The objective of this study was to review the published literature on pregnancy termination following a prenatal diagnosis of Down syndrome in the United States.
The objective of this study was to review the published literature on pregnancy termination following a prenatal diagnosis of Down syndrome in the United States.
A systematic search of US English-language articles (1995–2011) was conducted to identify primary research studies that reported data for pregnancies with definitive prenatal diagnosis of Down syndrome with subsequent pregnancy termination. Studies that provided indirect estimates of pregnancy termination, such as mathematical models, were excluded. The weighted mean termination rate was calculated across studies.
Twenty-four studies were accepted. The weighted mean termination rate was 67% (range: 61%–93%) among seven population-based studies, 85% (range: 60%–90%) among nine hospital-based studies, and 50% (range: 0%–100%) among eight anomaly-based studies. Evidence suggests that termination rates have decreased in recent years. Termination rates also varied with maternal age, gestational age, and maternal race/ethnicity.
This systematic review presents the largest synthesis of United States data on termination rates following a prenatal diagnosis of Down syndrome. Evidence suggests that termination rates are lower than noted in a previous review that was based on less contemporary studies and had an international focus. Heterogeneity across studies suggests that a summary termination rate may not be applicable to the entire US population. © 2012 John Wiley & Sons, Ltd.
Down syndrome is the most common chromosomal condition in liveborns and the most common genetic cause of intellectual disability.[1-3] Prenatal screening tests for Down syndrome are noninvasive and provide an estimate of the risk of an affected pregnancy, while definitive prenatal diagnosis is made by karyotyping cultured fetal cells obtained via an invasive procedure such as chorionic villus sampling (CVS) or amniocentesis.[4, 5]
The choice about whether or not to continue a pregnancy following a prenatal diagnosis of Down syndrome is a complex decision motivated by attitudes towards termination, socioeconomic factors, ultrasound findings, and other considerations. To date, the most comprehensive summary of termination rates for Down syndrome is an international systematic review conducted by Mansfield et al. This review included ten studies on Down syndrome published between 1980 and 1998 and determined that 92% of pregnancies were subsequently terminated (95% confidence interval [92%, 93%]). Although three of these studies were from the United States, they comprised only 77 of the 5035 patients (1.5%) in the analysis, and the most recent US study was published in 1988. Although this systematic review is frequently referenced, its applicability to the current US population is unclear.
In addition to their relevance to clinicians, quantitative estimates of termination rates following a prenatal diagnosis of Down syndrome are used for both research and practical purposes, from cost-effectiveness analyses to public health planning. For example, establishing whether or not termination rates are stable or variable over time (or across unique patient subgroups) may be useful for anticipating resource needs. Practitioners, researchers, and patients are in need of an updated summary of the most contemporary and relevant data on this topic. Therefore, the objective of this study was to conduct a systematic review of published literature on termination rates among women with a prenatal diagnosis of Down syndrome in the United States. In addition, we aimed to describe temporal trends and variations in subgroups (e.g. maternal age, gestational age, race/ethnicity).
Three electronic databases (PubMed, Cochrane, and EMBASE) were searched for English-language articles published between January 1995 and April 2011. Search terms related to pregnancy (e.g. prenatal), prenatal diagnostic procedures (e.g. amniocentesis, chorionic villus sampling), and pregnancy outcomes (e.g. termination) were limited to title/abstract fields (with wildcards) and relevant medical subject headings. Reference lists also were evaluated for relevant publications.
Publications were included if they (a) reported data on definitive prenatal diagnosis of Down syndrome (i.e. a denominator), and (b) reported data on subsequent pregnancy termination (i.e. a numerator) with or without other pregnancy outcomes, such as fetal demise, stillbirths, or live births. No restrictions were made based on the representativeness of the patient population; therefore, both population-based and nonpopulation-based studies were included. In addition, studies had to directly assess pregnancy termination from sources such as electronic medical records, patient charts, or registries; studies that provided estimates of pregnancy termination were excluded (e.g. model-based studies that estimated termination rates based on expected versus observed live births). Other exclusions (e.g. animal studies, review articles) are noted in Figure 1.
All abstracts were reviewed by the first author. If it was determined that a publication might contain relevant primary data or secondary information, then the full-text article was obtained. Full-text publications were reviewed by the first author, and a final list of accepted publications was circulated to all co-authors for review.
Data elements (see Table 1) were abstracted into an evidence table by the first author. A formal quality scoring system was not used, largely because of the lack of widely accepted formal quality scoring tools for epidemiologic studies that examine prevalence, incidence, and similar measures. In lieu of a formal scoring system, we attempted to evaluate study quality terms of (1) universal factors independent of subject matter, and (2) factors dependent on our specific research question. Most critical was study design and representativeness of the general population, and publications were grouped accordingly into population-based, hospital-based, anomaly-based studies. Other factors are presented in Table 1 and include sample size, gestational age at diagnosis, and thoroughness of study reporting on demographic characteristics. The final evidence table was circulated to all co-authors for review. Findings were qualitatively summarized using a narrative approach. Similar to Mansfield et al., weighted means were calculated using the number of pregnancies with a prenatal diagnosis of Down syndrome and the number of pregnancies subsequently terminated. Reporting of study methodology and results were guided by published standards for systematic reviews.[8, 9]
|Publication information||First author|
|Year of publication|
|Data source and methodology||Study location|
|Population-based vs hospital-based|
|Name of institution or registry|
|Relevant inclusion and exclusion criteria|
|Study population and demographics||Summary of study population|
|Maternal age (mean, SD, range, categorical data)|
|Ethnicity (categorical data)|
|Gestational age (mean, SD, range, categorical data)|
|Gestational age summary|
|Sample size and PND||Number of pregnancies|
|Number of pregnancies with prenatal screening (separated for first trimester and second trimester)|
|Number of pregnancies with a positive screen for Down syndrome|
|Number of pregnancies with PND (separated for diagnostic procedure: CVS, amniocentesis, PUBS)|
|Number of pregnancies with PND of Down syndrome (separated for diagnostic procedure: CVS, amniocentesis, PUBS)|
|Outcomes among pregnancies with PND of Down syndrome||Termination or elective abortion (n, %)|
|Fetal demise or spontaneous abortion (n, %)|
|Live birth (n, %)|
|Unknown outcome or loss to follow-up (n, %)|
|Subgroup analyses||Maternal age|
The search strategy identified 6570 publications, and ten additional publications were identified from reference lists. Among 1977 US abstracts, 308 publications were retrieved for full-text review, and 24 were accepted for inclusion (see Figure 1).
One of the most common reasons for exclusion was lack of data on prenatal diagnosis for aneuploidy or Down syndrome specifically. For example, many studies examined ultrasound-based prenatal diagnosis of physical abnormalities for which amniocentesis is not indicated. Other studies estimated the prevalence of Down syndrome based on live births, stillbirths, and terminations but failed to provide the total number of pregnancies with prenatal diagnosis. Among studies that presented data on prenatal diagnosis, many were excluded because of lack of longitudinal follow-up.
The 24 accepted publications were categorized based on study population (see Table 2), including population-based studies (e.g. statewide birth defect registries), hospital-based studies (e.g. single-institution studies), and anomaly-based studies (e.g. karyotyping following abnormal ultrasound). Two population-based studies presented data on the same study population; we present this study as a single row, and data from both publications were used to inform our abstraction.[10, 11]
|First author (Year)||State||Institution or registry name||Study years||Study population||Maternal age||% Caucasian||Gestational age||Down syndrome PND||Subgroup data|
|Kazerouni (2009)()||CA||California Chromosomal Defect Registry (CCDR)||2005–2007||Pregnant women in California undergoing second trimester triple screening||> 35 years, 13.7%||24||Second trimester (most 15–20 weeks)||466||No|
|Won (2005)()||CA||California Expanded AFP Screening Program, State of California Department of Health Service||1995–2000||Pregnant women with trisomy 21 identified by PND following high-risk serum screening results||—||—||Second trimester||1408||No|
|Bishop (1997)()||CA||California Birth Defects Monitoring Program (CBDMP), California's Department of Health Services Genetic Disease Branch (GDB)||1989–1991||Pregnant women with trisomy 21 identified by PND||—||—||Any time during pregnancy||531||No|
|Forrester (2002)()||HI||Hawaii Birth Defects Program||1986–1999||Pregnant women with autosomal abnormality||—||—||Any time during pregnancy||173||No|
|Forrester (1998, 1999)[10, 11]||HI||Hawaii Birth Defects Program||1987–1996||Pregnancies with one or more moderate-to-severe birth defects||> 35 years, 50.6%||30||Any time during pregnancy||131||Yes|
|Knight (2005)()||ME||Maine Bureau of Vital Records, Diagnostic Laboratories||2001–2003||Pregnant women who had an integrated serum screening test||> 35 years, 11.3%||98||Second trimester||15||No|
|Shaffer (2006)()||CA||University of California at San Francisco||1983–2003||Pregnant women with common aneuploidies identified by PND||> 35 years, 80.9%||47||Any time during pregnancy||449||Yes|
|Benn (1998)()||CT||University of Connecticut||1992–1996||All singleton pregnancies who had second trimester triple screening w/o prior CVS/amnio||> 35 years, 9.1%||73||Second trimester||27||No|
|Wray (2007)()||DC||Georgetown University Hospital||2002–2004||Pregnant women who had prenatal chromosome analysis||—||—||—||10||No|
|Caruso (1998)()||MA||Brigham and Women's Hospital||1972–1974 1979–1994||Pregnancies with a PND of Down syndrome before 24 weeks' gestation||—||—||Second trimester (< 24 weeks)||31||No|
|Britt (2000)()||MI||Wayne State University||1989–1998||Pregnant women with trisomy 21 identified by PND||> 38 years, 48%||—||Second trimester (< 24 weeks)||144||Yes|
|Kramer (1998)()||MI||Wayne State University||1989–1997||Viable pregnancies with trisomy 21 for which termination was available||—||61||First and second trimester (9–24 weeks)||145||Yes|
|Perry (2007)()||NY||University of Rochester Strong Memorial Hospital||1997–2005||Pregnant women with trisomy 21 identified by PND||> 35 years, 61%||87||First and second trimester (< 24 weeks)||59||Yes|
|Perni (2006)()||NY||New York Weill-Cornell Medical Center||2003–2004||Pregnant women who had first trimester screening||33.0 years (median)||73||First trimester (11–13 weeks)||22||No|
|Horger (2001)()||SC||University of South Carolina||1972–2000||Pregnant women undergoing amniocentesis||—||—||Second trimester||37||Yes|
|Shulman (2000)()||IL||University of Illinois at Chicago||NR||Fetal ‘space-suit hydrops’||22–39 years||—||First trimester (9-14 weeks)||5||No|
|Wickstrom (1996)()||IL||Northwestern University Medical School||1991–1994||Fetal isolated pyelectasis||—||—||Second and third trimester (15+ weeks)||1||No|
|Shipp (1998)()||MA||Brigham and Women's Hospital||1988–1996||Fetal isolated clubfoot||—||—||Second and third trimester (14+ weeks)||1||No|
|Tomlinson (1997)()||MI||Hutzel Hospital (Detroit)||1992–1994||Fetal isolated mild ventriculomegaly (11–15 mm)||—||—||Second and third trimester (after 14 weeks)||2||No|
|Koklanaris (2005)()||NY||Bellevue Hospital Ultrasound Unit||2001–2004||Fetal isolated echogenic intracardiac focus||—||1.8||Second trimester||2||No|
|Sivanandam (2006)()||NY||New York-Presbyterian Hospital||1994–2003||Fetal conotruncal malformations||—||—||Second and third trimester (16+ weeks)||3||No|
|Zyblewski (2009)()||SC||Medical University of South Carolina||1998–2006||Fetal critical congenital heart disease||—||59||Second and third trimester (24+ weeks)||21||No|
|Byers (2009)()||TX||University of Texas Medical Branch||1997–2007||Fetal intra-abdominal umbilical vein varix||18–43 years||—||Second and third trimester (18+ weeks)||1||No|
Of the seven publications from population-based studies, three were from California, three from Hawaii, and one from Maine (see Table 3). However, because of cross-coverage of time periods, only five unique, nonoverlapping populations are represented. The gestational age of patients differed across studies, as did the number of pregnancies with a prenatal diagnosis of Down syndrome, which ranged from 15 pregnancies over a 2-year period in Maine to 1408 pregnancies over a 5-year period in California. Termination rates ranged from a low of 61.4% in a relatively recent study from California (2005–2007) to a high of 93.3% in Maine (2001–2003). The weighted mean termination rate was 74.2% using the five unique, nonoverlapping populations and 67.0% based on the single most recent study from each of the three states.
|First author (Year)||State||Study years||Gestational age||Pregnancies with Down syndrome by PND||Termination following PND of Down syndrome|
|Kazerouni(2009) ()||CA||2005–2007||Second trimester (most 15–20 weeks)||466||286||61.4|
|Won (2005)()||CA||1995–2000||Second trimester||1408||1016||72.2|
|Bishop (1997)()||CA||1989–1991||Any time during pregnancy||531||469||88.3|
|Forrester (2002)()a||HI||1986–1999||Any time during pregnancy||173||138||79.8|
|Forrester (1998, 1999)[10, 11]a||HI||1987–1996||Any time during pregnancy||131||110||84.0|
|Knight (2005)()||ME||2001–2003||Second trimester||15||14||93.3|
|Weighted average of all studies (n = 6)||2724||2033||74.6|
|Weighted average of all studies from unique populations (n = 5)b||2593||1923||74.2|
|Weighted average of single most recent study from each state (n = 3)c||654||438||67.0|
Nine single-institution, hospital-based publications were identified, although only eight unique, nonoverlapping populations were represented (see Table 4). Similar to the population-based studies, the hospital-based studies were heterogeneous, making comparisons difficult. The number of pregnancies with a prenatal diagnosis of Down syndrome ranged from ten pregnancies over a 2-year period at Georgetown University Hospital (2002–2004) to 449 pregnancies over a 20-year period at the University of California at San Francisco (UCSF, 1983–2003). Termination rates following a prenatal diagnosis of Down syndrome varied from a low of 60.0% in the Georgetown study to a high of 89.6% at Wayne State University (WSU, 1989–1998). The weighted mean termination rate was 85.1% using the eight unique, nonoverlapping populations.
|First author (Year)||State||Institution||Study years||Study population||Gestational age||Pregnancies with Down syndrome by PND||Termination following PND of Down syndrome|
|Shaffer (2006)()||CA||University of California at San Francisco||1983–2003||Pregnant women with common aneuploidies identified by PND||Any time during pregnancy||449||391||87.1|
|Benn (1998)()||CT||University of Connecticut||1992–1996||All singleton pregnancies who had second trimester triple screening without prior CVS/amniocentesis||Second trimester||27||23||85.2|
|Wray (2007)()||DC||Georgetown University Hospital||2002–2004||Pregnant women who had prenatal chromosome analysis||Any time during pregnancy||10||6||60.0|
|Caruso (1998)()||MA||Brigham and Women's Hospital||1972–1974, 1979–1994||Pregnancies with a PND of Down syndrome before 24 weeks' gestation||Second trimester (< 24 weeks)||31||27||87.1|
|Britt (2000)()||MI||Wayne State University||1989–1998||Pregnant women with trisomy 21 identified by PND||Second trimester (< 24 weeks)||144||129||89.6|
|Kramer (1998)()||MI||Wayne State University||1989–1997||Viable pregnancies with trisomy 21 for which termination was available||First and second trimester (9–24 weeks)||145||126||86.9|
|Perni (2006)()||NY||New York Weill-Cornell Medical Center||2003–2004||Pregnant women who had first trimester screening||First trimester (11–13 weeks)||22||19||86.4|
|Perry (2007)()||NY||University of Rochester Strong Memorial Hospital||1997–2005||Pregnant women with trisomy 21 identified by PND||First and second trimester (< 24 weeks)||59||43||72.9|
|Horger 2001)()||SC||University of South Carolina||1972–2000||Pregnant women undergoing amniocentesis||Second trimester||37||25||67.6|
|Weighted average of all studies (n = 9)||924||789||85.4|
|Weighted average of all studies from unique populations (n = 8)a||779||663||85.1|
The eight studies that examined karyotyping following abnormal ultrasound findings were generally small studies with very narrow inclusion criteria (see Table 5). Specific ultrasound anomalies varied from soft markers to more severe abnormalities including hydrops and heart defects. Seven of the eight studies involved second and/or third trimester pregnancies beyond 14 weeks' gestation. The weighted mean termination rate for all eight studies was 50%, although this was heavily influenced by the largest study, which found a 38% termination rate among 21 pregnancies with critical heart defects. Other studies included three or fewer cases of prenatally diagnosed Down syndrome with termination rates ranging from 0%–100%.
|First author (Year)||State||Institution||Study years||Study population||Gestational age||Pregnancies with Down syndrome by PND||Termination following PND of Down syndrome|
|Shulman (2000)()||IL||University of Illinois at Chicago||NR||Fetal ‘space-suit hydrops’||First trimester (9-14 weeks)||5||5||100|
|Wickstrom (1996)()||IL||Northwestern University Medical School||1991–1994||Fetal isolated pyelectasis||Second and third trimester (15+ weeks)||1||1||100|
|Shipp (1998)()||MA||Brigham and Women's Hospital||1988–1996||Fetal isolated clubfoot||Second and third trimester (14+ weeks)||1||1||100|
|Tomlinson (1997)()||MI||Hutzel Hospital (Detroit)||1992–1994||Fetal isolated mild ventriculomegaly (11–15 mm)||Second and third trimester (after 14 weeks)||2||2||100|
|(after 14 weeks)|
|Koklanaris (2005)()||NY||Bellevue Hospital Ultrasound Unit||2001–2004||Fetal isolated echogenic intracardiac focus||Second trimester||2||1||50|
|Sivanandam Sivanandam(2006)()||NY||New York-Presbyterian Hospital||1994–2003||Fetal conotruncal malformations||Second and third trimester (16+ weeks)||3||0||0|
|Zyblewski (2009)()||SC||Medical University of South Carolina||1998–2006||Fetal critical congenital heart disease||Second and third trimester (24+ weeks)||21||8||38|
|Byers (2009)()||TX||University of Texas Medical Branch||1997–2007||Fetal intra-abdominal umbilical vein varix||Second and third trimester (18+ weeks)||1||0||0|
|Weighted average of all studies (n = 8)||36||18||50|
Only one study specifically examined temporal variations in termination rates. This hospital-based study from the University of South Carolina found that termination rates decreased from 78.6% (22 of 28) in 1972–1996 to 33.3% (3 of 9) in 1997–2000. Many of the remaining studies had overlapping study periods and clinical and geographic heterogeneity that precluded evaluation of temporal trends. However, the three population-based California studies, which presented data on mutually exclusive populations from different time periods, demonstrated a statistically significant decrease in termination rates over time, from 88.3% (1989–1991) to 72.2% (1995–2000) to 61.4% (2005–2007) (χ2 test for trend = 37.196, df = 2, p < 0.0001).[14-16]
Five studies presented data on termination rates based on maternal age, gestational age, and/or race/ethnicity[11, 17-20]. Four studies examined termination rates based on maternal age, including three studies that reported nonsignificant trends or statistically significant associations for higher termination rates with increasing maternal age. The fourth study by Britt et al.(2000) from WSU reported similar termination rates for women aged 38 years and older (90%) and those aged 37 years or younger (89%). However, an overlapping study from WSU by Kramer et al.(1998) documented that women who terminated tended to be older (mean 36.3 years for termination, 31.9 years for continuation, p = 0.003). Perry et al.(2007) reported similar findings, although the difference was not statistically significant (mean 36.1 years for termination, 32.3 years for continuation, p = 0.059). Forrester et al. (1999) reported that termination rates were higher for women in their 30s (30–34 years, 92% [56 of 61]; 35–39 years, 94% [17 of 18]) compared with women in their 20s (20–24 years, 0% [0 of 5]; 25–29 years, 70% [7 of 10]) or 40s (81% [30 of 37]).
Three hospital-based studies documented termination rates based on gestational age,[17, 18, 20] and higher termination rates were consistently associated with earlier gestational age. One study from WSU reported that 93% of women at 16 weeks' gestation or less chose to terminate the pregnancy versus 85% of those at 17 weeks' or greater gestation. The other WSU study using a related study population noted that patients who elected to terminate following a prenatal diagnosis of Down syndrome were earlier in the pregnancy compared with patients who elected to continue (15.3 weeks vs 19.0 weeks, p < 0.001). A third study from the University of Rochester also found that women who terminated the pregnancy tended to be at an earlier gestational age versus those who continued (18.2 weeks vs 19.1 weeks), although this difference was not statistically significant (p = 0.16).
Three studies evaluated racial and ethnic differences in termination rates.[11, 19, 20] A hospital-based study from UCSF found that termination following a prenatal diagnosis of Down syndrome varied by ethnic group, with Hispanic and Filipino women having lower termination rates (70% and 80%, respectively) than Caucasian (90%), African American (92%), and Asian (87%) women (χ2 across all five groups, p = 0.006). A population-based study from Hawaii reported that Pacific Islanders had the lowest termination rate (56%), followed by Filipinos (82%), Caucasians (84%), and Far East Asians (93%). Although the UCSF and Hawaiian studies examined the proportion of Caucasian women who decided to terminate following prenatal diagnosis, a study from the University of Rochester evaluated the proportion of women who terminated that were Caucasian. They found that 86% of women who terminated were Caucasian and 88% of women who continued were Caucasian (p = 0.33).
This systematic review provides the largest qualitative and quantitative summary of Down syndrome termination rates relevant to a contemporary population in the United States. High-quality data from population-based studies in three states indicate that the weighted mean termination rate following a prenatal diagnosis of Down syndrome was 67%, with a range of 61%–93%. Data from hospital-based studies in six states plus the District of Columbia suggest that the termination rate was 85%, with a range of 60%–90%. Finally, the mean termination rate among anomaly-based studies was 50% and ranged from 0%–100%. The evidence presented here also suggests that termination rates vary according to maternal age and race and gestational age.
The estimated termination rates following a prenatal diagnosis of Down syndrome presented in this review are appreciably lower than the 92% termination rate determined by Mansfield et al. We hypothesize that this difference is the result of lower termination rates in the United States in recent years. The three US studies reviewed by Mansfield et al. were published in the 1980s.[21-23] Evidence in our report from three population-based California studies supports the notion that US termination rates decreased during the 1990s and early 2000s. Other researchers have found international variation in termination rates following prenatal diagnosis of Down syndrome. A recent analysis of data from 20 registries in the International Clearinghouse for Birth Defects Surveillance and Research (ICBDSR, 1993–2004) found that Down syndrome termination rates per 10 000 live births were lower in North America than Europe and Australia. Thus, the termination rates presented in this report may be lower because we focused exclusively on US studies, while Mansfield's review was based on earlier studies and the termination rate for Down syndrome was heavily influenced by a single study from the UK.
Of note, the implementation of prenatal screening for Down syndrome has had a disproportionate impact on prenatal diagnosis among younger women, as demonstrated by findings from the England–Wales National Down Syndrome Cytogenetic Register (NDSCR) (1989–2008).[25, 26] Prenatal screening allows for more refined risk stratification among older women, all of whom would be considered at high risk based on age alone. Although prenatal screening has supplanted maternal age as the primary indication for prenatal diagnosis, the overall proportion of pregnancies diagnosed prenatally has remained relatively constant (60%–70% from 1989 through 2008). Historically, younger women at lower age-based risk were not routinely offered invasive diagnostic testing. However, prenatal screening reclassifies a subgroup of these patients as higher risk, and, consequently, the proportion of pregnancies with Down syndrome that are prenatally diagnosed among younger women has increased, from 3% in 1989/1990 to 43% in 2007/2008.
Changes in patient uptake of invasive diagnostic procedures directly affect the denominator used in calculating termination rates. Of note, women who elect to undergo invasive testing are a self-selecting population that may not be representative of all pregnant women. Women who choose to have an amniocentesis differ from women who decline amniocentesis with regard to several factors, including the need for certainty about a pregnancy with Down syndrome, post-screen anxiety level, perception of amniocentesis-associated miscarriage risk, attitudes about pregnancy termination, and religious beliefs. Evidence also suggests that utilization of CVS and amniocentesis has changed over time, with decreases beginning in the early 1990s following the widespread implementation of second trimester maternal serum screening and ultrasonography.[28, 29] A study by Nakata and colleagues (2010) suggests that this trend has reversed in recent years because of several factors: (a) the 2007 American College of Obstetrics and Gynecology (ACOG) Practice Bulletin on screening for fetal chromosomal abnormalities, which recommended that screening and invasive testing be made available to all pregnant women regardless of age, (b) increased access to first trimester screening options, and (c) the 2008 ACOG Practice Bulletin on invasive prenatal testing for aneuploidy, which stated that procedure-related risks were lower than previously suggested.
With these issues in mind, we considered explanations for a decrease in termination rates following a prenatal diagnosis of Down syndrome. First, the studies in this review provide evidence of a trend for lower termination rates among younger women. Because prenatal screening has resulted in an increase among younger women who elect to have prenatal diagnosis, this may have acted to decrease the overall termination rate. Second, although raising a child with Down syndrome has notable challenges, medical management has progressed in recent decades and individuals with Down syndrome are living longer and healthier lives.[30-34] In parallel with these medical advances have come additional educational, social, and financial support for families of children with Down syndrome.[35-37] Thus, some women may elect to have a prenatal diagnostic procedure not because they plan to terminate a pregnancy, but because they want to alleviate uncertainty, prepare emotionally, and/or make specific delivery plans.
Of note, none of the studies in this review presented data collected subsequent to the 2007 and 2008 ACOG Practice Bulletins on screening and invasive testing for aneuploidy, and it is unclear what impact, if any, these have had on termination rates. Historically, many women who would not terminate a pregnancy with Down syndrome have decided against invasive testing because of the procedure-related risk of miscarriage. The ACOG finding that these risks are lower than previously stated may help to make prenatal diagnosis a more realistic option for women who desire diagnostic certainty to make emotional and medical preparations. In this context, uptake of prenatal diagnosis would be expected to increase (as demonstrated by Nakata and colleagues (2010) and termination rates would be expected to decrease.
Inaccurate estimation of termination rates may affect other research studies. For example, many economic evaluations of prenatal screening and diagnosis programs that focus on the financial costs of conditions associated with physical and/or intellectual disabilities (e.g. Down syndrome, spina bifida) assume that the condition is severe enough that the vast majority of patients would elect to terminate a pregnancy with the condition. If termination rates are lower in actual practice, a proportion of the financial costs would fail to be averted and the cost-effectiveness of screening programs may be overestimated.
Although this is the largest systematic review of termination rates for Down syndrome in the United States, opportunities for additional epidemiologic studies are evident. All US states collect data on Down syndrome diagnoses via birth defect surveillance programs sponsored by the National Birth Defects Prevention Network. Although leveraging these data would provide additional insight into the broader US population, financial and practical obstacles exist. Although states typically include live births, fetal deaths, and elective terminations of Down syndrome, ascertainment of prenatal diagnoses requires expansion to prenatal diagnostic facilities, cytogenetic laboratories, and genetic counseling facilities, and not all states obtain data from these sources. Many programs also do not include elective terminations prior to 20 weeks' gestation.
Individual providers or hospital systems may consider collecting and analyzing their own data; although not widely generalizable and limited by sample size, findings would be applicable to their distinct patient populations. Whether a population-based or hospital-based approach is used, the ideal study design would follow a longitudinal cohort of women as soon as a pregnancy is identified (e.g. a pregnancy registry). Such an approach would allow for a broad description of screening uptake, diagnostic uptake, and decision-making following prenatal diagnosis and other pregnancy outcomes such as spontaneous miscarriage. Given that few studies reported data on factors that may affect termination rates, future studies should evaluate demographic variability in termination rates as permitted by the data.
This systematic review has limitations that should be acknowledged. Although some of these limitations pertain to our methodological approach, most are constraints inherent in the studies themselves. First, we noticed considerable variation across studies in terms of the indication for prenatal diagnosis, gestational age, geography, and study years. Although such clinical heterogeneity would typically preclude quantitative analysis, we wanted our findings to be comparable to those of Mansfield et al. However, this approach implies that a single summary termination rate is applicable to the entire United States and does not adequately address regional variations and other observed differences. Similarly, the studies in this review may not be representative of the US population as a whole. Among population-based and hospital-based studies, data were from eight states plus the District of Columbia. In addition, most anomaly-based studies had very narrow populations of women at advanced gestational age, which limits their generalizability.
Second, the target study population varied across studies. Although all studies reported the number of pregnancies with a prenatal diagnosis of Down syndrome, in most studies, the rate of uptake of prenatal diagnosis among all pregnant women (as opposed to those with a high-risk screening result, for example) was not presented.
Third, several studies examined a broad date range (e.g.13-year range from 1986–1999 in Forrester and Merz (2002) from Hawaii) and variation over time typically was unexamined. If temporal variations are present, then a broad study period may present a skewed estimate of current termination rates (e.g. only the most recent years of a 10-year study might be applicable).
Finally, as with any systematic review, the possibility for publication bias is a consideration (i.e. the studies identified in this review differ systematically from studies that were inadvertently missed). Although we did not include conference abstracts or unpublished data, we attempted to minimize publication bias by using a broad, comprehensive approach to identifying relevant studies. As noted previously, the studies in this review may not typify the broader US population, and population-based studies from states not represented in this review would provide additional understanding on this topic.
This systematic review presents the largest synthesis of data on termination rates following a prenatal diagnosis of Down syndrome in the United States. Our evidence suggests that termination rates are lower than noted in previous reports (67%–85% in this review vs 92% in Mansfield et al. and that termination rates vary with maternal age, maternal race/ethnicity, and gestational age. Evidence also suggests that termination rates have decreased in recent years, which may reflect progress in medical management for individuals with Down syndrome and advances in educational, social, and financial support for their families. Importantly, the range of termination rates observed across studies suggests that a single summary termination rate may not be applicable to the entire US population and would not adequately address regional and demographic differences among pregnant women.