SART CORS IVF registry: looking to the past to shape future perspectives

Abstract

Purpose

The SART CORS database is an informative source of IVF clinic-specific linked data that provides cumulative live birth rates from medically assisted reproduction in the United States (US). These data are used to develop best practice guidelines, for research, quality assurance, and post-market surveillance of assisted reproductive technologies. Here, we sought to investigate the key areas of current research focus (higher-order categories), discover gaps or underserved areas of ART research, and examine the potential application and impact of newer ART adjuvants, future data collection, and analysis needs.

Methods

We conducted a systematic review (PRISMA guidelines) to quantify unique output metrics of the SART CORS database. Included were SART member reporting clinics: full-length publications from 2004 to 2021 and conference abstracts from 2015 to 2021, the two key timepoints when the SART CORS database underwent transformative shifts in data collection.

Results

We found 206 abstracts presented from 2015 to 2021, 189 full-length peer-reviewed publications since 2004, with 654 unique authors listed on these publications. A total of 19 publications have been highly impactful, garnering over 100 citations at the time of writing. Several higher-order categories, such as endometriosis and tubal infertility, have few publications. The conversion of conference abstracts to full-length papers ranged from 15 to 35% from 2015 to 2021.

Conclusions

A substantial body of literature has been generated by analyzing the SART CORS database. Full-length publications have increased year over year. Some topic areas, such as endometriosis and tubal infertility, may be underrepresented. Conversion of conference abstracts to full-length publications has been low, indicating that more organizational support may be needed to ensure that research is methodologically sound and researchers supported to reach full publication status.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10815-022-02634-6.

Introduction

In 2018, 2.0% of all infants born in the USA were conceived using assisted reproductive technologies (ART) [1]. Utilization of ART is expected to increase, and a greater proportion of people will be born using these technologies in the future. A convergence of social trends—lowering barriers to access to care for vulnerable populations including fertility preservation for cancer, transgender individuals, and couples with recessive gene disorders and widening insurance coverage for the uninsured—combined with new technological innovations are accelerating the trends toward increased medically assisted reproduction. Ongoing, ever-evolving national databases to collect standardized data are thought to be necessary to continuously improve the delivery of healthcare.

The Society for Assisted Reproductive Technology Clinic Outcome Reporting System (SART CORS) database contains comprehensive data from more than 90% of all clinics providing ART in the USA (http://www.sart.org). Data for IVF treatment cycles are collected and verified by SART and reported to the Center for Disease Control and Prevention (CDC) in compliance with the Fertility Clinic Success Rate and Certification Act of 1992 (Public Law 102–493). SART CORS database provides ART treatment details starting in 2004 for research purposes, to persons or entities who have agreed to comply with SART guidelines. Stakeholders, including physicians, embryologists, data scientists, patient advocacy groups, third-party insurers, and legislatures, rely on these data to quantify the utilization of ART treatments, examine adverse effects, and reveal key insights into novel ART treatments through retrospective research questions. In 2014, the denominator for reporting ART outcomes changed, from reporting pregnancy per transfer to pregnancy per cycle start, and linked cycles (linking of a specific retrieval to a primary transfer either fresh or frozen/thaw) were initiated, allowing the calculation of cumulative live birth rates, which is considered to be the most contemporary way of reporting outcomes.

SART CORS is a unique database, annually validated for accuracy, veracity, and integrity of reported data. SART member clinics are expected to be compliant with the required guidelines, including prospective reporting, following ethical and best practices, quality assurance, and advertising guidelines. The engagement of these guidelines raises the standard and quality of data entered into the database, with the goal of using these data for research and analysis, ever-moving the goalpost toward practicing the highest standards of patient care through auditing and validation. For researchers, IVF registry-based studies provide large-scale sample sizes to test associations and trends with ample statistical power, which may otherwise go unrecognized in smaller datasets, as well as account for confounding variables that may inject unintentional bias and lead to unsubstantiated conclusions.

These data include hundreds of parameters, including how many IVF cycles are started, patient demographics, IVF success rates per cycle start, i.e., cumulative live birth rates, among many other data points. Reporting these data is a rigorous task that takes dedicated, highly trained embryologists and medical personnel to accurately enter and track such data into the database. However, it is not without limitations. SART CORS does not record socioeconomic status (i.e., annual income) or proxies for it (i.e., highest obtained educational level) which remain a significant confounding variable when studying social issues such as the impact of race or state insurance mandated care upon outcomes. Furthermore, there is no uniform registry of patients, so when patients and donors move from one clinic to another, the database records them as separate entities. SART member clinics notify SART prospectively, that is, within 4 days of any medicated IVF cycle being started, before the outcome of that cycle can be known, for the purpose of avoiding selection bias when reporting outcomes. This ensures a commitment to the highest standards of data collection. Lastly, SART CORS data fulfills the federally mandated reporting requirements for ART, specified by the Fertility Clinic Success Rate and Certification Act of 1992 (Public Law 102–493).

SART member IVF clinics make data available for research and publications for the benefit of other researchers and patients through continued participation in SART. In return, SART member IVF clinics are encouraged to use the national SART CORS database to address impactful retrospective research questions. Data is available for hypothesis-driven research from 2004 to present for all SART member clinics. SART member clinics who submit proposals to the research committee for approval receive a customized dataset of national data that can be analyzed to draw meaningful associations between patient and cycle variables with cycle outcomes following appropriate statistical modeling including controlling for confounding variables. The SART research committee is comprised of 12 clinicians and scientists in both academic and private practice that are appointed to a 6-year committee term. The research committee evaluates proposals based on study design, objective, rationale, planned methods of data analysis, and originality of research question. Jain et al. [4] previously reviewed the clinical impact that 30 years of data collection (starting in 1997) via the SART database has yielded. The database has been instrumental to validate that elective single embryo transfers result in lowered multiple pregnancy rates, link the economics of IVF and disparities to access and outcomes, illuminate powerful links between SART and state birth and cancer registries, and yield prediction models to inform contemporary clinical IVF practice to maximize birth rates with respect to individual patient and treatment characteristics. However, observational, retrospective studies require methodological expertise. Large datasets do not necessarily deal well with confounding (where people treated differ in their characteristics from those not treated). Big data can lead to big errors if it does not meet minimum standards for observational, retrospective studies [30].

The total output of the SART CORS database includes both full-length publications and conference abstracts, which undergo a less rigorous form of peer review, but serve an important function. Our goal was to investigate the total output of the SART CORS database to glean new insight into trends, gaps, and underserved areas and to speculate on potential areas of future growth for the database.

Methods

This review was performed using the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines to search three different scientific databases (PubMed, Web of Science, Scopus), with the search string: (“Society for Assisted Reproductive Technology Clinic Outcome* Reporting System” OR “SART CORS”) AND (“Assisted repro*” OR “clinic*” OR “fertility*” OR “IVF*”). Relevant publications were selected and analyzed. Inclusion criteria were peer-reviewed papers written in English whose major data source was the SART CORS database. Exclusion criteria included duplicates (citations present in more than one database), reviews, papers not written in English, and non-peer-reviewed papers. After the duplicates were removed, two investigators independently assessed the titles and abstracts of all articles. Studies that did not meet the inclusion criteria were excluded. Disagreements regarding inclusion were resolved by consensus or with the involvement of a third author. Our database search concluded on July 1, 2022, yielding (Web of Science: 93 results, PubMed: 23 results, Scopus: 802 results) 1018 total results.

We examined unique metrics that have not been previously quantified, for example, higher-order categories (topics for grouping purposes), number of unique authors, number of citations, and number of abstracts converted into full-length publications. We examined abstracts based on SART CORS data from 2015 on, as more contemporary practices (new fields) were captured for the first time going forward from 2014, which included PGT for aneuploidy screening, embryo banking, providing reasons for “freeze all” (i.e., avoidance of OHSS, genetic testing, or issues with endometrial receptivity) (Fig. 1).

Fig. 1.

PRISMA diagram for full-length articles and abstracts using the SART CORS database as the primary source

Results

Here, we present the body of literature that has been generated by analyzing the SART CORS database (Supplemental Table 1). We found that 206 abstracts have been presented from 2015 to 2021, 189 full-length peer-reviewed publications have been published since 2004, and 654 unique authors are listed on these publications. The number of publications has increased year over year (Fig. 2a). We grouped the full-length publications into 24 higher-order categories and 4 temporal subcategories to analyze the frequencies and reveal new insights into major investigative trends and possible gaps, under-studied, under-reported, or underserved areas. Endometriosis, tubal infertility, fertility preservation, and sperm donor categories had the least number of dedicated publications (Table 1; higher-order categories). ASRM conference abstracts ranged from 21 to 43 in the years studied here (Fig. 2b). The conversion rate of ASRM abstracts to full-length publications ranged from 15 to 35% (Fig. 2c). Analyzing citations per higher-order category reveals that although some topics are highly cited (we defined as over 100 citations), such as endometriosis, body mass index, and racial and ethnic disparities, there are relatively few publications in those higher-order categories (Fig. 2d). Supplemental Table 3 shows details for highly cited publications.

Fig. 2.

Collective impact of the SART CORS database

Table 1.

Higher-order categories for peer-reviewed publications from 2004 to 2021 using SART CORS database as a primary data source. (Subcategory percentage of total indicated in parentheses)

Discussion

Statistical databases in epidemiology and public health are important to identify the causes of disease, modes of transmission, appropriate control, and prevention measures and to prioritize and evaluate activities. Medical registry databases are a subset of public health databases. They ensure accountability, transparency, access, and democratize the ability of researchers and the public to identify medical practices associated with favorable outcomes. The large datasets collected from these databases facilitate research, which often leads to improvement in outcomes associated with best clinical and laboratory practices and may lead to more effective and cost-conscious health policy. The US Centers for Disease Control (CDC) was established in 1946 and has used statistical methods for epidemiological investigations since then, as its mission has expanded through the years to include environmental and occupational concerns. Advanced statistical methods for disease surveillance are used by policy makers to respond to developing public health problems and provide solutions, investigate changes in the patterns of data, as the basis for public health decisions, and to address health disparities.

Infertility impacts a large segment of society; estimates vary widely from 48 to180 million people worldwide causing serious consequences: economic, psychological, and physical stress, anxiety, depression, social isolation, marital discord, and divorce. Failure to conceive is a life crisis, with broad social and cultural implications. The average number of children a woman is expected to have over her lifetime now sits at 1.64 children per woman in the US [3]. This is the lowest rate recorded since the government began tracking these data in the 1930s, and it is below the so-called “replacement-level” fertility rate of 2.1. Despite this, infertility, sexual, and reproductive health are generally underfunded [31] and underserved areas of medicine.

IVF registries are available in Europe, the USA, Latin America, Japan, Africa, Canada, New Zealand, and Australia [5–7]; their strengths and limitations have been reviewed. These databases are in sharp contrast to SART CORS, which publicly reports preliminary, comprehensive, national clinic-specific linked datasets within 2 years and finalized results within 3 years after the data reporting occurs. All registries are imperfect and, as previously highlighted, have limitations [8]. As comprehensive as SART CORS is, there are missing data for some important required fields. For example, 39% of reported cycles are missing data identifying race/ethnicity details since 2006, despite it being a mandatory field [9]. These challenges may be addressed going forward by motivating more clinics to enter these data, or if still not recorded, then to employ imputation analysis. Imputation analysis is an acceptable method to overcome missing data by analyzing similar known data points and making informed extrapolations of the missing data based on an estimate of the distribution of the missing variable [10]. In addition, selected data fields concerning specific etiologies of infertility lack uniform granularity. For example, tubal factor and endometriosis offer only one data point in the SART dataset (yes/no for endometriosis listed under “reason for ART,” and in contrast to male factor, which offers detail about semen analysis, methods of sperm collection, and insemination technique. Enhanced uniform granularity of fields could provide greater definition and understanding of associations at the time of analysis. The differences in granularity may account for some of the underrepresentation of certain topic areas.

Data from the CDC report that annual rates of sexually transmitted diseases (STDs) continue to reach new heights in the USA and are predicted to continue to increase. In 2019, 2.5 million cases of the 3 most common STDs—chlamydia, gonorrhea, and syphilis—were reported, accounting for an approximately 30% increase since 2015. Cases of congenital syphilis experienced the sharpest increase (almost quadrupled) since 2015 [11]. Given this observation and the fact that STDs are a heavy contributor to tubal factor infertility, which accounts for up to 25–30% of female infertility [12], the limited research arising from SART CORS data [2] (Table 1, category tubal infertility relating to tubal ligation) highlights this as a potential underrepresented area.

It is important to consider the reasons why there is not more granularity and analysis resulting from SART CORS, as these data have the potential to influence significant health policy decisions. For example, only 30 states and the District of Columbia require sex education to be taught in public schools. Of these states, many emphasize an abstinence-only-until-marriage (AOUM) education. Data and research reflecting the significance of tubal infertility and the role STDs play in its manifestation have the potential to modify current educational standards, and the SART CORS database may play a future role in highlighting the societal burden of infertility related to a significantly increasing STD burden.

Endometriosis is a chronic abnormal cell growth and pain condition affecting ∼ 176 million women worldwide. In infertility treatment, most lines of evidence correlate the diagnosis of endometriosis with lower oocyte yields and fertilization, implantation, and pregnancy rates compared to all other infertility diagnoses [14]. However, the literature can be conflicting at times. In contrast to the previously mentioned studies, data from the SART registry shows similar implantation, pregnancy, and delivery rates for endometriosis and non-endometriosis patients [15]. The conflicting body of literature could arise from lack of granularity between the three established entities of endometriosis (peritoneal, ovarian, and deep nodular endometriosis) or stage (mild, moderate, severe). Just one publication focused exclusively on endometriosis.

The future of SART CORS

The SART CORS database is of utmost importance for monitoring the health and safety of children born after the introduction of new technologies in assisted reproduction. Clinical trials are not required to be performed for many lab-developed tests, and RCTs are necessarily limited in size and scope. The ART lab seems to be a source of tension between the aggressive marketing of potentially unproven IVF “add ons” and the urgent necessity of providing effective treatment interventions to a highly vulnerable class of patients those suffering from infertility. Certain new technologies, such as ovarian tissue cryopreservation, in vitro maturation, and artificial intelligence are increasing rapidly. Data input for post-market monitoring via the SART CORS database may seem burdensome to embryologists and clinical staff but is enormously important to follow ART pregnancies through birth, until these novel technologies can be confidently said to be safe and effective. New interventions must be monitored on a constantly shifting background of sexual health, education, and wellness policy, new embryo culture methods, and infertility treatment modalities. Current examples are new ovarian tissue cryopreservation and in vitro maturation technologies and the use of artificial intelligence technologies.

In vitro maturation technologies

Cryopreservation of ovarian tissue has been increasing globally, implemented in young women and girls diagnosed with cancer and unable to delay the onset of treatment, for transgender fertility preservation, and as a treatment option for infertile patients to avoid the side effects of taking exogenous gonadotropins. A full range of ART to support fertility options from all stages of oocytes available within ovarian tissue are being developed, such as in vitro follicle activation (IVA), oocyte, ovarian cryopreservation follicle culture, ovary, in vitro fertilization (IVF), in vitro growth (IVG), in vitro growth and maturation (IVGM), and in vitro maturation (IVM) [16].

Transplantation of cryopreserved-thawed ovarian cortex is successful and has resulted in the birth of over 130 babies [22]; it is routinely offered to safeguard the fertility of girls and young women at high risk of sterility or premature ovarian insufficiency [17] and is of growing importance to transgender care and counseling, pre-transition[37].

The effect of gender-affirming therapy on fertility remains with limited data [17, 18]. In transgender male youth (natal female), gonadotropin-releasing hormone (GnRH) agonists are often employed to prevent pubertal development in order to avoid feelings of gender dysphoria [19]. A unique challenge is presented clinically to determine the feasibility of OC without advancing gender dysphoria in the event GnRH agonists are discontinued.

A study observing oocyte cryopreservation (OC) cycles performed at SART member clinics between 2012 and 2016 found that 1.5% of OC cycles occurred in women under the age of 20 years old [20]. This absolute number was found to increase over the observation period [21]. Undoubtedly, IVF cycle numbers in persons under 20 years old will continue to rise as gender-affirming standards of care shift to prepubertal transitioning and as fertility preservation for transgender individuals is increasingly endorsed by medical associations [22]. Cases relevant to this situation emphasize the importance of the SART CORS database staying updated on emerging trends in both trans youth and oncofertility preservation patients, considering that only one study specifically examining oocyte cryopreservation was published, in our observance.

Artificial intelligence

One new intervention that is growing rapidly is the use of artificial intelligence technologies [23–25], and large datasets are becoming increasingly important for these advanced statistical methods. We think that it will be important to track the use of AI in reproductive medicine, as we do with ICSI, PGT, and other procedures, to answer to the fundamental principle “do no harm,” ensure quality standards are being met, and evidence-based medicine is being upheld. The potential benefits of AI are numerous, but so are the purported concerns. It is feared that algorithms may generalize poorly to different populations, create adverse economic conditions, make poor selection choices, and introduce a more paternalistic decision-making process [26]. It is also possible, for example, that some disadvantageous traits (such as increased large fetal size or cardiovascular complications) correlate with a higher chance of implantation or clinical pregnancy.

Artificial intelligence (AI) is reshaping healthcare in general and reproductive medicine in real time. AI, machine learning (ML), natural language processing (NLP), and deep learning (DL) enable us to identify infertility-related healthcare problems and solutions faster with more accuracy, using data patterns to make provider-informed clinical decisions. AI is able to analyze large amounts of data stored in the form of images, videos, clinical research trials, and electronic medical records (history and physical, medical claims, and so on) and can identify patterns and insights often undetectable by manual human senses. As IVF practitioners start adapting AI, how can we best monitor the use of this novel technology to ensure the highest standards of care are being met for our patients? Can a longitudinal database support discovery and monitor in real time the development of this nascent industry?

Importantly, AI has the potential to worsen healthcare inequality. We have a responsibility to design healthcare systems and tools that ensure fairness and equality are met, both in data science and in clinical studies, in order to deliver the best possible health outcomes for ART. In reproductive medicine, we run the risk of amplifying existing bias in infertility treatment worsening health inequities for certain groups of already vulnerable patients. Long-term follow-up of children born after AI for embryo selection (or any other use) can build transparency and trust—opening the “black box” of AI.

A separate but related question, should a diverse, high-quality dataset be made publicly available for studying embryo selection, implantation, or clinical pregnancy? If so, could SART CORS host it and be the gatekeeper of such a database? Now that many new AI models have been commercially developed, there is a need for external validation. The SART CORS database could evolve to include images and videos of embryo development with known outcomes for use by members for independent validation studies or to be accessed for a fee to support research priorities.

Reporting of long-term maternal and child outcomes

Monitoring long-term outcomes (LTO) of maternal health and those of future generations with respect to general health, congenital birth defects, cancer, mental health, imprinting disorders, sex chromosome abnormalities, and fetal origin of adult disease remain both a challenge and a sobering responsibility of our medical community. Countries outside the USA have addressed the issue of following LTO of pregnancies using nationally funded registries that have yielded invaluable insights into maternal and child health. One such example is the Danish National Birth Cohort established between 1996 and 2002 [27]. They recruited 100,000 mother/child pairs from all regions in Denmark during early pregnancy with prospective follow-up on child growth and development. Moreover, the Danish IVF Register established in 1994 covers information of all couples treated in Denmark with ART [28]. Unlike the USA, the Nordic countries are uniquely suited for this kind of LTO research effort because of their mandated and linked population-based registries of diseases, relatively homogenous demography, and uniform social conditions. The situation in the USA remains uniquely challenging despite our wealth and innovation. As a large heterogeneous, mobile, immigrant nation, the USA remains without national health insurance and without a uniform means of tracking births forward in time. There have been numerous efforts to address this challenge. This research approach of performing retrospective studies using de-identified datasets [29] has been remarkably productive in linking outcomes data with other registries (i.e., childhood cancer, birth defects, education, death index); however, these current methods are limited by nonuniform data standards, accuracy of identifying demographic information, lack of real-time clinical updated information as well as data validation. Linkage studies are also limited in their scope of inquiry into topics that investigators may wish to study such as the early onset of menopause, occurrence of endometriosis, or the study of those women who regret having undergone ART as a treatment of their infertility. Furthermore, as remarkable as this work has been, retrospective studies are further limited by incomplete data fields, lost to follow-up, and data entry misinterpretations and thus, are subject to confounders and bias. Efficacy of current contemporary treatments could benefit from greater rigor in study design and real-time investigation.

SART CORS has the potential to evolve as a novel integrated platform, which can provide the means to assess the value and possible maternal and/or neonatal long-term risks associated with newly developed future reproductive technologies. This added functionality for a real-time feedback loop that links patient-reported health measures with researchers who are designing and developing protocols for studies could close the gap to eventually move SART CORS from a retrospective registry engine to a prospective data collection and analysis tool. An evolved platform would create the infrastructure needed to perform unique, longitudinal birth cohort studies in the USA for decades to come. In such cohort studies, women undergoing ART for tubal disease and/or male factor could serve as a control for a woman’s fertility status. Furthermore, prospective cohort studies comparing maternal and childhood long-term health between fresh and frozen thaw cycles would be possible. This would allow SART CORS to expand the potential for critical information to be realized in shorter periods of time by stakeholders. Timelier analyses should translate into accelerated clinical and laboratory improvements in current and future approaches to ART. Furthermore, intergenerational datasets could be created that would be capable of providing insights for safe and effective current and future reproductive treatments.

In summary, the SART CORS database is an informative source of data for research, quality assurance, reporting, and post-market surveillance of novel ART technologies as performed in the USA. The database has enabled hundreds of investigators to describe novel insights that have actively impacted the clinical and laboratory practice of ART and allowed the field to continuously evolve and improve. Our analysis yielded some interesting insights that could be used by the organization to guide future internal policy. For example, although being highly cited, some higher-order topics (endometriosis, racial and ethnic disparities, and maternal obesity) had relatively few publications. This indicates that more organizational funding may need to be prioritized to these topics or research proposals in these topic areas may be specifically encouraged. Similarly, we identified a relatively low conversion rate of conference proceedings abstracts to full-length texts, indicating that more methodical support, organizational resources, or incentives may need to be implemented to gain the most value for the community from these texts. Organizational resources to provide biostatistical expertise and resources to appropriately analyze complicated and massive SART CORS datasets would be particularly helpful for investigators. We identified areas for the possible continued evolution of SART CORS in future database iterations, such as in vitro oocyte maturation technologies and integration of data collection regarding AI technologies. An area of database evolution could be to collect embryo images and videos with known outcomes, specifically to support the development or validation of AI and ML predictive models. With respect to data collection, the role of the data entry specialist is a special position requiring high skill and training. The validity and usefulness of the database hinges on these data; therefore, the data entry specialist is not just an entry-level position. Lastly, the development of a prospective platform that includes long-term maternal, neonatal, and child outcomes would provide the infrastructure and capability to perform unique, longitudinal birth cohort studies in the USA for decades to come.

Supplementary Information

Below is the link to the electronic supplementary material.

Declarations

Conflict of interest

CLC is the founder of ART Compass, a Fertility Guidance Technology, a big data and artificial intelligence software platform for IVF lab management. OT, MCA, and DBS have nothing to declare.