- Email: [email protected]
Oral contraceptives, tubal sterilization, and functional ovarian cyst risk
Oral Contraceptives, Tubal Sterilization, and Functional Ovarian Cyst Risk Victoria L. Holt, PhD, MPH, Kara L. Cushing-Haugen, MS, and Janet R. Daling, PhD OBJECTIVE: To determine whether current contraceptive method affects functional ovarian cyst risk, with emphasis on oral contraceptives (OCs) and tubal sterilization. METHODS: We conducted a case– control study of 18 –39year-old health maintenance organization enrollees with a functional ovarian cyst diagnosed between January 1, and June 30, 1994, and age-matched female controls randomly selected from enrollment files. In-person interviews as well as medical and pharmacy records were obtained for 78% of cases and 82% of controls; these analyses were based on 392 cases and 623 controls. Odds ratios (ORs) calculated with unconditional logistic regression were used to estimate the risk of a functional ovarian cyst diagnosis associated with current contraceptive method. RESULTS: In multivariable analyses adjusting for age, education, number of live births, and reference year, the overall OR was 0.72 (95% confidence interval [CI] 0.53, 0.99) for current OC use, compared with use of nonsurgical nonhormonal contraception or no contraception. The risk associated with use of 35 g ethinyl estradiol monophasic OCs (OR 0.69; 95% CI 0.44, 1.10) was slightly lower than that associated with less than 35 g ethinyl estradiol monophasic (OR 0.79; 95% CI 0.43, 1.47) or multiphasic OCs (OR 0.76; 95% CI 0.49, 1.19). Women with tubal sterilization had a substantially increased risk of a functional ovarian cyst diagnosis (OR 1.70; 95% CI 1.05, 2.75) compared with women using nonhormonal or no contraception. CONCLUSION: Our findings suggest that low-dose OC use has little or no effect on functional ovarian cyst likelihood. The increased risks we found associated with tubal sterilization merit further investigation. (Obstet Gynecol 2003; 102:252– 8. © 2003 by The American College of Obstetricians and Gynecologists.)
Functional ovarian cysts, although benign and often self-limited, may cause significant pelvic pain and menstrual dysfunction, and result in nearly 200,000 hospitalFrom the Department of Epidemiology, School of Public Health and Community Medicine, University of Washington; and Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington. Supported by grant 1 R01 HD-25959 from the National Institute of Child Health and Human Development.
252
izations annually among US reproductive-age women.1 Functional ovarian cyst risk has long been postulated to be affected by oral contraceptive (OC) use. Several studies conducted in the 1970s, when the steroid hormone content of OCs was higher than it is currently, found that the likelihood of a functional cyst diagnosis was markedly reduced among recent or current OC users.2–5 Results of more recent epidemiologic investigations of this issue have been mixed, with relative risks (RRs) associated with current monophasic or triphasic OC use ranging from 0.2 to 1.3.6,7 These two studies were limited by insufficient numbers of cases for conclusive findings, and neither assessed separately functional ovarian cyst risk associated with less than 35 g ethinyl estradiol and 35 g ethinyl estradiol OCs. Additionally, little information is available concerning other contraceptive methods in relation to functional ovarian cysts. A small Mexican cohort study indicated that tubal sterilization may be associated with increased cyst risk.8 If this method does increase the likelihood of functional ovarian cyst development, the protective effect of OCs seen in prior studies may be attributed, in part, to the inclusion in the comparison group of a subset of contraceptive users at relatively high disease risk. Our objective was to study whether current contraceptive method affects functional ovarian cyst risk. We conducted this interview-based, case– control study in a large health maintenance organization population of US women of reproductive age to provide new information about the risk of functional ovarian cysts associated with the multiphasic and very low-dose (less than 35 g ethinyl estradiol) monophasic OCs currently in widespread use, and to investigate further the relationships between the use of tubal sterilization and cyst risk.
MATERIALS AND METHODS This case– control study was conducted among enrollees of Group Health Cooperative, a large mixed model, consumer-owned health maintenance organization in
VOL. 102, NO. 2, AUGUST 2003 © 2003 by The American College of Obstetricians and Gynecologists. Published by Elsevier.
0029-7844/03/$30.00 doi:10.1016/S0029-7844(03)00572-6
western Washington State. We used Group Health Cooperative inpatient, outpatient, and radiology records to identify all 18 –39-year-old women with an ultrasound or surgical diagnosis of a functional ovarian cyst (International Classification of Diseases, 9th Revision, codes 620.0 – 620.2) 2 cm or greater in size between January 1, 1990, and June 30, 1994, excluding women diagnosed with theca-lutein cyst, cystadenoma, cystadenocarcinoma, polycystic ovaries, cystic teratoma, or endometrioma. Additional screening excluded potential patients who did not speak English, were pregnant, or had been enrolled in Group Health Cooperative less than 6 months at reference date, which was the date of cyst diagnosis. There were 750 eligible potential patients, 586 of whom agreed to participate in the study (78.1%). We used Group Health Cooperative enrollment lists to identify all 18 –39-year-old women who were enrollees between January 1, 1990, and June 30, 1994. We selected potential controls randomly from these lists, stratified by 5-year age groups (less than 20, 20 –24, 25–29, 30 –34, 35–39 years), with the number of controls selected in each age group proportional to the number of cases in that age group. Reference dates were assigned to potential controls according to the distribution of cases’ reference dates. After screening, we excluded potential control subjects if they did not speak English, had had a bilateral oophorectomy, were pregnant, or had been enrolled in Group Health Cooperative less than 6 months at reference date. There were 921 eligible potential control subjects, 757 of whom agreed to participate (82.2%). Study protocols were approved by the Fred Hutchinson Cancer Research Center and the Group Health Cooperative of Puget Sound human subjects review committees. For the analyses presented here, we excluded the following groups of women who were likely to have atypical ovarian function or contraception use: those with prior unilateral oophorectomy (37 cases, 13 controls) or hysterectomy (22 cases, eight controls) and those with infertility tests or treatment or 12 or more months of unprotected intercourse without conception (162 cases, 125 controls). Additionally, we excluded three cases and two controls with unreliable interview data. Some subjects met more than one exclusion criterion, and 392 cases (343 diagnosed by ultrasound and 49 surgically diagnosed) and 623 controls remained for these analyses. In 260 cases (66.3%), the presenting complaint at the visit at which the cyst was diagnosed was pelvic pain; in 35 cases (8.9%), other symptoms were the presenting complaint; and in the remaining 97 cases (24.7%), there were no symptoms, and the cyst was found incidentally.
VOL. 102, NO. 2, AUGUST 2003
After obtaining written informed consent, experienced female research interviewers who were Fred Hutchinson Cancer Research Center employees conducted in-person interviews in the participants’ homes or other location of participants’ choice, using a structured questionnaire that elicited a lifetime history of a variety of demographic, medical, and lifestyle factors, including contraceptive use, up to reference date. We used a life-events calendar and color photographs of all OC pills marketed in the United States to help participants recall dates and specific brands of OCs used. We defined current OC use (ie, reference month use) initially by subject self-report, excluding any OC use for treatment of cyst symptoms. We then confirmed self-reported OC use with Group Health Cooperative pharmacy data, resulting in categorization of 39 current OC users (11 cases, 28 controls) who did not know their current brand and dose, and recategorization of 21 current OC users (eight cases, 13 controls) whose current brand and dose in pharmacy data differed from interview data and who reported that they had not filled OC prescriptions outside of Group Health Cooperative. The exposures of interest in this study were current use of OCs or tubal sterilization, in comparison with use of nonsurgical, nonhormonal contraception (defined as condom, diaphragm, contraceptive foam or jelly, contraceptive sponge, and nonhormonal intrauterine devices), or no contraception (including withdrawal and periodic abstinence). Current OC use was subcategorized by type and dose as follows: greater than 35-g ethinyl estradiol monophasic, 35 g ethinyl estradiol monophasic, less than 35 g ethinyl estradiol monophasic, multiphasic (triphasic and biphasic), and progestin-only OCs. Tubal sterilization was defined as sterilization before reference date and was subcategorized as postpartum (occurring immediately after childbirth) or interval (occurring at any other time). Demographic differences between all functional ovarian cyst cases and controls were assessed using 2 tests (or Fisher exact test if any cell size was less than five), with P ⬍ .05 denoting significance. We used unconditional logistic regression to compute odds ratios (ORs) and 95% confidence intervals (CIs) for the associations between contraceptive method and functional ovarian cyst risk, with Stata Statistical Software 6.0 (Stata Corp., College Station, TX). We tested for the interaction of contraceptive method with smoking or prior functional ovarian cyst using the likelihood ratio test; no significant (P ⬍ .05) interaction was noted. We included potential confounding variables in multivariable models if they were matching variables or changed the OR for functional ovarian cyst risk associated with contraceptive
Holt et al
Contraception and Functional Ovarian Cysts
253
Table 1. Demographic, Behavioral, and Reproductive Characteristics of Functional Ovarian Cyst Cases and Controls at Reference Date Cases (n ⫽ 392) Age (y) ⬍20 20–24 25–29 30–34 35–39 Race/ethnicity White Black Asian Hispanic Native American Other Refused Marital status* Married Living as married Single Highest educational level* ⱕHigh school Some college College graduate Family income ($)* ⬍$15,000 15,000–29,999 30,000–45,000 ⬎45,000 Unknown/refused Cigarette smoking* Never Former Current Weight (kg)* ⬍56.5 56.5–⬍62.4 62.4–70.8 ⬎70.8 BMI (kg/m2)* ⬍20.8 20.8–⬍22.8 22.8–25.7 ⬎25.7 Pregnancies† 0 1 2 ⱖ3 Live births 0 1 ⱖ2 Induced abortions 0 1 ⱖ2
Controls (n ⫽ 623)
12 (3.1) 68 (17.3) 92 (23.5) 102 (26.0) 118 (30.1)
21 (3.4) 100 (16.1) 129 (20.7) 153 (24.6) 220 (35.3)
328 (83.7) 27 (6.9) 12 (3.1) 14 (3.6) 4 (1.0) 7 (1.8)
524 (84.2) 48 (7.7) 28 (4.5) 6 (1.0) 4 (0.6) 12 (1.9) 1
170 (43.4) 45 (11.5) 177 (45.2)
342 (54.9) 53 (8.5) 228 (36.6)
93 (23.7) 146 (37.2) 153 (39.0)
94 (15.1) 229 (36.8) 300 (48.2)
48 (12.3) 115 (29.6) 115 (29.6) 111 (28.5) 3
55 (8.9) 155 (25.2) 177 (28.7) 229 (37.2) 7
231 (58.9) 66 (16.8) 95 (24.2)
407 (65.3) 112 (18.0) 104 (16.7)
96 (24.5) 85 (21.7) 88 (22.4) 123 (31.4)
155 (24.9) 167 (26.8) 166 (26.6) 135 (21.7)
115 (29.3) 76 (19.4) 76 (19.4) 125 (31.9)
179 (28.7) 149 (23.9) 164 (26.3) 131 (21.0)
155 (39.5) 73 (18.6) 73 (18.6) 91 (23.2)
237 (38.0) 111 (17.8) 119 (19.1) 156 (25.0)
231 (58.9) 58 (14.8) 103 (26.3)
332 (53.3) 95 (15.2) 196 (31.5)
270 (68.9) 83 (21.2) 39 (9.9)
438 (70.3) 128 (20.5) 57 (9.1) (continued)
254
Holt et al
Table 1. Demographic, Behavioral, and Reproductive Characteristics of Functional Ovarian Cyst Cases and Controls at Reference Date (continued)
Contraception and Functional Ovarian Cysts
Spontaneous abortions† 0 1 ⱖ2 Ectopic pregnancies 0 ⱖ1 Stillbirths 0 ⱖ1
Cases (n ⫽ 392)
Controls (n ⫽ 623)
335 (85.5) 50 (12.8) 7 (1.8)
548 (88.0) 54 (8.7) 21 (3.4)
387 (98.7) 5 (1.3)
621 (99.7) 2 (0.3)
388 (99.0) 4 (1.0)
615 (98.7) 8 (1.3)
BMI ⫽ body mass index. Data are presented as n (%). * Significant difference by case– control status (P ⬍ .05). † Includes confirmed pregnancies only.
method by 10% or more.9 Age, education, number of live births, and reference year met these criteria. We carried out several analyses. To be consistent with prior research,6,7 in our primary analysis we defined cases as women diagnosed by ultrasound or surgical examination with functional ovarian cysts that were 2 cm or greater (mean cyst size 3.5 cm, range 2–17 cm) and we calculated the RRs associated with 1) current OC use and 2) tubal sterilization, compared with use of nonsurgical, nonhormonal contraception, or no contraception. Because not all diagnosed functional ovarian cysts indicate pathology or are of clinical importance, we also conducted a series of subanalyses limiting cases to those with more consequential cysts: 1) women with pelvic pain as the presenting complaint (n ⫽ 260), 2) women with cysts 5 cm or greater (n ⫽ 81), and 3) women with cysts that were surgically diagnosed and treated (n ⫽ 49). We conducted dose- and type-specific analyses of the associations between current OC use and diagnosis with a functional ovarian cyst 2 cm or greater, considering: 1) 35 g ethinyl estradiol monophasic OCs, 2) less than 35 g ethinyl estradiol monophasic, and 3) multiphasic OCs, compared with the use of nonsurgical, nonhormonal contraception, or no contraception. In these analyses, risks were not calculated for greater than 35 g ethinyl estradiol monophasic or progestin-only OCs because fewer than ten subjects were current users of either of these types of OCs. Similarly, we conducted additional analyses of the associations between tubal sterilization and risk of a functional ovarian cyst 2 cm or greater, considering separately postpartum and interval sterilization, and examining the effect of time since sterilization.
OBSTETRICS & GYNECOLOGY
Table 2. Risk of Functional Ovarian Cyst Diagnosis by Current Contraceptive Method
Nonsurgical, nonhormonal, or no contraception Oral contraceptives Tubal sterilization Other hormonal contraception†
Cases (n ⫽ 392)
Controls (n ⫽ 623)
OR*
95% CI
246 (62.8)
394 (63.2)
1.00
Reference
98 (25.0) 43 (11.0) 5 (1.3)
178 (28.6) 47 (7.5) 4 (0.6)
0.72 1.70
0.53, 0.99 1.05, 2.75
OR ⫽ odds ratio; CI ⫽ confidence interval. Data are presented as n (%). * Risks estimated by ORs using unconditional logistic regression controlling for subject age, education, number of live births, and reference year; compared with women using nonsurgical, nonhormonal contraception, or no contraception. † Includes contraceptive implants, injectable contraception, and hormonal intrauterine device.
RESULTS Women with functional ovarian cysts were less likely than control women to be married, have more than a high school education, and to have a household income over $45,000 (Table 1). Additionally, case subjects were more likely than control women to smoke, to weigh over 70.8 kg, and to have body mass index over 25.7 kg/m2. Reproductive histories varied somewhat by case status. Although the percentages of cases and controls with at least one confirmed pregnancy were quite similar (60.5% versus 62.0%), spontaneous abortion histories differed significantly. Over 87% of patients and 86% of controls had used OCs at some point in their lives. As shown in Table 2, current method of contraception varied by case status; patients were slightly less likely than controls to be current OC users (25.0% versus 28.6%), and more likely to have undergone tubal sterilization before reference date (11.0% versus 7.5%). In multivariable analyses controlling for age, education, number of live births, and reference year, the overall OR was 0.72 (95% CI 0.53, 0.99) for current OC use, compared with use of nonsurgical, nonhormonal contraception, or no contraception
(Table 2). Similar ORs were found associated with current OC use in subanalyses limited to patients presenting with pelvic pain (OR 0.72; 95% CI 0.50, 1.02) or patients diagnosed by surgery (OR 0.65; 95% CI 0.31, 1.38); the risk was somewhat greater (OR 0.86; 95% CI 0.50, 1.50) for patients with cysts 5 cm or greater in size. In multivariable analyses considering type and dose of OC, compared with use of nonsurgical, nonhormonal contraception, or no contraception, the functional ovarian cyst risk associated with use of 35 g ethinyl estradiol OCs (OR 0.69; 95% CI 0.44, 1.10; Table 3) was slightly lower than that associated with use of less than 35 g ethinyl estradiol monophasic (OR 0.79; 95% CI 0.43, 1.47) or multiphasic (OR 0.76; 95% CI 0.49, 1.19) OCs. In our primary analysis, tubal sterilization was associated with a substantially increased risk of functional ovarian cyst diagnosis (OR 1.70; 95% CI 1.05, 2.75). We found similar increases in risk associated with tubal sterilization in subanalyses limited to patients presenting with pelvic pain (OR 1.61; 95% CI 0.92, 2.85) or functional ovarian cysts 5 cm or greater (OR 1.47; 95% CI 0.55, 3.93), whereas the increase was greater for cases diagnosed by surgery (OR 3.15; 95% CI 1.13, 8.77).
Table 3. Risk of Functional Ovarian Cyst Diagnosis by Type of Currently Used Oral Contraceptives
Nonsurgical, nonhormonal, or no contraception Type of OC† ⬎35 g EE monophasic OC 35 g EE monophasic OC ⬍35 g EE monophasic OC Multiphasic OC Progestin-only OC
Cases (n ⫽ 392)
Controls (n ⫽ 623)
OR*
95% CI
246 (62.8)
394 (63.2)
1.00
Reference
2 (0.5) 34 (8.7) 19 (4.8) 40 (10.2) 1 (0.3)
2 (0.3) 67 (10.8) 30 (4.8) 67 (10.8) 3 (0.5)
0.69 0.79 0.76
0.44, 1.10 0.43, 1.47 0.49, 1.19
OC ⫽ oral contraceptive; EE ⫽ ethinylestradiol; other abbreviations as in Table 2. Data are presented as n (%). * Risks estimated by ORs using unconditional logistic regression controlling for subject age, education, number of live births, and reference year; compared with women using nonsurgical, nonhormonal contraception, or no contraception. Included in models but not shown above are women using tubal sterilization (43 cases, 47 controls) or other hormonal contraception (five cases, four controls). † Excludes users of unknown type OCs (two cases, nine controls).
VOL. 102, NO. 2, AUGUST 2003
Holt et al
Contraception and Functional Ovarian Cysts
255
Table 4. Risk of Functional Ovarian Cyst Diagnosis by Aspects of Tubal Sterilization
Nonsurgical, nonhormonal, or no contraception Type of sterilization Postpartum sterilization Interval sterilization Time since sterilization (y) ⬍5 ⱖ5
Cases (n ⫽ 392)
Controls (n ⫽ 623)
OR*
95% CI
246 (62.8)
394 (63.2)
1.00
Reference
12 (3.1) 31 (7.9)
10 (1.6) 37 (5.9)
2.55 1.50
1.04, 6.28 0.88, 2.57
17 (4.3) 26 (6.6)
18 (2.9) 29 (4.7)
1.73 1.68
0.85, 3.55 0.93, 3.05
Abbreviations as in Table 2. Data are presented as n (%). * Risks estimated by ORs using unconditional logistic regression controlling for subject age, education, number of live births, and reference year; compared with women using nonsurgical, nonhormonal contraception, or no contraception. Included in models but not shown above are women using oral contraceptives (98 cases, 178 controls) or other hormonal contraception (five cases, four controls).
Further investigation of the associations between tubal sterilization and overall cyst risk revealed that, compared with women using nonsurgical, nonhormonal contraception, or no contraception, risk elevation was higher among women who had undergone postpartum tubal sterilization (OR 2.55; 95% CI 1.04, 6.28; Table 4), and higher yet for those with postpartum sterilization at under age 30 (OR 3.54; 95% CI 1.24, 10.11; data not shown). Time since sterilization did not affect the tubal sterilization-associated increases in cyst risk (Table 4).
DISCUSSION In this large population-based, interview study, current OC use was associated overall with a modest decrease in risk of a functional ovarian cyst diagnosis. The 28% decrease in risk we found is smaller than the 40 –90% reductions in cyst occurrence among current or recent OC users reported in studies conducted in the 1970s.2–5 Although it has been hypothesized that the lower dose pills currently prescribed may have a diminished protective effect on cyst formation,10,11 the two epidemiologic studies that addressed this issue in the 1980s were inconclusive.6,7 In the first of these, a medical record review case– control study of 90 patients from the same health maintenance organization as the current study, Holt et al found a nonsignificant 20% decrease in functional ovarian cyst risk associated with current prescription of monophasic OCs of all ethinyl estradiol doses combined (RR 0.8; 95% CI 0.4, 1.8), compared with women without a current OC prescription.6 In that study, conducted when the average ethinyl estradiol dose in monophasic pills was higher than in the current study, the authors could not eliminate from the exposed group women who were prescribed OCs for treatment of cyst symptoms or determine definitively whether pre-
256
Holt et al
Contraception and Functional Ovarian Cysts
scribed pills were taken. Exposure measurement was likely to be more accurate in the current study, as all study subjects were interviewed, and details of and reasons for OC use were ascertained. The other previous epidemiologic study of this issue, a record-linkage cohort study of Maine Medicaid-recipient OC users with 32 cases, assessed the effects of OCs by dose and type. Lanes et al found large, although nonsignificant, decreases in functional ovarian cyst risk among women currently prescribed greater than 35 g ethinyl estradiol (RR 0.2; 95% CI 0.01, 1.3) or 35 g or less ethinyl estradiol OCs (RR 0.5; 95% CI 0.2, 1.3), compared with former OC users who had not filled an OC prescription for at least 3 months.7 In the current study, all but four current OC users used 35 g or less ethinyl estradiol OCs, precluding investigation of greater than 35 g ethinyl estradiol formulations, but enabling us to focus in more detail on the risks associated with very low-dose pills. We found that women using 35 g ethinyl estradiol monophasic OCs had a slightly lower OR than less than 35 g ethinyl estradiol monophasic or multiphasic OC users, among whom current OC use had no effect on functional ovarian cyst likelihood. The potential for increased cyst risk associated with current multiphasic OC use was first raised in the 1980s and has been of some interest since then.12,13 Neither of the abovedescribed epidemiologic studies found a significant increase in functional ovarian cyst risk associated with multiphasic OC use, but because of small sample sizes neither could rule out as much as a tripling of risk among women with this exposure.6,7 The current study, with nearly 400 cases and over 600 controls, allowed us to investigate with much more precision the associations between multiphasic OCs and functional ovarian cyst risk and rule out any substantial increase in risk associated with their use.
OBSTETRICS & GYNECOLOGY
In studies of diseases that may, in some circumstances, remain undiagnosed, the possibility exists that the exposure of interest affects diagnostic likelihood in the absence of a true effect on disease likelihood. For instance, if OC users are under more frequent medical surveillance than users of nonhormonal contraceptive methods, increased diagnostic opportunity may lead to the identification of incidental, asymptomatic cysts among OC users that may remain undiagnosed in nonusers, resulting in a spurious increase in the OR. The absence of a positive association between current OC use and cyst likelihood in our analyses provides some reassurance that this diagnostic bias did not exist. Additionally, we addressed the possibility that more complete diagnosis among OC users masked a truly substantial protective effect by including in one analysis only patients who were not diagnosed incidentally, that is, those who presented with pelvic pain. In this analysis, the association between current OC use and cyst risk was unchanged. We also found that the associations between current OC use and functional ovarian cyst risk were similar in other analyses limited to cysts whose characteristics may indicate more consequential pathology, further reassuring us that diagnostic bias was not responsible for our findings. Case– control studies that rely upon participant selfreport of past exposures may also be subject to recall bias, as it is plausible that case subjects could be more or less likely than controls to recall or report exposures they think might be relevant to the etiology of their condition. In this study, we used the Group Health Cooperative pharmacy database to confirm reference date OC use, creating an objective, prospective assessment of exposure, and eliminating the possibility of recall bias for OC use. We had no independent confirmation of selfreported tubal sterilization status, but the lack of a known association between tubal sterilization and functional ovarian cysts and the surgical nature of the procedure make it unlikely that substantial differential errors in recall existed for this exposure. One difference between our study and previous investigations of this issue was our use of a reference group consisting only of women currently using nonsurgical, nonhormonal contraception, or no contraception. If noncontraceptive using women are atypical in factors associated with functional ovarian cyst likelihood (such as infertility), their inclusion in the control group may bias study results. Consequently, we excluded from all analyses women with a history of infertility tests or treatment or at least 12 months of unprotected intercourse without conception. The major advantage of our reference group was that it allowed us to determine separately the risks associated with hormonal and surgical contraception, addressing the possibility that the pro-
VOL. 102, NO. 2, AUGUST 2003
tective effect associated with OCs seen in previous studies was in part a result of a heterogeneous reference group that included a subset of women with high disease risk, namely women with tubal sterilization. To our knowledge, only one prior study has examined the relationship between tubal sterilization and functional ovarian cyst risk. Similar to our findings, a Mexican cohort study by de Alba Quintanilla reported that twice as many sterilized women developed cysts as did women who had not been sterilized (24.1% versus 12.4%).8 Additionally, we found that the sterilizationassociated increases in functional ovarian cyst risk persisted or increased in subanalyses limited to women with painful or large cysts or those who were surgically diagnosed. Because tubal sterilization is the most commonly used contraceptive method in the United States,14 this association merits further investigation in larger studies including women 40 years and older who are more likely to have this exposure.
REFERENCES 1. Velebil P, Wingo PA, Xia Z, Wilcox LS, Peterson HB. Rate of hospitalization for gynecologic disorders among reproductive-age women in the United States. Obstet Gynecol 1995;86:764–9. 2. Royal College of General Practitioners. Oral contraceptives and health. Tunbridge Wells, United Kingdom: Pitman Medical Publishing, 1974. 3. Ory H. Functional ovarian cysts and oral contraceptives: Negative association confirmed surgically. JAMA 1974; 228:68–9. 4. Ramcharan S, Pellegrin FA, Ray R, Hsu JP. The Walnut Creek Contraceptive Drug Study. Vol III. Bethesda, Maryland: Center for Population Research, National Institute of Child Health and Human Development, United States Department of Health and Human Services, 1981. 5. Vessey M, Metcalfe A, Wells C, McPherson K, Westhoff C, Yeates D. Ovarian neoplasms, functional ovarian cysts, and oral contraceptives. Br Med J 1987;294:1518–22. 6. Holt VL, Daling JR, McKnight B, Moore D, Stergachis A, Weiss NS. Functional ovarian cysts in relation to the use of monophasic and triphasic oral contraceptives. Obstet Gynecol 1992;79:529–33. 7. Lanes SF, Birmann B, Walker AM, Singer S. Oral contraceptive type and functional ovarian cysts. Am J Obstet Gynecol 1992;166:956–61. 8. de Alba Quintanilla F. Functional ovary cysts in patients with and without tubal sterilization. Gin Obs Mex 2000; 68:345–8. 9. Maldonado G, Greenland S. Simulation study of confounder selection strategies. Am J Epidemiol 1993;138: 923–36.
Holt et al
Contraception and Functional Ovarian Cysts
257
10. Chiaffarino F, Parazzini F, La Vecchia C, Ricci E, Crosignani PG. Oral contraceptive use and benign gynecologic conditions. Contraception 1998;57:11–8. 11. Burkman RT. Oral contraceptives: Current status. Clin Obstet Gynecol 2001;44:62–72. 12. Caillouette JC, Koehler AL. Phasic contraceptive pills and functional ovarian cysts. Am J Obstet Gynecol 1987;156: 1538–42. 13. Grimes DA, Hughes JM. Use of multiphasic oral contraceptives and hospitalizations of women with functional ovarian cysts in the United States. Obstet Gynecol 1989; 73:1037–9.
258
Holt et al
Contraception and Functional Ovarian Cysts
14. Abma J, Chandra A, Mosher W, Peterson L, Piccinino LJ. Fertility, family planning, and women’s health: New data from the 1995 National Survey of Family Growth. Vital Health Stat 23 1997;19:1–114. Address reprint requests to: Victoria L. Holt, PhD, MPH, Fred Hutchinson Cancer Research Center, Division of Public Health Sciences (MP 474), 1100 Fairview Avenue N., PO Box 19024, Seattle, WA 98109; E-mail: [email protected]. Received January 8, 2003. Received in revised form March 17, 2003. Accepted April 24, 2003.
OBSTETRICS & GYNECOLOGY
Functional ovarian cysts, although benign and often self-limited, may cause significant pelvic pain and menstrual dysfunction, and result in nearly 200,000 hospitalFrom the Department of Epidemiology, School of Public Health and Community Medicine, University of Washington; and Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington. Supported by grant 1 R01 HD-25959 from the National Institute of Child Health and Human Development.
252
izations annually among US reproductive-age women.1 Functional ovarian cyst risk has long been postulated to be affected by oral contraceptive (OC) use. Several studies conducted in the 1970s, when the steroid hormone content of OCs was higher than it is currently, found that the likelihood of a functional cyst diagnosis was markedly reduced among recent or current OC users.2–5 Results of more recent epidemiologic investigations of this issue have been mixed, with relative risks (RRs) associated with current monophasic or triphasic OC use ranging from 0.2 to 1.3.6,7 These two studies were limited by insufficient numbers of cases for conclusive findings, and neither assessed separately functional ovarian cyst risk associated with less than 35 g ethinyl estradiol and 35 g ethinyl estradiol OCs. Additionally, little information is available concerning other contraceptive methods in relation to functional ovarian cysts. A small Mexican cohort study indicated that tubal sterilization may be associated with increased cyst risk.8 If this method does increase the likelihood of functional ovarian cyst development, the protective effect of OCs seen in prior studies may be attributed, in part, to the inclusion in the comparison group of a subset of contraceptive users at relatively high disease risk. Our objective was to study whether current contraceptive method affects functional ovarian cyst risk. We conducted this interview-based, case– control study in a large health maintenance organization population of US women of reproductive age to provide new information about the risk of functional ovarian cysts associated with the multiphasic and very low-dose (less than 35 g ethinyl estradiol) monophasic OCs currently in widespread use, and to investigate further the relationships between the use of tubal sterilization and cyst risk.
MATERIALS AND METHODS This case– control study was conducted among enrollees of Group Health Cooperative, a large mixed model, consumer-owned health maintenance organization in
VOL. 102, NO. 2, AUGUST 2003 © 2003 by The American College of Obstetricians and Gynecologists. Published by Elsevier.
0029-7844/03/$30.00 doi:10.1016/S0029-7844(03)00572-6
western Washington State. We used Group Health Cooperative inpatient, outpatient, and radiology records to identify all 18 –39-year-old women with an ultrasound or surgical diagnosis of a functional ovarian cyst (International Classification of Diseases, 9th Revision, codes 620.0 – 620.2) 2 cm or greater in size between January 1, 1990, and June 30, 1994, excluding women diagnosed with theca-lutein cyst, cystadenoma, cystadenocarcinoma, polycystic ovaries, cystic teratoma, or endometrioma. Additional screening excluded potential patients who did not speak English, were pregnant, or had been enrolled in Group Health Cooperative less than 6 months at reference date, which was the date of cyst diagnosis. There were 750 eligible potential patients, 586 of whom agreed to participate in the study (78.1%). We used Group Health Cooperative enrollment lists to identify all 18 –39-year-old women who were enrollees between January 1, 1990, and June 30, 1994. We selected potential controls randomly from these lists, stratified by 5-year age groups (less than 20, 20 –24, 25–29, 30 –34, 35–39 years), with the number of controls selected in each age group proportional to the number of cases in that age group. Reference dates were assigned to potential controls according to the distribution of cases’ reference dates. After screening, we excluded potential control subjects if they did not speak English, had had a bilateral oophorectomy, were pregnant, or had been enrolled in Group Health Cooperative less than 6 months at reference date. There were 921 eligible potential control subjects, 757 of whom agreed to participate (82.2%). Study protocols were approved by the Fred Hutchinson Cancer Research Center and the Group Health Cooperative of Puget Sound human subjects review committees. For the analyses presented here, we excluded the following groups of women who were likely to have atypical ovarian function or contraception use: those with prior unilateral oophorectomy (37 cases, 13 controls) or hysterectomy (22 cases, eight controls) and those with infertility tests or treatment or 12 or more months of unprotected intercourse without conception (162 cases, 125 controls). Additionally, we excluded three cases and two controls with unreliable interview data. Some subjects met more than one exclusion criterion, and 392 cases (343 diagnosed by ultrasound and 49 surgically diagnosed) and 623 controls remained for these analyses. In 260 cases (66.3%), the presenting complaint at the visit at which the cyst was diagnosed was pelvic pain; in 35 cases (8.9%), other symptoms were the presenting complaint; and in the remaining 97 cases (24.7%), there were no symptoms, and the cyst was found incidentally.
VOL. 102, NO. 2, AUGUST 2003
After obtaining written informed consent, experienced female research interviewers who were Fred Hutchinson Cancer Research Center employees conducted in-person interviews in the participants’ homes or other location of participants’ choice, using a structured questionnaire that elicited a lifetime history of a variety of demographic, medical, and lifestyle factors, including contraceptive use, up to reference date. We used a life-events calendar and color photographs of all OC pills marketed in the United States to help participants recall dates and specific brands of OCs used. We defined current OC use (ie, reference month use) initially by subject self-report, excluding any OC use for treatment of cyst symptoms. We then confirmed self-reported OC use with Group Health Cooperative pharmacy data, resulting in categorization of 39 current OC users (11 cases, 28 controls) who did not know their current brand and dose, and recategorization of 21 current OC users (eight cases, 13 controls) whose current brand and dose in pharmacy data differed from interview data and who reported that they had not filled OC prescriptions outside of Group Health Cooperative. The exposures of interest in this study were current use of OCs or tubal sterilization, in comparison with use of nonsurgical, nonhormonal contraception (defined as condom, diaphragm, contraceptive foam or jelly, contraceptive sponge, and nonhormonal intrauterine devices), or no contraception (including withdrawal and periodic abstinence). Current OC use was subcategorized by type and dose as follows: greater than 35-g ethinyl estradiol monophasic, 35 g ethinyl estradiol monophasic, less than 35 g ethinyl estradiol monophasic, multiphasic (triphasic and biphasic), and progestin-only OCs. Tubal sterilization was defined as sterilization before reference date and was subcategorized as postpartum (occurring immediately after childbirth) or interval (occurring at any other time). Demographic differences between all functional ovarian cyst cases and controls were assessed using 2 tests (or Fisher exact test if any cell size was less than five), with P ⬍ .05 denoting significance. We used unconditional logistic regression to compute odds ratios (ORs) and 95% confidence intervals (CIs) for the associations between contraceptive method and functional ovarian cyst risk, with Stata Statistical Software 6.0 (Stata Corp., College Station, TX). We tested for the interaction of contraceptive method with smoking or prior functional ovarian cyst using the likelihood ratio test; no significant (P ⬍ .05) interaction was noted. We included potential confounding variables in multivariable models if they were matching variables or changed the OR for functional ovarian cyst risk associated with contraceptive
Holt et al
Contraception and Functional Ovarian Cysts
253
Table 1. Demographic, Behavioral, and Reproductive Characteristics of Functional Ovarian Cyst Cases and Controls at Reference Date Cases (n ⫽ 392) Age (y) ⬍20 20–24 25–29 30–34 35–39 Race/ethnicity White Black Asian Hispanic Native American Other Refused Marital status* Married Living as married Single Highest educational level* ⱕHigh school Some college College graduate Family income ($)* ⬍$15,000 15,000–29,999 30,000–45,000 ⬎45,000 Unknown/refused Cigarette smoking* Never Former Current Weight (kg)* ⬍56.5 56.5–⬍62.4 62.4–70.8 ⬎70.8 BMI (kg/m2)* ⬍20.8 20.8–⬍22.8 22.8–25.7 ⬎25.7 Pregnancies† 0 1 2 ⱖ3 Live births 0 1 ⱖ2 Induced abortions 0 1 ⱖ2
Controls (n ⫽ 623)
12 (3.1) 68 (17.3) 92 (23.5) 102 (26.0) 118 (30.1)
21 (3.4) 100 (16.1) 129 (20.7) 153 (24.6) 220 (35.3)
328 (83.7) 27 (6.9) 12 (3.1) 14 (3.6) 4 (1.0) 7 (1.8)
524 (84.2) 48 (7.7) 28 (4.5) 6 (1.0) 4 (0.6) 12 (1.9) 1
170 (43.4) 45 (11.5) 177 (45.2)
342 (54.9) 53 (8.5) 228 (36.6)
93 (23.7) 146 (37.2) 153 (39.0)
94 (15.1) 229 (36.8) 300 (48.2)
48 (12.3) 115 (29.6) 115 (29.6) 111 (28.5) 3
55 (8.9) 155 (25.2) 177 (28.7) 229 (37.2) 7
231 (58.9) 66 (16.8) 95 (24.2)
407 (65.3) 112 (18.0) 104 (16.7)
96 (24.5) 85 (21.7) 88 (22.4) 123 (31.4)
155 (24.9) 167 (26.8) 166 (26.6) 135 (21.7)
115 (29.3) 76 (19.4) 76 (19.4) 125 (31.9)
179 (28.7) 149 (23.9) 164 (26.3) 131 (21.0)
155 (39.5) 73 (18.6) 73 (18.6) 91 (23.2)
237 (38.0) 111 (17.8) 119 (19.1) 156 (25.0)
231 (58.9) 58 (14.8) 103 (26.3)
332 (53.3) 95 (15.2) 196 (31.5)
270 (68.9) 83 (21.2) 39 (9.9)
438 (70.3) 128 (20.5) 57 (9.1) (continued)
254
Holt et al
Table 1. Demographic, Behavioral, and Reproductive Characteristics of Functional Ovarian Cyst Cases and Controls at Reference Date (continued)
Contraception and Functional Ovarian Cysts
Spontaneous abortions† 0 1 ⱖ2 Ectopic pregnancies 0 ⱖ1 Stillbirths 0 ⱖ1
Cases (n ⫽ 392)
Controls (n ⫽ 623)
335 (85.5) 50 (12.8) 7 (1.8)
548 (88.0) 54 (8.7) 21 (3.4)
387 (98.7) 5 (1.3)
621 (99.7) 2 (0.3)
388 (99.0) 4 (1.0)
615 (98.7) 8 (1.3)
BMI ⫽ body mass index. Data are presented as n (%). * Significant difference by case– control status (P ⬍ .05). † Includes confirmed pregnancies only.
method by 10% or more.9 Age, education, number of live births, and reference year met these criteria. We carried out several analyses. To be consistent with prior research,6,7 in our primary analysis we defined cases as women diagnosed by ultrasound or surgical examination with functional ovarian cysts that were 2 cm or greater (mean cyst size 3.5 cm, range 2–17 cm) and we calculated the RRs associated with 1) current OC use and 2) tubal sterilization, compared with use of nonsurgical, nonhormonal contraception, or no contraception. Because not all diagnosed functional ovarian cysts indicate pathology or are of clinical importance, we also conducted a series of subanalyses limiting cases to those with more consequential cysts: 1) women with pelvic pain as the presenting complaint (n ⫽ 260), 2) women with cysts 5 cm or greater (n ⫽ 81), and 3) women with cysts that were surgically diagnosed and treated (n ⫽ 49). We conducted dose- and type-specific analyses of the associations between current OC use and diagnosis with a functional ovarian cyst 2 cm or greater, considering: 1) 35 g ethinyl estradiol monophasic OCs, 2) less than 35 g ethinyl estradiol monophasic, and 3) multiphasic OCs, compared with the use of nonsurgical, nonhormonal contraception, or no contraception. In these analyses, risks were not calculated for greater than 35 g ethinyl estradiol monophasic or progestin-only OCs because fewer than ten subjects were current users of either of these types of OCs. Similarly, we conducted additional analyses of the associations between tubal sterilization and risk of a functional ovarian cyst 2 cm or greater, considering separately postpartum and interval sterilization, and examining the effect of time since sterilization.
OBSTETRICS & GYNECOLOGY
Table 2. Risk of Functional Ovarian Cyst Diagnosis by Current Contraceptive Method
Nonsurgical, nonhormonal, or no contraception Oral contraceptives Tubal sterilization Other hormonal contraception†
Cases (n ⫽ 392)
Controls (n ⫽ 623)
OR*
95% CI
246 (62.8)
394 (63.2)
1.00
Reference
98 (25.0) 43 (11.0) 5 (1.3)
178 (28.6) 47 (7.5) 4 (0.6)
0.72 1.70
0.53, 0.99 1.05, 2.75
OR ⫽ odds ratio; CI ⫽ confidence interval. Data are presented as n (%). * Risks estimated by ORs using unconditional logistic regression controlling for subject age, education, number of live births, and reference year; compared with women using nonsurgical, nonhormonal contraception, or no contraception. † Includes contraceptive implants, injectable contraception, and hormonal intrauterine device.
RESULTS Women with functional ovarian cysts were less likely than control women to be married, have more than a high school education, and to have a household income over $45,000 (Table 1). Additionally, case subjects were more likely than control women to smoke, to weigh over 70.8 kg, and to have body mass index over 25.7 kg/m2. Reproductive histories varied somewhat by case status. Although the percentages of cases and controls with at least one confirmed pregnancy were quite similar (60.5% versus 62.0%), spontaneous abortion histories differed significantly. Over 87% of patients and 86% of controls had used OCs at some point in their lives. As shown in Table 2, current method of contraception varied by case status; patients were slightly less likely than controls to be current OC users (25.0% versus 28.6%), and more likely to have undergone tubal sterilization before reference date (11.0% versus 7.5%). In multivariable analyses controlling for age, education, number of live births, and reference year, the overall OR was 0.72 (95% CI 0.53, 0.99) for current OC use, compared with use of nonsurgical, nonhormonal contraception, or no contraception
(Table 2). Similar ORs were found associated with current OC use in subanalyses limited to patients presenting with pelvic pain (OR 0.72; 95% CI 0.50, 1.02) or patients diagnosed by surgery (OR 0.65; 95% CI 0.31, 1.38); the risk was somewhat greater (OR 0.86; 95% CI 0.50, 1.50) for patients with cysts 5 cm or greater in size. In multivariable analyses considering type and dose of OC, compared with use of nonsurgical, nonhormonal contraception, or no contraception, the functional ovarian cyst risk associated with use of 35 g ethinyl estradiol OCs (OR 0.69; 95% CI 0.44, 1.10; Table 3) was slightly lower than that associated with use of less than 35 g ethinyl estradiol monophasic (OR 0.79; 95% CI 0.43, 1.47) or multiphasic (OR 0.76; 95% CI 0.49, 1.19) OCs. In our primary analysis, tubal sterilization was associated with a substantially increased risk of functional ovarian cyst diagnosis (OR 1.70; 95% CI 1.05, 2.75). We found similar increases in risk associated with tubal sterilization in subanalyses limited to patients presenting with pelvic pain (OR 1.61; 95% CI 0.92, 2.85) or functional ovarian cysts 5 cm or greater (OR 1.47; 95% CI 0.55, 3.93), whereas the increase was greater for cases diagnosed by surgery (OR 3.15; 95% CI 1.13, 8.77).
Table 3. Risk of Functional Ovarian Cyst Diagnosis by Type of Currently Used Oral Contraceptives
Nonsurgical, nonhormonal, or no contraception Type of OC† ⬎35 g EE monophasic OC 35 g EE monophasic OC ⬍35 g EE monophasic OC Multiphasic OC Progestin-only OC
Cases (n ⫽ 392)
Controls (n ⫽ 623)
OR*
95% CI
246 (62.8)
394 (63.2)
1.00
Reference
2 (0.5) 34 (8.7) 19 (4.8) 40 (10.2) 1 (0.3)
2 (0.3) 67 (10.8) 30 (4.8) 67 (10.8) 3 (0.5)
0.69 0.79 0.76
0.44, 1.10 0.43, 1.47 0.49, 1.19
OC ⫽ oral contraceptive; EE ⫽ ethinylestradiol; other abbreviations as in Table 2. Data are presented as n (%). * Risks estimated by ORs using unconditional logistic regression controlling for subject age, education, number of live births, and reference year; compared with women using nonsurgical, nonhormonal contraception, or no contraception. Included in models but not shown above are women using tubal sterilization (43 cases, 47 controls) or other hormonal contraception (five cases, four controls). † Excludes users of unknown type OCs (two cases, nine controls).
VOL. 102, NO. 2, AUGUST 2003
Holt et al
Contraception and Functional Ovarian Cysts
255
Table 4. Risk of Functional Ovarian Cyst Diagnosis by Aspects of Tubal Sterilization
Nonsurgical, nonhormonal, or no contraception Type of sterilization Postpartum sterilization Interval sterilization Time since sterilization (y) ⬍5 ⱖ5
Cases (n ⫽ 392)
Controls (n ⫽ 623)
OR*
95% CI
246 (62.8)
394 (63.2)
1.00
Reference
12 (3.1) 31 (7.9)
10 (1.6) 37 (5.9)
2.55 1.50
1.04, 6.28 0.88, 2.57
17 (4.3) 26 (6.6)
18 (2.9) 29 (4.7)
1.73 1.68
0.85, 3.55 0.93, 3.05
Abbreviations as in Table 2. Data are presented as n (%). * Risks estimated by ORs using unconditional logistic regression controlling for subject age, education, number of live births, and reference year; compared with women using nonsurgical, nonhormonal contraception, or no contraception. Included in models but not shown above are women using oral contraceptives (98 cases, 178 controls) or other hormonal contraception (five cases, four controls).
Further investigation of the associations between tubal sterilization and overall cyst risk revealed that, compared with women using nonsurgical, nonhormonal contraception, or no contraception, risk elevation was higher among women who had undergone postpartum tubal sterilization (OR 2.55; 95% CI 1.04, 6.28; Table 4), and higher yet for those with postpartum sterilization at under age 30 (OR 3.54; 95% CI 1.24, 10.11; data not shown). Time since sterilization did not affect the tubal sterilization-associated increases in cyst risk (Table 4).
DISCUSSION In this large population-based, interview study, current OC use was associated overall with a modest decrease in risk of a functional ovarian cyst diagnosis. The 28% decrease in risk we found is smaller than the 40 –90% reductions in cyst occurrence among current or recent OC users reported in studies conducted in the 1970s.2–5 Although it has been hypothesized that the lower dose pills currently prescribed may have a diminished protective effect on cyst formation,10,11 the two epidemiologic studies that addressed this issue in the 1980s were inconclusive.6,7 In the first of these, a medical record review case– control study of 90 patients from the same health maintenance organization as the current study, Holt et al found a nonsignificant 20% decrease in functional ovarian cyst risk associated with current prescription of monophasic OCs of all ethinyl estradiol doses combined (RR 0.8; 95% CI 0.4, 1.8), compared with women without a current OC prescription.6 In that study, conducted when the average ethinyl estradiol dose in monophasic pills was higher than in the current study, the authors could not eliminate from the exposed group women who were prescribed OCs for treatment of cyst symptoms or determine definitively whether pre-
256
Holt et al
Contraception and Functional Ovarian Cysts
scribed pills were taken. Exposure measurement was likely to be more accurate in the current study, as all study subjects were interviewed, and details of and reasons for OC use were ascertained. The other previous epidemiologic study of this issue, a record-linkage cohort study of Maine Medicaid-recipient OC users with 32 cases, assessed the effects of OCs by dose and type. Lanes et al found large, although nonsignificant, decreases in functional ovarian cyst risk among women currently prescribed greater than 35 g ethinyl estradiol (RR 0.2; 95% CI 0.01, 1.3) or 35 g or less ethinyl estradiol OCs (RR 0.5; 95% CI 0.2, 1.3), compared with former OC users who had not filled an OC prescription for at least 3 months.7 In the current study, all but four current OC users used 35 g or less ethinyl estradiol OCs, precluding investigation of greater than 35 g ethinyl estradiol formulations, but enabling us to focus in more detail on the risks associated with very low-dose pills. We found that women using 35 g ethinyl estradiol monophasic OCs had a slightly lower OR than less than 35 g ethinyl estradiol monophasic or multiphasic OC users, among whom current OC use had no effect on functional ovarian cyst likelihood. The potential for increased cyst risk associated with current multiphasic OC use was first raised in the 1980s and has been of some interest since then.12,13 Neither of the abovedescribed epidemiologic studies found a significant increase in functional ovarian cyst risk associated with multiphasic OC use, but because of small sample sizes neither could rule out as much as a tripling of risk among women with this exposure.6,7 The current study, with nearly 400 cases and over 600 controls, allowed us to investigate with much more precision the associations between multiphasic OCs and functional ovarian cyst risk and rule out any substantial increase in risk associated with their use.
OBSTETRICS & GYNECOLOGY
In studies of diseases that may, in some circumstances, remain undiagnosed, the possibility exists that the exposure of interest affects diagnostic likelihood in the absence of a true effect on disease likelihood. For instance, if OC users are under more frequent medical surveillance than users of nonhormonal contraceptive methods, increased diagnostic opportunity may lead to the identification of incidental, asymptomatic cysts among OC users that may remain undiagnosed in nonusers, resulting in a spurious increase in the OR. The absence of a positive association between current OC use and cyst likelihood in our analyses provides some reassurance that this diagnostic bias did not exist. Additionally, we addressed the possibility that more complete diagnosis among OC users masked a truly substantial protective effect by including in one analysis only patients who were not diagnosed incidentally, that is, those who presented with pelvic pain. In this analysis, the association between current OC use and cyst risk was unchanged. We also found that the associations between current OC use and functional ovarian cyst risk were similar in other analyses limited to cysts whose characteristics may indicate more consequential pathology, further reassuring us that diagnostic bias was not responsible for our findings. Case– control studies that rely upon participant selfreport of past exposures may also be subject to recall bias, as it is plausible that case subjects could be more or less likely than controls to recall or report exposures they think might be relevant to the etiology of their condition. In this study, we used the Group Health Cooperative pharmacy database to confirm reference date OC use, creating an objective, prospective assessment of exposure, and eliminating the possibility of recall bias for OC use. We had no independent confirmation of selfreported tubal sterilization status, but the lack of a known association between tubal sterilization and functional ovarian cysts and the surgical nature of the procedure make it unlikely that substantial differential errors in recall existed for this exposure. One difference between our study and previous investigations of this issue was our use of a reference group consisting only of women currently using nonsurgical, nonhormonal contraception, or no contraception. If noncontraceptive using women are atypical in factors associated with functional ovarian cyst likelihood (such as infertility), their inclusion in the control group may bias study results. Consequently, we excluded from all analyses women with a history of infertility tests or treatment or at least 12 months of unprotected intercourse without conception. The major advantage of our reference group was that it allowed us to determine separately the risks associated with hormonal and surgical contraception, addressing the possibility that the pro-
VOL. 102, NO. 2, AUGUST 2003
tective effect associated with OCs seen in previous studies was in part a result of a heterogeneous reference group that included a subset of women with high disease risk, namely women with tubal sterilization. To our knowledge, only one prior study has examined the relationship between tubal sterilization and functional ovarian cyst risk. Similar to our findings, a Mexican cohort study by de Alba Quintanilla reported that twice as many sterilized women developed cysts as did women who had not been sterilized (24.1% versus 12.4%).8 Additionally, we found that the sterilizationassociated increases in functional ovarian cyst risk persisted or increased in subanalyses limited to women with painful or large cysts or those who were surgically diagnosed. Because tubal sterilization is the most commonly used contraceptive method in the United States,14 this association merits further investigation in larger studies including women 40 years and older who are more likely to have this exposure.
REFERENCES 1. Velebil P, Wingo PA, Xia Z, Wilcox LS, Peterson HB. Rate of hospitalization for gynecologic disorders among reproductive-age women in the United States. Obstet Gynecol 1995;86:764–9. 2. Royal College of General Practitioners. Oral contraceptives and health. Tunbridge Wells, United Kingdom: Pitman Medical Publishing, 1974. 3. Ory H. Functional ovarian cysts and oral contraceptives: Negative association confirmed surgically. JAMA 1974; 228:68–9. 4. Ramcharan S, Pellegrin FA, Ray R, Hsu JP. The Walnut Creek Contraceptive Drug Study. Vol III. Bethesda, Maryland: Center for Population Research, National Institute of Child Health and Human Development, United States Department of Health and Human Services, 1981. 5. Vessey M, Metcalfe A, Wells C, McPherson K, Westhoff C, Yeates D. Ovarian neoplasms, functional ovarian cysts, and oral contraceptives. Br Med J 1987;294:1518–22. 6. Holt VL, Daling JR, McKnight B, Moore D, Stergachis A, Weiss NS. Functional ovarian cysts in relation to the use of monophasic and triphasic oral contraceptives. Obstet Gynecol 1992;79:529–33. 7. Lanes SF, Birmann B, Walker AM, Singer S. Oral contraceptive type and functional ovarian cysts. Am J Obstet Gynecol 1992;166:956–61. 8. de Alba Quintanilla F. Functional ovary cysts in patients with and without tubal sterilization. Gin Obs Mex 2000; 68:345–8. 9. Maldonado G, Greenland S. Simulation study of confounder selection strategies. Am J Epidemiol 1993;138: 923–36.
Holt et al
Contraception and Functional Ovarian Cysts
257
10. Chiaffarino F, Parazzini F, La Vecchia C, Ricci E, Crosignani PG. Oral contraceptive use and benign gynecologic conditions. Contraception 1998;57:11–8. 11. Burkman RT. Oral contraceptives: Current status. Clin Obstet Gynecol 2001;44:62–72. 12. Caillouette JC, Koehler AL. Phasic contraceptive pills and functional ovarian cysts. Am J Obstet Gynecol 1987;156: 1538–42. 13. Grimes DA, Hughes JM. Use of multiphasic oral contraceptives and hospitalizations of women with functional ovarian cysts in the United States. Obstet Gynecol 1989; 73:1037–9.
258
Holt et al
Contraception and Functional Ovarian Cysts
14. Abma J, Chandra A, Mosher W, Peterson L, Piccinino LJ. Fertility, family planning, and women’s health: New data from the 1995 National Survey of Family Growth. Vital Health Stat 23 1997;19:1–114. Address reprint requests to: Victoria L. Holt, PhD, MPH, Fred Hutchinson Cancer Research Center, Division of Public Health Sciences (MP 474), 1100 Fairview Avenue N., PO Box 19024, Seattle, WA 98109; E-mail: [email protected]. Received January 8, 2003. Received in revised form March 17, 2003. Accepted April 24, 2003.
OBSTETRICS & GYNECOLOGY