- Top of page
- Material and methods
The association of contraceptive methods, including oral contraceptives (OC), intrauterine devices (IUD) and tubal sterilization (TS), with overall and site-specific cancer were prospectively investigated in a cohort of 66,661 Chinese women in Shanghai, 76.7% of whom used contraception. During a median follow-up time of 7.5 years, 2,250 women were diagnosed with cancer. Ever-use of any contraceptive method was not associated with overall cancer risk [adjusted hazard ratio (HRadj) = 1.02, 95% CI, 0.92–1.12]. Use of any contraceptive method was associated with increased risk of rectal cancer (HRadj = 1.68, 95% CI, 1.08–2.62) and reduced risk of thyroid cancer (HRadj = 0.63, 95% CI, 0.38–1.04). Risk of gallbladder cancer increased with ever use of OC (HRadj = 2.38, 95% CI, 1.26–4.49). IUD use was associated with a possible reduced risk of thyroid cancer (HRadj = 0.64, 95% CI, 0.38–1.07). Longer duration of IUD use decreased risk for breast, thyroid and lung cancers. Ever having a TS was associated with increased uterine body cancer (HRadj = 2.50, 95% CI, 1.47–4.25) and decreased risk of stomach cancer (HRadj = 0.59, 95% CI, 0.39–0.91). We did not find any contraceptive method to be related to the risk of ovarian cancer but the analyses were based on few events. Although chance findings are a likely explanation for some of the associations found in our study, these findings suggest that various contraceptive methods or reproductive patterns may play a role in the etiology of cancer. © 2008 Wiley-Liss, Inc.
The use of contraception is one of the most important reproductive health issues among women. In 2005, according to the United Nations, over 660 million married or cohabitating women of reproductive age (15–49 years) used some type of contraceptive method,1 and 450 million used oral contraceptives (OC), intrauterine devices (IUD) or tubal sterilization (TS). The beneficial and harmful effects of these 3 most widely used contraceptive methods in relation to common cancers have been a major topic of medical research.2–4
There are over 100 million users of OC or exogenous reproductive hormones worldwide.5 Reproductive hormones, in particular estrogen and progestin, act at the cellular level through their hormone receptors on the cell to initiate essential biological responses and play a role in initiating carcinogenesis.6, 7 Therefore, the relationship between use of OC and cancers of the reproductive system has been investigated extensively.5, 8 While several studies have shown that OC use provides protection against ovarian and endometrial cancers,5, 8, 9 other studies have shown no effect10 or even an increase in risk of reproductive or breast cancers,11–13 particularly among pre-menopausal women.14 Epidemiological studies have also investigated the use of OC in relation to non-reproductive system cancers, including colon, rectal, lung, stomach, liver, gallbladder and thyroid cancers.11, 15–17 In the majority of studies, the use of OC was associated with a decreased risk of colorectal cancer,18–20 but an increased risk of liver cancer.21, 22
Another popular method of contraception is the IUD with approximately 150 million users worldwide,1 including 92 million users in China alone.23 Previous studies have mainly focused on the risk of cancers of the reproductive system and breast in relation to IUD use. Most previous studies24–26 and 2 recent meta-analyses have shown that IUD use may reduce the risk of endometrial cancer.3, 27 However, other studies have found no association.28, 29 Tubal sterilization (TS) is used by approximately 220 million women worldwide, 88 million of whom are in China.1 Several studies have reported a beneficial effect of TS on ovarian cancer risk30–32; however, the association between endometrial cancer and TS is inconclusive.31, 33, 34
Oral contraceptives, IUD and TS remain the methods of choice for women worldwide and their relation to the risk of neoplasia will continue to stay at the forefront of cancer research in order to resolve the continuing uncertainty. There is a substantial body of literature on OC use and cancer; however, relatively few cohort studies have examined the relationship between use of OC, IUD or TS and the risk of cancers of both reproductive and non-reproductive systems in developing countries, where the incidence rate of cancers differs substantially from that of developed countries.5, 12, 35 Based on the data from Shanghai Women's Health Study, a large, population-based, prospective cohort study, we examined the role of OC, IUD and TS use in the etiology of all cancers combined and in cancers of the breast, uterine body, ovary, thyroid, colon, rectum, liver, gallbladder, pancreas, stomach and lung.
Material and methods
- Top of page
- Material and methods
The Shanghai Women's Health Study is a population-based, prospective study of women 40–70 years of age at recruitment who resided in 7 urban districts of Shanghai, China. Details of the study have been described elsewhere.36 In brief, a total of 81,170 women were approached for the study between March 1997 and May 2000. In total, 74,942 women (92.3%) completed the baseline survey, which included an in-person interview and a self-administered questionnaire. Reasons for non-participation included refusal (n = 2,407), absence during the recruitment period (n = 2,073), and other miscellaneous reasons (n = 1,748). All cohort members have been followed by a combination of in-person surveys conducted every 2 years and record linkages to the Shanghai Cancer Registry and the Shanghai Vital Statistics Registry. Nearly all cohort members have been successfully followed; the response rates for first (2000–2002), second (2002–2004) and third (2004–2007) in-person follow-up surveys were 99.8%, 98.7% and 96.7%, respectively. Participants who reported being diagnosed with cancer (n = 2,440) at baseline or who were lost to follow-up immediately after study enrollment (n = 8), who had undergone hysterectomy or ovariectomy (n = 7,085) at baseline, and who used injectable contraceptives (n = 1,821) were excluded from the analysis. The total number of cancer cases that had also reported use of injectable contraceptives was small (n = 39), which prohibited an examination of the role of this method and its dose–response effect on site-specific cancers. As a result, we excluded injectable contraceptive users from this study. This left a total of 66,661 women for the present analysis. The study was approved by all relevant institutional review boards in the People's Republic of China and in the United States. All participants provided written, informed consent.
Incident cancer was documented in 2,250 participants. All incident cancer cases were verified by home visits and confirmed by medical chart review and, for most cases, by independent review of pathological slides by study pathologists. Cancer type was classified using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM).37 Clinical events of interest were cases of cancer of the stomach (code 151, n = 168), colon (code 153, n = 207), rectum (code 154, n = 136), liver (code 155, n = 95), gallbladder (code 156, n = 54), pancreas (code 157, n = 78), lung (code 162, n = 229), breast (code 174, n = 558), uterine body (code 182, n = 119), ovary (code 183, n = 94) and thyroid (code 193, n = 83). We did not investigate cervical cancer given its rarity (n = 24) in our study population.
In-person interviews were conducted at baseline to obtain information related to demographics, menstrual and reproductive history, disease and medication use, surgical history, physical activity and smoking and alcohol consumption habits. Weight and height were measured, and body mass index (BMI, kg/m2) and waist-to-hip ratios (WHR) were calculated. Smokers were defined as subjects who ever smoked at least one cigarette per day for more than 6 months prior to baseline. Alcohol drinkers were defined as women who ever drank alcohol more than thrice a week for 6 months prior to baseline.
Detailed information on contraceptive use was obtained. Participants were asked whether they had ever used any contraceptive method, including oral contraceptives, contraceptive injection, intrauterine devices (IUD) and tubal sterilization (TS). If participants had ever used these contraceptives, they were also asked for age at first use and age at last use for each method. For this study, age (years) at first use of a particular contraceptive was categorized into 2 groups: less than or equal to and greater than median age. Total duration (years) of use was defined as any continuous use of one specified method of contraception, which was calculated by subtracting age at first use from age at last use and was categorized into 2 groups: less than or equal to and greater than median distribution. The number of years since last use of oral contraceptives or IUD was calculated by subtracting age at last use from age at baseline, which was then categorized into 2 groups: less than or equal to and greater than median distribution. The median distribution was chosen because of the small sample size for site-specific cancer.
Statistical tests were based on 2-sided probability with a significance level of 0.05. Descriptive analyses, including age-adjusted prevalences and age-adjusted means of demographic characteristics, were generated according ever use of specific types of contraceptive method. Cox proportional hazards regression was applied to derive multivariate adjusted hazard ratios (HRadj) and their 95% confidence intervals (CI) for all-cancers combined, for specific cancers in relation to ever/never use of various contraceptive methods, and for patterns of use that included total duration of use, age at first use, and time since last use (for OC use).
In the regression analyses, we modeled age as a time scale in the cohort (from age at entry to age at exit).38 Covariates included education (categorized), age at menarche (continuous), number of live births (0, 1, 2 and ≥3), BMI (categorized into quartiles), regular exercise in the 5 years preceding the interview (yes/no), cigarette smoking (ever/never), menopausal status (pre-/post-), and first degree family history of any cancer (yes/no). The use of other contraceptive methods was added in the regression models for analyses of duration, age at first use, and time since last use of a particular contraceptive method. Only first primary cancers were included in the analysis. In the analysis of specific types of cancer, women with other types of cancer were censored at date of cancer diagnosis. All data management and statistical analyses were performed using the statistical software SAS.9.1 (SAS Institute, Cary, NC).
- Top of page
- Material and methods
In this large, population-based cohort, we found no evidence that contraceptive use in general was related to overall cancer risk. However, we did find both benefits and risks of using OC, IUD, and TS in relation to some common cancers.
Consistent with the Nurses' Health Study,39 we did not find that ever use of OC, in general, was associated with breast cancer risk. The borderline decreased risk observed in our study among women who were 29 years of age or younger when they began using OC is in contrast with previous findings of an increased risk among women who initiated OC use before age 3040 or ≤35 years41 and among premenopausal women.14 Inconsistent results have also been reported from other studies on OC use and breast cancer risk.5, 13 In our study, a statistically non-significant reduced risk of breast cancer was associated with IUD use, and the relationship was stronger with increasing duration of use or younger age at initiation. Few studies have investigated the relationship between breast cancer and IUD use and reported no increased risk with use of hormone-releasing IUD.3, 42
Several prospective studies, including the Norwegian-Swedish Women's Lifestyle and Health Cohort Study, the Nurses' Health Study, the Oxford Family Planning Association contraceptive study and reanalysis of data from 45 epidemiological studies5, 17, 31, 43 showed that long-term OC use was related to a reduced risk of ovarian cancer. In contrast, other studies have observed an increased risk of ovarian cancer associated with OC use in younger women (<40 years), and for the low-progestin or high-estrogen OC formulations.44–46 In this study and in a previously published case–control study conducted in the same population, we found a null association between ovarian cancer risk47 and age at initiation or time since last use of OC. However, we found risk of ovarian cancer increased in women who used OC for <2 years compared to women had who never used OC. In this study, we failed to find any evidence in support of previous findings that TS is associated with decreased risk of ovarian cancer.31 Our finding of a null association with TS is inconsistent with previous studies that have reported a lower risk of ovarian cancer with tubal ligation.31, 32, 48
It has been hypothesized that TS may alter the function of the ovaries by damaging the surrounding tissue, potentially disrupting the ovarian blood supply, and reducing levels of endogenous estrogen exposure.34, 49 However, some studies have found no change in hormone levels or ovarian blood flow of either the ovarian or uterine arteries following TS.50, 51 It has also been speculated that TS blocks the transportation of external carcinogenic agents from the lower reproductive organs to the ovaries and thus, prevents ovarian cancer.52 It is noteworthy that our analyses were based on few ovarian cancer events among women with TS, thus a chance finding cannot be ruled out.
A possible reduced risk of uterine body cancer with OC use, particularly with longer duration of use, was found in our study, which is in line with the majority of previous investigations,17, 43 including our previous case–control study.53 We found that the risk of uterine body cancer is positively associated with ever having TS, with greater number of years since having the procedure (≥33), and younger age at initiation of TS (<30 years), which is consistent with previous reports.33, 34 On the contrary, other studies have found a significant or non-significant reduced risk of endometrial cancer associated with TS.4, 28, 54 However, the biological causality between reproductive cancers and TS is still unclear.
There is very limited information pertaining to thyroid cancer risk in relation to OC use. Although a few studies have identified an increased risk of thyroid cancer with OC use,55, 56 several population-based, case–control studies57, 58 and 2 cohort studies59, 60 reported no association between OC use and thyroid cancer. To the best of our knowledge, no study has previously evaluated the association of thyroid cancer with use of other contraceptives. We found a reduced risk of thyroid cancer with use of any contraceptive method and with IUD use. Previous studies reported that a normal pregnancy results in a number of important physiological and hormonal changes that alter thyroid function.61, 62 Thus, the beneficial effects of contraceptive use, in general, and IUD use or TS, in particular, on thyroid cancer may be explained by the lack of physiological or hormonal changes in the thyroid glands due to prevented pregnancies and births.63, 64 However, it is also possible that women who had thyroid disorders avoided pregnancy because of concerns about pregnancy-related complications or pregnancy outcomes.
Previous findings on the associations between OC use and risk of colorectal cancer have been inconclusive. Our findings suggest that OC use may increase the risk of rectal cancer as found in The Melbourne Colorectal Cancer Study,65 or possibly among women with less than 2 years of use, users with later age at initiation (≥29 years), and past users (≥19 years since last use). We also found a possible positive association between colon cancer risk and ever use of OC, particularly with a shorter-term use, later age of OC initiation or less than 19 years since last use. This is consistent with Rosenblatt et al.'s finding of a positive association between colon cancer risk and OC use for >3 years among textile workers in Shanghai.66 In contrast, the Women's Health Study found a moderately reduced risk of colorectal cancer with OC use, regardless of the duration of use.67 Similarly, the Nurses' Health Study also showed that OC use for ≥96 months resulted in a 40% lower risk of colorectal cancer compared to never use.68 In addition, a recent review,11 a meta-analysis,69 and the Canadian Women's Cohort Study20 all reported a lower risk of colorectal cancer or cancer of the colon and rectum associated with OC use.69 However, several other studies have failed to find any association between OC use and colon, rectal70, 71 or colorectal cancer.72, 73 In contrast to the findings of the Ontario Women's Cohort Study, which reported a reduced risk of colorectal cancer related to TS,74, 75 we found no association for colon cancer and a possible positive association for rectal cancer with TS.
Our findings for increased risk of gallbladder cancer with OC use are consistent with reports from case–control studies,21, 76 but contrast with the findings of other studies.75, 77 Previous studies have reported that gallbladder disease or gallstones and gallbladder cancer share certain epidemiological features.78, 79 Studies have also found an increased risk of gallbladder disease with OC use.75, 79 We also found an increased risk of gallbladder cancer with short term use of TS (<33 years), however, it is difficult to compare our findings with other studies, since to date no similar studies have been conducted. Thus, the true association of TS and this cancer remains to be established.
Our findings of increased risk for stomach cancer in relation to ever use of IUD are in line with the findings of another of our studies in the same population.15 We cannot exclude the possibility that this finding is due to chance, and our result needs to be investigated in further studies.
The strengths of this study include the population-based study design, large sample size, high prevalence of IUD and TS use and detailed information on a wide range of potential confounders. Another advantage of this study is that we were able to investigate the effects of the most widely used methods of contraception, including their dose–response effect, in relation to common cancers. The second important strength of this study is that our results are not affected by the potential confounding from hormone replacement therapy (HRT) use, given that the majority of study participants were not exposed to exogenous hormones (only 3% of all participants). There is growing evidence that HRT plays a role in cancer risk,71, 80 and results from studies conducted in Western countries are likely to be confounded by HRT use given its high prevalence in Western populations.81, 82 As far as we know, the present study is the only one of its kind to investigate the risk of both reproductive and non-reproductive cancers in relation these common contraceptive methods based on a large, prospective, population-based cohort study.
Several limitations should be mentioned. Owing to the nature of self-reported OC, IUD and TS exposures, misclassification errors, particularly regarding OC use, are unavoidable, which may have biased associations of use of these methods and site-specific cancers towards the null. Low numbers of site-specific cancers led to less than optimal statistical power and prevented more comprehensive assessments. Another limitation is that we did not have information about the specific types of OC and IUD or about surgical methods used for TS. This prevented an assessment of subtypes of contraceptive methods that may have had different effects on carcinogenesis. Since most women in our study population were in their 50s and 60s at the time of the study interview, their OC use may have included mainly medium and low dose preparations. These preparations were prescribed worldwide between 1970 and 1980,5, 8 which overlaps the time that many of our participants used OCs. Copper-bearing IUDs23 and laparotomic tubal sterilization with non electric-cautery techniques have been widely used since the 1980s in China. Although our results may not be directly generalizable to all contraceptive users, our findings are relevant to women in countries with a similar make-up of available contraceptives.
To summarize, this population-based, prospective cohort study found that the most widely used contraceptive methods do not seem to affect the overall risk of developing cancer or risk of developing breast and gynecologic cancers among women in China. However, a few associations between specific contraceptive methods and selected cancer types were noted. Further research is warranted to determine the nature of these associations.