causes Control 1993; 4: Paoff K, Preston-Martin S, Mack W et al. A case-control study of maternal risk factors for thyroid cancer in young women (California, United States). Cancer Causes Control 1995; 6: Wingren G, Hatschek T, Axelson O. Determinants of papillary cancer of the thyroid. Am J Epidemiol 1993; 138: Hallquist A, Hardell L, Degerman A et al. Thyroid cancer: reproductive factors, previous diseases, drug intake, family history and diet. A case-control study. Eur J Cancer Prev 1994; 3: Akslen L A, Nilssen S, Kvale G. Reproductive factors and risk of thyroid cancer. A prospective study of women from Norway. Br J Cancer 1992; 65: Cohen C, De Rose P B, Campbell W G et al. Estrogen receptor status in malignant melanoma. Am J Dermatopathol 1990; 12: Walker M J, Ronan S G, Han M C et al. Interrelationship between histopathologic characteristics of melanoma and estrogen receptor status. Cancer 1991; 68: Yane K, Kitahori Y, Konishi N et al. Expression of the estrogen receptor in human thyroid neoplasms. Cancer Lett 1994; 84: DeLeo V, D Antona D, Lanzetta D. Thyrotropin secretion before and after ovariectomy in premenopausal women. Gynecol Endocrinol 1993; 7:

8 HYSTERECTOMY AND SUBSEQUENT RISK OF CANCER Zanetti R, Franceschi S, Rosso S et al. Cutaneous malignant melanoma in females: the role of hormonal and reproductive factors. Int J Epidemiol 1990; 19: Holly E A, Cress R D, Ahn D K. Cutaneous melanoma in women. III Reproductive and oral contraceptive use. Am J Epidemiol 1995; 141: Pukkala E. Cancer Risk by Social Class and Occupation. A survey of cancer cases among Finns of working age. Contributions to Epidemiology and Biostatistics, vol. 7. Basel: Karger, (Revised version received November 1996)