1360 Letters
November 1993 Am
J Obstet Gynecol
Table I. Adverse pregnancy outcome by hCG and AFP level Outcome
Prematurity Fetal defects (nonchromosomal) Pregnancy complications Spontaneous abortions and fetal death All adverse outcomes rorxr.
AFPand hCG elevated
Relative risk hCG elevated
Significance
0 4
10
17 7 15 4
4.54 3.28* 53.78* 1.52*
P < 0.0001 P = 0.0230 P < 0.0001 P = 0.5062
36 874
43 268
6 9
4.14
P < 0.0001
hCG normal
hCG elevated
17 8 1
I I
*r.;ot weighted. too few cases for stratification.
2. Dcutchman M. Advances in the diagnosis of first-trimester pregnancy problems. Am Fam Physician 1991;11(suppl 5):15-305. 3. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. .I Natl Cancer Inst 1959;28:719-47. 1. Goren R. Perez R, David '.1, Dar H. The association hetween unexplained levels of second-trimester maternal serum hCG elevation and pregnancy complications. Obstet Gynecol 1992;80:83-6. 5. Beekhuis JR, Van Lith '.1M, De Wolf BTHM, Mantingh A. Increased maternal serum a-fetoprotein and human chorionic gonadotropin in compromised pregnancies other than for neural tube defects and Down syndrome. Prenat Diagn 1992;12:643-7.
Does tertiary center perinatal care lower mortality In Infants < 1500 gm? More data needed To the Editors: Advances in neonatology and regionalization of perinatal care have been credited by Copper et al. (Copper RI., Goldenberg RI., Creasy RK, et al. A multicenter study of preterm birth weight and gestational age-specific neonatal mortality. AM J OBS'I ET GV:-;ECOI. 1993; 168:78-84). for lowering neonatal mortality rates in tertiary care centers. Neonatal mortality rates for infants at < 29 weeks' gestation, compared with the rates during the 1960s and 1970s from the University of Colorado, led the overall decline. I. " In addition, black premature infants were found, as many others have found, to have lower neonatal mortality rates than white premature infants. Then the authors ignored their own finding. They failed to connect a decrease in mortality with an increase in the number of black premature infants. Their population included 46% black births compared with 7% in the more recent Colorado study." The racial difference would be much greater for infants < 29 weeks' gestation, because births weights < 1500 gm occur three times more frequently among blacks than among whites. AJthough the numbers of births in various birth weight categories and weight stratifications by weight were not included in the study. one can estimate that approximately 70% of the newborns < 1500 gm were black.' More recent neonatal mortality rates from the National Institute of Child Health and Human Development Neonatal Network, another tertiary care
study that supplied birth weights, reported 62% black infants of this weight in their study.' When birth weight-specific and race-weight data for the tiniest births are left out of a study, selection bias and the extent of the bias can escape scrutiny or only be surmised. Lack of this information can help perpetuate the "high-risk fallacy," the notion that tertiary care centers invariably receive for treatment a majority of the high-risk mothers and newborns in a given population just because the centers were set up to do so.' Birth weights of extremely small newborns at the edge of viability are indispensable in studies of neonatal mortality, such as this one, for another reason. Gestational age estimates, when available (how many were not available in this study), can be ajourney into science fiction. Obstetric estimates derived from menstrual history, the best estimate available,'; become progressively worse as they deviate from term births." Biologic variation is probably the source of the error. Neonatal gestational estimates that might assess the accuracy of the obstetric estimates are also worse with premature infants." Finally, racial differences in neonatal gestational-age estimates have yet to be worked out." Black mothers may have shorter periods of gestation. \0
Samuel Sepkountz, MD Department of Pediatrics, University of Oklahoma Health Sciences Center. 5300 N. Meridian, Oklahoma City. OK 73112
REFERENCES I. Lubchenco LO, Searles DT, Brazie .IV. Neonatal mortality rate: relationship to birth weight and gestational age. .I Pediatr 1972;84:814-22. 2. Koops BL, Morgan L], Battaglia Fe. Neonatal mortality risk in relation to birth weight and gestational age: update. .I Pediatr 1982;10 1:969-77. 3. National Center for Health Statistics. Vital statistics of the United States. Volume I: natality. Hyattsville, Maryland: Public Health Service, 1982-1987. 4. Hack '.1, Hobar JD, Malloy MH, Tyson .IE, Wright E, Wright L. Very low birth weight outcomes of the National Institute of Child Health and Human Development Nconatal Network. Pediatrics 1991;87:587-92. 5. Lumley J. Does this work? Pediatrics 1987;79: I040-4. 6. Fetal extrauterine survivability, Report of the Committee on Fetal Extrauterine Survivability to The New York State Task Force on Life and the Law. Albany, New York: State of New York. 1988.
Volume 169, Number 5 Am J Obstet Gynecol
7. Kramer MS, McLean FH, Boyd ME, Usher RH. The validity of gestational age estimation by menstrual dating in term, preterm, and postterm gestation. JAMA 1988; 260:3306-8. 8. Yogman MW, Kraemer He, Kindlon D, TysonJE, Casey P, Gross RT. Identification of intrauterine growth retardation among low birth weight preterm infants. J Pediatr 1989; 115:799-807. 9. Spinnato JA, Sibai BM, Shaver DC, Anderson GD. Inaccuracy of Dubowitz gestational age in low birth weight infants. Obstet Gynecol 1984;63:491-5. 10. Hardy J B, Mellits ED. Relationship of low birth weight to maternal characteristics of age, parity, education and body size. In: Reed DM, Stanley FN, eds. The epidemiology of prematurity. Baltimore: Urban und Schwarzenberg, 1977:105-17.
Reply To the Editors: The major point of our article was that at
< 29 weeks' gestation, although gestational age was the major factor influencing neonatal mortality, race, plurality, and infant sex also had an effect, whereas at ~ 29 weeks birth weight appeared to have the major impact on neonatal mortality. Data that showed the neonatal mortality rate for all preterm gestational age and birth weight combinations and the survival by gestational age of delivery and by 100 gm increments in birth weight were displayed. We also displayed data that showed that at the median birth weight for gestational age, survival for preterm infants appears to have improved considerably over time. We believe these points were made clearly and were justified by the data presented. In contrast, we are not sure what point or points the letter writer is trying to make. He seems unhappy with our use of gestational age but fails to see the point that before 29 weeks, whatever the problems with gestational age measurement, it is a better predictor of survival than is birth weight. These comments ignore Methods, where we carefully described how the gestational age data were determined. If the writer is so concrete in his thinking that only data collected with a scale are considered acceptable, little will be gained from further discussion here. However, most of the world's researchers would agree that there is a reasonably good relationship between carefully determined gestational age and pregnancy outcome. Our results confirm this observation. The writer also implies bias on our part in the selection of patients, apparently not noticing that all inborn infants born at ~ 20 weeks were included. We did not present birth weight-specific data by race because that was not the purpose of this study. However, we did use race, birth weight, and gestational age in the regression models to determine the odds ratio of mortality associated with each of these characteristics. If this writer does not understand regression analysis but believes that we used this technique to "escape scrutiny," so be it. However, any statistical textbook should explain our choice of this analytic technique. We do not understand what "high-risk falla-
Letters
1361
cy" the letter writer is talking about. We made no claim about the centers involved in our study receiving the "majority of the high-risk mothers and infants in a population." If this is a battle the writer is waging with some real or imaginary foe, we hope he leaves us out of it. By the way, we are aware that our population (and that of the neonatal network) had more black infants than whites and that in nearly all studies being black is a risk factor for a very-early-gestational-age birth. However, the fact that the Denver population in 1961 had 7% blacks and ours had 70% does not imply an increase in the number of black premature infants over time, only a difference between two geographic areas. Even in populations where the racial mix has remained the same over time, it is apparent to all who have studied the matter that advances in neonatal care and regionalization have made a substantial contribution to increases in very early gestational age survival. Robert L. Goldenberg, MD, and Rachel L. Copper, MSN, CRN? Department of Obstetrics and Gynecology, University of Alabama, Uniuersits Station, Birmingham, AL 35294
Neurobiology of fetal breathing To the Editors: Derks et al. (Derks JB, Owiny J, Sadowsky
D, Ding XV, Wentworth R, Nathanielsz Pw, Effects of repeated administration of cocaine to the fetal sheep in the last days of pregnancy. AM.J OBSTET GYNECOL 1993; 168:719-23.) suggest that the induction of breathing activity in fetal sheep during labor may be the result of cocaine's action on the catecholaminergic system. This hypothesis is supported by shortened periods of rapideye-movement sleep during a continuous 90 minutes of infusion of cocaine to the fetus' and by the importance of dopamine in the control of wakefulness manifested by a reduction of reaction time and gap frequency. The fact that delay-dependent speeding of reaction time, reflecting motor readiness, is abolished by depletion of dopamine" supports the suggestion that the different fetal respiratory response to cocaine when the ewe was in labor may also reflect the maturation of the respiratory system in preparation for the extrauterine environment. Ernest H. Friedman, MD 1831 Forest Hills Blvd., East Cleveland, OH 44112-4313
REFERENCES
1. Burchfield DJ, Graham EM, Abrams RM, Gerhardt Kj. Cocaine alters behavioral states in fetal sheep. Dev Brain Res 1990;56:41-5. 2. Friedman EH, Monplaisirj. i-Dopa and narcolepsy [Letter and Reply]. Neurology 1992;42:1640-1.
Reply To the Editors: We agree with Friedman that the fetal
brain undergoes specific maturations in preparation for