Association of ethnicity with involuntary childlessness and perceived reasons for infertility: baseline data from the Study of Women's Health Across the Nation (SWAN).

In the United States, infertility (defined as failure to conceive after one year of regular unprotected intercourse) is estimated to affect 8-20% of couples ( 1 , 2 ). Involuntary childlessness is associated not only with the traditional factors affecting fertility, such as reproductive aging and history of tubal disease, but also with sociocultural and economic characteristics, including access to health care, treatment seeking behavior, and perception of nulliparity. Theories attempting to explain trends in childlessness have hypothesized that as countries become more developed, involuntary childlessness decreases due to a lower incidence of malnutrition and infectious diseases affecting the reproductive system. At the same time, however, rates of voluntary childlessness have been noted to escalate, likely due to increased educational and employment opportunities for women and delayed childbearing ( 3 ). The distinction between voluntary and involuntary childlessness begins to blur in some cases, as childbearing delayed beyond the mid-4 th decade of life may become involuntary infertility due to diminished ovarian reserve ( 4 ). A number of studies have investigated differences in childlessness and its involuntary component among various ethnicities. Mosher and Pratt used the data from the 1982 National Survey of Family Growth to analyze fecundity status and related factors ( 5 ). In this study, African-American women were twice as likely as Caucasians to have had pelvic inflammatory disease leading to infertility. While other studies have reported high rates of childlessness among African-Americans, this was at least partially attributed to social factors, such as less time spent in marriage ( 6 , 7 ). In addition, an emerging body of literature points toward reduced access to infertility care among US ethnic minorities ( 8 ). Taken together, the data imply that ethnic minorities have greater exposure to risk factors for infertility and have reduced access to treatment. A comprehensive understanding of involuntary childlessness and ethnicity necessitates a close look at differences in etiology of infertility. However, ethnic differences in women’s self-reported perception of their own infertility, and whether they know of its etiology at all, have rarely been examined. The Study of Women’s Health Across the Nation (SWAN) is an appropriate platform to elucidate the sociocultural factors and ethnic differences associated with involuntary childlessness and etiology of infertility as it represents a large, multiethnic cohort of women that collected detailed information on reproductive history.

The SWAN, a multiethnic, multi-center longitudinal study of women as they progress through menopausal transition, has been previously described ( 9 ). Screening surveys were administered by telephone to women selected from a sampling frame based on random digit dialing (Los Angeles, Pittsburgh, New Jersey) or household lists (Boston, Chicago, Detroit, Oakland). Of the eligible women, 3302 were recruited (51%). At each site, approximately half of the women were of one of the four purposely-sampled ethnic backgrounds (African American, Chinese, Japanese, or Hispanic), in addition to Caucasians. Eligibility criteria included: age 42-52, at least one menstrual period and no hormonal therapy within the prior three months, intact uterus and at least one ovary. Institutional Review Board (IRB) approval for SWAN was obtained from all appropriate institutions. History of infertility was ascertained by the question “Have you ever had a period of 12 months when you could not get pregnant although you were attempting to get pregnant or were letting yourself get pregnant?” The respondents were further probed by the question “Did a doctor give you a reason why you were not getting pregnant?” Those who responded affirmatively were asked to fill in a reason. Abstraction of this write-in variable was done by two reviewers independently (NS, AJP). Available responses were abstracted to fit into the known causes of infertility: unexplained, tubal factor, male factor, anovulation, diminished ovarian reserve, endometriosis and uterine factor. Some answers represented attributes not medically known to cause infertility (such as ‘tipped uterus’) and were coded as a distinct category. Disagreements were adjudicated (inter-observer correlation of 94%). The following dichotomous variables were created: use of fertility medications, history of eating disorders, and history of salpingitis. Menstrual regularity was ascertained and a oligomenorrhea variable was created (any history of amenorrhea ≥ 3 months not related to pregnancy or breast-feeding vs. none). Continuous oral contraceptive (OCP) use between the ages of 25 to 35 was also ascertained. Sexual practices were addressed by asking, ‘Who have you have generally had sex with over your adult lifetime?” All responses stating that a participant “never had sex” or had“sex with a woman” were used to create a corresponding variable: report of sexual preference for women or not sexually active vs. else (defined as preference for men or sex with men and women). This re-coded variable is referred to as “not sexually active or predominantly homosexual sex”. Race/ethnicity was self-identified. Marital status was dichotomized as ever married vs. never. Smoking was dichotomized as ever vs. never. Education was dichotomized based on college attendance. Socioeconomic status was categorized into 2 levels based on ability to pay for basic needs, such as food and shelter . Income was categorized into low income (<$35,000 annually), middle income ($35,000 – $74,999 annually), and high income ($75,000 or higher annually). Involuntary childlessness was used as a descriptor of self-reported unresolved infertility. Participants who reported a history of infertility and were nulliparous were deemed involuntarily childless. Perceived etiology of infertility was used as a descriptor of self-reported reasons for “not getting pregnant” as given to the participants by a physician. Of the total SWAN cohort of 3302 women, 153 women were excluded from this analysis because of missing pregnancy history and/or missing infertility history. Therefore, the final analytic sample included 3149 women. Data from 302 of these women (subjects who reported a history of infertility and stated a given etiology of infertility) were included in our analyses regarding self-reported etiology of infertility. Associations between demographic and clinical characteristics of the cohort by ethnicity were assessed using the Student t-test, ANOVA or the Kruskal-Wallis test for continuous variables and chi square or Fisher’s exact tests for categorical variables. To evaluate the independent association of ethnicity on the outcomes, multivariable logistic models were specified using model-building strategies suggested by Hosmer and Lemeshow ( 10 ). Sensitivity analyses were conducted by excluding those who reported history of continuous OCP use, and those who reported a history of fertility drug use. Lastly, a combined sensitivity analysis was performed, excluding all of the above-mentioned groups. All statistical tests used a two-tailed alpha of 0.05. All analyses were performed using SPSS 17.0 (SPSS Chicago, IL).

Table 1 describes the demographics and fertility determinants of the cohort by ethnicity. In almost every case, significant differences in known risk factors for infertility were reported among various ethnic groups. Prevalence of self-reported history of fallopian tube infection was highest among African- American women. Considerable differences in demographic variables were also evident, with Caucasian and Japanese women having been more likely to attend college, and along with Chinese women, more likely to possess private insurance. African-American women had the youngest median age at the time of their first birth, while Chinese and Japanese subjects were the oldest. One hundred and thirty-three subjects were involuntarily childless, defined by a reported history of infertility and nulliparity. African-Americans had the lowest prevalence of involuntary childlessness (1.8%) , while Caucasian and Japanese women had the highest prevalence (5.6% and 7.7%, respectively) . Results are similar when history of infertility alone is analyzed, although rates of infertility are higher than rates of involuntary childlessness. Women with involuntary childlessness were also more likely to report a higher income, attend college, hold private insurance, and report a history of fallopian tube infection ( table 2a ). Risk factors associated with history of infertility are demonstrated in table 3 . Multiple logistic regression demonstrated that ethnicity remained significantly associated with involuntary childlessness after adjustment for adult body mass index, measures of socio-economic status, history of fallopian tube infection, and oligomenorrhea ( table 2b ). Notably, African-American and Chinese women were less likely to be involuntarily childless as compared to Caucasian women (p = < 0.001 and 0.028, respectively). Additionally, the adjusted odds ratios of African-American and Chinese women for the outcome of involuntary childlessness were not sensitive to the variables ‘history of fertility drug use,’ ‘ever use of birth control pills,’ or ‘use of birth control pills to prevent pregnancy,’ as demonstrated by sensitivity analyses excluding these populations separately and combined (data not shown). In an attempt to clarify differences in reported infertility among various ethnic groups in the SWAN baseline cohort, further analyses were performed on a subset of data obtained only from women who described a history of infertility and who provided a reported etiology for their infertility (n = 302). Ethnicity was significantly associated with reported male factor infertility and reported attributes not known to cause infertility. Caucasian and Japanese subjects had the highest rates of reported male factor infertility and the lowest rates of reporting an etiology not known to cause infertility as compared to Chinese, African-American and Hispanic women ( table 4 ). Additionally, among women with a history of infertility, no significant difference in ethnicity was found among subjects who did and did not report a known etiology for their infertility (p = 0.069, data not shown). After adjustment for socio-economic status via multivariable analysis ( table 4 ), African-American women were almost 3 times more likely to report an etiology not known to cause infertility as compared to their Caucasian counterparts, and less likely to report male factor infertility.

In concert with extant literature, involuntary childlessness and perceived etiology of infertility in the SWAN cohort were profoundly intertwined with ethnicity, as evidenced by multivariate models demonstrating these associations after controlling for economic and other known risk factors for infertility. We found that African-American and Chinese women were less likely to be involuntarily childless as compared to Caucasian subjects. These results are in contrast to other studies reporting that African-American and Asian women have higher rates of infertility than their white counterparts ( 1 , 11 - 15 ). One possible reason for our departure from results of previous studies may be the changing demographic of African-American women today. Seventy-five percent of African-American subjects in SWAN attended college, 78% held private insurance, and 41% earned between $35,000 and $74,999 annually. Many of the reports quoting high rates of childlessness and tubal infections in African- American women were published in the 1970s and 1980s, when social and demographic characteristics of the African-American population were different than they are today ( 5 , 6 ). In the setting of the National Survey of Family Growth, Stephen and Chandra observed a greater prevalence of infertility among Hispanic and non-Hispanic black women ( 1 ). Although their study is one of the largest to date, their results have been debated ( 2 ). Stephen and Chandra defined infertility as existing in married women of reproductive age who were sexually active, not using contraception, and did not conceive a child within the year prior to screening. Significantly, desire for pregnancy or history of infertility prior to the 12 months’ timeline was not considered. Through construction of this infertility variable, true infertility rate may have been skewed by including unmarried women and sterilized women in the denominator but not in the numerator ( 2 ). In addition, differences in marital status and contraception use among various ethnicities would affect the reported fertility rate. In contrast to these findings, our data suggest that African-American women are less likely to suffer from involuntary childlessness as compared to Caucasian women. This inconsistency may be due to a number of factors, including choice of dependent variable. We chose to use involuntary childlessness as our outcome of interest. Defined as reported history of a 12 month period of failure to conceive despite attempts to become pregnant PLUS lifetime nulliparity, involuntary childlessness has a strong objective component (nulliparity) while still capturing stated attempts to become pregnant. Furthermore, unresolved infertility is reported in the literature to be affected by infertility treatment and, therefore, by those factors which affect access to healthcare and ART outcomes ( 16 , 17 ). Of note, our conclusions are similar even when using reported history of infertility alone as the dependent variable. Indeed, African-American women were less likely to report a history of infertility as compared to other ethnicities (p = 0.023, data not shown). A potential explanation may involve a previously reported finding that African-American adolescents, on average, initiate sexual activity at an earlier age compared to Caucasian adolescents ( 18 ).) In our study, African-American subjects were younger at the age of their first birth as compared to other ethnicities and may have completed childbearing earlier, thus mitigating age-related subfertility. Indeed, aging is a well documented risk factor for diminished ovarian reserve ( 19 ), and may be related to other etiologies of infertility such as cumulative risk for pelvic infection ( 20 ) and increasing incidence of endometriosis ( 21 ). Our second notable finding was a high prevalence in attributes not known to cause infertility as the explanation for “not getting pregnant” among African-American women. We have included a supplementary table detailing a few examples of such attributes. In our cohort, although Caucasian ethnicity was not related to whether or not a subject reported an etiology for infertility, it was significantly associated with attributes not known to cause infertility. African-American women were significantly more likely to report an etiology not known to cause infertility, as compared with Caucasian women. This may illustrate differences in physician attitude toward infertility among various ethnic groups. Medical providers may not take infertility in African-American women as earnestly, as it is sometimes perceived by medical providers as an issue mainly affecting Caucasian women ( 22 ). As such, providers may not perform a thorough work-up, inadvertently encouraging erroneous perceptions of infertility in their African-American patients. Finally, decreased access to care among African-American women may cause them to seek treatment from physicians who lack subspecialty qualification and are therefore potentially less effective in communicating a correct diagnosis ( 23 ). The strengths of our study lie in the large numbers and ethnic diversity of the SWAN cohort. Limitations of the study include the cross-sectional design, as in such analyses one may not ensure that a given exposure precedes the outcome. However, our primary outcome (childlessness) has been chosen to reflect an objective event (absence of children). Factors meriting further consideration include potential impact of cultural and/or language barriers on self-reported involuntary childlessness and etiology of infertility.