This study investigates whether experiences of physical or sexual violence over the life course affect ovarian function. We evaluated whether ovarian hormone levels indicative of menopausal changes are observed at an earlier age among women who reported experiences of violence during childhood, adolescence, or adulthood compared with women who reported no exposure to violence. Our hypothesis is that violence affects neuroendocrine activity, specifically hypothalamic-pituitary-adrenal (HPA) axis regulation of ovarian function, leading to altered age of onset of menopause.
Age at menopause, defined as 1 year without menstruation, is an important determinant of women’s postmenopausal health. Early age at menopause may be associated with elevated risk of cardiovascular disease, whereas late age at menopause is associated with elevated risk of breast cancer and endometrial cancer. 1–3 In the United States, the estimated mean age of menopause is 50–51 years. 4–6 Factors associated with the timing of menopause have not been well characterized. 7 Earlier age of menopause has been associated with shorter menstrual cycles, 4,8 nulliparity, 4,8–10 lower levels of education, 8,9,11,12 lower income, 11,12 blue-collar occupation, 9 being a woman of color, 8 earlier age at menarche, 13 and cigarette smoking, 8–11,14 whereas use of oral contraceptives, 12,13 irregular menstrual periods, 8,12 and multiple pregnancies 4,12,13 have been associated with later age at menopause.
Several investigations have reported that trauma can affect HPA axis function, the limbic system, and possibly gene expression for hormone receptors in the brain, with childhood sexual victimization being associated with chronic HPA axis activation. 15–22 The hypothalamus regulates menstrual function by secreting gonadotropin-releasing hormone in pulses, which stimulates the periodic release of both follicle-stimulating hormone (FSH) and luteinizing hormone (LH) from the pituitary gland. FSH is necessary for follicular maturation and growth, whereas LH stimulates estradiol (E2) secretion by the maturing follicle and helps to maintain the corpus luteum. 23 As women deplete their reserve of follicles capable of developing functioning granulose-thecal components necessary for steroid production and feedback inhibition of the pituitary, evidence of ovarian decline is observed by the profiles of high FSH and low E2. To our knowledge, no study has directly investigated whether violent experiences known to affect functioning of the HPA axis also affect age at onset of menopause.
Subjects and Methods
This study was based on an existing cohort of women participating in the Harvard Study of Moods and Cycles, a longitudinal study investigating the relation between ovarian function and depression among premenopausal women. Between 1995 and 1997, 976 women in the Greater Boston area were enrolled from a population-based, random sample of 6,222 women between 36 and 44 years of age (see Harlow et al24 for sampling details). We mailed the survey to the 907 active members of the longitudinal cohort in March 1999. A second questionnaire was mailed to 344 nonrespondents approximately 5–8 weeks later. Seven hundred thirty-six women completed the survey.
To be eligible for the current study, women had to be premenopausal, have information on hormone level measurements, and have completed the violence history. The total number of women meeting these criteria was 732.
Assessment of Ovarian Function
To determine whether ovarian hormone levels were indicative of perimenopausal changes, we obtained an early follicular phase hormone measurement, drawing blood between days 2 and 5 of each woman’s menstrual cycle. At the time of the blood draw, we assessed regularity of the menstrual cycle and events that could affect basal hormone levels, such as pregnancy, breastfeeding, and exogenous hormone use. Measurement of FSH and E2 were performed using Coat-A-Count immunoradioassays; E2 was measured in duplicate, and the levels were averaged. The World Health Organization Second International Reference Preparation (78/549) of human pituitary FSH was used as the standard for FSH. Intra-assay coefficients of variation were <4% for FSH and <7% for E2. Interassay coefficients of variation were <10% for FSH and <15% for E2. The lower levels of detection were 0.06 mIU/ml for the FSH assay and 8 pg/ml for the E2 assay. Laboratory staff were unaware of the study purpose and had no information on cohort members.
Assessment of Physical and Sexual Abuse
We assessed exposure to physical or sexual abuse during three life stages, childhood (up to 11 years of age), adolescence (12–18 years of age), and adulthood (≥19 years of age). Cohort members completed a self-administered questionnaire on experiences and threats of abuse, which we had modified from the Conflict Tactics Scale 25 and the Abuse Assessment Screen. 26 Women were asked questions about their lifetime exposure to experienced, witnessed, or threatened physical and sexual harm. Sexual harm was defined as any experience or fear of being forced to be sexual against one’s will, any sexual assault, or “more than a few times” when exposed to someone’s genitals. Physical harm was defined as “more than a few times” when the respondent reported having had any of the following: having been pushed, grabbed, or shoved; having had assailant throw something that could or did injure; having been hit with something that hurt the body; having been choked, burned, or hurt with hot water; having been spanked; having been physically attacked in some other way; or having witnessed any of these done to a member of her household. A final criterion for physical harm was a single experience or fear that she or someone she loved would be killed.
We measured associations between abuse and ovarian function by estimating the risk difference (RD) of high FSH and low E2 according to violence history (any harm vs none). We defined “high” FSH and “low” E2 according to the cutoff values for the upper third of the FSH distribution and lower third for E2 distribution by age group (36–40 and 41–45 years of age) among never-abused women. We also graphed the estimated mean change in hormone levels with age derived from linear regression models. We used natural log transformation to normalize the skewed distributions of FSH and E2 for these plots.
Preliminary analyses indicated that childhood and adolescent harm had similar associations with hormone levels; thus, we collapsed these exposure periods. Likewise, physical and sexual abuse each had similar associations with ovarian function; hence, we collapsed these exposures into a single abuse category.
We used linear regression to evaluate education (<graduate school/graduate school), economic well-being (ability to maintain standard of living if no income <2 months vs 2 or more months), current cigarette use (any/none), nulliparity (yes/no), body mass index (<25 vs ≥25 kg/m2), any current birth control pill use (yes/no), and age at menarche (≤11, 12–13, or ≥ 14 years of age) as potential confounders. 8,14,27 We defined confounding as a change in the abuse coefficients of 10% or more relative to the crude estimate. In addition, we adjusted for age at menarche because others had reported it as a confounder. 10,12,13,28
Table 1 presents sociodemographic characteristics of the 732 women who returned the lifetime abuse questionnaire and for whom we had hormone information. Four hundred thirty (59%) women reported some experience or fear of abuse. There was little difference in the distributions of age, race/ethnicity, monthly household income, nulliparity, current birth control pill use, early age at menarche, and body mass index among women who had ever vs never experienced or feared physical or sexual abuse. Women with abuse histories, however, were less likely to have a graduate degree, more likely to be a current cigarette smoker, and less likely to have assets other than wages that could support them beyond 2 months, compared with women who had never experienced abuse.
Table 2 presents the crude and adjusted estimates of RDs for the probability of having an FSH level in the highest tertile including and excluding current birth control use. Among women not using birth control who experienced abuse during childhood or adolescence there was a positive association between violence and high FSH [36–40 years of age, adjusted RD (RDadj) = 0.03, 95% confidence limits (CL) = −0.08, 0.13; 41–45 years of age, RDadj = 0.08, 95% CL = −0.03, 0.19], whereas among women who experienced first abuse during adulthood there was a decrease in the estimate for risk of high FSH levels, regardless of age (36–40 years of age, RDadj = −0.08, 95% CL = −0.33, 0.16; 41–45 years of age, RDadj = −0.13, 95% CL = −0.35, 0.09).
Table 3 presents the crude and adjusted estimates of RDs for the probability of having an E2 level in the lowest tertile including and excluding current birth control use. For women 36–40 years of age not currently using birth control, no association was apparent between violence before adulthood and low E2 level (RDadj = 0.00, 95% CL = −0.10, 0.10). Violence in adulthood, however, was associated with an 11% increase in the adjusted estimate of risk of low E2 relative to women reporting no violence in their lifetime (95% CL = −0.14, 0.36). There was an observed positive association for violence during any period in relation to ovarian function among women more than 40 years of age, regardless of life stage during first experience of abuse (before adulthood RDadj = 0.17, 95% CL = 0.06, 0.28; during adulthood RDadj = 0.23, 95% CL = 0.01, 0.46).
Figure 1 graphs the age-specific means of log FSH levels in relation to abuse history. We depict estimates for differences in mean log FSH level within each chronological age according to the period of first experience of abuse (childhood or adolescence; adulthood) as well as levels in those with no history of harm. Women with history of violence had age-specific FSH levels that were higher relative to never-abused women. For example, the log FSH levels of 40-year-old women who were exposed to abuse during childhood/adolescence or adulthood were approximately the same as for 45-year-old women with no history of abuse.
Figure 2 graphs the differences in log E2 levels in relation to abuse history. Consistent with progression toward menopause, age-specific log E2 levels were lower among women with childhood/adolescent or adulthood abuse histories relative to women without these experiences.
Because we selected the cohort to study the effect of depression, we were concerned whether abuse was an independent risk factor from depression. We used regression models to assess the role of depression as an effect modifier of the relation between abuse and ovarian function. Although depression had an independent effect on hormone levels, there was no evidence of synergy between depression and abuse history. Specifically, analyses of the subset of 424 never-depressed women found little or no difference in the estimates of adjusted RD. Therefore, the models presented do not include evaluation of effect measure modification of the abuse-hormone level relation by depression.
We observed more extreme levels of both FSH and E2 in relation to abuse history among premenopausal women more than 41–45 years of age, whereas little difference was seen for younger women. Yet a number of sources of error may have affected these findings.
First, if women who responded to the questionnaire were more likely to have an earlier menopause and a history of violence than women who did not complete the questionnaire, then the observed relation would be an overestimate. Because 81% of the cohort provided information and because women who did not respond to the questionnaire did not differ from respondents with respect to rates of both major depression and posttraumatic stress disorder as measured at baseline using the Structured Clinical Interview for DSM-III-R (Patient Edition), 29 the effect of such a bias was likely reduced. A second source of error could be that women with hormone levels in the highest tertile for FSH or lowest tertile for E2 had differential recall of abuse experiences. Women were unaware of their hormone levels when responding to the questions on physical and sexual harm or of the study hypothesis, however. Exposure to abuse may have also been underreported nondifferentially with respect to hormone levels. 30 This error would tend to dilute true differences. Also, we may have misclassified nonabused women as abused because of our broad definition of abuse. To evaluate this possibility, we conducted analyses with restricted definitions of abuse and obtained similar estimates of RD, but with less precision owing to the smaller number of exposed women. Finally, in characterizing the hormone levels of women at age at enrollment, rather than longitudinally, we assumed that the cross-sectional levels are representative of the longitudinal experience of women at 36 years of age as if they had been followed for 10 years.
Of note, for women who first experienced violence during adulthood there appeared to be an increase in mean FSH level, whereas the probability of high FSH was lower than for women with no history of abuse. One possible explanation for this finding could be that the mean hormone value was more sensitive to extreme hormone levels than the categorical measure.
Several physiological mechanisms support a plausible biological link between abuse, especially occurring early in life, and adult ovarian function. First, frightening experiences activate the HPA axis, 31 simultaneously increasing necessary survival processes, such as blood pressure, flow of blood to muscles, and circulating levels of glucose, while decreasing processes related to nonessential functions to conserve energy. 32 Activation of the HPA axis initiates secretion of corticotropin-releasing hormone, which, in turn, stimulates the anterior pituitary to release adrenocorticotropin-releasing hormone (ACTH). ACTH acts on the adrenal cortex and stimulates secretion of glucocorticoids. Glucocorticoids affect ovarian function by acting on the hypothalamus, the anterior pituitary gland, and the ovaries. 33–36 Specifically, their functions include inhibition of synthesis and release of gonadotropin-releasing hormone, FSH and LH, and modulation of oogenesis. They interact with receptors in the hippocampus that inhibit further release of corticotropin-releasing hormone and ACTH via a negative feedback mechanism and in essence, “turn off” the adaptation process. Prolonged or excessive levels of glucocorticoids, however, can damage the hippocampus and reduce the ability to “turn off” this process. The resultant damage to these feedback mechanisms results in HPA axis dysregulation and thus impairment of the neuroendocrine pathways influencing ovarian function. Experimental evidence supports this hypothesis; chronic elevation of glucocorticoids in pigs has been shown to suppress ovulation and maturation of oocytes. 36
To our knowledge, no epidemiologic study has directly studied associations between ovarian hormone levels and history of abuse. Notably, however, a prospective study by Bromberger et al8 found that psychosocial stressors were associated with a decrease in the median age of menopause of 1.5 years in African-American women (49.9 vs 48.4 years), but not in white women. They concluded that “African-American women may have a different ‘biological clock’ than white women, especially when under stress, or they may experience more stress of longer duration (p. 124).”
From a public health standpoint, determinants of increased ovarian aging are important to understand, given evidence of greater morbidity associated with early menopause. For example, the Nurses’ Health Study found that for each 1-year decrease in age at menopause there was an increased risk of coronary heart disease (RR = 1.03, 95% CL = 1.01, 1.05). 1 Other investigators found that delay in age at menopause was associated with a decrease in the age-adjusted hazard ratio (HR = 0.982; 95% CL = 0.968, 0.996) for cardiovascular mortality. 3 Our results support the idea that physical and sexual violence may potentially contribute to early onset of menopause, which may lead to increased risk of cardiovascular disease after menopause.
Special thanks to Lauren Wise, Harvard School of Public Health, and the women who participated in this study.
1. Hu FB, Grodstein F, Hennekens CH, Colditz GA, Johnson M, Manson JE, Rosner B, Stampfer MJ. Age at natural menopause and risk of cardiovascular disease. Arch Intern Med 1999; 159: 1061–1066.
2. Monninkhof EM, van der Schouw YT, Peeters PH. Early age at menopause and breast cancer: are leaner women more protected? A prospective analysis of the Dutch DOM cohort. Breast Cancer Res Treat 1999; 55: 285–291.
3. van der Schouw YT, van der Graaf Y, Steyerberg EW, Eijkemans JC, Banga JD. Age at menopause as a risk factor for cardiovascular mortality. Lancet 1996; 347: 714–718.
4. Whelan EA, Sandler DP, McConnaughey DR, Weinberg CR. Menstrual and reproductive characteristics and age at natural menopause. Am J Epidemiol 1990; 131: 623–632.
5. Greendale GA, Sowers MF. The menopause transition. Endocrinol Metab Clin North Am 1997; 26: 261–277.
6. Greendale GA, Lee NP, Arriola ER. The menopause. Lancet 1999; 353: 571–580.
7. Sowers MR. Menopause: Its Epidemiology. In: Goldman MB, Hatch MC, eds. Women and Health. San Diego: Academic Press, 2000; 1155–1168.
8. Bromberger JT, Matthews KA, Kuller LJ, Wing RR, Meilahn EN, Plantinga P. Prospective study of the determinants of age at menopause. Am J Epidemiol 1997; 145: 124–133.
9. Luoto R, Kaprio J, Uutela A. Age at natural menopause and sociodemographic status in Finland. Am J Epidemiol 1994; 139: 64–76.
10. Parazzini F, Negri E, LaVecchia C. Reproductive and general lifestyle determinants of age at menopause. Maturitas 1992; 15: 141–149.
11. Brambilla DJ, McKinlay SM. A prospective study of factors affecting age at menopause. J Clin Epidemiol 1989; 42: 1031–1039.
12. Stanford JL, Hartge P, Brinton LA, Hoover RN, Brookmeyer R. Factors influencing the age at natural menopause. J Chron Dis 1987; 40: 995–1002.
13. Khaw KT. Epidemiology of the menopause. Br Med Bull 1992; 48: 249–261.
14. Hardy R, Kuh D, Wadsworth M. Smoking, body mass index, socioeconomic status and the menopausal transition in a British national cohort. Int J Epidemiol 2000; 29: 845–851.
15. Heim C, Newport DJ, Heit S, Graham YP, Wilcox M, Bonsall R, Miller AH, Nemeroff CB. Pituitary-adrenal and autonomic responses to stress in women after sexual and physical abuse in childhood. JAMA 2000; 284: 592–597.
16. Horan DL, Hill LD, Schulkin J. Childhood sexual abuse and preterm labor in adulthood: an endocrinological hypothesis. Womens Health Issues 2000; 10: 27–33.
17. Heim C, Ehlert U, Hanker JP, Hellhammer DK. Abuse-related posttraumatic stress disorder and alterations of the hypothalamic-pituitary-axis in women with chronic pelvic pain. Psychosom Med 1998; 60: 309–318.
18. De Bellis MD, Lefter L, Trickett PK, Putnam FW. Urinary catecholamine excretion in sexually abused girls. J Am Acad Child Adolesc Psychiatry 1994; 33: 320–327.
19. De Bellis MD, Chrousos GP, Dorn LD, Burke L, Helmers K, Kling MA, Trickett PK, Putnam FW. Hypothalamic-pituitary-adrenal axis dysregulation in sexually abused girls. J Clin Endocrinol Metab 1994; 78: 249–255.
20. De Bellis MD, Baum AS, Birmaher B, Keshavan MS, Eccard CH, Boring AM, Jenkins FJ, Ryan ND. Developmental traumatology. Part I. Biological stress systems. Biol Psychiatry 1999; 45: 1259–1270.
21. Lemieux AM, Coe CL. Abuse-related posttraumatic stress disorder: evidence for chronic neuroendocrine activation in women. Psychosom Med 1995; 57: 105–115.
22. Teicher MH, Glod CA, Surrey J, Swett C. Early childhood abuse and limbic system ratings in adult psychiatric outpatients. J Neuropsychiatry Clin Neurosci 1993; 5: 301–306.
23. Wilson JD, Foster DW, Kronenberg HM, Larsen PR. Williams Textbook of Endocrinology. Philadelphia: WB Saunders, 1998.
24. Harlow BL, Cohen LS, Otto MW, Spiegelman D, Cramer DW. Prevalence and predictors of depressive symptoms in older premenopausal women. Arch Gen Psychiatry 1999; 56: 418–424.
25. Straus MA. Measuring family conflict and violence: the Conflicts Tactics Scale. J Marriage Family 1979; 41: 75–88.
26. McFarlane J, Parker B, Soeken K, Bullock L. Assessing for abuse during pregnancy: severity and frequency of injuries and associated entry into prenatal care. JAMA 1992; 267: 3176–3178.
27. Harlow SD, Ephross SA. Epidemiology of menstruation and its relevance to women’s health. Epidemiol Rev 1995; 17: 265–302.
28. Flint M. Is there a secular trend in age of menopause? Maturitas 1978; 1: 133–139.
29. Spitzer RL, Williams JBW, Gibbon M, First MB. Structured Clinical Interview for DSM-III-R: Patient Edition (SCID-P). Washington DC: American Psychiatric Press, 1990.
30. National Research Council. Understanding Violence Against Women. Washington DC: National Research Council, 1996.
31. Raber J. Detrimental effects of chronic hypothalamic-pituitary-adrenal axis activation. Mol Neurobiol 1998; 18: 1–22.
32. Evans RG, Hodge M, Pless IB. If not genetics, then what? Biological pathways and population health. In: Evans RG, Barer ML, Marmor TR, eds. Why Are Some People Healthy and Others Not? The Determinants of Health of Populations. New York: Walter de Gruyter, 1994; 161–188.
33. Webster EL, Torpy DJ, Elenkov IJ, Chrousos GP. Corticotropin-releasing hormone and inflammation. Ann NY Acad Sci 1998; 840: 21–32.
34. Yehuda R, Giller EL, Southwick SM, Lowy MT, Mason JW. Hypothalamic-pituitary-adrenal dysfunction in posttraumatic stress disorder. Biol Psychiatry 1991; 30: 1031–1048.
35. Schobitz B, Reul JMHM, Holsboer F. The role of the hypothalamic-pituitary-adrenocortical system during inflammatory conditions. Crit Rev Neurobiol 1994; 8: 263–291.
36. Yang JG, Chen WY, Li PS. Effects of glucocorticoids on maturation of pig oocytes and their subsequent fertilizing capacity in vitro
. Biol Reprod 1999; 60: 929–936.