Estrogen and Other Female Reproductive Risk Factors Are Not Strongly Associated With the Development of Rheumatoid

Arthritis in Elderly Women

Linda A. Merlino, James R. Cerhan, Lindsey A. Criswell, Ted R. Mikuls, and Kenneth G. Saag
Objective: Endogenous and exogenous reproductive hormones have been associated with rheumatoid arthritis (RA) in women, but data are inconsistent and no studies have assessed RA risk factors exclusively in elderly women. Methods: The authors examined the association between reproductive factors, exogenous hormone exposure, and RA in a prospective cohort study of 31,336 Iowa women who were aged 55 to 69 years at cohort baseline in 1986. Results: During 11 years of follow-up, 158 incident cases of RA were identied and validated. Age at last pregnancy (P trend .01) and age at menopause (P trend .03) were inversely associated with RA, whereas a history of polycystic ovary syndrome (relative risk [RR], 2.58; 95% condence interval [CI], 1.06 to 6.30), endometriosis (RR, 1.72; 95% CI, 0.93 to 3.18), and former use of hormone replacement therapy (RR, 1.47; 95% CI, 1.04 to 2.06) were positively associated with RA. In multivariate analysis models, a history of polycystic ovary syndrome remained the most consistent predictor of RA, whereas the RRs for other factors attenuated. Conclusion: Few reproductive factors showed a strong or statistically significant association with RA in elderly women. The association of polycystic ovary syndrome may be indicative of perturbations of endocrine-immune activity that may inuence the development of RA. This prospective cohort study adds to the understanding of the potential contribution of hormonal factors to the cause of RA in older women. Semin Arthritis Rheum 33:72-82. 2003 Elsevier Inc. All rights reserved. INDEX WORDS: Rheumatoid arthritis; estrogen; prospective cohort; reproductive risk factors.

HEUMATOID arthritis (RA) is an inammatory disorder that affects approximately 1% of the adult population in the United States (1-3).

The cause of this disease is not well understood, although there appear to be both genetic and nongenetic components (4-12). Because RA occurs
sor, Department of Medicine, University of Nebraska Medical Center, Omaha, NE; Kenneth G. Saag, MD, MSc: Associate Professor, Division of Rheumatology, The University of Alabama at Birmingham, Birmingham, AL. Supported by a Clinical Science Grant from the Arthritis Foundation; the Iowa Womens Health Study was supported by a National Cancer Institute Grant R01 CA 39741. Address reprint requests to: Kenneth G. Saag, MD, MSc, Division of Clinical Immunology and Rheumatology, Department of Medicine, University of Alabama at Birmingham, MEB 625, 1813 Sixth Ave S, Birmingham, AL 35294-3296. e-mail: ksaag@uab.edu 2003 Elsevier Inc. All rights reserved. 0049-0172/03/3302-0002$30.00/0 doi:10.1016/S0049-0172(03)00084-2

From the College of Public Health, The University of Iowa, Iowa City, IA; Department of Health Sciences Research, The Mayo Clinic, Rochester, MN; Division of Rheumatology, Rosalind Russell Medical Research Center for Arthritis, The University of California at San Francisco, San Francisco, CA; Department of Medicine, University of Nebraska Medical Center, Omaha, NE; and Division of Rheumatology, The University of Alabama at Birmingham, Birmingham, AL. Linda A. Merlino, MSc: Research Assistant, College of Public Health, The University of Iowa, Iowa City, IA; James R. Cerhan, MD, PhD: Associate Professor, Department of Health Sciences Research, The Mayo Clinic, Rochester, MN; Lindsey A. Criswell, MD, MPH: Associate Professor, Division of Rheumatology, Rosalind Russell Medical Research Center for Arthritis, The University of California at San Francisco, San Francisco, CA; Ted R. Mikuls, MD, MSPH: Assistant Profes72

Seminars in Arthritis and Rheumatism, Vol 33, No 2 (October), 2003: pp 72-82

ASSOCIATION OF ESTROGEN WITH RA

73

approximately twice as commonly in women as in men, it has been suggested that reproductive hormones play a role in disease pathogenesis (13). Support for a role of reproductive hormones has been strengthened by the common nding of RA remission during pregnancy and post-partum ares (14-17). However ,there have been many contradictory studies reporting either benecial or deleterious effects of exogenous hormone use on the risk of RA (18-27). Although traditionally considered to be a disease of middle age, RA incidence increases with age (1). Elderly-onset RA is most commonly dened as disease onset after the age of 60 years, and has putative features that may distinguish it from younger-onset RA. Compared with younger-onset disease, elderly-onset RA commonly involves the large joints, frequently is seronegative for rheumatoid factor, often has a high erythrocyte sedimentation rate, occasionally displays overlapping features with polymyalgia rheumatica, any may have poorer functional outcome (13). Although several studies have compared clinical features and outcomes of elderly-onset RA (28-30), previous investigators have not examined the risk factors for RA in elderly women. Our study used a large prospective populationbased cohort of older women to assess the role of selected gynecologic, reproductive, and exogenous hormonal factors postulated to contribute to the cause of RA. Based on previous studies (4), we a priori hypothesized that exogenous estrogens would be protective for RA and that surrogate factors for endogenous hormones would not be associated with RA (4).
METHODS

ity); and history of various medical conditions. The survey included extensive information on female reproductive history, including questions about infertility, all pregnancies and their outcomes, breastfeeding, and the use of birth control pills and hormone replacement therapy (HRT). Follow-up questionnaires were mailed in 1987 (91% response rate), 1989 (89%), 1992 (83%), and 1997 (79%). The 1992 questionnaire asked, Have you ever been told by a doctor that you have rheumatoid arthritis? and the 1997 questionnaire asked, Since July of 1992, were you diagnosed for the rst time by a doctor as having rheumatoid arthritis? Of the 35,635 women who responded to the 1992 and/or the 1997 surveys, 3171 women were identied as potential incident cases (1987 or later diagnosis), and 2012 were identied as potential prevalent cases (1986 or prior diagnosis). RA Case Validation and At-Risk Cohort Women with potential incident RA (RA onset after January 1, 1987), as ascertained on the 1992 and 1997 questionnaires (n 3171), were contacted by mail to conrm their yes (n 1702), not sure (n 1249), or missing response (n 220) response; to obtain additional arthritis-related data, including names and addresses of all physicians whom they had seen about their RA; and to obtain consent to release their medical records. Nonrespondents to this supplemental questionnaire were contacted by telephone to obtain the necessary information. For deceased subjects, next of kin were contacted. Requests for medical records were mailed to physicians for all consenting study participants. Details of the validation process and at-risk cohort have been described previously (36-38). Briey, a total of 1227 subjects reafrmed their initial positive response to the IWHS RA question, and 1186 had medical records reviewed to determine case status by using the ever (since cohort baseline) satisfaction of the 4 of 7 American College of Rheumatology (ACR) criteria (39,40). Although medical records were requested on all potential cases, any diagnosis of denite RA provided by a physician identied as Board-certied/eligible rheumatologist also was considered a validated case (41). We conservatively dened the date of RA diagnosis as the rst date of a symptom onset occurring after January 1, 1987. Cases with symptoms before

Study Cohort and Follow-up The Iowa Womens Health Study (IWHS), a population-based prospective cohort study, was established in 1986 by using a postal survey, and enrolled 41,836 (43% response rate) women who were age 55 to 69 years to evaluate risk factors for postmenopausal breast cancer (31,32). Characteristics of the study and the cohort have been described previously (31,33-35). Items reported on the baseline questionnaire included demographic data (education, martial status, usual occupation); weight history; reproductive history; lifestyle factors (smoking history, alcohol use, physical activ-

74

MERLINO ET AL

January 1, 1987 (ie, prevalent cases) were excluded, as were women with alternative diagnoses (gout, polyarticular arthritis, psoriatic arthritis, and so on). Based on the comprehensive record review and concordance by a combination of 2 rheumatologists, rheumatology advance-practice nurse, or rheumatology physician assistant, we validated 158 incident cases. RA diagnosis was based on cumulative satisfaction of ACR criteria (n 146; 92% of cases) and/or diagnosis by a respondents rheumatologist (n 139; 88% of cases). Eleven cases of RA were based on a Board-certied rheumatologists diagnosis only, 20 were based on satisfaction of ACR criteria only, and 127 met both criteria. From the original 41,836 cohort members, 30,362 women reported no history of RA on the 1992 and 1997 IWHS questionnaires. After the validation process, an additional 816 women who did not answer or who marked not sure on the IWHS questionnaire and then answered no on the validation questionnaire were not excluded, resulting in an at-risk cohort of 31,336 women. Women were considered at risk for from the date of return of the baseline questionnaire to September 30, 1992 (for women who died before or did not answer the 1997 questionnaire) or August 31, 1997 (remaining women). Statistical Analysis Risk ratios (RRs) trend tests, and 95% condence intervals (CIs) were used as the measure of association between the exposures of interest and RA and were estimated by using Cox proportional hazards regression (42). Ordered categorical variables based on a continuous scale were formed based on quartile cut points (age at menarche, age at menopause, number of ovulatory years, total number of pregnancies, age at last pregnancy) or tertile cut points (age at rst pregnancy, number of miscarriages). Univariate analyses were adjusted for age. For the multivariate analyses, 2 separate analyses were conducted because of nonapplicable data for several variables associated with pregnancies and nulliparous women. A multivariate model for each of the 2 separate analyses initially included all univariately modeled variables (at the P .15 signicance level). The rst model included all women and omitted the variable drug to stop lactation. A second model excluded nulligravid women and included the variable drug to stop lactation. The multivariable models were

adjusted for age and smoking status (never, former, current). Because of a prior strong association of smoking with RA (38), we assessed whether smoking (never smoked/ever smoked) modied the effects of age at menopause with RA as well as HRT with RA. We used stratied analysis (by smoking status) and also added 1-way interaction terms between smoking and estrogen exposure. All analyses were conducted by using the statistical analysis software program PHREG (SAS Institute, Cary, NC) (43).
RESULTS

We validated 158 incident RA cases in the cohort, providing a crude incidence rate of 47 per 100,000 person-years. There were no signicant differences at baseline between the cases and the noncases for age (mean, 61.4 years), race ( 99% white), education (42% high school graduates), or place of residence (34% live in cities with a population 10,000). Noncases were more likely to be widowed (P .03) and cases were more likely to be smokers (P .005). The average age of RA onset was 67.8 years (SD 4.9 years). The mean time from symptom onset to diagnosis was 1.1 years (SD 1.8 years), and 61% were positive for rheumatoid factor. The 3 ACR criteria related to degree of joint involvement (hand joints, symmetrical involvement, 3 or more joints involved) were satised by more than 98% of the RA patients, followed by morning stiffness (70%), erosions evident on radiographs (26%), and rheumatoid nodules (14%). The mean number of ACR criteria satised was 4.6 (SD 1.1). Results for age-adjusted association of RA with gynecologic and reproductive factors are presented in Table 1. No effect was seen for age at menarche, having undergone a hysterectomy or oophorectomy, or use of oral contraceptives. There was an inverse trend for later age at menopause (P trend .03), with women who underwent menopause after age 51 having an RR of 0.64 (95% CI, 0.41 to 1.00) compared with women who underwent menopause before age 45. There was an elevated risk for those with self-reported physician-diagnosed polycystic ovary syndrome (RR 2.58; 95% CI, 1.06 to 6.30) and a suggestive elevated risk for self-reported physician-diagnosed endometriosis (RR 1.72; 95% CI, 0.93 to 3.18). Former users, but not current users, of HRT had an elevated risk of RA (RR 1.47; 95% CI, 1.04 to

ASSOCIATION OF ESTROGEN WITH RA

75

Table 1: Age-Adjusted Relative Risks of Rheumatoid Arthritis in Elderly Women in Relation to Gynecologic History, IWHS, 1986 to 1997
P Value for Trend

Factor Age at menarche (yrs) 11 11 to 12 13 to 14 14 Age at menopause (yrs) 45 45 to 48 49 to 51 51 Number of ovulatory years 29.0 29.0 to 33.5 33.6 to 36.8 36.8 Hysterectomy No Yes Oophrectomy No Unilateral Bilateral Polycystic ovary syndrome No Yes Endometriosis No Yes Oral contraceptive use No Yes Use of HRT Never Former Current Duration of HRT use Never Former 5 yrs Former 5 yrs Current 5 yrs Current 5 yrs

Cases 5 68 66 18 43 36 34 35 43 31 44 28 101 57 110 14 30 151 5 144 11 126 32 86 55 17 86 45 9 8 7

Person-Years 9752 130,240 151,721 39,567 75,779 68,138 77,119 99,330 77,885 72,915 77,859 87,437 218,453 113,081 239,343 26,864 63,163 318,748 4,101 310,717 13,661 268,541 64,823 203,219 91,258 38,752 203,219 76,735 13,315 14,177 23,804

Age-Adjusted RR 1.00 1.03 0.85 0.90 1.00 0.95 0.80 0.64 1.00 0.79 1.05 0.60 1.00 1.07 1.00 1.14 1.00 1.00 2.58 1.00 1.72 1.00 1.00 1.00 1.47 1.02 1.00 1.42 1.65 1.30 0.70

95% CI Reference 0.42, 2.55 0.64, 2.10 0.33, 2.43 Reference 0.61, 1.49 0.51, 1.25 0.41, 1.00 Reference 0.50, 1.25 0.69, 1.60 0.37, 0.96 Reference 0.77, 1.48 Reference 0.66, 2.00 0.66, 1.50 Reference 1.06, 6.30 Reference 0.93, 3.18 Reference 0.67, 1.50 Reference 1.04, 2.06 0.61, 1.72 Reference 0.99, 2.04 0.83, 3.30 0.62, 2.69 0.32, 1.51

0.38

0.03

0.11

0.29

2.06) when compared with those who never used HRT. Based on stratied analyses, there was no confounding by smoking on age at menopause or for smoking with HRT use. Although there was a

signicant risk for RA in current smokers in the rst 2 quartiles of age at menopause (early menopause), formal tests of interaction were not significant (P .35). Current smokers who were former

76

MERLINO ET AL

Table 2: Age-Adjusted Relative Risks of Rheumatoid Arthritis in Elderly Women in Relation to Pregnancy History, IWHS, 1986 to 1997
AgeAdjusted RR 1.00 1.09 1.00 1.15 1.00 1.26 1.24 0.76 1.00 0.97 0.65 1.00 0.83 0.76 0.64 1.00 1.02 0.99 1.00 1.36 1.00 1.36 1.00 1.14 1.00 1.13 0.64 P-Value for trend

Factor History of infertility No Yes Ever pregnant No Yes Total number of pregnancies* 1 2 to 3 4 to 5 5 Age at rst pregnancy (yrs)* 20 20 to 25 25 Age at last pregnancy (yrs)* 25 25 to 29 30 to 34 34 Number of miscarriages* 0 1 1 Number of stillbirths* 0 0 Drug to stop lactation* No Yes Parity Nulliparous Parous Number of children breastfed 0 1 to 2 2
*Nulligravid women excluded. Nulliparous women excluded.

Cases 128 28 11 145 9 69 50 17 35 82 28 23 42 44 36 103 30 12 136 9 79 66 12 144 80 46 17

Person-Years 274,158 52,770 27,133 303,440 22,111 132,712 94,344 53,162 65,079 156,178 81,414 36,884 81,516 92,957 88,251 215,865 61,988 25,587 289,267 14,173 183,622 112,326 29,462 303,161 162,593 83,789 51,434

95% CI Reference 0.72, 1.66 Reference 0.62, 2.12 Reference 0.63, 2.53 0.61, 2.53 0.34, 1.71 Reference 0.65, 1.44 0.39, 1.08 Reference 0.50, 1.38 0.46, 1.26 0.38, 1.09 Reference 0.68, 1.53 0.54, 1.80 Reference 0.69, 2.68 Reference 0.97, 1.91 Reference 0.63, 2.05 Reference 0.78, 1.62 0.37, 1.09

0.30

0.10

0.10

0.99

0.24

HRT users also showed an elevated risk; however, interaction terms also were not signicant (P .09). Results for these analyses are limited by statistical power. Table 2 presents age-adjusted factors related to pregnancy history. Little effect was seen for his-

tory of infertility, ever being pregnant, number of miscarriages, parity, total number of pregnancies, or number of children breastfed. There was a nonsignicant positive association between RA and having had 1 or more stillbirths (RR 1.37; 95% CI, 0.69 to 2.68) and a trend toward positive

ASSOCIATION OF ESTROGEN WITH RA

77

Table 3: Multivariate-Adjusted Relative Risks of Rheumatoid Arthritis in Elderly Women, IWHS, 1986 to 1997
Factor All Women RR (95% CI) (N 27,184) Age at last pregnancy (yrs) 25 25 to 29 30 to 34 34 Nulligravid Drug to stop lactation No Yes Age at menopause (yrs) 45 45 to 48 49 to 51 51 Polycystic ovary syndrome No Yes Endometriosis No Yes Use of HRT Never Former Current 1.00 (Reference) 0.79 (0.47, 1.35) 0.70 (0.41, 1.19) 0.67 (0.39, 1.17) P-trend .14 0.70 (0.33, 1.44) Ever Pregnant Women* RR (95% CI) (N 24,378) 1.00 (Reference) 0.74 (0.43, 1.26) 0.63 (0.37, 1.08) 0.59 (0.33, 1.03) P-trend .06

1.00 (Reference) 1.57 (1.09, 2.26) 1.00 (Reference) 1.02 (0.64, 1.63) 0.93 (0.58, 1.50) 0.75 (0.47, 1.21) P-trend .22 1.00 (Reference) 2.51 (1.01, 6.24) 1.00 (Reference) 1.59 (0.82, 3.08) 1.00 (Reference) 1.25 (0.86, 1.82) 1.04 (0.60, 1.79) 1.00 (Reference) 1.01 (0.61, 1.67) 1.01 (0.62, 1.66) 0.80 (0.48, 1.31) P-trend .39 1.00 (Reference) 2.28 (0.83, 6.24) 1.00 (Reference) 1.48 (0.71, 3.07) 1.00 (Reference) 1.25 (0.85, 1.85) 0.96 (0.53, 1.72)

*Analyses excluded 2941 noncase and 11 case nulligravida women. Relative Risk adjusted for age, smoking status, and other factors in the column. Decline in denominator from overall sample size is caused by missing data for 1 or more variables.

association of RA and the drugs used to stop lactation (RR 1.37; 95% CI, 0.98 to 1.92). There was a suggestive protective trend for older age at rst pregnancy and older age at last pregnancy (P trend .10 for both factors). The results of the multivariate analyses are presented in Table 3. We rst analyzed data for all women, and simultaneously included all reproductive risk factors that had a univariately signicant (P .15) association with RA, except drug to stop lactation (cannot be included in a model with nulliparous women). In this model, all of the risk factors attenuated slightly, particularly age at menopause and the use of HRT, and only polycys-

tic ovary syndrome remained signicant at conventional levels among all women. We next focused on the subset of women who were ever pregnant, which allowed us to include drugs to stop lactation. In this model, all factors attenuated slightly, particularly age at menopause and the use of HRT. Drugs to stop lactation remained statistically signicant at conventional levels and the risk for polycystic ovary syndrome remained stable; however, the CIs became less precise (most likely because of the exclusion of nulliparous women. A common outcome of polycystic ovary syndrome is nulliparity, although only 1 woman with polycystic ovary syndrome was excluded).

78

MERLINO ET AL

DISCUSSION

In this prospective cohort of older Iowa women, a self-reported history of physician-diagnosed polycystic ovary syndrome was the factor most consistently associated with an elevated risk of RA, although this risk was small in absolute terms. Other reproductive risk factors were weaker and were less consistently and less signicantly associated with RA, suggesting that female reproductive factors are not likely to be major determinants of RA risk, at least in this cohort of older women. However, these weak associations in older women do not exclude the possibility of a stronger association of reproductive factors and RA in younger woman. Polycystic Ovary Syndrome and Endometriosis Polycystic ovary syndrome is characterized by hormonal abnormalities such as higher luteinizing hormone, estrone, testosterone, and androstenedione levels. Polycystic ovary syndrome may result in anovulation, infertility, menstrual dysfunction, and obesity (44-46). Although self-report of polycystic ovary syndrome was not validated in our cohort, another study found self-reported polycystic ovary syndrome comparable with that of physician report (47). In the IWHS cohort, 472 (1.35%) women self-reported polycystic ovary syndrome, a prevalence lower than 1 populationbased study but comparable with another (48). Elevated levels of androgens are thought to be immunosuppressive, as are elevated levels of estrogen (49). A history of endometriosis also produced a weakly positive, but nonsignicant, association with RA. This association attenuated after multivariable adjustment. Endometriosis has been associated with autoimmunity and reproductive failure (50-52). Although our results did not show an association between infertility and RA, both polycystic ovary syndrome and endometriosis result in reduced fecundity, which has been reported by other to be associated with RA (53). More recently, a cross-sectional study found high rates of autoimmune disorders, including RA, in women with endometriosis compared with the general population (54). The weak association of RA and endometriosis, in addition to the positive association of polycystic ovary syndrome and RA, suggests that endocrine perturbations could modulate immune function toward the development of RA. Alterna-

tively, the association of polycystic ovary syndrome with RA may represent a delay of the clinical onset of arthritis in susceptible women because of the decline of excess androgen production with aging. Age at Menopause and Other Endogenous Hormonal Factors An inverse association with later age at menopause and RA attenuated with multivariate analyses. Hormonal changes associated with menopause included elevated serum levels of luteinizing hormone and follicle-stimulating hormone, and decreased levels of estradiol, estrone, progesterone, and androstenedione (55). Goemaere et al (56) found that RA occurred most frequently around the menopausal transition. Thus, a delay in menopause could potentially protect against RA development. A prospective Norwegian study of RA showed a nonsignicant inverse risk with later age at menopause, ndings that are similar to ours. The investigators adjusted only for age, occupation, and country of residence (57); other important potential confounders (such as smoking) were not considered. In contrast to these positive studies, time since menopause was not a risk factor of RA in another large prospective cohort study involving nurses who were aged 30 to 55 years at enrollment (25). Of note, smoking is associated with both an early menopause and RA (57-59). We have previously reported a strong and consistent association between smoking and increased risk of RA in this cohort (38). Exogenous Estrogen Use Drugs administered to suppress lactation resulted in a slightly increased risk of development of RA in our cohort, which strengthened slightly in multivariate analysis. At the time these women took drugs to suppress lactation, the agents used for this purpose were mainly high-dose estrogens (60-62). Unfortunately, information of dose, formulation, frequency of exposure, or time of exposure that would allow us to further assess these effects are not available. Although this association is not well understood, Masi et al (63) have suggested that chronic subclinical physiologic imbalances between the endocrine, immune, and microvascular systems eventually progress to RA when susceptible individuals lack the ability to adapt to the imbalance in response to stressors (63). Short-

ASSOCIATION OF ESTROGEN WITH RA

79

term exogenous high-dose estrogen may initially contribute as such a stressor by altering the endocrine system. In our study, former users of HRT, but not current users, had an increased risk of RA in univariate analysis. This association weakened, however, after multivariate adjustment. Consistent with our nding, other studies have shown little effect of HRT on the development of RA. Hernandez-Avila et al (64), in a prospective cohort study of nurses, found a nonsignicant protective effect (RR 0.70; 95% CI, 0.5 to 1.2) for former users of HRT (in contrast to our slightly harmful effect nding) and a slight risk for current users. Spector et al (24) found no association with HRT use and RA in their retrospective cohort, a nding corroborated by others (19,22). Our study, although limited in power, did not show an association with the development of RA in elderly women and use of oral contraceptives. Given the age of the subjects and the relative unavailability of oral contraceptives during their peak childbearing years in the late 1930s to early 1960s, our study population had a very low use of oral contraceptives. Consistent with our study, several studies of oral contraceptives have shown little association with RA (20,21,23,27), whereas other investigations suggest oral contraceptives reduce the risk of RA (12,26,65). In addition to the low prevalence of use of oral contraceptives by our respondents, differing study methodologies may also explain some of these differences. Pregnancy and Breastfeeding In our study, later age at last pregnancy was protective for RA incidence; later age at rst pregnancy also showed an inverse trend. To reduce colinearity in the model, we chose age at last pregnancy for multivariate modeling because this variable encompassed all pregnancies. Although no studies have shown an association between age at last pregnancy and RA, other investigators have examined age at rst or last birth and found no association with RA (25,26,57). Null associations between the number of live births and RA have been previously reported (23,25,57), although an increased risk of RA with more than 3 children was seen in 1 case-control study (26). An N-shaped risk was reported in yet a third study, a prospective cohort (66). If valid, the association between pregnancy

and RA could result directly from endocrineimmune disruption or from an immunologic reaction to paternal fetal DNA (67,68). As previously noted, it has been well established that RA often partially remits during pregnancy and ares in the postpartum period (15,69). These clinical observations during pregnancy have led to the theory that prolactin might be involved in disease initiation (14,70,71). Similar to other cohort studies, we found no association between parity and RA (25,27,66), whereas at least 1 investigator has reported an increased risk of RA with parity (26). Our study did not show an association with number of children breastfed and risk of RA. Information needed to gain further insight into this association, such as the duration of breastfeeding, was not available in our cohort. It remains to be established whether sex-hormone variations associated with pregnancy and breastfeeding might be causing disruptions to immune signaling (72), causing a defective hypothalamic-pituitary-adrenal axis (73), or having some other effect leading to a heightened risk of RA in older women. Strengths and Limitations Several strengths of this study are it populationbased prospective cohort design, comprehensive follow-up, and extensive exposure measures. Although we did not clinically examine cases, our strict validation criteria and the large noncase comparison group minimized the effects of case misclassication. This study also has limitations that inuence our ability to draw conclusions. The baseline survey had a 43% response rate, which may limit the ability to generalize. However, respondents and nonrespondents to the baseline survey showed only minor demographic differences (31), and have had similar cancer morbidity (including breast and endometrial cancer) and overall mortality rates (34), with the exception of smoking-related cancers. We had 80% power to detect RRs of at least 2.00, with an exposure prevalence as low as 8%. Expose prevalence of 8% inuenced the precision of our estimates for some factors, which is particularly important in the context of null ndings. Given this limitation in power, we were not able to look at specic clinical subgroups of RA (ie, rheumatoid-factor positive or rheumatoid-factor negative study participants). Selection bias caused by more frequent physician

80

MERLINO ET AL

visits is an unlikely factor resulting in an RA diagnosis in this cohort. Our review of more than 1000 records indicates that the vast majority of study participants saw their physicians specically because of the pain associated with the disease. We also examined socioeconomic factors, which may inuence medical care access, and none were associated with RA. It is still possible that patients seeking care for RA were also more likely to engage in other health behaviors; therefore, this form of selection bias may have occurred. For example, the association of drugs taken to suppress lactation and RA in the elderly might represent differential access to specialist care. Furthermore, despite the richness of our dataset, we did not have information of the duration, timing, formulation, or reason for use of exogenous hormones. Moreover, all of our measures of endogenous hormone exposure were surrogate self-report measures, not direct measures of hormonal levels. Lastly, the subjects

in this study were mainly white elderly women from the state of Iowa and the results may not be generalized to younger women or to other racial groups or geographic regions.
SUMMARY

Except for polycystic ovary syndrome, there were few female reproductive factors that showed a strong or statistically signicant association with the development of RA in elderly women in our study. The association of polycystic ovary syndrome and RA in older women may be indicative of perturbations of endocrine-immune activity that may inuence the development of RA or may represent a protective effect, delaying onset until an older age. Although further work is needed to conrm or to refute our ndings, we suggest that hormonal and gynecologic factors likely play only a modulating role in the development of late-onset RA.

REFERENCES
1. Linos A, Worthington JW, OFallon WM, Kurland LT. The epidemiology of rheumatoid arthritis in Rochester, Minnesota: A study of incidence, prevalence, and mortality. Am J Epidemiol 1980;111:87-98. 2. Lawrence RC, Helmick CG, Arnett FC, Deyo RA, Felson DT, Giannini EH, et al. Estimates of the prevalence of arthritis and selected musculoskeletal disorders in the United Sates [see comments]. Arthritis Rheum 1998;41:778-99. 3. Hochberg MC. Changes in the incidence and prevalence of rheumatoid arthritis in England and Wales, 1970-1982. Semin Arthritis Rheum 1990;19:294-302. 4. Silman AJ. Epidemiology of rheumatoid arthritis. APMIS 1994;19:294-302. 5. Wolfe F, Kleinheksel SM, Kahn MA. Familial vs sporadic rheumatoid arthritis: A comparison of the demographic and clinical characteristics of 956 patients. J Rheumatol 1998;15: 400-4. 6. Silman AJ, MacGregor AJ, Thomson W, Holligan S, Carthy D, Farhan A, et al. Twin concordance rates for rheumatoid arthritis: Results from a nationwide study. Br J Rheumatol 1993;32:903-7. 7. Aho K, Koskenvuo M, Tuominen J, Kaprio J. Occurrence of rheumatoid arthritis in a nationwide series of twins. J Rheumatol 1986;13:899-902. 8. Silman AJ, Ollier W, Holligan S, Birrell F, Adebajo A, Asuzu MC, et al. Absence of rheumatoid arthritis in a rural Nigerian population. J Rheumatol 1993;20:618-22. 9. Brighton SW, de la Harpe AL, van Staden DJ, Badenhurst JH, Myers OL. The prevalence of rheumatoid arthritis in a rural African population. J Rheumatol 1988;15:405-8. 10. Solomon L, Robin G, Valkenburg HA. Rheumatoid arthritis in an urban South African Negro population. Ann Rheum Dis 1975;34:128-35. 11. Lander ES, Schork NJ. Genetic dissection of complex traits [published erratum appears in Science 1994;266:353]. Science 1994;265:2037-48. 12. Seldin MF, Amos CI, Ward R, Gregersen PK. The genetics revolution and the assult on rheumatoid arthritis. Arthritis Rheum 1999;42:1071-9. 13. van Schaardenburg D, Hazes JM, de Boer A, Zwinderman AH, Meijers KA, Breedveld FC. Outcome of rheumatoid arthritis in relation to age and rheumatoid factors at diagnosis. J Rheumatol 1993;20:45-52. 14. Brennan P, Silman A. Breast-feeding and the onset of rheumatoid arthritis. Arthritis Rheum 1994;37:808-13. 15. Barrett JH, Brennan P, Fiddler M, Silman AJ. Does rheumatoid arthritis remit during pregnancy and relapse postpartum? Results from a nationwide study in the United Kingdom performed prospectively from late pregnancy [see comments]. Arthritis Rheum 1999;42:1219-27. 16. Barrett JH, Brennan P, Fiddler M, Silman A. Breastfeeding and postpartum relapse in women with rheumatoid and inammatory arthritis. Arthritis Rheum 2000;43:1010-5. 17. Silman A, Kay A, Brennan P. Timing of pregnancy in relation to the onset of rheumatoid arthritis. Arthritis Rheum 1992;35:152-5. 18. Brennan P, Bankhead C, Silman A, Symmons D. Oral contraceptives and rheumatoid arthritis: Results from a primary care-based incident case-control study. Semin Arthritis Rheum 1997;26:817-23. 19. Carette S, Marcoux S, Gingras S. Postmenopausal hormones and the incidence of rheumatoid arthritis. J Rheumatol 1989;16:911-3. 20. del Junco DJ, Annegers JF, Luthra HS, Coulam CB, Kurland LT. Do oral contraceptives prevent rheumatoid arthritis? JAMA 1985;254:1938-41. 21. Hernandez-Avila M, Liang MH, Willett WC, Stampfer MJ, Colditz GA, Rosner B, et al. Oral contraceptives, replace-

ASSOCIATION OF ESTROGEN WITH RA

81

ment oestrogens and the risk of rheumatoid arthritis. Br J Rheumatol 1989;28(suppl 1):31, discussion 42-5. 22. Koepsell TD, Dugowson CE, Nelson JL, Voight LF, Darling JR. Non-contraceptive hormones and the risk of rheumatoid arthritis in menopausal women. Int J Epidemiol 1994; 23:1248-55. 23. Linos A, Worthington JW, OFallon WM, Kurland LT. Case-control study of rheumatoid arthritis and prior use of oral contraceptives. Lancet 1983;1:1299-300. 24. Spector TD, Brennan P, Harris P, Studd JW, Silman AJ. Does estrogen replacement therapy protect against rheumatoid arthritis? J Rheumatol 1991;18:1473-6. 25. Hernandez Avila M, Liang MH, Willett WC, Stampfer MJ, Colditz GA, Rosner B, et al. Reproductive factors, smoking, and the risk for rheumatoid arthritis. Epidemiology 1990; 1:285-91. 26. Jorgensen C, Picot MC, Bologna C, Sany J. Oral contraception, parity, breast feeding, and severity of rheumatoid arthritis. Ann Rheum Dis 1996;55:94-8. 27. Vessey MP, Villard-Mackintosh L, Yeates D. Oral contraceptives, cigarette smoking and other factors in relation to arthritis. Contraception 1987;35:457-64. 28. Pease CT, Bhakta BB, Devlin J, Emery P. Does the age of onset of rheumatoid arthritis inuence phenotype? A prospective study of outcome and prognostic factors. Rheumatology (Oxford) 1999;38:228-34. 29. Deal CL, Meenan RF, Goldenberg DL, Anderson JJ, Sack B, Pastan RS, et al. The clinical features of elderly-onset rheumatoid arthritis. A comparison with younger-onset disease of similar duration. Arthritis Rheum 1985;28:987-94. 30. Terkeltaub R, Esdaile J, Decary F, Tannenbaum H. A clinical study of older age rheumatoid arthritis with comparison to a younger onset group. J Rheumatol 1983;10:418-24. 31. Folsom AR, Kaye SA, Prineas RJ, Potter JD, Gapstur SM, Wallace RB. Increased incidence of carcinoma of the breast associated with abdominal adiposity in postmenopausal women. Am J Epidemiol 1990;131:794-803. 32. Sellers TA, Kushi LH, Potter JD, Kaye SA, Nelson CL, McGovern PG, et al. Effect of family history, body-fat distribution, and reproductive factors on the risk of postmenopausal breast cancer [published erratum appears in N Engl J Med 1992;327:1612]. N Engl J Med 1992;326:1323-9. 33. Folsom AR, Mink PJ, Sellers TA, Hong CP, Zheng W, Potter JD. Hormonal replacement therapy and morbidity and mortality in a prospective study of postmenopausal women. Am J Public Health 1995;85(8 pt 1):1128-32. 34. Bisgard KM, Folsom AR, Hong CP, Sellers TA. Mortality and cancer rates in nonrespondents to a prospective study of older women: 5-year follow-up. Am J Epidemiol 1994;139: 990-1000. 35. Folsom AR, Kaye SA, Sellers TA, Hong CP, Cerhan JR, Potter JD, et al. Body fat distribution and 5-year risk of death in older women [published erratum appears in JAMA 1993;269: 1254]. JAMA 1993;269:483-7. 36. Mikuls TR, Saag KG, Cerhan JR, Merlino LA, Criswell LA. Coffee, tea, and caffeine consumption and risk of rheumatoid arthritis: Results from the Iowa Womens Health Study. Arthritis Rheum 2002;46:83-91. 37. Cerhan JR, Saag KG, Criswell LA, Merlino LA, Mikuls TR. Blood transfusion, alcohol use and anthropometric risk

factors for rheumatoid arthritis in older women. J Rheum 2002;29:246-54. 38. Criswell LA, Merlino LA, Cerhan JR, Mikuls TR, Mudano AS, Burma M, et al. Cigarette smoking is a major risk factor for elderly onset rheumatoid arthritis: Results from the Iowa Womens Health Study. Am J Med 2002;112:465-71. 39. MacGregor AJ, Silman AJ. A reappraisal of the measurement of disease occurrence in rheumatoid arthritis. J Rheumatol 1992;19:1163-5. 40. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classication of rheumatoid arthritis. Arthritis Rheum 1988;31:315-24. 41. MacGregor AJ, Bamber S, Silman AJ. A comparison of the performance of different methods of disease classication for rheumatoid arthritis. Results of an analysis from a nationwide twin study. J Rheumatol 1994;21:1420-6. 42. Cox D. Regression models and life tables (with discussion). J R Stat Soc Ser B 1972;34:187-220. 43. SAS/STAT [computer program]. Changes and enhancements through release 6.12. Version 6.12. Cary, NC:SAS Institute, 1997. 44. Young RL, Goldzieher JW. Clinical manifestations of polycystic ovarian disease. Endocrinol Metab Clin North Am 1988;17:621-35. 45. McKenna TJ. Pathogenesis and treatment of polycystic ovary syndrome. N Engl J Med 1988;318:558-62. 46. Hunter MH, Sterrett JJ. Polycystic ovary syndrome: Its not just infertility. Am Fam Physician 2000;62:1079-88, 1090. 47. Gammon MD, Thompson WD. Polycystic ovaries and the risk of breast cancer. Am J Epidemiol 1991;134:818-24. 48. Anderson KE, Sellers TA, Chen PL, Rich SS, Hong CP, Folsom AR. Association of Stein-Leventhal syndrome with the incidence of postmenopausal breast carcinoma in a large prospective study of women in Iowa. Cancer 1997;79:494-9. 49. Cutolo M. Do sex hormones modulate the synovial macrophages in rheumatoid arthritis? Ann Rheum Dis 1997;56: 281-4. 50. Gleicher N, el-Roeiy A, Conno E, Friberg J. Is endometriosis an autoimmune disease? Obstet Gynecol 1987;70: 115-22. 51. Gleicher N, Pratt D. Abnormal (auto)immunity and endometriosis. Int J Gynaecol Obstet 1993;40(suppl):S21-7. 52. Lucena E, Cubillos J. Immune abnormalities in endometriosis compromising fertility in IVF-ET patients. J Reprod Med 1999;44:458-64. 53. Nelson JL, Koepsell TD, Dugowson CE, Voigt LF, Daling JR, Hansen JA. Fecundity before disease onset in women with rheumatoid arthritis [see comments]. Arthritis Rheum 1993;36:7-14. 54. Sinaii N, Cleary SD, Ballweg ML, Nieman LK, Stratton P. High rates of autoimmune and endocrine disorders, bromyalgia, chronic fatigue syndrome and atopic diseases among women with endometriosis: A survey analysis. Hum Reprod 2002;17:2715-24. 55. Overlie I, Moen MH, Morkrid L, Skjaeraasen JS, Holte A. The endocrine transition around menopausea ve years prospective study with proles of gonadotropines, estrogens, androgens and SHBG among healthy women. Acta Obstet Gynecol Scand 1999;78:642-7. 56. Goemaere S, Ackerman C, Goethals K, De Keyser F,

82

MERLINO ET AL

Van der Straeten C, Verbruggen G, et al. Onset of symptoms of rheumatoid arthritis in relation to age, sex and menopausal transition. J Rheumatol 1990;17:1620-2. 57. Brun JG, Nilssen S, Kvale G. Breast feeding, other reproductive factors and rheumatoid arthritis. A prospective study. Br J Rheumatol 1995;34:542-6. 58. Cooper GS, Sandler DP, Bohlig M. Active and passive smoking and the occurrence of natural menopause. Epidemiology 1999;10:771-3. 59. Kato I, Toniolo P, Akhmedkhanov A, Koenig KL, Shore R, Zeleniuch-Jacquotte A. Prospective study of factors inuencing the onset of natural menopause. J Clin Epidemiol 1998;51:1271-6. 60. Jeffcoate TNA, Lister UM, Hargreaves B, Roberts H. Ethinyl oestradiol. Br Med J November 1948:809-12. 61. Tyson JE. A high-dose estrogen for lactation suppression. Obstet Gynecol 1966;27:729-32. 62. Hodge C. Suppression of lactation by stilboestrol. Lancet 1967;2:286-7. 63. Masi AT, Bijlsma JW, Chikanza IC, Pitzalis C, Cutolo M. Neuroendocrine, immunologic, and microvascular systems interactions in rheumatoid arthritis: Physiopathogenetic and therapeutic perspectives. Semin Arthritis Rheum 1999;29:65-81. 64. Hernandez-Avila M, Liang MH, Willett WC, Stampfer MJ, Colditz GA, Rosner B, et al. Exogenous sex hormones and the risk of rheumatoid arthritis [see comments]. Arthritis Rheum 1990;33:947-53. 65. Hazes JM, Dijkmans BC, Vandenbroucke JP, de Vries RR, Cats A. Reduction of the risk of rheumatoid arthritis among women who take oral contraceptives [see comments]. Arthritis Rheum 1990;33:173-9.

66. Heliovaara M, Aho K, Reunanen A, Knekt P, Aromaa A. Parity and risk of rheumatoid arthritis in Finnish women. Br J Rheumatol 1995;34:625-8. 67. Nelson JL, Hughes KA, Smith AG, Nisperos BB, Branchaud AM, Hansen JA. Maternal-fetal disparity in HLA class II alloantigens and the pregnancy-induced amelioration of rheumatoid arthritis. N Engl J Med 1993;329:466-71. 68. van der Horst-Bruinsma IE, de Vries RR, de Buck PD, van Schendel PW, Breedveld FC, et al. Inuence of HLA-class II incompatibility between mother and fetus on the development and course of rheumatoid arthritis of the mother. Ann Rheum Dis 1998;57:286-90. 69. Lansink M, de Boer A, Dijkmans BA, Vandenbroucke JP, Hazes JM. The onset of rheumatoid arthritis in relation to pregnancy and childbirth. Clin Exp Rheumatol 1993;11:171-4. 70. Brennan P, Ollier B, Worthington J, Hajeer A, Silman A. Are both genetic and reproductive associations with rheumatoid arthritis linked to prolactin? Lancet 1996;348:106-9. 71. Brennan P, Hajeer A, Ong KR, Worthington J, John S, Thomson W, et al. Allelic markers close to prolactin are associated with HLA-DRB1 susceptibility alleles among women with rheumatoid arthritis and systemic lupus erythematosus. Arthritis Rheum 1997;40:1383-6. 72. Chikanza IC, Grossman AB. Neuroendocrine immune responses to inammation: The concept of the neuroendocrine immune loop. Baillieres Clin Rheumatol 1996;10:199-225. 73. Da Silva JA. Sex hormones and glucocorticoids: Interactions with the immune system. Ann N Y Acad Sci 1999;876: 102-17, discussion 117-8.