Gravidity and Fertility Desires of Female Cancer Survivors and Age-Matched Controls: Using the National Survey of Family Growth

Abstract

Purpose:

This study analyzes differences in gravidity, fertility desires, and fertility treatment utilization among female cancer survivors relative to age-matched female controls with no history of cancer.

Methods:

This cross-sectional study utilized four data sets from the National Survey of Family Growth (NSFG) ranging from 2011 to 2019. Out of 16,172 participants, 310 cancer survivors and 620 age-matched controls were identified. Secondary analysis assessed subgroups of survivors based on age at diagnosis, categorizing those diagnosed at age less than 20 years old as childhood cancer survivors (CCSs). Statistical analysis involved frequencies (%), chi-square tests, and logistic regression models to compare reproductive history, fertility desires, and treatments sought among groups.

Results:

There was no significant difference in gravidity between female cancer survivors and age-matched controls. Gravid CCSs were more likely to want a/nother baby than gravid controls (p = 0.017). Nulligravid subjects had higher rates of fertility desires than gravid women, except in the CCS group. There were no significant differences in seeking fertility treatment between survivor groups and controls, except for higher rates of in vitro fertilization treatment in survivors diagnosed before their first pregnancy (3.4% vs. 0.3%, p = 0.017).

Conclusion:

Overall rates of gravidity are similar between female cancer survivors and non-survivors. However, there are differences in CCSs’ desires for fertility. These findings emphasize the importance of thorough fertility preservation counseling and proactive consideration of fertility desires from the time of diagnosis through survivorship (particularly, for women diagnosed in childhood).

Introduction

In recent years, advancements in the detection and treatment of cancer have led to increased survival rates and a growing population of cancer survivors. In fact, as of January 2022, there were an estimated 1,015,440 female cancer survivors aged 0–49 years old in the United States. ¹ For many of these women, the journey to parenthood post-cancer is influenced by lasting adverse effects of cancer therapy on their reproductive health.^2,3 This is further complicated by the gradual increase in mean age of first-time mothers in the United States—notably rising from 21.4 years old in 1970 to 27.1 in 2020.^4,5 These demographic shifts may further impact the reproductive needs of cancer survivors.

To conceive and carry a pregnancy naturally, a female cancer survivor must have sufficient ovarian follicular reserve, a properly functioning hypothalamic–pituitary–ovarian axis, a uterus capable of supporting fetal development, and organ systems that can adequately adapt to the demands of pregnancy. ⁶ Unfortunately, cancer therapy can impact each of these critical components, with the extent of the impact depending on factors such as the type and dosage of treatment received, as well as the patient’s age and pubertal development at the time of exposure. ⁷ For example, the Childhood Cancer Survivor Study demonstrated increased infertility diagnosis in survivors compared to their siblings, while also finding similar rates of seeking care for infertility. ⁸ Childhood cancer survivors (CCSs) were less likely than their siblings to be prescribed fertility-enhancing medications, such as clomiphene, in spite of insufficient evidence for decreased success rates. Outcomes in assisted reproductive technology (ART) measures for this population were not different than those in their siblings, though there were differences in the secondary analyses based on treatment regimens. ⁹ These studies suggest perhaps reproductive endocrinologists perceive these patients to have less likelihood of success with treatments or that the survivors themselves are less likely to undergo treatment after extensive and discouraging counseling. The overall rate of premature ovarian insufficiency (POI) in women who received cancer treatment (excluding surgery such as oophorectomies) as children is believed to be around 8% in comparison to 0.8% in their female siblings. ¹⁰ Given that adverse effects of different therapies can range anywhere from subfertility to POI, it is imperative that patients of all ages be proactively counseled about their options for individualized fertility preservation prior to starting treatment in order to significantly improve the chances of preserving optimal fertility potential. ¹¹

This study uses the National Survey of Family Growth (NSFG) database, which is a nationally representative, de-identified questionnaire of the U.S. household population, which sources data through face-to-face interviews with women aged 15–49 regarding pregnancy, births, infertility, and reproductive health. ¹² Although there are limitations in using data from a source without the specific aims of our study, we aim to use the NSFG to describe differences in gravidity, fertility desires (as the NSFG asks all respondents if they “would like to have a/nother baby in the future”), and self-reported fertility treatment histories of a diverse group of female cancer survivors as compared to an age-matched cohort of women without a history of cancer. We create a subgroup of subjects who are defined as CCSs based on age of diagnosis at 20 years or younger. Given the aforementioned literature describing both the gonadotoxicity of cancer treatments and the subjects of Childhood Cancer Survivor Study rates of fertility desires and treatment, we hypothesize the cohort of cancer survivors will have decreased gravidity, similar or increased desire for fertility, and similar rates of treatment—while CCSs are hypothesized to have even lesser gravidity and lower rates of treatment. This study, with the NSFG’s generalizable sample of American women, affirms certain findings of published research, tests their applicability to subgroups of survivors (nulliparous and parous), and demonstrates the importance of expectant and comprehensive fertility counseling.

Methods

In this cross-sectional study, four public-use datasets (2011–2013, 2013–2015, 2015–2017, and 2017–2019) from the NSFG were downloaded, combined, and loaded to SPSS® (version 26).^13–16 The respondent files included women aged 15–44, all of whom were eligible for inclusion in the study. Of note, the 2017–2019 data included women aged 45–49, all of whom were excluded to avoid bias.

Respondents who indicated “yes” to ever having had cancer were identified; those with the additional variable of their cancer diagnosis type were included. A disproportionately large number of women reported a diagnosis of cervical cancer—this raised suspicion for inaccurate self-reporting, likely by women who had required therapeutic interventions such as loop electrosurgical excision procedures or cold knife cones for high-grade cervical dysplasia. To avoid bias of the results, all women reporting cervical cancer were excluded. The remaining cancer survivors were included as the survivor cohort. Next, the remaining women without a cancer history were randomized. The SPSS case-control matching function was used to match survivors to controls by age at time of study (tolerance set to zero). Matched controls were removed, and the process was repeated in order to create a cohort of controls at a 2:1 ratio with survivors. The age-matching was confirmed with a two-tailed t-test comparing the ages of the women in each cohort. The cancer survivors were sub-characterized for secondary analysis based on their age at diagnosis, less than 20 years old being considered CCSs.

Fertility history of survivors and controls was compared, including gravidity, age at first pregnancy, and age at last pregnancy. These variables were then compared among subgroups of survivors versus controls. Furthermore, the timing of diagnosis for survivors in relation to their pregnancies was analyzed. Next, desire for future fertility was assessed. The NSFG asks respondents, “(Looking to the future, do/If it were possible, would) you, yourself, want to have (a/nother) baby at some time (after this pregnancy is over/in the future)?” Nulligravid women were compared with gravid women overall; survivors and controls were then compared. The subgroups of survivors were analyzed, as well, in addition to cancer diagnosis and pregnancy timing for gravid survivors. Finally, fertility counseling and treatment were assessed, including having been to a provider about becoming pregnant, receiving only advice, undergoing fertility testing, taking fertility-enhancing medications, having surgery for “blocked tubes,” undergoing artificial insemination, and undergoing in vitro fertilization (IVF).

Statistics included frequencies (%), chi-square tests and logistic regression models, as appropriate, with odds ratios and 95% confidence intervals. A significance value (α) of 0.05 was used for all analyses. Models that were adjusted for demographics are reported as such. When analyzing CCSs, the age-matching was no longer in effect between cohorts; therefore, results are controlled for age in all analyses. Demographic differences are described and are not considered confounding as the population is nationally representative and therefore, it is presumed these differences are present in the general population of survivors, which we aim to describe.

Results

The 2011–2019 combined dataset included data for 21,717 American women (Fig. 1). After exclusions for the 2015–2017 dataset (which did not include cancer type as a variable), 16,172 subjects remained. There were 503 respondents who indicated “yes” to ever having had cancer and included information on their diagnoses. Of these, 193 identified their cancer type as being cervical and were therefore excluded (see Methods). Therefore, 310 cancer survivors were included in this study. There remained 15,668 valid controls who did not report a cancer history. Cases of survivors were randomly age-matched at a 1:2 ratio with controls for a total number of 930 subjects (620 controls) included in this study. The resulting mean age at time of survey of both groups was 35.1 years old, and both groups had a standard deviation of 7.37 years (p = 1.00); therefore, the matching process was shown to be successful.

FIG. 1.

Flowchart demonstrating the subject selection process from public NSFG datasets. NSFG, National Survey of Family Growth.

Table 1 demonstrates the demographics of cancer survivors and the frequency of cancer types. Cancer survivors were more likely to be White (67.1% of survivors vs. 46.6% of controls, p < 0.0001) and equally as likely to have health insurance as controls (p = 0.240). The most common cancers reported were melanoma (21.9%), breast (16.5%), and ovarian (14.2%). As shown in Table 2, gravidity was not significantly different between survivors and controls in the overall cohort: 21.9% of controls and 24.2% of survivors were nulligravid (p = 0.200). The mean ages at first pregnancy for survivors and controls were 21.7 years (standard deviation [SD] 5.8) and 22.9 years (SD 5.8), respectively (p = 0.006). CCSs had a mean age of 20.6 years (SD 5.7) at first pregnancy (vs. controls, p = 0.009). The mean ages at last pregnancy for survivors and controls were 28.4 years (SD 6.1) and 29.1 years (SD 5.7), respectively (p = 0.134). Of gravid survivors, timing of pregnancy and cancer were analyzed (N = 235): 24.7% (58) were diagnosed before their first pregnancy, 28.1% (66) between pregnancies or in the same year as their first/last pregnancy, and 47.2% (111) were diagnosed after their last pregnancy.

Table 1.

Demographics of Controls vs. Survivors and Cancer Diagnoses of Survivors (N = 930)

	Controls (N = 620)	All cancer survivors (N = 310)	p-Value	CCSs (N = 78)	p-Value
	Mean (SD, range)	Mean (SD, range)		Mean (SD, range)
Age	35.1 (7.37, 16–44)	35.1 (7.37, 16–44)	1.0	29.71 (8.03,16–44)	<0.0001
Race	N (frequency)	N (frequency)		N (frequency)
White	289 (46.6%)	208 (67.1%)	<0.0001	50 (64.1%)	0.012
Black/African American	139 (22.4%)	47 (15.2%)		15 (19.2%)
Hispanic	144 (23.23%)	41 (13.2%)		8 (10.3%)
Other	48 (7.7%)	14 (4.5%)		5 (6.4%)
Current insurance status	N (frequency)	N (frequency)		N (frequency)
Private insurance	327 (53.4%)	168 (54.2%)	0.240	35 (44.9%)	0.073
Public insurance	124 (20.3%)	74 (23.9%)		25 (32.1%)
No insurance	161 (26.3%)	68 (21.9%)		18 (23.1%)
Unknown	8	0		0

Cancer diagnosis	N (frequency)	N (frequency)
Bone	5 (1.6%)	4 (5.1%)
Brain/CNS	6 (1.9%)	4 (5.1%)
Breast cancer	51 (16.5%)	1 (1.3%)
Colon	7 (2.3%)	2 (2.6%)
Endometrial	19 (6.1%)	6 (7.7%)
Head and neck	2 (0.6%)	1 (1.3%)
Leukemia	18 (5.8%)	12 (15.4%)
Liver	1 (0.3%)	1 (1.3%)
Lung	1 (0.3%)	1 (1.3%)
Lymphoma	14 (4.5%)	4 (5.1%)
Melanoma	68 (21.9%)	13 (16.7%)
Neuroblastoma	3 (1.0%)	3 (3.8%)
Other	26 (8.4%)	8 (10.3%)
Ovarian	44 (14.2%)	9 (11.5%)
Rectal	3 (1.0%)	0 (0.0%)
Renal	6 (1.9%)	2 (2.6%)
Skin (non-melanoma)	8 (2.6%)	1 (1.3%)
Skin (unknown type)	4 (1.3%)	0 (0.0%)
Soft tissue	2 (0.6%)	2 (2.6%)
Stomach	5 (1.6%)	1 (1.3%)
Throat	1 (0.3%)	1 (1.3%)
Thyroid	16 (5.2%)	2 (2.6%)

CCS, childhood cancer survivors; CNS, Central Nervous System.

Table 2.

Gravidity of Controls, Survivors, and Childhood Cancer Survivors (CCS) Subgroup, as Well as Diagnosis Timing Regarding Pregnancy for Gravid Survivors

Gravidity	Controls (620)	All survivors (310)	CCSs (78)
Nulligravid	21.9% (136)	24.2% (75)	38.5% (30)
1 pregnancy	13.2% (82)	15.8% (49)	17.9% (14)
2 pregnancies	21.9% (136)	16.5% (51)	12.8% (10)
3+ pregnancies	42.9% (266)	43.5% (135)	30.8% (24)
p-Value	Ref.	0.376	0.468 a
Diagnosis timing (gravid only)		All survivors (235)	CCSs (48)
Before the first pregnancy		24.7% (58)	77.1% (37)
Between/same year		28.1% (66)	20.8% (10)
After the last pregnancy		47.2% (111)	2.1% (1)

^aAge-adjusted regression for CCSs vs. controls.

Of the survivors, 78 (25.2%) were diagnosed before age 20 and considered to be CCSs. When evaluating this sub-cohort (which is inherently not age-matched with controls), gravidity was the same as controls (age-matched p = 0.468). CCSs were more likely to have been diagnosed before their first pregnancy (37/48, 77.1%).

When asked if they would desire a/nother baby in the future, 51.0% of cancer survivors said “yes” versus 46.4% of controls (p = 0.189) (Table 3). Overall, nulligravid women were more likely than gravid women to want a/nother baby (p < 0.0001). Survivor status did not change this finding (p = 0.275). However, a disproportionately higher number of gravid CCSs than gravid controls desired future fertility (66.7% vs. 37.5%, age-corrected p = 0.015).

Table 3.

Desire for A/Nother Baby Based on Survivor Status and Gravidity

	Controls (620)	All survivors (310)	p-Value	CCSs (78)	p-Value
Overall	46.4% (282)	51.0% (156)	0.189	72.7% (56)	0.033 a
Nulligravid	78.6% (103)	71.6% (53)	0.260	82.8% (24)	0.695 a
Gravid	37.5% (179)	44.4% (103)	0.080	66.7% (32)	0.015 a

^aAge-adjusted regression for childhood cancer survivors (CCSs) vs. controls.

When asked if they “have been to a provider about becoming pregnant,” 12.7% of survivors said “yes” versus 10.9% of controls (p = 0.417). Table 4 shows the rates of fertility services sought by survivors and subgroups compared to healthy controls. There were no significant differences between any survivor group compared to controls, aside from the subgroup of patients who were diagnosed before their first pregnancy who reported higher rates of IVF treatment: 3.4% of these survivors (2/58) underwent IVF versus 0.3% (2/620) of controls (p = 0.017).

Table 4.

Fertility Treatments Sought by Controls, Survivors, and Two Subgroups of Survivors

	Controls (620)	All survivors (310)	CCSs (78)	Diagnosed before first pregnancy (58)
Seen a provider	10.9% (67)	12.7% (39)	13.5% (10)	13.8% (8)
Advice	8.4% (52)	10.3% (32)	9.0% (7)	12.1% (7)
Testing	7.4% (46)	7.1% (22)	7.7% (6)	8.6% (5)
Medications	5.6% (35)	6.5% (20)	6.4% (5)	6.9% (4)
Surgery for blocked tubes	2.3% (14)	2.3% (7)	1.3% (1)	1.7% (1)
Artificial insemination	2.4% (15)	2.6% (8)	0.0% (0)	0.0% (0)
IVF	0.3% (2)	1.3% (4)	1.3% (1)	3.4% (2) a

^aDenotes statistical significance. p = 0.017.

IVF, in vitro fertilization.

Discussion

This population-based study of female cancer survivors ages (15–44) assesses reproductive health and desires utilizing the NSFG, a large dataset characteristic of U.S. households. It provides valuable insight beyond the limitations of previous studies. This cohort is an important addition to the current literature, which consists mostly of case-control studies from single institutions with greater potential for selection bias and geographical influence.

Our cohort demonstrates a higher rate of White cancer survivors, which is consistent with previous reports of a statistically significant association between race and cancer survival. ¹⁷ In order to assure that prevalent socioeconomic determinants of health are also considered, the demographics of patients’ insurance status were analyzed. Though one may presume a patient’s cancer history is likely to serve as a precursor for a future of high health care costs, our analysis showed no significant difference in insurance coverage between cancer survivors and controls.

The most common cancer diagnosis in this cohort was melanoma, followed by breast and ovarian. These seem to be disproportionate to current U.S. incidence rates, with melanoma making up 21.9% of this cohort and having an incidence of 7.8 per 100,000 U.S. women less than 50 years old; whereas, breast cancer makes up 16.5% of this cohort with an incidence of 48.9 per 100,000.^18,19 This may be due to selection bias, as the sources of these statistics are different, or due to the high rate of melanoma survival inflating the populations of survivors available to survey. This may also contribute to the racial differences, given the increased incidence of skin cancer in White women. ²⁰

Our data revealed no significant difference in gravidity between female survivors (all cancer types) and controls, in contradiction to our hypothesis. Notably, this is also not concordant with data suggesting lower pregnancy rates in cancer survivors compared to the general population.^21,22 The subgroup of CCSs also had non-different gravidity rates, which demonstrates preserved fertility in this population, likely for those who received treatments with low gonadotoxicity.

About a quarter (28.1%) of survivors were diagnosed between pregnancies or in the same year as a pregnancy. This underscores the importance of routine screenings during pregnancy-related engagement in the health care system. As expected, CCSs (diagnosed before age 20) are more likely to have had cancer prior to their first pregnancy. However, unexpectedly, CCSs have a lower mean age at first pregnancy than controls (20.6 years vs. 22.9 years). This may not be clinically significant—or, as suggested by prior literature, it may be due to earlier attempts at pregnancy driven by awareness of their potential for POI and decreasing fertility, or conversely, a mistaken belief that they cannot become pregnant.^23–26 It is important for providers to counsel about the likelihood of retained fertility to ensure those who do not desire children seek contraceptive options and do not rely on false beliefs of definite infertility.

Our study also includes data regarding fertility desires among the participants. The results suggest that special attention should be taken when considering the fertility desires of CCSs. Female cancer survivors overall are as likely as controls to desire a/nother baby in the future. Gravid CCS, however, are more likely to want a/nother baby than controls. These findings suggest that women diagnosed earlier in life have an increased desire for another baby, but only when they have had at least one pregnancy. The explanation may be due to a proportion of nulligravid women in both cohorts who do not desire children for a variety of reasons such as lifestyle, cost of raising children, and careers. It is also possible that the desire to have a/nother baby is shadowed by the acceptance of some women who know they cannot, though the survey question is posed as “if it were possible.” This may signify diminished fertility in this population, such as POI/early menopause, which can be secondary to gonadotoxic treatments. This can be monitored through Anti-Müllerian Hormone measurement, which would be expected to decline at a quicker rate than in a non-survivor. ²³

Regarding health care sought for help becoming pregnant, survivors and controls reported similar rates (12.7% and 10.9%, respectively). As seen in other reports, there were no differences in the levels or types of fertility treatment sought based on this survey. However, very few women in both groups underwent IVF, limiting the power of this particular finding. Prior studies purported bias in offering ART to CCSs despite insufficient evidence for decreased success.^8,9,27 With an understanding of the complexity of their health history and treatment gonadotoxicity, it is possible that survivors have “given up” on future fertility and may not seek fertility care due to fear of futility, fear about the potential dangers of pregnancy, risk of recurrence, or genetic risks to their children.^24,26,27 However, it is suggested that decreased success with ART may be found only in certain CCS subgroups: those who received cranial or pelvic radiation therapy, but not in those who received alkylating chemotherapies. ⁹ A study by Howards et al. using the Furthering the Understanding of Cancer, Health, and Survivorship in Adult study had equivalent findings: cancer survivors had similar utilization rates of ART services when compared to age-matched controls. ²⁸ This study excluded women who were not Black or White in race and those who were previously sterilized, but incorporated variables on treatment history and barriers. Unlike our study, it did not analyze CCSs separately or assess the temporal relationship between diagnosis and pregnancy. These results, however, are corroborative and complementary, strengthening the evidence of a mismatch between infertility risk and infertility treatment in the U.S. population of female cancer survivors. Reasons for this “mismatch” should be further explored so they can be addressed in the future.

Limitations of this data include the recall bias intrinsic to any patient-based information source, as well as the nonspecific nature of NSFG questions, which were not written or validated for survivors of cancer. Thus, we are unable to state rates of fertility or success with ART for any specific cancer survivor populations. This cohort includes up to 26% skin cancer survivors, for example, who may be less subject to gonadotoxic therapy or counseling typically received by cancer survivors of other types. A larger sample size may only have been possible through a study targeted specifically to cancer survivors. There are limitations in the ability to assess temporal relationships between cancer treatment and fertility treatments, prevalence of cancer types different from the general population, as well as the specific cancer treatment regimens and their gonadotoxicity level. An additional dataset from 2022 to 2023 was published by the NSFG after this analysis; a subsequent study comparing older versus newer data may provide further insight.

Conclusions

In conclusion, our population-based study of the NSFG data finds that overall rates of gravidity are similar in cancer survivors when compared to non-survivors. Gravid CCSs have a disproportionately higher desire for fertility when compared to controls. These results underscore the importance of candid yet hopeful counseling for all reproductive-age women undergoing cancer treatment. Moreover, the population of CCSs proves to be especially susceptible to unfulfilled fertility desires. It is imperative that these girls and women are offered the full spectrum of fertility-preserving measures and monitoring as they endure their diagnoses and progress through survivorship.

Authors' Contributions

A.K.: Conceptualization, data curation, formal analysis, methodology, visualization, writing (original draft and revisions). C.V.: Visualization, writing (original draft and revisions). M.A.S.: Conceptualization, data curation, writing—original draft. J.M.: Conceptualization, supervision, writing (original draft and revisions). V.G.-L.: Conceptualization, supervision, writing (original draft and revisions).