Stillbirth, defined as fetal death at 20 weeks of gestation or greater, affects approximately 1 in 160 pregnancies in the United States.1 Despite the relative frequency of the condition, the optimal evaluation for the determination of the cause of death remains unclear because stillbirth can be caused by a large number of heterogeneous, often uncommon conditions. In addition, many cases are multifactorial or may be associated with risk factors rather than causes. The approach to determine the cause(s) of stillbirth currently recommended by the American College of Obstetricians and Gynecologists is quite extensive2; however, the yield of such an approach must be weighed against the cost, patient and health system burden, and potential for false-positive results leading to harm in subsequent pregnancies. Thus, our purpose was to assess the utility of diagnostic tests for the evaluation of stillbirth in a large, well-characterized, geographically, racially, and ethnically diverse population-based cohort of U.S. stillbirths.
MATERIALS AND METHODS
This was a secondary analysis of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Stillbirth Collaborative Research Network study. The Stillbirth Collaborative Research Network was a multicenter, population-based cohort that attempted to enroll all women delivering stillborn fetuses and a representative sample of live births between March 2006 and September 2008 in five specific catchment areas. These catchment areas were defined by maternal residence in five geographic areas defined by county lines including all of Rhode Island and portions of Massachusetts, Georgia, Utah, and two locations in Texas. Recruitment occurred at 59 hospitals, which, in total, average more than 80,000 deliveries per year to catchment area residents.
Study design, methodology, and sample size calculations have been previously described in detail.3,4 Stillbirth was defined as birth at or after 20 weeks of gestation with Apgar scores of 0 at 1 and 5 minutes and no signs of life on direct observation. Fetal deaths at an estimated gestational age of 18 or 19 weeks based on uncertain dating criteria were also enrolled so as to include all potential cases occurring at 20 weeks of gestation or later. Gestational age was determined using data from assisted reproductive technology, first day of last menstrual period, and ultrasonography.4 Deliveries resulting from termination of a live fetus were excluded. The study was approved by the institutional review boards of each clinical site and the Data Coordinating and Analysis Center, and all mothers gave written informed consent.
Enrolled mothers with stillbirth were interviewed, their medical records abstracted, and, if they consented, their fetuses underwent autopsy, placental pathology, and attempted karyotype.3 The maternal interviews were largely conducted in-hospital and gathered information regarding social history, reproductive history, pregnancy complications, psychosocial data, personal and family medical history, drug exposure, and feelings regarding the index pregnancy. The fetal autopsy and placental examinations were performed by perinatal pathologists using systematic protocol and diagnostic criteria that had been determined before initiation of the study.5,6 Additional clinically indicated testing was performed at the discretion of the primary clinician, although the Stillbirth Collaborative Research Network physicians informed the treating physicians of a series of analyses including antibody screen, syphilis serology, toxicology screen, Kleihauer-Betke or other screen for fetal–maternal hemorrhage, and antiphospholipid antibodies.3,7 When possible, maternal blood, fetal umbilical blood, placental tissue, and fetal tissue were collected. These samples were stored for additional diagnostic testing that was not performed at the time of delivery. These tests included antibody screen, syphilis serology, parvovirus serology, anticardiolipin antibodies, fructosamine, and chromosomal microarray. For the current analysis, only stillbirths with complete fetal autopsy and placental histopathology were included. Results of karyotype and microarray were combined and considered collectively as “genetic testing results.” For example, if a test did not have a karyotype result but had an abnormal microarray, it was recorded as an abnormal or positive genetic test. Determination of the presence of antiphospholipid antibodies was performed with anticardiolipin antibody and β-2 glycoprotein antibody testing as previously reported.8 We used this measure for our analysis of antiphospholipid antibody status.
A cause of death for each stillbirth was previously determined using the Initial Causes of Fetal Death instrument developed by the Stillbirth Collaborative Research Network.9 This evidence-based system allows for potential etiologies of stillbirth to be classified as a probable cause, possible cause, or as a present condition based on the best available scientific evidence. A probable cause of death was defined as one with a high likelihood of causing the stillbirth. A possible cause was one for which there is reasonable certainty it is involved in the pathophysiologic pathway leading to stillbirth.9 Potentially important conditions that occurred but did not meet criteria for probable or possible causes of death were recorded as present. This breakdown reflects the inherent uncertainty that exists in the determination of a potential cause of stillbirth. The Initial Causes of Fetal Death system divides potential causes into broad categories including: maternal medical conditions, obstetric complications, placental conditions, infection, fetal genetic or structural abnormalities, hypertensive disorders of pregnancy, cord accidents, and other conditions that do not fit into the previously mentioned categories. Detailed information regarding the criteria for each category has been published.9
To determine how often each diagnostic test was useful, we developed an a priori definition of “usefulness.” We defined a useful test as one that helped to establish a probable or possible cause of death as designated by Initial Causes of Fetal Death or one that excluded a suspected cause of death based on the clinical scenario. The clinical narrative for each participant was reviewed to assess clinical characteristics that may indicate a potential cause of death. After this, each diagnostic test for potential causes of stillbirth including fetal autopsy, placental pathology, and laboratory assay results was reviewed. A description of the criteria used to designate an abnormal or positive test has been previously published.4 Tests that established a probable or possible cause of death, and thus were considered a pertinent positive, were those that either 1) confirmed a probable or possible cause of death that was suspected based on the clinical scenario or 2) identified an unanticipated cause. Some tests excluded a suspected cause based on the clinical situation and these were considered a pertinent negative.
A schematic of the approach to determine whether a test was useful in the evaluation of potential causes of stillbirth is shown in Figure 1. For example, a negative test for fetal–maternal hemorrhage was considered useful as a pertinent negative if there was vaginal bleeding, sinusoidal fetal heart rate tracing, or fetal hydrops. However, a negative test for fetal–maternal hemorrhage was not considered useful if there was no suspicion of fetal–maternal hemorrhage based on available data. A similar approach was used by Korteweg et al10 to assess antepartum stillbirths in the Netherlands. Assessment of utility for each diagnostic test was performed independently by two Stillbirth Collaborative Research Network physicians with subsequent discussion and agreement on each patient.
The percent of patients in which a test was useful was calculated to determine which components of the stillbirth evaluation were of the highest yield in determination of a probable or possible cause of death or to eliminate a suspected cause. Confidence intervals were calculated using Poisson methodology to determine statistical significance. We performed additional analysis after stratification by clinical scenario available at the time of delivery to determine which tests were most helpful for patients presenting with specific clinical characteristics. These included fetal growth restriction or small for gestational age by birth weight, maternal hypertensive disorders, suspected fetal anomalies, obstetric complications including preterm labor, chorioamnionitis and preterm prelabor rupture of membranes, and intrapartum stillbirth. These were chosen because they represented the most common clinical presentations in our cohort. To simulate real-world conditions, we included participants in each clinical scenario based only on information available at the time of delivery. For example, participants included in the fetal growth restriction or fetal anomaly groups were those that were suspected by ultrasonogram or gross examination of the fetus rather than those identified after fetal autopsy. Accordingly, these scenarios should not be considered causes of death and did not use Initial Causes of Fetal Death. We also assessed the utility of tests by gestational age using the following categories to comport with our prior analyses: 20.0–23 6/7, 24.0–27 6/7, 28.0–31 6/7, 32.0–36 6/7, and 37.0 weeks of gestation or greater.4 An additional analysis was performed for patients with stillbirth that occurred during the intrapartum period.
The Stillbirth Collaborative Research Network identified 953 eligible women with a total of 972 stillbirths. Of these women, 290 were not included; 126 of these were not approached and 164 declined participation. This resulted in 663 women (70%) with 676 stillbirths enrolled. Five hundred sixty women (84%) consented to fetal autopsy (complete or partial) and of these, 500 women with 512 stillbirths consented to complete fetal autopsy and were included in the current analysis. The demographics of this population have been previously described.4
A probable Initial Causes of Fetal Death cause of death was identified in 312 (60.9%) and a possible cause in 78 (15.2%). More than one probable or possible cause was identified in 161 (31.4%) of these stillbirths.4 Of the 122 (23.8%) participants without a probable or possible cause, 105 (20.5%) had a condition present and 17 (3.3%) had no Initial Causes of Fetal Death condition present.
If a test was not performed at the time of presentation, results from stored maternal serum samples were used when possible. Genetic testing with karyotype, microarray, or both was performed in 96.7% of participants. Antiphospholipid antibody testing was performed in 90.8% of participants and greater than 90% had glucose testing, antibody screen, syphilis, and parvovirus serology (Table 1). Testing for fetal–maternal hemorrhage was performed in only 43.6% of participants because it could not be completed on stored maternal samples.
Table 1 demonstrates the percent of participants in which each diagnostic test was helpful. The test that was useful in the largest proportion of participants was placental pathology (64.6%, 95% confidence interval [CI] 57.9–72.0). Fetal autopsy was the next most useful test in 42.4% (95% CI 36.9–48.4) of stillbirths. Genetic testing, including both karyotype and microarray, was useful for several participants (11.9%, 95% CI 9.1–15.3) followed by testing for antiphospholipid antibodies (11.1%, 95% CI 8.4–14.4). The remaining tests were helpful in a relatively small subset of participants. Most tests were useful primarily in cases of pertinent positive results. Tests that were most often useful as pertinent negatives were tests for fetal–maternal hemorrhage and antiphospholipid antibodies.
The utility of tests stratified by clinical scenario is displayed in Table 2. For those participants affected by growth restriction, the most useful tests were placental pathology (88.7%) and fetal autopsy (79.2%). Abnormal placental findings often included substantial fibrin deposition and fetal thrombotic vasculopathy. Antiphospholipid antibody screening was also useful in a substantial portion of the growth-restricted participants (32.1%), often as a pertinent negative result in early-onset or severe growth restriction. Genetic testing, including both karyotype and microarray, was helpful in 26.4% of participants, the majority of which had suspected fetal anomalies based on antenatal ultrasonography or abnormal maternal serum screening.
Placental pathology and fetal autopsy were again the most useful tests (90.0% and 50.0%, respectively) in pregnancies affected by hypertensive disorders. Placental histologic abnormalities were similar to those noted in fetal growth restriction. Screening for antiphospholipid antibodies was also useful in many of these pregnancies (28.0%) including those with early-onset hypertensive disorders before 34 weeks of gestation.
For pregnancies with suspected fetal anomalies, based on antenatal ultrasonography or genetic screening results, fetal autopsy was the most useful test (90.3%) followed by genetic testing (87.1%). Examples of participants in which these tests were not helpful were those in which the fetus was too macerated for adequate examination or when karyotype and microarray did not yield results. Placental pathology was also helpful in a large proportion of suspected malformation participants (41.9%), albeit less frequently than for other scenarios. Placental pathology was helpful because many of these participants had findings consistent with placental insufficiency such as syncytial knots or infarct as well as possible abruption.
Tests that were most useful in pregnancies affected by preterm labor, chorioamnionitis, or preterm prelabor rupture of membranes included placental pathology (80.5%) and fetal autopsy (44.2%). Genetic testing and testing for fetal–maternal hemorrhage were both helpful in 5.2% of participants in this cohort because they identified genetic anomalies or fetal–maternal hemorrhage in participants in which these findings were not otherwise expected. Examples of pertinent positive results include chorioamnionitis in the placenta membranes and cord and pneumonia or positive bacterial cultures identified by fetal autopsy.
Of the 512 stillbirths included, 95 occurred during the intrapartum period. In this group, placental pathology and fetal autopsy remained the most useful tests in 85.3% and 43.2%, respectively. Testing for fetal–maternal hemorrhage was also useful in 10.5% of participants, a higher proportion than that observed in some other clinical scenarios. Testing for antiphospholipid antibodies, genetic testing, and glucose screening was of less utility in this group.
There were 115 participants in which no clinical clues were present on admission. These patients did not have elements from their history or presentation that aroused suspicion for a potential cause of death. In this group, the most useful tests were placental pathology (78.3%), fetal autopsy (49.6%), genetic testing (13.9%), and antiphospholipid antibody testing (11.3%).
After stratification by gestational age, fetal autopsy and placental pathology remained the most helpful tests at all gestations (Fig. 2). Fetal autopsy was the most useful at early gestational ages, often revealing evidence of sepsis in participants with preterm prelabor rupture of membranes and chorioamnionitis as described previously. Placental pathology became more useful with increasing gestational age, whereas other tests did not vary significantly by gestational age at stillbirth.
The results of our subanalysis by clinical presentation were compiled into an algorithm to guide clinical testing (Fig. 3). We included each clinical scenario and recommended tests with the highest yield, defined as greater than 10%, in identifying a probable or possible cause of death. Beyond this we recommended further testing at the discretion of the clinician.
Based on these data, we conclude that placental pathologic examination and fetal autopsy are the most useful diagnostic tests for the evaluation of potential causes of stillbirth. Overall, these tests were helpful in either confirming or excluding a potential cause of stillbirth in 64.6% and 42.4% of participants, respectively. Genetic testing and screening for antiphospholipid antibodies also were useful in a meaningful proportion (over 10%) of stillbirths. Other tests were helpful only in a relatively small proportion of participants (less than 5%).
Our results are consistent with those obtained in a large observational study of 1,025 stillbirths in the Netherlands.10 Using a similar approach with the TULIP classification system, Korteweg et al10 noted placental pathology to be useful in 95.7% of participants. Autopsy was useful in 72.6% and cytogenetic analysis in 29.0% of stillbirths.10 Like with the current study, other tests were useful only in a small proportion of participants. We speculate that the relatively higher utility of placental histology and fetal autopsy noted in the Dutch cohort is the result of the very conservative criteria for cause of death inherent in the Initial Causes of Fetal Death (relative to TULIP) classification system.
Others also have noted placental histology and autopsy to be important diagnostic tools for the evaluation of stillbirth. Autopsy provides new information that may alter counseling and recurrence risk in up to half of stillbirths.11–17 Similarly, placental pathology can provide data regarding most causes of stillbirth and many lesions are strongly associated with an increased risk of stillbirth.15,18,19 A recent study noted a stepwise increase in the identification of cause of death with the addition of placental histology and autopsy.20 In a study of 104 stillbirths, clinical scenario and laboratory testing identified a probable cause of death in 24% of participants. This increased to 61% with placental histology and 74% with autopsy.20
Karyotype also is important with abnormal results noted in 8–13% of stillbirths.21,22 Microarray appears to be superior to karyotype as a result of a lower failure rate because it can be done on nonviable tissue and identify copy number changes not apparent with conventional cytogenetic analysis.22 Information available at the time of delivery can inform the work-up of stillbirth. For example, genetic testing was useful in 87% of participants with anomalies but only 5% with obstetric complications. Although a rate of 5% is lower than other clinical cohorts, it still exceeds the estimated risk of roughly 1 in 200 to 1 in 300 at which genetic screening is generally recommended based on antenatal screening results. Thus, testing in such circumstances may be desirable if feasible.
In the United States, it is estimated that fewer than 50% of stillbirths are evaluated by fetal autopsy.23,24 Barriers to the uptake of these evaluations may include patient misconceptions about the logistics of autopsy and a perception that the evaluation will not “bring their child back” or make any difference. If families still decline autopsy, partial autopsy, radiographs, and postmortem magnetic resonance imaging remain useful adjunctive options, although these were not evaluated in this study.25,26 Other problems include reluctance on the part of physicians to perform autopsy counseling and a relative shortage of perinatal pathologists with appropriate expertise.
Our study had several limitations. Not all tests were performed on all stillbirths and in some participants, tests were noninformative. In addition, the definition of a useful test is somewhat subjective, may vary among observers, and is dependent on the classification system used. Also, generalizability may be limited. The study also had numerous strengths. It included large numbers of participants from a population-based sample with considerable geographic, racial, and ethnic diversity. Also, fetal autopsy and placental pathology were performed by perinatal pathologists who underwent central training and used a uniform protocol.
We used a conservative approach to determining utility. In a theoretical sense, all tests have some utility with regard to the exclusion of potential causes of stillbirth. However, potential benefits must be weighed against expense, patient, and health system burden and potential of a false-positive result. Recommendations by the American College of Obstetricians and Gynecologists for the evaluation of stillbirth include perinatal autopsy, placental histopathology, karyotype, complete blood count, syphilis screen, screen for maternal–fetal hemorrhage, parvovirus B-19 serology, thyroid-stimulating hormone, lupus anticoagulant, and anticardiolipin antibodies.2 Other tests are advised in selected patients.
Based on our results as well as those from the Dutch cohort, the algorithm in Figure 3 emphasizes what we have found to be the most useful tests for the evaluation of potential causes of stillbirth. Our study is unique because it incorporates the presenting clinical scenario into the decision-making regarding which tests to order for stillbirth. This was performed to use all available clinical information to develop a high-yield approach to stillbirth evaluation. Performing fetal autopsy and placental histopathology is justified in all stillbirths if possible given the high yield. Genetic testing and screening for antiphospholipid antibodies will be indicated in many women based on the clinical scenario. Remaining tests can be performed in selected patients based in part on testing results20 and available resources as outlined in Figure 3. Additional research will help to further refine the most accurate and cost-effective approach to the evaluation of stillbirth.
1. Gregory ECW, MacDorman MF, Martin JA. Trends in fetal and perinatal mortality in the United States, 2006–2012. NCHS data brief, no 169. Hyattsville (MD): National Center for Health Statistics; 2014.
2. Management of stillbirth. ACOG Practice Bulletin No. 102. American College of Obstetricians and Gynecologists. Obstet Gynecol 2009;113:748–61.
3. Parker CB, Hogue CJ, Koch MA, Willinger M, Reddy UM, Thorsten VR, et al. Stillbirth Collaborative Research Network: design, methods and recruitment experience. Paediatr Perinatal Epidemiol 2011;25:425–35.
4. Stillbirth Collaborative Research Network Writing Group. Causes of death among stillbirths. JAMA 2011;306:2459–68.
5. Pinar H, Koch MA, Hawkins H, Heim-Hall J, Abramowsky CR, Thorsten V, et al. The stillbirth collaborative research network postmortem examination protocol. Am J Perinatol 2012;29:187–202.
6. Pinar H, Koch MA, Hawkins H, Heim-Hall J, Shehata B, Thorsten V, et al. The Stillbirth Collaborative Research Network (SCRN) placental and umbilical cord examination protocol. Am J Perinatol 2011;28:781–92.
7. Stillbirth Collaborative Research Network Writing Group. Association between stillbirth and risk factors known at pregnancy confirmation. JAMA 2011;306:2469–79.
8. Silver RM, Parker CB, Reddy UM, Goldenberg R, Coustan D, Dudley DJ, et al. Antiphospholipid antibodies in stillbirth. Obstet Gynecol 2013;122:641–57.
9. Dudley DJ, Goldenberg R, Conway D, Silver RM, Saade GR, Varner MW, et al. A new system for determining the causes of stillbirth. Obstet Gynecol 2010;116:254–60.
10. Korteweg FJ, Erwich JJ, Timmer A, van der Meer J, Ravisé JM, Veeger NJ, et al. Evaluation of 1025 fetal deaths: proposed diagnostic workup. Am J Obstet Gynecol 2012;206:53.e1–53.e12.
11. Bonetti LR, Ferrari P, Trani N, Maccio L, Laura S, Giuliana S, et al. The role of fetal autopsy and placental examination in the causes of fetal death: a retrospective study of 132 cases of stillbirths. Arch Gynecol Obstet 2011;283:231–41.
12. Ernst LM. A pathologist's perspective on the perinatal autopsy. Semin Perinatol 2015;39:55–63.
13. Faye-Petersen OM, Guinn DA, Wenstrom KD. Value of perinatal autopsy. Obstet Gynecol 1999;94:915–20.
14. Kidron D, Bernheim J, Aviram R. Placental findings contributing to fetal death, a study of 120 stillbirths between 23 and 40 weeks gestation. Placenta 2009;30:700–4.
15. Michalski ST, Porter J, Pauli RM. Costs and consequences of comprehensive stillbirth assessment. Am J Obstet Gynecol 2002;186:1027–34.
16. Pásztor N, Keresztúri A, Kozinsky Z, Pál A. Identification of causes of stillbirth through autopsy and placental examination reports. Fetal Pediatr Pathol 2014;33:49–54.
17. Tellefsen CH, Vogt C. How important is placental examination in cases of perinatal deaths? Pediatr Dev Pathol 2011;14:99–104.
18. Petersson K, Bremme K, Bottinga R, Hofsjö A, Hulthén-Varli I, Kublickas M, et al. Diagnostic evaluation of intrauterine fetal deaths in Stockholm 1998–99. Acta Obstet Gynecol Scand 2002;81:284–92.
19. Pinar H, Goldenberg RL, Koch MA, Heim-Hall J, Hawkins HK, Shehata B, et al. Placental findings in singleton stillbirths. Obstet Gynecol 2014;123:325–36.
20. Miller ES, Minturn L, Linn R, Weese-Mayer DE, Ernst LM. Stillbirth evaluation: a stepwise assessment of placental pathology and autopsy. Am J Obstet Gynecol 2016;214:115.e1–6.
21. Korteweg FJ, Bouman K, Erwich JJ, Timmer A, Veeger NJ, Ravisé JM, et al. Cytogenetic analysis after evaluation of 750 fetal deaths. Proposal for diagnostic workup. Obstet Gynecol 2008;111:865–74.
22. Reddy UM, Page GP, Saade GR, Silver RM, Thorsten VR, Parker CB, et al. Karyotype versus microarray testing for genetic abnormalities after stillbirth. N Engl J Med 2012;367:2185–93.
23. Heuser CC, Hunn J, Varner M, Hossain S, Vered S, Silver RM. Correlation between stillbirth vital statistics and medical records. Obstet Gynecol 2010;116:1296–301.
24. Silver RM, Varner MW, Reddy U, Goldenberg R, Pinar H, Conway D, et al. Work-up of stillbirth: a review of the evidence. Am J Obstet Gynecol 2007;196:433–44.
25. Whitby EH, Offiah AC, Cohen MC. Initial experiences of a minimally invasive autopsy service. A report of the techniques and observations in the first 11 cases. Pediatr Dev Pathol 2015;18:24–9.
26. Woodward PJ, Sohaey R, Harris DP, Jackson GM, Klatt EC, Alexander AL, et al. Postmortem fetal MR imaging: comparison with findings at autopsy. AJR Am J Roentgenol 1997;168:41–6.