Our data come from the North Carolina Early Pregnancy Study.[3
] Briefly, the North Carolina Early Pregnancy Study (1982–1985) was a prospective cohort study designed to provide basic information on reproductive events including very early pregnancy loss [3
], ovulation [5
], and day-specific fertility [6
]. Women who were planning to become pregnant were recruited from local communities and enrolled upon cessation of birth control methods. Women were excluded if they had a serious chronic illness, or if they or their partners had a history of fertility problems. Women collected first-morning urine specimens and completed daily record cards either until they became clinically pregnant or until 6 months had passed with no clinically-apparent pregnancy. Record cards included information on vaginal bleeding. All participants gave written informed consent, and the study protocol was approved by the Institutional Review Board of the National Institute of Environmental Health Sciences.
First-morning urine specimens were assayed for estrone 3-glucuronide (E1
G), pregnanediol 3-glucuronide (PdG), luteinizing hormone (LH), and human chorionic gonadotrophin (hCG). E1
G and PdG were measured by direct radioimmunoassay.[7
] The specimens were analyzed in duplicate or triplicate and the geometric means of the daily replicates were divided by the corresponding creatinine concentration to adjust for variations in dilution. LH was measured through an immunofluorometric assay that detects both intact LH and the LH β-subunit.[9
] We employed a highly-sensitive immunoradiometric assay to quantify hCG in the urine samples.[10
] Pregnancy was defined as a rise in hCG that exceeded 0.025ng/ml for three consecutive days.[3
] A pregnancy was categorized as an early loss if the initial rise in hCG was followed by a decline to baseline with menstrual-like bleeding beginning within 42 days of the previous menstrual period. The day of ovulation for each cycle was defined based on the rapid drop in the estrogen-to-progesterone ratio [11
], using an algorithm validated by LH measurements.[12
] This information combined with the participant’s self-reported vaginal bleeding allowed us to define follicular phase length as the time from the first day of the last menstrual period up to, but not including, the estimated day of ovulation.[5
] Luteal phase length was defined as the time from the day after ovulation up to but not including the first day of the next menstrual cycle.
We based our analyses on sets of three consecutive ovulatory menstrual cycles from a given woman (“triads”). We created two types of triads (). One set was centered on early loss cycles that were preceded by a non-conception cycle and followed by a cycle that resulted in either a clinical conception or nonconception. The second set of triads were identical except that the middle cycle was a non-conception cycle instead of an early loss cycle. The first cycle provided a baseline for evaluating possible within-woman effects of the middle cycle on the third cycle. The proportion of third cycles that resulted in a clinical conception was similar in the two sets of triads.
Figure 1 Depiction of the analytic design used to assess the influence of early pregnancy loss on the subsequent menstrual cycle. Two types of triads were created and within-woman baseline adjusted measures were obtained by subtracting the value of a summary variable (more ...)
We characterized the crucial aspects of hormone secretion using a set of summary variables that had been previously defined ().[13
] These summary variables describe hormone secretion up through luteal day 6, prior to the occurrence of a new implantation (which usually occurs after luteal day 6).[14
] Among the conceptions in the third cycle of the triads, only one implanted on luteal day 6. Our results did not change if this triad was excluded, and so it is included.
Definitions of summary urinary hormone variables
For each woman and each variable, we subtracted the hormone or phase length value in cycle 1 of the triad from the value in cycle 3, creating a baseline-adjusted summary value for each triad. We compared the baseline-adjusted values between the two types of triads and tested for statistical significance using the two-sample non-parametric Wilcoxon’s rank sum test. All p-values are two-sided.
Some hormone and menstrual cycle data were missing because the participant did not provide a urine sample on a given day or because daily hormone analyses had been performed on only an a priori subset of cycles. One woman was missing information on bleeding at the start of her cycle, such that we were unable to calculate variables that depended on a specific time window in the follicular phase (mid-follicular E1G, mid-follicular PdG, and follicular phase length).
Of the 221 participants in the North Carolina Early Pregnancy Study, 44 women had a total of 48 very early pregnancy losses. An early loss was included in the analysis if a full triad could be formed around it. We considered a cycle unobserved if the woman’s first cycle in the study began after menstrual-cycle-day 6. Thirteen early losses occurred in first cycles and therefore did not have a pre-loss comparison cycle. Two losses occurred in the last cycle under observation and were also excluded. Four women had two very early losses; we included only the first in the analysis. Six triads were missing information on ovulation (but in no case was a post-loss cycle anovulatory). This left 23 early loss triads for analysis.
The comparison group of non-conception triads was assembled by selecting the earliest eligible triad of cycles in the study (women could have contributed up to six months of cycles). Women who had an early loss were not eligible to contribute a non-conception triad except for one woman who had an early loss in her last cycle in the study. Of the 178 remaining women, 91 became pregnant before providing two observed cycles and 7 were missing a known day of ovulation in the first or last cycles of the triad. This left a total of 80 non-conception triads for analysis.
Demographic characteristics were obtained from the enrollment questionnaire. Gravidity was defined as the number of pregnancies conceived before enrollment. Body mass index was calculated as weight (in kilograms) divided by height (in meters) squared and categorized according to the World Health Organization criteria.[15