The same phenomenon was evident for neonatal death (Table 3). In Canada, the rate ratio for neonatal death at preterm relative to term gestation was 26.0 (95% CI 23.8–28.4), whereas in the United States it was 18.6 (95% CI 18.2–19.0, P value for difference in rate ratios<.001) for the menstrual-based gestational age estimate and 25.5 (95% CI 25.0–26.1) for the clinical estimate (P value for difference in ratio ratios=.68). Similarly, the rate ratio for perinatal death at preterm relative to term gestation was 27.8 (95% CI 26.3–29.3) in Canada, similar to that in the United States when based on the clinical estimate (rate ratio 26.5, 95% CI 26.1–26.9, P value for difference in rate ratios=.06) but not when based on menstrual dates (rate ratio 18.9, 95% CI 18.7–19.2, P value for difference in rate ratios<.001).
Stillbirth, neonatal mortality, and perinatal mortality rate ratios at mild (34–36 weeks) and moderate (32–33 weeks) preterm gestation were not significantly different between Canada and the United States when gestational age was based on the clinical estimate but significantly lower in the United States when based on menstrual dates (Table 4). Thus, the perinatal mortality rate ratio was 25.8 (95% CI 23.1–28.7) at 32–33 weeks compared with that at term in Canada and 24.8 (95% CI 24.2–25.5) in the United States when gestational age was based on the clinical estimate (P value for difference in rate ratios=.44). The rate ratio for perinatal mortality at 32–33 weeks compared with that at term gestation in the United States, with gestation based on menstrual dates, was 16.2 (95% CI 15.8–16.6), which was significantly different from that observed in Canada (P value for difference in rate ratios<.001). At very preterm gestation, however, this pattern did not hold; rate ratios for perinatal death at less than 28 weeks and at 28–31 weeks (compared with term gestation) were substantially and significantly higher in Canada compared with the United States, irrespective of how gestational age was estimated (Table 4). Thus, rate ratios for perinatal death at 28–31 weeks of gestation were 56.0 (95% CI 51.5–61.0) in Canada, 47.9 (95% CI 46.9–48.9) in the United States when based on the clinical estimate of gestation, and 34.3 (95% CI 33.6–35.0) in the United States when based on menstrual dates (P value for difference between the Canadian and the two U.S. rate ratios<.001 and <.001). This was true of stillbirth rate ratios and neonatal mortality rate ratios at very preterm gestation as well. Nevertheless, stillbirth, neonatal mortality, and perinatal mortality rate ratios at very preterm gestation were significantly higher in the United States (and closer to the Canadian estimates) when preterm birth was based on clinical estimates rather than on menstrual dates (Table 4).
Figure 3 shows incidence rates of birth and perinatal death in Canada and the United States based on the fetuses-at-risk approach. Birth rate patterns were more alike between Canada and the United States when gestational age for the latter was based on the clinical estimate of gestation. At the extremes of gestation, U.S. mortality rates based on the clinical estimate of gestation were closer to the Canadian rates.
Births from California constituted a majority of the live births excluded from the U.S. study population due to missing information on the clinical estimate of gestation (95.7% of live births excluded in 1995 and 98.6% of live births excluded in 2002 were from California). Menstrual dates showed that the rate of preterm birth among excluded live births was lower than in the remaining study population (10.8% and 10.3% in 1995 and 2002, respectively, among those with missing information on the clinical estimate of gestation compared with 11.5% and 12.3%, respectively, in the remaining population). The rate of postterm birth (based on menstrual dates) among those with missing information on the clinical estimate of gestation was 8.7% in 1995 and 7.5% in 2002, compared with 8.6% in 1995 and 6.6% in 2002 in the remaining population.
Our study shows that preterm and postterm birth rates in the United States are closer in magnitude to those in Canada (and other industrialized countries) than routine reports suggest. To a large extent, the observed variation in rates of preterm birth is a consequence of differences in the method of gestational age estimation. Preterm and postterm birth rates in Canada in 2002 were 7.6% and 1.0%, respectively, whereas those in the United States in 2002 were 10.1% and 1.0%, respectively, when the clinical estimate of gestation was used to define gestational age. Non-Hispanic whites had preterm and postterm birth rates of 9.5% and 1.0%, respectively, and non-Hispanic blacks had rates of 14.2 and 1.0%, respectively, based on the clinical estimate. The U.S. rates excluded a substantial fraction of live births with missing information on the clinical estimate of gestation (primarily those from California). Because California has a relatively lower rate of preterm birth based on menstrual dates, we speculate that the inclusion of these live births would have further lowered the rate of preterm birth in the United States
Menstrual dates are prone to error for several reasons, including irregularity of menstrual cycles, delayed ovulation, bleeding in early pregnancy, unrecognized miscarriage, and poor recollection or related errors.20,21 These problems, along with data collection and transcription errors, inflate the rates of preterm and postterm birth in large perinatal datasets (eg, the postterm birth rate in the United States in 2003 was 6.4% based on menstrual dates3 compared with 1.5% in Australia5 and 0.8% in Canada,4 where gestational age is based on the clinical estimate).
Misclassification of term births as preterm births is also responsible for the well-known bimodality observed in the birth weight distribution at preterm gestation (Ananth CV. Menstrual versus clinical estimate of gestational age: temporal trends and variability in indices of perinatal outcomes in the United States, in press),22,23 besides contributing to the inflated preterm birth rate. The clinical estimate of gestation, on the other hand, is typically based on information from early ultrasonography or examination of the newborn infant, as well as menstrual dates. It thus represents the clinician's best estimate of gestational age and reduces the potential for misclassification. The bimodality of birth weight is not evident at preterm gestation when gestational age is based on the clinical estimate (Ananth, in press).23 The most serious drawback of the clinical estimate of gestation appears to be the absence of specification and standardization with regard to the mode of ascertainment (ie, whether from menstrual dates or early ultrasonography or neonatal examination). Also, the accuracy of the clinical estimate of gestation is limited by the availability of an early ultrasound measurement and various measurement errors.
One feature of gestational age estimation using menstrual dates is worthy of note. Errors in menstrual dates in large perinatal databases lead to an artifactually higher rate of preterm birth relative to preterm birth rates based on a clinical estimate of gestation. This is in contradistinction to menstrual estimates of preterm birth rates in hospital-based studies, which are lower than preterm birth rates based on early ultrasound estimates of gestational age.24,25 Menstrual dates in hospital-based studies contain errors due to delayed ovulation (more than 14 days after the last normal menstrual period), which is more frequent than early ovulation.26,27 Menstrual dates in vital statistics data sets, on the other hand, appear to be more seriously compromised because of additional errors in recall, data collection, and transcription, which misclassify a larger fraction of term births as preterm births.
Relative mortality rates also confirm that the clinical estimate of gestation is preferable for international comparisons of preterm birth and other gestational age–based outcomes. The rate ratio for perinatal mortality at preterm compared with term gestation is comparable between the United States and Canada when preterm birth in the United States is based on the clinical estimate of gestation but not when based on menstrual dates. The lower mortality among preterm births (relative to term births) in the United States when preterm birth is based on menstrual dates suggests that menstrual dating is responsible for the misclassification of some term births as preterm births. A somewhat different mortality pattern was observed at very preterm gestation (Table 4), which could indicate differences in the clinical estimates in Canada compared with the United States or true differences in perinatal mortality at early gestation, perhaps due to more aggressive clinical management in the United States
We had expected that relative mortality comparisons between Canada and the United States would be more informative when based on the fetuses-at-risk approach, given the causal nature of this latter model.18,19 In hindsight, it is apparent that this expectation was misplaced; the large denominator that characterizes this calculation at preterm gestation (ie, all fetuses at risk at that gestation) is insensitive to the relatively small amount of misclassification that occurs. On the other hand, the traditional calculation of gestational age–specific perinatal mortality, which uses births at a particular gestation as the denominator, is very sensitive to misclassified births since there are relatively few births occurring at preterm gestation.
Perinatal reports from Canada and other industrialized countries have embraced changes in the method of gestational age estimation. This initially led to demonstrated changes in the gestational age distribution (less even digit preference for weeks of gestation, eg, 37 weeks versus 36 weeks) and a decline in rates of preterm birth.28 The United States, on the other hand, has continued to report preterm birth rates using menstrual dates (while simultaneously collecting information on the clinical estimate of gestation). Aside from international comparisons, the continued use of menstrual-based gestational age in surveillance and research reports may have important implications for the United States even within its own borders. For instance, recent declines in preterm birth rates among non-Hispanic blacks in North Carolina (based on menstrual dates) have been interpreted as an artifact of improvements in data quality.29 However, the decline in preterm birth among non-Hispanic blacks in the United States based on the clinical estimate of gestation (14.8% in 2000 and 14.2% in 2002, Fig. 1) and other evidence related to trends in preterm birth subtypes (Joseph KS, Ananth CV. Re: “Preterm delivery rates in North Carolina: are they really declining among non-Hispanic African Americans?” [letter]. Am J Epidemiol 2005;161:1181)30 suggests that the decline may not be an artifact. The National Center for Health Statistics, cognizant of the advantages and shortcomings of the clinical and menstrual-based estimates of gestation, has recently proposed a new algorithm31 that further edits menstrual-based estimates of gestational age using the clinical estimate of gestation. Whereas the merits of alternative estimates of gestational age deserve scrutiny and debate, the purpose of international comparisons appears best served by the clinical estimate of gestation, because that is the only estimate collected in many countries (including Canada). Nevertheless, the absence of a clinical estimate of gestation for births from California means that national temporal trends of preterm birth, postterm birth, and related indices in the United States have to be based on menstrual estimates of gestational age.
Although differences between menstrual-based and clinical estimates of gestation are well recognized, the discrepancy between the high rates of preterm and postterm birth in the United States and the more modest rates in other industrialized countries, such as Canada, has never before been explained on the basis of differences in the method of gestational age ascertainment. Any change in the method of gestational age estimation in the United States would also require a reevaluation of the Healthy People 2010 goal for the preterm birth rate (currently set at 7.6%32), given the artificial nature of the reduction in preterm birth that would follow a change in definition of gestational age. Another issue of relevance to clinicians is the issue of fetal growth standards. Whereas clinicians typically work with the clinical estimate of gestation, the U.S. national fetal growth standard33 was created using the menstrual-based estimate of gestational age. If the U.S. fetal growth standard is recreated based on the clinical estimate of gestation, this will impact the identification and consequent management of small-for gestational-age and large-for-gestational-age fetuses and infants.
In summary, our study shows substantial differences in rates of preterm and postterm birth in Canada and the United States when gestational age in the United States is based on menstrual dates. The differences in preterm birth rates are reduced and those in postterm birth are eliminated when U.S. rates are based on the clinical estimate of gestation. More importantly, the rate ratio for perinatal death at preterm gestation compared with term gestation in the United States is very similar to that in Canada when gestational age is based on the clinical estimate. Our findings suggest that U.S. data on gestational age and its derivative indicators should be based on the clinical estimate, especially for all international comparisons.
1. Morrison JC. Preterm birth: a puzzle worth solving. Obstet Gynecol 1990;76 suppl:5S–12S.
2. Creasy RK, Merkatz IR. Prevention of preterm birth: clinical opinion. Obstet Gynecol 1990;76 suppl:2S–4S.
3. Martin JA, Hamilton BE, Sutton PD, Ventura SJ, Menacker F, Munson ML. Births: final data for 2003. Natl Vital Stat Rep 2005;54:1–116.
5. Laws PJ, Sullivan EA. Australia's mothers and babies 2003. Sydney, Australia: Australian Institute of Health and Welfare's (AIHW) National Perinatal Statistics Unit; 2005.
6. Buitendijk S, Zeitlin J, Cuttini M, Langhoff-Roos J, Bottu J. Indicators of fetal and infant health outcomes. Eur J Obstet Gynecol Reprod Biol 2003;111 suppl:S66–77.
7. Craig ED, Mantell CD, Ekeroma AJ, Stewart AW, Mitchell EA. Ethnicity and birth outcome: New Zealand trends 1980–2001. Part 1. Introduction, methods, results and overview. Aust N Z J Obstet Gynaecol 2004;44:530–6.
8. Kramer MS, Demissie K, Yang H, Platt RW, Sauve R, Liston R. The contribution of mild and moderate preterm birth to infant mortality. JAMA 2000;284:843–9.
9. McDorman MF, Atkinson JO. Infant mortality statistics from the linked birth/infant death data set—1995 period data. Mon Vital Stat Rep 1998;46 suppl;1–22.
10. Martin JA, Hamilton BE, Sutton PD, Ventura SJ, Menacker F, Munson ML. Births: final data for 2002. Natl Vital Stat Rep 2003;52:1–113.
11. Fair M, Cyr M. The Canadian Birth Data Base: a new research tool to study reproductive outcomes. Health Rep 1993;5:281–90.
12. Joseph KS, Kramer MS. Recent trends in infant mortality rates and proportions of low-birth-weight live births in Canada. CMAJ 1997;157:535–41.
14. Sepkowitz S. International rankings of infant mortality and the United States' vital statistics natality data collecting system: failure and success. Int J Epidemiol 1995;24:583–8.
15. Breslow NE, Day NE. Statistical methods in cancer research. Volume I –The analysis of case- control studies. IARC Sci Publ 1980;32:5–338.
16. Yudkin PL, Wood L, Redman CWG. Risk of unexplained stillbirth at different gestational ages. Lancet 1987;1:1192–4.
17. Kramer MS, Liu S, Luo Z, Yuan H, Platt RW, Joseph KS. Analysis of perinatal mortality and its components: time for a change? Am J Epidemiol 2002;156:493–7.
18. Joseph KS. Incidence-based measures of birth, growth restriction and death can free perinatal epidemiology from erroneous concepts of risk. J Clin Epidemiol 2004;57:889–97.
19. Joseph KS. Theory of obstetrics: the fetuses-at risk approach as a causal paradigm. J Obstet Gynaecol Can 2004;26:953–6.
20. Treloar AE, Behn BG, Cowan DW. Analysis of gestational interval. Am J Obstet Gynecol 1967;99:34–45.
21. Saito M, Yazawa K, Hashiguchi A, Kumasaka T, Nishi N, Kato K. Time of ovulation and prolonged pregnancy. Am J Obstet Gynecol 1972;112:31–8.
22. Battaglia F, Frazier T, Hellegers A. Birth weight, gestational age, and pregnancy outcome, with special reference to high birth weight–low gestational age infant. Pediatrics 1966;37:417–22.
23. Mustafa G, David RJ. Comparative accuracy of clinical estimate versus menstrual gestational age in computerized birth certificates. Public Health Rep 2001;116:15–21.
24. Kramer MS, McLean FH, Boyd ME, Usher RH. The validity of gestational age estimation by menstrual dating in term, preterm, and postterm gestations. JAMA 1988;260:3306–8.
25. Goldenberg RL, Davis RO, Cutter GR, Hoffman HJ, Brumfield CG, Foster JM. Prematurity, postdates, and growth retardation: the influence of use of ultrasonography on reported gestational age. Am J Obstet Gynecol 1989;160:462–70.
26. 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.
27. Savitz DA, Terry JWN Jr, Thorp JM Jr, Siega-Riz AM, Herring AH. Comparison of pregnancy dating by last menstrual period, ultrasound scanning, and their combination. Am J Obstet Gynecol 2002;187:1660–6.
28. Joseph KS, Kramer MS. Recent versus historical trends in preterm birth in Canada. CMAJ 1999;161:1409.
29. Vahratian A, Buekens P, Bennett TA, Meyer RE, Kogan MD, Yu SM. Preterm delivery rates in North Carolina: are they really declining among non-Hispanic African Americans? Am J Epidemiol 2004;159:59–63.
30. 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.
31. Qin C, Dietz P, England L, Martin J, Flowers L. Trends in preterm delivery by race: the effects of data editing. Workshop on measurement of gestational age: challenges to conducting research and surveillance. Washington, DC: U.S. Department of Health and Human Services. Centers for Disease Control and Prevention; 2005.
32. U.S. Department of Health and Human Services. Healthy People 2010. 2nd ed. With Understanding and Improving Health and Objectives for Improving Health. 2 vols. Washington, DC: U.S. Government Printing Office; 2000.
© 2007 by The American College of Obstetricians and Gynecologists.
33. Alexander GR, Himes JH, Kaufman RB, Mor J, Kogan M. A United States national reference for fetal growth. Obstet Gynecol 1996;87:163–8.