From the *Department of Preventive Medicine and Public Health, Creighton University School of Medicine, Omaha, Nebraska; and †Epidemiology Branch, National Institutes of Environmental Health Sciences, National Institutes of Health, Department of Health and Human Services, Research Triangle Park, North Carolina.
Submitted 2 March 2007; accepted: 25 May 2007; posted 20 July 2007.
Supported by the Intramural Research Program of the NIH, National Institutes of Environmental Health Sciences.
Correspondence: Aimin Chen, Department of Preventive Medicine and Public Health, Creighton University School of Medicine, 2500 California Plaza, Omaha, NE 68178. E-mail: firstname.lastname@example.org.
Background: A recent report described an association between low maximum diastolic blood pressure (DBP) during pregnancy and perinatal death (stillbirth and death in the first week combined). The authors did not account for gestational length, a strong predictor of perinatal death.
Methods: We studied 41,089 singleton pregnancies from the U.S. Collaborative Perinatal Project (1959–1966).
Results: We observed an association between low maximum DBP and elevated risk of perinatal death. However, this association disappeared after accounting for reverse causation related to gestational length. At any given gestational week, women whose offspring ultimately experienced perinatal death did not have significantly lower maximum DBP than women whose offspring survived the perinatal period. When accounting for the trend of increasing DBP during late pregnancy through gestational-age-specific DBP standardized score, we saw no association between low diastolic blood pressure and perinatal death.
Conclusions: Low maximum maternal DBP during pregnancy is a post hoc correlate of perinatal death, not a true risk factor.
A recent U.K. study suggested that babies whose mothers had low diastolic blood pressure (DBP) during pregnancy had an increased risk of perinatal death.1 A similar finding had been reported in an earlier study among a segment of women enrolled in the U.S. Collaborative Perinatal Project.2 Both studies used as a marker the highest recorded value of DBP during pregnancy. However, because blood pressure normally decreases in the early second trimester and then increases in late pregnancy,3 the maximum recorded DBP will depend in part on gestational length, a strong predictor of perinatal death. Furthermore, the maximum value within any given set of measurements becomes higher as the number of measurements increases. This may result in a tendency for shorter pregnancies to have a lower maximum blood pressure.
To assess the association between low DBP and perinatal death, we reanalyzed the full data from the Collaborative Perinatal Project, with particular attention to the problems involved in using as a predictor the maximum DBP over the whole pregnancy.
The Collaborative Perinatal Project was a prospective pregnancy cohort of 58,760 pregnancies from 48,197 women recruited in 12 academic medical centers in the U.S. between 1959 and 1966.4 Women were enrolled during pregnancy (median gestation at enrollment: 21 weeks; interquartile range: 15–28 weeks).
We excluded 623 multifetal pregnancies and 3432 pregnancies lost before 20 weeks. For 9759 woman contributing more than one singleton birth, we retained only the earliest. We further excluded 302 pregnancies with gestational length (based on the date of last menstrual period) that was missing or recorded as being below 20 or above 51 weeks (allowing for some degree of error in dating). We additionally excluded 1623 pregnancies with no valid maternal DBP measurement and 22 women with no DBP measurement higher than 40 mm Hg or lower than 110 mm Hg. Among women with at least one measurement within the above range, we excluded any value <40 or >110 mm Hg. Although we included women with hypertension during pregnancy in most analyses, we excluded those with reported history of hypertension before pregnancy and those who developed proteinuria during pregnancy (n = 1910). These restrictions left 41,089 singleton births for analysis.
Blood pressure was measured by sphygmomanometer at registration, at each subsequent prenatal visit, and at the time of labor. For diastolic blood pressure, either Korotkoff phase 4 (muffling of the sound) or phase 5 (disappearance of the sound) was used.5 If a woman had more than one blood pressure measurement during a specific gestational week, we used only the earliest valid measurement within that gestational week. Perinatal death was defined as the combination of fetal death after 20 weeks of gestation and death within the first 7 days after birth.
We started by using the highest DBP value during pregnancy as a predictor of perinatal death, as in previous studies.1,2 We then checked whether maximum DBP depended on the length of gestation and the number of measurements. By comparing the maximum DBP up to a given week between women who ultimately experienced perinatal death and women who did not, we assessed whether a low maximum DBP predicted perinatal death. Finally, to account for the trend of increasing DBP in the third trimester of pregnancy, we calculated a gestational-age-specific DBP z-score, and used percentiles of mean z-score to predict perinatal death. In the logistic regression models, we adjusted for race, maternal age, parity, socioeconomic status, smoking during pregnancy, and maternal prepregnancy body mass index (BMI) (covariates categorized as in Table 1). We additionally adjusted for study center (Boston, Buffalo, New Orleans, New York Columbia, Baltimore, Richmond, Minneapolis, New York Metropolitan, Portland, Philadelphia, Providence, and Memphis). We used R 2.3.16 for graphs and splines, and SAS 9.17 for statistical analysis.
Among the 41,089 babies in the analysis sample, 671 were stillborn (1.6%) and 563 died within the first 7 days of life (1.4%), yielding a total of 1234 perinatal deaths (3.0%). Black race, older age, higher parity, lower socioeconomic status, smoking, and obesity were all associated with higher perinatal mortality, as was a shorter gestation (Table 1).
The mean ± SD gestational age at delivery was 39 ± 3 weeks. The mean number of blood pressure measurements per woman was 7 ± 4 (range, 1–29). Women whose baby died perinatally had shorter gestation (32 ± 7 weeks; range, 20–47), and fewer blood pressure measurements (5 ± 4; range, 1–26).
As expected, DBP decreased in the second trimester and increased during the third trimester (not shown). The mean maximum DBP increased with gestational length and also with increasing number of blood pressure measurements (from 75 mm Hg with 2 measurements to 83.5 mm Hg with 13 or more measurements). The effect of gestational length and number of measurements on mean DBP was, on the other hand, limited.
Similar to the findings of Steer et al,1 the relationship between maximum DBP and perinatal mortality was U-shaped, with highest mortality at the lowest maximum DBP (Fig. 1).
Women with lower maximum DBP had, on average, a shorter gestation and fewer BP measurements. Both of these factors could contribute to biasing any inference based on Figure 1, as suggested by the fact that, at any gestational age, a low maximum DBP up to that point was not predictive of perinatal death (Fig. 2).
When using gestational-age-specific mean DBP z-score percentiles, we saw little evidence of elevated risk of perinatal death among women with low z scores (Fig. 3). To obtain more precise estimates, we defined the lowest and highest 10% of z-score as “low” and “high” DBP. The perinatal death rate was 2.4% for the lowest and 5.5% for the highest decile. The corresponding adjusted ORs were 0.88 (95% confidence interval = 0.67−1.16) and 2.08 (1.68–2.59), respectively, compared with a DBP z-score between the 40th and 59th percentiles.
The results persisted when including DBP measurements with values <40 or >110 mm Hg, using mean DBP, restricting to women registered early in pregnancy or with DBP ≤90 mm Hg, or using different categorizations of mean DBP z-score (data not shown).
In this analysis, low maximum maternal DBP during pregnancy appeared to be a post hoc correlate of perinatal death. At no point in pregnancy was a low maximum DBP predictive of perinatal death, suggesting that such a marker would not be a useful clinical predictor of risk. The association between low maximum DBP across the whole pregnancy and perinatal death likely reflects bias due to differential-by-outcome mean lengths of gestation and the corresponding differential mean number of BP measurements. Thus, our analysis suggests that the previous findings1,2 likely resulted from an artifact due to the failure to account for reverse causation related to gestational length.
Using mean gestational-age-specific DBP z-score should produce a marker of DBP that eliminates the artifactual correlation between maximum DBP and gestational length. Our analysis based on the z-score showed no important association between lower DBP during pregnancy and perinatal mortality, although the lowest 1% of z-score came close to statistical significance. Whether this may have clinical relevance in a subgroup of women remains to be determined. The higher mortality with higher maternal blood pressure (Fig. 3) is consistent with the well-known risks of hypertensive disorders in pregnancy.8
Previous studies have suggested that maternal hypotension during pregnancy (often defined as BP ≤110/65 mm Hg) may be associated with reduced uteroplacental perfusion, prematurity, and low birth weight.9–13 However, these studies often included few women or lacked adjustment for potential confounders.14 Zhang and Klebanoff15 argued that the adverse outcomes in women with low DBP were due to their characteristics, such as low prepregnancy BMI and low social status, rather than to low DBP per se.
This was a large prospective cohort with multiple blood pressure measurements. However, our study has limitations. Recruitment occurred between 1959 and 1966, and perinatal death has substantially decreased since. Nonetheless, in replicating the analysis of Steer et al1 on this large cohort, we found an association when using maximum DBP throughout pregnancy, which disappeared after taking into consideration the fact that shorter gestations tended to have a low maximum DBP.
Overall, our analysis suggests that a low maternal blood pressure during pregnancy is not predictive of increased risk of perinatal death at the population level, and that maximum DBP throughout pregnancy is subject to bias as a predictor of poor perinatal outcome.
We thank Clarice R. Weinberg for invaluable advice on data analysis and writing, and Dale P. Sandler and Matthew P. Longnecker for comments on an earlier draft of the manuscript.
1. Steer PJ, Little MP, Kold-Jensen T, et al. Maternal blood pressure in pregnancy, birth weight, and perinatal mortality in first births: prospective study. BMJ
2. Friedman EA, Neff RK. Hypertension-hypotension in pregnancy. Correlation with fetal outcome. JAMA
3. Hermida RC, Ayala DE, Iglesias M. Predictable blood pressure variability in healthy and complicated pregnancies. Hypertension
4. Niswander KR, Gordon M. The Women and Their Pregnancies: the Collaborative Perinatal Study of the National Institute of Neurological Diseases and Stroke
. Washington, DC: U.S. Goverment Print Office; 1972.
5. Friedman EA, Neff RK. Pregnancy Hypertension: A Systematic Evaluation Of Clinical Diagnostic Criteria
. Littleton, MA: PSG Publishing Co.; 1977.
6. R Development Core Team. R: A Programming Environment for Data Analysis and Graphics, 2006. Available at: www.r-project.org
. Accessed June 15, 2007.
7. SAS Institute I. SAS/STAT 9.1 User's Guide
. Cary, NC: SAS Publishing; 2004.
8. ACOG Technical Bulletin. Hypertension in pregnancy. Number 219–January 1996 (replaces no. 91, February 1986). Committee on Technical Bulletins of the American College of Obstetricians and Gynecologists. Int J Gynaecol Obstet
9. Kunzel W. [Hypotensive disorders during pregnancy as a cause of fetal hazard (author's transl)]. Z Geburtshilfe Perinatol
10. Grunberger W, Parschalk O, Fischl F. [Treatment of hypotension complicating pregnancy improves fetal outcome]. Med Klin
11. Grunberger W, Leodolter S, Parschalk O. [Pregnancy hypotension and fetal outcome]. Fortschr Med
12. Grunberger W, Leodolter S, Parschalk O. Maternal hypotension: fetal outcome in treated and untreated cases. Gynecol Obstet Invest
13. Grunberger W, Leodolter S, Parschalk O. [Pregnancy-connected hypotension - the influence of mineralo-corticoids on fetal nutrition (author's transl)]. Geburtshilfe Frauenheilkd
14. Ng PH, Walters WA. The effects of chronic maternal hypotension during pregnancy. Aust N Z J Obstet Gynaecol
15. Zhang J, Klebanoff MA. Low blood pressure during pregnancy and poor perinatal outcomes: an obstetric paradox. Am J Epidemiol
© 2007 Lippincott Williams & Wilkins, Inc.