Is infertility resolution associated with a change in women’s well-being?

Abstract

STUDY QUESTION

Is giving birth associated with improved subjective well-being among involuntarily childless women?

SUMMARY ANSWER

Resolution of infertility is associated with increased life satisfaction and self-esteem, but not with a decrease in depressive symptoms.

WHAT IS KNOWN ALREADY

Cross-sectional data and studies of treatment-seekers show that infertility is associated with lower subjective well-being. Childless women with infertility tend to report lower subjective well-being than women who experience secondary infertility, but a prospective study using a random sample of involuntarily childless women over time has not previously been conducted.

STUDY DESIGN, SIZE, DURATION

The sample for the current study includes all women without children who met medical criteria for infertility or perceived a fertility problem (N = 283) at baseline and who were interviewed in both waves (3 years apart) of the National Survey of Fertility Barriers (NSFB), in a random-digit dialing telephone survey. It is therefore possible to explore here whether there are differences in the association of infertility resolution and subjective well-being among women who do and do not perceive themselves as having a fertility problem.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Depressive symptoms (as measured by the Center for Epidemiologic Studies—Depression Scale), self-esteem (as measured by a modified version of the Rosenberg Self-esteem Scale) and life satisfaction (as measured by a modified version of the Satisfaction with Life Scale) were assessed for all 283 participants at both waves. For all three variables, change scores of 47 involuntarily childless women who resolved their infertility through a live birth were compared to the scores for the 236 women who remained childless. A number of variables shown to be associated with subjective well-being among infertile women were included as controls.

MAIN RESULTS AND THE ROLE OF CHANCE

No relationship between infertility resolution and change in depressive symptoms was observed (b = −0.04; P > 0.05). Involuntarily childless women who resolved their infertility improved in self-esteem (b = 0.74; P < 0.01) and life satisfaction (b = 1.06; P < 0.01).

LIMITATIONS, REASONS FOR CAUTION

Women were measured at only two time points. Only 47 women had a live birth between waves. While it is common practice to make causal interpretations based on panel data, such interpretations should be made with caution. In addition, the NSFB was conducted in the USA where medical expenditures are high and most fertility treatment expenses are not covered by insurance. Thus it may not be possible to generalize the findings to other modern industrialized societies.

WIDER IMPLICATIONS OF THE FINDINGS

Knowing that resolution of infertility is associated with improved subjective well-being is important for infertile couples and infertility professionals alike.

STUDY FUNDING/COMPETING INTEREST(S)

This research was supported in part by NICHD grant R01-HD044144 and NIGMS grant P20-GM109097 from the National Institutes of Health. The authors have no competing interests.

Introduction

Infertility is defined medically as 12 months or more of unprotected, heterosexual intercourse without conception (American Society for Reproductive Medicine, 2008). Infertility is a common experience; the proportion of women who experience medically defined infertility is 7–15.5% in a given year depending upon the specific measurement procedures (Thoma et al., 2013). Lifetime prevalence is estimated to be between 6.6 and 24.6% (Boivin et al., 2007), but these estimates do not take into account the ‘hidden infertile’ (Greil et al., 2010b), referring to women who meet the medical criteria for infertility but who do not describe themselves as having tried to conceive. When these women are taken into account, the percentage of women who have ever met the medical criteria at some point throughout their reproductive lifespan is substantially higher (51.8%) (Greil et al., 2011). Although the lay public understands the term ‘infertility’ to refer to a permanent inability to give birth (Miall, 1994), infertility as medically defined does not imply permanent involuntary childlessness. Indeed, the majority of women who experience infertility do become mothers at some point; thus, their infertility is ‘resolved’ before their reproductive years have passed (Abma and Martinez, 2006).

Motherhood is an important status and valued identity among women in the USA (McQuillan et al., 2008; Katz-Wise et al., 2010). Few Americans want to or expect to be childless (Abma and Martinez, 2006), and most women still desire at least two children (Hayford, 2009; Morgan and Rackin, 2010). Motherhood is seen as a sign of adulthood among many American women (Parry and Shinew, 2004), and most women rate motherhood as their most important life role (Reitzes and Mutran, 2002). If motherhood is a desired and valued social status, then infertility can be understood as a barrier to achieving an important life goal for women (Greil et al., 2014). Thus, we would expect that infertility should be associated with lower subjective well-being (SWB) among women and that the successful resolution of infertility through giving birth should be associated with higher SWB. Although it appears reasonable to expect the resolution of infertility to improve the SWB of infertile women, there is also evidence to suggest that the resolution of infertility (i.e. childbirth) is itself associated with increased depressive symptoms and reduced SWB (Evenson and Simon, 2005; Hansen, 2012; Sejbaek et al., 2015; Baetschmann et al., 2016).

Women who experience infertility often report that it becomes the central theme in their lives: they are unable to move forward with life transitions and they continuously grieve the mother-child relationship that never came to pass (Johansson and Berg, 2005). Even though there is no specific loss for those who remain involuntarily childless, they often report deep feelings of loss and mourn for children that they never had (Daniluk, 2001). A number of recent studies have addressed the impact of infertility on quality of life (QoL) (Monga et al., 2004; Chachamovich et al., 2007, 2010; Wischmann et al., 2012; Heredia et al., 2013; Mousavi et al., 2013). SWB can perhaps be regarded as the subjective dimension (as opposed to the ‘objective’ dimension, which refers to social, economic and health indicators) of QoL and refers to a self-reported personal sense of contentment, fulfillment or happiness (Diener and Lucas, 1999; Ferriss, 2001; Costanza et al., 2007; Camfield and Skevington, 2008). SWB is often described as having an affective dimension that taps into mood states and a cognitive dimension that reflects self-appraisal of life circumstances (Diener, 1984). In this article, we use a measure of depressive symptoms as an indicator of negative mood states and a measure of life satisfaction to assess cognitive well-being. We also include a measure of self-esteem; self-esteem reflects overall positive evaluation of the self and is a key component of a healthy self-concept (Rosenberg, 1979, 1990; Cast and Burk, 2002).

We are aware of approximately a dozen studies that attempt to examine the effects of infertility resolution on the SWB of women following infertility. In general, these studies conclude that infertility resolution is associated with increased SWB, but they have some methodological limitations. Most of these studies have been cross-sectional (McQuillan et al., 2003, 2007; Jacob et al. 2007; Kjaer et al., 2011; McCarthy and Chiu, 2011; Ben-Shlomo et al., 2016). The majority either fail to distinguish between involuntarily childless women who became mothers and women with secondary infertility or limit their sample to involuntarily childless women only (See Kjaer et al., 2011 for an exception). Thus, it is difficult to draw conclusions from these studies about the consequences of infertility resolution. Furthermore, any cross-sectional analysis is vulnerable to the possibility that some unmeasured variable is confounding the results. Some studies have been retrospective, asking women to reflect on their experience of having been involuntarily childless and of becoming mothers (Ulrich and Weatherall, 2000; Olshansky, 2003). These studies depend on recall of past events and lack a comparison group.

Other studies have been longitudinal, following up with infertile patients a number of years after the baseline study to see whether treatment success results in changes in SWB (Verhaak et al., 2005, 2007b; Habroe et al., 2007; Wirtberg et al., 2007; Schanz et al., 2011; Wischmann et al., 2012). These studies have also found that infertility resolution is associated with increased SWB (See review by Gameiro and Finnigan, 2017). These studies examine change over time prospectively but necessarily limit their focus to treatment seekers only. In fact, samples are typically limited to the minority of infertile women who have undergone some form of ART. Thus, it is not possible to generalize to the large number of infertile women who do not receive treatment; neither is it possible to separate out the effects of infertility from the effects of infertility treatment. Some researchers have found that SWB among women who sought fertility treatment is lower than that of their counterparts who did not see a doctor (Abbey et al., 1992; Greil et al., 2011), suggesting that infertility treatment may influence SWB in addition to infertility itself. McCarthy and Chiu (2011), however, found that infertile women who received infertility treatment had higher scores on life satisfaction than women who were not treated. Because ART is a time-consuming, invasive and stressful process (Verhaak et al., 2007a), studies showing that failed ART is associated with decreased SWB compared to successful ART cannot determine the extent to which decrease in SWB is attributable to failure to conceive and how much is attributable to failure to conceive after ART.

In this study, we use prospective two-wave panel data from a nationally representative sample to explore the impact of infertility resolution through a live birth on within-person changes in SWB while controlling for personal characteristics previously shown to be associated with SWB. The use of a population sample means that we can generalize beyond clinic patients to infertile women in general and that we can statistically separate the effects of infertility and its resolution from the effects of infertility treatment. The use of panel data means that we can look at within-person changes, thus controlling for all possible time constant confounds. The use of prospective data means that we are less dependent on subjects’ recall. Using a subsample of infertile women who were childless at Wave 1 of the National Survey of Fertility Barriers (NSFB), we are able to examine how infertility resolution (live birth compared to no live birth) is associated with within-person change between waves across three dimensions of well-being: depression, self-esteem and life satisfaction.

Social surveys take various approaches to measuring fertility problems. From a clinical medical perspective, the most desirable measure of fertility problems is clinical diagnosis, but a diagnostic measure requires seeing a doctor or receiving treatment (Dick et al., 2003). Thus, this approach cannot be used in a population-based survey that includes women who do not seek treatment. Using a medical definition of infertility as 12 months of regular, unprotected intercourse without conception (American Society for Reproductive Medicine, 2008), some researchers create ‘constructed’ measures out of contraception and birth histories (e.g. Mascarenhas, et al., 2012; Chandra et al., 2013). Other researchers rely on self-reports, simply asking women if they have experienced a period of regular, unprotected intercourse of 12 months or some other duration that did not result in conception (e.g. Larsen, 2005; Greil et al., 2010a). Estimates of infertility based on self-reports have been found to be an efficient and acceptable method for estimating infertility at the community level (Gurunath et al., 2011). Two studies provide evidence that estimates of infertility-based self-reports roughly match estimates from calendar data (Joffe, 1989; Dick et al., 2003). Self-reports have also been found to be accurate when women are asked to recall past episodes of infertility (Chandra et al., 2013).

Several surveys have taken yet another approach and rely upon self-reports in the form of perceptions about one’s own fertility as distinct from medically defined infertility. For example, the National Survey of Reproductive and Contraceptive Knowledge asked, ‘How likely do you think it is that you are infertile or will have difficulty getting [a woman] pregnant?’ (Polis and Zabin, 2012). The National Survey of Family Growth (NSFG) asked respondents, ‘As far as you know, would you, yourself, have any difficulty getting pregnant (again) or carrying (a/another) baby (after this pregnancy)?’ (Chandra et al., 2013). Measures of perceived fertility do not perfectly reflect measures of medically defined infertility. Studies find that many women who meet medical criteria for infertility do not perceive themselves as infertile (Abbey et al.,. 1994; White et al. 2006; Loftus, 2009; Passet-Wittig et al., 2016) and that women may perceive a problem even if they do not meet medical criteria for infertility (Johnson et al., 2018). In a study using the NSFB, McQuillan et al. (2017) found that both self-perception and meeting medical criteria are significantly associated with depressive symptoms when both measures are included in a regression model. They recommend that researchers use both measures when this is feasible. Because the NSFB includes both a measure of infertility based on meeting medical criteria and a measure based on self-perception, it is possible to explore here whether there are differences in the association of infertility resolution and SWB among women who do and do not perceive themselves as having a fertility problem.

Materials and Methods

Data and sample

The NSFB conducted telephone interviews with a probability-based sample of 4787 US women aged 25–45 years during the years 2004–2007 (Wave 1) with follow-up interviews 3 years after the initial interview (Wave 2). Using random digit dialing, the sampling procedure selected women between the ages of 25 and 45 years. The sample was restricted to this age group because the NSFG study found that infertility for women younger than 25 years only accounted for a small percentage of cases across the USA (only 3%). The investigators wanted a sample that was representative of women in the USA with a landline telephone. The survey also over-sampled high minority census tracks to ensure sufficient numbers of nonwhite women for subgroup analyses. Interviewing was conducted by the Survey Research Center at the Pennsylvania State University and the Bureau of Sociological Research at the University of Nebraska-Lincoln. The same interviewer training material and interviewer guides were used at both sites, and Institutional Review Board approval was secured at both sites before data collection began. The data may be accessed at: https://www.icpsr.umich.edu/icpsrweb/DSDR/studies/36902#bibcite. The NSFB has screening questions to select all women who were at risk for infertility or who previously had infertility, and 1/10th of those with no issue with or risk of infertility for comparison. The response rate to the eligibility screening questions was 54%. Women with a high school degree or less were slightly under-sampled compared to the NSFG, a large in-person survey with a response rate of 90%. The analytic sample for the current study includes all women without children who met medical criteria for infertility or perceived a fertility problem and who were interviewed in both waves of the NSFB (N = 283). Women who were avoiding pregnancy and those who expressed that they did not want to have a baby were excluded from the analysis, as were women who had been sterilized and women who were breastfeeding during the period when they were having regular intercourse with conception. We included women who said they were ‘okay either way’ about becoming pregnant because women who experience infertility do not respond to fertility intentions/behaviors survey measures in the same way that fecund women might; Shreffler et al. (2016) found that infertile women reported lower fertility intentions but stronger fertility desires than fecund women and suggested that the discrepancy may occur because women who do not think they can get pregnant are unsure how to answer survey questions about whether they are trying to get pregnant or the likelihood they will become pregnant within a short timeframe. We used multiple imputation to preserve as many cases as possible for analysis.

We have two reasons for focusing on infertile women with no children at Wave 1. First, there is evidence that infertility alone, which can occur before or after having a child, has a weaker association with psychological distress (McQuillan et al., 2003) and life satisfaction (McQuillan et al., 2007) than involuntary childlessness. It thus makes sense to focus on the infertility experiences of women who initially have no children. Furthermore, focusing on involuntarily childless women who either do or do not become mothers takes full advantage of the longitudinal nature of our sample by making our research design quasi-experimental. Examining change scores allows us to hold within-person differences constant as we look at the effect of an exogenous event (having or not having a child) that either occurs or does not occur between waves. Gameiro and Finnigan (2017) have argued on the basis of a meta-analysis of studies on infertility resolution and SWB among ART patients that it is a failure to achieve parenthood goals rather than involuntary childlessness that is the source of distress among women with unresolved infertility. It should be noted that, by limiting our sample to childless women, we have given up the opportunity to address this question.

Measurement

Dependent variables

The three dependent variables for this analysis are depressive symptoms, self-esteem and life satisfaction. Depressive symptoms were measured using a 10-item modified version of the Center for Epidemiologic Studies—Depression Scale (CES-D; Radloff, 1977; Andresen et al., 1994). The scale included questions such as ‘In the past two weeks…I was bothered by things that don’t usually bother me;’ ‘I felt depressed;’ and ‘My sleep was restless.’ Items were coded or reverse-coded so that high scores indicate high levels of depression. Cronbach’s alpha for the CES-D scale in the NSFB is 0.78.

The measure of self-esteem included three items (‘I feel that I do not have much to be proud of;’ ‘I am a person of worth at least equal to others;’ and ‘All in all, I am inclined to feel I am a failure’) drawn from the Rosenberg Self-esteem Scale (Rosenberg, 1979). Response categories included strongly agree, agree, disagree or strongly disagreed. The items were coded or reverse coded so that a higher score indicates greater self-esteem in the scale, created by taking the mean of responses. To minimize respondent burden, the NSFB survey developers shortened scales whenever possible. The three items included from the original Rosenberg Self-esteem Scale had the highest loadings on one factor and Cronbach’s alpha of 0.77 on a pilot survey of Midwestern women and therefore were selected for the NSFB survey. Cronbach’s alpha within the NSFB is 0.76. In the current analyses, we created a change score for self-esteem by subtracting Wave 1 values from Wave 2 values.

To measure life satisfaction, we used a modified version of the Satisfaction with Life Scale created by Diener and Diener (1995) that has desirable psychometric properties across a variety of ages to measure life satisfaction. Respondents were asked whether they strongly agreed, agreed, disagreed or strongly disagreed with the following statements: ‘In most ways, my life is close to ideal;’ ‘I am satisfied with my life;’ ‘If I could live my life over, I would change almost nothing;’ and ‘So far, I have gotten the important things I want in life.’ This unidimensional scale has an acceptable internal consistency (α = 0.75).

Focal independent variables

A dummy variable indicated whether women had a live birth between waves. Women who gave birth were coded as 1; all others were coded as 0. As noted above, we limited our sample to infertile women who were childless at Wave 1. We included women who met medical criteria for infertility as well as women who perceived themselves to have a fertility problem in order to see whether the use of one or the other of these measures has an influence on the association between infertility resolution and SWB. Women were considered to have met medical criteria for infertility at Wave 1 if they responded ‘yes’ to either question: ‘Was there ever 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?’ or if they ever became pregnant after at least 12 months of having unprotected intercourse without conception as measured in a pregnancy history module. Thus, it was necessary to construct meeting medical criteria from a combination of the two questions plus a more in-depth pregnancy history.

Perception of a fertility problem at Wave 1 was measured by the questions: ‘Do you think of yourself as someone who has, has had or might have trouble getting pregnant?’ and ‘Do you think of yourself as someone who has or has had fertility problems?’ Women were considered to perceive a fertility problem if they answered ‘yes’ or ‘maybe’ to either question. Women who answered ‘no’ to both questions were considered not to have perceived themselves as having a fertility problem. Infertility was therefore operationally defined as either meeting medical criteria for infertility or perceiving a fertility problem. We then classified women as falling into one of three groups: met medical criteria only, perceived a problem only and both met medical criteria and perceived a problem. The three infertility identification groups were included in the multivariate analysis to see whether the infertility identification group was related to the dependent variables. ‘Met medical criteria only’ was used as the reference category. Treatment between waves was assessed with the question: ‘Are you [still] seeking medical help to have a baby?’ This is a binary variable in which women who were in treatment at the time of the Wave 1 interview were coded as ‘1’ and those who were not were coded as ‘0.’

Control variables

We controlled for a number of variables that have been shown to be related to SWB among infertile women. The importance of motherhood scale was created by combining the following items: ‘Having kids is important to my feeling complete as a woman;”’ ‘I always thought I’d be a parent;”’ ‘I think life will be or is more fulfilling with children;”’ and ‘It is important for me to have children.’ The possible response choices were: strongly agree, agree, disagree and strongly disagree. A fifth item included in the scale was, ‘How important is raising children?’ with possible response choices: very important, important, somewhat important and not important. This is a unidimensional scale with a Cronbach’s alpha of 0.86. The scale was coded so that a higher score meant a greater value placed on motherhood. We computed the mean of the variables and then mean-centered the scale for the multivariate analysis. This variable was measured at Wave 1.

Traditional gender ideology was measured by a two-item index created by averaging responses to the following Likert scale items: ‘It is much better for everyone if the man earns the main living and the woman takes care of the home and family’ and ‘If a husband and a wife both work full-time they should share household tasks equally.’ Responses were scored so that higher scores correspond to less egalitarian attitudes. The importance of leisure item was based on the question ‘How important is each of the following to you in your life?’ One of the choices was ‘having leisure time to enjoy my own interest.’ The possible response choices were very important, important, somewhat important and not important, and the variable was recoded from not important to very important. The religiosity scale was developed by the creators of the NSFB and includes the four items: ‘How often do you attend religious services?’ ‘How often do you pray?’ ‘How close do you feel to God most of the time?’ and ‘In general, how much would you say your religious beliefs influence your daily life?’ The scale was standardized, mean-centered and coded so that a higher value indicates greater religiosity. Cronbach’s alpha of this measure is 0.77.

Age at Wave 1 was measured in years. All women were between the ages of 25 and 45 years at Wave 1. Marital status at Wave 1 was assessed via the question, ‘What is your current marital status?’ We created binary indicators for married, cohabiting, formerly married (divorced, separated or widowed) and single. ‘Married’ was treated as the reference category for the multivariate analyses. Education at Wave 1 was measured in years. Economic hardship at Wave 1 was developed from three items: ‘During the last 12 months, how often did it happen that you: had trouble paying the bills?’ ‘…did not have enough money to buy food, clothes or other things in your household needed?’ and ‘…did not have enough money to pay for medical care?’ Possible responses were never, not very often, fairly often and very often. The scale was computed using the mean of the items. We measured household income in dollars and logged it to reduce skew. This variable was constructed based on responses to a question about which category of income best described the family’s total income and the dollar value of the midpoints of the categories was used in the scale. Race/ethnicity was measured using the two standard Census questions. Individuals who reported multiple races were classified giving first priority to identification as ‘Hispanic’ and second priority to identification as ‘Black.’ Dummy variables were constructed for Black, Hispanic and ‘other race’ compared to White.

Statistical analyses

Because the three dependent variables—depressive symptoms, self-esteem and life satisfaction—are continuous, we used ordinary least squares regression in Stata (StataCorp LLC, College Station, TX, USA) to determine whether having a live birth between waves was associated with each of these three aspects of SWB. A value of P < 0.05 was considered as significant. We conducted separate analyses for each of the three variables. In each model, the dependent variable was the change score, computed by subtracting the Wave 1 score from the Wave 2 score. We included all the control variables described above in order to eliminate possible confounds from the analyses. We included a measure of the Wave 1 value (mean-centered) in each analysis to control for the fact that women started with differing levels of SWB at baseline. Women who have higher SWB to start with cannot increase as much as those who start with lower SWB; including baseline SWB controls for this potential confound.

Change-score analysis is an effective way to control for exogenous differences among women who did and did not experience a live birth between waves (Johnson, 2005). Change-score models are used when the goal is to assess the effect of events or experiences on a change from time 1 to time 2 for individuals. Change-score analysis controls for all time-invariant variables even if they are not measured (Allison, 1994; Johnson, 2005). Change-score analysis also decreases the chance of bias from measurement error, which can be problematic when using a lagged dependent variable technique (Johnson, 2005).

Results

Bivariate statistics (scores and SDs for continuous variables and percentages for categorical variables) of study variables are included in Table I, which compares participants by whether or not they resolved their infertility (e.g. had a live birth). Slightly fewer than 20% (N = 47) of women who were involuntarily childless at Wave 1 had a live birth by Wave 2. Twelve percent of the women in the sample received fertility treatment between waves; those who received treatment were more likely to have a live birth between waves. Involuntarily childless women who had a live birth were more likely than those who did not have a live birth to meet medical criteria for infertility and perceive a fertility problem, and they were less likely to meet medical criteria for infertility without perceiving a fertility problem. Bivariate analyses show that there were significant changes in self-esteem for women who had a live birth between waves, but not in depressive symptoms or life satisfaction. Involuntarily childless women who had a live birth were less likely than those who did not to meet criteria for clinical depression at Wave 1 (not shown). Women who resolved their infertility between waves by having a live birth were more likely than women who did not to score high on the importance of motherhood scale and less likely to score high on the traditional gender ideology and religiosity scales. Involuntarily childless women who had a live birth were more likely than those who did not to be younger, to be married, to identify with ‘other race’ and to be more educated. The fact that there were so many differences between groups on variables that have been previously shown to be related to SWB among infertile women suggests the importance of including these variables as controls in our multiple regression models.

Table I.

Descriptive statistics for 283 infertile childless women according to resolution of infertility through live birth.

	Total sample (N = 283)		No live birth (N = 236)		Live birth between waves (n = 47)
Variables	Score/%	SD	Score/%	SD	Score/%	SD	P-value
Subjective well-being outcome variables
Depressive symptoms
CES-D scale W1a	7.12	5.78	7.34	6.1	6.04	3.8
CES-D scale W2b	7.3	6.52	7.41	7	6.77	3.12
CES-D scale change	0.18	5.73	0.08	5.94	0.72	4.7
Self-esteem
Self-esteem scale W1	7.4	1.83	7.41	1.89	7.37	1.47
Self-esteem scale W2	7.44	2.31	7.35	2.39	7.9	1.63
Self-esteem scale change	0.04	1.97	−0.06	1.98	0.53	1.77	*
Life satisfaction
Life satisfaction scale W1	7.71	2.5	7.58	2.45	8.34	2.91
Life satisfaction scale W2	7.46	2.86	7.29	2.92	8.29	2.43	*
Life satisfaction scale change	−0.25	2.77	−0.29	2.57	−0.05	3.59
Focal independent variables
Birth between waves	17%
Infertility identification groups							*
Medical criteria only	29%		31%		15%
Perceived fertility problem only	27%		27%		29%
Medical criteria + Perceived fertility problem	44%		42%		57%
Fertility treatment between waves	12%		9%		29%		**
Control variables
Importance of motherhood scale (mean)	2.8	0.93	2.71	0.9	3.26	0.79	***
Traditional gender ideology scale (mean)	3.56	1.34	3.56	1.31	3.58	1.5
Importance of leisure scale (mean)	3.4	0.85	3.47	0.75	3.08	1.09	*
Religiosity scale	−1.5	4.99	−1.36	4.79	−2.24	5.69
Age	35.53	7.85	36.44	7.81	30.97	4.43	***
Union status							*
Married	51%		47%		71%
Cohabiting	11%		11%		8%
Formerly married	12%		13%		6%
Single	26%		29%		14%
Race/ethnicity							*
White	69%		72%		53%
Black	13%		13%		12%
Hispanic	7%		7%		8%
Race-other	12%		9%		27%
Education (years)	14.92	3.47	14.74	3.52	15.81	2.75	*
Economic hardship scale (mean)	4.69	2.81	4.66	2.91	4.83	2.21
Log-household income	10.76	0.95	10.74	0.95	10.87	0.96

^{a a}W1 = Wave 1

^{b b}W2 = Wave 2

^* P < 0.05, ^**P < 0.01, ^***P < 0.001. Student’s t tests for scale variables; chi square tests for categorical variables. CES-D, Center for Epidemiologic Studies—Depression Scale. Data source: Wave 1 and Wave 2 of National Survey of Fertility Barriers.

Only 35 (12%) of the involuntarily childless women in our sample received treatment between waves. This marks the women in our sample as quite different from the clinic patients examined in other studies of the association between infertility resolution and SWB. Table II displays some characteristics of the women who received treatment between waves. Most (60%) of those who received treatment between waves did not experience a live birth. Eighty-five percent of the women who had experienced a live birth between waves attributed the resolution of their infertility (e.g. pregnancies resulting in live births) to treatment. Among those who received treatment, women who resolved their infertility through giving birth experienced an increase in life satisfaction, while women who remained childless experienced a decrease.

Table II.

Characteristics of infertility status, treatment and well-being by infertility resolution for 35 women who received any treatment between waves.

	Total: any treatment (n = 35)		No resolution (60%)		Birth (40%)
	Score/%	SD	Score/%	SD	Score//%	SD	P-value
Infertility identification groups							*
Medical criteria only	36%		51%		12%
Perceived fertility problem only	7%		0%		19%
Medical criteria + Perceived fertility problem	57%		49%		69%
Treatment type between waves
IVF	58%		64%		47%
Drug	48%		37%		67%
Insemination	25%		16%		40%
Other treatment	7%		6%		9%
Treatment resulted in pregnancy	42%		15%		85%		*
Treatment resulted in birth	33%		0%		85%
CES-D scale change	1.35	4.34	0.71	5.04	2.36	2.97
Self-esteem scale change	−0.13	2.39	−0.41	2.42	0.31	1.68
Life satisfaction scale change	−0.58	3.75	−1.36	3.48	0.66	1.72	*

^* P < 0.05, ^**P < 0.01, ^***P < 0.001. Student’s t tests for scale variables; chi square rests for categorical variables. Data Source: Wave 1 and Wave 2 of National Survey of Fertility Barriers.

Multiple regression analyses of the association of infertility resolution through live birth compared to no live birth on changes in depression, self-esteem and life satisfaction across three years using change score analyses are presented in Table III. In the multivariate models, we did not find that infertility resolution was associated with changes in depressive symptoms between waves (Model 1), but having a live birth was associated with significantly greater improvements in previously involuntarily childless women’s self-esteem (Model 2) and life satisfaction (Model 3) between waves. Whether or not women received treatment between waves was not associated with changes in SWB. Whether women met medical criteria for infertility only, perceived themselves as having a fertility problem only or both met criteria and perceived a fertility problem was not associated with changes in depressive symptoms or life satisfaction. However, women who both met medical criteria for infertility and perceived themselves as having a fertility problem experienced greater declines in self-esteem than women who met medical criteria only but who did not perceive themselves as having a fertility problem. The negative signs for the B’s for the Wave 1 values of the dependent variables mean that involuntary childless women with higher self-esteem and higher life satisfaction were less likely to experience an increase in these domains of SWB 3 years later, while involuntarily childless women with more depressive symptoms were likely to experience more symptoms over time. Women with a more traditional gender ideology experienced less improvement in self-esteem between waves compared to those with a less traditional gender ideology. Higher religiosity at Wave 1 was associated with modest improvements in self-esteem and life satisfaction between waves. Women who were no longer married reported a greater increase in depressive symptoms compared to married women. Both Black women and women who reported ‘other race’ experienced significant declines in life satisfaction over time. Women with higher incomes reported greater increases in life satisfaction than women with lower incomes.

Table III.

Infertility resolution and change in subjective well-being (depressive symptoms, self-esteem and life satisfaction).

	Model 1 Depressive symptoms		Model 2 Self-esteem		Model 3 Life satisfaction
	b	SE	b	SE	b	SE
Focal independent variables
Gave birth between waves	−0.04	(0.82)	0.74*	(0.30)	1.06*	(0.47)
Infertility groups (ref = medical criteria only)
Perceived fertility problem only	0.35	(0.82)	−0.24	(0.31)	0.02	(0.39)
Medical criteria+ Perceived fertility problem	0.42	(0.88)	−0.54*	(0.25)	−0.07	(0.35)
Fertility treatment between waves	0.52	(0.81)	−0.29	(0.37)	−0.19	(0.51)
Wave 1 subjective well-being variables
CES-D scale	−0.41***	(0.08)
Self-esteem scale			−0.52***	(0.09)
Life satisfaction scale					−0.49***	(0.08)
Control variables
Importance of motherhood scale	0.37	(0.51)	0.12	(0.15)	−0.15	(0.19)
Traditional gender ideology scale	0.20	(0.35)	−0.21*	(0.10)	0.03	(0.15)
Importance of leisure scale	−0.36	(0.47)	−0.01	(0.15)	−0.04	(0.21)
Religiosity scale	−0.01	(0.08)	0.06*	(0.03)	0.07+	(0.04)
Age	−0.03	(0.06)	0.02	(0.02)	0.01	(0.02)
Union status (ref = married)
Cohabiting	0.67	(1.15)	0.00	(0.33)	0.25	(0.47)
Ever married	2.36+	(1.21)	−0.61	(0.44)	0.01	(0.52)
Single	0.63	(0.96)	0.39	(0.27)	0.40	(0.38)
Race/ethnicity (ref = white)
Black	1.78	(1.14)	0.04	(0.29)	−0.60+	(0.35)
Hispanic	−0.17	(1.23)	−0.24	(0.38)	0.31	(0.41)
Race-other	−0.16	(0.94)	0.07	(0.39)	−0.95	(0.65)
Education (years)	−0.14	(0.13)	0.01	(0.05)	−0.02	(0.05)
Economic hardship scale	−0.20	(0.17)	0.00	(0.06)	−0.13	(0.08)
Log-household income	−0.14	(0.63)	0.40+	(0.23)	0.35	(0.27)
Intercept	7.03	(8.25)	−0.63	(2.77)	0.75	(3.59)
N	283		283		283
R 2	0.16		0.23		0.21

SEs in parentheses. ⁺P < 0.10, ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 ordinary least squares (OLS) regression. Data source: Wave 1 and Wave 2 of National Survey of Fertility Barriers.

Discussion

The primary goal of this study was to explore whether resolution of infertility through a live birth compared to no live birth is associated with changes in SWB among women who met criteria for and/or perceived a fertility problem and had no children at Wave 1. Findings suggest that while having a live birth is associated with higher self-esteem and life satisfaction, depressive symptoms may not be as malleable or quick to respond. It is not surprising that depressive symptoms are less closely associated than self-esteem or life satisfaction with the resolution of infertility since we know that there are many factors, including personality traits and mental health conditions, which are associated with depression. Further, childbirth can increase the risk of depression for some women (Evenson and Simon, 2005; Sejbaek et al., 2015), which may have suppressed the effect for other women in the sample.

Our findings are at odds with those of several studies that have found that infertility resolution is associated with a reduction in depressive symptoms (Mindes et al., 2003; Verhaak et al., 2005; Verhaak et al., 2007b; Johansson et al., 2010; Gameiro et al., 2014). It is possible that these dissimilarities may be the result of differences in samples and methodological approaches. None of these studies employed a representative sample that included nontreatment seekers. Only two of these studies (Verhaak et al., 2005, 2007b) were longitudinal in nature, comparing the same women at different points in time, and they looked only at IFV/ICSI patients. Our sample included only 35 women who received treatment between waves and only 21 who received ART. It may be that the stresses associated with fertility treatment, and ART in particular, are such that the association between nonresolution of infertility and depressive symptoms is strongest among women who have undergone unsuccessful treatment. We did not find that receiving fertility treatment between waves was associated with a change in depressive symptoms, but this could be due to the small number of women who received treatment between waves. Research on a sample of involuntarily childless women that includes relatively equal proportions of those who did and did not seek treatment may help to clarify this. Only one of the studies that found a relationship between infertility resolution and depressive symptoms (Mindes et al., 2003) controlled for characteristics known to be associated with depression. It seems unlikely, however, that the presence or absence of controls is the source of the divergence of our findings from those of the other studies, given that we did not find a bivariate relationship between change in depressive symptoms and the resolution of infertility (see Table I).

This is, to our knowledge, the first longitudinal study of the association between infertility resolution and life satisfaction that employs a well-established measure of life satisfaction among a representative population and controls for other characteristics known to be associated with life satisfaction. We are aware of two studies that specifically examined the association between infertility resolution and life satisfaction. In a cross-sectional study that is consistent with our findings, Kuivasaari-Pirinen et al. (2014) compared 296 women who had had successful ART to 209 women who had had unsuccessful ART and found that women with successful ART reported greater life satisfaction. This study, however, had several methodological shortcomings. First, because it was a cross-sectional study, it is impossible to know if the two groups of women differed in life satisfaction before experiencing ART. Second, the study did not control for characteristics of the women that are known to be associated with life satisfaction. Third, the researchers used a self-developed scale that had not been previously validated. The only longitudinal study of which we are aware (Schanz et al., 2011) reported that successful treatment for infertility did not have an effect on life satisfaction 5 years after baseline. This difference in findings as compared with ours might be due to differences in measurement or time frame. The Schanz et al. (2011) study used a domain-specific measure of life satisfaction rather than a global measure, did not control for other relevant variables and did not report significance levels. Additionally, the longer time frame of 5 years might influence differential results; it is possible that differences in life satisfaction between those who have a live birth and those who do not resolve themselves eventually.

We are aware of two longitudinal studies that address the association between infertility resolution and self-esteem, and both are in accord with our findings. In a study of 123 clinic patients who had not previously given birth, Mindes et al. (2003) reported that women who remained childless 6–12 months after the initial interview reported lower self-esteem. In a 10-year follow-up study of infertile women who presented at a fertility clinic, Wischmann et al. (2012) found that those who had become mothers had higher self-esteem than those who did not become mothers. Hence, our study supports the findings of other studies with regard to infertility resolution and self-esteem. Several studies have found that infertility resolution is associated with sense of coherence (SOC) (Habroe et al., 2007; Johansson et al., 2010). Unfortunately, we were unable to investigate the relationship between infertility resolution and SOC because the NSFB does not include a measure of SOC.

As noted in the introduction, we compared three infertility identification groups of women: women who met medical criteria for infertility but did not perceive a fertility problem, women who perceived a fertility problem but who did not meet medical criteria for infertility and women who both met medical criteria and perceived a fertility problem. Infertility identification group was not associated with the magnitude of changes in depressive symptoms or life satisfaction, but women who both met medical criteria for infertility and perceived themselves as having a fertility problem experienced greater declines in self-esteem than women who met medical criteria only but who did not perceive a fertility problem. This finding has some important methodological implications. Some researchers might be inclined to argue that women who perceive a fertility problem but do not meet medical criteria for infertility are not ‘really’ infertile and should not be included in studies of infertility. Yet, our findings suggest that when one is studying SWB and infertility via survey research, there seems to be no reason to regard studies that use self-perception measures as inherently inferior to studies that measure infertility based on meeting medical criteria. The fact that women who both meet medical criteria for infertility and perceive a problem experience declining self-esteem relative to other groups suggests that both the objective state of infertility and its subjective interpretation are relevant for SWB.

This study uncovered some other interesting findings as well. Women with more traditional gender ideology experienced less improvement in self-esteem between waves. Religiosity was associated with positive changes in self-esteem and life satisfaction even after controlling for whether or not an involuntary childless woman had a live birth. This may be due to enhanced social support that religious affiliation membership can provide (Morton et al., 2017), to differences in perceptions of the experience of infertility (Ventura et al., 2007) or to the general association of religion with mental health (Hackney and Sanders, 2005). In any case, the finding that certain attitudes and beliefs are associated with improved SWB suggests that counseling may have an impact on infertile women (Gameiro and Finnigan, 2017). It is well beyond the scope of this paper to discuss the extensive literature on coping with infertility and infertility counseling.

Additionally, we found that women who were no longer married reported a greater increase in depressive symptoms compared to married women. Both Black women and women who reported ‘other race’ experienced significant declines in life satisfaction over time. Women with higher incomes reported greater increases in life satisfaction than women with lower incomes. The fact that certain personal characteristics appear to be associated with SWB suggest that, when possible, studies of the association between infertility resolution and SWB should control for these characteristics.

This study has some limitations. First, we examined change among women in our sample over the relatively short time span of 3 years. It is unclear what changes we might have seen were we able to observe these women for a longer period of time or to observe them across multiple intervals. There is some evidence that the negative effects of failing to resolve infertility (e.g. remaining involuntarily childless) decline or disappear over time (Verhaak et al., 2007b; Wischmann et al., 2012; Kuivasaari-Pirinen et al., 2014). Furthermore, because we measured receipt of treatment between waves, it is possible that some women who received treatment more than 3 years ago were inadvertently omitted from the ‘received treatment’ category. In addition, the relatively small number of involuntarily childless women who received treatment may have made it difficult to detect significant associations between treatment and the relationship between infertility resolution and SWB.

It would be interesting to compare differences in SWB among women depending upon the strength of their desire to have a baby. As noted above in the Materials and Methods section, we included both women trying to become pregnant and those who reported ambivalence (e.g. ‘okay either way’ about getting pregnant). We suspect that in doing so, our findings are likely conservative estimates because women who are ambivalent about becoming pregnant may be less likely to experience a change in SWB due to prolonged infertility. It would also be interesting to explore whether the association between resolution of infertility via adoption and SWB is similar to or different from the association between live birth and SWB, but there were only four women in the sample who adopted, so we were unable to examine this research question. Thus, there is a need for replication with a larger sample of involuntarily childless women.

Survey measurement limitations also restricted our analyses. It would be informative to examine these associations within the context of adverse birth outcomes (e.g. premature birth, delivery complications, maternal and child health problems, etc.) because women who give birth following infertility and treatment are at greater risk for adverse birth outcomes (Joffe and Li, 1994; Welmerink et al., 2010), which might in turn affect SWB. Unfortunately, the survey did not include these data. Question order in the survey could also influence survey responses, because the SWB items were asked after the fertility problem questions, possibly prompting respondents to think that they should have worse SWB if they perceive a fertility problem. Yet, consistency in question order between waves should limit the impact of question order on SWB change. Another potential limitation stems from the fact that the NSFB was conducted in the USA, where medical expenditures are high, where many people do not have health insurance, and where most fertility treatment expenses are not covered by insurance. A study of eight fertility practices computed that the mean cost of infertility treatment over an 18-month period was $15 388, and the mean cost for IVF was $24 373 (Katz et al., 2011). Thus, it may not be possible to generalize findings based on US data to other modern industrialized societies. Finally, while it is a common practice to make causal interpretations based on panel data, such interpretations should be made with caution.

Despite limitations, this study makes important contributions to our understanding of the relationship between infertility resolution and SWB. This is the first prospective study to examine the association between infertility resolution and SWB using a nationally representative sample that includes women who seek and do not seek treatment for infertility and controlling for personal characteristics known to be associated with SWB. Improving understanding of the associations between infertility and women’s well-being over time suggests important implications for health care and mental health practitioners working with infertile women and couples. Infertile women and practitioners alike can take comfort from the knowledge that the successful resolution of infertility is associated with improvement in some subjective indicators of QoL.

Authors’ roles

K.M.S. conceptualized the current study and wrote the first draft of the manuscript. A.L.G. assisted with the study design and substantially revised the manuscript. S.M.T. conducted the statistical analyses and assisted with drafting of the manuscript. J.M. played an instrumental role in supervising data collection, helped to conceptualize the analyses and assisted with revisions of the manuscript. All authors have approved the final version of the manuscript.

Funding

NICHD (#R01-HD044144); NIGMS (#P20-GM109097); National Institutes of Health.

Conflict of interest

The authors have no conflicts of interest to declare.