A prospective cohort study of a woman's own gestational age and her fecundability

Abstract

STUDY QUESTION

What is the magnitude of the association between a woman's gestational age at her own birth and her fecundability (cycle-specific probability of conception)?

SUMMARY ANSWER

We found a 62% decrease in fecundability among women born <34 weeks of gestation relative to women born at 37–41 weeks of gestation, whereas there were few differences in fecundability among women born at later gestational ages.

WHAT IS KNOWN ALREADY

One study, using retrospectively collected data on time-to-pregnancy (TTP), and self-reported data on gestational age, found a prolonged TTP among women born <37 gestational weeks (preterm) and with a birthweight ≤1500 g. Other studies of women's gestational age at birth and subsequent fertility, based on data from national birth registries, have reported a reduced probability of giving birth among women born <32 weeks of gestation.

STUDY DESIGN, SIZE, DURATION

We used data from a prospective cohort study of Danish pregnancy planners (‘Snart-Gravid’), enrolled during 2007–2011 and followed until 2012. In all, 2814 women were enrolled in our study, of which 2569 had complete follow-up.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Women eligible to participate were 18–40 years old at study entry, in a relationship with a male partner, and attempting to conceive. Participants completed a baseline questionnaire and up to six follow-up questionnaires until the report of pregnancy, discontinuation of pregnancy attempts, beginning of fertility treatment, loss to follow-up or end of study observation after 12 months.

MAIN RESULTS AND THE ROLE OF CHANCE

Among women born <34 gestational weeks, the cumulative probability of conception was 12, 28 and 48% within 3, 6 and 12 cycles, respectively. Among women born at 37–41 weeks of gestation, cumulative probability of conception was 47, 67 and 84% within 3, 6 and 12 cycles, respectively. Relative to women born at 37–41 weeks' gestation, women born <34 weeks had decreased fecundability (fecundability ratio (FR) 0.38, 95% confidence interval (CI): 0.17–0.82). Our data did not suggest reduced fecundability among women born at 34–36 weeks of gestation or at ≥42 weeks of gestation (FR 1.03, 95% CI: 0.80–1.34, and FR 1.13, 95% CI: 0.96–1.33, respectively).

LIMITATIONS, REASONS FOR CAUTION

Data on gestational age, obtained from the Danish Medical Birth Registry, were more likely to be based on date of last menstrual period than early ultrasound examination, possibly leading to an overestimation of gestational age at birth. Such overestimation, however, would not explain the decrease in fecundability observed among women born <34 gestational weeks. Another limitation is that the proportion of women born before 34 weeks of gestation was low in our study population, which reduced the precision of the estimates.

WIDER IMPLICATIONS OF THE FINDINGS

By using prospective data on TTP, our study elaborates on previous reports of impaired fertility among women born preterm, suggesting that women born <34 weeks of gestation have reduced fecundability.

STUDY FUNDING/COMPETING INTEREST(S)

The study was supported by the National Institute of Child Health and Human Development (R21-050264), the Danish Medical Research Council (271-07-0338), and the Health Research Fund of Central Denmark Region (1-01-72-84-10). The authors have no competing interests to declare.

Introduction

Improvements in neonatal care during the 1980s have led to increasing numbers of preterm born infants (birth <37 weeks of gestation) surviving to reach reproductive age (Villadsen, 2008). Survivors of preterm birth may have an elevated risk of long-term adverse health outcomes, including chronic respiratory symptoms (Anand et al., 2003; Jaakkola et al., 2006; Saigal et al. 2007; Harju et al., 2014), neurodevelopmental disorders (Hack et al., 2002; Saigal et al., 2007; Moster et al., 2008), higher blood pressure (de Jong et al., 2012; Parkinson et al., 2013) and insulin resistance and diabetes (Hofman et al., 2004; Kaijser et al., 2009; Crump et al., 2011). Abbreviated gestation may also be associated with poor fertility. Infant girls born preterm have increased levels of follicle-stimulating hormone (FSH) and luteinizing hormone (LH) up to 3 months after birth, as well as delayed follicular development, compared with girls born at term. This phenomenon is thought to indicate an insufficient maturation of the hypothalamic–pituitary–ovarian axis at preterm birth (Tapanainen et al., 1981; Kuiri-Hanninen et al., 2011). Among women aged 23–37 years, reduced fertility (measured as registered births of the woman in national birth registries) has been reported for those born before 32 gestational weeks compared with women born at term (Swamy et al., 2008; deKeyser et al., 2012). A Danish cross-sectional analysis of 21 786 women who gave birth did not find prolonged time-to-pregnancy (TTP) among women born preterm compared with women born at term, except for women born preterm with a birthweight ≤1500 g. The authors suggested that the longer TTP among such women might be attributable to very preterm birth (Nohr et al., 2009). The study did not estimate fecundability (i.e. the cycle-specific probability of conception).

Reproductive history tends to recur within families, as shown for preterm birth (Swamy et al., 2008; Boyd et al., 2009; Shah et al., 2009; Bhattacharya et al., 2010), low birthweight (Shah et al., 2009), spontaneous abortion (Zhang et al., 2010; Kolte et al., 2011) and parity (Murphy and Knudsen, 2002; Goodman and Koupil, 2009). Thus, it is reasonable to hypothesize the existence of familial recurrence of decreased fecundability. With this hypothesis, reproductive outcomes of a woman's mother may be markers of the mother's fecundability, with a possible influence on the fecundability of her daughter. Thus, maternal reproductive history may confound the association between gestational age at birth and fecundability in the daughter. These factors were not controlled in previous studies.

We conducted a prospective cohort study among pregnancy planners in Denmark to examine the association between gestational age at birth and fecundability, while controlling for potential confounding by maternal reproductive history.

Subjects and Methods

Study population

Data for this study originated from a population-based prospective cohort study of Danish pregnancy planners (‘Snart-Gravid’), initiated in 2007 (Mikkelsen et al., 2009). Women eligible to participate were Danish residents, 18–40 years old at study entry, in a relationship with a male partner, attempting to conceive, and not receiving fertility treatment. Eligible participants completed a baseline questionnaire and bi-monthly follow-up questionnaires for an observational period of up to 12 months. Participants were enrolled during 2007–2011 and follow-up concluded in 2012.

There were 5512 potential participants for this study. We excluded women who provided only baseline data, had already entered the study once, had implausible or insufficient information on date of last menstrual period (LMP), had been adopted or with missing data on adoptive status, or were born after a non-singleton gestation or with missing data on multiplicity of gestation. We also excluded women born before 1 January 1978, because information about the specific gestational age at birth was not recorded in the Danish Medical Birth Registry (DMBR) until this date. The final study population comprised 2814 women (Fig. 1).

Figure 1.

Study flow chart.

Assessment of gestational age at birth

After giving consent, participants provided their Civil Personal Registration (CPR) number, a unique personal identifier assigned to all Danish citizens at birth or time of immigration, enabling linkage of persons in national health registries (Pedersen, 2011). We collected data on participants' gestational age at birth from the DMBR, which contains computerized health records of over 99% of hospital-based or home live births and stillbirths in Denmark since 1973. Data are reported to the registry by midwives attending the birth (Kristensen et al., 1996; Knudsen and Olsen, 1998). In the DMBR, gestational age at birth was recorded in full weeks (since 1978) and estimated from the woman's LMP, adjusted by results of an ultrasound examination, if performed. Use and results of ultrasound examinations were not recorded in the DMBR. To our knowledge, the earliest report on the use of prenatal ultrasound examination in Denmark considered the years 1989–1990 (Jorgensen, 1993). At that time, around 20% of pregnant women did not receive an ultrasound examination, suggesting that a non-negligible proportion of values of gestational age were determined solely by LMP during the birth years of our cohort (Jorgensen, 1993). The participants' gestational ages at birth ranged from 28 to 44 completed weeks. We defined gestational age <37 weeks as preterm, 37–41 weeks as term, and ≥42 weeks as post-term (Wilcox, 2010).

Assessment of time-to-pregnancy

The event of interest was pregnancy, regardless of outcome. At baseline, participants reported the number of months that they had already attempted to become pregnant and the date of the LMP. In each follow-up questionnaire, participants reported the date of their LMP and whether they were currently pregnant or had experienced a pregnancy termination (spontaneous abortion, therapeutic abortion or ectopic pregnancy) since the last follow-up. TTP, defined as the number of menstrual cycles at risk for pregnancy, was estimated using the following formula: (days of pregnancy attempt at study entry/days of usual cycle length) + (((LMP date from the most recent follow-up questionnaire − date of study entry)/days of usual cycle length) + 1) (Wise et al., 2010). Participants contributed cycles at risk until they reported a pregnancy, started fertility treatment, gave up pregnancy attempts, were lost to follow-up, or until the end of the observation period of 12 months, whichever came first. Women with an unknown reason for not completing the study were considered lost to follow-up and censored at the time of last follow-up questionnaire completion.

Assessment of covariates

Measures of maternal reproductive health such as history of difficulty conceiving, spontaneous abortion, preterm birth and lifetime parity were considered to be markers of the participant's mother's fecundability. Data on participant's mother's age and marital status at time of the participant's delivery, history of preterm birth, and lifetime parity were obtained from the DMBR via linkage with the participant's CPR number. Data on mother's history of preterm birth included siblings born since 1973 at a gestational age <37 completed weeks. Mother's lifetime parity was assessed by combining mother's parity recorded in the DMBR with number of children identified in the registry and using the maximum value in the analyses. From the DMBR we also obtained data on the participant's weight at birth. From the Danish National Registry of Patients (DNRP), we obtained data on mother's hospital diagnoses of hypertension (diagnosis codes 400–404 and 637.00), pre-eclampsia (637.03, 637.04, 637.09, 637.19 and 637.99) and diabetes (249, 250 and 634.74) during pregnancy with the participant. The diagnoses were coded according to the International Classification of Diseases, 8th revision. From the ‘Snart-Gravid’ baseline questionnaire we obtained data from each participant on her mother's and father's educational level, mother's smoking status during pregnancy with the participant, mother's history of difficulty conceiving and spontaneous abortion, and the following information on the participant: age at study entry, age at menarche, menstrual cycle regularity, gravidity, parity, history of ≥12 months attempting a pregnancy, and number of cycles of attempted pregnancy at the time of study entry.

Ethical approval

The ‘Snart-Gravid’ study was approved by the Danish Data Protection Board (record no. 2013-41-1922) and by the Institutional Review Board at Boston University. Consent was obtained from all participants before completion of questionnaires.

Data analysis

According to the World Health Organization, birth before 32 full gestational weeks is defined as very preterm, birth at 32–33 weeks as moderately preterm and birth at 34–36 weeks as late preterm (March of Dimes, PMNCH, Save the Children, WHO, 2012). Based on these standards and the conventional definitions of preterm birth (<37 gestational weeks), term birth (37–41 gestational weeks) and post-term birth (≥42 gestational weeks) (Wilcox, 2010), we examined the distribution of baseline characteristics according to categories of gestational age at birth (<34, 34–36, 37–41 and ≥42 weeks) among women lost to follow-up and compared it with the distribution among women with complete follow-up. We also examined the distribution of characteristics among all women included in our analyses. To examine the cumulative probability of conception by gestational age, we calculated Kaplan–Meier estimates, allowing for delayed entry and censoring at loss to follow-up, discontinuation of pregnancy attempts, initiation of fertility treatment, or reaching the end of the observation period (Hosmer et al., 2008). We examined fecundability according to the predefined categories of gestational age as well as 1-week categories of gestational age at birth (<32, each completed week 32–42, and ≥43 weeks, with 40 gestational weeks as the reference group) by calculating fecundability ratios (FR) with 95% confidence intervals (CI). FRs were calculated by proportional probabilities regression modeling, and represent ratios of cycle-specific probabilities of conception comparing exposed with unexposed women (Weinberg and Wilcox, 2008). To account for women whose pregnancy attempts started before study entry, cycles before study entry were left-truncated. Thus, a woman contributed cycles observed only after study entry, but these were corrected for the number of cycles attempting pregnancy before study entry (Wise et al., 2010).

Potential confounders were selected based on available literature of associations with gestational age at birth (Mercer et al., 1999; Shah and Bracken, 2000; Sibai et al., 2000; Kallen 2001; Ray et al., 2001; Buchmayer et al., 2004; Ekholm et al., 2005; Fadl et al., 2010; Eidem et al., 2011; Weintraub et al., 2011; deKeyser et al., 2012; Yanit et al., 2012; Messerlian et al., 2013), and their potential effect on fecundability (Weinberg et al., 1989; Ekholm et al., 2005; Ye et al., 2010; deKeyser et al., 2012). Considering that other reproductive outcomes tend to cluster in families, markers of mothers' fecundability, i.e. history of difficulty conceiving, history of spontaneous abortion, history of preterm birth, and lifetime parity, may be causally associated with daughters' fecundability. In addition, medical conditions such as hypertension, pre-eclampsia and diabetes may be associated with maternal impaired fertility (Basso et al., 2003; Trogstad et al., 2009; Whitworth et al., 2011), and thus, may influence the fecundability of the daughter. On this basis, we adjusted for participant's year of birth (continuous); mother's age (<20, 20–24, 25–29 and ≥30 years), marital status (married, unmarried or divorced/widowed), smoking status during pregnancy with the participant (yes/no), hypertension (yes/no), pre-eclampsia (yes/no), and diabetes during pregnancy with the participant (yes/no); and mother's and father's educational level (9th–10th grade or Upper Secondary School/equivalent) in Model 1. We further adjusted for mother's history of difficulty conceiving (yes/no), spontaneous abortion (yes/no), preterm birth (yes/no) and lifetime parity (1, 2–3 or ≥4) in Model 2.

We assessed the potential non-linear relation between gestational age at birth and fecundability using restricted cubic splines. Measures of gestational age that are determined from the LMP may be overestimated, compared with measures based on ultrasound examination (Tunon et al., 1996; Savitz et al., 2002). To assess the potential influence of misclassification of gestational age in the DMBR, we subtracted 1 week from each value of gestational age and repeated the analyses for 1-week categories of gestational age (<34, each gestational week 34–42, and ≥43 weeks, with 40 weeks' gestation as the reference group). Finally, in other sensitivity analyses, we restricted to women with no more than three cycles of attempted pregnancy at study entry to assess associations among participants with the highest fecundability.

The proportion of missing values ranged from 4.8 to 17.1% for the variables obtained from registries, and from 0.1 to 35.2% for variables from the self-administered questionnaires (Supplementary Table SI). On the premise that data were missing at random, we used multiple imputation by chained equations (MICE, Stata version 12.0) to impute missing values. This approach included all substantive variables used in the analyses, and generated five data sets. Because there were over 35% missing values of one variable included in the study (father's educational level), we generated forty imputed datasets and repeated the main analysis, yielding results that were close to those based on five datasets (White et al., 2011). For this reason, we considered using five imputed datasets to be sufficient for this and other analyses.

Analyses were conducted using Stata version 12.0 (StataCorp., TX, USA), and SAS version 9.2 (Cary, NC, USA).

Results

During the observation period, 245 women (8.7%) were lost to follow-up. These women were slightly younger (mean age at study start 25.7 years versus 26.6 years), but had a similar distribution of gestational age at birth as women with complete follow-up. Among women born <34 gestational weeks and lost to follow-up, a greater proportion had attempted to become pregnant for more than three cycles at study entry, and a greater proportion had a mother or a father with a maximum of 10 years of education, and a mother who was 20–24 years old, or unmarried at time of delivery of the participant. Fewer had a mother with a history of difficulty conceiving and a history of preterm birth, compared with women born <34 gestational weeks who completed the study.

Among the 2814 participants, 19 (0.7%) had been born <34 weeks, 89 (3.2%) at 34–36 weeks, 2463 (87.5%) at 37–41 weeks and 243 (8.6%) at ≥42 weeks of gestation. The proportion of women born preterm was similar to those reported in other studies of preterm birth in Scandinavia in the period, which ranged from 4.4 to 4.7% (Swamy et al., 2008; Boyd et al., 2009; deKeyser et al., 2012). Compared with women born at 37–41 weeks, women born <34 weeks of gestation were less likely to have irregular cycles, to have been pregnant or to be parous, more likely to have a history of ≥12 months attempting a pregnancy, and more likely to have attempted pregnancy for more than three cycles at study entry. They were also more likely to have a mother who was 20–24 years old at delivery, married, who smoked during pregnancy, was diagnosed with pre-eclampsia, had a history of difficulty conceiving, spontaneous abortion, or preterm birth, and a parity of at least four children (Table I).

Table I.

Characteristics of 2814 participants and their mothers according to four categories of gestational age.

Characteristic	Gestational age, weeks
	<34	34–36	37–41	≥42
No. of women, n (%)	19 (0.7)	89 (3.2)	2463 (87.5)	243 (8.6)
Mean age in years (s.e.)	25.1 (0.6)	26.6 (0.3)	26.5 (0.1)	26.3 (0.2)
Mean weight at birth in grams (s.e.)	1572 (102.5)	2476 (51.8)	3326 (9.6)	3638 (29.4)
Mean age at menarche in years (s.e.)	12.8 (0.4)	12.5 (0.1)	12.9 (0.0)	12.8 (0.1)
Irregular menstrual cycles, %	21.1	14.6	28.2	27.6
Gravidity ≥1, %	15.8	37.1	33.4	39.1
Parity ≥1, %	10.5	24.7	20.0	24.7
History of ≥12 months attempting a pregnancy, %	31.6	11.2	8.9	5.4
No. of cycles of attempted pregnancy at study entry, %
0–1	42.1	46.1	47.2	49.0
2–3	21.1	20.2	23.3	18.5
4–11	36.8	33.7	29.5	32.5
Mother's age at time of delivery, %
<20	0.0	5.6	4.1	3.3
20–24	47.4	37.1	32.5	38.7
25–29	26.3	25.8	38.9	38.3
≥30	26.3	31.5	24.5	19.8
Mother's marital status at time of delivery, %
Married	73.7	57.3	65.3	63.4
Unmarried	21.1	40.5	30.9	33.3
Divorced/widowed	5.3	2.3	3.8	3.3
Mother's education, 9th–10th grade, %	57.9	60.7	57.9	60.9
Father's education, 9th–10th grade, %	79.0	70.8	67.3	69.1
Mother smoked during pregnancy, %	52.6	49.4	34.7	26.8
Mother had hypertension, %	0.0	1.1	0.9	0.8
Mother had pre-eclampsia, %	10.5	10.1	2.0	1.2
Mother had diabetes, %	0.0	4.5	0.5	0.0
Mother's history of difficulty conceiving, %	26.3	14.6	14.6	15.2
Mother's history of spontaneous abortion, %	42.1	37.1	24.5	25.1
Mother's history of preterm birth, older sibs, %	26.3	16.9	3.7	2.1
Mother's history of preterm birth, all sibs, %	42.1	22.5	6.1	3.7
Mother's lifetime parity, %
1	5.3	14.6	10.6	8.6
2–3	68.4	73.0	76.9	78.2
≥4	26.3	12.4	12.5	13.2

s.e., standard error.

Kaplan–Meier estimates for the cumulative probability of conception were 12% (95% CI: 0–31%), 28% (95% CI: 0–50%), and 48% (95% CI: 11–69%) within 3, 6, and 12 cycles, respectively, for women born <34 weeks of gestation, and 47% (95% CI: 43–49%), 67% (95% CI: 65–70%), and 84% (95% CI: 82–85%) within 3, 6, and 12 cycles, respectively, for women born at 37–41 weeks of gestation. Crude FRs, presented in Table II, were 0.37 (95% CI: 0.17–0.81) for women born <34 weeks, 1.05 (95% CI: 0.82–1.34) for women born at 34–36 weeks and 1.11 (95% CI: 0.94–1.30) for women born at ≥42 weeks of gestation, relative to women born at 37–41 weeks' gestation. Results were similar after adjusting for year of birth and mothers' socio-demographic and medical characteristics, and when we further adjusted for the markers of mothers' reproductive health. Adjusted FRs for each completed gestational week at birth, presented in Table III, did not indicate a material association with fecundability for any category of gestational age, except for women born <34 weeks of gestation. Within the category of <34 weeks of gestation, we found similar effect estimates for women born in the three subcategories <32, 32 and 33 weeks of gestation. The smaller numbers within these subcategories gave broader confidence intervals than the combined category, and these confidence intervals individually included a wider range of parameter values. Nonetheless, the pattern of effect estimates was similar for the categories below 34 weeks of gestation, indicating that the observed effect was not limited to either subcategory. The smoothed relation between fecundability and gestational age at birth, throughout the range from 28 to 44 completed weeks, was modeled using restricted cubic splines, and is shown in Fig. 2. Using 40 weeks as the reference point, the smoothed curve indicates increasing fecundability with increasing gestational age at birth from 28 weeks until about 35 weeks, and is then nearly level with only small fluctuations from the reference value through the highest gestational ages.

Table II.

Fecundability by four categories of gestational age, N = 2814.

Gestational age, weeks	No. of women	No. of cycles	No. of pregnancies	Unadjusted model		Adjusted Model 1		Adjusted Model 2
				FR	95% CI	FR	95% CI	FR	95% CI
<34	19	109	6	0.37	0.17–0.81	0.39	0.18–0.84	0.38	0.17–0.82
34–36	89	371	60	1.05	0.82–1.34	1.04	0.80–1.34	1.03	0.80–1.34
37–41	2463	9845	1571	1	Reference	1	Reference	1	Reference
≥42	243	877	150	1.11	0.94–1.30	1.13	0.96–1.33	1.13	0.96–1.33

Model 1: Adjusted for participant's year of birth, mother's age, mother's marital status, mother's and father's educational level, mother's smoking during pregnancy, mother's hypertension, mother's pre-eclampsia, and mother's diabetes during pregnancy with the participant.

Model 2: Model 1 + mother's history of difficulty conceiving, mother's history of spontaneous abortion, mother's history of preterm birth and mother's lifetime parity.

FR, fecundability ratio; CI, confidence interval.

Table III.

Fecundability according to gestational age in weeks, N = 2814.

Gestational age, weeks	No. of women	No. of cycles	No. of pregnancies	Unadjusted model		Adjusted Model 1		Adjusted Model 2
				FR	95% CI	FR	95% CI	FR	95% CI
<34	19	109	6	0.37	0.17–0.81	0.39	0.18–0.85	0.38	0.17–0.83
<32	11	70	4	0.38	0.15–0.98	0.40	0.15–1.03	0.40	0.15–1.04
32	4	24	1	0.31	0.05–2.09	0.32	0.05–2.21	0.30	0.04–2.08
33	4	15	1	0.42	0.06–2.79	0.43	0.06–2.82	0.39	0.06–2.54
34	15	61	11	1.14	0.65–2.02	1.15	0.63–2.11	1.12	0.61–2.06
35	24	94	19	1.19	0.78–1.82	1.18	0.77–1.82	1.17	0.75–1.80
36	50	216	30	0.94	0.62–1.42	0.93	0.61–1.42	0.94	0.62–1.42
37	134	566	80	0.96	0.77–1.20	0.97	0.77–1.22	0.97	0.76–1.22
38	267	1083	159	0.91	0.74–1.11	0.91	0.74–1.12	0.90	0.74–1.11
39	472	1836	308	1.04	0.91–1.17	1.05	0.93–1.20	1.05	0.92–1.19
40	1105	4481	711	1	Reference	1	Reference	1	Reference
41	485	1879	313	1.01	0.89–1.15	1.02	0.90–1.16	1.02	0.90–1.16
42	209	765	128	1.11	0.92–1.32	1.14	0.95–1.36	1.14	0.95–1.37
≥43	34	112	22	1.09	0.71–1.66	1.12	0.74–1.70	1.11	0.73–1.69

Model 1: Adjusted for participant's year of birth, mother's age, mother's marital status, mother's and father's educational level, mother's smoking during pregnancy, mother's hypertension, mother's pre-eclampsia, and mother's diabetes during pregnancy with the participant.

Model 2: Model 1 + mother's history of difficulty conceiving, mother's history of spontaneous abortion, mother's history of preterm birth and mother's lifetime parity.

FR, fecundability ratio; CI, confidence interval.

Figure 2.

Association between gestational age at birth and fecundability, fitted by restricted cubic splines. The dashed lines indicate the 95% confidence interval (CI). The reference level for the fecundability ratio (FR) was 40 weeks of gestation. The curves were adjusted for participant's year of birth; mother's age, marital status, smoking status, hypertension, pre-eclampsia, diabetes, history of difficulty conceiving, spontaneous abortion, preterm birth and lifetime parity; and mother's and father's educational level. Five knot points were located at 33, 34, 38, 40 and 42 weeks' gestation.

In a sensitivity analysis, we subtracted 1 week from each value of gestational age, assuming that it was overestimated in the registry. The adjusted FR for women born <34 weeks according to this categorization was 0.64 (95% CI: 0.40–1.04), thus still markedly reduced compared with women born at 40 weeks of gestation (Supplementary Table SII).

To examine whether our results were influenced by having included women with up to 11 cycles of pregnancy attempt time at study entry, we repeated the analysis after restricting to women with ≤3 cycles of attempt time (n = 1971). The fully adjusted FRs in this analysis were 0.33 (95% CI: 0.13–0.86) for women born <34 weeks, 1.06 (95% CI: 0.78–1.45) for women born at 34–36 weeks and 1.17 (95% CI: 0.98–1.41) for women born ≥42 weeks of gestation.

Discussion

In this study of 2814 Danish pregnancy planners, fecundability was 62% lower among women born <34 weeks than women born at 37–41 weeks of gestation. This result was not explained by measured maternal characteristics, including markers of reproductive health. Fecundability did not appear to be different among women born at 34–36 weeks or ≥42 weeks of gestation.

Data on gestational age at birth, obtained from the DMBR for women born during 1978–1992, inevitably have a degree of measurement error. In a study based on 1662 Danish births occurring in the period 1982–1987, the level of agreement between data on gestational age in the DMBR and the medical record was estimated to be 43% (Kristensen et al., 1996). For the majority of discrepancies, gestational age at birth was recorded as 1 week later in the DMBR than evaluated by the investigators from the medical record, indicating an underreporting of preterm birth in the registry. In the medical records, determination of gestational age at birth was based on date of LMP in 64% of cases, on ultrasound examination in 35% of cases, and on clinical examination in 1% (Kristensen et al., 1996). This suggests that in our study, gestational age was likely to have been determined primarily by the LMP-based method, which moves the distribution of gestational age toward higher values compared with ultrasound examination (Tunon et al., 1996; Savitz et al., 2002). When we re-defined the categories of gestational age by subtracting 1 week from each value, the adjusted FR for women born <34 gestational weeks was 0.64 (95% CI: 0.40–1.04). Thus, measurement error of gestational age may have contributed to a decrease in FR, but even after considering this, our data indicated that women born <34 weeks had a 36% reduction in fecundability compared with women born at 40 weeks of gestation. Misclassification of gestational age in a woman's birth record would be unlikely to be related to subsequent TTP, implying that such misclassification would be non-differential. More than 96% of pregnancies in our study were detected by home pregnancy tests (Wise et al., 2011), suggesting that our results were not influenced by differential recognition of pregnancy by gestational age of the women.

It is plausible that our study of pregnancy planners attracted women who were already struggling to conceive. If women born <34 weeks' gestation and with previous reproductive problems entered the study out of concern for their fecundability, the FR for such women would be biased downward (Rothman, 2002). Nonetheless, participation was unlikely to be associated with gestational age at birth, because studying gestational age was not a stated objective of the ‘Snart-Gravid’ study, nor was there much information in the literature about an association of gestational age with infertility.

Our study included pregnancy planners only, thus excluding women with high fecundability who had an unintended pregnancy. To examine whether our results were partly attributable to a selection of women with prolonged pregnancy attempts, we restricted to women with ≤3 cycles of pregnancy attempt time at study entry, and obtained similar results, suggesting that inclusion of women trying to conceive for >3 cycles did not introduce substantial bias.

A greater proportion of women born <34 gestational weeks and lost to follow-up had tried to become pregnant for >3 cycles at study entry than women born <34 gestational weeks with complete follow-up. This difference implies that fecundability among women born <34 weeks may be lower than what we observed. This result, however, was based on only five women born <34 weeks and lost to follow-up. Finally, small numbers of women at the extreme ends of the distribution of gestational age reduced the precision of the associated estimates.

Overall, our results correspond to findings from previous studies. Based on the Danish National Birth Cohort, Nohr et al. reported an OR for a TTP >12 months versus <6 months of 1.8 (95% CI: 1.1–3.1) among women born preterm with a birthweight ≤1500 g, compared with women born at term with birthweights of 3001–4000 g (Nohr et al., 2009). There were no substantial differences in probability of prolonged TTP among women born preterm or term with approximately the same birthweights, suggesting that preterm birth was not associated with prolonged TTP. Preterm birth, however, was merely defined as birth <37 weeks' gestation; because a birthweight ≤1500 g is likely to be related to very preterm birth, the possibility that very preterm birth influenced later TTP was not ruled out.

Further, Norwegian and Swedish historical registry based cohort studies have examined associations between a woman's gestational age and her later pregnancy resulting in a birth, as recorded in national birth registries. Ekholm et al. reported a hazard ratio (HR) for reproducing of 0.89 (95% CI: 0.74–1.07) for women born <32 weeks; when stratifying by women's age at the time of delivering their first child, HR decreased to 0.71 (95% CI: 0.50–1.01) for women ≥25 years old, whereas there was little association among women who gave birth at younger ages (Ekholm et al., 2005). DeKeyser et al. found a HR for reproducing of 0.69 (95% CI: 0.45–1.05) among women born <27 completed weeks, and HR of 0.81 (95% CI: 0.75–0.88) among women born <32 completed weeks of gestation (deKeyser et al., 2012). This study included women from the other Swedish study (Ekholm et al., 2005). Swamy et al. reported a relative risk (RR) for reproducing of 0.78 (95% CI: 0.65–0.93) among women born at 22–27 gestational weeks, and RR of 0.89 (95% CI: 0.86–0.93) among women born at 28–32 gestational weeks (Swamy et al., 2008). Finally, Moster et al. reported a RR for reproducing of 0.9 (95% CI: 0.6–1.2) among women born at 23–<28 gestational weeks, and RR of 0.9 (95% CI: 0.8–1.0) among women born at 28–<31 weeks (Moster et al., 2008). This study included women from the other Norwegian study (Swamy et al., 2008). Lower fertility among women born preterm, as suggested by these studies, may not entirely reflect decreased fecundability; it could be partly attributed to altered mating patterns, since individuals born preterm are less likely than those born at term to be cohabiting or married (Lindstrom et al., 2007; Moster et al., 2008). In contrast, our results cannot be explained by mating patterns related to preterm birth, since we only considered women in stable relationships. Further, these studies considered the number of registered births, which is not a sensitive indicator of fecundability; e.g. conceptions ending in a miscarriage will not contribute to such a measure of fertility. In contrast to previous studies, we assessed fecundability in 1-week categories of gestational age, from <32 to ≥43 weeks. The FRs for the gestational weeks <32, 32 and 33 were imprecise due to a low number of women in these categories, however, the effect estimates all ranged from 0.30 to 0.40, consistent with a deleterious effect of early gestational age on fecundability of approximately the same magnitude. On this basis, we chose to combine these categories into one category (<34 gestational weeks). Our data did not indicate a notable decrease in fecundability among women born after 34 weeks.

It is biologically plausible that preterm birth is associated with subsequent impaired fecundability, although the underlying pathways remain difficult to disentangle. At delivery, the infant is separated from its sources of maternal and placental hormones, leading to large increases in infant gonadotrophin levels (i.e. FSH and LH) and increased ovarian follicular maturation, particularly during the first 3–6 months of life (Speroff et al., 1999). However, FSH levels are 10–20 times higher, and LH levels 3–4 times higher in the first post-natal weeks among girls born preterm compared with girls born at term (Tapanainen et al., 1981; Kuiri-Hanninen et al., 2011). This increase is prolonged and follicular development is delayed relative to full-term girls (Kuiri-Hanninen et al., 2011), suggesting immaturity of reproductive organs and the hypothalamic–pituitary–ovarian axis at preterm birth. Although speculative, it seems plausible that such abnormalities may be related to impaired fecundability.

The link between preterm birth and later fecundability also could be established in fetal life. According to the ‘developmental origins of health and disease’ hypothesis, adverse environmental stimuli during the prenatal or early post-natal period may induce permanent alterations in physiology, metabolism and the functioning of endocrine axes, predisposing the individual to adult diseases (Gluckman and Hanson, 2004, Gluckman et al., 2008). Preterm birth may be a fetal response to an adverse intrauterine environment (Impey and Child, 2012); hence, factors operating in the prenatal period may explain the relation between preterm birth and later fecundability. Adolescent girls born small-for-gestational age (a different measure of a suboptimal intrauterine milieu) have reduced uterine and ovarian size, and anovulation or lower ovulation rate compared with girls with an appropriate weight for their gestational age at birth (Ibanez et al., 2000, 2002), indicating a relation between early life events and later fertility. To consider the potential influence of maternal environmental factors, we controlled for mother's smoking and medical conditions during pregnancy; however, controlling for these factors did not materially alter our estimates of association. We also considered whether potential hereditary factors, i.e. markers of maternal fecundability with a possible influence on fecundability of the daughter, might contribute to the observed association, but we found no evidence of this.

In conclusion, using prospective data on TTP, we found a pronounced decrease in fecundability among women born <34 weeks of gestation. We hesitate to infer a causal relation between early birth and lower fecundability, but our finding does augment results from previous studies that reported reduced fertility among women born preterm.

Supplementary data

Supplementary data are available at http://humrep.oxfordjournals.org/.

Authors' roles

All authors contributed to the design of the study. C.W. wrote the drafts of the paper, and C.W. and A.H.R. performed the statistical analyses. All authors contributed to the interpretation of the study results, and reviewed and approved the final manuscript.

Funding

The study was supported by the National Institute of Child Health and Human Development (R21-050264), the Danish Medical Research Council (271-07-0338) and the Health Research Fund of Central Denmark Region (1-01-72-84-10).

Conflict of interest

None declared.