Skip to main content
NCBI home page
Human Reproduction (Oxford, England) logoLink to Human Reproduction (Oxford, England)
. 2021 May 21;36(8):2331–2338. doi: 10.1093/humrep/deab107

Trends in time-to-pregnancy in the USA: 2002 to 2017

Michael L Eisenberg 1,✉, Marie E Thoma 2, Shufeng Li 1, Alexander C McLain 3
PMCID: PMC8289323  PMID: 34021350

Abstract

STUDY QUESTION

Has there been there a temporal change in time-to-pregnancy (TTP) in the USA.

SUMMARY ANSWER

Overall, TTP was stable over time, but a longer TTP for women over 30 and parous women was identified.

WHAT IS KNOWN ALREADY

Fertility rates in the USA have declined over the past several years. Although these trends have been attributed to changing reproductive intentions, it is unclear whether declining fecundity (the biologic ability to reproduce measured by TTP in the current report) may also play a role. Indeed, trends based on declining sperm quality and higher utilisation of infertility treatment suggest fecundity may be falling.

STUDY DESIGN, SIZE, DURATION

This cross-sectional survey data from the National Survey of Family Growth was administered from 2002 to 2017. The surveys are based on nationally representative samples of reproductive-aged women in the USA. Interviews were conducted in person or through computer-assisted self-administration of sensitive questions.

PARTICIPANTS/MATERIALS, SETTING, METHODS

The study included women who self-reported time spent trying to become pregnant allowing utilisation of the current duration approach to estimate the total duration of pregnancy attempt (i.e. TTP). In all, 1202 participants were analysed over each study period. To estimate a TTP distribution overall and by parity, we used a piecewise constant proportional hazards model that accounts for digit preference. Accelerated-failure-time regression models, which were weighted to account for the sampling design, were used to estimate time ratios (TRs). Models were adjusted for age, BMI, race, education, relationship status, parity, pelvic inflammatory disease treatment and any reproductive problems.

MAIN RESULTS AND THE ROLE OF CHANCE

Of the participants analysed, the average age was 31.8 and BMI was 28.6, which was similar across the survey periods. Relationship status was the only demographic characteristic that changed over time. All other variables remained constant across the study periods. Overall, TRs comparing TTP between 2002 and 2017 increased slightly (TR: 1.02, 95% CI: 0.99, 1.04). When stratified by parity, parous women had a longer TTP over the later years of the study (TR: 1.04, 95% CI: 1.01, 1.06). TTP remained constant for nulliparous women. Similarly, TTP also increased over time for women over age thirty (TR: 1.02, 1.00, 1.05) but not for women under age thirty.

LIMITATIONS, REASONS FOR CAUTION

Small changes in data collection over time may have impacted the findings. We accounted for this in sensitivity analyses using imputed data. Overall, TRs were slightly attenuated using the imputed data, but represented similar patterns to the original data. Results for parous women and women over 30 remained consistent in the sensitivity analyses.

WIDER IMPLICATIONS OF THE FINDINGS

Consistent with reports of falling fertility rates and sperm counts, this study suggests parous and older couples in the USA may be taking longer to become pregnant. Although trends were suggestive of a small overall increase in TTP, particularly for parous women and women over age thirty, additional data are needed to attempt to understand these trends given the societal, economic and public health implications related to fecundity.

STUDY FUNDING/COMPETING INTEREST(S)

Funding was provided by National Institutes of Health grant R03HD097287 to A.C.M. There are no competing interests.

TRIAL REGISTRATION NUMBER

N/A.

Keywords: fertility, infertility, pregnancy, parity, time-to-pregnancy

Introduction

There has been some evidence that fecundity, defined as the biological capacity to reproduce regardless of intention, is changing over time. Between 2017 and 2019, fertility rates (i.e. average number of children born to a woman over her reproductive life) declined by 3% in the USA with similar trends across the most common races/ethnicities (Martin et al., 2019, 2020). Furthermore, fertility rates have declined by approximately 18% in the past three decades (Martin et al., 2013, 2020). While some of the decline has been attributed to behavioural factors (e.g. contraceptive use, delayed childbearing) and changing childbearing intentions (Mathews and Hamilton, 2002; lton, 2016; Khandwala et al., 2017), biological factors affecting fecundity or the biological capacity to reproduce, have also been suggested. For example, a recent meta-analysis reported that sperm counts have declined by 50% over the past 40 years in Western countries (Levine et al., 2017).

With declines in fertility rates and semen quality combined with increasing utilisation of ARTs in Western countries (Sunderam et al., 2018; Martin et al., 2019), there is increasing concern that infertility is becoming more prevalent. While no biomarker of fecundity exists, investigators have used time-to-pregnancy (TTP) as a proxy at the population level, whereby shorter times represent more fecund couples (Joffe, 2000; Smarr et al., 2017). A number of study designs have been applied to assess TTP, each with their strengths and limitations. (Weinberg et al., 1993, 1994; Slama et al., 2006). The most common approach for population-based estimates of TTP is a retrospective pregnancy-based assessment in which couples who have had a pregnancy are asked about number of months of unprotected intercourse prior to pregnancy. This design often relies on long recall and excludes couples who have never been pregnant, thus overestimating fecundity (Cooney et al., 2009; Smarr et al., 2017). Prospective TTP studies include couples at the start of their pregnancy attempt and follow them to monitor the occurrence of pregnancy. This design can include nulligravid women and get detailed prospective information; however, they are costly and are often limited to couples who plan their pregnancies in advance and therefore may not be representative of the general population. A more recent approach proposed for estimating TTP, known as the current duration approach (Weinberg and Gladen, 1986; Keiding et al., 2002), overcomes some of these limitations. The current duration approach includes couples at risk of pregnancy at the time of interview and queries on the amount of time at risk of pregnancy, or trying to become pregnant. This approach overcomes some of the limitations of prior studies by including nulligravid women and couples who may not have planned to become pregnant in advance of the survey (i.e. started in the month of interview). Given its recent use in the literature, we are unaware of other studies that have applied this approach to examine trends in TTP.

Studies in other countries have not found temporal declines in fecundity, but the time period under evaluation was not current and they often relied on a retrospective pregnancy-based design (Joffe, 2000; Scheike et al., 2008). Methodological concerns of the existing studies remain, and the question of whether human fecundity is changing remains controversial and unanswered (Smarr et al., 2017). Most importantly, the implications of changes in fertility may have social, economic and public health consequences, such as a declining work force and tax base or may serve as an indicator of broader societal health (Smarr et al., 2017). The current study assessed trends in TTP estimated from self-reported current durations of pregnancy attempts in the USA between 2002 and 2017.

Materials and methods

We analysed data from the National Survey of Family Growth (NSFG), a multistage probability sample designed to be nationally representative of the population of USA women aged 15–44. The NSFG, which began in 1973 as a periodic survey, gathers information on family life, marriage and divorce, pregnancy, infertility, use of contraception and men’s and women’s health. In 2006, NSFG changed to a continuous sample design (i.e. fieldwork takes place continually over 48 weeks each year) (Lepkowski et al., 2010). All subjects provided consent for participation. The survey was available in both English and Spanish. The current analysis was categorised as exempt from approval by the Stanford Institutional Review Board.

We examined cycle 6 (2002) and compared it to survey years that employed a continuous sample design across time intervals (2006–2010, 2011–2013, 2013–2015, 2015–2017). Overall response rates were 80% in 2002, 78% in 2006–2010, 73% in 2011–2013, 71% in 2013–2015 and 67% in 2015–2017. Descriptive characteristics were compared across survey years and differences across study periods were assessed by chi-square tests. Women who were sexually active and not pregnant and who reported not using birth control were asked, ‘Is the reason you are not using a method of birth control now because you, yourself, want to become pregnant as soon as possible?’ If yes, they were asked, ‘How long have you been trying to become pregnant?’ The women’s self-reported time trying to become pregnant at the time of interview, referred to as their current duration, are used to provide inference on the total duration of pregnancy attempt (i.e. TTP) via the current duration approach (Keiding et al., 2002; Thoma et al., 2013). In the current duration approach, women are included before they become pregnant (i.e. right censored data) and women with longer pregnancy attempts are likely to be over-represented in the sample relative to the population of attempting women (i.e. length bias). To deal with these features, the current duration approach applies survival methods that rely on the theory of backwards recurrence time (Yamaguchi, 2003), which allows for the estimation of the unobserved total trying time from the observed current duration, or time trying to become pregnant at interview. The current duration approach requires that the beginning of pregnancy attempts is generated at a constant rate and, given the covariates, the distribution of TTP is independent of calendar time. Note the latter assumption is given the covariate data, which can include the NSFG cycle or the year of interview, so it does not preclude us from assessing trends in TTP over time. It should be noted that the current duration values should not be interpreted as TTP values before estimation. It is only after the application of the survival framework that the survival curves and time ratios (TRs) from the regression models can be interpreted as a TTP-like distribution. Earlier work demonstrates that 10.5% of the NSFG sample is at risk for pregnancy and 3.8% is both at risk and currently trying for pregnancy (Thoma et al., 2013).

To estimate a TTP, we used a piecewise constant proportional hazards model that accounts for digit preference (McLain et al., 2014). TRs were estimated by accelerated-failure-time (AFT) regression models with a generalised gamma distribution and weighted to account for the NSFG sampling design, similar to previous studies (Louis et al., 2013; Thoma et al., 2013). The TR can be interpreted as a comparison between the rates of change over time. For example, for categorical predictors (e.g. survey years), a TR of 1.5 would mean that the outcome occurs 50% faster in survey B compared to survey A (referent). For a continuous variable (e.g. years), a TR of 1.5 would mean that the outcome occurs 50% faster for each additional year. Models were adjusted for covariates at the time pregnancy is being attempted: age categories (20–24, 25–29, 30–34, 35–39, 40–44), BMI categories (underweight: <18, ideal weight: 18–24.9, overweight: 25–29.9, obese: 30–34.9, severely obese: ≥35), race/ethnicity (Hispanic, non-Hispanic White, non-Hispanic Black, non-Hispanic other), years of education categories (0–11, 12, 13–15, ≥16), relationship status (married, living with partner, not living with partner), parity (parous, nulliparous), pelvic inflammatory disease (PID) treatment (yes/no) and any reproductive problems (yes/no), which have been associated with fertility in NSFG (Chandra et al., 2013; Martinez et al., 2018). Models for pregnancy history were not additionally adjusted for parity. Adjusted models were additionally stratified by pregnancy history (never pregnant, pregnant without live birth, pregnant with live birth) and age at interview.

Sensitivity analyses were conducted to account for changes in coding instructions over the different study periods. These changes resulted in differences in the distribution of 0 and 1 current duration values coded over time whereby women trying less than one month were instructed to be coded as 1 (2002–2011) or 0 (2013–2017). To assess the impact of this on our findings, we imputed the 0 and 1 values based on the distribution of these values in the most recent data years (2015–2017). This was repeated 50 times, the AFT regression models were estimated for each imputed dataset, and summarised using standard imputation methods (Rubin, 2004). Additionally, we repeated analyses using the 2006–2010 survey as the reference group given the change after the 2002 survey to the continuous sampling design and exclusion of women who reported being non-surgically sterile from being asked questions on trying for pregnancy in 2006 and subsequent survey years. Finally, we examined a continuous measure of month of interview (rather than survey year) to account for the heterogeneity in timing of interview within survey periods. Analyses were performed using SAS, Stata (v16) or R software (version 3.6 R project for statistical computing), depending on the analysis.

Results

Of the 1202 participants analysed, the average age was 31.8 and BMI was 28.6. Age, BMI, education, income, racial/ethnic composition, parity, percent poverty level and history of PID treatment, gynaecologic disorders and infertility treatment showed generally similar distributions across survey periods, with the exception of relationship status (Table I), which showed increases in cohabitation over time.

Table I.

Characteristics of cohort stratified by National Survey for Family Growth time period.

2002 2006 2011 2013 2015 Total P a
Total N (unweighted) 7643 12 279 5601 5699 5554 36 776
Total N (weighted) 61 560 715 61 754 741 60 887 363 61 491 766 72 218 086 317 912 671
N Current duration sample 263 360 184 206 189 1216
Age (years) Mean age 32.1 (0.4) 31.4 (0.4) 31.3 (0.9) 33.1 (0.7) 31.0 (0.5) 31.8 (0.3) 0.12
20–24 11.4% 13.3% 17.1% 9.0% 16.2% 13.3%
25–29 22.3% 30.2% 25.3% 23.0% 23.9% 24.8%
30–34 29.8% 22.8% 22.2% 24.7% 26.5% 25.1% 0.31
35–39 23.1% 20.9% 24.9% 23.8% 27.1% 24.1%
40–44 13.4% 12.8% 10.4% 19.6% 6.3% 12.7%
BMI (kg/m2) Mean 27.0 (0.3) 29.9 (1.1) 28.1 (0.9) 28.8 (1.1) 29.0 (0.7) 28.6 (0.4) 0.003
<18 0.1% 1.3% 0.1% 0.5% 0.0% 0.4%
18–24.9 45.0% 36.7% 41.3% 40.5% 35.4% 39.7%
25–30 24.9% 23.2% 27.5% 23.6% 22.3% 24.3% 0.60
30–35 12.2% 15.9% 11.4% 16.7% 18.2% 15.0%
35+ 16.2% 22.9% 19.8% 18.7% 24.1% 20.3%
Race/ethnicity Hispanic 18.8% 15.8% 21.7% 17.9% 23.7% 19.7% 0.51
NH White 64.9% 62.0% 61.7% 58.5% 59.0% 61.0%
NH Black 11.0% 13.8% 8.7% 9.5% 10.2% 10.5%
NH other 5.3% 8.3% 7.9% 14.1% 7.2% 8.9%
Education (years) 0–11 17.3% 16.2% 17.9% 7.4% 12.7% 13.8%
12 21.7% 16.6% 18.5% 20.5% 23.5% 20.3%
13–15 24.1% 25.3% 25.5% 25.7% 25.7% 25.3% 0.62
16+ 36.9% 42.0% 38.1% 46.4% 38.1% 40.6%
Relationship status Married 82.3% 78.5% 63.0% 70.0% 70.6% 72.3%
Living with partner 10.7% 14.2% 28.8% 23.1% 19.6% 19.8% 0.02
Not living with partner 7.1% 7.3% 8.3% 6.9% 9.8% 7.9%
Reproductive history Never pregnant 30.8% 33.5% 37.2% 35.2% 34.5% 34.4%
Pregnant/no live births 18.2% 20.9% 14.3% 13.5% 11.3% 15.3% 0.67
Pregnant/live births 51.1% 45.5% 48.5% 51.3% 54.2% 50.3%
Pelvic inflammatory treatment No 95.0% 98.1% 98.7% 97.2% 95.6% 96.9% 0.47
Gynaecologic disordersb No 59.9% 61.8% 70.4% 72.1% 65.3% 66.5% 0.25
% of poverty levelc >300% 53.8% 57.2% 48.7% 51.6% 44.3% 50.9% 0.58
150–299% 22.0% 20.7% 21.5% 22.4% 30.9% 23.7%
<150% 24.2% 22.0% 29.8% 26.0% 24.7% 25.5%
Treatment for fertilityd Never 72.5% 74.9% 84.5% 77.2% 77.3% 77.5% 0.46
Current 10.5% 8.6% 6.9% 10.6% 11.0% 9.6%
Ever 17.1% 16.4% 8.6% 12.2% 11.8% 13.0%
Open in a new tab

NH, non-Hispanic.

Cells represent percentages except for mean, standard deviation when listed.

a

Based on a chi-square test for categorical variables or ANOVA for continuous variables.

b

Ever received a diagnosis of ovarian cysts, fibroids, endometriosis, or ovulation/menstruation problems.

c

Family income relative to the 2001 poverty levels defined by the U.S. Census Bureau.

d

Based on self-reported current or ever use of medical help for pregnancy and type of services sought.

Based on unadjusted analyses, temporal increases in TTP were minimal (TR: 1.02, 95% CI 0.99–1.04, P-value for trend test = 0.12, Table II). While no temporal trend was seen for women who were never pregnant (TR: 1.00, 95% CI 0.97–1.03) or pregnant without a live birth (TR 1.02, 95% CI 0.98, 1.02), parous women reported a significantly longer time trying to conceive over the study period (TR: 1.04, 95% CI 1.01–1.06). These trends remained after adjustment for covariates in the model. TRs were also estimated for each survey period compared to 2002 and showed similar patterns to the linear trend (per year) estimates overall and by parity. In general, parous women from later years reported longer times trying to conceive (Fig. 1). TTP remained longer for study periods on or after 2011 after adjustment among parous women using the 2002 cycle as a referent (TR: 1.59, 95% CI: 1.09, 2.33 for 2011–2013; TR: 1.63; 95% CI: 1.06, 2.50 for 2013–2015; TR: 1.56, 95% CI: 1.07, 2.27 for 2015–2017; Fig. 1). Similarly, for women over the age of 30, compared to the 2002 cycle, TRs were longer in later survey years (TR: 1.42, 95% CI: 1.01, 1.99 for 2006; TR: 1.29; 95% CI: 0.88, 1.98 for 2011–2013; TR 1.69, 95% CI 1.16, 2.45 for 2013–2015; TR: 1.35, 95% CI: 0.92, 1.98 for 2015–2017). Overall, women over 30 also reported longer time trying to conceive over the study period (TR: 1.02, 95% CI 1.00, 1.05). In contrast, no trend was identified for women under 30 (TR: 1.01, 95% CI 0.98, 1.03) (Table II, Fig. 2). Using the 2006–2010 cycle as a referent showed similar temporal trends with some attenuation (Supplementary Table SI) given the different reference period.

Table II.

Unadjusted and adjusted1 trends in time-to-pregnancy overall and by parity, 2002–2017.

2002 2006–2010 2011–2013 2013–2015 2015–2017 Linear trend P-value2
Overall Unadjusted Ref 1.06 (0.83, 1.37) 1.23 (0.89, 1.70) 1.35 (1.02, 1.78) 1.11 (0.83, 1.49) 1.02 (0.99, 1.04) 0.12
Adjusted1 Ref 1.15 (0.91, 1.47) 1.24 (0.94, 1.63) 1.26 (0.96, 1.64) 1.19 (0.91, 1.55) 1.01 (0.99, 1.03) 0.12
Stratification by parity/gravidity
 Never pregnant Unadjusted Ref 0.66 (0.45, 0.99) 0.69 (0.43, 1.11) 0.99 (0.67, 1.45) 0.89 (0.58, 1.35) 1.01 (0.98, 1.03) 0.70
Adjusted1 Ref 0.79 (0.52, 1.20) 0.73 (0.46, 1.18) 1.08 (0.71, 1.65) 0.92 (0.57, 1.49) 1.00 (0.97, 1.03) 1.00
 Pregnant without live births Unadjusted Ref 1.28 (0.75, 2.17) 1.56 (0.75, 3.24) 0.83 (0.45, 1.53) 0.87 (0.42, 1.79) 0.98 (0.94, 1.03) 0.40
Adjusted1 Ref 1.29 (0.82, 2.02) 1.69 (0.88, 3.23) 1.21 (0.73, 1.98) 1.24 (0.68, 2.29) 1.02 (0.98, 1.06) 0.25
 Pregnant with live births Unadjusted Ref 1.22 (0.88, 1.70) 1.64 (1.11, 2.42) 1.72 (1.16, 2.56) 1.27 (0.83, 1.93) 1.03 (1.00, 1.06) 0.04
Adjusted1 Ref 1.34 (0.94, 1.93) 1.59 (1.09, 2.33) 1.63 (1.06, 2.50) 1.56 (1.07, 2.27) 1.04 (1.01, 1.06) 0.01
Stratification by age (in years)
 Age ≤ 30 Unadjusted Ref 0.94 (0.69, 1.29) 1.17 (0.79, 1.75) 1.12 (0.80, 1.58) 1.05 (0.73, 1.50) 1.01 (0.99, 1.03) 0.42
Adjusted1 Ref 0.88 (0.64, 1.21) 1.10 (0.75, 1.61) 1.02 (0.71, 1.46) 0.99 (0.71, 1.40) 1.01 (0.98, 1.03) 0.56
 Age > 30 Unadjusted Ref 1.27 (0.89, 1.82) 1.27 (0.86, 1.88) 1.50 (1.02, 2.21) 1.19 (0.80, 1.77) 1.02 (0.99, 1.05) 0.17
Adjusted1 Ref 1.42 (1.01, 1.99) 1.29 (0.88, 1.89) 1.69 (1.16, 2.45) 1.35 (0.92, 1.98) 1.02 (1.00, 1.05) 0.04
Open in a new tab
1

Adjusted for age in years (20–24, 25–29, 30–34, 35–39, 40–44), BMI (kg/m2) (underweight: <18, ideal weight: 18–24.9, overweight: 25–29.9, obese: 30–34.9, severely obese: ≥35), race/ethnicity (Hispanic, non-Hispanic White, non-Hispanic Black, non-Hispanic other), years of education (0–11, 12, 13–15, ≥16), relationship status (married, living with partner, not living with partner), pelvic inflammatory disease (PID) treatment (yes/no) and any gynaecologic disorders (yes/no for ever diagnosed with ovarian cysts, fibroids, endometriosis or ovulation/menstruation problems). For stratified models, adjustment was the same with the exception of the stratified variable.

2

Test of linear trend.

Figure 1.

Figure 1.

Open in a new tab

Estimation of the proportion of women not yet pregnant as a function of the number of months trying for pregnancy among (A) nulliparous women and (B) parous who reported currently trying and their duration of pregnancy attempt.

Figure 2.

Figure 2.

Open in a new tab

Estimation of the proportion of women not yet pregnant as a function of the number of months trying for pregnancy among (A) women aged ≤30 and (B) women aged >30 who reported currently trying and their duration of pregnancy attempt.

Unadjusted TTP trends were re-run using imputed data in sensitivity analyses (Supplementary Table SII). In general, patterns were consistent with the primary analysis with some attenuation (Overall, TR: 1.02, 95% CI: 0.99, 1.04; Parous, TR: 1.03, 95% CI: 1.00, 1.06; Nulliparous, TR: 1.00, 95% CI: 0.97, 1.04). Finally, linear trend patterns were also consistent when fitting the models using interview month (coded as number of months since January 1960/2012) rather than survey period, with a slight attenuation in unadjusted point estimates for overall (TR: 1.01, 95% CI: 1.00, 1.03) and parous women (TR: 1.03, 95% CI: 1.00, 1.05).

Discussion

The fertility rate in the USA (births per women) has declined in the USA over the past century (Hamilton et al., 2015). While changing reproductive behaviours (e.g. contraception, delayed childbearing) certainly contribute, the role of biological capacity to reproduce (i.e. fecundity) remains unknown in the decline of the fertility rate. The current data suggest temporal trends in TTP were generally stable with a small increase over time; however, women who have had a prior birth or were over 30 years of age showed a decline in fecundity (i.e. took longer to conceive) over the past 15 years in the USA. These trends held after adjustment for demographic and health characteristics, many of which did not change over time, suggesting other factors may be contributing to these patterns. In addition, these patterns were consistent with sensitivity analyses accounting for small changes in survey design and heterogeneity in time of interview within surveys.

This is the first study to examine trends in TTP in the USA using a current duration approach. Our findings contrast with TTP trends from prior studies in European countries comparing earlier study periods, which demonstrated an increase in fecundity or no change in TTP in earlier time periods. These previous studies employed a retrospective pregnancy-based TTP design, which asks couples who have been pregnant or had a live birth how long they had been trying to become pregnant. Thus, our findings for parous women may be a more comparable group to evaluate in relation to these other studies. A Swedish study examined data from over 800 000 primiparous pregnant women between 1983 and 2002 and identified a decline in infertility based on reported TTP (Scheike et al., 2008). Another retrospective study of over 3000 Danish twins born between 1931 and 1952 also identified an increase in female fecundity and a decline in severe infertility in men (Jensen et al., 2005). A retrospective survey in Britain identified an increase in fertility (i.e. shorter TTP) over time from the 1960s to the 1990s (Joffe, 2000). A Chinese study, which examined the number of married couples with children as well as retrospectively reported TTP, also identified an increase in fecundity from 1980 to 2003 (Tian et al., 2016). In contrast, one study by Joffe et al. using data from five Western European countries found relative stable patterns of TTP between 1953 and 1993 (Joffe et al., 2013). In addition to study design and geographic differences, the time period of our study is more current than the prior literature. In the USA, other studies using a different measure from NSFG data found declines in infertility over time, contrasting with our findings, but no change in impaired fecundity (Chandra et al., 2013). In this context, a couple is defined as infertile at the time of interview if, during the previous 12 months or longer, they were continuously married or cohabiting, were sexually active each month, had not used contraception, and had not become pregnant. Impaired fecundity was defined by a women classified as being nonsurgically sterile, reporting difficulty becoming pregnant, or other responses suggesting a long interval without conception. However, prior trends based on this definition of infertility have been questioned (Barnhart, 2006; Guzick and Swan, 2006; Olive and Pritts, 2006; Thornton and Goldman, 2006).

In contrast to prior studies, the advantage of a current duration approach is that it provides current, rather than retrospective, estimates of fecundity and includes all women in the sample, regardless of prior pregnancy status. Specifically, respondents report how long they have currently been trying to become pregnant rather than recall how long it took them to conceive prior pregnancies, which may have occurred over long intervals. Moreover, the current duration sample is able to include infertile couples who will never conceive in its estimation. In contrast, other methodologies that only include pregnant women will underestimate TTP as those who cannot conceive will be excluded. In addition, the current duration approach relaxes some assumptions of prior methodology (e.g. assuming all women who are not at risk of pregnancy are fecund) and includes only women at risk pregnancy at the time of survey and then accounts for length-biased sampling in the analysis. This method has been shown to provide reliable estimates of TTP using NSFG data (Louis et al., 2013; Thoma et al., 2013; Kasman et al., 2018); however, it may be more accurate for short rather than longer time periods of recall.

While the annual TR trend for parous women was modest, compounded over longer periods and extrapolated to the entire population, the longer time trying to conceive becomes substantial. For example, the estimated TR for parous women over 5, 10 and 15 years changes are 1.21 (95% CI 1.05–1.34), 1.48 (95% CI 1.10–1.79) and 1.80 (95% CI 1.16–2.40), respectively. Estimating the average TTP is challenging due to a number of factors (Smarr et al., 2017). However, conservatively estimating that the average TTP for parous women in 2002 is 3 months (1 month shorter than the median found in the Longitudinal Investigation of Fertility and the Environment (LIFE) Study (Buck Louis et al., 2012)), the compounded growth from 2002 to 2017 would increase the average TTP to 5.4 months (95% CI 3.5–7.2 months). Given that 62% of the 3.79 million births in 2018 (i.e. the last year for which US birth data is available) were to parous women (Martin et al., 2019) and approximately 55% of pregnancies are intended (Finer and Zolna, 2016), parous women in the USA spend over 3.1 million (95% CI 0.61–5.4 million) extra months trying to conceive compared to 15 years ago.

There are limitations that warrant mention. First, the data rely on participant reporting of intentions and of time trying to become pregnant, which could be subject to bias. However, unlike retrospective TTP designs, the recall in current duration data is over their current attempt versus recalling the TTP of a prior pregnancy. Prior studies have validated recall of TTP and have documented relatively good validity for shorter periods of recall (Zielhuis et al., 1992; Joffe et al., 1995). The analysis is limited to women who reported they were actively trying for pregnancy, which may not represent all those at risk for pregnancy. However, it does represent a particular group who may be having regular unprotected intercourse to evaluate TTP, which is consistent with populations evaluated in prospective TTP study designs. In addition, reporting of TTP was based on female responses, despite TTP being a couple-dependent measure; however, couples were not queried in the NSFG survey. Further, intentions, or respondents identifying as ‘trying’ to become pregnant, may differ across demographic groups (Greil et al., 2010). In addition, certain populations are not represented in the survey, such as homeless individuals not in shelters or incarcerated individuals. As noted earlier, survey questions changed slightly between 2002 and later years. Specifically, for the 2002 cycle, women who said they were not surgically sterile, but were reported to be unable to physically have children, could be included in the current duration sample. From 2006 onward, women in this category were not asked the current duration questions, making comparisons between time periods imperfect. However, we would anticipate that if the women who reported they were unable to physically have children were included in the sample, it would have resulted in a longer TTP over later survey periods than what we found in our study. Thus, our findings on trends may be somewhat attenuated. The current duration approach also relies on modelling assumptions, such as a nonvarying rate of couples starting their pregnancy attempt (i.e. stationarity) and independence of the TTP distribution over calendar time given the covariates. We tested the robustness of our findings to the latter assumption using the month of interview analysis which had similar results. Finally, these trends may be influenced by greater utilisation of infertility treatment, or varying types of fertility treatment (i.e. use of ovulation kits, adoption of more effective methods, change in guidelines, testing practices) over time. Indeed, the percentage of women reporting infertility treatment increased with time. We did not control for infertility treatment in the analysis, because this could potentially bias the findings. Methods to handle infertility treatment as a competing risk are more appropriate, but are still being developed (Duron et al., 2013). Future studies examining trends employing different methods to handle infertility treatment are warranted.

Nevertheless, the current study identified stable trends in TTP over the last 15 years overall but longer TTP among parous women and women over 30. Given the differences, additional data are needed to attempt to understand these trends given the economic, social and public health implications of fecundity.

Supplementary data

Supplementary data are available at Human Reproduction online.

Data availability

All data is available at https://www.cdc.gov/nchs/nsfg/index.htm.

Authors’ roles

Study design: M.L.E., M.E.T., A.C.M.; Data analysis: S.L., M.E.T., A.C.M.; Manuscript drafting: M.L.E., M.E.T., A.C.M.; Critical revision of manuscript: M.L.E., M.E.T., S.L., A.C.M.

Funding

Funding was provided by National Institutes of Health grant R03HD097287 to A.C.M.

Conflict of interest

The authors report no conflicts of interest.

Supplementary Material

deab107_Supplementary_Data
Click here for additional data file. (24.3KB, zip)

References

  1. Barnhart KT. The challenge and enjoyment of the interpretation of epidemiologic data. Fertil Steril 2006;86:527–528; discussion 534. [DOI] [PubMed] [Google Scholar]
  2. Buck Louis GM, Sundaram R, Schisterman EF, Sweeney AM, Lynch CD, Gore-Langton RE, Chen Z, Kim S, Caldwell KL, Barr DB.. Heavy metals and couple fecundity, the LIFE Study. Chemosphere 2012;87:1201–1207. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Chandra A, Copen CE, Stephen EH.. Infertility and impaired fecundity in the United States, 1982-2010: data from the National Survey of Family Growth. Natl Health Stat Rep 2013;67:1–18, 1 p following 19. [PubMed] [Google Scholar]
  4. Cooney MA, Buck Louis GM, Sundaram R, McGuiness BM, Lynch CD.. Validity of self-reported time to pregnancy. Epidemiology 2009;20:56–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Duron S, Slama R, Ducot B, Bohet A, Sorensen DN, Keiding N, Moreau C, Bouyer J.. Cumulative incidence rate of medical consultation for fecundity problems–analysis of a prevalent cohort using competing risks. Hum Reprod 2013;28:2872–2879. [DOI] [PubMed] [Google Scholar]
  6. Finer LB, Zolna MR.. Declines in unintended pregnancy in the United States, 2008-2011. N Engl J Med 2016;374:843–852. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Greil AL, McQuillan J, Johnson K, Slauson-Blevins K, Shreffler KM.. The hidden infertile: infertile women without pregnancy intent in the United States. Fertil Steril 2010;93:2080–2083. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Guzick DS, Swan S.. The decline of infertility: apparent or real? Fertil Steril 2006;86:524–526; discussion 534. [DOI] [PubMed] [Google Scholar]
  9. Hamilton BE, Martin JA, Osterman MJ, Curtin SC, Matthews TJ.. Births: final data for 2014. Natl Vital Stat Rep 2015;64:1–64. [PubMed] [Google Scholar]
  10. Jensen TK, Joffe M, Scheike T, Skytthe A, Gaist D, Christensen K.. Time trends in waiting time to pregnancy among Danish twins. Hum Reprod 2005;20:955–964. [DOI] [PubMed] [Google Scholar]
  11. Joffe M, Holmes J, Jensen TK, Keiding N, Best N.. Time trends in biological fertility in Western Europe. Am J Epidemiol 2013;178:722–730. [DOI] [PubMed] [Google Scholar]
  12. Joffe M, Villard L, Li Z, Plowman R, Vessey M.. A time to pregnancy questionnaire designed for long term recall: validity in Oxford, England. J Epidemiol Community Health 1995;49:314–319. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Joffe M. Time trends in biological fertility in Britain. Lancet 2000;355:1961–1965. [DOI] [PubMed] [Google Scholar]
  14. Kasman AM, Thoma ME, McLain AC, Eisenberg ML.. Association between use of marijuana and time to pregnancy in men and women: findings from the National Survey of Family Growth. Fertil Steril 2018;109:866–871. [DOI] [PubMed] [Google Scholar]
  15. Keiding N, Kvist K, Hartvig H, Tvede M, Juul S.. Estimating time to pregnancy from current durations in a cross-sectional sample. Biostatistics 2002;3:565–578. [DOI] [PubMed] [Google Scholar]
  16. Khandwala YS, Zhang CA, Lu Y, Eisenberg ML.. The age of fathers in the USA is rising: an analysis of 168 867 480 births from 1972 to 2015. Hum Reprod 2017;32:2110–2116. [DOI] [PubMed] [Google Scholar]
  17. Lepkowski JM, Mosher WD, Davis KE, Groves RM, Van Hoewyk J.. The 2006-2010 National Survey of Family Growth: sample design and analysis of a continuous survey. Vital Health Stat 2 2010;150:1–36. [PubMed] [Google Scholar]
  18. Levine H, Jorgensen N, Martino-Andrade A, Mendiola J, Weksler-Derri D, Mindlis I, Pinotti R, Swan SH.. Temporal trends in sperm count: a systematic review and meta-regression analysis. Hum Reprod Update 2017;23:646–659. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Louis JF, Thoma ME, Sorensen DN, McLain AC, King RB, Sundaram R, Keiding N, Buck Louis GM.. The prevalence of couple infertility in the United States from a male perspective: evidence from a nationally representative sample. Andrology 2013;1:741–748. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Martin JA, Hamilton BE, Osterman MJK.. Births in the United States, 2018. NCHS Data Brief 2019;346:1–8. [PubMed] [Google Scholar]
  21. Martin JA, Hamilton BE, Osterman MJ, Curtin SC, Matthews TJ.. Births: final data for 2012. Natl Vital Stat Rep 2013;62:1–68. [PubMed] [Google Scholar]
  22. Martin JA, Hamilton BE, Osterman MJK, Driscoll AK.. Births: final data for 2018. Natl Vital Stat Rep 2019;68:1–47. [PubMed] [Google Scholar]
  23. Martin JA, Hamilton BE, Osterman MJK.. Births in the United States, 2019. NCHS Data Brief 2020;387:1–8. [PubMed] [Google Scholar]
  24. Martinez GM, Daniels K, Febo-Vazquez I.. Fertility of men and women aged 15-44 in the United States: National Survey of Family Growth, 2011-2015. Natl Health Stat Rep 2018;113:1–17. [PubMed] [Google Scholar]
  25. Mathews TJ, Hamilton BE.. Mean age of mother, 1970-2000. Natl Vital Stat Rep 2002;51:1–13. [PubMed] [Google Scholar]
  26. Mathews TJ, Hamilton BE.. Mean age of mothers is on the rise: United States, 2000-2014. NCHS Data Brief 2016;232:1–8. [PubMed] [Google Scholar]
  27. McLain AC, Sundaram R, Thoma M, Buck Louis GM.. Semiparametric modeling of grouped current duration data with preferential reporting. Stat Med 2014;33:3961–3972. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Olive DL, Pritts EA.. Estimating infertility: the devil is in the details. Fertil Steril 2006;86:529–530; discussion 534. [DOI] [PubMed] [Google Scholar]
  29. Rubin DB. Multiple Imputation for Nonresponse in Surveys. Hoboken, New Jersey: John Wiley & Sons; 2004. [Google Scholar]
  30. Scheike TH, Rylander L, Carstensen L, Keiding N, Jensen TK, Stromberg U, Joffe M, Akre O.. Time trends in human fecundability in Sweden. Epidemiology 2008;19:191–196. [DOI] [PubMed] [Google Scholar]
  31. Slama R, Ducot B, Carstensen L, Lorente C, de La Rochebrochard E, Leridon H, Keiding N, Bouyer J.. Feasibility of the current-duration approach to studying human fecundity. Epidemiology 2006;17:440–449. [DOI] [PubMed] [Google Scholar]
  32. Smarr MM, Sapra KJ, Gemmill A, Kahn LG, Wise LA, Lynch CD, Factor-Litvak P, Mumford SL, Skakkebaek NE, Slama R. et al. Is human fecundity changing? A discussion of research and data gaps precluding us from having an answer. Hum Reprod 2017;32:499–504. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Sunderam S, Kissin DM, Crawford SB, Folger SG, Boulet SL, Warner L, Barfield WD.. Assisted reproductive technology surveillance—United States, 2015. MMWR Surveill Summ 2018;67:1–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Thoma ME, McLain AC, Louis JF, King RB, Trumble AC, Sundaram R, Buck Louis GM.. Prevalence of infertility in the United States as estimated by the current duration approach and a traditional constructed approach. Fertil Steril 2013;99:1324–1331.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Thornton KL, Goldman MB.. Impact of subgroup analysis on estimates of infertility. Fertil Steril 2006;86:531–533; discussion 534. [DOI] [PubMed] [Google Scholar]
  36. Tian X, Jiang J, Wang J, Zhang S.. Improved fecundity in Northern China: a secular trend from 1980 to 2003. PLoS One 2016;11:e0165097. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Weinberg CR, Baird DD, Rowland AS.. Pitfalls inherent in retrospective time-to-event studies: the example of time to pregnancy. Stat Med 1993;12:867–879. [DOI] [PubMed] [Google Scholar]
  38. Weinberg CR, Baird DD, Wilcox AJ.. Sources of bias in studies of time to pregnancy. Stat Med 1994;13:671–681. [DOI] [PubMed] [Google Scholar]
  39. Weinberg CR, Gladen BC.. The beta-geometric distribution applied to comparative fecundability studies. Biometrics 1986;42:547–560. [PubMed] [Google Scholar]
  40. Yamaguchi K. Accelerated failure-time mover-stayer regression models for the analysis of last-episode data. Sociol Methodol 2003;33:81–110. [Google Scholar]
  41. Zielhuis GA, Hulscher ME, Florack EI.. Validity and reliability of a questionnaire on fecundability. Int J Epidemiol 1992;21:1151–1156. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

deab107_Supplementary_Data
Click here for additional data file. (24.3KB, zip)

Data Availability Statement

All data is available at https://www.cdc.gov/nchs/nsfg/index.htm.

Articles from Human Reproduction (Oxford, England) are provided here courtesy of Oxford University Press

RESOURCES