Abstract
Nearly one-third to one-half of U.S. women meets the medical criteria for infertility at some point in their reproductive lives. Yet many do not view lack of conception as problematic. Why might some women self-identify as having a fertility problem but others do not? Using two waves of the National Survey of Fertility Barriers, we conducted binary and longitudinal multinomial logistic regression to answer this question. Results suggest that only a portion of women actually experience infertility as a “spoiled identity” or as disruptive to their lives. Rather, consistent with symbolic interactionist perspectives, there is evidence that infertility symptoms (i.e., not conceiving) depend upon interpretations and definitions of the situation. Multiple patterns of self-identification over time (identity non-adopters, maintainers, adopters, and relinquishers) suggest an indeterminate association between illness and impacts on the self, even for a condition that is highly medicalized in the United States.
There is a robust sociological tradition of research about the illness experience, including the relationship between illness and the self and the ways individuals come to define their problem as something that needs medical attention (Charmaz 1995; Karp 1994; Parsons 1970; Whitt and Meile 1985; Zola 1973). In line with a broader symbolic interactionist perspective (Blumer 1969), one major contribution from prior research is that the self-concept undergoes deconstruction and reconstruction as individuals negotiate and interpret what the illness means to them. David Karp (1994) and Kathy Charmaz (1995), in particular, elaborate this self-concept process through their respective studies on depression and chronic physical illness. They identify various stages of illness careers, and show “how persons arrive at illness definitions and then reconstruct their identities accordingly” (Karp 1994:7).
Much of the research on identity and illness examines conditions that are profoundly debilitating (e.g., severe depression, multiple sclerosis, diabetes, chronic fatigue syndrome, etc.). There is little analysis of how individuals interpret conditions that may be chronic, stressful, and potentially life-altering, but not necessarily life-threatening or incapacitating. These less life-threatening conditions, however, have become more salient with the expansion of medicalization. Scholars increasingly point to the driving role of patients-as-consumers (Conrad and Leiter 2004), to self-medicalization (Conrad 2005), and to a growing desire to treat even “mild symptoms and benign infirmities” (Barsky and Borus 1995:1931). At the same time, there are potentially a number of different, culturally available interpretations for illness symptoms (Whitt and Meile 1985)—especially those symptoms that are ambiguous. For infertility, solutions can include adoption or having fewer children than desires. It is thus crucial for scholars and health practitioners to understand more about the process of defining oneself as someone with a problem that might be helped by medical treatment.
In addition to infertility, several contemporary illnesses (e.g., attention deficit hyperactivity, baldness, and acne) are also not life-threatening, do not usually result in debilitating impairment, and are unlikely to receive medical attention unless people initiate medical contact. Infertility is associated with suffering for many individuals and couples (Becker 2000; Greil 1991b; Sandelowski 1993). Infertility is also highly medicalized in the United States. The American Society for Reproductive Medicine (ASRM; 2009) defines infertility as 12 months of unprotected heterosexual intercourse without conception. Many medical practitioners specialize in treating infertility. Nearly one-third to one-half of U.S. women meets the medical criteria for infertility at some point during their reproductive careers (Gurunath et al. 2011; Johnson et al. 2018). Despite the high prevalence rate, only some of the women who meet the medical criteria also subjectively perceive that they have a fertility problem (Abbey, Andrews, and Halman 1994; Greil, Leyser-Whalen, et al. 2011; Loftus 2009; White et al. 2006; Wilson 2014). The difference between meeting the criteria for and perceiving infertility raises the following question: Why might some women subjectively perceive a problem, while others do not, even when they share the same set of symptoms for a predominantly medicalized condition?
To answer this question, we first examined factors associated with whether or not women self-identify as having a fertility problem (among those who meet medical criteria for infertility). We next classified women into one of four “self-identification patterns” based on whether they self-identified as having a fertility problem at both waves, only Wave 1, only Wave 2, or at neither wave. Finally, we explored what social contexts or characteristics could explain membership in the four patterns of self-identification. The results indicate that four different self-identification patterns exist, including women who never see their lack of conception as a problem. These empirical findings are contrary to popular images (largely based upon women who seek medical help) that infertility generally leads women to desperation and high distress (Franklin 1990; Madeira 2012), or to automatically internalize infertility into their sense of self (as discussed in Johnson and Fledderjohann 2012). The patterns identified using longitudinal survey data are consistent with recent qualitative research indicating an indeterminate relationship between infertility and the self, contingent on varying social locations and how individuals define the situation (e.g. Leyser-Whalen et al. 2018).
Infertility and Identity
To guide the exploration of infertility and identity, we draw concepts from sociological theories and studies of illness, identity, and the self. Much of this work has examined chronic illness as a form of “biographical disruption” whereby “the structures of everyday life and the forms of knowledge which underpin them are disrupted” (Bury 1982:169). Disruptions of everyday life undermine the taken-for-granted unity of the body and self; they also produce a sense of losing control over one’s body and one’s ability to project a desired self-image both internally and to others (Charmaz 1995). Depending on the extent of impairment and of disruption, individuals may further experience a loss of self: “a crumbling away of their former self-images without simultaneous development of equally valued new ones” (Charmaz 1983:168). Bodily changes associated with chronic illness influence the range of feasible possible future selves (both desired and feared), which refers to individuals’ imagining who they might become in the future (Markus and Nurius 1986; Vignoles et al. 2008). Some illnesses threaten one’s self-concept more generally, as Charmaz (1983, 1995) describes, while others may threaten specific identities as “parts of a self” comprised of “internalized role expectations” (Stryker and Burke 2000:284, 286).
Infertility has been likened to chronic illness and disability. Arthur L. Greil (1991a:22) examined infertility as a “secret stigma” and an “invisible” disability. Because it is a medicalized condition with no outward symptoms and usually no meaningful physical impairment for the afflicted individual(s), the disruption of unity between body and self comes largely from perceiving a barrier to a desired outcome (having a child) and the process of self-labeling. Yet, infertility can involve “long-term limitations on an individual’s ability to perform social roles and normal daily activities” (Greil 1991a:17). For women and couples who want (more) children but cannot have them because of biomedical barriers, infertility may produce a “spoiled identity.” In his classic work, Erving Goffman (1963) argued that when an individual possesses a stigma or “an undesired differentness from what [was] anticipated” (p. 5), they may be considered socially “tainted” or “discredited”—considered less than others by social comparison. This creates a gap between one’s desired versus actual identity, thereby producing a “spoiled identity.” This spoiled identity comes from not only social recognition of the undesired difference, but also personal recognition and internalization that one’s identity is discredited.
In this vein, Arthur L. Greil (1991b) showed that the married women in his study viewed infertility as more than just a “mechanical failure” of the body (p. 53): Most could not separate their bodily failure to reproduce from their sense of self as a woman and a wife. Thus, infertility may threaten a specific desired or future identity of biological parenthood, or it may pose a larger threat to one’s identity as a woman or man. Jeni Loftus and Angie L. Andriot (2012:228) describe infertility as a “failed life course transition” that leads to a “period of destabilizing the self.” Ralph Matthews and Anne Martin Matthews (1986:641) suggest that for most people, parenthood is such a central identity that the “transition to non-parenthood” is both real and stressful for the infertile, despite the fact that no objective change in status occurs. Infertility also poses a strong threat to gender identity for both men and women, given the strong connections between reproductive and sexual capacity, on one hand, and gender, on the other (Barnes 2014; Becker 2000; Loftus and Andriot 2012). It is important to note here that the aptness of the chronic illness/disability frame for infertility relies on whether or not someone desires to become a biological parent. In the absence of such desire (or an acute desire for voluntary childlessness), lack of conception might be a relief or an expected outcome.
Despite the potential for infertility to be life-altering, prior research utilizing large, population-based samples shows that not all women who meet the medical criteria will self-identify as having a fertility problem and vice versa (Abbey et al. 1994; Greil, Leyser-Whalen, et al. 2011; Loftus 2009; White et al. 2006; Wilson 2014). Marguerite Sandelowski (1993:62) posited that men and women could vary with regard to their definitions of infertility and that different definitions led them to either “distance themselves from or lead them to accept” an infertile identity. The definition that truly affected the individual’s sense of self was a “phenomenological” definition that led to internalizing images of being “incapable, abnormal, defective”—that is, a “spoiled identity” (Goffman 1963; Greil 1991a, 1991b). In this view, internalization of an identity as having a fertility problem is neither automatic nor necessary. For some women, infertility becomes a “landmark event” (Sandelowski 1993:65) marking off “before” and “after” moments in their lives. For others, infertility might be a fleeting condition “cured” through pregnancy, adoption, choosing a child-free life, or some other means of resolution.
Prior research has supported the notion that adopting a sick role and internalizing an illness identity is not an automatic result of experiencing symptoms. Rather, individuals must interpret symptoms based on cues from their personal and social environment in order to self-identify a problem (Karp 1994; Whitt and Meile 1985; Zola 1973). Therefore, a key mediating factor between symptoms and self-identifying a problem as an illness is the definition of the situation—a core element of the symbolic interaction perspective (Blumer 1969). As Hugh P. Whitt and Richard L. Meile (1985:684) argued in their study of mental illness symptoms, we need to view individuals as “confronting problematic experiences which they seek to understand, explain, and deal with within a situational context.” More broadly, David L. Altheide (2000) argued for a renewed emphasis on contextualizing identities as emergent from situations and shaping future definitions of the self. Applying the symbolic interactionist insights about contexts to the specific case of infertility, we argue here that the individual (or couple) draws on personal and social cues to interpret their symptoms as (un)wanted and (un)desirable. The process of interpretation leads people toward or away from self-identifying a fertility problem. One major interpretive cue here, addressed more below, is the desire to have a child and become a parent. Absent this desire, infertility symptoms may be interpreted as non-problematic or even a relief at not becoming pregnant after a long duration of unprotected intercourse.
Guided by the overall theoretical framework of symbolic interactionism, we focus on five constructs that likely shape interpretation of the situation: (1) social location, (2) perceived vulnerability, (3) perceived life course disruption, (4) interactions with others with power to define the situation, and (5) resolution of the problem. It is also important to note here that because we were using survey data not specifically designed to measure perceptions and interpretations directly, we used available measures that serve as proxies for these. Below we describe indicators of each construct and how one might expect them to be associated with self-identification as having a fertility problem.
Social Location
Social location (e.g., social class, race/ethnicity) strongly affects the experience of infertility in the United States (Bell 2014; Greil, McQuillan, et al. 2011). Because most forms of fertility treatment in the United States are concentrated in a private medical market (Conrad and Leiter 2004), socioeconomic status (SES) is a key determinant of medical help-seeking (Stephen and Chandra 2000). For example, Ann V. Bell (2014) found that many women of lower SES must learn to cope with (potentially unresolved) infertility whereas many women of higher SES must learn to cope with (potentially never-ending) treatment of infertility. Because many women of lower SES lack access to fertility treatments, they “try to dissociate themselves and their lives from their infertility” (Bell 2014:114). Furthermore, because low SES women have repeated experiences of adversity and little control over their lives, infertility can be similar to other hardships as opposed to uniquely life-altering. In contrast, higher SES women experience more control over reproduction and their lives more generally (Edin and Kefalas 2005; Roberts 1997; Stern 2005), and thus may find infertility a significant aberration that consumes their time and their minds through the “treadmill of treatment” (Bell 2014:115).
Media depictions of infertility cast the imagined patient as white, middle class, and married (Bell 2014; Greil, McQuillan, et al. 2011). Black and Hispanic women are usually stereotyped as “hyperfertile” (Ikemoto 1996), despite evidence that they have higher overall rates of impaired fecundity than white women (Chandra and Stephen 2010; Greil, McQuillan, et al. 2011). Controlling images of the “strong black woman” may also keep women from self-identifying as infertile because weakness, fragility, and illness are associated with “traditional white norms of femininity” (Beauboeuf-LaFontant 2007:31). Therefore, we expect higher SES women and white women to be more likely to ever self-identify as having a fertility problem and to maintain this identity over time.
Perceived Life Course Disruption
For some women, lack of conception after repeated unprotected heterosexual intercourse may not pose a problem. For others, it may be highly disruptive of their life course goals (Exley and Letherby 2001). Marriage is often a normative trigger for childbearing (White et al. 2006); therefore, married women may experience infertility as more disruptive than unmarried women; however, variations in the meanings of childbearing and marriage by race and social class could shape the association of marital status with self-identifying a problem (Cherlin 2010). Women who are trying to conceive when they experience an infertility episode are more likely to be distressed, and more likely to self-identify as having a fertility problem than women who were not actively trying (Greil et al. 2010; Johnson and Fledderjohann 2012). Additionally, not all women desire children. They may not be at a point in their life course where having a child is salient, or they may not want children at all. We expect that ever-married women, women who were trying to get pregnant when they experienced infertility, and those who desire children will be more likely to self-identify as having a fertility problem than women in comparison groups.
Perceived Vulnerability
Women’s perceived vulnerability to fertility problems can also affect their self-identification as having a problem. For many women, being over 30 years of age sounds “an alarm” (Sandelowski 1993:59) about achieving pregnancy within a certain culturally expected timeframe for childbearing. Age offers both a cultural and a biological marker for women, as fecundity begins to gradually decline around age 31 for most women (Broekmans et al. 2007). Having had a prior successful pregnancy and delivery (i.e., secondary infertility vs. primary infertility) could also shape women’s perceptions of their fertility even if they experience difficulty later on (Greil, Johnson, et al. 2011). Self-assessed health—a subjective, yet highly reliable, cognitive rating of one’s overall health (Jylha 2009)—may factor into women’s perception of their fertility: Women with poorer perceived health might assume they will have decreased ability to conceive or may have had a conversation with their physician about improving their preconception health (Waggoner 2017). We expect that women with primary infertility (difficulty conceiving a first pregnancy), women with poorer self-assessed health, and older women will be more likely to ever self-identify as having a fertility problem than women in the reference categories.
Interactions with Others
A crucial premise of symbolic interaction is that people’s definitions of situations are contingent on others’ perceptions of them and their situation (Blumer 1969). Two types of others that are especially relevant are significant others and authoritative others. Significant others have their own expectations about the importance of having/not having children that influence women’s in/fertility expectations (Greil et al. 2013; Thomson 1997). We expect that women whose spouses/partners want (more) children will be more likely to self-identify as having a fertility problem than those whose partners do not want (more) children. Authoritative others, such as medical practitioners, have professional power to diagnose infertility, determine if treatment is appropriate, and to provide assistance to those seeking help (Pescosolido 1992). A medical diagnosis of infertility can also provide a “clear message” that “things will never be the same again” (Jutel 2011:26). Some women push for diagnosis, having already perceived a problem (Becker and Nachtigall 1992; Leyser-Whalen et al. 2018). Others perceive a problem only after receiving a diagnosis (Leyser-Whalen et al. 2018). Although we cannot untangle the direction of this relationship, we expect that women who have sought medical help for pregnancy will be more likely to self-identify than those who have not.
Resolving the Problem
Resolving infertility can mean many things: having a biological child, adopting or fostering, stepparenting, or choosing a child-free lifestyle. Because of the cultural emphasis on biological kinship (Becker 2000; Letherby 1999) and the conflation of infertility with permanent childlessness (Greil 1991b), we expect that women who have had a biological child will be less likely to self-identify as having a fertility problem than those who have not had a child. It is unclear how social parenthood might relate to women’s self-identification. Women who are mothers but have not given birth have resolved the problem of involuntary childlessness but might still view their bodies as functionally problematic and thus self-identify as having a fertility problem.
Method
Data and Sample
We used the National Survey of Fertility Barriers (NSFB), a random-digit-dialing telephone survey with a probability-based sample of 4,787 U.S. women aged 25 to 45 years. Wave 1 was conducted on a rolling basis from 2004 to 2007. Wave 2 follow-up interviews were conducted with all women who could be reached three years after their initial interview (2008–2010) for a total of 2,136 participants (58 percent of those sought for re-interview). Methodological information and data can be accessed at: https://www.icpsr.umich.edu/icpsrweb/DSDR/studies/36902#bibcite. Because this study relied on de-identified, publicly available, secondary data, it was deemed exempt from review requirements for research on human subjects by the Institutional Review Board (IRB #14-561459UE).
Our analyses relied on two subsamples. First, we selected all heterosexual women in Wave 1 for whom we had complete data on all variables of interest and who ever met the medical criteria for infertility (one year of regular, heterosexual sex without conception). Women were considered infertile at Wave 1 if they responded “yes” to either question: “Was there ever a time when you were trying to get pregnant but did not conceive within 12 months?” or “Was there ever a time when you regularly had sex without birth control for a year or more without getting pregnant?” Cases with missing data were excluded through listwise deletion. The final analytic sample (n = 2,341) contained 95.8 percent of all women who met medical criteria for infertility in Wave 1. For the second part of the analysis, we selected all heterosexual women who met medical criteria for infertility at Wave 1 and for whom we had complete data across both waves (n = 727). Of the original Wave 1 sample (n = 4,796), 3,724 were selected to be re-interviewed again three years later. These were women who had future fertility expectations, had experienced fertility barriers, or were childless at Wave 1. Due to sample attrition, 57.4 percent of the targeted respondents were interviewed at Wave 2. Given the selection criteria for re-interview, the respondents who participated in both waves as a whole do differ on certain indicators compared to women who just completed Wave 1 (e.g., Census region, marital status, race/ethnicity, number of children, age, education, having no health insurance in the last three years). It is important to address the differential attrition for the current study. All women who met the medical criteria for infertility were sought for re-interview. Within this group, we found that those who responded to Wave 2 were significantly more likely to self-identify as infertile. The overrepresentation of women who self-identify does not necessarily cause problems for interpreting the results below because we are interested in all of the identity trajectories, not just women who self-identify. It may, however, lead to an underestimation of the percentages of women in the “relinquisher” and “non-adopter” groups. Therefore, we are cautious regarding generalizing the representation in the four groups.
Measures
Outcomes
Preliminary testing of survey items for the NSFB indicated that many women are disturbed by the designation “infertile.” For this reason, the NSFB research team followed the lead of Antonia Abbey et al. (1994) and asked respondents, not whether they identified as being “infertile” but whether they identified as “having a fertility problem.” The outcomes of interest were (1) whether a woman self-identified as having had a fertility problem at Wave 1 and (2) self-identification subgroups across waves. We considered someone to have self-identified as having a fertility problem if she answered “yes” or “maybe” to either of the following: “Do you think of yourself as someone who has, has had, or might have trouble getting pregnant?” or “Do you think of yourself as someone who has or has had fertility problems?” We decided to include both “yes” and “maybe” responses based on theoretical and empirical considerations. First, women may be inclined to report “maybe” to the survey question given the stigma of infertility. Or they may have a hunch about difficulty getting pregnant, but this has not yet been solidified through medical help-seeking. Thus, we believe the “maybe” responses are conceptually closer to “yes” than “no.” In a sensitivity analysis, we examined regression models using both the more liberal (yes and maybe) and more restricted definitions (yes only) and found that results were substantively similar. Less than 4 percent of women reported “maybe” for the question about trouble getting pregnant, and less than 2 percent reported maybe to the question about having a fertility problem. Given this, we opted to include both “yes” and “maybe” responses.
We developed subcategory names during the analysis, based on a substantive descriptive of the self-identification pattern. We labeled women who answered “no” to both waves as identity non-adopters because they never acquired a self-identification as infertile, despite meeting the medical criteria. Women who answered either “yes” or “maybe” at both waves were characterized as identity maintainers: They consistently maintained the same self-identification across waves. Women who answered either “yes” or “maybe” only at Wave 2 were characterized as identity adopters because they adopted or acquired a fertility problem identity over time. Those who self-identified only at Wave 1 but not Wave 2 were characterized as identity relinquishers because although they initially self-identified, they dropped this identity over time. These subgroups are represented as follows (Table 1).
Self-identification Subgroups.
Social location
Two measures addressed SES: a scale measuring income to need and an indicator of whether women currently had private health insurance (=1) or not (=0). The income to need scale was derived by dividing family income by the poverty line for a given individual’s family at the time of the interview. Due to sensitivity to income questions, family income was first constructed as an ordinal scale ranging from 1 (less than $5,000) to 12 ($100,000+). We then substituted the midpoint of each category for the category value in order to convert this into a continuous scale. A high score on the income to need scale indicated a greater ratio of income to need and thus higher SES. Participants could select multiple categories for race/ethnicity. Those who reported multiple races were classified giving first priority to a non-white identification. Because of small cell counts, especially for subsample 2, we collapsed the categories into white (=1) and non-white (=0).
Perceived life course disruption
Relationship status was measured by a binary variable indicating whether respondents had ever been married (never married = 1; ever married = 0) at Wave 1. Finer distinctions (e.g., married vs. cohabiting, currently married vs. divorced, divorced vs. separated, etc.) were not possible due to sample size limitations. We considered adding a second variable indicating whether or not respondents entered a new first union between waves. This was not feasible because there were only eight women in the category Trying to get pregnant at the time of an initial episode of infertility (=1) who were compared with “not trying” or “ok either way” (=0). All women had experienced an episode of infertility by the first wave; we also measured whether they had an episode between waves (=1) or not (=0). Current desire for a baby was measured at each wave. Responses were collapsed into agree/strongly agree indicating wanting a(nother) baby (=1) and disagree/strongly disagree indicating not wanting a(nother) baby (=0).
Perceived vulnerability to infertility
Age was initially measured in years but recoded to over thirty (=1) versus thirty and below (=0) at Wave 1. In preliminary models, we also included a measure of whether women were over thirty by Wave 2, to assess change into this category. We did not find any significant effects, so this variable was omitted in the final analysis to streamline findings. Women who had experienced primary infertility (=1) were compared with those who had experienced secondary infertility (=0). Self-assessed health was measured from high (4 = excellent health) to low (1 = poor health).
Social cues and interactions
Perceptions of partner desire to have a baby were measured using a Likert-type scale with responses ranging from “definitely yes” to “definitely no.” Responses of “probably yes” or “definitely yes” were recoded as “1” (partner would like a baby), and responses of “probably no” or “definitely no” were recoded as “0” (partner would not like a baby). Medical contact with a physician was assessed at both waves. Women who said they spoke to a doctor specifically about fertility-related issues were coded as making medical contact (=1); all others were coded as not (=0).
Resolution of the fertility problem
Biological mothers (=1) were compared to women who were either not mothers or were non-biological (i.e., social) mothers (=0) at Wave 1. Because we included all women who had ever met the medical definition of infertility, women who were biological mothers at Wave 1 could have potentially experienced primary or secondary infertility, or both, in their reproductive career, but still have had a biological child by the time they were interviewed for the NSFB. Thus, biological motherhood status is distinct from type of infertility episode. We also indicated whether a woman had a live birth between waves (=1) or not (=0). We also included indicators for social motherhood (e.g., step, adoption, or fostering) status (yes = 1; no = 0) at Wave 1, and whether a woman became a social mother (=1) or not (=0) between waves.
Analytic Strategy
We first estimated descriptive statistics for all variables of interest for Wave 1 and combined Wave 1 and 2 subsamples. We then proceeded to conduct logistic regression analysis to examine the binary dependent variable of whether women ever self-identified as having a fertility problem (=1) or not (=0). Finally, we used multinomial logistic regression to compare what factors are associated with membership in the non-adopter, adopter, and relinquisher categories compared to women who maintained a self-identification as having a fertility problem for both waves.
Results and Discussion
Descriptive Statistics
Among medically infertile women in Wave 1, less than half (44.1 percent) self-identified as having a fertility problem (Table 2). For women who completed both survey waves, more than one third (37.2 percent) were “non-adopters” who did not self-identify at either wave, and almost half (45.3 percent) were “maintainers” who continued to identify as having a fertility problem across both waves. Substantially, fewer women (7.9 percent) were “adopters” who gained a self-identification as having a fertility problem between waves, or they (9.5 percent) were “relinquishers” who dropped their self-identification between waves. Notably, non-adopters and maintainers accounted for 82.5 percent of the Wave 1 and 2 subsample, indicating consistent internal identity over time for the majority of women.
Descriptive Statistics of Heterosexual Women with Infertility in the NSFB.
Note. Weighted by NSFB final weights. NSFB = National Survey of Fertility Barriers.
Binary Logistic Regression Analysis
Table 3 provides binary logistic regression results for the associations of measures for the five constructs (social location, perceived life course disruption, perceived vulnerability, social cues and interactions, and resolution of the problem) with self-identification as having a fertility problem (Wave 1 sample only). At least one measure of each construct was associated with self-identifying as having a fertility problem. Table 3 includes the regression results, hypothesized direction, and interpretation of the association. In terms of social location, women with higher scores on income to need (i.e., higher SES) had higher odds of self-identifying as having a fertility problem (odds ratio [OR] = 1.09).
Logistic Regression Results Predicting Self-identification as Having a Fertility Problem, Wave 1 (n = 2,341).
Note. Weighted by NSFB final weights. OR = odds ratio; NSFB = National Survey of Fertility Barriers.
p < .05. **p < .01. ***p < .001.
Regarding perceived life course disruption, women who were trying to become pregnant when they experienced an infertility episode had odds of self-identifying more than three times (OR = 3.13) those of women who were not trying or were “ok either way.” Additionally, women who currently wanted a(nother) baby had higher odds (OR = 1.82) of self-identifying than women who did not want a(nother) baby. Marital status (ever vs. never married) was not associated with self-identifying.
One measure of perceived vulnerability to infertility was associated with self-identifying: Women with higher self-assessed health had lower odds (OR = 0.69) of self-identifying compared to women who self-rated their health as poorer. In contrast to prior research, age and primary infertility were not associated with self-identification. In supplemental analyses not shown here, we also found no association between advanced maternal age (35 years and above) and self-identification.
One measure of social cues and social interactions was associated with self-identification: For women who talked to a doctor about their fertility, the odds of self-identifying was almost eight times higher (OR = 7.61) than for women who had not talked to a doctor about pregnancy. The size of this effect was substantially higher than any other variable in the analysis. With cross-sectional data, it is impossible to determine whether women who self-identify are simply more likely to talk to a doctor or whether talking to a doctor contributes to self-identifying. One recent study found that both directions are possible, but that more women first self-identify then see a doctor than the reverse (Leyser-Whalen et al. 2018). Finally, turning to measures of resolution of the fertility problem, women who had given birth at or before Wave 1 had substantially lower odds (OR = 0.33) of self-identifying compared to women who were not biological parents. In the next section, we examine Wave 1 and 2 stability and change in self-identification as having a fertility problem.
Multinomial Logistic Regression Analysis
Table 4 displays results from the multinomial regression model of the two-wave self-identification patterns: non-adopters, adopters, and relinquishers compared to maintainers (the base category). The dependent variable measures within-person stability (i.e., identity maintainer or non-adopter) or change (i.e., identity adopter or relinquisher) and therefore provides a more stringent assessment of factors associated with claiming a fertility problem identity than the cross-sectional analyses in Table 3.
Multinomial Logistic Regression Results Comparing Self-identification Subgroups Across Waves (n = 727).
Note. Base category is “Maintainers.” Variables measured at W1 unless otherwise indicated.
p < .05. **p < .01. ***p < .001.
Race/ethnicity was the only social location measure associated with the self-identification subgroups. White non-Hispanic women had lower odds of relinquishing or not-adopting compared to maintaining a self-identity across both survey waves. Two measures of perceived life course disruption were associated with self-identification subgroups. Trying to get pregnant (Wave 1) at the time of experiencing an infertility episode was a strong consistent predictor across all subgroups: Women who were “trying” had a lower odds of being in all other subgroups (adopters, relinquishers, or non-adopters; OR = 0.12, 0.21, and 0.10) respectively, compared to maintaining a self-identity as having a fertility problem at both waves. Women who wanted a baby (at Wave 2) also had reduced odds of being in the identity non-adopter subgroup compared to the identity maintainer group (OR = 0.36).
Regarding measures of perceived vulnerability, women reporting higher self-assessed health had a higher odds of not-adopting compared to maintaining a self-identity (OR = 1.53). In terms of social cues and interactions, women who talked to a doctor at Wave 1 had reduced odds of adopting, relinquishing, and not-adopting compared to maintaining a self-identity (OR = 0.21, 0.35, and 0.07, respectively). However, women who talked to a doctor at Wave 2 had a higher odds of adopting a fertility problem identity compared to maintaining a self-identity (OR = 3.22). We discuss this latter finding below in relation to recent research on self versus health care provider labeling women as having a fertility problem.
There are several ways to resolve fertility problems. Table 4 includes four measures of parenthood (biological parent at Wave 1, giving birth between waves, being a social parent at Wave 1, and becoming a social parent between waves) compared to not being a parent at either wave, adjusted for wanting children. Therefore, the women in the comparison group are not mothers, did not become mothers, and do not want a child. Women who were social parents at Wave 1 had lower odds of relinquishing compared to maintaining a fertility problem identity (OR = 0.11). Women who became social parents between survey waves had higher odds of relinquishing a fertility problem identity compared to maintaining one (OR = 8.03). This suggests that social parenthood possibly helped some women to let go of their fertility problem identities, but not other women.
Conclusion
There is a significant body of work on the illness-self relationship, but much of this has examined conditions that are profoundly debilitating, thereby affecting individuals’ capacity to function on a day-to-day basis and requiring a reevaluation of one’s self-concept (Charmaz 1995; Karp 1994). At the same time, insights from symbolic interaction point to the centrality of defining a situation and interacting with others as part of interpretative processes of making sense of a particular experience or phenomenon (Blumer 1969). Drawing on the symbolic interactionist perspectives of illness and the self, and in light of empirical findings that there is a disjuncture between medical and subjective definitions of infertility (Greil, Leyser-Whalen, et al. 2011; Wilson 2014), we questioned the dominant portrayal of infertility as necessarily producing a “spoiled identity” for U.S. women. Rather, we acknowledge the myriad influences in individual lives as we asked: What factors might be associated with self-identifying as having a fertility problem among a sample of women who meet medical criteria for infertility?
The NSFB data clearly show that only some women who meet the medical criteria for infertility actually think of themselves as someone with a fertility problem (44.1 percent self-identified in Wave 1). Therefore, for many women there is a strong disjuncture between meeting medical criteria for infertility and actually internalizing an infertile identity. When we looked at self-identification trajectories over time, we found that less than half of women maintained a self-identification across both survey waves (45.3 percent). A substantial portion of women also never self-identified (37.2 percent). A smaller number of women either adopted or relinquished identities between waves (7.9 and 9.5 percent, respectively). Therefore, our results indicate that many women do not fully internalize infertility as a consistent part of their identity, that there is notable variation in women’s subjective perceptions of infertility, and that these perceptions may shift over time depending on life circumstances.
Maintainers (self-identified at both waves) appeared to fit popular images of infertile women (Bell 2014; Greil, McQuillan, et al. 2011; Madeira 2012). They were more likely to be non-Hispanic white and to see a physician early on; they appeared to perceive infertility as disruptive to their lives because of a greater intent to get pregnant and greater desire for a child. Maintainers thus appear to reflect the “I am infertile” group that Sandelowski (1993) described in her earlier qualitative work; these women were likely to have strongly incorporated infertility into their sense of self. Although maintainers fit the dominant cultural image of the infertile woman, it is important to stress that they represented fewer than half of women in the NSFB sample (45.3 percent). The second-largest group in the sample was identity non-adopter (37.2 percent). This group of women appeared to have had fewer cues encouraging them to define their lack of conception as problematic. Overall, non-adopters had lower fertility desires in terms of both intent to get pregnant when they experienced infertility and desire for a baby, especially by Wave 2. They were also much less likely to interact with medical professionals. Identity adopters and relinquishers were substantially smaller subgroups in the NSFB sample (7.9 and 9.5 percent, respectively). Identity adopters had a similar social location as maintainers; however, their infertility episode at Wave 1 was apparently not interpreted as a self-defining experience, as they were less likely than maintainers to say they were trying to get pregnant when they experienced infertility. Yet, it may have become one over time, as suggested by the fact that they initially were much less likely to talk to a doctor about the problem at Wave 1, but were more likely to do this by Wave 2 compared with maintainers. Identity relinquishers differed from maintainers in terms of race/ethnicity, talking to a doctor, and resolution through social parenthood. Given that relinquishers were more likely to be non-white and social parents, it is important to consider the cultural tradition of “other mothering” that has “been central to the institution of Black motherhood” (Collins 1990:119). Overall, relinquishers presumably “let go” of their identity over time because (1) they did not have strong or active desires for pregnancy or (2) they resolved their infertility through social parenting.
We also found moderate support for the relevance of each of the five constructs we examined, but none stood out as the main driver of fertility problem self-identification. Rather, there appears to be a more complex set of variables that shape the development of infertile women’s self-identities. Moreover, we see incongruence between factors that fertility specialists versus the women in our sample focus on, such as age. Whereas the medical profession puts strong emphasis on the relationship between aging and fertility (Broekmans et al. 2007), there was no association between age and odds of self-identification. Among women who met criteria for infertility, the few factors that were associated with stability and change in self-identity (i.e., race/ethnicity, trying to get pregnant, wanting a baby, self-assessed health, talking to a doctor, and being/becoming a social parent) overlap somewhat, but only partially, with factors associated with fertility-specific distress and infertility help-seeking. Indeed, similar to recent work by Ophra Leyser-Whalen et al. (2018), it is unclear if talking to a doctor reflects or creates self-identification of a fertility problem. Combined with research on variations in psychosocial responses to infertility (for a summary, see Greil, Slauson-Blevins, and McQuillan 2010), the current study highlights how self-identifying as having a fertility problem reflects more than having an illness. It is important to determine how women define their situation rather than assuming that lack of conception after a year of unprotected intercourse is a problem. Thus, these different perspectives may need to be taken more into consideration in a clinical setting to aid in patient–provider communication.
Most notably, our results indicate that there is not a universal pattern for the relationship between meeting medical criteria for infertility and self-identification as someone having a fertility problem. For some women (i.e., maintainers), infertility may result in a “spoiled identity” (Goffman 1963; Greil 1991a, 1991b). For others, self-identifying as having a fertility problem is more transitory, or even non-existent, as evidenced by the relinquishers and non-adopters in our analysis. The patterns of coefficients for the “trying to conceive” and “wanting a baby” indicators also suggest that some women who meet medical criteria may be more properly considered “child free” because they were not trying to become pregnant when they experienced infertility and had less desire for a child at that moment. Thus, the findings lend further support to the view that there is an indeterminate relationship between illness and the self, contingent on interpreting symptoms through individual definitions of the situation.
Our findings also have broader implications beyond the specific case of infertility. Prior research has emphasized the uncertainty of the illness career, the interpretive process that eventually transforms people into patients who seek medical help, and the adaptation to illness over time (Charmaz 1983, 1995; Karp 1994; Whitt and Meile 1985; Zola 1973). Although each of these approaches emphasizes uncertainty and indeterminacy, they typically show how individuals do eventually come to recognize symptoms as part of a larger medical problem. In contrast to this, we emphasize uncertainty about whether individuals ever recognize symptoms as problematic, much less as indicative of an illness that potentially re-defines the self. This is especially salient when the illness itself is highly open to interpretation. Therefore, even highly medicalized conditions are subjectively perceived and interpreted based on personal, familial, and contextual cues. Insights here can extend to other cases where a condition has become predominantly medicalized, but where there are multiple possible interpretations of symptoms. Other such examples include conditions such as female sexual dysfunction, anxiety, or food intolerances. For instance, Mayo Clinic (2019) describes the condition of female sexual dysfunction, its symptoms, and possible treatments, but also notes that it is only considered a problem if it bothers the individual in question or is associated with stress in their relationship with their partner, thereby recognizing the highly subjective nature of the disorder.
Limitations
There are some limitations in the current study. Even with two waves of data, there are challenges to establishing temporal order. For example, we conceptualized talking to a doctor as a cue to self-identifying as having a problem, but women could self-identify and then make medical contact. More waves of data collected at shorter intervals could further illuminate temporal order. In addition, while our theoretical focus has been on perceptions and interpretations, we were unable to measure these directly. For example, trying to become pregnant is not a direct measure of perceived life course disruption, but a proxy that women trying to get pregnant would experience lack of conception as disruptive to their life course plans. This, however, is one trade-off in using survey data to explore broader patterns among a large sample of women. Finally, couple influences are important in shaping responses to infertility (Johnson and Johnson 2009); thus another shortcoming of this study is that we only indirectly included the influence of male partners through women’s perception of their partner’s fertility desires. A more direct analysis of partner influence is warranted, especially given the centrality of social interaction in the symbolic interaction paradigm (Blumer 1969) and the need for more social science research on men and infertility (as reviewed in Slauson-Blevins and Johnson 2016).
Conceptualizing infertility as an illness that may or may not be perceived as a problem by those who meet medical criteria gives rise to new questions and directions for future research on the relationship between illness and the self, especially in terms of exploring variable trajectories of illness as it relates to self-concept. Future research should seek to better understand how these women’s self-identification processes unfold through more in-depth qualitative analysis. Additionally, we argue that cross-sectional analyses of psychosocial responses to illness more generally, and infertility more specifically, cannot adequately capture certain patterns because women’s experiences are not static. Thus, longitudinal work is needed on dynamic responses over time. This is also more consistent with an interactional approach to thinking about the self-illness relationship.
Footnotes
Authors’ Note
This is a substantially revised version of a paper presented at the 2014 annual meeting of the Eastern Sociological Society, Baltimore, MD.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for research, authorship, and/or publication of this article: This research was supported by a grant (R01-HD044144; David Johnson [PI]) titled “Infertility: Pathways and Psychosocial Outcomes,” funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD).