- Email: [email protected]
Trends in twin preterm birth subtypes in the United States, 1989 through 2000: Impact on perinatal mortality
American Journal of Obstetrics and Gynecology (2005) 193, 1076.e1–1076.e9
www.ajog.org
Trends in twin preterm birth subtypes in the United States, 1989 through 2000: Impact on perinatal mortality Cande V. Ananth, PhD, MPH,a,b,* K. S. Joseph, MD, PhD,c Kitaw Demissie, MD, PhD,b Anthony M. Vintzileos, MDd Division of Epidemiology and Biostatistics, Department of Obstetrics, Gynecology, and Reproductive Sciences, Robert Wood Johnson Medical School,a University of Medicine and Dentistry of New Jersey, New Brunswick, NJ; Department of Epidemiology, School of Public Health,b University of Medicine and Dentistry of New Jersey, Piscataway, NJ; Perinatal Epidemiology Research Unit, Departments of Obstetrics and Gynaecology, and of Pediatrics, Dalhousie University,c Halifax, Canada; Division of Maternal-Fetal Medicine,d Department of Obstetrics, Gynecology, and Reproductive Sciences, Robert Wood Johnson Medical School, University of Medicine and Dentistry of New Jersey, New Brunswick, NJ Received for publication February 8, 2005; revised June 1, 2005; accepted June 7, 2005
KEY WORDS Preterm delivery Trends Premature rupture of membranes Labor induction Cesarean delivery Twins Perinatal mortality
Objective: We examined trends in twin preterm birth !37 weeks following ruptured membranes (ROM), medically indicated preterm birth, and preterm birth following spontaneous onset of labor (PTL). We further examined whether the changes in preterm birth subtypes were associated with trends in twin perinatal mortality. Study design: We carried out a retrospective cohort study of 1,172,405 twin live births and stillbirths delivered in the US between 1989 and 2000. Trends in preterm birth subtypes and perinatal mortality (stillbirths at R22 weeks plus neonatal deaths within 28 days) were examined through ecological logistic regression models after adjusting for confounders. Results: Twin preterm birth among whites increased from 46.6% in 1989 to 1990 to 56.7% in 1999 to 2000, and from 56.1% to 61.0% among blacks over the same period. Medically indicated preterm birth increased by 50% (95% CI 49-52) among whites, and by 33% (95% CI 29-36) among blacks. PTL increased by 24% among whites, but remained fairly unchanged among blacks between the two periods. Preterm birth following ROM also did not change between the 2 periods among whites, but declined by 7% among blacks. Perinatal mortality among twin births declined by 41% (95% CI 38-44) among whites, and by 37% (95% CI 32-42) among blacks between 1989 and 1990 and 1999 and 2000. This mortality decline was most closely associated with the increase in medically
Presented at the Twenty-Fifth Annual Meeting of the Society for Maternal-Fetal Medicine Meeting, Reno, NV, February 7-12, 2005. Dr Ananth is partially supported through a grant (R01-HD038902) from the National Institutes of Health. Dr Joseph is supported by a Clinical Research Scholarship from the Dalhousie University Faculty of Medicine, Nova Scotia, Canada, and is a Peter Lougheed New Investigator of the Canadian Institutes of Health Research. * Reprint requests: Cande V. Ananth, PhD, MPH, Division of Epidemiology and Biostatistics, Department of Obstetrics, Gynecology, and Reproductive Sciences, UMDNJ-Robert Wood Johnson Medical School, 125 Paterson Street, New Brunswick NJ 08901-1977. E-mail: [email protected] 0002-9378/$ - see front matter Ó 2005 Mosby, Inc. All rights reserved. doi:10.1016/j.ajog.2005.06.088
1076.e2
Ananth et al indicated preterm birth among whites, and with the decrease in preterm birth following ROM among blacks. Conclusion: Temporal trends in twin preterm birth varied substantially based on underlying subtypes and race. The increase in medically indicated preterm birth is associated with a large reduction in perinatal mortality. Ó 2005 Mosby, Inc. All rights reserved.
Preterm birth (!37 weeks) remains one of the leading causes of adverse perinatal outcomes, including neonatal and infant mortality. Compared with infants delivered at term, preterm infants also suffer greater neurodevelopmental problems and increased morbidity resulting from respiratory complications. The rate of preterm birth among twins has increased dramatically in recent years1-4 in the US1,2 and other industrialized countries,3,4 largely because of increased obstetric intervention (ie, labor induction and cesarean delivery).5 In addition, increases in assisted reproduction technologies, and in older maternal age, have also affected these trends. With the possible exception of France6,7 and Finland,8 most other industrialized countries have experienced a temporal increase in overall preterm birth rates.2-4,9-13 The increase in preterm birth rates has been puzzling, and often regarded as a failure of clinical and community-based efforts to reduce preterm birth.14 Preterm birth is a multifactorial condition with at least 3 distinct clinical pathways: preterm birth resulting from ruptured membranes; preterm birth resulting from medical/obstetric intervention; and preterm birth following spontaneous onset of labor. We recently showed that preterm birth among singleton whites in the US has increased largely because of a concurrent increase in medically indicated preterm birth, whereas among singleton blacks, preterm birth rates declined because of a decrease in spontaneous preterm birth and preterm birth following ruptured membranes.15 Further, we showed that the temporal increase in preterm birth among singletons was associated with a reduction in perinatal mortality, with the largest reductions in mortality being associated with medically indicated preterm birth.15 We carried out a study of trends in twin preterm birth subtypes because it is unclear whether the associations seen among singleton births15 also occurred among twins. We hypothesized that medically indicated preterm birth among twins would have increased over the last decade in the US, and that these increases would have led to a large and substantial decline in twin perinatal mortality.
Material and methods United States twin birth cohort composition We used publicly available data from the US twin natality data files from 1989 through 2000 to examine
temporal trends in preterm birth subtypes, and linked natality and infant mortality data files 1989 through 1991 and 1995 through 2000, to examine the impact of preterm birth trends on perinatal mortality.16 The data contained in these files were obtained from live birth and fetal and infant death certificates, and were assembled and compiled by the National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention.17 Because the National Center for Health Statistics did not link the live birth data files to their corresponding infant death files for 1992 through 1994, analysis of neonatal mortality in the 1992 through 1994 period could not be performed. Gestational age data, reported in completed weeks, on vital statistics records were predominantly based on the date of the last menstrual period, calculated as the difference between the date of delivery and the date of the last menstrual period. In a small fraction of births (!5%) that did not contain the date of the menstrual period, or when the menstrual-based gestational age was grossly incompatible with birth weight, a clinical estimate of gestation, also contained on vital records, was used instead.17 Imputation of gestational age (in completed weeks) was also performed by the NCHS National Center for Health Statistics for births that contained a valid month and year for the last menstrual period, but had a missing date.18 The replacement of menstrual estimate of gestational age by a clinical estimate, as well as imputation of missing gestational age, was performed consistently for all years (1989 through 2000).
Preterm birth classification Trends in twin preterm birth !37 weeks were assessed after stratifying births by underlying clinical subtype. Preterm birth was sequentially classified into 3 distinct clinical subtypes based on the following hierarchy: preterm birth following rupture of membranes; medically indicated preterm birth, defined as a preterm birth which followed iatrogenic intervention (labor induction or a primary or repeat cesarean); and preterm birth following spontaneous onset of labor, defined as preterm birth that was neither associated with ruptured membranes nor medically indicated. The determination of preterm birth subtype was based on 4 check-box items on the live birth, and fetal, and infant death certificates.17 The item ‘‘ruptured membranes’’ accompanied by delivery at !37 weeks was first
Ananth et al used to identify twins who delivered preterm following ruptured membranes. The items ‘‘primary cesarean,’’ ‘‘repeat cesarean,’’ or ‘‘labor induction,’’ accompanied by delivery at !37 weeks were combined to identify twins who delivered preterm because of medical indications. The item corresponding to labor induction represents an attempt to induce labor, regardless of its success.
Perinatal mortality We examined trends in twin perinatal mortality rates between 1989 and 2000 within the 3 clinical subtypes of preterm birth. Perinatal mortality was defined as stillbirths at R22 weeks plus death to liveborn twins within the first 28 days after birth, and was expressed per 1000 total twin births. Changes in cause-specific neonatal mortality rates between 1989 and 1990 and 1999 and 2000 were also assessed. For this analysis, the cause of death assigned on the mortality data was based on the International Classification of Diseases, ninth revision (ICD-9) for the 1989 and 1990 data, and the 10th revision (ICD-10) for the 1999 and 2000 periods. The following cause of death categories were examined: congenital anomalies (ICD-9 740-759; ICD-10 Q00-Q99); maternal complications (ICD-9 761; ICD-10 P01); complications of the placenta, cord, or membrane (ICD-9 762; ICD-10 P02); short gestation or low birth weight (ICD-9 765; ICD-10 P05 and P07); intrauterine hypoxia or asphyxia (ICD-9 768; ICD-10 P20 and P21); respiratory distress syndrome (ICD-9 769; ICD-10 P22); infections during the perinatal period (ICD-9 771; ICD-10 P23, P35, P36, P37, and P39); and sudden infant death syndrome (ICD-9 798.0; ICD-10 R95). Causes of stillbirth were not examined because such data were unavailable.
1076.e3 regression model will likely overestimate the RR. We, therefore, converted the odds ratio obtained from logistic models to RR.19 The contribution of trends in the 3 clinical subtypes of preterm birth to trends in perinatal mortality was also assessed. For this analysis, with perinatal deaths as the outcome, we aggregated the data by each US state for the 2 periods (1989-1990 and 1999-2000). In each of the 100 strata (50 states times 2 periods), we separately determined the proportion for each of the 3 preterm birth subtypes (preterm birth proportion was an independent variable in these regression models). Adjustment for confounding because of maternal age !20 and R35 years, nulliparity, maternal education !12 years of schooling, and unmarried status was accomplished using the proportion of each of these confounders in each of the 100 strata. The ecological design of the study was necessary in order to avoid confounding by indication, which occurs when attempting to estimate the effect of a clinical intervention20 using individual-level analysis and a nonexperimental study design. Such confounding is likely because labor induction and cesarean delivery are typically used in pregnancies where fetal compromise is suspected, and because antepartum stillbirths are often delivered by inducing labor. We regressed the preterm birth variable (in addition to state, the period variable, and confounders) against perinatal death. From these models, we obtained the RR and 95% CI denoting the change in perinatal mortality rates between 1989 and 2000 before and after adjustment for temporal changes in preterm birth subtypes. This analysis was then repeated for stillbirths and neonatal deaths separately. We also repeated all analyses for preterm birth !32 weeks.
Data exclusions Statistical analysis Changes in the rates of preterm birth !37 weeks and preterm birth subtypes among twins between 1989 and 1990 and 1999 and 2000 among white and black women were examined first, and quantified using relative risks (RR) with 95% CIs. In a similar fashion, changes in perinatal mortality rates were assessed by comparing mortality rates across the 2 periods (1989-1990 and 1999-2000) using RRs. Given the population-based nature of our study, we present CI of effect measures as a means of assessing the precision of the estimate. We fit logistic regression models with preterm birth as the dependent variable, and an indicator for period as the independent variable (1989-1990 served as the reference). These models also included maternal age, parity, maternal education, and marital status in order to provide adjusted estimates of the temporal change in preterm birth and its subtypes. Because preterm birth rates among twins are high, odds ratio from the logistic
Of all US twin births delivered in the US between 1989 and 2000, we sequentially excluded those with missing gestational age (n Z 14,110), gestational age !22 completed weeks (n Z 19,535), and pregnancies that ended at R44 weeks (n Z 3392). We additionally excluded 52,050 twin births to non-black and non-white women. After all exclusions, 1,172,405 twin live births and stillbirths occurring between 1989 and 2000 remained for analysis (951,778 white; 220,627 black). Data on labor induction were missing in a sizable proportion (over 25%) of delivery records from Louisiana, and so data from this state were excluded in subanalyses pertaining to preterm birth subtype. This exclusion was necessary because the missing information was needed for the classification of preterm birth by subtype (ie, medically indicated preterm birth). However, for overall analysis of preterm birth trends (ie, without consideration of preterm birth subtypes), such data were retained.
1076.e4
Ananth et al
Figure 1 Temporal changes in preterm birth (!37 weeks), preterm birth following ruptured membranes, medically indicated preterm birth, and spontaneous preterm birth rates (left), and temporal changes in perinatal mortality rates (right), US whites, 1989 through 2000.
This study was approved by the ethics review committee of the Institutional Review Board of the UMDNJ-Robert Wood Johnson Medical School, NJ.
Results Among whites, the rate of preterm birth (!37 weeks) among twins increased by 31%, from 43.5% in 1989 to 56.8% in 2000 (Figure 1). Similarly, the preterm birth rate among black twins increased by 16%, from 52.8% in 1989 to 61.2% in 2000 (Figure 2). Changes in sociodemographic determinants and complications during pregnancy between 1989 and 1990 and 1999 and 2000 are shown in Table I. During the periods 1989 and 1990 and 1999 and 2000, the proportion of women aged R35 years increased by 81% among whites, and by 69% among blacks (Table I). Preexisting/gestational diabetes also increased between these 2 periods among both whites and blacks, while an increase in the rate of chronic hypertension was evident only among whites. Rates of labor induction and, to a lesser extent, cesarean delivery also increased among both whites and blacks between 1989 and 1990 and 1999 and 2000.
The overall increase in preterm birth (!37 weeks) among both whites and blacks was largely driven by a concurrent relative increase in the rate of medically indicated preterm birth (Table II). Between 1989 and 1990 and 1999 and 2000, rates of medically indicated preterm birth !37 weeks (adjusted for maternal age, nulliparity, maternal education, and marital status) increased by 50% (RR 1.50, 95% CI 1.49-1.52) among whites, and by 33% (RR 1.33, 95% CI 1.29-1.36) among blacks. Adjusted rates of spontaneous preterm birth among whites increased by 24% (RR 1.24, 95% CI 1.221.26), but remained virtually unchanged among blacks during the 2 periods. Adjusted rates of preterm birth following ruptured membranes remained unchanged among whites, but declined by 7% among blacks between the 2 periods. Preterm birth !32 weeks increased by 22% and 5% among whites and blacks, respectively, between 1989 and 1990 and 1999 and 2000. An increase in overall preterm birth !32 weeks between 1989 and 1990 and 1999 and 2000 was also evident in the analyses after adjustments for maternal age, nulliparity, maternal education, and marital status. A 22% increase was
Ananth et al
1076.e5
Figure 2 Temporal changes in preterm birth (!37 weeks), preterm birth following ruptured membranes, medically indicated preterm birth, and spontaneous preterm birth rates (left), and temporal changes in perinatal mortality rates (right), US blacks, 1989 through 2000.
Table I
Changes in maternal characteristics of twin births in the US, 1989-1990 and 1999-2000 White women
Black women
Rate per 100 Maternal characteristics
1989-1990
1999-2000
Relative risk (95% CI)
Maternal age !20 years Maternal age R35 years Primigravida Maternal education !12 years Unmarried status Chronic hypertension Pregnancy-induced hypertension Preexisting/gestational diabetes Cesarean delivery Labor induction
7.2 12.7 17.1 18.3 17.2 0.8 7.5 2.5 55.3 6.9
6.8 23.0 18.0 14.4 19.5 0.9 9.0 3.6 57.3 14.7
0.94 1.81 1.05 0.79 1.14 1.22 1.22 1.45 1.04 2.13
observed among whites, and a 5% increase among blacks. Adjusted rates of medically indicated preterm birth increased by 29% and 20% among whites and blacks, respectively, between the 2 periods. Spontaneous preterm birth and preterm birth following ruptured
(0.92-0.97) (1.77-1.85) (1.03-1.07) (0.78-0.80) (1.12-1.16) (1.13-1.32) (1.19-1.25) (1.38-1.52) (1.03-1.05) (2.08-2.18)
Rate per 100 1989-1990
1999-2000
Relative risk (95% CI)
16.8 8.8 12.1 27.0 64.6 1.8 6.8 1.9 51.6 4.5
14.4 13.9 12.2 21.6 65.8 1.9 8.1 3.1 55.8 10.7
0.84 1.69 1.02 0.75 1.06 1.06 1.20 1.63 1.09 2.39
(0.80-0.87) (1.60-1.77) (0.97-1.06) (0.72-0.77) (1.03-1.09) (0.95-1.18) (1.14-1.27) (1.45-1.84) (1.06-1.13) (2.24-2.54)
membranes (!32 weeks) declined among both whites and blacks between 1989 and 1990 and 1999 and 2000. Among both white and blacks, perinatal mortality rates among twin births declined between 1989 and 1990 and 1990 and 2000 even as preterm birth rates increased
1076.e6 Table II
Ananth et al Changes in preterm birth rates !37 and !32 weeks: twin live births in the US, 1989-1990 and 1999-2000 White women
Black women
Preterm birth rate (%) Preterm birth subtypes
1989-1990
1999-2000
Adjusted RR (95% CI)
Number of twin births Preterm birth !37 weeks Ruptured membranes Medically indicated Spontaneous preterm birth Preterm birth !32 weeks Ruptured membranes Medically indicated Spontaneous preterm birth
144,330 46.6 6.2 24.1 16.2 10.7 2.5 4.7 3.4
183,921 56.7 4.8 35.2 16.6 10.7 1.8 6.1 2.8
1.22 0.98 1.50 1.24 1.22 0.75 1.29 0.86
(1.21-1.23) (0.95-1.01) (1.49-1.52) (1.22-1.26) (1.19-1.24) (0.71-0.79) (1.25-1.33) (0.82-0.89)
Preterm birth rate (%) 1989-1990
1999-2000
Adjusted RR (95% CI)
36,440 56.1 7.6 26.5 21.9 19.2 3.9 7.8 7.4
39,927 61.0 6.2 35.3 19.5 18.0 3.2 9.5 5.3
1.11 0.93 1.33 1.02 1.05 0.81 1.20 0.75
(1.10-1.12) (0.87-0.99) (1.29-1.36) (0.99-1.05) (1.02-1.09) (0.75-0.88) (1.14-1.26) (0.71-0.79)
Relative risks were adjusted for the confounding effects caused by maternal age, parity, maternal education, and marital status.
Table III Changes in perinatal mortality and association with changes in preterm birth !37 and !32 weeks: twin births in the US, 1989-1990 and 1999-2000 White women
Black women
Perinatal mortality* 1989-1990 All births Adjusted for confoundersy
32.7 –
1999-2000 19.4 –
Relative risk (95% CI)
Perinatal mortality* 1989-1990
1999-2000
Relative risk (95% CI)
0.59 (0.56-0.62) 0.61 (0.58-0.64)
43.8 –
27.9 –
0.63 (0.58-0.68) 0.64 (0.59-0.69)
Association between preterm birth !37 weeks and perinatal mortality Adjusted for confounders plus Preterm birth following ROM 77.3 52.9 0.62 (0.56-0.69) Medically indicated preterm birth 44.1 25.2 0.69 (0.66-0.73) Spontaneous preterm birth 63.3 33.4 0.61 (0.59-0.63)
101.7 47.1 77.3
75.7 31.2 48.6
0.68 (0.58-0.79) 0.66 (0.58-0.74) 0.61 (0.54-0.69)
Association between preterm birth !32 Adjusted for confounders plus Preterm birth following ROM Medically indicated preterm birth Spontaneous preterm birth
177.2 120.6 199.9
137.9 86.5 151.9
0.69 (0.60-0.78) 0.68 (0.61-0.76) 0.69 (0.62-0.76)
weeks and perinatal mortality 170.6 156.6 253.3
126.2 102.9 165.5
0.66 (0.62-0.71) 0.59 (0.55-0.63) 0.63 (0.58-0.67)
All relative risks express the period change in perinatal mortality (1999-2000 vs 1989-1990). * Perinatal mortality rates (crude) are expressed per 1000 total twin births. Perinatal mortality rate for each of the 3 preterm birth subtype category was calculated as the number of perinatal deaths in each preterm birth subtype category (eg, medically indicated preterm birth), divided by the total number of preterm births within that subtype (eg, medically indicated preterm birth). y Relative risks were adjusted for the confounding effects caused by maternal age, parity, maternal education, and marital status.
(Figures 1 and 2). Crude perinatal mortality rate declined by 41% and 37% among whites and blacks, respectively, between 1989 and 1990 and 1999 and 2000 (Table III). Changes in medically indicated preterm birth rates were most closely associated (inversely) with changes in perinatal mortality rates. For instance, among whites, the adjustment for the increase in medically indicated preterm birth between 1989 and 1990 and 1999 and 2000 (Table II) was associated with a decline in the period effect for perinatal mortality from RR 0.61 (95% CI 0.58-0.64) to RR 0.69 (95% CI 0.660.73). In other words, had medically indicated preterm birth (!37 weeks) among whites not increased by 50%, perinatal mortality among whites would have declined
by 31% and not 39%. Changes in preterm birth following ruptured membranes and spontaneous preterm birth among whites were not associated with substantial changes in perinatal mortality among whites (Table III). Among blacks, the decline in perinatal mortality was most strongly associated with a concurrent decline in preterm birth (!37 weeks) following ruptured membranes. In other words, had preterm birth following ruptured membranes not declined by 7%, perinatal mortality would have declined only by 32% (RR 0.68, 95% CI 0.58-0.79) and not 36% (RR 0.64, 95% CI 0.590.69). Similar patterns of associations between preterm birth !32 weeks and perinatal mortality were observed.
Ananth et al
1076.e7
Table IV Changes in cause-specific neonatal mortality among preterm births (!37 weeks): twin births in the United States, 19891990 and 1999-2000 White women Preterm birth !37 weeks subtypes
Black women
Neonatal mortality 1989-1990
1999-2000
Adjusted RR (95% CI)
Neonatal deaths resulting from respiratory distress syndrome Ruptured membranes 16.8 5.9 0.40 (0.28-0.58) Medically indicated 6.4 2.2 0.34 (0.27-0.44) Spontaneous labor 11.8 2.5 0.23 (0.17-0.31) Neonatal deaths resulting from congenital malformations Ruptured membranes 8.3 3.6 Medically indicated 6.2 3.3 Spontaneous labor 4.2 1.9
0.47 (0.30-0.79) 0.60 (0.48-0.75) 0.49 (0.33-0.72)
Neonatal deaths resulting from complications of short gestation/low birthweight Ruptured membranes 4.5 14.0 3.26 (2.06-5.16) Medically indicated 0.8 1.7 2.61 (1.56-4.38) Spontaneous labor 5.2 10.3 1.97 (1.54-2.52)
Neonatal mortality 1989-1990
1999-2000
Adjusted RR (95% CI)
22.8 12.1 16.4
10.4 3.0 5.2
0.49 (0.29-0.84) 0.35 (0.24-0.53) 0.42 (0.29-0.62)
6.6 5.1 4.0
5.6 2.7 1.2
0.98 (0.39-2.41) 0.63 (0.38-1.02) 0.29 (0.13-0.65)
10.8 3.0 7.6
24.9 3.6 19.6
2.65 (1.59-4.34) 1.49 (0.85-2.60) 3.31 (2.31-4.74)
All relative risks express the period change in neonatal mortality (1999-2000 vs 1989-1990). Cause-specific neonatal mortality rates are expressed per 1000 twin live births delivered preterm. Cause-specific neonatal mortality rate for each of the 3 preterm birth subtype categories was calculated as the number of neonatal deaths with a particular cause of death in each preterm birth subtype category (eg, medically indicated preterm birth), divided by the total number of preterm births within that subtype (eg, medically indicated preterm birth). Relative risks were adjusted for the confounding effects caused by maternal age, parity, maternal education, and marital status.
The association between increases in each preterm birth subtype and neonatal mortality were also generally similar to associations with perinatal mortality. Cause-specific neonatal mortality rates within each preterm birth subtype are shown in Table IV. There were significant declines in neonatal deaths caused by respiratory distress syndrome, and congenital malformations and increases in neonatal deaths caused by complications that accompanying short gestational age and low birth weight. Neonatal mortality from other causes, including those caused by maternal complications, complications of placenta, cord, and membranes, intrauterine hypoxia or birth asphyxia, infections during the perinatal period, and sudden infant death syndrome in relation to preterm birth subtypes remained unchanged (data not shown).
Comment Despite extensive research, preterm birth continues to pose an enigma in term of its etiology and prevention. Our study indicates that, among twin births in the US, the temporal increase in preterm birth among both white and black women was largely driven by a concurrent temporal increase in medically indicated preterm birth. This increase in medically indicated preterm birth was less steep among twins delivered at !32 weeks than among those delivered at !37 weeks. Perinatal mortality rates declined, and as hypothesized, the decline was strongly associated with increases in medically indicated
preterm birth among whites. Among blacks the decline in perinatal mortality rates was more closely associated with a decline in the preterm birth following ruptured membranes. Finally, the increases in medically indicated preterm birth !37 and !32 weeks were greater among whites compared with blacks. Although spontaneous preterm birth among whites increased, the temporal increase in this subtype was not associated with changes in perinatal mortality. Our previously cited study of singleton births in the US showed that the increase in preterm birth among whites was largely driven by an increase in medically indicated preterm birth, whereas among blacks, preterm birth rates declined because of a decrease in spontaneous preterm birth and preterm birth following ruptured membranes.15 It is not surprising that the increase in preterm birth among twins was largely driven by a concurrent increase in medically indicated preterm birth. There has been a temporal increase in older maternal age and in women with a history of subfertility or infertility among women with twin gestation.22-24 Increases in obstetric complications and associated obstetric interventions are to be expected. Arguably, the most interesting observation in our study pertains to trends in the medically indicated preterm birth. The increase in preterm birth rates in most industrialized countries has sometimes been deemed a ‘‘failure’’ of efforts to reduce preterm birth. Obstetric intervention, primarily labor induction and to a lesser extent, cesarean delivery, has increased among twin pregnancies.2,5,10,12 Most of the increase in such
1076.e8 obstetric intervention has occurred at R34 weeks, and to a lesser extent, at R32 weeks.2,5 With dramatic advances in neonatal care, as well as increased use of corticosteroids for acceleration of functional fetal maturity, preterm labor induction, and preterm cesarean delivery are being increasingly used to reduce stillbirth and neonatal death and associated morbidity in cases where serious fetal, and less commonly, maternal compromise is evident. This claim is supported by our study, which shows that the largest reduction in perinatal mortality among twins was associated with increases in medically indicated preterm birth. Thus, the increase in medically indicated preterm birth among both whites and blacks may therefore be viewed as beneficial insofar as such intervention reduces stillbirth and neonatal mortality. This population-based viewpoint does not necessarily apply to individual cases, however; note the increases in neonatal deaths due to short gestation and low birth weight. Also, further study needs to be directed at addressing whether population rates of other serious neonatal morbidity among twins have decreased as well. The classification of preterm birth subtypes in our study was based on the assumption that the subtypes are uniquedand distinctdwith regards to underlying etiologies. In reality, the different clinical subtypes of preterm birth share common determinants, and hence, an increase in a particular subtype may result in a decrease in another subtype. For instance, an increased awareness of the signs and symptoms of preterm labor may lead to an increase in spontaneous preterm birth at the expense of preterm birth following ruptured membranes. Although this assumption may have some implications in the interpretation of our findings, the differences, and potential overlap, between preterm birth subtypes may also indicate that etiologic studies, and preterm birth prevention programs, should examine preterm birth both in terms of its overall frequency, and also by subtype in order to more fully recognize and understand potential changes. This is particularly true given the general failure of preterm birth prevention programs, and also the recent identification of progesterone as a promising medical intervention for preventing recurrent preterm birth.25,26 The method for identifying preterm birth subtypes in our study may have resulted in some misclassification. This is especially true because the vital statistics data do not indicate whether labor preceded or followed rupture of membranes. Similarly, some cases identified as medically indicated preterm births may in fact have been in labor before the obstetric intervention. Nevertheless, the impact of such misclassification is mitigated by the use of the same algorithm for identifying the different subtypes across the study period, as well as a focus on relative, rather than absolute, changes in the rate of each subtype.
Ananth et al Errors in menstrual estimate of gestational age are likely to have affected our results to some extent.27,28 The clinical conditions examined here, such as premature rupture of membranes, labor induction, etc, noted on vital statistics certificates may be subject to some errors, including those caused by transcription. Also, the data do not distinguish gestational diabetes from preexisting diabetes. The cause of (neonatal) death in vital statistics data may be incomplete because only a single cause of death is assigned. Furthermore, the impact of race-specific preterm birth trends on postneonatal mortality remains unexplored, and may be worthy of future investigations. Another limitation of our analysis was our inability to adjust for nonindependent observations, namely, births within the same twin set. The NCHS did not identify the 2 twin fetuses within a sibship in these data sets. Such nonadjustment is likely to result in imprecise estimates of the variance (ie, larger confidence intervals) in individual level analyses, although the RRs themselves will remain unaffected.21 However, this limitation may be less effected by the ecologic analyses examining the association between population trends in preterm birth and changes in perinatal mortality. Despite these shortcomings, the large populationbased study allows generalizability of the results. The study also enabled us to detect important differences in the rates of clinical subtypes of twin preterm births in the US, and their impact on perinatal mortality, against a background of persistent black-white disparity in preterm birth. The ecological design minimizes possible bias caused by confounding by indication compared with a more traditional individual-level analysis. In summary, although the preterm birth rate among twins has been increasing in the US, the increase is largely associated by a concurrent increase in medically indicated preterm birth. The decline in perinatal mortality was most strongly associated with the increase in medically indicated preterm birth among whites, whereas among blacks, the decline in perinatal mortality was more closely associated with a decline in preterm birth following rupture of membranes. Our study also demonstrates a strong racial disparity in trends in twin preterm birth by clinical subtype and in perinatal mortality. The overall message is an encouraging one, however, because perinatal mortality rates among twins in the US appear to have declined substantially as a result of iatrogenic increases in preterm birth among twins.
References 1. Martin JA, Hamilton BE, Ventura SJ, Menacker F, Park MM. Births: final data for 2000. Natl Vital Stat Rep 2002;50:1-101. 2. Ananth CV, Joseph KS, Kinzler WL. The influence of obstetric intervention on trends in twin stillbirths in the United States. J Mat-Fetal Neonat Med 2004;15:380-7.
Ananth et al 3. Joseph KS, Kramer MS, Marcoux S, Ohlsson A, Wen SW, Allen A, et al. Determinants of preterm birth rates in Canada from 1981 through 1983 and from 1992 through 1994. N Engl J Med 1998;339: 1434-9. 4. Kramer MS, Platt R, Yang H, Joseph KS, Wen SW, Morin L, et al. Secular trends in preterm birth: a hospital-based cohort study. JAMA 1998;280:1849-54. 5. Joseph KS, Allen AC, Dodds L, Vincer MJ, Armson BA. Causes and consequences of recent increases in preterm birth among twins. Obstet Gynecol 2001;98:57-64. 6. Papiernik E, Zeitlin J, Rivera L, Bucourt M, Topuz B. Preterm birth in a French population: the importance of births by medical decision. BJOG 2003;110:430-2. 7. Breart G, Blondel B, Tuppin P, Grandjean H, Kaminski M. Did preterm deliveries continue to decrease in France in the 1980s? Paediatr Perinat Epidemiol 1995;9:296-306. 8. Olsen P, Laara E, Rantakallio P, Jarvelin MR, Sarpola A, Hartikainen AL. Epidemiology of preterm delivery in two birth cohorts with an interval of 20 years. Am J Epidemiol 1995;142: 1184-93. 9. Demissie K, Rhoads GG, Ananth CV, Alexander GR, Kramer MS, Kogan M, et al. Trends in preterm birth and neonatal mortality among blacks and whites in the United States from 1989 to 1997. Am J Epidemiol 2001;154:307-15. 10. Ananth CV, Joseph KS, Smulian JC. Trends in twin neonatal mortality in the United States: influence of obstetric intervention. Am J Obstet Gynecol 2004;190:1313-21. 11. Roberts CL, Algert CS, Raynes-Greenow C, Peat B, HendersonSmart D. Delivery of singleton preterm infants in New South Wales, 1990-1997. Aust NZ J Obstet Gynaecol 2003;43: 32-7. 12. Kogan MD, Alexander GR, Kotelchuck M, MacDorman MF, Buekens P, Martin JA, et al. Trends in twin birth outcomes and prenatal care utilization in the United States, 1981-1997. JAMA 2000;284:335-41. 13. Branum AM, Schoendorf KC. Changing patterns of low birthweight and preterm birth in the United States, 1981-98. Paediatr Perinat Epidemiol 2002;16:8-15. 14. Goldenberg RL, Jobe AH. Prospects for research in reproductive health and birth outcomes. JAMA 2001;285:633-9. 15. Ananth CV, Joseph KS, Oyelese Y, Demissie K, Vintzileos AM. Trends in preterm birth and perinatal mortality among singletons: United States 1989 through 2000. Obstet Gynecol 2005;105:1084-91.
1076.e9 16. National Center for Health Statistics. 1989-99 linked birth/infant death data set. NCHS CD-ROM series 20, No 3-18. Hyattsville, MD: United States Department of Health and Human Services, Centers for Disease Control and Prevention. 17. Taffel SM, Ventura SJ, Gay GA. Revised US certificate of birthdnew opportunities for research on birth outcome. Birth 1989;16:188-93. 18. Taffel S, Johnson D, Heuse R. A method of imputing length of gestation on birth certificates. Vital Health Stat 2 1982;93:1-11. 19. Zhang J, Yu KF. What’s the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA 1998;280:1690-1. 20. Wen SW, Kramer MS. Uses of ecologic studies in the assessment of intended treatment effects. J Clin Epidemiol 1999;52:7-12. 21. Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generalized estimating equation approach. Biometrics 1988;44: 1049-60. 22. Cunningham GF, Gant NF, Leveno KJ, Gilstrap LC, Hauth JC, Wenstrom KD. Williams obstetrics, 21st ed. New York, NY: McGraw-Hill Professional; 2001. 23. Misra DP, Ananth CV. Infant mortality among singletons and twins in the United States during 2 decades: effects of maternal age. Pediatrics 2002;110:1163-8. 24. Bergh T, Ericson A, Hillensjo T, Nygren KG, Wennerholm UB. Deliveries and children born after in-vitro fertilisation in Sweden 1982-95: a retrospective cohort study. Lancet 1999;354:1579-85. 25. Meis PJ, Klebanoff M, Thom E, Dombrowski MP, Sibai B, Moawad AH, et al, for the National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. Prevention of recurrent preterm delivery by 17 alpha-hydroxyprogesterone caproate. N Engl J Med 2003;348:2379-85. 26. da Fonseca EB, Bittar RE, Carvalho MH, Zugaib M. Prophylactic administration of progesterone by vaginal suppository to reduce the incidence of spontaneous preterm birth in women at increased risk: a randomized placebo-controlled double-blind study. Am J Obstet Gynecol 2003;188:419-24. 27. Gjessing HK, Skjaerven R, Wilcox AJ. Errors in gestational age: evidence of bleeding early in pregnancy. Am J Public Health 1999;89:213-8. 28. Yang H, Kramer MS, Platt RW, Blondel B, Breart G, Morin I, et al. How does early ultrasound scan estimation of gestational age lead to higher rates of preterm birth? Am J Obstet Gynecol 2002;186:433-7.
www.ajog.org
Trends in twin preterm birth subtypes in the United States, 1989 through 2000: Impact on perinatal mortality Cande V. Ananth, PhD, MPH,a,b,* K. S. Joseph, MD, PhD,c Kitaw Demissie, MD, PhD,b Anthony M. Vintzileos, MDd Division of Epidemiology and Biostatistics, Department of Obstetrics, Gynecology, and Reproductive Sciences, Robert Wood Johnson Medical School,a University of Medicine and Dentistry of New Jersey, New Brunswick, NJ; Department of Epidemiology, School of Public Health,b University of Medicine and Dentistry of New Jersey, Piscataway, NJ; Perinatal Epidemiology Research Unit, Departments of Obstetrics and Gynaecology, and of Pediatrics, Dalhousie University,c Halifax, Canada; Division of Maternal-Fetal Medicine,d Department of Obstetrics, Gynecology, and Reproductive Sciences, Robert Wood Johnson Medical School, University of Medicine and Dentistry of New Jersey, New Brunswick, NJ Received for publication February 8, 2005; revised June 1, 2005; accepted June 7, 2005
KEY WORDS Preterm delivery Trends Premature rupture of membranes Labor induction Cesarean delivery Twins Perinatal mortality
Objective: We examined trends in twin preterm birth !37 weeks following ruptured membranes (ROM), medically indicated preterm birth, and preterm birth following spontaneous onset of labor (PTL). We further examined whether the changes in preterm birth subtypes were associated with trends in twin perinatal mortality. Study design: We carried out a retrospective cohort study of 1,172,405 twin live births and stillbirths delivered in the US between 1989 and 2000. Trends in preterm birth subtypes and perinatal mortality (stillbirths at R22 weeks plus neonatal deaths within 28 days) were examined through ecological logistic regression models after adjusting for confounders. Results: Twin preterm birth among whites increased from 46.6% in 1989 to 1990 to 56.7% in 1999 to 2000, and from 56.1% to 61.0% among blacks over the same period. Medically indicated preterm birth increased by 50% (95% CI 49-52) among whites, and by 33% (95% CI 29-36) among blacks. PTL increased by 24% among whites, but remained fairly unchanged among blacks between the two periods. Preterm birth following ROM also did not change between the 2 periods among whites, but declined by 7% among blacks. Perinatal mortality among twin births declined by 41% (95% CI 38-44) among whites, and by 37% (95% CI 32-42) among blacks between 1989 and 1990 and 1999 and 2000. This mortality decline was most closely associated with the increase in medically
Presented at the Twenty-Fifth Annual Meeting of the Society for Maternal-Fetal Medicine Meeting, Reno, NV, February 7-12, 2005. Dr Ananth is partially supported through a grant (R01-HD038902) from the National Institutes of Health. Dr Joseph is supported by a Clinical Research Scholarship from the Dalhousie University Faculty of Medicine, Nova Scotia, Canada, and is a Peter Lougheed New Investigator of the Canadian Institutes of Health Research. * Reprint requests: Cande V. Ananth, PhD, MPH, Division of Epidemiology and Biostatistics, Department of Obstetrics, Gynecology, and Reproductive Sciences, UMDNJ-Robert Wood Johnson Medical School, 125 Paterson Street, New Brunswick NJ 08901-1977. E-mail: [email protected] 0002-9378/$ - see front matter Ó 2005 Mosby, Inc. All rights reserved. doi:10.1016/j.ajog.2005.06.088
1076.e2
Ananth et al indicated preterm birth among whites, and with the decrease in preterm birth following ROM among blacks. Conclusion: Temporal trends in twin preterm birth varied substantially based on underlying subtypes and race. The increase in medically indicated preterm birth is associated with a large reduction in perinatal mortality. Ó 2005 Mosby, Inc. All rights reserved.
Preterm birth (!37 weeks) remains one of the leading causes of adverse perinatal outcomes, including neonatal and infant mortality. Compared with infants delivered at term, preterm infants also suffer greater neurodevelopmental problems and increased morbidity resulting from respiratory complications. The rate of preterm birth among twins has increased dramatically in recent years1-4 in the US1,2 and other industrialized countries,3,4 largely because of increased obstetric intervention (ie, labor induction and cesarean delivery).5 In addition, increases in assisted reproduction technologies, and in older maternal age, have also affected these trends. With the possible exception of France6,7 and Finland,8 most other industrialized countries have experienced a temporal increase in overall preterm birth rates.2-4,9-13 The increase in preterm birth rates has been puzzling, and often regarded as a failure of clinical and community-based efforts to reduce preterm birth.14 Preterm birth is a multifactorial condition with at least 3 distinct clinical pathways: preterm birth resulting from ruptured membranes; preterm birth resulting from medical/obstetric intervention; and preterm birth following spontaneous onset of labor. We recently showed that preterm birth among singleton whites in the US has increased largely because of a concurrent increase in medically indicated preterm birth, whereas among singleton blacks, preterm birth rates declined because of a decrease in spontaneous preterm birth and preterm birth following ruptured membranes.15 Further, we showed that the temporal increase in preterm birth among singletons was associated with a reduction in perinatal mortality, with the largest reductions in mortality being associated with medically indicated preterm birth.15 We carried out a study of trends in twin preterm birth subtypes because it is unclear whether the associations seen among singleton births15 also occurred among twins. We hypothesized that medically indicated preterm birth among twins would have increased over the last decade in the US, and that these increases would have led to a large and substantial decline in twin perinatal mortality.
Material and methods United States twin birth cohort composition We used publicly available data from the US twin natality data files from 1989 through 2000 to examine
temporal trends in preterm birth subtypes, and linked natality and infant mortality data files 1989 through 1991 and 1995 through 2000, to examine the impact of preterm birth trends on perinatal mortality.16 The data contained in these files were obtained from live birth and fetal and infant death certificates, and were assembled and compiled by the National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention.17 Because the National Center for Health Statistics did not link the live birth data files to their corresponding infant death files for 1992 through 1994, analysis of neonatal mortality in the 1992 through 1994 period could not be performed. Gestational age data, reported in completed weeks, on vital statistics records were predominantly based on the date of the last menstrual period, calculated as the difference between the date of delivery and the date of the last menstrual period. In a small fraction of births (!5%) that did not contain the date of the menstrual period, or when the menstrual-based gestational age was grossly incompatible with birth weight, a clinical estimate of gestation, also contained on vital records, was used instead.17 Imputation of gestational age (in completed weeks) was also performed by the NCHS National Center for Health Statistics for births that contained a valid month and year for the last menstrual period, but had a missing date.18 The replacement of menstrual estimate of gestational age by a clinical estimate, as well as imputation of missing gestational age, was performed consistently for all years (1989 through 2000).
Preterm birth classification Trends in twin preterm birth !37 weeks were assessed after stratifying births by underlying clinical subtype. Preterm birth was sequentially classified into 3 distinct clinical subtypes based on the following hierarchy: preterm birth following rupture of membranes; medically indicated preterm birth, defined as a preterm birth which followed iatrogenic intervention (labor induction or a primary or repeat cesarean); and preterm birth following spontaneous onset of labor, defined as preterm birth that was neither associated with ruptured membranes nor medically indicated. The determination of preterm birth subtype was based on 4 check-box items on the live birth, and fetal, and infant death certificates.17 The item ‘‘ruptured membranes’’ accompanied by delivery at !37 weeks was first
Ananth et al used to identify twins who delivered preterm following ruptured membranes. The items ‘‘primary cesarean,’’ ‘‘repeat cesarean,’’ or ‘‘labor induction,’’ accompanied by delivery at !37 weeks were combined to identify twins who delivered preterm because of medical indications. The item corresponding to labor induction represents an attempt to induce labor, regardless of its success.
Perinatal mortality We examined trends in twin perinatal mortality rates between 1989 and 2000 within the 3 clinical subtypes of preterm birth. Perinatal mortality was defined as stillbirths at R22 weeks plus death to liveborn twins within the first 28 days after birth, and was expressed per 1000 total twin births. Changes in cause-specific neonatal mortality rates between 1989 and 1990 and 1999 and 2000 were also assessed. For this analysis, the cause of death assigned on the mortality data was based on the International Classification of Diseases, ninth revision (ICD-9) for the 1989 and 1990 data, and the 10th revision (ICD-10) for the 1999 and 2000 periods. The following cause of death categories were examined: congenital anomalies (ICD-9 740-759; ICD-10 Q00-Q99); maternal complications (ICD-9 761; ICD-10 P01); complications of the placenta, cord, or membrane (ICD-9 762; ICD-10 P02); short gestation or low birth weight (ICD-9 765; ICD-10 P05 and P07); intrauterine hypoxia or asphyxia (ICD-9 768; ICD-10 P20 and P21); respiratory distress syndrome (ICD-9 769; ICD-10 P22); infections during the perinatal period (ICD-9 771; ICD-10 P23, P35, P36, P37, and P39); and sudden infant death syndrome (ICD-9 798.0; ICD-10 R95). Causes of stillbirth were not examined because such data were unavailable.
1076.e3 regression model will likely overestimate the RR. We, therefore, converted the odds ratio obtained from logistic models to RR.19 The contribution of trends in the 3 clinical subtypes of preterm birth to trends in perinatal mortality was also assessed. For this analysis, with perinatal deaths as the outcome, we aggregated the data by each US state for the 2 periods (1989-1990 and 1999-2000). In each of the 100 strata (50 states times 2 periods), we separately determined the proportion for each of the 3 preterm birth subtypes (preterm birth proportion was an independent variable in these regression models). Adjustment for confounding because of maternal age !20 and R35 years, nulliparity, maternal education !12 years of schooling, and unmarried status was accomplished using the proportion of each of these confounders in each of the 100 strata. The ecological design of the study was necessary in order to avoid confounding by indication, which occurs when attempting to estimate the effect of a clinical intervention20 using individual-level analysis and a nonexperimental study design. Such confounding is likely because labor induction and cesarean delivery are typically used in pregnancies where fetal compromise is suspected, and because antepartum stillbirths are often delivered by inducing labor. We regressed the preterm birth variable (in addition to state, the period variable, and confounders) against perinatal death. From these models, we obtained the RR and 95% CI denoting the change in perinatal mortality rates between 1989 and 2000 before and after adjustment for temporal changes in preterm birth subtypes. This analysis was then repeated for stillbirths and neonatal deaths separately. We also repeated all analyses for preterm birth !32 weeks.
Data exclusions Statistical analysis Changes in the rates of preterm birth !37 weeks and preterm birth subtypes among twins between 1989 and 1990 and 1999 and 2000 among white and black women were examined first, and quantified using relative risks (RR) with 95% CIs. In a similar fashion, changes in perinatal mortality rates were assessed by comparing mortality rates across the 2 periods (1989-1990 and 1999-2000) using RRs. Given the population-based nature of our study, we present CI of effect measures as a means of assessing the precision of the estimate. We fit logistic regression models with preterm birth as the dependent variable, and an indicator for period as the independent variable (1989-1990 served as the reference). These models also included maternal age, parity, maternal education, and marital status in order to provide adjusted estimates of the temporal change in preterm birth and its subtypes. Because preterm birth rates among twins are high, odds ratio from the logistic
Of all US twin births delivered in the US between 1989 and 2000, we sequentially excluded those with missing gestational age (n Z 14,110), gestational age !22 completed weeks (n Z 19,535), and pregnancies that ended at R44 weeks (n Z 3392). We additionally excluded 52,050 twin births to non-black and non-white women. After all exclusions, 1,172,405 twin live births and stillbirths occurring between 1989 and 2000 remained for analysis (951,778 white; 220,627 black). Data on labor induction were missing in a sizable proportion (over 25%) of delivery records from Louisiana, and so data from this state were excluded in subanalyses pertaining to preterm birth subtype. This exclusion was necessary because the missing information was needed for the classification of preterm birth by subtype (ie, medically indicated preterm birth). However, for overall analysis of preterm birth trends (ie, without consideration of preterm birth subtypes), such data were retained.
1076.e4
Ananth et al
Figure 1 Temporal changes in preterm birth (!37 weeks), preterm birth following ruptured membranes, medically indicated preterm birth, and spontaneous preterm birth rates (left), and temporal changes in perinatal mortality rates (right), US whites, 1989 through 2000.
This study was approved by the ethics review committee of the Institutional Review Board of the UMDNJ-Robert Wood Johnson Medical School, NJ.
Results Among whites, the rate of preterm birth (!37 weeks) among twins increased by 31%, from 43.5% in 1989 to 56.8% in 2000 (Figure 1). Similarly, the preterm birth rate among black twins increased by 16%, from 52.8% in 1989 to 61.2% in 2000 (Figure 2). Changes in sociodemographic determinants and complications during pregnancy between 1989 and 1990 and 1999 and 2000 are shown in Table I. During the periods 1989 and 1990 and 1999 and 2000, the proportion of women aged R35 years increased by 81% among whites, and by 69% among blacks (Table I). Preexisting/gestational diabetes also increased between these 2 periods among both whites and blacks, while an increase in the rate of chronic hypertension was evident only among whites. Rates of labor induction and, to a lesser extent, cesarean delivery also increased among both whites and blacks between 1989 and 1990 and 1999 and 2000.
The overall increase in preterm birth (!37 weeks) among both whites and blacks was largely driven by a concurrent relative increase in the rate of medically indicated preterm birth (Table II). Between 1989 and 1990 and 1999 and 2000, rates of medically indicated preterm birth !37 weeks (adjusted for maternal age, nulliparity, maternal education, and marital status) increased by 50% (RR 1.50, 95% CI 1.49-1.52) among whites, and by 33% (RR 1.33, 95% CI 1.29-1.36) among blacks. Adjusted rates of spontaneous preterm birth among whites increased by 24% (RR 1.24, 95% CI 1.221.26), but remained virtually unchanged among blacks during the 2 periods. Adjusted rates of preterm birth following ruptured membranes remained unchanged among whites, but declined by 7% among blacks between the 2 periods. Preterm birth !32 weeks increased by 22% and 5% among whites and blacks, respectively, between 1989 and 1990 and 1999 and 2000. An increase in overall preterm birth !32 weeks between 1989 and 1990 and 1999 and 2000 was also evident in the analyses after adjustments for maternal age, nulliparity, maternal education, and marital status. A 22% increase was
Ananth et al
1076.e5
Figure 2 Temporal changes in preterm birth (!37 weeks), preterm birth following ruptured membranes, medically indicated preterm birth, and spontaneous preterm birth rates (left), and temporal changes in perinatal mortality rates (right), US blacks, 1989 through 2000.
Table I
Changes in maternal characteristics of twin births in the US, 1989-1990 and 1999-2000 White women
Black women
Rate per 100 Maternal characteristics
1989-1990
1999-2000
Relative risk (95% CI)
Maternal age !20 years Maternal age R35 years Primigravida Maternal education !12 years Unmarried status Chronic hypertension Pregnancy-induced hypertension Preexisting/gestational diabetes Cesarean delivery Labor induction
7.2 12.7 17.1 18.3 17.2 0.8 7.5 2.5 55.3 6.9
6.8 23.0 18.0 14.4 19.5 0.9 9.0 3.6 57.3 14.7
0.94 1.81 1.05 0.79 1.14 1.22 1.22 1.45 1.04 2.13
observed among whites, and a 5% increase among blacks. Adjusted rates of medically indicated preterm birth increased by 29% and 20% among whites and blacks, respectively, between the 2 periods. Spontaneous preterm birth and preterm birth following ruptured
(0.92-0.97) (1.77-1.85) (1.03-1.07) (0.78-0.80) (1.12-1.16) (1.13-1.32) (1.19-1.25) (1.38-1.52) (1.03-1.05) (2.08-2.18)
Rate per 100 1989-1990
1999-2000
Relative risk (95% CI)
16.8 8.8 12.1 27.0 64.6 1.8 6.8 1.9 51.6 4.5
14.4 13.9 12.2 21.6 65.8 1.9 8.1 3.1 55.8 10.7
0.84 1.69 1.02 0.75 1.06 1.06 1.20 1.63 1.09 2.39
(0.80-0.87) (1.60-1.77) (0.97-1.06) (0.72-0.77) (1.03-1.09) (0.95-1.18) (1.14-1.27) (1.45-1.84) (1.06-1.13) (2.24-2.54)
membranes (!32 weeks) declined among both whites and blacks between 1989 and 1990 and 1999 and 2000. Among both white and blacks, perinatal mortality rates among twin births declined between 1989 and 1990 and 1990 and 2000 even as preterm birth rates increased
1076.e6 Table II
Ananth et al Changes in preterm birth rates !37 and !32 weeks: twin live births in the US, 1989-1990 and 1999-2000 White women
Black women
Preterm birth rate (%) Preterm birth subtypes
1989-1990
1999-2000
Adjusted RR (95% CI)
Number of twin births Preterm birth !37 weeks Ruptured membranes Medically indicated Spontaneous preterm birth Preterm birth !32 weeks Ruptured membranes Medically indicated Spontaneous preterm birth
144,330 46.6 6.2 24.1 16.2 10.7 2.5 4.7 3.4
183,921 56.7 4.8 35.2 16.6 10.7 1.8 6.1 2.8
1.22 0.98 1.50 1.24 1.22 0.75 1.29 0.86
(1.21-1.23) (0.95-1.01) (1.49-1.52) (1.22-1.26) (1.19-1.24) (0.71-0.79) (1.25-1.33) (0.82-0.89)
Preterm birth rate (%) 1989-1990
1999-2000
Adjusted RR (95% CI)
36,440 56.1 7.6 26.5 21.9 19.2 3.9 7.8 7.4
39,927 61.0 6.2 35.3 19.5 18.0 3.2 9.5 5.3
1.11 0.93 1.33 1.02 1.05 0.81 1.20 0.75
(1.10-1.12) (0.87-0.99) (1.29-1.36) (0.99-1.05) (1.02-1.09) (0.75-0.88) (1.14-1.26) (0.71-0.79)
Relative risks were adjusted for the confounding effects caused by maternal age, parity, maternal education, and marital status.
Table III Changes in perinatal mortality and association with changes in preterm birth !37 and !32 weeks: twin births in the US, 1989-1990 and 1999-2000 White women
Black women
Perinatal mortality* 1989-1990 All births Adjusted for confoundersy
32.7 –
1999-2000 19.4 –
Relative risk (95% CI)
Perinatal mortality* 1989-1990
1999-2000
Relative risk (95% CI)
0.59 (0.56-0.62) 0.61 (0.58-0.64)
43.8 –
27.9 –
0.63 (0.58-0.68) 0.64 (0.59-0.69)
Association between preterm birth !37 weeks and perinatal mortality Adjusted for confounders plus Preterm birth following ROM 77.3 52.9 0.62 (0.56-0.69) Medically indicated preterm birth 44.1 25.2 0.69 (0.66-0.73) Spontaneous preterm birth 63.3 33.4 0.61 (0.59-0.63)
101.7 47.1 77.3
75.7 31.2 48.6
0.68 (0.58-0.79) 0.66 (0.58-0.74) 0.61 (0.54-0.69)
Association between preterm birth !32 Adjusted for confounders plus Preterm birth following ROM Medically indicated preterm birth Spontaneous preterm birth
177.2 120.6 199.9
137.9 86.5 151.9
0.69 (0.60-0.78) 0.68 (0.61-0.76) 0.69 (0.62-0.76)
weeks and perinatal mortality 170.6 156.6 253.3
126.2 102.9 165.5
0.66 (0.62-0.71) 0.59 (0.55-0.63) 0.63 (0.58-0.67)
All relative risks express the period change in perinatal mortality (1999-2000 vs 1989-1990). * Perinatal mortality rates (crude) are expressed per 1000 total twin births. Perinatal mortality rate for each of the 3 preterm birth subtype category was calculated as the number of perinatal deaths in each preterm birth subtype category (eg, medically indicated preterm birth), divided by the total number of preterm births within that subtype (eg, medically indicated preterm birth). y Relative risks were adjusted for the confounding effects caused by maternal age, parity, maternal education, and marital status.
(Figures 1 and 2). Crude perinatal mortality rate declined by 41% and 37% among whites and blacks, respectively, between 1989 and 1990 and 1999 and 2000 (Table III). Changes in medically indicated preterm birth rates were most closely associated (inversely) with changes in perinatal mortality rates. For instance, among whites, the adjustment for the increase in medically indicated preterm birth between 1989 and 1990 and 1999 and 2000 (Table II) was associated with a decline in the period effect for perinatal mortality from RR 0.61 (95% CI 0.58-0.64) to RR 0.69 (95% CI 0.660.73). In other words, had medically indicated preterm birth (!37 weeks) among whites not increased by 50%, perinatal mortality among whites would have declined
by 31% and not 39%. Changes in preterm birth following ruptured membranes and spontaneous preterm birth among whites were not associated with substantial changes in perinatal mortality among whites (Table III). Among blacks, the decline in perinatal mortality was most strongly associated with a concurrent decline in preterm birth (!37 weeks) following ruptured membranes. In other words, had preterm birth following ruptured membranes not declined by 7%, perinatal mortality would have declined only by 32% (RR 0.68, 95% CI 0.58-0.79) and not 36% (RR 0.64, 95% CI 0.590.69). Similar patterns of associations between preterm birth !32 weeks and perinatal mortality were observed.
Ananth et al
1076.e7
Table IV Changes in cause-specific neonatal mortality among preterm births (!37 weeks): twin births in the United States, 19891990 and 1999-2000 White women Preterm birth !37 weeks subtypes
Black women
Neonatal mortality 1989-1990
1999-2000
Adjusted RR (95% CI)
Neonatal deaths resulting from respiratory distress syndrome Ruptured membranes 16.8 5.9 0.40 (0.28-0.58) Medically indicated 6.4 2.2 0.34 (0.27-0.44) Spontaneous labor 11.8 2.5 0.23 (0.17-0.31) Neonatal deaths resulting from congenital malformations Ruptured membranes 8.3 3.6 Medically indicated 6.2 3.3 Spontaneous labor 4.2 1.9
0.47 (0.30-0.79) 0.60 (0.48-0.75) 0.49 (0.33-0.72)
Neonatal deaths resulting from complications of short gestation/low birthweight Ruptured membranes 4.5 14.0 3.26 (2.06-5.16) Medically indicated 0.8 1.7 2.61 (1.56-4.38) Spontaneous labor 5.2 10.3 1.97 (1.54-2.52)
Neonatal mortality 1989-1990
1999-2000
Adjusted RR (95% CI)
22.8 12.1 16.4
10.4 3.0 5.2
0.49 (0.29-0.84) 0.35 (0.24-0.53) 0.42 (0.29-0.62)
6.6 5.1 4.0
5.6 2.7 1.2
0.98 (0.39-2.41) 0.63 (0.38-1.02) 0.29 (0.13-0.65)
10.8 3.0 7.6
24.9 3.6 19.6
2.65 (1.59-4.34) 1.49 (0.85-2.60) 3.31 (2.31-4.74)
All relative risks express the period change in neonatal mortality (1999-2000 vs 1989-1990). Cause-specific neonatal mortality rates are expressed per 1000 twin live births delivered preterm. Cause-specific neonatal mortality rate for each of the 3 preterm birth subtype categories was calculated as the number of neonatal deaths with a particular cause of death in each preterm birth subtype category (eg, medically indicated preterm birth), divided by the total number of preterm births within that subtype (eg, medically indicated preterm birth). Relative risks were adjusted for the confounding effects caused by maternal age, parity, maternal education, and marital status.
The association between increases in each preterm birth subtype and neonatal mortality were also generally similar to associations with perinatal mortality. Cause-specific neonatal mortality rates within each preterm birth subtype are shown in Table IV. There were significant declines in neonatal deaths caused by respiratory distress syndrome, and congenital malformations and increases in neonatal deaths caused by complications that accompanying short gestational age and low birth weight. Neonatal mortality from other causes, including those caused by maternal complications, complications of placenta, cord, and membranes, intrauterine hypoxia or birth asphyxia, infections during the perinatal period, and sudden infant death syndrome in relation to preterm birth subtypes remained unchanged (data not shown).
Comment Despite extensive research, preterm birth continues to pose an enigma in term of its etiology and prevention. Our study indicates that, among twin births in the US, the temporal increase in preterm birth among both white and black women was largely driven by a concurrent temporal increase in medically indicated preterm birth. This increase in medically indicated preterm birth was less steep among twins delivered at !32 weeks than among those delivered at !37 weeks. Perinatal mortality rates declined, and as hypothesized, the decline was strongly associated with increases in medically indicated
preterm birth among whites. Among blacks the decline in perinatal mortality rates was more closely associated with a decline in the preterm birth following ruptured membranes. Finally, the increases in medically indicated preterm birth !37 and !32 weeks were greater among whites compared with blacks. Although spontaneous preterm birth among whites increased, the temporal increase in this subtype was not associated with changes in perinatal mortality. Our previously cited study of singleton births in the US showed that the increase in preterm birth among whites was largely driven by an increase in medically indicated preterm birth, whereas among blacks, preterm birth rates declined because of a decrease in spontaneous preterm birth and preterm birth following ruptured membranes.15 It is not surprising that the increase in preterm birth among twins was largely driven by a concurrent increase in medically indicated preterm birth. There has been a temporal increase in older maternal age and in women with a history of subfertility or infertility among women with twin gestation.22-24 Increases in obstetric complications and associated obstetric interventions are to be expected. Arguably, the most interesting observation in our study pertains to trends in the medically indicated preterm birth. The increase in preterm birth rates in most industrialized countries has sometimes been deemed a ‘‘failure’’ of efforts to reduce preterm birth. Obstetric intervention, primarily labor induction and to a lesser extent, cesarean delivery, has increased among twin pregnancies.2,5,10,12 Most of the increase in such
1076.e8 obstetric intervention has occurred at R34 weeks, and to a lesser extent, at R32 weeks.2,5 With dramatic advances in neonatal care, as well as increased use of corticosteroids for acceleration of functional fetal maturity, preterm labor induction, and preterm cesarean delivery are being increasingly used to reduce stillbirth and neonatal death and associated morbidity in cases where serious fetal, and less commonly, maternal compromise is evident. This claim is supported by our study, which shows that the largest reduction in perinatal mortality among twins was associated with increases in medically indicated preterm birth. Thus, the increase in medically indicated preterm birth among both whites and blacks may therefore be viewed as beneficial insofar as such intervention reduces stillbirth and neonatal mortality. This population-based viewpoint does not necessarily apply to individual cases, however; note the increases in neonatal deaths due to short gestation and low birth weight. Also, further study needs to be directed at addressing whether population rates of other serious neonatal morbidity among twins have decreased as well. The classification of preterm birth subtypes in our study was based on the assumption that the subtypes are uniquedand distinctdwith regards to underlying etiologies. In reality, the different clinical subtypes of preterm birth share common determinants, and hence, an increase in a particular subtype may result in a decrease in another subtype. For instance, an increased awareness of the signs and symptoms of preterm labor may lead to an increase in spontaneous preterm birth at the expense of preterm birth following ruptured membranes. Although this assumption may have some implications in the interpretation of our findings, the differences, and potential overlap, between preterm birth subtypes may also indicate that etiologic studies, and preterm birth prevention programs, should examine preterm birth both in terms of its overall frequency, and also by subtype in order to more fully recognize and understand potential changes. This is particularly true given the general failure of preterm birth prevention programs, and also the recent identification of progesterone as a promising medical intervention for preventing recurrent preterm birth.25,26 The method for identifying preterm birth subtypes in our study may have resulted in some misclassification. This is especially true because the vital statistics data do not indicate whether labor preceded or followed rupture of membranes. Similarly, some cases identified as medically indicated preterm births may in fact have been in labor before the obstetric intervention. Nevertheless, the impact of such misclassification is mitigated by the use of the same algorithm for identifying the different subtypes across the study period, as well as a focus on relative, rather than absolute, changes in the rate of each subtype.
Ananth et al Errors in menstrual estimate of gestational age are likely to have affected our results to some extent.27,28 The clinical conditions examined here, such as premature rupture of membranes, labor induction, etc, noted on vital statistics certificates may be subject to some errors, including those caused by transcription. Also, the data do not distinguish gestational diabetes from preexisting diabetes. The cause of (neonatal) death in vital statistics data may be incomplete because only a single cause of death is assigned. Furthermore, the impact of race-specific preterm birth trends on postneonatal mortality remains unexplored, and may be worthy of future investigations. Another limitation of our analysis was our inability to adjust for nonindependent observations, namely, births within the same twin set. The NCHS did not identify the 2 twin fetuses within a sibship in these data sets. Such nonadjustment is likely to result in imprecise estimates of the variance (ie, larger confidence intervals) in individual level analyses, although the RRs themselves will remain unaffected.21 However, this limitation may be less effected by the ecologic analyses examining the association between population trends in preterm birth and changes in perinatal mortality. Despite these shortcomings, the large populationbased study allows generalizability of the results. The study also enabled us to detect important differences in the rates of clinical subtypes of twin preterm births in the US, and their impact on perinatal mortality, against a background of persistent black-white disparity in preterm birth. The ecological design minimizes possible bias caused by confounding by indication compared with a more traditional individual-level analysis. In summary, although the preterm birth rate among twins has been increasing in the US, the increase is largely associated by a concurrent increase in medically indicated preterm birth. The decline in perinatal mortality was most strongly associated with the increase in medically indicated preterm birth among whites, whereas among blacks, the decline in perinatal mortality was more closely associated with a decline in preterm birth following rupture of membranes. Our study also demonstrates a strong racial disparity in trends in twin preterm birth by clinical subtype and in perinatal mortality. The overall message is an encouraging one, however, because perinatal mortality rates among twins in the US appear to have declined substantially as a result of iatrogenic increases in preterm birth among twins.
References 1. Martin JA, Hamilton BE, Ventura SJ, Menacker F, Park MM. Births: final data for 2000. Natl Vital Stat Rep 2002;50:1-101. 2. Ananth CV, Joseph KS, Kinzler WL. The influence of obstetric intervention on trends in twin stillbirths in the United States. J Mat-Fetal Neonat Med 2004;15:380-7.
Ananth et al 3. Joseph KS, Kramer MS, Marcoux S, Ohlsson A, Wen SW, Allen A, et al. Determinants of preterm birth rates in Canada from 1981 through 1983 and from 1992 through 1994. N Engl J Med 1998;339: 1434-9. 4. Kramer MS, Platt R, Yang H, Joseph KS, Wen SW, Morin L, et al. Secular trends in preterm birth: a hospital-based cohort study. JAMA 1998;280:1849-54. 5. Joseph KS, Allen AC, Dodds L, Vincer MJ, Armson BA. Causes and consequences of recent increases in preterm birth among twins. Obstet Gynecol 2001;98:57-64. 6. Papiernik E, Zeitlin J, Rivera L, Bucourt M, Topuz B. Preterm birth in a French population: the importance of births by medical decision. BJOG 2003;110:430-2. 7. Breart G, Blondel B, Tuppin P, Grandjean H, Kaminski M. Did preterm deliveries continue to decrease in France in the 1980s? Paediatr Perinat Epidemiol 1995;9:296-306. 8. Olsen P, Laara E, Rantakallio P, Jarvelin MR, Sarpola A, Hartikainen AL. Epidemiology of preterm delivery in two birth cohorts with an interval of 20 years. Am J Epidemiol 1995;142: 1184-93. 9. Demissie K, Rhoads GG, Ananth CV, Alexander GR, Kramer MS, Kogan M, et al. Trends in preterm birth and neonatal mortality among blacks and whites in the United States from 1989 to 1997. Am J Epidemiol 2001;154:307-15. 10. Ananth CV, Joseph KS, Smulian JC. Trends in twin neonatal mortality in the United States: influence of obstetric intervention. Am J Obstet Gynecol 2004;190:1313-21. 11. Roberts CL, Algert CS, Raynes-Greenow C, Peat B, HendersonSmart D. Delivery of singleton preterm infants in New South Wales, 1990-1997. Aust NZ J Obstet Gynaecol 2003;43: 32-7. 12. Kogan MD, Alexander GR, Kotelchuck M, MacDorman MF, Buekens P, Martin JA, et al. Trends in twin birth outcomes and prenatal care utilization in the United States, 1981-1997. JAMA 2000;284:335-41. 13. Branum AM, Schoendorf KC. Changing patterns of low birthweight and preterm birth in the United States, 1981-98. Paediatr Perinat Epidemiol 2002;16:8-15. 14. Goldenberg RL, Jobe AH. Prospects for research in reproductive health and birth outcomes. JAMA 2001;285:633-9. 15. Ananth CV, Joseph KS, Oyelese Y, Demissie K, Vintzileos AM. Trends in preterm birth and perinatal mortality among singletons: United States 1989 through 2000. Obstet Gynecol 2005;105:1084-91.
1076.e9 16. National Center for Health Statistics. 1989-99 linked birth/infant death data set. NCHS CD-ROM series 20, No 3-18. Hyattsville, MD: United States Department of Health and Human Services, Centers for Disease Control and Prevention. 17. Taffel SM, Ventura SJ, Gay GA. Revised US certificate of birthdnew opportunities for research on birth outcome. Birth 1989;16:188-93. 18. Taffel S, Johnson D, Heuse R. A method of imputing length of gestation on birth certificates. Vital Health Stat 2 1982;93:1-11. 19. Zhang J, Yu KF. What’s the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA 1998;280:1690-1. 20. Wen SW, Kramer MS. Uses of ecologic studies in the assessment of intended treatment effects. J Clin Epidemiol 1999;52:7-12. 21. Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generalized estimating equation approach. Biometrics 1988;44: 1049-60. 22. Cunningham GF, Gant NF, Leveno KJ, Gilstrap LC, Hauth JC, Wenstrom KD. Williams obstetrics, 21st ed. New York, NY: McGraw-Hill Professional; 2001. 23. Misra DP, Ananth CV. Infant mortality among singletons and twins in the United States during 2 decades: effects of maternal age. Pediatrics 2002;110:1163-8. 24. Bergh T, Ericson A, Hillensjo T, Nygren KG, Wennerholm UB. Deliveries and children born after in-vitro fertilisation in Sweden 1982-95: a retrospective cohort study. Lancet 1999;354:1579-85. 25. Meis PJ, Klebanoff M, Thom E, Dombrowski MP, Sibai B, Moawad AH, et al, for the National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. Prevention of recurrent preterm delivery by 17 alpha-hydroxyprogesterone caproate. N Engl J Med 2003;348:2379-85. 26. da Fonseca EB, Bittar RE, Carvalho MH, Zugaib M. Prophylactic administration of progesterone by vaginal suppository to reduce the incidence of spontaneous preterm birth in women at increased risk: a randomized placebo-controlled double-blind study. Am J Obstet Gynecol 2003;188:419-24. 27. Gjessing HK, Skjaerven R, Wilcox AJ. Errors in gestational age: evidence of bleeding early in pregnancy. Am J Public Health 1999;89:213-8. 28. Yang H, Kramer MS, Platt RW, Blondel B, Breart G, Morin I, et al. How does early ultrasound scan estimation of gestational age lead to higher rates of preterm birth? Am J Obstet Gynecol 2002;186:433-7.