Oral contraceptive type and functional ovarian cysts Stephan F. Lanes, PhD: Brenda Birmann, BA: Alexander M. Walker, MD, DrPH:·b and Suanne Singer, BAc Chestnut Hill and Boston, Massachusetts, and Augusta, Maine
OBJECTIVE: We tested the hypothesis that multiphasic, low-dose monophasic, and high-dose monophasic oral contraceptives share a common protective effect against functional ovarian cysts. STUDY DESIGN: We conducted a cohort study using the automatic files of Maine Medicaid to assemble a population of 7462 women between the ages of 15 and 44 who were prescribed an oral contraceptive between Jan. 1, 1987, and Dec. 31,1988. We included as cases 32 women with a principal diagnosis of a functional ovarian cyst confirmed by medical records as being > 20 mm in diameter. RESULTS: At comparison with the absence of an oral contraceptive prescription, we observed decreasing rates of functional ovarian cysts among women prescribed multiphasic pills (rate ratio 0.91, 95% confidence interval 0.3000 to 2.31), low-dose monophasic pills with :535 flog estrogen (rate ratio 0.52, 95% confidence interval 0.17 to 1.33), and high-dose monophasic pills with >35 ILg estrogen (rate ratio 0.24, 95% confidence interval 0.01 to 1.34). CONCLUSIONS: The protective effect of oral contraceptives against functional ovarian cysts reported previously for high-dose monophasic pills may be attenuated with newer pills of lower hormonal potency. (AM
J
QBSTET GVNECOl
1992;166:956-61.)
Key words: Oral contraceptives, functional ovarian cysts Oral contraceptives were introduced in the United States in 1960, and the most popular formulations through the mid-1960s were combination pills containing progestin and >50 ~g estrogen." 2 Besides being remarkably effective at preventing pregnancy, these high-dose combination pills also conferred certain health benefits, such as reducing the risk of functional ovarian cysts.s" Publicity arose concerning possible adverse effects, however, and 1967 saw the introduction of low-dose formulations containing :535 flog estrogen. 1 Among the unintended effects of the early oral contraceptives was an increased risk of thromboembolic disease, which epidemiologic studies suggested was related to hormonal dose." Possibly as a result of mounting concern about adverse effects, the popularity of the high-dose pills declined through the 1970s. I, 2 This trend continued into the 1980s with the introduction in 1984 of multiphasic pills, which not only lowered hormone doses but also varied the dosage over the 21day medication period in an attempt to mimic more closely the natural hormone cycle.' In 1987, Caillouette and Koehler' reported on seven women in their clinical practice who were diagnosed with functional ovarian cysts soon after starting to use a multiphasic contraceptive. The hypothesized relation between multiphasic pills and functional ovarian cysts From Epidemiology Resources Inc." the Harvard School of Public H ealth/ and the Maine Health Information Center.' Supported in part by G.D. Searle & Co. Received f or publication June 4, 1991; revised September 3, 1991 ; accepted September 13, 1991. Reprint requests: Stephan Lanes, Epidemiology R esources Inc., One Newton Executive Park, Newton, MA 02162. 6/1/33774
956
has biologic plausibility because multiphasic pills may not suppress ovarian function to the same degree as pills with higher hormone doses. 8.10 Initial investigation by Grimes and Hughes" showed that national trends in sales of multiphasic oral contraceptives were not correlated with hospitalization rates for functional ovarian cysts. Such correlations are of limited use in investigating etiologic relations, however, and even though no relation was apparent, the investigators cautioned that "this indirect assessment can neither confirm nor refute the postulated association ."" To date, there have been no measurements of the incidence of functional ovarian cysts among women using multiphasic oral contraceptives. We report here the rate of functional ovarian cysts associated with use of multiphasic pills, low-dose monophasic pills, and high-dose monophasic formulations of oral contraceptives, and we contrast these rates with one another and with the rate of cysts associated with nonuse of oral contraceptives.
Material and methods We used the automated files of Maine Medicaid to identify a cohort of 9461 women who were prescribed an oral contraceptive from Jan. I, 1987, to Dec. 31 , 1988. We used Medicaid claims data to exclude from the primary analyses women who were not of reproductive age, who had cancer or ovarian dysfunction, or who were given drugs that might interfere with hormone metabolism. These drugs included fertility drugs and hormones, antipsychotics, anticonvulsants, and antitubercular drugs. After exclusions, the study population comprised 7462 women between the ages of 15
Oral contraceptive type and ovarian cysts 957
Volume 166 Number 3
and 44 years who contributed person time to the analysis from Jan. I, 1987, to Dec. 31,1988, during periods of Medicaid eligibility and when they were not classified as pregnant. To ensure that pregnancy periods were excluded, we excluded periods beginning 36 weeks before until 8 weeks after a delivery and 24 weeks before until 8 weeks after an abortion. We sought to ascertain first occurrences during the study period of symptomatic functional ovarian cysts >20 mm in diameter. We identified cases from Medicaid billing claims and medical records for ultrasonographic procedures, hospitalizations, and outpatient surgical procedures. We reviewed medical records for hospitalizations and for outpatient surgical procedures that listed an ovarian cyst as a primary or secondary Medicaid diagnosis, but we accepted as cases only those women whose cyst was indicated by the medical record as the primary diagnosis. A diagnostic ultrasonographic procedure is sometimes associated in Medicaid claims data with a presumptive diagnosis. For example, an ultrasonographic procedure performed to rule out ovarian cancer might list ovarian cancer as the billing diagnosis even though the primary diagnosis might be an ovarian cyst. To identify cases diagnosed by ultrasonography but not listed in the claims file, we reviewed radiologists' reports for any pelvic ultrasonographic procedure that had a claim date from 3 months before until 1 week after a medical claim with any diagnosis of functional ovarian cyst. Again, we accepted as cases only those women whose cyst was the principal diagnosis of the ultrasonographic procedure. We defined as cases therefore women with a principal diagnosis of a functional ovarian cyst >20 mm in diameter. Acquisition and initial processing of Medicaid claims data, elicitation of hospital and clinic participation, and medical record abstraction was conducted by Maine Health Information Center. We computed incidence rates as the number of functional ovarian cysts diagnosed between Jan . 1, 1987, and Dec. 31, 1988, divided by the sum of the person time contributed by the study population over the same period. Oral contraceptives were grouped into four types: multiphasic pills, low-dose monophasic pills (~35 fLg estrogen) , high-dose monophasic pills (>35 fLg estrogen), and progestin-only pills. Person time was allocated to three temporal categories of drug use: active prescription, recent prescription , and no prescription for an oral contraceptive. Women were considered to have an active prescription beginning on the date of the prescription and continuing for the number of days implied by the amount of medication supplied. Both 21-day and 28-day prescriptions were treated identically as 1 month of medication. We classified person time as recent use during the 30-day period after an active prescription. Any eligible day that did not fall within an active or recent prescription was assigned to
the category of no prescription. Each prescription initiated an active prescription classification that truncated the classification in effect at the time of the prescription. We compared first the rate of functional ovarian cysts during an active prescription for each type of oral contraceptive with the rate during no prescription for an oral contraceptive. Then, to describe differences between pills, we computed effect estimates for each type of oral contraceptive, using multiphasic pills as the reference category. We considered as potential confounding variables age (15 to 24, 25 to 34, 35 to 44), a prescription for antibiotics and duration of oral contraceptive use (~1 month, > 1 to 3 months, >3 to 6 months, >6 months). In the computation of duration of continuous oral contraceptive use, a prescription was considered continuous if it occurred within 5 days after the end of the previous prescription. We assessed and controlled for confounding effects of these variables by standardizing the rate in the reference category according to the covariate distribution in the index (i.e., exposed) category.12 We also fit regression models to the data under the hypothesis that the observed number of cases follows a Poisson distribution. 13 The results from the regression models were virtually identical to those obtained from the simpler standardized analyses. We measured effects using incidence rate ratios, and we quantified the precision of each rate ratio estimate by computing 95% confidence intervals according to the exact mid-p method. 14 Results
The study population of 7462 women contributed 179,088 person months, from which we excluded 31,676 person months as pregnancy periods and 50,934 person months as periods of Medicaid ineligibility. We sought medical records for 169 women in the study population, and we obtained records for 155 women. We did not obtain records for 14 women either because we could not find a medical record corresponding to the Medicaid claim (n = 6) or because the claim was from a medical facility from which no records were available (n = 8). Medical records were unavailable for women whose records were located at facilities that were out-of-state (n = 2) or closed (n = 1) or the facilities (or their representatives) decided not to participate for religious (n = 3) or other (n = 2) reasons. Of the 155 women for whom we obtained records, we confirmed 32 women as cases and 113 women as noncases. In addition, we did not include as cases 10 women for whom the medical records contained insufficient information to classify one way or the other according to our case definition. To assess prescription claims as an indicator of drug use, we computed effect estimates on pregnancy for each oral contraceptive category (Table I). After comparison with periods when no oral contraceptive was
958
Lanes et al.
March 1992 Am J Obstet Gynecol
Table I. Rate ratio estimates for pregnancy comparing each oral contraceptive category with no oral contraception
I Rate ratio*
No prescription Recent prescription Active prescription Progestin -only Multiphasic ,;;35 fLg estrogen >35 fLg estrogen
1.00 0.65 0.35 0.44 0.32 0.36 0.34
95% Confidence interval
Reference category 0.55-0.77 0.31-0.39 0.1l-1.77 0.27-0.39 0.32-0.42 0.27-0.43
*Rate ratios standardized to age distribution of index (i.e., "exposed") category.
Table II. Incidence rates of functional ovarian cysts by oral contraceptive category
I~:e11 Person months IRate No prescription Recent prescription Active prescription Progestin-only Multiphasic ,;;35 fLg estrogen >35 fLg estrogen
19 2 II
o
5 5 I
42,738.0 8,872.8 44,053.2 219.6 13,196.4 22,251.6 8,385.6
(X 10,000)
4.4 2.3 2.5 0.0 3.8 2.2 1.2
prescribed, these data show an approximate 65% reduction in the pregnancy rate for women prescribed multiphasic, low-dose monophasic, and high-dose monophasic pills and a 56% reduction for women taking progestin-only pills. Pregnancy rates declined with increasing duration of use, and there was a 75% reduction in the pregnancy rate after 6 months of continuous oral contraceptive prescriptions. Table II describes the distribution of cases, person time, and incidence rates of functional ovarian cysts according to oral contraceptive category. The distribution of person time was divided nearly evenly between an active prescription (46%) and no prescription (45%) for an oral contraceptive, with the remaining time classified as recent use of oral contraceptives. Among types of oral contraceptives, the distribution was similar to national market shares 2 : 50% low-dose monophasic pills, 30% multiphasic pills, 18% high-dose monophasic pills, and 1 month and :53 months, then increases slightly with increasing duration.
Comment In this population of women prescribed oral contraceptives, the incidence rate of functional ovarian cysts was greater during periods when no oral contraceptive was prescribed than it was during an active prescription for either monophasic pills or multiphasic pills. For users of oral contraceptives, however, this preventive effect against functional ovarian cysts appears to be
Oral contraceptive type and ovarian cysts
Volume 166 Number 3
Table Iii. Rate ratio estimates for functional ovarian cysts comparing each oral contraceptive category with no oral contraceptio~
I
No prescription Active prescription: Multiphasic $35 ILg estrogen >35 ILg estrogen
Rate ratio*
95% Confidence.interoal
1.00
Reference category
0.91 0.52 0.24
0.30-2.31 0.17-1.33 0.01-1.34
*Rate ratios standardized to age distribution of index (i.e., "exposed") category.
greater for monophasic pills-especially high-dose pills-than for multiphasic pills. Two limitations of this study are its small size in relation to the frequency of functional ovarian cysts and the possibility of selective diagnosis of cysts among users of multiphasic oral contraceptives. Misclassification of oral contraceptive use probably resulted in some underestimation of the effects of oral contraceptives on functional ovarian cysts. Because the inaccuracy of prescription data as a marker of drug use should apply similarly to each oral contraceptive category and to cases and noncases alike, misclassification of this type would be expected to diminish any apparent association. 15 To investigate the extent of misclassification of oral contraceptive use, we examined the intended effects of oral contraceptives in reducing pregnancy rates. Compared with no oral contraceptive prescription, the pregnancy rate was reduced by approximately 65% for women with prescriptions for multiphasic, low-dose monophasic, or high-dose monophasic pills and about 56% for women using progestin-only pills. In addition, there was a trend toward decreasing pregnancy rates with increasing duration of oral contraceptive use, so that the pregnancy rate declined by about 75% after 6 months of use. The similar efficacy of monophasic and multiphasic pills,16 the lower efficacy of progestin-only pills,16 and the trend of increasing efficacy with duration of use 17 • 18 have been suggested previously but cannot account entirely for the apparent efficacy of oral contraceptives being less than expected. Although the contraceptive effect is apparent in these data, the relatively large number of pregnancies suggests that a lack of compliance resulted in substantial underestimation of oral contraceptive efficacy, perhaps by as much as 35%. Nevertheless, this form of misclassification would not cause the appearance of an association where none existed and would not explain the association between oral contraceptive type and functional ovarian cysts. Another potential source of exposure misclassification derives from the possibility that oral contraceptives
959
Table IV. Incidence rates of functional ovarian cysts by oral contraceptive category and duration of exposure Duration (mo)
$1
>1-3
>3-6
>6
Oral contraceptive
Rate (x 10,000)
Progestin-only Multiphasic $35 ILg estrogen >35 ILg estrogen
0 2 2 1
70.8 3,969.6 6,566.4 2,230.8
0.0 5.0 3.0 4.5
TOTAL
5
12,837.6
3.9
Progestin-only Multiphasic $35 ILg estrogen >35 ILg estrogen
0 1 2 0
72.0 4,144.8 6,842.4 2,482.8
0.0 2.4 2.9 0.0
TOTAL
3
13,542.0
2.2
Progestin-only Multiphasic $35 ILg estrogen > 35 ILg estrogen
0 1 0
52.8 2,818.8 4,759.2 1,861.2
0.0 3.5 2.1 0.0
TOTAL
2
9,492.0
2.1
Progestin-only Multiphasic $35 ILg estrogen >35 ILg estrogen
0 1 0 0
24.0 2,263.2 4,082.4 1,809.6
0.0 4.4 0.0 0.0
8,179.2
1.2
TOTAL
I
were prescribed to treat a functional ovarian cyst. We examined oral contraceptive histories of all exposed cases to determine if any women were first prescribed the oral contraceptive type within a few months of the diagnosis. We identified one such case who was prescribed high-dose monophasic pills shortly before being diagnosed with a functional ovarian cyst. If the cyst was suspected and treated before the diagnosis, then this case should have been classified as having no oral contraceptive prescription. The effect of this bias would be to overestimate the incidence rate of ovarian cysts in the category of high-dose monophasic pills. Thus therapeutic use of oral contraceptives to treat ovarian cysts would not explain high-dose monophasic pills being associated with the lowest rate of ovarian cysts. Misclassification of disease status is another potential source of error in this study. In particular, functional ovarian cysts are difficult to diagnose, and there may be a large group of occult cases that does not come to medical attention. Moreover, suspicion about a possible relation between multiphasic pills and functional ovarian cysts' could have resulted in cases being diagnosed preferentially among users of multiphasic pills. We addressed this issue by attempting to ascertain only symptomatic cases. We note also that physicians performing diagnostic procedures such as uitrasonographic examinations would not necessarily be aware of a patient'S oral contraceptive history. Although we cannot rule out
960
Lanes et al.
a diagnostic bias, two lines of evidence run counter to this hypothesis. First, a suspicion concerning ovarian cysts and multiphasic pills should not apply to low-dose monophasic pills, because there has been no suggestion of a relation between ovarian cysts and low-dose monophasic pills. Thus the finding of a higher rate of cysts among users of low-dose monophasic pills than among users of high-dose monophasic pills should not be due to diagnostic bias. Second, any diagnostic bias for users of multiphasic pills should exist as soon as multiphasic pills are prescribed and might be expected to diminish with continued contraceptive use as the susceptible population is diagnosed and switched from multiphasic pills to other forms of contraception. During the first month of use, however, the results show little difference between types of oral contraceptives in the rate of cysts (Table IV). After the first month of use, the rate of cysts declines for users of monophasic contraceptives. For users of multiphasic pills, however, the rate of cysts declines briefly, but then increases with increasing duration of use. Thus the difference between pill types in the risk of functional ovarian cysts is enhanced with increasing duration of use. Rather than indicating a diagnostic bias, these data suggest that, like the pregnancy data, categories of longer duration of oral contraceptive use provide a more accurate indication of pharmacologic effects. We controlled by either restriction or analytic means for potential confounding effects of age and duration of oral contraceptive use, for a variety of drugs that might affect hormone metabolism, and for the effects of cancer and ovarian dysfunction. Smoking data were unavailable from Medicaid, and we did not control for a possible effect of smoking. Smoking has been reported to be associated with an increased risk of ovarian cystS. '9 The reported effect was small, however, with the rate of cysts among current smokers being 50% greater than the rate among never-smokers. This difference could account for the difference in the rate of cysts between high-dose and low-dose monophasic pills or between low-dose monophasic and multiphasic pills, but it could not account for the threefold difference between multiphasic and high-dose monophasic pills even if everyone using multiphasic pills smoked and no one using high-dose monophasic pills smoked. It is possible that the effects of smoking are greater than has been estimated, but it is also unlikely that there is a great disparity in the prevalence of smoking among users of different types of oral contraceptives. One reason that we restricted the cohort to women using oral contraceptives and did not include a comparison group of nonusers was to reduce differences between comparison groups with regard to extraneous risk factors (i.e., confounding). Within the cohort of oral contraceptive users, we selected as the primary
March 1992 Am J Obstet Cynecol
reference category periods of no prescription because this category provided the most statistically stable effect estimates. If the protective effect of oral contraceptives lasts for longer than 1 month, then the rate observed here for the category of no prescription underestimates the true risk in the absence of oral contraceptives. This source of error, combined with the likelihood that some women did not use their prescribed oral contraceptives or did not use them properly, would bias rate ratio estimates toward a finding of no association. Thus the small decline in the rate of functional ovarian cysts observed for women prescribed multiphasic pills compared with no oral contraceptive is likely to underestimate the protective effect of multiphasic pills. Nevertheless, the possibility that multiphasic pills increase the risk of cysts compared with no oral contraceptive cannot be dismissed. This is only the first study to investigate the risk of functional ovarian cysts in relation to multiphasic and other types of oral contraceptives. In addition, functional ovarian cysts are uncommon, and the small number of cysts in this study renders the incidence rates unstable and the effect estimates imprecise. Consequently, any or all of the differences between contraceptive types in the rate of ovarian cysts could be due to chance. To the extent that the categories of oral contraceptives used in this analysis correspond to different levels of ovarian suppression, however, this possibility becomes increasingly remote. The issue is best addressed, of course, by additional studies. We thank Maine Medicaid for providing access to their data, the Maine Health Information Center for facilitating the acquisition and processing of Medicaid data and medical records, and Maggie Satterfield and Noreen Manzo for their assistance in conducting the research. REFERENCES I. Piper JM, Kennedy DL. Oral contraceptives in the United
2. 3. 4. 5. 6.
7. 8.
States: trends in content and potency. Int J Epidemiol 1987;16:215-21. Annegers JF. Patterns of oral contraceptive use in the United States. Br J Rheumatol 1989;28(suppl 1):4850. Ory H. Functional ovarian cysts and oral contraceptives: negative association confirmed surgically. JAMA 1974; 228:68-9. Ylikorkala O. Ovarian cysts and hormonal contraception [Letter]. Lancet 1977;2: 1101-2. Vessey M, Metcalfe A, Wells C, McPherson K, Westhoff C, Yeates D. Ovarian neoplasms, functional ovarian cysts, and oral comraceptives. BMJ 1987;294:1518-20. Inman WHW, Vessey MP, Westerholm B, Engelund A. Thromboembolism and the steroidal contem of oral contraceptives: a report to the Committee on Safety of Drugs. BMJ 1970;2:203-9. CaiIIouette JC, Koehler A. Phasic contraceptive pills and functional ovarian cysts. AM J OBSTET GYNECOL 1987; 156: 1538-42. Westcombe R, Ellis R, Fotherby K. Suppression of ovu-
Oral contraceptive type and ovarian cysts
Volume 166 Number 3
9. 10. II .
12. 13.
lation in women using a triphasic oral contraceptive. Br J Fam Plann 1987;13:127-32. Ketting E. The relative reliability of oral contraceptives: findings of an epidemiologic study. Contraception 1988;37:343-8. Kovacs GT, Riddoch G, Duncombe P, et al. Inadvertent pregnancies in oral contraceptive users. Med J Aust 1989; 150:594-51. 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. Miettinen OS. Components of the crude risk ratio. Am J Epidemiol 1972;96:168-72. Rothman KJ. Modern epidemiology. Boston : Little Brown, 1986.
961
14. Miettinen OS. Comment. J Am Stat Assoc 1974;69: 380-2. 15. Copeland KT, Checkoway HA, McMichael AJ, Holbrook RH . Bias due to misclassification in the estimation of relative risk. Am J Epidemiol 1977; 105:488-95. 16. Olin BR, Hebel SK, eds. Drug Facts and Comparisons. St Louis: Lippincott, Co, 1989: 107b. 17. Tietze C. The clinical effectiveness of contraceptive methods. AM J OBSTET GYNECOL 1959;78:650-1. 18. Potter RG. Additional measures of use e ffectiveness of contraception. Milbank Q 1963;41:400-18. 19. Wyshak G, Frisch RE. Albright TE. Albright NL. Schiff I. Smoking and cysts of the ovary. 1m] FertiI1988;33:398404 .
Bound volumes available to subscribers Bound volumes of the AMERICAN JOURNAL OF OBSTETRICS AND GYNECOLOGY are available to subscribers (only) for the 1992 issues from the Publisher, at a cost of $68.00 for domestic, $95.76 for Canada, and $91.00 for international for Vol. 166 (January-June) and Vol. 167 (July-December). Shipping charges are included. Each bound volume contains a subject and author index and all advertising is removed . Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram with the JOURNAL name, volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact Mosby- Year Book, Inc., Subscription Services, 11830 Westline Industrial Drive, St. Louis, MO 63146-3318, USA ; phone (800) 325-4177, ext. 4351, or (314) 453-4351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular JOURNAL subscription.