Birth Outcomes by Infertility Diagnosis Analyses of the Massachusetts Outcomes Study of Assisted Reproductive Technologies (MOSART)

The descriptive statistics of the 7,354 ART cycles/deliveries are given in Table I by maternal and paternal characteristics, Table II by health and ART treatment parameters, and Table III by pregnancy and birth outcomes for singletons and Table IV for twins. As shown in Table I , the parental characteristics differed significantly across the eight diagnosis groups, with mothers and fathers in the diminished ovarian reserve group being oldest (40.8 years for mothers and 41.2 years for fathers), most likely to be white (90.7% and 89.3%, respectively), and with the most education (82.5% and 76.2%, respectively, with a bachelor degree or graduate school). The youngest parents were in the ovulation disorders group (mothers, 33.8 years and fathers, 35.9 years). The tubal factors group was the most racially and ethnically diverse, with the highest proportions of mothers and fathers who were Hispanic (7.7% and 6.4%, respectively), Black (8.1% and 8.4%, respectively), or Asian (6.6% and 6.2%, respectively); the uterine factor group also had a high proportion of Black mothers and fathers (5.1% for both). Table II shows that the health and ART treatment parameters differed substantially by diagnosis group. Chronic hypertension was present in more than 4% of pregnancies with ovulation disorders, diminished ovarian reserve, and uterine factors; women with ovulation disorders also had the highest proportion of diabetes mellitus (4.5%). Autologous sperm were used for more than 78% of diagnosis groups, and autologous oocytes for more than 92% of diagnosis groups (except diminished ovarian reserve, 37.8%, and other factors, 75.9%). In more than 85% of all pregnancies, fresh embryos were used. Despite significant differences in the number of embryos transferred (the uterine factor group was most likely to have three or more embryos transferred, and the ovulation disorders group most likely to have one embryo transferred), plurality at the 6-week ultrasound was only marginally different across groups, and at birth did not differ significantly. At the 6-week ultrasound, about 67% of pregnancies were singletons, 31% twins, and 2% triplets or higher order multiples; at birth, these proportions were 73% singletons and 27% twins. The univariate comparison of pregnancy and birth outcomes by the diagnosis groups is shown in Table III for singletons and Table IV for twins. Women with the diagnosis of diminished ovarian reserve had the highest rate of pregnancy hypertension (21.1% for singletons and 37.4% for twins), followed by women with the diagnosis of ovulation disorders (14.7% for singletons and 27.8% for twins), who also had the highest rate of gestational diabetes (13.5% for singletons and 15.2% for twins). The rates of bleeding diagnoses (uterine bleeding, abruption placenta, excess bleeding in labor, and placenta previa) were low (ranging from 1.4–2.7% overall for singletons and 1.5–2.8% for twins, and 0–4.8% for individual diagnoses for singletons and 0–7.7% for twins) and did not differ significantly across the eight diagnosis groups. There were substantial differences in rates of prenatal hospital utilization across the diagnosis groups. The highest rates of emergency room visits were among women with the diagnosis of tubal factors pregnant with singletons (16.0%) and women with the diagnosis of uterine factors pregnant with twins (26.9%). The highest rates of observational stays were among women with ovulation disorders pregnant with singletons (14.3%) and women with endometriosis pregnant with twins (23.5%). Antenatal hospitalizations were highest among women with the diagnosis of endometriosis (7.2% for singletons) and uterine factors (7.1% for singletons and 26.9% for twins). The rates of breech or malpresentation were 6.8% for singletons, with the highest rate for women with uterine factors (18.1%). Among twins, the overall rate of breech or malpresentation was 32.3%, with a range of 29.7–39.5% for individual diagnoses. Mode of delivery by diagnosis group varied significantly among singletons, but not among twins. Singletons born to women with the diagnosis of uterine factor had the lowest rate of vaginal birth (30.6%) and the highest rate of primary cesarean birth (52.8%). Only 18.9% of twins were born vaginally overall, and ranged from 14.9–23.1% for individual diagnoses. There were substantial unadjusted differences in length of gestation, preterm delivery, birthweight, and low birthweight outcomes across diagnosis groups. Within singleton births, length of gestation averaged more than 38 weeks, with the exception of the uterine factors group (37.7 weeks). Preterm birth ranged from 9.4% to 14.8%, with male factor and endometriosis having the lowest rates (9.4% and 9.5%, respectively), and diminished ovarian reserve, ovulation disorders, and other factors having the highest rates (13.2%, 14.2%, and 14.8%, respectively). Singleton birthweights averaged more than 3,200 grams for all diagnosis groups, with no significant differences in mean birthweight zscore or percent small-for-gestation or large-for-gestation. Low birthweight ranged from 4.4% to 12.5%, with endometriosis and male factor having the lowest rates (4.4% and 6.7%, respectively), and ovulation disorders (9.9%), other factors (10.0%), and uterine factors (12.5%) having the highest rates. Within twin births, length of gestation averaged more than 35 weeks, with the exception of the ovulation disorders group (34.9 weeks) and the uterine factors group (34.5 weeks). Preterm birth ranged from 57.0% to 68.5%, with male factor and endometriosis having the lowest rates (57.0% and 59.1%, respectively), and ovulation disorders (68.5%) and diminished ovarian reserve (64.2%) having the highest rates. Twin birthweights averaged more than 2,400 grams for all diagnosis groups, with the exceptions of ovulation disorders (2,363 grams) and uterine factors (2,318 grams). There were no significant differences in mean birthweight zscore or percent small-for-gestation or large-for-gestation. Low birthweight ranged from 42.1% to 54.5%, with endometriosis (42.1%) and other factors (44.8%) having the lowest rates, and ovulation disorders (54.5%), uterine factors (53.8%) having the highest rates. The models of adverse outcomes by infertility diagnosis are shown in Table V . Compared to pregnancies with the diagnosis of male factor (the reference diagnosis group), the risks of prematurity and gestational diabetes were increased with ovulation disorders; the risk for pregnancy hypertension was decreased with endometriosis; and the risks of prematurity increased and large-for-gestational age birthweight decreased with the diagnosis of other factors. Also compared to pregnancies with the diagnosis of male factor, six out of the other seven diagnoses were associated with increased risks for prenatal hospitalizations, including emergency room visits (diminished ovarian reserve), and hospital admissions (endometriosis, ovulation disorders, tubal factors, uterine factors, and other factors). The risk of primary cesarean delivery was twice as likely for women with the diagnosis of uterine factors.

This analysis indicates that although the infant outcomes of LBW, SGA, and LGA were generally similar across diagnosis groups, specific diagnoses had greater risks for morbidities, including prematurity, gestational diabetes, emergency room visits, observational stays, hospital admissions, and primary cesarean delivery compared to pregnancies with the diagnosis of male factor, even after adjustment for baseline demographic factors, ART treatment parameters, and plurality. Women with diminished ovarian reserve were older and were more likely to have assisted hatching. Women with ovulation disorders (which includes women with polycystic ovarian syndrome) were younger, were more likely to have diabetes (both preexisting and gestational), and to deliver preterm. Women with uterine factors were more likely to have bleeding diagnoses and to deliver by primary cesarean. The characteristics and outcomes by diagnosis groups confirmed prior reports 14 , 15 – 19 , 22 . One of the central questions in ART research is whether differences seen in outcomes between ART pregnancies and spontaneous conceptions are a function of the ART treatment or the underlying infertility. The results of this study suggest that within this ART population, the differences seen between the diagnostic groups in obstetric and perinatal outcomes were consistent with their underlying infertility. For example, when compared with women whose infertility diagnosis was male factor, women with a diagnosis of uterine factors had higher rates of hospital admissions (AOR 2.68) and cesarean delivery (AOR 1.96) both of which could be directly associated with a compromised uterus (18). Women with ovulation disorders had a higher risk of gestational diabetes (AOR 1.77) as expected by the high proportion of patients with polycystic ovarian syndrome in this group 18 . As reported previously 15 , 19 , singleton birthweight was lower in the ovulation disorders group (as mean birthweight and percent SGA). Emergency room visits were more common in women with diminished ovarian reserve (AOR 1.45), reflecting the higher percentage of older women in this diagnostic category. Interestingly, but somewhat unexpectedly, patients with endometriosis had a lower rate of pregnancy hypertension (AOR 0.61). The results suggest that the underlying infertility diagnosis could have a direct effect on the course and complications of the pregnancy. Findings from Denmark 7 , 8 , 13 , Norway 6 , 10 , 11 , and Australia 12 have all shown that subfertility (whether treated or untreated) is associated with reductions in birthweight, length of gestation, and fetal growth, suggesting the etiology is the underlying pathology not the infertility treatment. Our findings of reduced singleton birthweight with the diagnosis of uterine factors we report are in accord with Gibbons et al 22 , who evaluated national ART singleton outcomes in 2004–06 by infertility diagnosis. Contrary to their results, we did not find an excess of LBW by diagnosis, but our study design, populations, and analyses differed: they limited their population to singletons and compared IVF, donor oocyte, and gestational carrier groups; we included both singletons and twins, excluded gestational carriers, and adjusted for donor oocyte and semen, and plurality, as well as other covariates. The MOSART study, which includes linking ART cycles to the vital records and hospital utilization data, represents the first time these datasets have been linked using direct identifiers from both datasets. ART national surveillance summaries are limited to birth outcomes reported by the patient herself or her obstetric provider 34 – 36 . Prior studies 37 , 38 have relied on linkages between ART cycles and vital records using only maternal and infant dates of birth, or probabilistic algorithms 39 . Although there is a high degree of comparability between the SART CORS and vital records 40 , our study design assures more accurate linkage between ART treatment cycles, vital records, and the hospital discharge birth data, and a more complete picture of perinatal outcomes. This study uses retrospective data from several centralized datasets and although this is advantageous to achieve large numbers, it has the disadvantage that data entered into the SART CORS system is not as rigorously controlled as data collected for a prospective research study. This is nowhere more critical than in the diagnostic criteria that are used for defining the data entry into SART CORS. The criteria used to define diagnosis in SART CORS are not as strict as might be optimal for a prospective study and different clinics may define these more or less rigorously. Thus, for example, endometriosis can include any severity of endometriotic lesions; ovulation disorders could be either polycystic ovarian syndrome or oligoovulatory patients. Despite these concerns, our data show consistency in the definition of diseases with the expected demographics of patients in these groups (see Table I ). The groups of Other and Unexplained are the most likely to have a combination of patients with differing conditions and may also contain patients who are incorrectly classified as has been shown by validation procedures discussed earlier, and therefore the results for these two groups should be interpreted with caution. Although this study has several unique advantages over prior ART research, it is also subject to a number of limitations. The SART CORS is a clinical and surveillance database, and its use for research is secondary. Although a sample of the data is validated annually, there is still the possibility of error in data abstraction from the medical record and data entry into the online database. Likewise, the primary purpose of vital records is civil registration, with public health research and surveillance being secondary uses. A recent analysis of the quality of medical and health data from the 2003 birth certificate revision compared to the hospital medical record data reported that levels of agreement were high for some items, but only substantial or moderate for most 41 . Another limitation of this analysis is that it only includes women in Massachusetts. There may be significant demographic and outcome differences in patients in other regions of the country and with other healthcare systems, potentially limiting the generalizability of our findings. This study is also limited by the small sample size within specific diagnosis groups, particularly uterine factors, in which there were only 98 pregnancies. With the addition of the remaining five years of this study (2009–13), it is expected that the number of pregnancies with this diagnosis will at least double; this larger study population will also permit new analyses for women with more than one infertility diagnosis, as well as ART cycles which resulted in higher-order multiple births (triplets and quadruplets). We also plan on including BMI with the 2011–13 data. In summary, this analysis demonstrates differences in parental characteristics, ART treatment parameters, and adverse obstetric outcomes by diagnosis, but no substantial clinical differences in low birthweight or extremes of fetal growth for gestation (SGA or LGA).

Data were obtained from two sources 1) the SART CORS online database, a system that contains cycle-based ART data from the majority of US ART clinics and 2) the birth certificates, fetal death records, and hospital utilization data in the Massachusetts-based Pregnancy to Early Life Longitudinal (PELL) data system, an ongoing population-based project that compiles data from statewide vital statistics systems. The study took place under a Memorandum of Understanding executed between SART and the three entities that participate in the PELL project: Boston University School of Public Health, the Massachusetts Department of Public Health, and the Centers for Disease Control and Prevention. Human subjects approval was obtained from all entities and participating Universities. The study was also approved by the SART Research Committee. The project is known as the Massachusetts Outcome Study of Assisted Reproductive Technology, or MOSART. The SART CORS contains comprehensive data from more than 90% of all clinics performing ART in the US. Data were collected and verified by SART and reported to the Centers for Disease Control and Prevention in compliance with the Fertility Clinic Success Rate and Certification Act of 1992 (Public Law 102–493). SART maintains HIPAA compliant business associates agreements with reporting clinics. In 2004, following a contract change with CDC, SART gained access to the SART CORS data system for the purposes of conducting research. The national SART CORS database for 2004–08 contains 642,927 ART treatment cycles. The database includes information on demographic factors; ART diagnoses and treatment parameters; ART treatment and pregnancy outcomes. The data in the SART CORS are validated annually 28 , 29 with some clinics having on site visits for chart review based on an algorithm for clinic selection. During each visit, data reported by the clinic were compared with information recorded in patients’ charts. In 2010, records for 2,070 cycles at 35 clinics were randomly selected for full validation, along with 135 embryo banking cycles 9 . The full validation included review of 1,352 cycles for which a pregnancy was reported; of which 446 were multiple-fetus pregnancies. Nine out of 10 data fields selected for validation were found to have discrepancy rates of ≤5%. The exception was the diagnosis field, which had a discrepancy rate of 18%. For approximately 20% of the discrepancies, a single wrong diagnosis was reported, mainly the diagnoses of ‘other’ or ‘unexplained,’ instead of a specific cause. For another 50% of the discrepancies, multiple causes of infertility were found in the medical record, but only a single cause was reported. The PELL data system has linked information on more than 99% of all births and fetal deaths in Massachusetts from 1998–2008 to corresponding hospital utilization data (hospital admissions, observational stays, and emergency room visits) for individual women and their children. The Massachusetts Department of Public Health (MDPH) and the Massachusetts Center for Health Information and Analysis are the custodians of the PELL data. PELL is a relational data system composed of individual databases linked together by randomly-generated unique IDs for mother and infant. The PELL data system is housed at MDPH. We constructed the MOSART database by linking the SART CORS and PELL data systems for all children born in Massachusetts hospitals to Massachusetts resident women between July 1, 2004 and December 31, 2008. The starting date was chosen based on the availability of SART CORS data (January 1, 2004) to allow us to capture any births associated with ART and the end date reflected the latest available data from both SART and PELL when we began the study. PELL data from July 1, 2004 and December 31, 2008 included 282,971 women with 334,152 deliveries resulting in 342,035 live births and fetal deaths which were linked to 42,649 ART cycles among 18,439 women from SART CORS. A deterministic five phase linkage algorithm methodology was implemented 24 . The matching was based on common information in both records on the baby’s date of birth, mother’s date of birth, her first name and last name; and father/partner’s last name. We linked 9,092 deliveries for a linkage rate of 89.7% overall and 95.0% for deliveries in which both mother’s zip code and clinic were located in Massachusetts. Of the 9,092 linked pregnancies, there were 60 pregnancies with fetal deaths, which were excluded. Also excluded were cycles with more than one infertility diagnosis (1,601 pregnancies), and cycles resulting in a higher-order birth (99 triplet or quadruplet pregnancies). The 7,354 ART cycles which resulted in live births (5,339 singleton pregnancies and 2,015 twin pregnancies of at least 20 weeks gestation and birthweights of at least 350 grams) were categorized by eight mutually-exclusive infertility diagnoses assigned by the treating clinic (male factor, endometriosis, ovulation disorders, diminished ovarian reserve, tubal factors, uterine factors, other factors, and unexplained). Diagnoses are defined for data entry to SART CORS as follows: male factor is the presence of abnormal semen parameters or function; endometriosis is the presence of any stage of endometriosis whether treated or untreated; ovulation disorders can have several differing definitions including multiple cysts affecting fertility, oligoovulation, or anovulation; diminished ovarian reserve is currently defined as high follicle stimulating hormone or estradiol in the early follicular stage as measured on a clomiphene challenge test, or reduced ovarian volume, but could also have been defined by advanced maternal age for some earlier cycles in our cohort; tubal factor is any condition affecting the patency of the Fallopian tubes; uterine factor includes any uterine abnormality. The category of other factors could have included immunologic, chromosomal, cancer, and any other conditions not listed in the previously defined categories. Unexplained is intended to be an absence of any defined male and female diagnoses. We examined a range of pregnancy and birth outcomes, including maternal morbidity (pregnancy hypertension, gestational diabetes), prenatal hospital utilization (including emergency room visits, observational stays, and hospital admissions), delivery complications (including primary cesarean delivery), and birth outcomes (including preterm birth, low birth weight, small-for-gestational age and large-for-gestational age birthweight). The reference group was ART pregnancies with male factor as the infertility diagnosis. This reference group has been used on other studies of ART outcomes 14 , 22 because it suggests the absence of fertility issues for the female partner. Length of gestation was calculated by using the SART CORS outcome date minus date of transfer and adding 14 days and the cycle day of transfer. This effectively was the outcome date minus the date of conception (or fertilization) plus 14 days 25 , 30 , 31 . All other pregnancy and birth outcomes were obtained from the birth certificate data in PELL. Preexisting maternal conditions (diabetes mellitus and chronic hypertension), pregnancy hypertension, and gestational diabetes were identified in PELL from either the birth certificate or the hospital discharge delivery record (ICD-9 codes of 648.0 or 250 for diabetes mellitus; 401, 402, 403, 404, or 405 for chronic hypertension; 642 for pregnancy-related hypertension; 648.8 for gestational diabetes). Prenatal hospitalization usage was identified in PELL by linkage of the maternal longitudinal hospital utilization data with birth certificate records, and categorized as emergency room visits, observational stays, or hospital admissions. Maternal and paternal demographic factors (age, race and ethnicity, and education) were obtained from the birth certificate in PELL; ART treatment parameters (diagnoses; oocyte source, state, and number retrieved; embryos cryopreserved; semen source; micromanipulation; embryo state; and plurality at 6-week ultrasound) were obtained from the SART CORS database. Birth weights at each gestational age are normally distributed. A z-score (or standard deviation score) is the deviation of the value for an individual from the mean value of the reference population divided by the standard deviation for the reference population 32 . Birthweight z-scores were calculated to evaluate adequacy of weight-for-age using population-based standards, as recommended by Land 33 and modeled as continuous and categorical variables. We generated gender-, race/ethnicity-, and gestation-specific birthweight means and standard deviations using Massachusetts data for all live births from 1998–2008. Infants with z-scores of ≤−1.28 (below the 10 th percentile for gestation) were classified as small-for-gestational age, and those with Z-scores of ≥1.28 (above the 90 th percentile for gestational age) were classified as large-for-gestational age. We compared the maternal and paternal characteristics, the health and ART treatment parameters, and the pregnancy and birth outcomes across the eight infertility diagnoses using analysis of variance for continuous variables and χ 2 for categorical variables, stratified by plurality at birth (singleton, twin). Pregnancy and birth outcomes odds ratios and 95% confidence intervals were further computed from multivariate logistic regression models, adjusting for maternal and paternal demographic factors; ART treatment parameters; maternal preexisting medical conditions; and plurality at birth. The models for primary cesarean delivery were additionally adjusted for breech/malpresentation and cephalopelvic disproportion, and excluded women with prior cesarean delivery. All models were also initially adjusted for maternal pregnancy morbidities (pregnancy hypertension and gestational diabetes), but because the odds ratios and confidence intervals did not change substantially, these factors were not retained in the final models (results not shown). Body mass index (BMI, kg/m 2 ) was not included in these analyses because it was only added to the SART CORS in 2007, and will be not included in the Massachusetts birth certificate until 2012. All analyses were performed using the Statistical Package for the Social Sciences, version 19.0 (IBM SPSS, Inc., Chicago, IL, USA, 2010).