Life style and cancer: Effect of divorce


Dear Sir,

In a recent issue of the International Journal of Cancer, Lambe and co-workers1 investigated the risk of breast cancer in women who had had children with different partners. The results were negative, in contrast to some but not all earlier studies, which have tested the hypothesis that women who have children with different men develop diverse antibodies against fetal antigens and are protected from breast cancer.2, 3 Few studies using other cancer sites have been carried out on the effects of divorce, even though it is a major life event attributable to diverse lifestyle and socioeconomic factors. In a series of studies on Norwegian women, cervical and lung cancers were found to be increased among divorced women, whereas cancers of the thyroid, endometrium, colorectum and breast were decreased, in addition to melanomas and hematologic malignancies.4, 5 Even less is known about the male consequences of divorce and life with a divorced partner. Again, Norwegian data suggest that divorced men may be at risk of prostate cancer, yet the results were of borderline significance.6

Due to the lack of data on cancer risks after divorce (we use “divorce” only as a technical term because in Sweden many couples who have children are not formally married), we used the latest update of the Swedish Family-Cancer Database to study the effects of divorce in 1 particular type of situation, and risks were calculated as follows: (i) for women who had had children with 2 or more men, compared with those who had only 1 father to their children, or who had only 1 child; and (ii) for men from such unions, i.e., for the last fathers of the couple's common children, without consideration of whether or not the men had been divorced. The reason for selecting the last rather than the first male partner was that they were the ones who continued living with the women under study. Our hypotheses were that tobacco- and alcohol-related and sexually transmitted cancers were more common in women who had children with different partners, and similar findings were assumed for their partners. A further hypothesis was that divorced women were economically deprived and would show a decrease in affluent lifestyle cancers such as colon cancer and melanoma.

The 2001 update of the nation-wide Swedish Family-Cancer Database, covering over 10.2 million individuals and 0.76 million invasive cancers, was used to calculate site-specific cancer risks among women and their partners.7 The Database includes offspring born since 1932 and their biologic parents; it is linked to the nation-wide Swedish Cancer Registry from the years 1958–1998, which has been described in detail in this journal.8 The following ICD-7 codes were pooled: upper aerodigestive cancer (larynx, lip, mouth and pharynx) and leukemia (leukemia, polycythemia vera and myelofibrosis). Rectal cancer was divided into cancers of the anus and mucosal rectum. Skin cancer was squamous cell carcinoma because basal cell carcinoma of the skin has not been registered in the Cancer Registry. The study included only couples who had the last child of the particular woman together. Follow-up was started at birth of the couple's last child or January 1, 1958, whichever came latest. Follow-up was terminated on death, emigration or the closing date of the study, December 31, 1998.

Standardized incidence ratios (SIRs) were calculated as the ratio of observed (O) to expected (E) number of cases. The expected numbers were calculated from 5-year age, sex, tumor type, region (large cities and the rest of the country), period and socioeconomic status (manual workers, blue collar worker, professionals and other) specific standard incidence rates for all women or men. The data on residential region and socioeconomic status were obtained from a national census. For breast, endometrial and ovarian cancers, further adjustments were made for parity and age at first birth, both calculated from the Database.9 Confidence intervals (95% CIs) were calculated assuming a Poisson distribution.10 Cancer sites were included in the study if a total of more than 1,000 cases were recorded, or if any of the comparisons were significant.

SIRs are shown in Table I in 3 groups of women, those with 2 or more children with 1 partner or more partners, and those with 1 child only. Among women who had had children with a single husband, all the SIRs were close to unity even though some were formally significant because of the large numbers. The SIRs were 0.87 for lung, 0.88 for laryngeal and 0.93 for cervical cancers; the highest SIR was 1.03 for melanoma. When a woman had had 1 child, the SIRs were close to 1.00, i.e., they ranged from 0.99 to 1.03 (Table I). Women who had had children with 2 or more men showed both increased and decreased SIRs. The highest SIRs were 1.64 for cervical, 1.55 for laryngeal, 1.48 for anal and 1.33 for lung cancers. Even increases at the upper aerodigestive tract, other female genitals, bladder and endocrine glands were significant. The SIRs for cancer of the esophagus, small intestine, kidney, skin and connective tissue and for Hodgkin's disease were not changed. All the remaining cancer sites showed a decrease in risk, including colon (0.90), rectum (0.84), breast (0.90), endometrium (0.88) ovary (0.89), non-Hodgkin's lymphoma (0.88) and multiple myeloma (0.79). The data for these reproductive sites were further adjusted for parity and age at first childbirth; the SIRs changed toward unity but breast and endometrial cancers remained significant: breast SIR 0.95 (95% CI 0.91–0.99), endometrium 0.88 (95% CI 0.79–0.97) and ovary 0.95 (95% CI 0.88–1.03).

Table I. SIR for Cancers in Wife1
Wife cancer sitesFirst child and last child have same parentsFirst child and last child have different fatherFamily has only one child
  • 1

    O, observed; E, expected; SIR, standardized incidence ratio. Bold face shows that the lower confidence interval exceeds 1.00.

Upper aerodigestive tract1,7241,789.130.960.921.01391326.471.201.081.321,0931,092.401.000.941.06
Small intestine784759.041.030.961.11113132.780.850.701.02420425.180.990.901.08
Cervix uteri9,14510,525.510.870.850.893,4032,077.341.641.581.694,6734,618.161.010.981.04
Other female genital1,4871,554.670.960.911.01355282.831.261.131.39912916.501.000.931.06
Nervous system7,4467,320.431.020.991.041,1901,303.300.910.860.973,5353,547.271.000.961.03
Connective tissue1,3461,316.891.020.971.08210236.570.890.771.01679681.531.000.921.07
Non-Hodgkin's lymphoma5,1064,989.901.021.001.05781884.240.880.820.952,5322,544.850.990.961.03
Hodgkin's disease1,0231,004.991.020.961.08162177.600.910.781.06635637.411.000.921.08
Multiple myeloma2,4332,347.331.041.001.08326410.900.790.710.881,3071,307.771.000.951.05

For men who had fathered all the children of a woman, only the SIR (1.03) of melanoma was increased (Table II). The risk at many sites was decreased, most for esophagus (0.89) and lung (0.90). Among men who had fathered the last child of a woman who had had a previous child or children with another partner, only 2 sites showed a decrease (melanoma and skin, both 0.89), whereas many cancers were in excess, including those of the esophagus (1.51), lung (1.42), anus (1.41), larynx (1.39), liver (1.37) and pancreas (1.34). In 1-child families, all the SIRs ranged between 0.97 and 1.10. The highest SIR of 1.10 was for testis cancer, which could have been the reason why the family had only had 1 child.

Table II. SIR for Cancers in Husband1
Husband cancer siteFirst child and last child have same parentsFirst child and last child have different fatherFamily has only 1 child
  • 1

    For abbreviations, see Table I footnote. Bold face shows that the lower confidence interval exceeds 1.00.

Upper aerodigestive tract4,9005,360.850.910.890.941,001769.811.301.221.382,8112,581.341.091.051.13
Small intestine1,0461,092.660.960.901.02183153.
Other male genital773785.400.980.921.05122111.501.090.911.30370368.101.010.911.11
Nervous system7,0757,324.760.970.940.991,2941,103.961.171.111.243,1253,
Connective tissue1,6211,673.910.970.921.02278241.791.151.021.29769752.301.020.951.10
Non-Hodgkin's lymphoma7,1907,306.290.980.961.011,1231,019.941.101.041.173,2443,230.771.000.971.04
Hodgkin's disease1,5731,698.790.930.880.97350267.971.311.171.45783739.231.060.991.13
Multiple myeloma3,4923,485.981.000.971.04482463.291.040.951.141,6801,704.730.990.941.03

This study appeared to distinguish through divorce 2 types of lifestyle patterns, which were associated with both female and male cancers, one in families that involved 1 father and the other in families that involved more fathers. Cancer patterns in the 1-child families fell between these 2 types; women had somewhat lower than average risks for most cancers and men somewhat higher than average risks. However, the SIRs for women and men in these 1-child families were so close to unity that we do not discuss their results further. In families in which all the children had the same parents, cancer rates were average or lower than average. The lower than average rates were found for cancers associated with known strong environmental factors, such as lung and larynx (tobacco), upper aerodigestive tract, esophagus and liver, particularly for men (tobacco and/or alcohol) and cervical and anal cancer (human papillomavirus and multiple sexual partners). The data suggest that in these “stable” families, exposure to lifestyle-related carcinogenic factors was lower than average. The only exception was the small increase in the risk of melanoma (SIR 1.03) that was found both in women and men, probably relating to the relative affluence of these families and their long common life history, even for exposure to sun.

In families in which the woman had had children with more than 1 man, a difference in cancer risk was observed for women and men. For men many cancers were in excess; the highest increase was in the lifestyle-related cancers discussed above, which conformed to the study hypothesis. For women lifestyle-related cancers (lung, larynx, cervix, anus but not liver) were increased, but for a larger number of cancer sites a decrease was observed. The apparent protection was observed for common cancers such as breast, ovary, endometrium, colorectum, stomach, pancreas and skin melanoma, and for lymphoma and leukemia. We showed additionally that much of the effect for breast, ovarian and endometrial cancer could be explained by reproductive factors (parity and age at first birth), but for most other sites no such obvious explanation exists, and our data were adjusted for socioeconomic factors. These data were consistent with the Norwegian study4 suggesting that the protective effects were real. The likelihood that women who have children with multiple partners eventually remain sole supporters of the family is large, and such families are economically deprived. Compared with other women in the same social class, this may result in a lifestyle of less obesity, more physical activity and less solar exposure, which are protective against some of the cancers for which the risk was decreased. This was also partially consistent with our initial hypothesis.

However, we have no way of refuting the fetal antigen hypothesis; in fact, the decreased risk in women with multiple partners for non-Hodgkin's lymphoma and squamous cell skin cancer, both immune function-related malignancies,11, 12 would be in line with this hypothesis. Why the last father of the children of these women cannot enjoy any of these protective effects remains enigmatic and may relate to further life events and perhaps broken marriages for these men.