A growing body of literature, starting as early as the mid-1990s, demonstrates that neonatal morbidity and mortality continues to decline, even at term, and reaches a nadir at 39–40 weeks of gestation.1–71–71–71–71–71–71–7 As such, there have been multiple efforts across the United States by professional societies, regulatory bodies, and statewide quality collaboratives to lower the rate of nonmedically indicated delivery in the early term (37 0/7–38 6/7 weeks of gestation).8–148–148–148–148–148–148–14
There is concern that 39-week delivery policies have the potential to increase the rate of term stillbirth if early-term deliveries are being shifted later in gestation, especially if policies are misapplied to higher risk pregnancies. There have been conflicting findings in the literature. Ehrenthal et al10 found a threefold increased risk of stillbirth in the early term at a single institution; however Oshiro et al12 found a twofold reduced risk. Furthermore, several studies have shown no statistically significant change.11,1511,15
Given these contradictory findings, our study analyzes the association between state-level early-term delivery trends and term stillbirth rates. A multistate analysis provides a large sample size and the ability to look at both high- and low-risk pregnancies. We used state-level data from the Centers for Disease Control and Prevention to look at reductions in early-term delivery from 2005 to 2011 and term stillbirth rates to answer three questions. First, have states reduced the rate of early-term delivery? Second, did states reduce early-term deliveries for both high- and low-risk pregnancies? Third, did states with the highest reductions in early-term deliveries experience an increased rate of term stillbirth?
MATERIALS AND METHODS
This is a retrospective descriptive analysis of variation in term delivery timing and stillbirth from 2005 to 2011 based on birth certificate and fetal death data. We used publically available data aggregated at the state level. All singleton, term deliveries (37 weeks of gestation or more) were included. We excluded any deliveries without a known gestational age (less than 1% of deliveries). The rate of early-term delivery (37 0/7–38 6/7 weeks of gestation) was calculated as a percentage of total term deliveries. We then calculated the relative decrease in the rate of early-term delivery from 2005 to 2011 per state.
We calculated the total rate of reduction in early-term deliveries as well as the reduction in higher risk subgroups (women with hypertension or diabetes). These two conditions were chosen a priori because they are prevalent and available on birth certificate data. Both pregestational and gestational diabetes were grouped together as were chronic hypertension and gestational hypertension (including preeclampsia). We calculated the correlation between each state's overall reduction in early-term delivery and reductions for these higher risk subgroups using Pearson correlation coefficients weighted by each state's population.
In 2003, a revised version of the U.S. Standard Birth Certificate was introduced. In 2005, 12 states were using the revised form and by 2011, 36 states and the District of Columbia were using the form. Gestational age is common across forms and would not have been affected by the revision. Hypertension was reported similarly and is comparable across versions. Diabetes was separated into pregestational and gestational with the 2003 revision; however, it is comparable to the previous version when combined.16 Of note, there is a known 30% increase in the rate of diabetes once states switch to the 2003 version, thought to be secondary to improved reporting.16
Fetal death data were similarly obtained from publically available Centers for Disease Control and Prevention data aggregated at the state level. We analyzed all stillbirths at term (37 weeks of gestation or more). We excluded all fetal deaths with an unknown gestational age (less than 1% of deaths). We analyzed the overall stillbirth rates as well as those among women with hypertension or diabetes. Similar to birth certificate data, there was also a 2003 revision to the fetal death data. By 2011, 28 states and the District of Columbia were using the revised fetal death data.17
Birth and fetal death data are publically available through 2013. However, starting in 2012 the Centers for Disease Control and Prevention suppressed reporting of any cell that was less than 10. Stillbirth is rare and we needed stratification by gestational week to aggregate all term births; thus, we were unable to obtain state-level data beyond 2011. We therefore used data only through 2011; we analyzed overall trends in the United States in 2012 and 2013 as a sensitivity analysis.
We ranked states by their relative reduction in early-term deliveries between 2005 and 2011. We grouped states into high reduction states (those that reduced early-term birth by more than the national average) and low reduction states (those that reduced early-term birth by less than the national average). Cochrane-Armitage testing for linear trend was used to analyze stillbirth rates over time in the overall United States as well as in the high and low reduction states separately. Additionally, we analyzed the correlation between a state's relative reduction in early-term delivery between 2005 and 2011 and the relative change in term stillbirth using a Pearson correlation coefficient weighted by the state's population.
Power was calculated post hoc. For the total United States, we had 80% power to detect a 6% increase in the overall rate of term stillbirth, a 24% increase in stillbirth rate for women with diabetes, and a 28% increase for women with hypertension. Among high reduction states, we had 80% power to detect a 9% increase in the rate of term stillbirth, a 36% increase in women with diabetes, and a 36% increase in women with hypertension.
All data were deidentified and we were exempt from institutional review board approval at our hospital. SAS 9.3 was used for all statistical analysis, and a P value of <.05 was considered statistically significant.
In 2005, there were 3,533,233 term, singleton births, of which 1,123,467 (31.8%) occurred in the early term. By 2011, there were 3,429,172 term, singleton births of which only 978,294 (28.5%) occurred in the early term corresponding to a 3.3% absolute reduction and 10.3% relative reduction. Figure 1 displays the trends over time in the United States. State-level data are displayed in Table 1.
There was also a decline in early-term delivery for higher risk pregnancies. In 2005, there were 127,083 singleton term deliveries to women with diabetes, of which 53,899 took place in the early term (42.4%). By 2011, there were 177,166 singleton, term deliveries to women with diabetes of which 65,014 (36.7%) occurred in the early term, corresponding to a 5.7% absolute reduction and 13.5% relative reduction. Similarly, for women with hypertension, 67,540 of 146,748 deliveries (46.0%) occurred in the early term in 2005 as compared with 72,689 of 167,374 (43.4%) in 2011, corresponding to a 2.6% absolute reduction and 5.6% relative reduction. States with the largest reduction in overall early-term delivery had the largest reductions in early-term delivery for women with diabetes (r=0.78; P<.01) and women with hypertension (r=0.80; P<.01).
Table 2 displays the population characteristics of the overall United States, the high reduction states, and the low reduction states over the 2005 to 2011 time period. As shown, the rate of diabetes increased from 3.6% to 5.2% in the United States between 2005 and 2011. Some of this increase is likely the result of a known increase in reporting with the revised birth certificates.16 The rate of hypertension also increased from 4.2% to 4.9%. Increases in diabetes and hypertension were similar in low and high reduction states.
The overall rate of term stillbirth did not change in the United States between 2005 and 2011 (Table 3). In 2005 there were 4,341 term stillbirths (123/100,000 births) as compared with 4,449 (130/100,000 births) in 2011 (P=.189 for linear trend). There was no trend in the high reduction states alone (127/100,000 to 133/100,000; P=.986). Among low-risk women (those without diabetes or hypertension), there was no change in stillbirth (105/100,000 in 2005 to 110/100,000 in 2011; P=.167) nor was there a significant trend for women with hypertension. However, for women with diabetes, there was a significant increase from 302 stillbirths (238/100,000 births) in 2005 to 532 (300/100,000 births) in 2011 (P=.010); this increase was similar for women with diabetes in both the low and high reduction states. Figure 2 displays the early-term birth rates and term stillbirth rates per year for the high and low reduction states.
We also looked at the correlation between state-level reduction in early-term delivery and state-level changes in the term stillbirth rate from 2005 to 2011. There was no correlation at the state level between overall reduction in early-term delivery and term stillbirth (r=0.21; P=.13) (see Appendix 1, available online at http://links.lww.com/AOG/A705). Also, there was not a correlation between a state's reduction in early-term delivery for women with diabetes and their term stillbirth rate (r=−0.18; P=.24) (Appendix 2, available online at http://links.lww.com/AOG/A706) nor for women with hypertension (r=0.27; P=.09) (Appendix 3, available online at http://links.lww.com/AOG/A707).
We were only able to capture state-level trends through 2011 as a result of the limitations in publically available data as described in the “Methods.” However, at a national level there was no change when including 2012 and 2013 data; early-term delivery continued to decline (28.2% of term deliveries in 2012 and 27.2% of term deliveries in 2013) and there was no change in the term stillbirth rate (123/100,000 births in 2012 and 126/100,000 in 2013). For women with diabetes, term stillbirth rates did not have continued increases (285/100,000 in 2012 and 279/100,000 in 2013).
Among term births in the United States from 2005 to 2011, there was a reduction in early-term deliveries. Wide variation was noted in this reduction across states, ranging from a 25.5% relative reduction to a 3.9% relative increase. States with a greater overall reduction in early-term births had a greater reduction for higher risk pregnancies (women with diabetes and hypertension). However, there was no association with term stillbirth trends; there was no change in the overall term stillbirth rate in the United States nor was there an increase for the high reduction states alone (those with a greater reduction than the national average). Additionally, there was no correlation between a state's overall reduction in early-term deliveries and any changes in its term stillbirth rate.
We did find that there was 25% increase in the rate of term, singleton stillbirth among women with diabetes in the United States (from 238/100,000 in 2005 to 300/100,000 in 2011; P=.010 for linear trend). States with the highest reductions in early-term delivery had the highest reductions in early-term delivery for women with diabetes, suggesting that clinicians may be misapplying early-term delivery policies to high-risk women. However, the increase in term stillbirth for women with diabetes appears to be independent of state-level reductions in early-term delivery; there was a similar increase in high and low reduction states, and there was no correlation at the state level between reduction in early-term delivery for women with diabetes and changes in the term stillbirth rate. Thus, it is possible that women with diabetes are experiencing more stillbirths independent of changes in early-term delivery policy. There was an increase in the proportion of women with diabetes coded on birth certificates, which may reflect changes in birth certificate coding,16 changes in diagnostic criteria, or the growing burden of obesity and type 2 diabetes in the U.S. population.18,1918,19 As a result of these demographic shifts, women with diabetes coded on birth certificates may be a more ill population than in previous years and thus experiencing increased rates of stillbirth across all gestational ages. Further research on the causes of stillbirth in women with diabetes is needed to fully understand these findings.
Our findings are similar to others who were underpowered but found no difference in stillbirth trends after implementing policies to reduce early-term delivery.11,1511,15 Unlike Ehrenthal et al10 (n=24,028), who found a large increase in stillbirth, and Oshiro et al12 (n=160,404) who found a large decrease in term stillbirth, we found neither a statistical nor a clinically meaningful difference. Our sample size was greater than either of these studies, thus potentially more stable to any random trends, which are possible when analyzing a rare outcome.
Our study has a number of strengths. First, it has a large sample size and is powered to detect trends in rare outcomes. Second, we looked at trends in overall, low-, and high-risk deliveries. If the analysis was limited to low-risk stillbirths, potential unintended consequences of 39-week delivery policies may have been overlooked, yet we found no trends in term stillbirths in high reduction states for either high- or low-risk pregnancies.
Our study is not without limitations. We were not able to gather information on state-level early-term delivery policies; it is possible that policies were not causal. However, given the steady, significant decline in early-term deliveries, it seems most probable that this decrease was secondary to an increased awareness of early-term morbidity and policies to reduce delivery in the early term. Second, to obtain a large sample size, we sacrificed clinical detail. We were not able to say whether the prolongation of the pregnancies, particularly the high-risk pregnancies, was appropriate; however, this was not the aim of this study. We also were not able to examine gestational age more finely than the week; it is possible that shifts in early-term delivery were from the late 38th week to the early 39th week and may not have been particularly clinically meaningful. Third, we were unable to obtain details about the type and onset of hypertensive disorders. The trends, however, were the same when pregestational hypertension alone was analyzed. It is possible that changing definitions of diseases could have affected our results. The increased rate of diabetes, for example, is at least in part attributable to better reporting with the revised birth certificate data. However, these revisions took place for both birth and fetal death data and thus would not be anticipated to affect the overall rates.
Overall, it is reassuring that the reduction in early-term deliveries does not appear to be associated with an increase in the term stillbirth rate. However, our findings should be interpreted with some caution. First, we did not include late preterm delivery. It is possible that states with large reductions in early-term delivery are also increasing late preterm delivery, although it seems unlikely that the trends in stillbirth for this population would be significantly different. Second, it is possible that there is a subgroup for which 39-week delivery policies are affecting the stillbirth rate. We chose not to look at gestational age subgroups, for example 37 0/7- to 38 6/7-week stillbirths alone. Rather, we were interested in overall term stillbirth rates. In the absence of any trends in overall term stillbirths, the benefit of looking at specific gestational age subgroups seems limited. Third, it is possible that our study did not detect small changes; we were powered to find a 6% increase in the rate of term birth across the United States. Last, we were only able to capture state level data through 2011. However, at a national length there was no change in the term stillbirth rate when including data through 2013.
Overall, our study is reassuring that the decrease in the rate of early-term birth in the United States between 2005 and 2011 was not associated with changes in the rate of term stillbirth. Continued surveillance of stillbirth rates, for both high- and low-risk pregnancies, at a state and national level, however, is warranted.
1. Tita AT, Landon MB, Spong CY, Lai Y, Leveno KJ, Varner MW, et al.. Timing of elective repeat cesarean delivery at term and neonatal outcomes. N Engl J Med 2009;360:111–20.
2. Consortium on Safe Labor, Hibbard JU, Wilkins I, Sun L, Gregory K, Haberman S, et al.. Respiratory morbidity in late preterm births. JAMA 2010;304:419–25.
3. Chiossi G, Lai Y, Landon MB, Spong CY, Rouse DJ, Varner MW, et al.. Timing of delivery and adverse outcomes in term singleton repeat cesarean deliveries. Obstet Gynecol 2013;121:561–9.
4. Reddy UM, Ko CW, Raju TN, Willinger M. Delivery indications at late-preterm gestations and infant mortality rates in the United States. Pediatrics 2009;124:234–40.
5. Fleischman AR, Oinuma M, Clark SL. Rethinking the definition of “term pregnancy”. Obstet Gynecol 2010;116:136–9.
6. Morrison JJ, Rennie JM, Milton PJ. Neonatal respiratory morbidity and mode of delivery at term: influence of timing of elective caesarean section. Br J Obstet Gynaecol 1995;102:101–6.
7. Clark SL, Miller DD, Belfort MA, Dildy GA, Frye DK, Meyers JA. Neonatal and maternal outcomes associated with elective term delivery. Am J Obstet Gynecol 2009;200:156.e1–4.
8. Nonmedically indicated early-term deliveries. Committee Opinion No. 561. American College of Obstetricians and Gynecologists. Obstet Gynecol 2013;121:911–5.
9. Specifications manual for Joint Commission national quality core measures [version 2011A]. Oakbrook Terrace (IL): The Joint Commission; 2010.p. PC-1–242.
10. Ehrenthal DB, Hoffman MK, Jiang X, Ostrum G. Neonatal outcomes after implementation of guidelines limiting elective delivery before 39 weeks of gestation. Obstet Gynecol 2011;118:1047–55.
11. Oshiro BT, Kowalewski L, Sappenfield W, Alter CC, Bettegowda VR, Russell R, et al.. A multistate quality improvement program to decrease elective deliveries before 39 weeks of gestation. Obstet Gynecol 2013;121:1025–31.
12. Oshiro BT, Henry E, Wilson J, Branch DW, Varner MW; Women and Newborn Clinical Integration Program. Decreasing elective deliveries before 39 weeks of gestation in an integrated health care system. Obstet Gynecol 2009;113:804–11.
13. Main E, Oshiro B, Chagolla B, Bingham D, Dang-Kilduff L, Kowalewski L. Elimination of non-medically indicated (elective) deliveries before 39 weeks gestational age (California Maternal Quality Care Collaborative Toolkit to Transform Maternity Care) developed under contract #08-85012 with the California Department of Public Health; Maternal, Child and Adolescent Health Division. 1st ed. White Plains (NY): March of Dimes; 2010.
14. Donovan EF, Lannon C, Bailit J, Rose B, Iams JD, Byczkowski T; Ohio Perinatal Quality Collaborative Writing Committee. A statewide initiative to reduce inappropriate scheduled births at 36(0/7)–38(6/7) weeks' gestation. Am J Obstet Gynecol 2010;202:243.e1–8.
15. Little SE, Robinson JN, Puopolo KM, Mukhopadhyay S, Wilkins-Haug LE, Acker DA, et al.. The effect of obstetric practice change to reduce early term delivery on perinatal outcome. J Perinatol 2014;34:176–80.
18. International Association of Diabetes and Pregnancy Study Groups Consensus Panel, Metzger BE, Gabbe SG, Persson B, Buchanan TA, Catalano PA, et al.. International association of diabetes and pregnancy study groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy. Diabetes Care 2010;33:676–82.
19. Albrecht SS, Kuklina EV, Bansil P, Jamieson DJ, Whiteman MK, Kourtis AP, et al.. Diabetes trends among delivery hospitalizations in the U.S., 1994–2004. Diabetes Care 2010;33:768–73.
Supplemental Digital Content
© 2015 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.