Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
48 Health Effects of Contraception Nancy C. Lee, Herbert B. Peterson, and Susan Y. Chu INTRODUCTION Until the 1960s rhythm and barrier contraceptives were the only methods of birth control widely available to couples desiring to plan the number and spacing of their children. In the 1 960s oral contraceptives (OCs) were introduced and new efficacious intrauterine devices (IUDs) became widely available, so that the choice of effective methods of contraception increased substantially. Later, in the 1970s, female and male sterilization techniques became much more widely ac- cepted and used. Couples were then able to choose from several different temporary and permanent methods of contraception and to switch from one to another. Worldwide, family planning programs expanded, and the prevalence of contraceptive use increased. As these methods of contraception became more widely used, anecdotal re- ports of adverse health effects associated with their use began to appear. Since the late 1960s and early l970s, epidemiologic studies have more rigorously evaluated the health effects associated with the use of different contraceptive methods. Most of these studies have been conducted in the United States and Europe. In the process researchers have recognized that different contraceptive methods have Nancy C. Lee and Herbert B. Peterson are deputy chief and chief, respectively, of the Epidemiologic Studies Branch, Division of Reproductive Health, Center for Chronic Disease Prevention and Health Promotion of the Centers for Disease Control. Susan Y. Chu is epidemiologist in the Special Projects Section, Surveillance Branch, AIDS Program, Center for Infectious Diseases, of the Centers for Disease Control. 48
IlEALTH EFFECTS OF CO=RACE"ION 49 important beneficial health effects, in addition to the desired effect of preventing pregnancy. Although much research is still needed, especially targeted to the developing world, a large body of information is now available to assess the health effects of the venous contraceptive methods. The various contraceptive methods have health risks, but pregnancy itself has attendant risks of morbidity and mortality. In 1983, Ory et al. attempted to quantify the mortality risks associated with using the various methods of contra- ception. Those risks were compared with the risks associated with using no method of contraception, which are actually the mortality risk associated with pregnancy. These estimates are presented in Table 1. Using no method of contraception cames a higher cumulative risk of death than using any contracep- tive method except that of OCs by older women who smoke. The mortality risks associated with using no contraception and with using OCs are higher in older women than in younger women. For all other contraceptive methods, the mortal- ity risk does not appear to vary by age. Although the estimates are not presented in Table 1, vasectomy involves no mortality risk for women and virtually none for the male partner (Ory et al., 1983~. Below we will present in detail the health effects of the venous widely available methods of contraception, limiting our discussion to those methods that are considered moderately to highly effective. Most epidemiologic and clinical studies of the health effects of contraceptives have been earned out in developed counties. We recognize the difficulty in generalizing these results to the special health and cultural situations in the less developed countnes. Furthermore, the TABLE 1 Estimated Cumulative Number of Deaths per 100,000 Nonsterile Women Aged 17-44, Aunbutablc to Contraceptive Method, by Age Group Age Group Method 17-34 35-44 Total (17-44) No method 179 270 449 PilVnonsmoker 20 230 250 PilUsmoker 127 845 972 IUD 22 20 42 Condom 17 4 21 Diaphragm/spennicide 24 25 49 Rhythm 31 32 63 Tubal sterilizations - 4 Not available by age group. Source: Modified from Ory et al., 1983, p. 36, Figure 16.
50 ~E, PETERSON, AND CHU effects of the venous contraceptive methods on the risk of diseases generally limited to less developed countries have had very little characterization. Because of the expanding role of family planning programs and contraceptive use in these countries, studies to evaluate the health effects of contraceptives in various regions and cultures are needed. Modern contraceptive methods vary substantially in how effectively they prevent pregnancies. Because pregnancy itself has attendant health risks and benefits, the rates of accidental pregnancy associated with the various methods of contraception are one important aspect to consider when measuring the health effects of these methods. In 198? Trussell and Kost published their comprehen- sive review assessing failure rates for each method of contraception. After reviewing all available studies, they estimated the rate of failure (i.e., accidental pregnancy) in the first year of use associated with "perfect)' use of each con~ace~ five. They called this estimate the "lowest expected" failure rate; this rate should reflect the frequency of failures caused by the contraceptive itself. They also reported for each method a "typical" failure rate, defined as the rate of accidental pregnancy in the first year of use among typical couples who use that method. The typical failure rate is determined both by failures as a result of imperfect use of a contraceptive and by failures directly related to the method itself. Most of the typical failure rates were derived from national surveys of U.S. women. The authors summarized these rates, by contraceptive method, in a single table (Trussell and Kost, 1987, p. 271~. We present a modified version of that table here Cable 2~. In our table and throughout the text we have substituted the term method failure rate for lowest expectedfailure rate and userfailure rate for typicalfailure rate to follow more closely the terminology used in much of the existing litera- ture. Note that the method failure rates are consistently low for most modern contraceptives. However, user failure rates vary widely, a function of the degree of acceptability and compliance required for successful use of each method. ORAL CONTRACEPTIVES OCs, a highly effective method of birth control, are available in two types. Combination OCs, the most widely used, consist of both an estrogen and a progestin component. Most combination OCs contain a fixed daily dose of an estrogen and progestin and are talcen for 21 of 28 days of each menstrual cycle (Hatcher et al., 1988~. Recently introduced, phasic combination OCs contain varying doses of the estrogen and progestin components throughout the menstrual cycle. The second type of OC is the progestin-only pill (often called the minipill), which contains only a progestin. Combination OCs with fixed doses of estrogen and progestin have been used much more frequently than phasic or progestin-only pills; hence, most epidemiologic studies on the health effects of OCs are essen- tially studies of the effects of this type of OC.
HE'UTH EFFECTS OF CONTRA CEPrION 5 ~ TABLE 2 Method Failure and User Failure Rata During the First Year of Use of a Contraceptive Method, United States Contraceplivc Method Percentagc of Women E'cpenencing an Accidental Pregnancy in the First Year of Use Method Failures User Failurc. Pin Combination 0.1 Progestin~y 0.5 IUD 6 Medicated 1 Non m cd ic at ed 2 Injectable progesiin DMPA 0.3 0.3 NET 0.4 0.4 Implants ~ORP~9 0.2 0.2 Diaphragm with spcnnicidc 3 18 Condom 2 12 SpcmucidesC 3 21 Sponge (Todays) Nulliparous 5 18 Parous >8 >28 Gp/spemucide 5 18 Periodic abstinence 20 Ovulation method 8 Symp~ermal 6 Calendar 10 Posto~rulation 2 Withdrawal 4 18 Female sterilization 0.2 0.4 Male sterilization 0.1 0.15 Among couples who use a method perfectly (both consistently and correctly), an estimate of the percentage expected to c~cperience an accidents] pregnancy during the first year if they do not stop use for any reason. Among typical couples who use a method, the percentage who c~penence an accidental pregnancy during Tic first year if they do not stop use for any reason. Idcludes failures due to the method itself as well as failures due to imperfect use of the method. Roams, creams, jellies, and vaginal suppositones. Source: Modified from Tmsscll and Kost, 1987, p. 271.
52 ~E, PETERSON, AND CHU OCs prevent pregnancy chiefly by~inhibiting ovulation in almost all menstrual cycles, although OC-related changes in the cervical mucus and endometrium may also have contraceptive effects. Failure rates associated with OC use are low. The method failure rate for combination OCs is 0.1 percent per year and for progestin-only OCs is 0.5 percent per year (Table 2~. The user failure rate cannot be readily determined for the two types of OCs separately, but it is about 3 percent per year for any type of OC. Because this user failure rate was derived from data where the vast majority of women used combination OCs, the figure probably is closest to the user failure rate for combination OCs. Most experts believe that the progestin-only pill has a higher user failure rate. The health risks and benefits of OC use have been extensively studied and documented (Ory, 1982; Ory et al., 1983; Stadel, 1986; Prentice and Thomas, 1987~. For a recent extensive review and list of references, see Stadel (1986) or Prentice and Thomas (1987~. Because OCs are highly effective at preventing any pregnancy, they appear to decrease greatly the risk of ectopic pregnancy. Results from a large case-control study of ectopic pregnancies conducted in the United States showed that current OC users had a relative riisk of ectopic pregnancy of 0.1 (95 percent confidence interval, 0.1~.2) compared with women who were using no contraceptive method (Ory, 1981~. Noncontraceptive Benefits An important benefit from OC use is a reduction in risk of two serious reproductive system cancers, endometrial and ovarian cancers. This reduction has been documented in at least 9 and 11 epidemiologic studies, respectively C1he Cancer and Steroid Hormone Study [CASH], 1987a, 1987b). Although the theoretical mechanisms that may explain these protective effects are quite differ- ent for the two types of cancer, the magnitude and characteristics of the protective effects are similar. The most detailed characterization of these protective effects comes from the CASH Study, a large case-control study conducted in the United States by the Centers for Disease Control, with support from the National Institute of Child Health and Human Development (CASH, 1987a, 1987b). OC use was associated with a 40 percent reduction in the risk of endometrial cancer as well as a 40 percent reduction in the risk of ovarian cancer, regardless of the specific formulation of combination OC used. The effect appeared to persist long after OC use had been discontinued; furthermore, protection increased with increasing cumulative duration of OC use. The protective effect of OCs on endometrial cancer is most likely related to direct effects on the endometrium. Among current OC users the carcinogenic effect of unopposed estrogen on the endometrium is probably reduced because combination OCs contain both estrogen and progestin. The continued protection
HEALTH EFFECTS OF CO=RACE~ION 53 seen among past OC users is less well understood. Perhaps the combination of estrogen and progestin irreversibly changes endometrial cells so that they are not susceptible to carcinogens or to malignant transformation (CASH, 1987a). Suppression of ovulation and suppression of pituitary secretion of gonadotropins have both been postulated as mechanisms by which OCs protect against ovarian cancer (Weiss, 1982~. Two other factors that provide protection from ovarian cancer, increasing parity and breastfeeding (Gwinn et al., submitted), may also derive their protective effects from one of these two proposed mechanisms. Available epidemiologic studies do not provide sufficient information to choose one of these postulated mechanisms over the other. Fourteen epidemiologic studies have found a decreased risk of benign breast disease (RED) associated with OC use, including both case-control and cohort studies (Stadel, 1986~. Evidence suggests that OCs decrease the risk of f~brocys- tic disease and fibroadenoma diagnosed by biopsy as well as the risk of breast lumps observed clinically but not biopsied. Results from a large cohort study conducted in the United Kingdom by the Oxford Family Planning Association have provided especially useful information about the relationship between OC use and BBD (Brinton et al., 1981~. The decreased risk of BBD seen among women who use OCs occurs primarily among current or recent users who have used them for 2 years or longer. The relative risk among women who have used OCs for more than 2 years compared with nonusers is about 0.6 for fibrocystic disease and about 0.5 for unbiopsied breast lumps. The relative risk of fibroade- noma among women who have used OCs for less than 2 years is about 0.4, which is essentially the same as the relative risk of 0.3 among women who have used OCs for 2 or more years. The decreased risk of BBD does not persist among past OC users who have not used OCs for more than 1 year. Many epidemiologic studies have found that a history of BBD increases a woman's risk of breast cancer. Even though OCs decrease the risk of BBD, epidemiologic studies have not found that OCs decrease the risk of breast cancer, as might be suggested by the OCs-BBD relationship. The most likely explana- tion for this paradox is that OCs probably decrease the risk of the large proportion of BBD that is not closely linked to breast cancer risk but do not decrease the risk of the types of BBD that increase a woman's risk of breast cancer (Stadel, 1986~. Clearly, more information about the interrelationship between OC use, histologic types of BBD, and breast cancer is needed. Seven epidemiologic studies have found that current or recent OC use reduces the risk of pelvic inflammatory disease (PID) (Stadel, 1986~. On average, these studies have found that the risk of PID among OC users is about 40 percent lower than the risk among women using no contraceptive method. The most detailed analysis of this issue comes from the U.S.-based Women's Health Study, a large hospital-based case-control study conducted from 1976 to 1978 (Rubin et al., 1982~. This study found that the overall relative risk of PID
54 r FE, PETERSON, AND CIlU among current OC users was about 0.5 compared with women using no contra- ception. However, this protective effect was limited to women who had been using OCs for at least 1 year. Furthermore, women not currently using OCs but who had used them in the past were no longer protected. Two mechanisms have been proposed to explain how OCs protect against the development of PID. First, OCs may change cervical mucus so that it prevents pathogenic organisms from ascending into the upper genital tract. Second, because OCs reduce menstrual blood flow, a decreased amount of medium may be available for bacterial growth (Rubin et al., 1982~. Most of the studies of the relationship between OCs and PID have been hospital-based studies, so that case groups in these studies consisted of women hospitalized for PID. Because many women diagnosed with acute PID are not hospitalized, findings about women who are may not be generally extended to women who develop asymptomatic PID or symptomatic PID that does not require hospitalization (Washington et al., 1985~. Specifically, gonorrhea may be an important cause of PID that requires hospitalization, whereas other bacterial etiologies such as chlamydia may cause much of the PID among women who do not require hospitalization. If OCs only protect against the bacterial etiologies of PID that are likely to require hospitalization, using OCs may not protect against some important causes of PID. Little epidemiologic evidence exists to clarify this question. Results from a large cohort study in the United Kingdom have provided clear evidence that OC use decreases the risk of iron-deficiency anemia, in both current and past OC users (Royal College of General Practitioners, [RCGPl, 1970~. The protective effect provided by current OC use is probably due to the decrease in menstrual blood flow routinely seen among OC users. An increase in iron reserves probably accounts for the persistence of the decreased risk in past users. In countries where the prevalence of iron-deficiency anemia is high, this benefit to OC users may be especially important (Stadel, 1986~. Three epidemiologic studies have found that OC use decreases the risk of functional ovarian cysts, including follicular, granulosa lutein, and theca lutein cysts (Stadel, 1986~. Decrease in risk appears to be confined to current OC users and is probably related to the suppression of ovulation that occurs during OC use. Evidence from a case-control study that used data collected from the Oxford Family Planning Association cohort study suggests that OC use protects a woman from developing uterine fibroids (Ross et al., 1986~. The risk of fibroids de- creased with increasing duration of OC use: each 5 years of OC use contributed another 17 percent reduction in fibroid risk. The mechanism of the protective effect is still speculative. The authors proposed that circulating estrogens, either exogenous or endogenous, may promote the formation of fibroids and that the decreased risk associated with OC use may be explained by the modifying effect of the progestins in OCs.
I!EALTH EFFECTS OF CO=RACE~ION 55 Adverse Health Effects Cardiovascular Effects Most epidemiologic evidence suggests that OC use increases the risk of cardio- vascular disease, in particular the risk of venous thromboembolism, myocardial infarction (MI), and stroke (Stadel, 1986~. However, the risk of serious illness or death from cardiovascular disease that can be attributed to OC use is apparently concentrated among certain groups of women, primarily older women and women who smoke cigarettes. At least 11 case-control and 4 cohort studies have found that OC use increases the risk of venous thromboembolism (Vessey, 1980~. Results from those studies have shown that current OC use increases the risk of venous thromboembolism, although the increased risk does not appear to persist among past users. Further- more, the risk among current users remains constant with increasing duration of OC use. The risk of both superficial and deep vein thrombosis among current OC users is directly related to the estrogen content of OCs: the higher the estrogen content of the OC, the greater the risk of venous thromboembolism (Stadel, 1986~. The pathogenesis of venous thromboembolism among OC users probably involves an increase in the size of intravascular clots formed in response to thrombotic stimuli, most likely a result of estrogen-induced decreases in an- tithrombin III and plasminogen activators. Unlike the associations between OC use and MI and stroke, available studies have not found any interrelationship between OCs, venous thromboembolism, and cigarette smoking. The increased risk of venous thromboembolism is an important source of illness attributable to OC use but is a very infrequent cause of mortality (Stadel, 1986~. In contrast to the low attributable risk of death from venous thromboembolism associated with OC use, the increased risk of MI and stroke observed in women currently using OCs has been demonstrated to be an important source of the mortality risk attributable to OCs (Stadel, 1986~. Current OC use increases the risk of MI, thrombotic stroke, and hemorrhagic stroke. The risk of MI and stroke associated with current OC use is strongly influenced by age and by the presence of other cardiovascular risk factors, such as cigarette smoking, hypertension, and diabetes. For example, the risk of MI that is attributable to OCs among nonsmok- ing women 30 to 39 years of age is about 4 cases per 100,000 current users per year, but it increases to about 185 cases per 100,000 current users per year among women aged 40 to 44 years who smoke heavily (Table 3~. The risk attributable to past OC use appears also to be concentrated among older women and older women who smoke heavily. The risk of adverse cardiovascular events among current OC users appears to be directly related to the estrogen content of the OCs; although less conclusive, risk may also be related to progestin content. The pathogenesis of MI and stroke among current OC users may be related both to the
56 [ARE, PETERSON, AND CHU TABLE 3 Current Use of Oral Contraceptives (OCs), Cigarette Smoking, and Risk of Myocardial farcuon (MI) MIs per 100,000 Current OC Users MIs per 100,000 per Year Cigarettes Women per Year Relative Attributable Age per say OC Users Nonusers Riska Risk 30-39 yr All women 11 4 3 7 0-14 6 2 3 4 >15 30 1 1 3 19 40-44 yr All women 89 22 4 67 0-14 47 12 4 35 >15 246 61 4 185 aRelauve risk of ~ for OC users compared with nonusers. Source: Modified from Stadel, 1986, p. 17. intravascular coagulation system and to the effects of increased blood pressure and metabolic changes. Current OC use has been found to elevate blood pressure slightly in most women about 1 to 2 mm Hg diastolic and 5 mm Hg systolic (Stadel, 1986~. OC use leads to approximately a threefold to sixfold increased risk of overt hyperten- sion. This risk has been observed to increase with increasing age and with increasing duration of OC use. Whether other risk factors for hypertension may be related to the increased risk attributable to OC use has not be established. Metabolic Effects The progestin component of OCs has been found to decrease the concentration of high-density lipoprotein-cholesterol (HDL-C), whereas the estrogen compo- nent has been found to increase HDL-C concentration (Stadel, 1986~. lIence, the effects of different OC formulations on HDL-C concentration apparently depend on the specific estrogen-progestin content. A U.S. study provided information about 10 combination OCs: 3 lowered HDL-C concentrations, 2 had no effect, and 5 increased HDL-C levels (Bradley et al., 1978~. If the progestin component has a strong anti-estrogen effect, the tendency of the estrogen component to increase HDL-C concentration may be overpowered (Stadel, 1986~. Current OC use has been found to decrease glucose tolerance among most women, although this decrease appears to be small, unrelated to duration of use,
HEALTH EFFECTS OF CONTRACTION 57 and only additive to the effects of other risk factors for impaired glucose tolerance (Stadel, 1986~. This decrease in glucose tolerance is directly related to the estrogen content of the OCs, although there may be a relationship to the progestin content also. More important, OCs containing relatively small amounts of estro- gen-~ 50 micrograms of ethinyl estradiol-do not appear to decrease glucose tolerance to any appreciable extent. Neoplastic Diseases Epidemiologic studies clearly indicate that OC use increases the risk of hepato- cellular adenoma (HCA), a rare, benign neoplasm of the liver. Although benign, HCA can cause serious abdominal hemorrhage and death, with a death-to-case ratio of approximately ~ percent (Rooks et al., 1979~. Among women of repro- ductive age who have never used OCs or who used them for a short time, HCA develops at an annual rate of about 1.0 to 1.3 per million women 16 to 44 years of age. However, OC use is a strong risk factor for HCA, with a relative risk greater than 100 among women who have used OCs for 3 or more years compared with women who have used OCs for 1 year or less. The increased risk appears to be directly related to the duration of use, the age of the user, and the estrogen content of the OC. However, the absolute risk of HCA associated with OC use is small because of the rarity of the tumor. Among women who have used OCs for 5 years or longer, the attributable risk is estimated to be about 2 cases of HCA per 100,000 users per year (Stadel, 1986~. Epidemiologic studies have well demonstrated that OC use protects women from developing endometrial and ovarian cancers. However, the effect OCs may have on the risk for developing certain other malignancies remains unclear. Hepatocellular carcinoma and malignant melanoma have both been associated with OC use, although the strength of the associations has not been great. For this reason as well as the fact that those tumors are quite rare, the public health impact of a true positive association would not be great for either type of malignancy. Conversely, the debate about whether OC use increases the risk of cervical and breast cancers remains heated; some studies have found no effect on cancer risk, while others have found disturbing increases in risk. Because breast and cervical cancers are two of the most common cancers affecting women, contraceptive providers and epidemiologists feel an urgent need to resolve these discrepant results. However, as will be discussed subsequently, the possibility of a quick resolution to the controversies is unlikely. Both case-control and cohort studies have assessed the relationship between OC use and malignant melanoma. Generally, they have not found substantial increases in the risk of melanoma associated with OC use (Stadel, 1986~. Certain studies do suggest that O{ s may increase the risk among certain subgroups of women, especially those who have used OCs for a long time (Ramcharan et al., 1981; Bain et al., 1982; Holly et al., 1983; Beral et al., 1984~. Few studies have
58 l:FE, PETERSON, AND CIlU adequately addressed the risk by histologic type of melanoma. Future studies of this association will have 'several issues to consider, including the rarity of the tumor among women, the different histologic subtypes of melanoma, and the potentially confounding effects of exposure to sunlight. The association between OC use and benign liver tumors, as well as a number of case reports of liver cancer among OC users, have led to theoretical concerns that OC use might increase the risk of malignant liver tumors. Three case-control studies published since 1983 have found increased risks of hepatocellular carci- noma among OC users (Henderson et al., 1983; Forman et al., 1986; Neuberger et al., 1986~. Generally, the increased risk has been confined to women with a history of long-term OC use. However, each of the studies had few women in the case group ~30) and had methodological problems that may have biased the results. In developed countries hepatocellular carcinoma is extremely rare among reproductive-aged women. In the United States in 1982 only 59 women died from liver cancer among approximately 52 million women aged 15 to 44 years (Na- hona1 Center for Health Statistics, 1982~. Hence, even if OC use substantially increases the relative risk of liver cancer, the attributable risk would still be very low. In many developing countries -liver cancer is a much more common problem, primarily because of the relationship between hepatocellular carcinoma and chronic hepatitis B virus infection, which has a high prevalence in some regions. In those areas the possible interrelationships between OC use, hepatitis B infection, and liver cancer are more troublesome. Currently, the World Health Organization (WHO) is conducting a multicenter case-control study to address the relationship between OC use and liver cancer. Data are being collected from three developing countries with high rates of hepatitis B infection and liver cancer. It is hoped that results from this study will shed light on the relationship between OC use and this · ~ . serious malignancy. The potentially positive association between OC use and cervical cancer has added importance when considered in the setting of less developed countries. Surveillance information from developing countries, although sometimes frag- mentary and incomplete, suggests that cancer of the cervix is the most frequent malignancy among women in those countries (Lunt, 1984~. Unfortunately, screen- ing efforts in those countries usually reach very limited segments of the female population. Unlike the situation in developed countries where provision of contraceptive services is usually accompanied by routine Papanicolaou (Pap) screening for cervical cancer, family planning programs in developing countries often do not have the resources to provide Pap screening for their clients. To date, no definite causal relationship has been established between OC use and cervical cancer. Of 15 major epidemiologic studies, ~ have found no in- creased risk of cervical neoplasia and 7 have found significantly increased risks overall or increases among certain subgroups of users (Piper, 1985; Brinton et al.,
HEALTH EFFECTS OF CONTRACEPTION 59
60 ~ FE, PETERSON, AND CHU studies of the OC-breast cancer relationship were not large enough to examine the relationship according to specific OC formulations and among certain high-risk subgroups of women. To date, the CASH Study is the largest study designed to assess the risk of breast cancer associated with OC use (CASH, 1986~. The study was a population- based case-control study conducted in eight regions of the United States from 1980 to 1982. Detailed information about reproductive history, contraceptive use, family history of cancer, and personal characteristics and habits was col- lected from 4,711 women with newly diagnosed breast cancer and from 4,676 control women. Case and control women were from 20 to 54 years of age. Compared with women who had never used OCs, women who had used OCs had a relative risk of breast cancer of 1.0 (0.9-1.1~; even women who had used OCs for 15 or more years had no increased risk. None of the 12 OC formulations most commonly used in the United States were associated with a statistically signifi- cant increased risk of breast cancer. Analyses also showed no increase in risk among high-risk subgroups of women, including women with a history of BED or a family history of breast cancer or those who were nulliparous or older at first term pregnancy. Nonetheless, despite many studies indicating that OC use apparently does not increase the overall risk of breast cancer, controversy remains concerning whether long-term OC use, use at an early age, or use before the first term pregnancy might increase breast cancer risk (Skegg, 1988~. Several recent studies have found increases in risk among certain subgroups of women. An analysis pub- lished in 1983 suggested that women who used certain OC formulations classified as "high progestin" before the age of 25 appeared more likely to develop pre- menopausal breast cancer than women who did not use OCs before age 25 (Pike et al., 1983~. In the same year a report from England suggested that women with long-term OC use before their first child was born had an increased risk of premenopausal breast cancer compared with women who did not use OCs before the birth of their first child (McPherson et al., 1983~. A 1985 report described results from an analysis using a subset of data from the CASH Study that replicated the analyses from the two 1983 reports (Stadel et al., 1985~. In this analysis 2,088 women who had breast cancer and were from 20 to 44 years old were compared with 2,065 control women in the same age group. No significant increase or decrease in risk was observed among women who used high-progestin OCs before age 25, even when duration of use exceeded 6 years. Similarly, women who used OCs before their first term pregnancy did not have an increased risk, even when duration of use exceeded 4 years. In 1986 two additional conflicting reports were published (Meirek et al., 1986; Paul et al., 1986~. Analyses from a case-control study conducted in Scandinavia found a risk of premenopausal breast cancer among women with long-term OC use and use before the first term pregnancy (Meirek et al., 1986~. The investiga- tors reported a twofold increase in risk of breast cancer among women who used
HEALTH EFFECTS OF CONTRACTION 6 ~ OCs for as long as 12 years. In contrast; analyses from a New Zealand case- control study did not find elevations in risk associated with OC use, even among women who had used OCs for a long time and who started taking them at an early age or among women who used OCs before their first term pregnancy (Paul et al., 1986~. A recently published analysis from the CASH Study examined in detail the risk of breast cancer in relation to early use of OCs (Schlesselman et al., l9X8~. The study found no evidence that use of OCs in the distant past increased breast cancer risk though age 54. Parous women who had used OCs for more than 6 years before their first term pregnancy, at 1~14 years after that pregnancy, were found to have a risk of breast cancer relative to nonusers of 1.1 (0.3-3.9~. Among nulliparous women with more than 6 years of OC use, the relative risk of breast cancer at 1~14 years after they last used OCs was 0.6 (0.1-3.7~. Epidemiologists continue trying to disentangle the reasons for the discrepan- cies among the published studies (McPherson et al., 1986; McPherson and Drife, 1986; Anonymous, 1986; Schlesselman et al., 1987; Skegg, 1988~. McPherson et al. (1986) have suggested that any possible risk of breast cancer associated with OC use at early ages may not become apparent for at least 20 years. Women born in the 1940s who used OCs in the 1960s will reach the ages of highest risk for breast cancer by the year 2000. If OC use at young ages is a risk factor for breast cancer, and if at least 20 years is required between exposure to OCs and the diagnosis of breast cancer, then these women might be expected to be diagnosed with breast cancer some time during the next 20 years. Under these assumptions, researchers may not be able to detect such a relationship at the present time. The conflicting reports about the possible relationship between OC use and breast cancer risk, particularly use atom early age and before first term pregnancy, are confusing and troublesome. However, the preponderance of epidemiologic studies suggests that, overall, OCs do not increase the risk of breast cancer. If they do increase the risk in certain subgroups of women, the increase is not great and is generally confined to long-term users. Most experts believe that current recommendations for OC use should not be changed because of the continuing OC-breast cancer controversy. As women who used OCs in the 1960s reach the ages of highest risk for breast cancer, investigators should vigorously pursue opportunities to continue studying the long-term effects of OC use on breast cancer risk. Other Effects Several epidemiologic studies published in the 1970s reported that using OCs increased a wombats risk of gallbadder disease. Results from two large cohort studies from the United Kingdom published in 1982 better clarified the relation- ship between OCs and gallbladder disease (Layde et al., 1982; RCGP, 1982~. Each study found that the increased risk appeared to be confined to recent users
62 "E, PETERSON, AND CHU and to short-term users. No differences in risk were found with differing estrogen or progestin doses. The authors of both studies concluded that using OCs apparently accelerated the development of gallbladder disease among susceptible women, rather than increased the overall lifetime risk. A twofold to threefold increase in the prevalence of chlamydia trachomatis infection of the cervix among OC users has been demonstrated in 12 epidemiol- ogic studies (Washington etal., 1985~. Three reasons for this increase in risk have been suggested: (1) greater sexual activity among OC users than nonusers, (2) enhanced detection of chlamydial infection because of increased cervical ectro- pion induced by OC use, and (3) a direct effect of OC use on the risk of chlamydial infection. Three of the epidemiologic studies measured sexual activ- ity and found no increase among OC users; furthermore, several epidemiologic studies have not found an increased risk of gonococcal infection among OC users (Washington et al., 1985~. Whether or not it is merely easier to detect chlamydia infection among women who use OCs cannot be easily assessed and remains unresolved. INTRAUTERINE DEVICES The first IUDs that were widely used, such as the Lippes Loops and Saf-T- Coil7M, were made of inert plastic. In the 1970s a second generation of IUDs was introduced in which the plastic IUD acted as a carrier for metal or hormonal substances (Piotrow et al., 1979~. These medicated devices included the Copper- 7, Copper-T series, and the Progestasert7M. More recently a new and improved copper IUD, the Copper-T 380A, was developed; it may soon be the major IUD available in most countries (Treiman and Liskin, 1988~. The IUD is a highly effective contraceptive, with method failure rates of about 1 percent per year for medicated IUD s and 2 percent per year for nonmedicated ones (Table 2~. Because of undetected IUD expulsion, the user failure rate for IUDs is estimated to be somewhat higher, about 6 percent per year. Because IUDs apparently prevent bow intrauterine and ectopic pregnancies, the overall risk of ectopic pregnancy among IUD users is probably decreased compared with women who use no contraception. A 1981 study reported that current IUD users had a relative risk of ectopic pregnancy of 0.4 (0.3~.6) compared with women using no contraception (Ory, 1981~. However, about 5 percent to 15 percent of IUD-associated pregnancies are ectopic as IUDs seem to be more effective at preventing intrauterine pregnancies. No major noncontraceptive health benefits are linked to IUD use. Progester- one-elaborating IUDs tend to decrease menstrual blood loss and dysmenorrhea, which can be viewed as a benefit for those women in whom it occurs (Hatcher et al., 1988).
IlEALTH EFFECTS OF CONTRACTION 63 Adverse Health Effects Four major health risks have been associated with IUD use: (1) spontaneous abortion, which may rarely progress to septic abortion; (2) uterine perforation; (3) PID; and (4) tubal infertility. Based on information from studies done primarily in the United States and other developed countries, estimates for the annual mortality risk attributable to IUD use are between 1 and 2 deaths per 100,000 women (Ory et al., 1983~. Mortality rates from IIJD use may be somewhat higher in developing countries because of delays in treating complications or lack of access to medical facilities. Although pregnancy rates are low among IUD users, complications may occur in women who do become pregnant. If the IUD is left in place, the chance of spontaneous abortion is 50 percent. If the IUD is removed, this rate drops to 25 percent (Hatcher et al., 1988~. Septic abortion is a rare but sometimes fatal complication (10 per 100,000) that may occur if the IUD is left in place after the first trimester of pregnancy (Ory et al., 1983~. If the IUD is removed as soon as the pregnancy is diagnosed, most instances of septic abortion can be prevented. Perforation of the uterus may occur when an IUD is inserted, although this injury is often undetected and usually not serious. The incidence of perforation is unknown but is probably less than 1 percent (Hatcher et al., 1988~. Very rarely the IUD is extruded into the abdominal cavity, requiring laparotomy to remove it. The risk of perforation during insertion increases substantially in two subgroups of women: lactating women and women within ~ weeks of delivery, because they have softer uterine musculature; at these times the risk of perforation may be increased substantially (Heart~vell and Schlesselman, 1983~. Most epidemiologic studies of the issue have found that IUD users have an increased risk of PID (Grimes, 1987~. This finding contrasts with all other modern methods of temporary contraception, which seem to protect a user from developing PID. PID is usually but not always the result of an STD, such as gonorrhea or chlamydial infection. The overall incidence of PID vanes greatly in different populations and depends mainly on sexual behaviors. The incidence of PID among IUD users also varies substantially. In a Scandinavian study of women with gonorrhea, 24 percent of IUD users developed PID (Ryden et al., 1979~. Results from a large cohort study of married women in the United Kingdom showed a PID rate of 0.15 per 100 woman-years (Vessey et al., 1981~. Clinical trials that determined rates of IUD removal because of PID found rates about eight times greater than the study from the United Kingdom, up to 1.22 per 100 woman-years (Siven and Stern, 1979~. The Dalkon ShieldlM, an IUD used in the 1970s but no longer available, has been associated with a high risk of PID in several studies (Grimes, 1987~. Women using IUD types other than the Dalkon Shield have been found to have about 1.5
64 BEE, PETERSON, AND CHU to 2.0 times greater risk of PID than women using no contraceptive method. Most studies that have looked at the relationship of PID risk to the timing of IUD insertion have found that much of the increased risk is probably confined to the fast few months after insertion. A recently published analysis has found that IUD users who are in mutually monogamous sexual relationships probably have little increased risk of PID associated with their IUD use (Lee et al., 1988~. Much of the increased risk of PID seen among IUD users may be confined to women who are at increased risk for developing STDs. One of the most important and serious sequellae of PID is tubal infertility. Results from a large cohort study of women with surgically confirmed PID have demonstrated that these women have an increased risk of subsequent tubal infer- tility Angstrom, 1987~. The data showed that, among the women with PID who were subsequently exposed to a chance of pregnancy, infertility occurred in 11 percent after one episode of PID, 23 percent after two episodes, and 54 percent after three episodes. Until recently little epidemiologic evidence directly linked fUD use with infertility. Results from most clinical trials of IUDs found that more than 70 percent of women conceive within 12 months of IUD removal (Liskin and Fox, 1982). A large cohort study from the United Kingdom found that within 2 years of removal 92 percent of former users had given birth (Vessey et al., 1978~. How- ever, two U.S. case-control studies reported that the risk of tubal infertility among nulliparous women who had ever used IUDs was about twice that of those who had never used them (Daring et al., 1985; Cramer et al., 1985~. Both of these case-control studies found different risks of tubal infertility associated with different types of IUDs (Daring et al., 1985; Cramer et al., 1985~. Women who had used the Dalkon Shield had the highest risk of tubal infertility. The lowest risk was seen among women who had used copper IUDs, while women who used the Lippes Loop or Saf-T-Coil IUDs had intermediate risks. Although these findings were consistent between the two studies, the numbers of users of specific IUD types were small, and the differences may have been due to chance alone. The incidence of IUD-associated infertility is unknown, although it is probably less than the incidence of IUD-associated PID. The increased risk of infertility is presumably related to the increased PID risk associated with IUD use, even if PID is never recognized clinically. Cramer et al. (1985) found that women who reported having only one sexual partner had no increased risk of tubal infertility associated with IUD use. BARRIER METHODS Because of the potential for preventing transmission of STDs, such as acquired immune deficiency syndrome, researchers are focusing new attention on barrier
HEALTlI EFFECTS OF CO=RACE~ION 65 methods of contraception-condoms, diaphragms, spermicides, and sponges. Generally, barrier methods are less effective at preventing pregnancy than are OCs, IUDs, sterilization, injectables, and implants. Further, the effectiveness of barrier methods is extremely user dependent; improper use by unmotivated users can result in high failure rates. Side effects are minimal; consequently, the main risks associated with barrier methods are complications from unintentional preg- nancy (Ory et Al., 1983~. Discontinuation of use of barrier methods is common, both because they have higher risks of failure and because they may be obtrusive and less convenient to use (Sherns et al., 1984~. But an important advantage of barrier methods is their protection against venous STDs. Barner methods also protect against acute PID and the resulting tubal infertility by keeping organisms that cause PID-such as Neisseria gonorrhocae and Chlamydia trachomatis from infecting the lower genital tract and from ascending into the upper genital tract (Kelaghan et al., 1982; Cramer et al., 19X7~. Condoms Condoms are a safe, reversible method of birth control increasingly being used to prevent both pregnancy and STDs. The estimated method failure rate for condom use (without spermicide3 is 2 percent per year, while the user failure rate is 12 percent (Table 2~. The effectiveness of condoms as both a contraceptive and a disease prophylactic depends on consistent and proper use. If couples begin to rely solely on condoms for protection against both pregnancy and infection, they could experience a substantial increase in the number of accidental pregnancies from incorrect or inconsistent use (Trussell and Kost, 19873. As condom use increases, acceptability and compliance will be important to monitor. In vitro studies have demonstrated that latex condoms are effective barriers to herpes simplex virus type 2 (Conant et al., 1984; Judson et al., in press), Ch- lamydia trachomatis (Judson et al., in press), cytomegalovirus (Katznelson et al., 1984), and human immunodeficiency virus (HIV) (Conant et al., 1986; Rietmeijer et al., 1988~. Natural membrane condoms may give less protection against viral STDs than latex ones, which contain smaller pores (Centers for Disease Control, 1988~. Because using condoms prevents semen deposition, their use should also reduce transmission of organisms that may be present in semen, such as Neisseria gonorrhoeae, hepatitis B virus, Trichomonas vaginalis, and HIV (Stone et al., 1986~. Epidemiologic data relating to condom use and the prevention of STDs are limited. Several studies have shown that condom users, their partners, or both have a lower frequency of gonorrhea (Barlow, 1977; Hooper et al., 1978~. Pre- liminary epidemiologic data suggest that using condoms may protect against sexual transmission of HIV infection (Fischl et al., 1987; Centers for Disease Control, 1987~. Recent studies of prostitutes have also suggested a protective
66 ~E, PETERSON, AND ClIU relationship between condom use and HIV infection (Smith and Smith, 1986; Centers for Disease Control, 1987; Mann et al., 1987~. However, these studies were based on observational data, and condom users are likely to differ from nonusers in important characteristics that may be related to the risk of acquisition of STDs (Feldblum and Fortney, 1988~. Although conclusive epidemiologic evidence is lacking, available data suggest that the use of condoms is an effective preventive measure against STDs (Horsburgh et al., 1987~. Failure of condoms to protect against STDs is probably explained by user failure more often than by product failure (Centers for Disease Control, 1988~. The use of Spermicides is considered a useful adjunct to condoms because spermi- cides can inhibit in vitro the growth of a broad range of sexually transmitted pathogens (Cramer et al., 1987; Francis and Chin, 1987~. Spermicides and the Contraceptive Sponge Spermicides are chemical agents that inactivate sperm in the vagina before the sperm can move into the upper genital tract. Spermicides in use today include nonoxynol-9, octoxynol-9, and menfegol, although the latter is not available in the United States. Spermicides containing mercuric compounds, rejected as unsafe by the U.S. Food and Drug Administration (FDA), are still marketed in some Latin American countries and elsewhere (Sherris et ad., 1984~. Inert carriers of the spermicide include foams, jellies, creams, and suppositories; newer carriers include foaming suppositories (Neo Samnoon~) and snermicide-imor~n~tPa sponges (TodaylM). ~r ~ ~--I- _= _w _ The type of spermicidal agent or carrier has little effect on failure rates. The method failure rate is estimated to be about 3 percent per year; however, the user failure rate, at 21 percent per year, is much higher (Table 2), largely because use must be premeditated. Furthermore, because sperrnicides are messy and can be irritating to mucous membranes, women often stop using them (Sherris et al., 1984~. Protection from pregnancy, however, can be increased considerably if Spermicides are used in conjunction with other barrier methods such as dia- phragms or condoms. Among nulliparous women the failure rates for the contraceptive sponge are comparable to the failure rates for the diaphragm; the method failure rate is 5 percent per year and the user failure rate is 18 percent per year (Table 2~. However, among parous women the method failure rate is ~ percent and the user failure rate is 28 percent. Both laboratory evidence and clinical evidence suggest that Spermicides pro- tect against various STDs (Stone et al., 1986~. In vitro studies have shown that Spermicides inhibit the growth of Neisseria gonorrhocae (Singh et al., 1972; Cowan and Cree, 1973) and inactivate heIpes simplex virus type 2 (Singh et al., 1976) and HIV (Hicks et al., 1985~. Although epidemiologic evidence is sparse, the virucidal effects of Spermicides may provide some protection against cervical
HEALTH EFFECTS OF CO=RACE~ION 67 cancer, since the cancer may be initiated or promoted by a sexually transmitted virus such as human papilloma virus (Spring and Gruber, 1985~. Several epi- demiologic studies have found that spermicides protect users from gonorrhea infections Hick et al., 1982; Quinn and O'Reilly, 1985), especially when used with condoms or diaphragms (Austin et al., 1984~. A clinical trial among high- risk women using contraceptive sponges impregnated with nonoxynol-9 demon- strated protection against chlarnydial and gonorrhea! infections (Rosenberg et al., 1987~. One epidemiologic study suggested a connection between spermicides used near the time of conception or during pregnancy and the risk of congenital defects (Jick et al., 1981~. Since that report, several larger and better designed studies have failed to confirm the association (Hugging et al., 1982; Mills et al., 1982; Shapiro et al., 1982; Cordero and Layde, 1983~. Additional health risks have been associated with the contraceptive sponge, a newer type of sperrnicidal carrier. Sponge users may be at increased risk of vaginal candidiasis, possibly because nonoxynol-9 allows candidal overgrowth (Rosenberg et al., 1987~. Of greater health consequence is a possible association between using the contraceptive sponge and developing toxic shock syndrome (TSS); sponge users have a relative risk of 10.5 (2.1-52.7) compared with women using no barrier method (Schwartz et al., 19893. Because TSS is an extremely rare disease, the absolute risk of sponge users developing TSS is quite small; however, because TSS is a serious illness, physicians and sponge users should be aware of the symptoms and alert to the possibility (Faich et al., 1986; Schwartz et al., 1989~. Postpartum women and women who have had TSS should not use the sponge, and women should never leave the sponge in the vagina for more than 30 hours (Reingold, 1986~. Diaphragm With Spermicide Diaphragms used with spermicidal cream or jelly are a safe method of contra- ception and, when used correctly and consistently, can be an effective contracep- tive. Because diaphragms/spermicides are inconvenient to use, only highly moti- vated women will experience low failure rates. Although the method failure rate is estimated to be about 3 percent per year, the user failure rate is much higher, about 18 percent per year (Table 2~. Clinicians should consider how motivated their patients are before advising diaphragm use (Sherris et al., 1984~. As with other barrier methods, diaphragm/spermicide use may have several noncontraceptive health benefits. This method appears to protect against gonor- rhea, PID, and tubal infertility Dick et al., 1982; Kelaghan et al., 1982; Cramer et al., 1987~. Several case-control studies found cervical dysplasia and cervical neoplasia less common among diaphragm users than among other women (Wright et al., 1978; Harris et al., 1980; Celentano et al., 1987~. Because diaphragms are almost always used with spermicides and spermicides alone may protect against
68 LEE, PETERSON, AND CHU various STDs, the separate protective effects of diaphragms on specific conditions may be difficult to measure (Stone et al., 1986~. As with the contraceptive sponge, diaphragms have been associated with a significantly higher risk of TSS (Hymowitz, 1981; Baehler et al., 1982; Schwartz et al., 1989~. In one case-control study diaphragm users had a relative risk of TSS of 1 1.7 (2.5-56.1) compared with nonusers; however, the estimated absolute risk of TSS associated with diaphragm use was low, about 2.25 cases per 100,000 users per year (Schwartz et al., 1989~. A less serious but more frequent complica- tion associated with diaphragm use is urinary tract infections; the relative risks for diaphragm users developing a urinary tract infection compared with nonusers ranges from 2.0 to 3.0 (Foxman and Frerichs, 1985; Fihn et al., 1985; Vessey et al., 1987~. Better fitted diaphragms and urination after intercourse may reduce the risk of urinary tract infections (Foxman and Frerichs, 1985~. Women at particular risk for urinary tract infections, such as those with recurrent infections, should probably consider other contraceptive methods. Cervical Caps Cervical caps are cup-shaped devices, held in place by suction, that fit over the cervix (Hatcher et al., 1988~. When used with spermicides, their effectiveness is comparable to that of diaphragms, with a method failure rate of about 5 percent per year and a user failure rate of 18 percent per year (Table 21. Cervical caps have not been widely used; hence, little published research is available. Reports from small studies suggest that about half the women who want to use the cap cannot be fitted. Many other women discontinue use because of odor problems, difficulty in insertion or removal, or dislodgement during intercourse (Ory et al., 1983~. Advantages are that the cap can be left in place longer than 24 hours and that it can be inserted many hours or even a day or two before coitus (Sherns et al., 1984~. Although potential health risks and benefits associated with the cervical cap are probably similiar to those associated with the diaphragm, available data are too limited to document them (Hatcher et al., 1988~. LONG-ACTING METHODS Several long-acting contraceptive methods now available are highly effective, convenient to use, and give protection from pregnancy for 1 month to 5 years. All contain some kind of progestin. Disturbance of the menstrual cycle is the most common side effect of all progestin-containing contraceptives and is the major reason women stop using them. Injectables Two long-acting injectable progestins~epot-medroxprogesterone acetate (DMPA; Depo-Provera~) and norethindrone enanthate (NET; Noristerat~)
IlEALTI-] EFFE=5 OF CONTRA CE"ION 69 are the most widely used and studied of several injectable contraceptives avail- able. DORA is approved for contraception in at least 90 countries, NET in more than 40 Kristin et al., 1987~. At present neither is licensed for use as a contracep- tive in the United States. Injectables are highly effective: the estimated method failure rate is 0.3 percent per year for DMPA, given every 90 days, and 0.4 percent per year for NET, given at 8- or 12-week intervals (Table 2~. Because minimum compliance is necessary, the user failure rates are approximately equal to the method failure rates. Injectables prevent pregnancy in several ways. Possible mechanisms include inhibiting ovulation; thickening cervical mucus; and altering the en- dometrial lining, which inhibits implantation (Liskin and Quillin, 1982~. Considerable controversy regarding possible cancer risks has been associated with DMPA and NET, chiefly stemming from data from animal studies suggest- ing that DORA use could increase the risk of breast and endometrial cancers (WHO, 1986a). Although the applicability of these animal models is debatable and human studies have either failed to demonstrate these relationships or are inconclusive (Liang et al., 1983; WHO, 1986a; Lee et al., 1987), the possibility of increased risk of cancer among DMPA users was one reason the FDA did not approve DMPA for contraceptive use in the United States (Sun, 1984~. In 1986 preliminary results were published from an ongoing case-control study conducted by WHO concerning the relationship between DMPA and breast cancer (WHO, 1986a). This is the largest epidemiologic study of the DMPA-breast cancer relationship published to date. Women who had ever used DMPA had a risk of breast cancer of 1.0 (0.7-1.5~; no increase in risk was seen even for long- term users. Conversely, a recent study in Costa Rica found that DMPA users had an elevated risk of breast cancer of 2.6 (1.4~.7~; however, no dose-response effect was found (Lee et al., 1987~. Because of the small number of DMPA users in the study and the lack of a dose-response relationship, the authors considered the results to be inclusive. Another analysis from the same WHO study found that the risk of invasive cervical cancer was slightly higher for long-term DMPA users than for never- users (relative risk of 1.4 [0.9-2.21~; however, results were considered inconclu- sive because of the small number of long-term DMPA users as well as the lack of a consistent trend with duration of use (WHO, 1986a). Results from the Costa Rica study showed no increase in risk of either invasive cervical cancer or carcinoma in situ associated with DMPA use (Oberle et al., 1988~. Although DMPA and NET have yet to be conclusively linked to any human cancer, epidemiologic studies have been hindered by small sample sizes and short durations of exposure. Whether injectable contraceptives are associated with any cancer will be resolved only after additional data have accumulated (WHO, 1986a). Reported metabolic effects of DMPA and NET use include changes in blood pressure as well as changes in insulin, cholesterol, and triglyceride levels; how- ever, the findings are inconsistent and have shown no clear clinical significance
70 LEE, PETERSON, AND CHU (WHO, 1986b; Liskin et al., 1987~. Concern over cholesterol levels among women who use DMPA or NET is mainly due to the reported cardiovascular problems associated with OC use, generally thought to be caused by the estrogen component of the pill (Rosenfield et al., 1983~. Lower levels of HDL-C have been reported among women who were taking OCs with a high progestin potency or a progestin alone (Bradley et al., 1978~. Most studies of DMPA users have found either no change in total cholesterol and triglycerides or a decrease, a possibly beneficial effect (Liskin et al., 1987~. However, a few studies have reported an increase in cholesterol with longer use of DMPA or a decrease in HDL-C, both possibly adverse effects. The only study involving NET also found a decrease in HDL-C (Fotherby et al., 1982~. The clinical significance of these decreases in HDL-C is uncertain; to date, no associations between DMPA or NET use and increased incidence of cardiovascular disease have been reported. Trials are under way in five countries to measure changes in lipid metabolism in DMPA and NET users (Liskin et al., 1987~. Unlike combination OCs, DMPA and NET appear to have little effect on the coagulation system (Liskin et al., 1987~. Most laboratory studies have found little change in the coagulation and fibrinolytic systems that affect blood clotting. The most consistently reported side effect of DMPA and NET use is a change in menstrual pattem: more than two-thirds of DMPA users and one-half of NET users have no regular menstrual cycles during the first year they use these contraceptives (Liskin et al., 1987~. Amenorrhea or irregular, unpredictable bleeding episodes are the most common problems (WHO, 1978, 1987) and the principal reasons that women stop using injectables. Amenorrhea, more frequent with DMPA than NET, is probably the most upsetting change because women cannot be sure they are not pregnant (Liskin et al., 1987~. Prolonged bleeding or spotting can be disturbing, especially if local customs restrict the activities of menstruating women. The occurrence of heavy bleeding requiring medical atten- tion is rare, found among about 0.5 percent of users. In fact, because bleeding is often lighter than normal, increased hemoglobin levels among some users have been reported (WHO, 1986b). Studies of children exposed to DMPA from breast milk have found no measur- able consequences, even when followed up to 10 years after exposure (Liskin et al., 1987~. This is not an unexpected finding, for very little of the hormone is transmitted in breast milk. Neither DMPA nor NET appears to have any permanent effect on fertility (Liskin et al., 1987~. However, resumption of ovulation and fertility is delayed; ovulation can be inhibited for 4 to 9 months or more after the last injection (Pardthaisong et al., 1980; Affandi et al., 1987~. Although return to fertility is comparable to that seen with OC and IUD use, because of fears about permanent infertility, some family planning programs use injectables only among higher- parity women (Liskin et al., 1987~. In fact, injectable progestins may protect women from PID, a major cause of infertility, by causing changes in cervical mucus (Gray, 1985~.
HEALTlI EFFECTS OF CONTRA CEPI ION 71 Injectable progestins may also protect against endometrial and ovarian can- cers. Excess estrogen is known to increase the risk of endometnal cancer, while progestins mitigate this effect (Rosenfield et al., 1983~. Further, several studies have demonstrated a negative association between the use of combination OCs and endometnal cancer, a protective effect thought to be caused by the progestin component. Preliminary results from a WHO case-control study found that DMPA users had a relative risk of 0.3 (0.04-2.4) compared with never-users; however, data were based on just 52 women with endometrial cancer, only one of whom had ever used DMPA (WHO, 1986a). Even fewer data exist on the association between ovarian cancer ano EMMA use. However, several studies have demonstrated a negative association between OC use and ovarian cancer, perhaps due to the effects of anovulaiaon (Rosenfield et al., 1983~. Because injectable progestins usually prevent ovulation, there is at least a theoretical possibility that use of DMPA or NET might reduce the risk of ovarian cancer. Preliminary results from the WHO study support this possibility: DMPA users had a relative risk of ovarian cancer of 0.7 (0.3-1.7) compared with never users, based on 7 exposed cases and 74 exposed controls (WHO, 1986a). Several new types of injectables are currently in use or under development. Injectable microspheres and microcapsules release hormone at a more constant rate, and preliminary trials have found fewer side effects than with DMPA or NET (Liskin et al., 1987~. However, as with all progestational agents, irregular men- strual bleeding is a common side effect. Estrogen-progestin combinations that are injected monthly are being used in Latin America and China; preliminary results have shown highly effective protection and fewer problems with bleeding irregu- larities. Implants The NOR PLANT subdermal implant system is another highly effective, long-acting progestational contraceptive method (Liskin et al., 1987~. It is cur- rently being considered by the FDA for use in the United States. Developed by the Population Council, NOR PLAN implants come in two forms. The first and more widely used system, NORPLANT@, consists of six silastic rods, each about 1 inch long and 0.1 inch in diameter, containing 35 milligrams of the progestin levonorgestrel. The newer system, NORPLANT-2~, consists of two rods, each containing 70 milligrams of levonorgestrel. With both systems, rods are im- planted under the skin, usually in the upper arm; the hormone is then released at a continuous, slow rate. Insertion and removal require a minor surgical procedure performed by trained health care personnel; however, once implanted, no further action is required for several years. Both systems provide long-te'.~' protection from pregnancy. NOR PLANT is highly effective for 5 years, NORPLANT-2~ for at least 3 years. In the first year of use the average reported failure rate is 0.2 percent (Table 2~. However, the failure rate may be slightly higher among obese women. The mechanisms for
72 LEE, PETERSON, AND CHU pregnancy prevention are probably similar to those for the injectable contracep- tives. The rate of ec topic pregnancy among NORPLAN~ users, about 1.5 per 1,000 woman-years, is about the same as the rate among women using copper and unmedicated IUDs (Liskin et al., 1987~. Because the rate is lower than that for women using no contraception, NORPLAN~ users can be considered to be protected from ectopic pregnancy. As with injectables, the most common side effect of implants is disturbance of the menstrual cycle. Bleeding irregularity, the most frequent reason that women discontinue use of the method (Sarah et al., 1987; Pasquale et al., 1987), causes 2 percent to 7 percent of women to stop using implants in the first year (Liskin et al., 1987~. However, episodes of abnormal bleeding diminish with duration of use (Sarah et al., 1987; Diaz et al., 1987), and the implants can be removed immedi- ately if there are any disturbing side effects. Transient ovarian cysts occur in a small percentage of women using NORPLAN1~ (Sarah et al., 1987; Diaz et al., 1987~; although the cysts regress eventually, they have been a reason for implant removal (Pasquale et al., 1987~. Permanent infertility does not appear to be a concern; several studies have shown that fertility quickly returns after implants are removed (Diaz et al., 1987; Affandi etal., 1987~. Generally, no changes in several metabolic indicators have been found secon- dary to NORPLAN~ use, including liver function, carbohydrate metabolism, blood coagulation, blood pressure, or body weight (Liskin et al., 1987~. Studies of lipid metabolism in NOR PLANT users have produced conflicting results. How- ever, no epidemiologic studies of cardiovascular complications among NORPLANT~ users have been completed. TUBAL STERILIZATION since the 1930s more than 95 million women worldwide have undergone tubal sterilization, making it the most widely used contraceptive method in the world (Liskin et al., 1985~. The popularity of tubal sterilization continues to grow rapidly in many countries, which suggests that the estimate of the number of women who have had the procedure may be a conservative one. Currently, more than 15 percent of married women of reproductive age have undergone tubal sterilization in countries as diverse as Brazil, Panama, South Korea, Thailand, and the United States. In China an estimated 27 percent of women have undergone tubal sterilization. Tubal sterilization has been very popular in the United States, Latin America, the Caribbean, and in much of Asia. Africa and the Middle East have the lowest prevalence of tubal sterilization. Tubal sterilization procedures can be characterized by (1) timing with respect to pregnancy, (2) the surgical approach to the fallopian tubes, and (3) the method of tubal occlusion. Tubal sterilization can be performed shortly after vaginal
HEALTH EFFECTS OF CO~RACE~ION 73 delivery, concurrently with cesarean section, at the time of induced abortion, or remote from pregnancy termination. In the latter instance the procedure is known as interval tubal sterilization. Surgery for most tubal sterilizations requires an abdominal incision. For either postpartum or interval sterilizations, an abdominal incision 2 to 5 centimeters long, termed a minilaparotomy, allows the surgeon access to the fallopian tubes (Ross et al., 1985~. Laparoscopic tubal sterilization is becoming increasingly popular, particularly in developed countries. For laparoscopic procedures the abdominal incision is usually only about 1 centimeter long. Two general methods of inserting the laparoscope are used. In the first, termed closed laparoscopy, a surgeon uses an in~sufflating needle, a sharp trocar, or both to enter the abdominal cavity before inserting the laparoscope. An alternative technique called open laparoscopy attempts to minimize the risk of intra-abdominal injury caused by the insufflating needle or trocar by combining features of minilaparotomy and closed laparo- scopy. Although open laparoscopy requires more time than closed laparoscopy, it reduces the chance of trauma to major blood vessels and the gastrointestinal tract and thus is inherently safer, especially for facilities that do not have the equipment and personnel necessary to handle serious surgical complications. Alternative surgical approaches include the vaginal approach and the tran- scervical approach. Although the vaginal approach produces no visible scar, it is used less frequently because of the increased risk of pelvic infection (Ross et al., 1985~. A transcervical approach, involving instillation of a sclerosing agent into the fallopian tubes, is widely used only in China. The fallopian tubes can be occluded by ligation (with or without resection), by coagulation using unipolar or bipolar current, or by mechanical occlusion with silastic bands or clips. Tubal occlusions for postpartum laparotomies are gener- ally done by ligation. All of these techniques can be used for interval sterilization, but the methods of occlusion for laparoscopy are limited to the use of coagulation or mechanical devices. Unipolar coagulation involves use of an electrical current that passes into the tube and through the patient to a ground plate attached to the patient. This system, which has resulted in fatal thermal bowel injuries (Peterson et al., 1981a), has largely been replaced in the United States by a bipolar coagula- tion technique that includes only the tissue in the grasping forceps as an integral part of the electrical circuit, thereby reducing the risk of serious thermal injuries (Hulka et al., 1987). With the exception of conventional laparotomy, all surgical approaches and methods of tubal occlusion can be safely and effectively performed using local anesthesia. Because of the hazards inherent in the use of general anesthesia, the World Federation of Health Agencies for the Advancement of Voluntary Surgical Contraception (1984) advocates using local anesthesia for uncomplicated proce- dures. A pregnancy that is identified after sterilization may have occurred because of
74 ~E, PETERSON, AND CHU conception before sterilization or because the operation was not successful. Con- ception before sterilization, called a luteal-phase pregnancy, is estimated to occur in about 2 to 3 per 1,000 interval tubal sterilizations (Chi and Feldblum, 1981~. The most effective means of reducing the risk of luteal-phase pregnancy is to perform the sterilization before the estimated date of ovulation (Grubb and Peter- son, 1985~. Rates of true sterilization failure vary by method of tubal occlusion, surgical expertise, and other factors, such as whether a woman has tubal disease. Overall rates are estimated to be between 0.2 percent and 0.4 percent per year (Table 2~. No studies currently available permit rigorous comparison of failure rates by method of tubal occlusion, though one such study with the potential to do so is in progress in the United States. Since 1978 a multicenter cohort study conducted by the Centers for Disease Control has enrolled approximately 12,000 women under- going tubal sterilization. These women are currently in various stages of a planned 5-year follow-up. Final study results will not, however, be available until the early 1990s. When sterilization fails, the relative likelihood of ectopic gestation appears to be increased. Further, the risk seems to vary by method of tubal occlusion, with highest risk of ectopic pregnancy following electrocoagulation (Bhiwandiwala et al., 19823. The absolute likelihood of ectopic pregnancy after sterilization is actually reduced, however, relative to the risk a woman would incur when not using any contraception or using an IUD (DeStefano et al., 1982~. Studies from around the world suggest that tubal sterilization is a remarkably safe surgical procedure; however, more women die from tubal sterilizations than men die from vasectomies. Case-fatality rates for tubal sterilization have been estimated in reports from Europe, the United States, India, and Bangladesh. A French survey of approximately 100,000 laparoscopic procedures (including those for purposes other than sterilization) estimated a case-fatality rate of 20 per 100,000 procedures (Mintz, 1977~. A prospective study of laparoscopic proce- dures from the United Kingdom reported a case-fatality rate of 10 per 100,000 procedures (Chamberlain and Brown, 1978~. The case-fatality rate of tubal sterilization in U.S. hospitals has been estimated to be approximately 4 per 100,000 procedures (Peterson et al., 1982a). In a report from India the case- fatality rate from postpartum tubal sterilization was 7 per 100,000 procedures (Indian Council of Medical Research, 1982~. In two studies from Bangladesh the case-fatality rate was estimated to be between 12 and 19 per 100,000 procedures (Grimes et al., 1982a, 1982b). Worldwide, most sterilization-attributable deaths are caused by complications related to use of anesthesia, although deaths have occurred from hemorrhage and thermal injury (Peterson et al., 19839. In the United States 11 of 29 women whose deaths were attributable to sterilization suffered complications of general anesthe- sia (Peterson et al., 1983~. Other reports from the United Kingdom and France have also identified complications of general anesthesia as the leading cause of
IlEALTH EFFECTS OF CO=RACE~ION 75 laparoscopic-associated, sterilization-attributable death (Mintz, 1977; Chamber- lain and Brown, 1978~. Even in countries where general anesthesia is not often used, complications of anesthesia are still a major cause of stenlization-attribut- able death (Grimes et al., 1982a; Ross et al., 1985~. Reports regarding nonfatal complications attributable to tubal sterilization have varied by study design, definition, and classification of complications. Although comparing complication rates between studies is difficult, such studies indicate that major morbidity associated with tubal sterilization is uncommon and that safety varies with surgical and anesthetic technique. One large multicenter study, which included data from 23 countries, reported a complication rate of 2.0 percent for the approximately 7,000 women undergoing laparoscopy with silastic band application (Mum ford et al., 1980~. This rate was approximately twice that experienced by the approximately 5,000 women undergoing minilaparotomy procedures (0.8 percent). The WHO has reported the only randomized multicen- ter, multinational study. In that study approximately 800 women had minilapa- rotomies, and approximately 800 underwent laparoscopy using electrocoagula- tion. Major complications occurred in 1.5 percent of women undergoing minila- parotomy and in 0.9 percent of women undergoing laparoscopy. Minor complica- tions were reported in 11.6 percent and 6.0 percent, respectively (WHO, 1982a). The WHO also reported on a multicenter study of postpartum sterilization in which major complications occurred in only 0.3 percent of the patients (WHO, 1982b). This rate is identical to that reported from a large multicenter study of postpartum minilaparotomy conducted in India (Indian Council of Medical Re- search, 1982~. In that study major complication rates associated with minilapa- rotomy did not substantially vary between those minilaparotomy procedures performed postpartum, those performed in association with early pregnancy ter- mination, or those performed as an interval procedure. By contrast, women undergoing interval laparoscopy had a complication rate of 1.2 percent, and women undergoing interval colpotomy (via a vaginal incision) had an overall complication rate of 5.7 percent. In a U.S. multicenter prospective study of laparoscopic tubal sterilization, 1.7 percent of women experienced at least one of six standard intra-operative or postoperative complications (DeStefano et al., 1983a). The most frequent major complication (1.1 percent) was unintended major surgery. In some instances this unintended surgery occurred because of incidental pathology identified during laparoscopy or because of technical limitations of laparoscopy itself. Thus, not all complications were directly caused by the laparoscopic procedure per se. The risk of complications was increased twofold or more by the following factors: obesity, lung disease, diabetes mellitus, previous abdominal or pelvic surgery, and a history of PID. Although the overall major complication rate attribu cable to tubal sterilization is less than 2 percent, such complications can result in serious injury or death. The likelihood of serious injury appears to vary somewhat by surgical approach
76 LEE, PETERSON, AND ClIU and tubal occlusion technique. Most minilaparotomy complications are not serious, with minor hemorrhage, minor wound infections, and uterine perfora- tions reported most frequently (Liskin et al., 1985~. By contrast, laparoscopy is more likely to result in serious complications. While rare, perforation injuries to major blood vessels or the gastrointestinal tract can occur (Peterson et al., 1982b). Open laparoscopy should markedly reduce the risk of these complications. One comparative study assessing complications among 1,112 women undergoing open laparoscopy and 288 women undergoing conventional laparoscopy failed to demonstrate a difference in complication rates (Bhiwandiwala et al., 1985~. However, to detect a difference between open and closed laparoscopy, an epi- demiologic study would have to involve thousands of women to have sufficient study power to detect a difference. Complication rates also may vary with respect to method of tubal occlusion. For example, mesosalpingeal hemorrhage occurs more often from silastic band application than from alternative techniques. Unipolar coagulation increases the risk of thermal bowel injury. Although this type of injury likely occurs in fewer than 1 per 1,000 cases, more than 100 nonfatal thermal bowel injuries and 3 deaths from the use of unipolar instruments have been reported in the United States (Peterson et al., 1981a). Because of the number of women involved, any important long-term sequelae of tubal sterilization would have major impact. On balance, the existing literature has failed to demonstrate any important negative long-term health effects of tubal sterilization. Most concerns regarding potential long-term sequelae have dealt with four issues: (1) menstrual abnormalities, (2) hysterectomy following tubal sterilization, (3) regret, and (4) ectopic pregnancy. We addressed the last concern previously in the section on sterilization efficacy. Before 1980 most studies found an increase in menstrual abnormalities follow- ing tubal sterilization, the so-called post-tubal syndrome. Such studies, however, had major methodological limitations, including: (1) failure to identify and ac- count for biases inherent in their retrospective design, (2) failure to use compari- son groups, (3) failure to consider the effects of prior contraceptive use, (4) failure to consider women without menstrual changes or with improvements in menstrual symptoms, and (5) failure to evaluate information regarding presterilization men- strual patterns. Since 1980 most reports have had fewer methodological limita- tions and have failed to support the existence of a post-tubal syndrome. One study has reported data from a series of approximately 10,000 women undergoing tubal sterilization at 64 institutions in 27 countries (Bhiwandiwala et al., 1983~. Six menstrual symptoms were evaluated before and after sterilization, with most women reporting no changes 1 year after sterilization. When menstrual changes did occur, about as many women experienced improvement in symptoms as experienced deleterious change. A substantial percentage of the observed changes were attributable to cessation of OC or IUD use rather than to the
HEALTH EFFECTS OF CO=RACE~ION 77 sterilization procedure per se. No significant differences in menstrual patterns after sterilization were found among the various tubal occlusion techniques re- viewed. Another report concerned a sample of approximately 1,500 women selected from 45 centers in 24 countries (Fortney et al., 1983~. The most important determinants of menstrual patterns after sterilization were the patterns at the time of sterilization. Women with abnormal patterns before sterilization were three times more likely to experience change than women with normal cycles; many of these women experienced change toward more normal cycles. Several other reports from the United States and the United Kingdom that assessed menstrual patterns 1 or 2 years after sterilization failed to find support for the existence of a post-tubal syndrome. In the United States the Centers for Disease Control used data from a multicenter prospective study in which approxi- mately 2,500 women were followed for 2 years after sterilization (DeStefano et al., 1983b). Each woman served as her own control; her menstrual function at the follow-up interview was compared with her menstrual function at the preopera- tive interview. Except for menstrual pain among women who underwent unipolar electrocoagulation, the prevalence of adverse menstrual function after tubal ster- ilization did not increase. Further, approximately 50 percent of women with adverse menstrual function preoperatively had an improvement by 2 years after tubal sterilization. In a report from England 551 women undergoing interval sterilization were studied (FouLkes and Chamberlain, 1985~. No increase in menstrual abnormali- ties among sterilized women compared with nonsterilized women was found after a year of follow-up. Both sterilized and nonsterilized women were more likely to experience heavy and prolonged menstrual bleeding if they had been using OCs before sterilization. Thus, discontinuation of OCs was more likely to result in abnormalities of menstruation than was sterilization per se. Two studies, one from the United Kingdom (Vessey et al., 1983) and one from the United States (DeStefano et al., 1985), reported on menstrual function among women more than 2 years after their sterilizations. The study from the United Kingdom compared Me experience of 2,243 women undergoing tubal sterilization with 3,551 women whose husbands underwent vasectomy. Three years after sterilization hospital referral or hospitalization for menstrual complaints was nearly equal for women undergoing sterilization and for the comparison Coup. After 6 years, sterilized women had a slight, but not statistically significant, increased risk for hospital referral or admission for menstrual complaints. The study from the United States was principally designed to assess the long- term impact of OC use (DeStefano et al., 1985~. In that analysis 719 women who underwent tubal sterilization were compared with 1,083 women whose partners underwent vasectomy. At follow-up, with intervals longer than 2 years, the tubal sterilization group had a significantly increased risk of abnormal menstrual cycles
78 r FE, PETERSON, AND CHU and combinations of two or more adverse menstrual outcomes. The likelihood of menstrual abnormalities was greatest, however, among the group of sterilized women who had presterilization menstrual complaints. To date, no study has been reported that includes long-term follow-up for sterilized and nonsterilized women in a manner that allows comparison of specific menstrual symptoms and specific methods of tubal occlusion. The Centers for Disease Control in the United States is conducting such a study, but it will not be completed until the early 1990s. Whether tubal sterilization increases the risk of subsequent hysterectomy is uncertain. Concerns that it does have been based, in part, on the possible existence of the post-tubal syndrome we noted earlier. Whether or not a post- tubal syndrome exists, the perception that it does may increase the likelihood that menstrual abnormalities after sterilization are managed surgically. Further, once a woman has been sterilized, either she or her physician may resort more quickly to surgical management of any gynecologic problem. These possibilities were assessed in the previously mentioned cohort study from the United Kingdom (Vessey et al., 1983~. After both 3 years and 6 years of follow-up, hospital referrals leading to hysterectomy were equally likely in women who had under- gone tubal sterilization and in wives of men who had undergone vasectomy. In Canada a population-based study using data from a health insurance plan studied 4,374 women aged 25 to 44 who underwent tubal ligation in 1974 (Cohen, 1987~. As a comparison group, a random sample of 10,000 women registered in the insurance plan in 1974 was chosen. After 2 years of follow-up, no difference was found in adverse gynecologic outcomes between the two groups. However, beginning with the 2-year follow-up and increasing for up to 9 years, women ages 25 to 29 undergoing tubal sterilization had a statistically significant 60 percent increase in the probability of subsequent hysterectomy, even after controlling for other risk factors. For women aged 30 and over, tubal sterilization did not increase the likelihood of subsequent hysterectomy. The results from the two cohort studies with longer-term follow-up argue for continuing surveillance of sterilization sequelae beyond a 1- or 2-year period. Published reports indicate that most women are satisfied with their decision to undergo sterilization. An overview of recent worldwide studies indicates that 2 percent to 13 percent of women interviewed 6 months to 6 years postoperatively expressed regret at having been sterilized; only 1 percent to 3 percent of those women seriously considered or underwent sterilization reversal procedures (Liskin et al., 1985~. However, the prevalence of regret is likely to vary substantially by definition of regret. One survey of U.S. women found that 26 percent responded affirmatively when asked, "If it were possible for you to have another baby, would you, yourself, like to have one?" (Henshaw and Singh, 1986~. In that same survey 10 percent of women responded affirmatively when asked, "As things look to you just now, if the operation could be safely reversed, that is changed back, would you want to have it reversed?"
IIEALTI! EFFECTS OF CONTRA CE~ION 79 Careful presterilization counseling is likely to minimize the chance of postster- ilization regret. To maximize effectiveness, such counseling should focus on the reasons women might regret having been sterilized. A variety of studies have indicated that younger women, usually reported as less than 30 years of age, are more likely to regret having been sterilized (Grubb et al., 1985~. In many developed countries divorce and remarriage are common reasons women regret sterilization and request its reversal. A multicenter prospective study conducted by the Centers for Disease Control in the United States has found that only 2 percent of women regretted having a tubal sterilization 1 year after the procedure, compared with 2.7 percent after 2 years (Grubb et al., 1985~. Five years after the procedure approximately 4 percent regretted it (unpublished data). Because the possibility for major life changes (e.g., death of a spouse or child) increases as women age, the longer the time since the procedure, the more likely women may be to regret having been sterilized. Thus, length of follow-up must be considered when comparing rates of regret in various studies. In another study conducted in the United States, Shain et al. (1986) compared the likelihood that women undergoing tubal sterilization will regret it relative to the likelihood that women whose husbands underwent vasectomy will have regrets. At 1-year follow-up, only 2 of 234 (0.9 percent) women who had undergone tubal ligation compared with 2 of 154 (1.3 percent) women whose husbands underwent vasectomy expressed regret at having had the procedure. As the authors indicated, however, dissatisfaction with sterilization may involve various degrees of intensity, ranging from occasional concerns to desire for reversal. Further, these feelings are not static. When the authors used a broader measure of satisfactionidissatisfaction as an outcome, women who had undergone tubal sterilization expressed significantly more positive feelings during follow-up interviews than did women whose husbands underwent vasectomies. This was attributed to a disproportionate prevalence of female control over reproductive decision making among women undergoing tubal sterilization. The authors concluded that counseling of the couple and the ability to exercise control over one's body are important predictors of sterilization regret. Sterilization Reversal The proportion of sterilized women who seek reversal is small but nontrivial (Ross et al., 1985~. Highly sophisticated microsurgical techniques for tubal reanastomosis that increase the likelihood of successful reversal have been devel- oped. These techniques are generally limited to developed countries When available, such procedures are expensive and require laparotomy under general anesthesia. The likelihood of successful reversal depends on the method of tubal occlusion originally used. In general, 50 percent to 70 percent of women under- going reversal surgery achieve intrauterine pregnancy (Liskin et al., 1985~. How
80 ~E, PETERSON, AND CHU ever, most reports are from the selected series of highly skilled surgeons using sophisticated microsurgical techniques. VASECTOMY Vasectomy is cutting and occluding the vas deferens to prevent sperm trans- port in semen. Although generally considered to be safe, simple, and highly effective, vasectomy is not widely popular in most of the world. In fact, the majority of the estimated 41 million men currently using vasectomy for contra- ception reside in four countries the United States, United Kingdom, China, and India (Gallen et al., 1986~. By comparison, more than 95 million women world- wide have undergone tubal sterilization. Among regions, Africa is the least active in providing vasectomy services, Asia the most active, and Latin America moder- ately active (Ross and Huber, 1983~. The limited popularity of vasectomy has been attributed to motivational factors and access to services (Gallen et al., 1986~. In addition, female sterilization has become more popular than male sterilization in some countries (Ross and Huber, 1983~. Vasectomy is a minor surgical procedure that usually takes 5 to 10 minutes to perform (Ross et al., 1985~. The procedure consists of isolating the vas deferens and then using one of three methods to occlude it: ligation, coagulation, or clip application. Ligation is the most widely used approach. If ligatures are tied too snugly, resulting in vas transection, or too loosely, resulting in incomplete occlu- sion, sperm can leak into surrounding tissues, resulting in sperm granuloma formation. Coagulation of the vas mucosa is an alternative method of occlusion that minimizes damage to the vas muscularis. Some believe that this method reduces the likelihood of sperm granuloma formation. Clips are rarely used for vas occlusion. Several clips must be applied to achieve maximum effectiveness, and, because they are not absorbable, the vasectomized man can sometimes feel the clips. Local anesthesia without premedication is the anesthetic technique of choice for most vasectomies (Association for Voluntary Sterilization, 1983~. The strong preference for local anesthesia is based on the higher risk of morbidity and mortality associated with general anesthesia (Peterson et al., 1981b, 1983~. A suIvey of selected outpatient facilities conducted in the United States in 1980 indicated that the type of anesthesia used varied substantially by facility type (Kendrick et al., 1985~. Freestanding surgical facilities reported performing 29 percent of their vasectomies using general anesthesia; conversely, 96 percent of the facilities not classified as freestanding reported using local or regional anes- thesia. Outside the United States, particularly in developing countries, almost all vasectomies are performed using local anesthesia (Ross et al., 1985~. Vasectomy failure rates have ranged from 0 percent to 2.2 percent, with most studies reporting fewer than 1 percent (Liskin et al., 1983) (Table 2~. However,
HEALTH EFFECISOFCON7RACEPTION 81 few data are available that allow for a comparative assessment of failure rates associated with the venous methods of vas occlusion. Most vasectomy failures can be attributed to unprotected coitus shortly after vasectomy or spontaneous reanastomosis of the vast Unprotected coitus before sperm has been cleared from the male reproductive tract can result in pregnancy, although the duration of male fertility following vas occlusion remains uncertain (Lewis et al., 1984~. Sperm may be absent within 1 to 10 weeks or more after vasectomy, depending in part on the frequency of ejaculation (Liskin et al., 1983~. The Association for Voluntary Sterilization recommends using contraceptives for the first 15 ejaculations or 6 weeks after vasectomy (Ross et al., 1985~. Most Due vasectomy failures occur when fistulous tracks develop through a sperm granuloma, resulting in a spontaneous reanastomosis of the vas (Liskin et al., 1983~. The reanastomosis usually occurs within several months after vasec- tomy, although late reanastomosis up to 3 years after vasectomy has been docu- mented, both after ligation and coagulation methods. In a study of 14,047 men the wives of 6 men became pregnant between 16 months and 3 years after vasectomy (Philp et al., 1984~. Other causes of vasectomy failure include operating on the wrong structure and congenital duplication of the vas (Liskin et al., 1983~. Mortality The risk of death attributable to vasectomy is quite low. A detailed review of sterilization-attributable deaths in Bangladesh occurring in 1979 and 1980 identi- fied seven deaths attributable to vasectomy, for a vasectomy-attributable mortal- ity rate of 31 per 100,000 procedures (Grimes et al., 1982a). Scrotal infection and sepsis caused most of the deaths. Subsequently, the use of sterile gloves during all vasectomies was recommended. Follow-up surveillance in 1981 identified no vasectomy-attributable deaths among approximately 14,000 men (Grimes et al., 1982b). The Association for Voluntary Sterilization has recorded only two vasectomy-attributable deaths associated with more than 160,000 procedures performed in the international programs it has supported (Ross et al., 1985~. Vasectomy-attributable deaths in the United States are quite rare. While there has been no systematic effort to identify such deaths in the United States, at least one vasectomy-attributable death is known to have occurred (District Court of Ap- peals of Florida, Dunn v. Campbell, 1964~. Morbidity The most common complications of vasectomy are swelling of the scrotal tissue, bruising, and pain. As many as 50 percent of men undergoing the procedure may experience these complaints, which generally subside within 1 to 2 weeks without treatment (Liskin et al., 1983~. Hematoma formation and infection, which can be serious but generally are not, occur much less frequently.
82 ~FE, PETERSON, AND CIIU Hematoma formation generally occurs in fewer than 1 percent of vasectomized men, but it has been reported in as many as 4 percent. Physicians surveyed in the United States in 1983 reported a hematoma rate of 2 percent (Hendrick et al., 1987~. In that survey the hematoma rate was significantly higher among physi- cians performing 1 to 10 vasectomies per year (4.6 percent) than among those performing 11 to 50 vasectomies (2.4 percent) or more than 50 vasectomies (1.6 percent). Infection after vasectomy is uncommon, usually occurring in fewer than 2 percent of men, although rates of infection as high as 6 percent have been reported (Liskin et al., 1983~. Most infections occur at the site oilskin incisions or around skin sutures. In rare instances, such as those reported in Bangladesh, infection can result in sepsis and death (Grimes et al., 1982a). Careful aseptic technique and sterile gloves are likely to reduce the risk of infection (Association for Voluntary Sterilization, 1983~. Epididymitis, manifested as swelling and tenderness of the epididymis, is generally reported in fewer than 1 percent of men undergoing vasectomy (Liskin et al., 1983~. This complication has been attributed to a hydrostatic pressure increase within the epididymis after vas occlusion (Association for Voluntary Sterilization, 1983~. Bacterial infection as an etiology of the epididymitis is uncommon (Schmidt, 1975~. Epididymiiis may develop several months after vasectomy, but it is most likely to occur soon after the procedure. Symptoms usually subside within a week (Ross et al., 1985~. One report (Massey et al., 1984) compared vasectomized and nonvasectomized men with respect to the likelihood of developing epididymitis. After vasectomy, 2.6 percent of men experienced epididymitis versus 1.1 percent of men who had not had the proce- dure. However, differences between the two groups in the rate of epididymitis occurred only within the first year after vasectomy. Sperm granulomas nodules containing sperm, epithelial cells, and lympho- cytes represent an inflammatory response to sperm leaking into surrounding tissue. Most such granulomas are small and clinically not important, but they can result in pain at the site of vas occlusion or in the epididymis. The incidence of sperm granulomas is not known. However, granulomas at the vasectomy site have been found in 15 percent to 40 percent of sterilization reversal procedures; granulomas in the epididymis have been found in 10 percent to 50 percent of these procedures (Liskin et al., 1983~. Symptomatic sperm granulomas are substan- tially less frequent. Follow-up studies report symptomatic granulomas in only 0 percent to 3 percent of vasectomies (Ross~ et al., 1985~. In such instances a tender nodule at the site of vas occlusion or in the epididymis may cause discomfort, although the discomfort usually subsides spontaneously (Schmidt and Morris, 1973~. As already noted, sperm granulomas can cause vasectomy failure if fistulous tracks develop through the granuloma. Ligation for vas occlusion is more likely to result in sperm granuloma formation than use of coagulation (Schmidt and Free, 1978~. However, coagulation results in more complete
HEALTH EFFECTS OF CO=RACE~ION 83 occlusion of the vas, and this occlusion may increase the likelihood of an epididy- mal granuloma with resultant epididymal obstruction (Silber, 1978~. Long-Term Safety A series of major epidemiologic investigations have been conducted in the l980s to assess the long-term safety of vasectomy. In general, those studies are remarkably consistent in failing to identify any long-term health risks attributable to vasectomy. Studies of long-term complications associated with vasectomy were stimulated primarily by two concerns: (1) evidence that 50 percent to 70 percent of men develop antisperm antibodies following vasectomy (Ansbacher, 1971; Schulman et al., 1972) and (2) reports of increased atherosclerosis in cynomolgus monkeys after vasectomy compared with nonvasectomized monkeys (Alexander et al., 1974~. These findings are the basis for the hypothesis that antisperm antibodies form circulating immune complexes that might collect in arterial walls and contribute to the development of atherosclerosis (Alexander and Anderson, 1979~. At least six epidemiologic studies specifically designed to assess this hypothesis have since indicated that vasectomy does not increase the incidence of MI in the 10 years following vasectomy (Goldacre et al., 1978, 1979; Walker et al., 1981; Petitti et al., 1982; Massey et al., 1984; Perrin et al., 1984~. Subsequent to these reports, Rosenberg et al. (1986) evaluated data from a hospital-based, case- control study to assess whether vasectomy was associated with a subsequent increase in the risk of MI 10 or more years after surgery and whether any effect was more pronounced in men already predisposed to MI. The analysis supported findings from previous reports indicating that vasectomy did not increase the risk of MI within 10 years after surgery. Further, the authors found no overall increase in risk 10 or more years after surgery and no additional risk among men already predisposed to MI because of the presence of other risk factors for coronary artery disease. In summary, the available epidemiologic studies provide strong evidence that vasectomy does not increase the likelihood of MI. As noted, much of the concern regarding the relationship between vasectomy and atherosclerosis was precipitated by reports of worsening atherosclerosis among vasectomized monkeys. Recently, the investigators who first reported this find- ing have presented data that do not support their first report. Their new data suggest that vasectomy does not increase the extent of atherosclerosis among cynomolgus monkeys, even in those found to be hyperresponsive to an athero- genic diet (Clarkson et al., 1988~. Several epidemiologic studies have assessed the relationship between vasec- tomy and a variety of other diseases associated with alterations of the immune system. In a historical cohort study conducted in the United States, 10,590 vasectomized men from four cities were interviewed, along with a neighborhood control for each vasectomized man (Massey et al., 1984~. Any report of disease
84 f FE, PETERSON, AND CHU potentially related to vasectomy via an immunopathologic mechanism was vali- dated by medical record review. Except for epididymitis/orchitis, the incidence of diseases, including those with a known or suspected immunopathologic basis, was similar for vasectomized men and paired controls. This finding was true not only for individual diseases but also for broader categories of diseases in which immunopathologic mechanisms might operate. Two other studies have failed to show an increase in the incidence of hospitalized illness for vasectomized men compared with nonvasectomized men (Walker et al., 1981; Petitti et al., 1982~. The sole exception was that in one of the studies vasectomized men had higher rates of hospitalization for diseases of the genito-urinary system during the early postvasectomy period (Walker et al., 1981~. One recent report (Kronmal et al., 1988) raises additional concern regarding the relationship between vasectomy and subsequent genito-urinary tract disease by suggesting that vasectomy may increase the likelihood of urolithiasis. The age-adjusted relative risk for calculi among men who had undergone vasectomy was statistically significantly increased by approximately 70 percent. Although based on a study designed to evaluate the relationship between vasectomy and coronary artery disease, these findings are supported by another report mentioned previously (Walker et al., 1981~. By contrast, another study found that the incidence of hospitalization for genito-urinary diseases was not significantly different between vasectomized and nonvasectomized men (Petitti et al., 1983~. One possible biologic explanation for a relationship between vasectomy and urolithiasis centers around the production of antisperm antibodies. Whether antibody production might increase the tendency for formation of urinary calculi is not known. The possibility that vasectomized men may have an increased risk for genito-urinary disease warrants further evaluation, but available data are generally reassuring. Vasectomy Reversal Changes in life situation may lead men to regret their decision to undergo vasectomy. Surveys indicate that the prevalence of such regret worldwide is low (Liskin et al., 1983~. Vasectomy reversal requires a high level of surgical expertise, is expensive, and is relatively unavailable in many parts of the world. Microsurgical techniques are necessary for optimal results. Furthermore, the original vasectomy technique may affect the likelihood of success of any subse- quent reversal attempt. Because of the wide variety of factors influencing the likelihood of successful reversal, the success rate in the aggregate is difficult to estimate. Most surgeons report 30 percent to 60 percent success rates for vasec- tomy reversal, whereas some report even higher rates (Liskin et al., 1983~. Even if the theoretical reversal rate remains consistently high, the technical difficulty of the surgery, with the attendant requirements for skilled microsurgeons, makes reversal impractical or unavailable for men in much of the world (Ross et al., 1985~.
HEALTH EFFECTS OF CONTRACEPTION 535 REFERENCES Affandi, B., S. S. I. Santoso, Djajadilaga, et al. 1987 Pregnancy after removal of Norplant~implants contraceptive. Contraception 36:203-209. Alexander, N. J., and D. J. Anderson 1979 Vasectomy: consequences of autoimmunity to sperm antigens. Fertility & Sterility 32:253-260. Alexander, N. J., B. J. Wilson, and B. D. Patterson 1974 Vasectomy: immunological effects on rhesus monkeys and men. Fertility & Sterility 25:149-156. Anonymous 1986 Oral contraceptives and breast cancer. Uncut 2:665 666. Ansbacher, R. 1971 Sperm-agglutinating and sperm-immobilizing antibodies In vasectorruzed men. Fer- tility & Sterility 22:629-632. Association for Voluntary Sterilization 1983 Minutes prepared for a Vasectomy Senunar. Science Committee, New York, June 6. Austin, H., W. C. Louv, and W. J. Alexander 1984 A case-control study of spermicides and gonorrhea. Journal of the American Medical Association 251:2822-2824. BaeHer, E. A., W. P. Dillon, T. J. Cumbo, et al. 1982 Prolonged use of a diaphragm and toxic shock syndrome. Fertility & Sterility 38:248-250. Bain, C., C. H. Hennekens, F. E. Speizer, et al. 1982 Oral contraceptive use and malignant melanoma. Journal of the National Cancer Institute 68:537-539. Barlow, D. 1977 The comedown and gcnonhea. L~ncet 2:811-813. Beral, V., S. Evans, H. Shaw, et al. 1984 Oral contraceptive use and malignant melanoma in Australia. British Journal of Cancer 50:681~85. Bhiw&ndiwala, P. P., S. D. Mumford, and P. J. Feldblum 1982 A comparison of different laparoscopic sterilization occlusion techniques In 24,439 procedures. American Journal of Obstetrics and Gynecology 144:319-331. Menstrual pattern changes following laparoscopic sterilization with different occlu sion techniques: a review of 10,004 cases. American Journal of Obstetrics and Gynecology 145:684-694. Bhiwandiwala, P. P., S. D. Mumford, and K. I. Kennedy 1985 Canparison of the safety of open and conventional laparoscopic sterilization. Obstet- rics and Gynecology 66:391-394. Bradley, D. D., J. Wingerd, D. B. Petitti, et al. 1978 Serum high density lipoprotein cholesterol in women using oral contraceptives, estro- gens, and progestins. New England Journal of Medicine 299:17-20. Bnnton, L. A., M. P. Vessey, R. Flavel, et al. 1981 Risk factors for benign breast disease. American Journal of Epidemiology 113:203-214. Bnnton, L. A., G. R. Huggins, H. F. I=ham, et al. 1986 I=ng-term use of oral contraceptives and risk of invasive cervical cancer. Interna- tional Journal of Cancer 38:330344. 1983
86 LEE, PETERSON, AND CHU The Cancer and Steroid Hormone Study of the Centers for Disease Control and the National Institute of Child Health and Human Development 1986 Oral-contraceptive use and the risk of breast cancer. New England Journal of Medicine 315:405-411. The Cancer and Steroid Hormone Study of the Centers for Disease Control and the National Institute of Child Health and Human Development 1987a Combination oral contraceptive use and the risk of endometnal cancer. Journal of the American Medical Association 257:796-800. The Cancer and Steroid Hormone Study of the Centers for Disease Control and the National Institute of Child Health and Human Development 1987b The reduction in risk of ovarian cancer associated with oral-contracep~ve use. New England Journal of Medicine 316:65~655 Celentano, D. D., A. C. Klassen, C. S. Weissman, et al. 1987 The role of contraceptive use in cervical cancer: the Maryland cervical cancer case- control study. American Journal of Epidemiology 126:592 604. Centers for Disease Control 1987 Antibody to human irnmunodeficiency virus in female prostitutes. Morbidity and Mortality Weekly Report 36:157-161. 1988 Condoms for prevention of sexually transmitted diseases. Morbidity and Mortality Weekly Report 37:133-137. Chamberlain, G., and J. C. Brown, eds. 1978 Gynaecological Laparoscopy: The Report of the Working Party of the Confulential Inquiry info Gynaecological Laparoscopy. London: The Royal College of Obstetri- cians and Gynaecologists. Chi, I.-C., and P. J. Feldblum 1981 Luteal phase pregnancies in female sterilization patients. Contraception 23:10. Oaricson, T. B., N. J. Alexander, and T. M. Morgan 1988 Atherosclerosis of cynomolgus monkeys hyper- and hyporesponsive to dietary cho- lesterol: lack of an effect of vasectomy. Arteriosclerosis 8:488-498. Cohen, M. M. 1987 Long-term risk of hysterectomy after tubal sterilization. American Journal of Epi demiology 125:410-419. Conant, M. A., D. W. Spicer, and C. D. Smith 1984 Herpes simplex virus transmission: condom studies. Sexually Transmitted Diseases 11 :94-95. Conant, M., D. Hardy, J. Sematinger, et al. 1986 Condoms prevent transmission of AlDS-associated retrovirus. Journal of the Ameri- can Medical Association 255:1706. Cordero, J. F., and P. M. Layde 1983 Vaginal spermicides, chromosomal abnormalities and limb reduction defects. Family Planning Perspective 15:16. Cowan, M. E., and G. E. Cree 1973 A note on the susceptibility of N. gonorrhocae to contraceptive agent Nonyl-P. British Journal of Venereal Diseases 49:65-66. Cramer, D. W., I. Schiff, S. C. Schoenbaum, et al. 1985 Tubal infertility and the intrauterine device. New England Journal of Medicine 312:941-947. Cramer, D. W., M. B. Goldman, I. Schiff, et al. 1987 The relationship of tubal infertility to barrier method and oral contraceptive use. Journal of the American Medical Association 257:244~2450.
IlEALTH EFFECTS OF CONTRA CAPTION 557 Dating, J. R., N. S. Weiss, B. J. Metch, et al. 1985 Primary tubal infertility in relation to the use of an intraulennc device. New England Journal of Medicine 312:937-941. DeStefano, F., H. B. Peterson, P. M. Layde, et al. 1982 Risk of ectopic pregnancy following tubal sterilization. Obstetrics and Gynecology 60:32~330. DeStefano, F., J. R. Greenspan, R. C. Dicker, et al. 1983a Complications of interval laparoscopic tubal sterilization. Obstetrics and Gynecology 61:153-158. DeStefano, F., C. M. Huezo, H. B. Peterson, et al. 1983b Menstrual changes after tubal sterilization. Obstetrics and Gynecology 62:673-681. DeStefano, F., J. A. Perlman, H. B. Peterson, et al. 1985 Long-term risk of menstrual disturbances after tubal sterilization. American Journal of Obstetrics and Gynecology 152:835-841. Diaz, S., M. Pavez, P. Miranda, et al. 1987 Long-term follow-up of women treated with Norplant,M implants. Contraception 35:551-567. District Court of Appeals of Florida 1964 Dunn v. Campbell, Flonda, 166 SO. 2d217. July 1, Second Distnct. Ebeling, K., P. Nischan, and C. Schindler 1987 Use of oral contraceptives and risk of invasive cervical cancer in previously screened women. International Journal of Cancer 39:427~30. Faich, G., K. Pearson, D. Fleming, et al. 1986 Toxic shock syndrome and the vaginal contraceptive sponge. Journal of the American Medical Association 255:21~218. Feldblum, P. J., and J. A. Fortney 1988 Condoms, speTmicides, and the transmission of human immunodeficiency virus: a review of the literature. American Journal of Public Health 78:52-54. Finn, S. D., R. H. Latham, P. Roberts, et al. 1985 Association between diaphragm use and urinary tract infection. Journal of the American Medical Association 254:24~245. Fischl, M. A., G. M. Dickinson, G. B. Scott, et al. 1987 Evaluation of heterosexual partners, children, and household contacts of adults with AIDS. Journal of the Americar: Medical Association 257:64~44. Forman, D., T. J. Vincent, and R. Doll 1986 Cancer of the liver and the use of oral contraceptives. British Medical Journal 292:1357-1361. Fortney, J. A., L. P. Cole, and K. I. Kennedy 1983 A new approach to measuring menstrual pattern change after sterilization. American Journal of Obstetrics and Gynecology l4/:83~836. Fothe~by, K., I. Trayner, I. Howard, et al. 1982 Effect of injectable norethisterone oenanthate ~origest) on blood lipid levels. Contra- ception 2S:435~46. Foulkes, J., and G. Chamberlain 1985 Effects of sterilization on menstruation. Southern Medical Journal 78:544. Foxman, B., and R. R. Frerichs 1985 Epidemiology of urinary tract infections. I: Diaphragm use and sexual intercourse. American Journal of Public Health 75:1308-1313. Francis, D. P., and J. Chin 1987 1be prevention of acquired immunodeficiency syndrome in the United States. Jour- nal of the American Medical Association 257:1357-1366.
8~3 ~E, PETERSON, AND CHU Gallen, M. E., L Liskin, and N. Kak 1986 Men-new focus for family planning programs. Population Reports Series J. No. 33. Goldacre, M. J., J. A. Clarke, M. A. Heasman, et al. 1978 Follownp of vasectomy using medical record linkage. American Journal of Epidemi- ology 108: 176-180. Goldacre, M., M. Vessey, J. Clarke, et al. 1979 Record linkage study of morbidity following vasectomy. Pp. 567-579 in I. H. Lepow and R. Crazier, eds. Vasectomy: Immunologic and Pathophysiologic Effects in Anunals and Man. New York: Academic Press. Gray, R. H. 1985 Reduced risk of pelvic inflammatory disease with injectable contraceptives. Lancet 1:1046. Grimes, D. A. 1987 Intrauterine devices and pelvic inflammatory disease: recent developments. Contra- ception 36:97-109. Grimes, D. A., H. B. Peterson, M. J. Rosenberg, et al. 1982a Sterilization-aunbutable deaths in Bangladesh. International Journal of Gynaecology and Obstetrics 20:149-154. Grimes D. A., A. P. Sattenhwaite, R. W. Rochat, et al. 1982b Deaths fran contraceptive sterilization in Bangladesh: rates, causes, and prevention. Obstetrics and Gynecology 60:635~40. Grubb, G. S., and H. B. Peterson 1985 Luteal phase pregnancy and tubal stenlization. Obstetrics and Gynecology 66:784-788. Grubb, G. S., H. B. Peterson, P. M. Layde, et al. 1985 Regret after decision to have a tubal sterilization. Fertility & Sterility 44:248-253. Gwinn, M. L., N. C. Lee, P. H. Rhodes, et al. n.d. Pregnancy, breast feeding, and oral contraceptives and the risk of epithelial ovarian cancer. Journal of Clinical Epidemiology. (Submitted for publication) Hams, R. W. C., L. A. Brinton, R. H. Cowdell, et al. 1980 Characteristics of women with dysplasia or carcinoma in situ of the cervix uteri. British Journal of Cancer 42:35~369. Hatcher, R. A., F. Guest, F. Stewart, et al. 1988 Contraceptive technology: 1988-1989. New York: Printed Matter, Inc./Irvington Publisher, Inc. Heartwell, S. F., and S. Schlesselman 1983 Risk of uterine perforation among users of intrauterine devices. Obstetrics and Gynecology 61:31-36. Henderson, B. E., S. Preston-Martin, H. A. Edmondson, et al. 1983 Hepatocellular carcinoma and oral contraceptives. 48:437 410. Henshaw, S. K., and S. Singh British Journal of Cancer 1986 Sterilization regret among U.S. couples. Family Planning Perspectives 18:23~240. Hicks, D. R., L. S. Martin, J. P. Getchell, et al. 1985 Inactivation of HTLV-m/LAV-infected cultures of normal human lymphocytes by nonoxynol-9 in vitro. Lancet 1422-1423. Holly, E. A., N. S. Weiss, and J. M. Liff 1983 Cutaneous melanoma in relation to exogenous hormones and reproductive factors. Journal of the National Cancer Institute 70:827-831. Hopper, R. R., G. H. Reynolds, O. G. Jones, et al. 1978 Cohort study of venereal disease. I. The risk of gonorrhea transmission from infected women to men. American Journal of Epidemiology 108:13~144.
HEALTH EFFECTS OF CONTRACEPTION 539 Horsburgh, C. R., J. M. Douglas, and F. M. LaForce 1987 Preventive strategies in sexually transmitted diseases for the primary care physician. Journal of the American Medical Association 258:815-821. Huggins, G., M. Vessey, R. Flavd, et al. 1982 Vaginal spennicides and outcome of pregnancy: findings in a large cohort study. Contraception 25:219. Hulks, J. F., H. B. Peterson, M. Surrey, ct al. 1987 American Association of Gynecologic Laparoscopists' 1985 membership surrey. Journal of Reproductive Medicine 32:732-73S. Hymowitz, E. E. 1981 Toxic shock syndrome and the diaphragm. New England Journal of Medicine 305:834. Indian Council of Medical Research 1982 Collaborative Study on Scquelae of Tubal Sterilization. Monograph. New Delhi. Twin, K. L., L. Rosero-Bixby, M. W. Oberlc, ct al. 1988 Oral contraceptives and cervical cancer risk in Costa Rica: Detection bias or causal association? Journal of He American Medical Association 259:59 64. Tick, H., A. M. Walker, K. J. Rothman, et al. 1981 Vaginal spennicides and congenital disorders. Journal of the American Medical Association 245:1329-1332. Tick, H., M. T. Hannan. A. Sterzachis, et al. 1982 Vaginal spermicides and gonorrhea. Journal of the American Medical Association 248:1619-1621. Judson, F. N., J. M. Enret, G. M. Bodin, et al. in In vitro evaluations of condoms with and without nonoxynol-9 as physical and chemical barriers against Chlamydia trachomatis, herpes simplex virus type 2, and human ~mmunodeficiency Virus. Set lly Transmitted Diseases (Suppl.) Katmelson, S., W. L. Drew, and ~ Mintz 1984 Efficacy of the condom as a barrier to the transmission of cytomegalo virus. Journal of Infectious Diseases 150:155-1S7. Kelaghan, J., G. ~ Rubin, H. W. Ory, et al. 1982 gamer-method contraceptives and pelvic inflammatory disease. Journal of the Ameri- can Medical Association 248:180187. Kenduck, J. S., E. P. Rhodenhiser, G. L. Rubin, et al. 1985 Charactenstics of vasectomy performed in selected outpatient facilities in the United States, 1980. Journal of Reproductive Medicine 30:93~938. Kendnclc, J. S., B. Gonzales, D. H. Huber, ct al. 1987 Complications of vasectomy in the United States. Journal of Family Practice 25:245-248. Krorunal, R. A., J. N. Krieger, J. W. Kennedy, et al. 1988 Vasectomy and urolithiasis. Lancet 1:22-23. Layde, P. M., M. P. Vessey, and D. Yeates 1982 Risk factors for gallbladder disease: a cohort study of young women attending family planning clinics. Journal a] Epidemiology and Corrvrumity Health 36:27~278. Ike, N. C., L. Rosero Bixby, M. W. Oberle, et al. 1987 A case control study of breast cancer and hormonal contraception in Costa Rica. Journal of the National Canecr Institute 79:1247-1254. Lee, N. C., G. L. Rubin, and R. Borucki 1988 The intrauterine device and pelvic inflammatory disease revisited: new results from the Women's Health Study. Obstetrics and Gynecology 72:1~. Lewis, E. L., C. K. Brazil, and J. W. Overstrcet 1984 Human sperm function in the ejaculate following vasectomy. Fertility & Sterility 42:895-898. press
90 ~ FE, PETERSON, AND CH[J Liang, A. P., A. G. Levenson, P. M. Layde, et al. 1983 Risk of breast, uterine Corpus, and ovarian cancer in women receiving medro~cypro- gesterone injections. Journal of the American Medical Association 249:2909-2912. Liskin, L., and G. Fox 1982 IUDS: an appropriate contraceptive for many women. Population Reports Series B. No. 4. Liskin, L., and W. F. Quillin 1982 Long-acung progestins: promise and prospects. Population Reports Series K, No. 2. Liskin, L., J. M. Pile, and W. F. Quillin 1983 Vasectomy-safe and simple. Population Reports Series D, No. 4. Liskin, L., W. Rinehart, R. Blackburn, et al. 1985 Minilaparotomy and laparoscopy: safe, effective and widely used. Population Re- ports Series C, No. 9. Liskin, L., R. Blackbum, and R. Ghani 1987 Honnmal contraception: new long-acting methods. Population Reports Series K, No. 3. Lunt, R. 1984 Worldwide early detection of cervical cancer. Obstetrics and Gvn~oloa~v f~1~7nP~711 Mann, J., T. C. Quinn, and P. Plot 1987 Condom use and HIV infection among prostitutes-Zaire (Issuer). New England Journal of Medicate 316:345. Massey, F. J., G. S. Bemstein, W. M. O'Fallon, et al. 1984 Vasectomy and health: results from a large cohort study. Journal of the American Medical Association 252:1023-1029. McPherson, K., and J. O. Dnfe 1986 The pill and breast cancer: Why the unce~nty? British Medical Journal 293:709-710. McPherson, K., A. Neil, M. P. Vessey, et al. 1983 Oral contraceptives and breast cancer. Lancet 2:1414-1415. McPherson, K., P. A. Coope, and M. P. Vessey 1986 Early oral contraceptive use and breast cancer: theoretical effects of latency. Journal of Epidemiology and Community [lealth 40:289-294. Meirik, O., E. Lund, H.-O. Adami, et al. 1986 Oral contraceptive use and breast cancer in young women. Locket 2:650 654. Mills, J. L., E. E. Harley, G. F. Reed, et al. 1982 Are spermicides teratogenic? Journal of the American Medical Association 248:2148. Mintz, M. 1977 _,, - vies, v - ., it, A. Risk and prophylaxis in laparoscopy: a survey of 100,000 cases. Journal of Renro- ductive Medicine 18:269-272. Mumford, S. D., P. P. Bhiwandiwala, and I.-C. Chi 1980 Laparoscopic and minilaparotomy female sterilization compared in 15,167 cases. Lancet 2:10~1070. National Center for Health Statistics 1982 Vital Statistics of the United States. Vol. 2. Mortality Part A. Hyattsville, Md.: National Center for Health Statistics. Neuberger, J., D. Forman, R. Doll, et al. 1986 Oral contraceptives and hepatocellular carcinoma. British Medical Journal 292:1355-1357. Oberle, M. W., L. Rosero-Bixby, K. L Irwin, et al. 1988 Cervical cancer risk and use of depot-medroxyprogesterone acetate in Costa Rica. International Journal of Epidemiology 17:718-723. Ory, H. W. 1981 The Women's Health Study. Ectopic pregnancy and intrauterine contraceptive de vices: new perspectives. Obstetrics and Gynecology 57:137-144. , . . .
HEALTH EFFECTS OF CONTRACEPTION 91 1982 the noncontraceptive health benefits from oral contraceptive use. Family Planning Perspective 14:182-184. Only, H. W., J. D. Forrest, and R. Lincoln 1983 Making Choices: Evaluating the Health Risks and Benef Is of Birth Control Methods. New York: lye Alan Guttrnacher Institute. Pardthaisong, T., R. H. Gray, and E. B. McDaniel 1980 Return of fertility after discontinuation of depot-medroxyprogesterone acetate and intra-uterine devices in Northem Thailand. Lancet 1:509-512. Pasquale, S. A., V. Brandeis, R. I. Cruz, et al. 1987 NorplantTM contraceptive implants: rods versus capsules. Contraception 36:305-316. Paul, C., D. C. G. Skegg, G. F. S. Spears, et al. 1986 Oral contraceptives and breast cancer: a national study. British Journal of Medicine 293:723-726. Perrin, E. B., J. S. Woods, and N. Tsukasu, et al. 1984 Long-tenn effect of vasectomy on coronary heart disease. American Journal of Public Health 74:128-132. Peterson, H. B., H. W. Ory, J. R. Greenspan, et al. 1981a Deaths associated with laparoscopic sterilization by unipolar coagulating devices, 1978-1979. American Journal of Obstetrics and Gynecology 139:141. Peterson, H. B., D. A. Grimes, W. Cates, et al. 1981b Comparative risk of death from induced abortion at 12 weeks gestation performed with local vs. general anesthesia. American Journal of Obstetrics and Gynecology 141:763. Peterson, H. B., F. DeStefano, J. R. Greenspan, et al. 1982a Mortality risk associated with tubal sterilization in United States hospitals. American Journal of Obstetrics and Gynecology 143:125-129. Peterson, H. B., J. R. Greenspan, and H. W. Ory 1982b Death following puncture of the aorta during laparoscopic sterilization. Obstetrics and Gynecology 59:133. Peterson, H. B., F. DeStefano, G. L. Rubin, et al. 1983 Deaths attributable to tubal sterilization in the United States, 1977-1981. American Journal of Obstetrics and Gynecology 146: 131-136. Petitii, D. B., R. Klein, H. Kipp, et al. 1982 A survey of personal habits, symptoms of illness, and histories of disease in men with and without vasectomies. American Journal of Public Health 72:476-480. Petitti, D. B., R. Klein, H. Kipp, et al. 1983 Vasectomy and the incidence of hospitalized illness. Journal of Urology 129:76~762. Philp, T., J. Guillebaud, and B. Budd 1984 Late failure of vasectomy after 2 documented analyses showing azoosperrnic semen. British Medical Journal 289:77-79. Pike, M. C., B. E. Henderson, M. D. Krailo, et al. 1983 Breast cancer in young women and the use of oral contraceptives: possible modifying effects of formulation and age at use. Lancet 2:92~930. Piotrow, P. T., W. Rinehart, and J. C. Schmidt 1979 IUDs-update on safety, effectiveness, and research. Population Reports Series B. No. 3. Piper, I. N. 1985 Oral contraceptives and cervical cancer. Gynecologic Oncology 22:1-14. Prentice, R. L., and D. B. Thomas 1987 On the epidemiology of oral contraceptives and disease. Advances in Cancer Re- search 49:285~01.
92 [FE, PETERSON, AND CHU Quinn, R. W., and K. R. O'Reilly 1985 Contraceptive practices of women attending the sexually transmitted diseases clinic in Nashville, Tennessee. Sexually Transmitted Diseases 12:99-102. Rameharan, S., F. A. Pellegrin, R. Ray, et al. 1981 We Walnut Creek contraceptive drug study. A prospective study of the side effects of oral eontraeeptives, Vol. m. Washington, D.C.: U.S. Government Printing Office. Reingold, A. ~ 1986 Toxic shock syndrome and the contraceptive sponge. Journal of the American Medical Association 255:242-243. Rietmeijer, C. A. M., J. W. Krebs, P. M. Feorino, et al. 1988 Condoms as physical and chemical barriers against human immunodefieieney virus. Journal of the American Medical Association 259:1851-1853. Rooks, J. B., H. W. Ory, K. G. Ishak, et al. 1979 Epidemiology of hepatoeellular adenoma: the role of oral contraceptive use. Journal of the American Medical Association 242:644 648. Rosenberg, L., P. J. Sehwingel, D. W. Kaufmann, et al. 1986 Ihe risk of myoeardial infarction 10 or more years after vasectomy in men under 55 years of age. American Journal of Epidemiology 123:1049-1056. Rosenberg, M. J., W. Rojanapithayakom, P. J. Feldblum, and J. E. Higgins 1987 Effect of the contraceptive sponge on ehlamydial infection, gonorrhea, and eandidiasis. Journal of the American Medical Association 257:2308-2312. Rosenfield, A., D. Maine, R. Roehat, et al. 1983 We Food and Drug Administration and medroxyprogesterone acetate. Journal of the American Medical Association 249:2922-2928. Ross, J. A., and D. H. Huber 1983 Acceptance and prevalence of vasectomy in developing countnes. Studies in Family Planning 14:67-72. Ross, J. A., S. Hong, and D. H. Huber 1985 Voluntary Sterilization: An International Fact Book New York: Association for Voluntary Sterilization. Ross, R. K., M. C. Pike, M. P. Vessey, et al. 1986 Risk factors for uterine fibroids: reduced risk associated with oral contraceptives. British Medical Journal 293:359-362. Royal College of General Practitioners 1970 Oral eontraeepiives and health. London: Pitman Medical. Royal College of General Practitioners Oral Contraception Study 1982 Oral eontraeepiives in gallbladder disease. Lancet 2:957-959. Rubin, G. L., H. W. Ory, and P. M. Layde 1988 1982 Oral contraceptives and pelvic inflammatory disease. American Journal of Obstetrics and Gynecology 144:630-635. Ryden, G., L. Fahraeus, L. Molin, et al. 1979 Do contraceptives influence the incidence of acute pelvic inflammatory disease in women with gonorrhea? Contraception 20:149-157. Salah, M., A. M. Ahmed, M. Abo-Eloyoun, et al. 1987 Five-year experience with NorplantlM implants in Assiut, Egypt. Contraception 35:543-550. Sehlesselman, J. J., B. V. Stadel, P. Mullay, et al. 1987 Consistency and plausibility in epidemiologic analyses: application to breast cancer in relation to use of oral contraceptives. Journal of Chronic Diseases 40:1033-1039. Breast cancer in relation to early use of oral contraceptives: no evidence of a latent effect. Journal of the American Medical Association 259:1828-1833.
IlEALTW EFFECTS OF CONTRA CEPI ION 93 Schmidt, S. S. 1975 Complications of vas surgery. Pp. 78-88 in J. J. Sciarra, C. Markland, and J. J. Speidel, eds. Control of Male Fertility. Hagerstown, Md.: Harper & Row. Schmidt, S. S., and N. J. Free 1978 Cite bipolar needle for vasectomy. 1. E'cpenence with the first 1,000 cases. Fertility & Sterility 29:676~580. Schmidt, S. S., and R. R. Moms 1973 Spermatie granuloma: the complication of vasectomy. Fertility & Sterility 24:941-947. Schulman, S., E. Zappi, U. Ahmed, et al. 1972 Immunologic consequences of vasectomy. Contraception 5:269-278. Schwartz, B., S. Gaventa, C. V. Broome, et al. 1989 Nonmenstrual toxic shock syndrome associated with bamer contraceptives: report of a ease-control study. Review of Infectious Diseases 11 (Suppl. 1):S43-S49. Shain, R. N., W. B. Miller, and A. E. C. Holden 1986 Married wanen's dissatisfaction with tubal sterilization and vasectomy at first-year follow-up: effects of perceived spousal dominance. Fertility & Sterility 45:808-819. Shapiro, S., D. Stone, O. P. Heinonen, et al. 1982 Birth defects and vaginal spermieides. Journal of the American Medical Association 247:2381. Shems, J. D., S. H. Moore, and G. Fox 1984 New developments in vaginal contraception. Population Reports Series H. No. 7. Silber, S. J. 1978 Vasectomy and vaseetcxny reversal. Fertility & Sterility 29:125-140. Singh, B., J. C. Cutler, and A. M. Utidjian 1972 II. Effect in vitro of vaginal contraceptive and noncontracepiive preparation on Treponema pallidum and Neisseria gonorrhoeae. British Journal of Venereal Dis- eases 48:57~4. Singh, B., B. Postie, and J. C. Cutler 1976 Virucial effect of certain chemical contraceptives on type 2 herpes virus. American Journal of Obstetrics and Gynecology 126:422~25. sivin, I., and J. Stem 1979 Long acting, more effective Copper T IUD's: a summary of U.S. experience, 197~1975. Studies in Family Planning 10:263-281. Skegg, D. C. 1988 Potential for bias in ease-control studies of oral contraceptives and breast cancer. American Journal of Epidemiology 127:205-212. Smith,G.L.,andK.F.Smith 1986 Lack of HIV infection and condom use in licensed prostitutes (Lever). Lancet 2:1392. Spring, S. B., and J. Gruber 1985 Relationship of DNA viruses and cervical carcinoma. Journal of the National Cancer Institute 75:589-590. Stadel, B. 1986 Oral contraceptives and the occurrence of disease: clinical overview. Pp. 3~1 in A. T. Gregoire and R. G. Blye, eds. Coniraceptive Steroids~harmacology and Safety. New York: Plenum Press. Stadel, B. V., G. L. Rubin, L. A. Webster, et al. 1985 Oral contraceptives and breast cancer in young women. Lancet 2:97~973. Stone,K.M.,D. A.Grimes,and L.S.Magder 1986 Personal protection against sexually transmitted diseases. American Journal of Ob stetrics and Gynecology 155:18~188. Sun, M. 1984 Panel says Depo-Provera not proved safe. Science 226:95~951.
94 ~E, PETERSON, AND ClIU Swan, S. H., and D. B. Petitti 1982 A rewew of problems of bias and confounding in epidemiologic studies of cervical neoplasia and oral contraceptives. American Journal of Epidemiology 115:1~18. Swenson, I., A. R. Khan, and F. A. Jahan 1980 A randomized, single blind comparative trial of norethindrone enanthate and depot- medroxyprogesterone acetate in Bangladesh. Contraception 2:207-215. Treirnan, K., and L. Iiskin 1988 IUDs a new look. Population Reports Series B. No. 5. Trussell, J., and K. Kost 1987 Contraceptive failure in the United States: a critical review of the literature. Studies in Family Planning 18:237-283. Vessey, M. P. 1980 Female hormones and vascular disease-an epidemiological overview. British Jour- nal of Family Planning 6:1-12. Vessey, M. P., N. H. Wnght, K. McPherson, et al. ~ 1978 Fertility after stopping different methods of contraception. British Medical Journal (6108) February:265-267. , D. Yeates, R. Flavel, et al. ~, . . ~.. . .. . . . . . . ^. .. . ~. Vessey, M. P., 1981 Pelvic inflammatory disease and the intrauterine device: findings in a large cohort study. British Medical Journal 282:855-857. Vessey, M., G. Huggins, M. Lawless, et al. 1983 Tubal stenlization: findings in a large prospective study. British Journal of Obstet- rics and Gynaecology 90:203-209. Vessey, M. P., M. A. Metcalfe, K. McPherson, et al. 1987 Urinary tract infection in relation to diaphragm use and obesity. International Journal of Epidemiology 16:441~44. Walker, A. M., H. Jick, J. R. Hunter, et al. 1981 Vasectomy and nonfatal myocardial infarction. Lancet 1:13-15. Washington, A. E., S. Gove, J. Schachter, et al. 1985 Oral contraceptives, chlamydia trachomatis infection, and pelvic inflammatory dis ease: a word of caution about protection. Journal of the American Medical Associa tion 253:224~2250. Weiss, N. S. 1982 Ovary. Pp. 871-880 in D. Schottenfeld and J. F. Fraumeni, Jr., eds. Cancer Epidemi ologyand Prevention. Philadelphia: W. B. Saunders. Westrom, L. 1987 Pelvic inflammatory disease. Bacteriology and sequelae. Contraception 36:111-128. World Federation of Health Agencies for the Advancement of Voluntary Surgical Contraception 1984 Safety of Voluntary Surgical Contraception. New York. World Health Organization Task Force on Long-Acting Systemic Agents for the Regulation of Fertility Multinational comparative clinical evaluation of two long-acting injectable contracep- tive steroids: norethisteronc oenanthate and medroxyprogesterone acetate. 2. Bleed- ing patterns and side effects. Contraception 17:395~06. World Health Organization Task Force on Female Sterilization 1982a ' ' ' - ~ World Health 1982b 1978 mmllaparotomy or laparoscopy tor sterll~zat~on: a multicenter, multinational random- ized study. American Journal of Obstetrics and Gynecology 143:645~52. Organization Task Force on Female Sterilization Randomized comparative study of culdoscopy and minilaparotomy for surgical contra- ception in women. Contraception 26:587-593.
"~~ EFFECTS OF C ~93 Wodd Hat Org~i~d~ 198~ ~~ed~=eslemne (SPA) and Manor: memorandum _ s WHO mung. 8~ Iffy -~ IDA C~;~ ~:37~S2. Wodd Heat 0~1-on Task fog on ~ng~c~g Agent for Feat Region 198~ ~ sideman of Became de-~ed~x~mgesm~ne aceme, l5-g ~me- mo~, ~ undem~shed lacing then. B~ ~ -~6 O~- '^ ~87~94. Wow Heat OF Task ate ~ -At physic Acts for Feat Regular 1987 A ma ~= ~ ~m~mdve cb~c~ ~ ~ de~t-medmx~esm~ne ~ ~ an_ ~ ~- ~ lag ~ If: Hi. Bang prams. ~f~ 33:591~10. Wail, N. H., a. P. Valley, B. Kiowas, ~ a. 1978 Ne~ssia ad dysplasia of He cent umd ad coon: a Passive prove each of He ~aph~gm. Bri'bA WHIZ ~r 38:~3~9.
