- Email: [email protected]
Do growth-retarded premature infants have different rates of perinatal morbidity and mortality than appropriately grown premature infants?
Do Growth-Retarded Premature Infants Have Different Rates of Perinatal Morbidity and Mortality Than Appropriately Grown Premature Infants? JEANNA M. PIPER, MD, ELLY M-J. XENAKIS, BYRON D. ELLIOTT, MD, MICHAEL
A common belief of obstetricians and neonatologists is that growth-retarded fetuses are “stressed” by their intrauterine environment and that this stress accelerates their maturation so that growth-retarded preterm infants have fewer complications of prematurity than their appropriately grown peers.‘,* Based on this concept, pregnancies complicated by growth retardation might benefit from planned preterm delivery to avoid From the Department of Obstetrics and Gynecology, the University of Texas Health Science Center at San Antonio, San Antonio, Texas.
87, NO.
2, FEBRUARY
1996
MCFARLAND,
MD,
D. BERKUS, MD, AND ODED LANCER, MD
Objective: To determine if perinatal morbidity and mortality differ in growth-retarded, small for gestational age @GA), premature infants and appropriate for gestational age (AGAl infants. Methods: All consecutive, singleton, nondiabetic, preterm pregnancies delivered over a B-year period were analyzed. Infants were categorized as SGA (at (M below the tenth percentile) or AGA (11th to the 89th percentiles), then stratified by birth weight and gestational age categories. Perinatal morbidity and mortality were examined. Results: We studied 4183 preterm deliveries, 1012 of them SGA and 3171 of them AGA. Overall, we found significantly higher rates of fetal and neonatal death in the SGA group. Stratification by gestational age revealed significantly higher rates of neonatal death for the SGA group compared with the AGA group in each gestational age category. Overall, comparison also revealed significantly tigher rates of fetal heart rate abnormality in the SGA group but no difference in neonatal sepsis, birth trauma, cesarean delivery, hyaline membrane disease, or congenital anomalies. Conclusion: Growth-retarded premature infants have a significantly higher risk of morbidity and mortality, both before and after delivery, than do appropriately grown infants. (Obstet Gynecol 2996;87:169-74)
VOL.
MD, MELINDA
the possibility of fetal death. Before proceeding to action based on this idea, however, we must critically examine the evidence to support or refute it. Conflicting findings on different aspects of this issue have been reported. Neonatal death rates have been reported to be decreased,3,4 unchanged,5,6 and increased7-‘* m ’ small for gestational age (SGA) infants in comparison with appropriately grown infants. Neonatal morbidity has been reported to be both reduced’ and elevated4,r3; however, fetal death rates have been reported uniformly as increased in SGA pregnancies. 10,11,74 The purpose of this study was to compare growth-retarded premature infants with appropriately grown preterm infants at the same gestational age with respect to perinatal morbidity and mortality.
Materials and Methods All pregnancies delivered during a 15-year period (July 1970 to June 1985) at the teaching hospital of Bexar County in San Antonio, Texas, were analyzed individually on discharge by obstetric faculty physicians for historical, demographic, delivery, and outcome information, then entered into our computerized data base. Maternal characteristics included age, gravidity, parity, race, marital status, preexisting medical problems, and complications of pregnancy. Delivery characteristics included gestationa age at delivery, mode of delivery, trauma, fetal heart rate (FHR) abnormality, and type of anesthesia and/or analgesia. Fetal and neonatal data included death, birth weight, Apgar scores, respiratory problems, neonatal infection, and congenital anomalies. Gestational age at delivery was based on the mother’s last menstrual period confirmed by early examination, fetal heart tone auscultation, ultrasonography, or neo-
0029-7844/96/$15.00 SSDI 0029-7844(95)00400-9
169
natal assessment. Gestational age was revised only if ultrasonographic examination was incompatible with menstrual dating. Obstetric assessment of gestational age was used in all cases, with neonatal assessment of dating used only for confirmation. All infants were examined systematically for anomalies at delivery and during their nursery stay by obstetric and pediatric physicians. All suspected anomalies were confirmed by obstetric or neonatal faculty members following diagnostic evaluation. Neonatal respiratory disease was differentiated based on clinical presentation and radiographic findings. The diagnosis of hyaline membrane disease was based on persistent oxygen requirement and characteristic radiographic appearance. Testing for gestational diabetes was perform& in all women with risk factors (such as obesity, family history, prior macrosomic infant, and prior gestational diabetes). From this data set, all singleton pregnancies resulting in the delivery of an infant at 24-36 weeks gestation were selected. We excluded pregnancies complicated by diabetes (preexisting or gestational) and pregnancies with unknown or conflicting dating criteria (neonatal dating incompatible with obstetric dating). Pregnancies with birth weight at or below the tenth percentile (SGA) were then compared with those with birth weights between the 11th and 89th percentiles (appropriate for gestational age [AGAI). Birth percentiles were based on population-specific growth standards derived from 80,000 consecutive contemporaneous live-born, singleton, nonanomalous infants (20-45 weeks’ gestation) in the local inner city, economically compromised population (82% Mexican-American, 11% white, 7% black). All pregnancies were analyzed for perinatal mortality and congenital anomaly rates. Only live births were used for comparison of neonatal death, FHR abnormality, Apgar scores, mode of delivery, neonatal infection, and birth trauma. Because of the potential correlation between congenital anomalies and fetal growth restriction (FGR), fetal death and perinatal mortality (expressed as number per 1000 births) as well as neonatal death (expressed as number per 1000 live births) were examined before and after exclusion of anomalous fetuses. Hyaline membrane disease rates were compared between SGA and AGA pregnancies. In addition, because of the lengthy period of study required to compile this information (15 years), the population was compared by 5-year blocks for variance within the study interval. Categorical data between SGA and AGA pregnancies were compared using J analysis. Comparison was also made after stratification by gestational age and birth weight using Mantel-Haenszel 2 analysis. Stepwise logistic regression analyses were performed with fetal death, neonatal death, and perinatal death as the de-
170 Piper et al
Growth Retardation and Preterm
pendent variables, and abruptio placentae, FGR, placenta previa, chorioamnionitis, maternal medical complications, maternal hypertension, birth trauma, FHR abnormality, congenital anomalies, 5-minute Apgar score less than 7, respiratory problems, and cesarean delivery as the independent variables. Results are displayed as odds ratio (OR) with 95% confidence interval (CI).
Results More than 75,000 singleton deliveries occurred over the 15-year study period. The population was largely Mexican-American (75%) and of low socioeconomic status. After exclusions for inadequate dating criteria (n = 281, 6.3%), 4183 nondiabetic pregnancies delivering before 37 weeks’ gestation remained for analysis, 1012 of whom were SGA and 3171 AGA. In the SGA pregnancies, 728 (72%) resulted in surviving infants, 170 (17%) resulted in fetal deaths, and 114 (11%) resulted in deaths in the neonatal period. Among the AGA pregnancies, 2633 (83%) of the infants survived, 218 (7%) resulted in fetal death, and 320 (10%) newborns died in the neonatal period. Overall, the SGA pregnancies had a significantly higher perinatal mortality rate than the AGA pregnancies (281 versus 170, OR 1.91 [95% CI 1.61-2.261). Both the fetal death rate (168 versus 69, OR 2.74 [95% CI 2.20-3.411) and neonatal death rate (135 versus 108, OR 1.29 [95% CT 1.02-1.631) were significantly higher for the SGA group than the AGA group. Anomaly rates did not differ significantly between SGA and AGA pregnancies (14 versus 13%). By eliminating malformed cases, correction of fetal, neonatal, and perinatal mortality rates revealed perinatal mortality of 262 versus 159, OR 1.89 (95% CI 1.57-2.27), fetal mortality of 169 versus 75, OR 2.52 (95% CI 1.993.18), and neonatal mortality of 112 versus 91, OR 1.27 (95% CI 0.96-1.67). There were no significant differences in gestational age distribution, proportion of SGA fetuses, or perinatal mortality rates between the three 5-year blocks within the study period. Both SGA and AGA groups included deliveries of 24-36 weeks’ gestation (Table 1). Stratification of fetal deaths by gestational age showed a uniform distribution in the SGA group, whereas for the AGA group, fetal deaths decreased as gestational age increased (Table 1). There was an excess of fetal deaths noted in the AGA group before 27 weeks (92 of 218,42% of total AGA fetal deaths). Stratification of neonatal mortality by gestational age revealed a significant difference between SGA and AGA pregnancies (overall OR 7.07 [95% CI 4.9-10.11) (Figure 1A). Furthermore, examination of the individual categories identified a significantly higher neonatal death
Obstetrics & Gynecology
Table 1. Distribution by Gestational Age Categories Gestational age (wk)
Total pregnancies SGA AGA Fetal deaths SGA AGA Neonatal deaths SGA AGA
24 -26
27-28
29-30
31-32
33-34
35-36
44 (4%) 265 (8%)
44 (4%) 291 (9%)
60 (6%) 362 (11%)
109 (11%) 670 (21%)
205 (20%) 809 (26%)
550 (55%) 774 (25%)
25 (15%) 92 (42%)
22 (13%) 36 (17%)
25 (15%) 31 (14%)
30 (18%) 29 (13%)
31 (18%) 16 (7%)
37 (21%) 14 (7%)
18 (16%) 114 (35%)
18 (16%) 108 (34%)
18 (16%) 51 (16%)
25 (22%) 25 (8%)
21 (18%) 15 (5%)
14 (12%) 7 (2%)
SGA = small for gestational age; AGA = appropriate for gestational age. The data are presented as number of total pregnancies, fetal deaths, or neonatal deaths in each gestational age category, and percentage of the total number of pregnancies or deaths in the SGA or AGA group (percentage out of cases in that line).
rate for SGA pregnancies in every gestational age category (Table 2). However, stratification by birth weight (Figure 1B) revealed no significant difference in neonatal mortality rates between SGA and AGA pregnancies either overall or by individual categories (Table 3). Comparison of perinatal morbidity also revealed significant differences. The diagnosis of FHR abnormality was more common in the SGA pregnancies (10.5 versus 8%, OR 1.34 195% CI 1.03-1.7511 and the need for cesarean approached significance (26 versus 23%, OR 1.19 195% CI 0.99-1.421). Apgar score less than 7 at 5 minutes (21.9 versus 19.5%, OR 1.15 [95% CI 0.95-1.3911, birth trauma (1.7 versus 2.6%‘), and neonatal sepsis (7.8 versus 6.2%) did not differ significantly between SGA and AGA infants. Based on the strong relationship between measures of
A
age-size and respiratory disease, we stratified our SGA and AGA pregnancies by both gestational age and birth weight categories for comparison of the rates of hyaline membrane disease (Figure 2). Comparison by gestational age categories revealed a significantly higher rate of hyaline membrane disease in the SGA infants (OR 1.70 [95% CI 1.29-2.231) compared with their AGA peers (Figure 2A). In contrast, comparison by birth weight categories revealed a significantly lower rate of hyaline membrane disease in the SGA pregnancies (OR 0.75 [95% CI 0.58-0.951) (Figure ZR). To evaluate the true net effect of each of the independent variables on outcome, logistic regression analysis was performed individually for fetal death, neonatal death, and perinatal mortality (Table 41. The independent variables analyzed for fetal death were presence or absence of growth retardation, anomalies, chorioamnio-
T--__-
B /’1;
1,000
ISGP IAGP
Figure 1. Neonatal mortality (NNM) rates in small for gestational age (SGA) and appropriate for gestational age (AGA) pregnancies compared by gestational age (A) and birth weight (B). Neonatal death rates in SGA pregnancies were significantly higher by gestational age (P < .05) but not by birth weight.
NBOO M po: 1000 200
2426 27.20 20.30 3142 33.34 35.36
G’ESTATIONAL AGE VOL. 87, NO. 2, FEBRUARY
1996
BIRTHWEIGHT (g) Piper et al
Grozvth Retardation and
Preterm
171
Table 2. Neonatal
Mortality
Rates by Gestational
Age
rate/1000 live births
Gestational age (wk)
SGA
AGA
OR (95% CI)
24-26 27-28 2930 3132 3s34 35-36
947 818 514 316 121 27
659 424 154 39 19 9
9.32 W-51.1) 6.13 (1.9-22) 5.81 (2.7-12.8) 11.40(5.9-22.2) 7.10 (3.4-14.9) 3.02 (1.1-8.3)
SGA = small for gestational age; AGA = appropriate for gestational age; OR = odds ratio; CI = confidence interval.
nitis, abruptio placentae, placenta previa, maternal medical complications, and maternal hypertension. For both neonatal death and perinatal mortality, additional variables entered into the logistic regression model were 5-minute Apgar score less than 7, cesarean delivery, birth trauma, FHR abnormality, and respiratory problems. Only chorioamnionitis proved to be a significant contributor to all three dependent variables; however, several other factors were contributora for one or more logistic regression analyses (Table 4). Growth retardation was a significant risk factor for fetal death and approached statistical significance for neonatal death (P = .06); thus, it was an overall contributor to perinatal mortality (OR 2.17 195% CI 1.8-2.61). Chorioamnionitis also contributed to both fetal and neonatal mortality. Placental complications (abruption and placenta previa) increased risk of fetal death but did not alter neonatal mortality. Low Apgar scores, anomalies, and neonatal respiratory disease were significantly associated with neonatal death. Cesarean delivery was associated with a protective effect against neonatal and perinatal death.
Discussion The key findings of our study to be discussed individually are as follows: 1) Growth-retarded preterm infants Table 3. Neonatal
Mortality
Rates by Birth Weight
Categories SGA
AGA
Birth weight (g)
n
NMR
n
NMR
P
-aQo 500-749 750-999 1000-1249 1250-1499 1500-1999 ZOO0
15 39 40 91 78 327 252
933 821 475 253 167 37 4
3 78 168 198 200 755 1553
795 607 364 150 50 10
NS NS NS NS NS NS
S-GA = small for gestational age; AGA = appropriate for gestational age; NMR = neonatal mortality rate/lWiJ live births; NS = not significant.
172 Piper et al
Growth Retardation and Preterm
have no survival advantage over appropriately grown infants of the same gestational age; 2) growth retardation is an independent predictor of perinatal mortality; 3) the rates of hyaline membrane disease are significantly higher in SGA infants than in appropriately grown infants of the same gestational age; and 4) the rates of hyaline membrane disease are significantly lower in the SGA infants compared with appropriately grown infants of the same birth weight. Our analysis of more than a thousand growthretarded preterm pregnancies revealed significantly higher fetal death rates, neonatal death rates, and perinatal mortality rates in the growth-retarded pregnancies compared with appropriately grown pregnancies. Stratification by gestational age also displayed significantly higher neonatal death rates for the SGA infants. This finding is compatible with prior reports of decreased survival in preterm growth-retarded infants However, comparison of our by gestational age.7,9~‘o~12 data by birth weight categories showed no difference in neonatal mortality between SGA and AGA infants. This finding is also compatible with previous reports of unchanged or improved outcome in preterm SGA infants compared by birth weight only,3” although ours is the first report of morbidity and mortality compared by both gestational age and birth weight in a single large population of preterm infants. The percentage of SGA infants in our study population may seem high (1012 of 4183, 24%), but it is important to recognize that our study cohort spec&ally excludes diabetic pregnancies and large for gestational age infants, but includes fetal deaths and fetuses with congenital anomalies (specifically excluded from birth percentile curve calculations) because of their obvious impact in the consideration of morbidity and mortality. Accumulation of a population of this size required collection over an extended period of time (15 years, 1970-1985); however, analysis by 5-year blocks revealed no significant change in population characteristics over that time. Growth retardation was an independent predictor of fetal death, neonatal death, and perinatal mortality. Other factors asblociated with fetal death were placental problems and intrauterine infection. Additional factors associated with neonatal mortality were chorioamnionitis, neonatal respiratory disease, S-minute Apgar score less than 7, and congenital anomalies. Cesarean delivery was associated with decreased neonatal mortality, probably because of more intensive therapy in these cases. Of interest was our finding of congenital anomalies in 13-14% of our preterm deliveries, whether SGA or AGA, much higher than the 2-4% rates quoted for United States births overall.” We expected congenital anomalies to be more frequent in SGA pregnancies but this proved not to be the case. Undiagnosed anom-
Obstetrics & Gynecology
Figure 2. Hyaline membrane disease (HMD) rates in small for gestational age (SGA) and appropriate for gestational age (AGA) infants compared by gestational age (A) and birth weight (8). Comparison by gestational age revealed an overall significantly higher rate of hyaline membrane disease in the SGA infants compared with the AGA infants (P < .05). In contrast, when compared by birth weight categories, the rate of hyaline membrane disease was lower in the SGA than in the AGA pregnancies (P < .05). 24-26 27-28
29-30 31-32 33-34 35-35
Gestational
Age (weeks)
alies are always a concern, but every infant in this preterm population was examined specifically for malformations and, if anything, one would anticipate that the SGA infants were scrutinized even more scrupulously than their AGA peers. Thus, we conclude that pregnancies delivered preterm are at high risk of congenital anomalies (13-14%), regardless of fetal growth pattern. In our evaluation of neonatal morbidity, comparison by gestational age showed significantly higher rates of hyaline membrane disease in the SGA infants than in the AGA infants. A commonly held belief is that the stress of growth retardation accelerates pulmonary maturation so that preterm delivery is not as great a risk for a growth-retarded pregnancy.lJ2 Our data clearly do not support such a relationship. Indeed, at any given point Table
4. Factors Associated with and Perinatal Death
Abruptio placentae Growth retardation Placenta previa Chorioamnionitis Congenital anomalies 5-min Apgar <7 Respiratory problems Cesarean delivery
Risk of Fetal, Neonatal,
Fetal death
Neonatal death
Perinatal death
4.45 (3.3-6.1) 2.85 (2.3-3.6) 2.42 (1.6-3.8) 1.81 (1.2-2.7) 0.56 (0.4-0.8)
0.87 (0.5-1.4) 1.31 (1.0-1.07) 0.86 (0.5-1.6) 2.05 (1.3-3.1) 1.96 (1.5-2.6)
2.80 (2.1-3.8) 2.17 (1.8-2.6) 1.72 (1.1-2.6) 2.13 (1.6-2.9) 1.22 (0.96-l .5)
18.5 (14.3-24.0) 1.92 (1.5-2.5)
7.12 (5.8-8.7) 1.54 (1.3-1.9)
0.60 (0.4-0.8)
0.66 (0.5-0.8)
Logistic regression, expressed as odds ratio (95% confidence interval).
VOL.
87, NO.
2, FEBRUARY
1996
600
500.750-l&w1250.1500. >1999 749 ss9 124s 1499 lBS@
Birthweight
(grams)
in pregnancy, the SGA infant is placed at higher risk by preterm delivery than a comparable AGA pregnancy because of higher morbidity and mortality. On initial evaluation, our finding of significantly lower rates of hyaline membrane disease in the SGA infants compared by birth weight would appear to create a conflict. However, this is easily explained by the fact that within each birth weight category, the AGA infants were of much lower gestational age than the SGA infants. Thus, the SGA infants were protected by their higher gestational age, whereas stratification by gestational age removed the protection, allowing the full effect of growth retardation to be seen. This highlights the need for comparison by both birth weight and gestational age to draw adequate conclusions regarding the effects of growth retardation on perinatal morbidity. Although comparison by both birth weight and gestational age are essential for research purposes, based on our data, we feel that gestational age is the better predictor of neonatal outcome. Likewise, comparison of neonatal mortality by both gestational age and birth weight is essential to determine the impact of growth retardation. We found significantly higher mortality in SGA infants than in AGA infants when compared by gestational age but no difference when compared by birth weight. Prior reports compared outcome based on either birth weight or gestational age, but not both. Indeed, the reports of higher neonatal survival rates in SGA infants were based on comparison by birth weight only.3,4 As described above for hyaline membrane disease, it is reasonable that comparison by birth weight alone could
Piper et al
Growth Retardation
and Preterm
173
show increased survival for SGA infants because an 800-g SGA infant of 29 weeks’ gestation would be compared with an 800-g AGA infant of 24-25 weeks gestation. Thus, when compared by birth weight, the benefit of being less premature outweighs the risk from growth retardation, whereas comparison by gestational age allows the mortality risk of growth retardation to be observed clearly. The authors who compared by gestational age found either no difIerencesf6 or poorer neonatal outcome7*9S9”0*12 in the SGA infants. However, any stratification was restricted in these studies by limitations of sample size, varying from 19 to 182 preterm growth-retarded infants. The magnitude of the preterm growth-retarded population reported here (1012 pregnancies) allowed significant differences to be obtained after stratification, not only in summary, but also within the individual categories. We can thus report with confidence that growth-retarded premature infants have a uniformly higher risk of death in the neonatal period, as well as a higher risk of fetal death and overall perinatal mortality. This increased risk of neonatal death for preterm SGA infants continues to outweigh the risk of fetal demise at all preterm gestational ages; therefore, elective preterm delivery is not indicated for growth retardation.
References 1. Preterm and post-term pregnancy and fetal growth retardation. In: Cunningham FG, MacDonald PC, Gant NF, Leveno KJ, Gilstrap LC Ill, eds. Williams obstetrics. 19th ed. Norwalk, Connecticut: Appleton & Lange, 19X+853-90. 2. Leake R. Growth disorders. In: Taeusch HW, Ballard RA, Avery ME, eds. Schaffer and Avery’s diseases of the newborn. 6th ed. Philadelphia: WB Saunders:23642. 3. Bhargava SK, Bhargava V, Kumari S, Madhavan S, Ghosh S. Outcome of babies with severe intra-uterine growth retardation: Maternal factors, congenital malformation, mortality and survival pattern. Indian J Med Res 1964$2:367-74. 4. Starfield 8, Shapiro S, McCormick M, Brass D. Mortality and morbidity in infants with intrauterine growth retardation. J Pediatr 1982;101:978-83. 5. Laurin J, Persson PH, Polberger S. Perinatal outcome in growth retarded pregnancies dated by ultrasound. Acta Obstet Gynecol *and 1987;66:337-43.
174
Piper
et al
Growth Retardation and Preterm
6. T&erg AJ, Walther FJ, Pena IC. Mortality, morbidity, and outcome of the small-for-gestational-age infant. Semin Perinatol 1988;12:8494. 7. Balcazar H, Haas JD. Retarded fetal growth patterns and early neonatal mortality in a Mexico City population. Bull Pan Am Health Organ 1991;12:55-63. 8. D&son PC, Abell DA, Be&her NA. Mortality and morbidity of fetal growth retardation. Aust N Z J Obstet Gynaecol 1981;21:6972. 9. Kmps BL, Morgan LJ, Battaglia FC. Neonatal mortality risk in relation to birthweight and gestational age: Update. J Pediatr 1982;101:969-77. 10. Perry CP, Harris RE, DeLemos RA, Null DM Jr. Intrauterine growth-retarded infants: Correlation of gestational age with matemal factors, mode of delivery, and perinatal survival. Obstet Gynecol 1976;48:182-6. 11. Williams RL, Creasy RK, Cunningham GC, Hawes WE, Norris FD, Tashiro M. Fetal growth and perinatal viability in California. Obstet Gynecol 1982;59:624-32. 12. Heionen K, Matilainen R, Koski H, Launiala D. Intrauterine growth retardation (IUGR) in preterm infants. J Perinat Med 1985;12:171-9. 13. Low JA, Boston RW, Pan&am SR. Fetal asphyxia during the intrapartum period in intrauterine growth-retarded infants. Am J Obstet Gym01 1972;113:351-7. 14. Myers SA, Ferguson R. A population study of the relationship between fetal death and altered fetal growth. Obstet Gynecol 1989;74:3%31. 15. Simpson JL. Genetic counseling and prenatal diagnosis. ln: Gabbe SG, Niebyl JR, Simpson JL, eds. Obstetrics normal and problem pregnancies. 2nd ed. New York: Churchill Livingstone, 1991:26998.
Address reprint requests to:
Jemna M. Piper, MD University of TexasHealth ScienceCenter Department of Obstetrics and Gynecology 7703 Floyd Curl Drive San Antonio, TX 78284-7836
Received July 27, 1995. Received in revisedform October 10, 1995.
Accepted October 24, 1995. Copyright 0 1996 by The American College of Obstetricians Gynecologists. Published by Elsevier Science Inc.
and
Obstetrics & Gynecology
A common belief of obstetricians and neonatologists is that growth-retarded fetuses are “stressed” by their intrauterine environment and that this stress accelerates their maturation so that growth-retarded preterm infants have fewer complications of prematurity than their appropriately grown peers.‘,* Based on this concept, pregnancies complicated by growth retardation might benefit from planned preterm delivery to avoid From the Department of Obstetrics and Gynecology, the University of Texas Health Science Center at San Antonio, San Antonio, Texas.
87, NO.
2, FEBRUARY
1996
MCFARLAND,
MD,
D. BERKUS, MD, AND ODED LANCER, MD
Objective: To determine if perinatal morbidity and mortality differ in growth-retarded, small for gestational age @GA), premature infants and appropriate for gestational age (AGAl infants. Methods: All consecutive, singleton, nondiabetic, preterm pregnancies delivered over a B-year period were analyzed. Infants were categorized as SGA (at (M below the tenth percentile) or AGA (11th to the 89th percentiles), then stratified by birth weight and gestational age categories. Perinatal morbidity and mortality were examined. Results: We studied 4183 preterm deliveries, 1012 of them SGA and 3171 of them AGA. Overall, we found significantly higher rates of fetal and neonatal death in the SGA group. Stratification by gestational age revealed significantly higher rates of neonatal death for the SGA group compared with the AGA group in each gestational age category. Overall, comparison also revealed significantly tigher rates of fetal heart rate abnormality in the SGA group but no difference in neonatal sepsis, birth trauma, cesarean delivery, hyaline membrane disease, or congenital anomalies. Conclusion: Growth-retarded premature infants have a significantly higher risk of morbidity and mortality, both before and after delivery, than do appropriately grown infants. (Obstet Gynecol 2996;87:169-74)
VOL.
MD, MELINDA
the possibility of fetal death. Before proceeding to action based on this idea, however, we must critically examine the evidence to support or refute it. Conflicting findings on different aspects of this issue have been reported. Neonatal death rates have been reported to be decreased,3,4 unchanged,5,6 and increased7-‘* m ’ small for gestational age (SGA) infants in comparison with appropriately grown infants. Neonatal morbidity has been reported to be both reduced’ and elevated4,r3; however, fetal death rates have been reported uniformly as increased in SGA pregnancies. 10,11,74 The purpose of this study was to compare growth-retarded premature infants with appropriately grown preterm infants at the same gestational age with respect to perinatal morbidity and mortality.
Materials and Methods All pregnancies delivered during a 15-year period (July 1970 to June 1985) at the teaching hospital of Bexar County in San Antonio, Texas, were analyzed individually on discharge by obstetric faculty physicians for historical, demographic, delivery, and outcome information, then entered into our computerized data base. Maternal characteristics included age, gravidity, parity, race, marital status, preexisting medical problems, and complications of pregnancy. Delivery characteristics included gestationa age at delivery, mode of delivery, trauma, fetal heart rate (FHR) abnormality, and type of anesthesia and/or analgesia. Fetal and neonatal data included death, birth weight, Apgar scores, respiratory problems, neonatal infection, and congenital anomalies. Gestational age at delivery was based on the mother’s last menstrual period confirmed by early examination, fetal heart tone auscultation, ultrasonography, or neo-
0029-7844/96/$15.00 SSDI 0029-7844(95)00400-9
169
natal assessment. Gestational age was revised only if ultrasonographic examination was incompatible with menstrual dating. Obstetric assessment of gestational age was used in all cases, with neonatal assessment of dating used only for confirmation. All infants were examined systematically for anomalies at delivery and during their nursery stay by obstetric and pediatric physicians. All suspected anomalies were confirmed by obstetric or neonatal faculty members following diagnostic evaluation. Neonatal respiratory disease was differentiated based on clinical presentation and radiographic findings. The diagnosis of hyaline membrane disease was based on persistent oxygen requirement and characteristic radiographic appearance. Testing for gestational diabetes was perform& in all women with risk factors (such as obesity, family history, prior macrosomic infant, and prior gestational diabetes). From this data set, all singleton pregnancies resulting in the delivery of an infant at 24-36 weeks gestation were selected. We excluded pregnancies complicated by diabetes (preexisting or gestational) and pregnancies with unknown or conflicting dating criteria (neonatal dating incompatible with obstetric dating). Pregnancies with birth weight at or below the tenth percentile (SGA) were then compared with those with birth weights between the 11th and 89th percentiles (appropriate for gestational age [AGAI). Birth percentiles were based on population-specific growth standards derived from 80,000 consecutive contemporaneous live-born, singleton, nonanomalous infants (20-45 weeks’ gestation) in the local inner city, economically compromised population (82% Mexican-American, 11% white, 7% black). All pregnancies were analyzed for perinatal mortality and congenital anomaly rates. Only live births were used for comparison of neonatal death, FHR abnormality, Apgar scores, mode of delivery, neonatal infection, and birth trauma. Because of the potential correlation between congenital anomalies and fetal growth restriction (FGR), fetal death and perinatal mortality (expressed as number per 1000 births) as well as neonatal death (expressed as number per 1000 live births) were examined before and after exclusion of anomalous fetuses. Hyaline membrane disease rates were compared between SGA and AGA pregnancies. In addition, because of the lengthy period of study required to compile this information (15 years), the population was compared by 5-year blocks for variance within the study interval. Categorical data between SGA and AGA pregnancies were compared using J analysis. Comparison was also made after stratification by gestational age and birth weight using Mantel-Haenszel 2 analysis. Stepwise logistic regression analyses were performed with fetal death, neonatal death, and perinatal death as the de-
170 Piper et al
Growth Retardation and Preterm
pendent variables, and abruptio placentae, FGR, placenta previa, chorioamnionitis, maternal medical complications, maternal hypertension, birth trauma, FHR abnormality, congenital anomalies, 5-minute Apgar score less than 7, respiratory problems, and cesarean delivery as the independent variables. Results are displayed as odds ratio (OR) with 95% confidence interval (CI).
Results More than 75,000 singleton deliveries occurred over the 15-year study period. The population was largely Mexican-American (75%) and of low socioeconomic status. After exclusions for inadequate dating criteria (n = 281, 6.3%), 4183 nondiabetic pregnancies delivering before 37 weeks’ gestation remained for analysis, 1012 of whom were SGA and 3171 AGA. In the SGA pregnancies, 728 (72%) resulted in surviving infants, 170 (17%) resulted in fetal deaths, and 114 (11%) resulted in deaths in the neonatal period. Among the AGA pregnancies, 2633 (83%) of the infants survived, 218 (7%) resulted in fetal death, and 320 (10%) newborns died in the neonatal period. Overall, the SGA pregnancies had a significantly higher perinatal mortality rate than the AGA pregnancies (281 versus 170, OR 1.91 [95% CI 1.61-2.261). Both the fetal death rate (168 versus 69, OR 2.74 [95% CI 2.20-3.411) and neonatal death rate (135 versus 108, OR 1.29 [95% CT 1.02-1.631) were significantly higher for the SGA group than the AGA group. Anomaly rates did not differ significantly between SGA and AGA pregnancies (14 versus 13%). By eliminating malformed cases, correction of fetal, neonatal, and perinatal mortality rates revealed perinatal mortality of 262 versus 159, OR 1.89 (95% CI 1.57-2.27), fetal mortality of 169 versus 75, OR 2.52 (95% CI 1.993.18), and neonatal mortality of 112 versus 91, OR 1.27 (95% CI 0.96-1.67). There were no significant differences in gestational age distribution, proportion of SGA fetuses, or perinatal mortality rates between the three 5-year blocks within the study period. Both SGA and AGA groups included deliveries of 24-36 weeks’ gestation (Table 1). Stratification of fetal deaths by gestational age showed a uniform distribution in the SGA group, whereas for the AGA group, fetal deaths decreased as gestational age increased (Table 1). There was an excess of fetal deaths noted in the AGA group before 27 weeks (92 of 218,42% of total AGA fetal deaths). Stratification of neonatal mortality by gestational age revealed a significant difference between SGA and AGA pregnancies (overall OR 7.07 [95% CI 4.9-10.11) (Figure 1A). Furthermore, examination of the individual categories identified a significantly higher neonatal death
Obstetrics & Gynecology
Table 1. Distribution by Gestational Age Categories Gestational age (wk)
Total pregnancies SGA AGA Fetal deaths SGA AGA Neonatal deaths SGA AGA
24 -26
27-28
29-30
31-32
33-34
35-36
44 (4%) 265 (8%)
44 (4%) 291 (9%)
60 (6%) 362 (11%)
109 (11%) 670 (21%)
205 (20%) 809 (26%)
550 (55%) 774 (25%)
25 (15%) 92 (42%)
22 (13%) 36 (17%)
25 (15%) 31 (14%)
30 (18%) 29 (13%)
31 (18%) 16 (7%)
37 (21%) 14 (7%)
18 (16%) 114 (35%)
18 (16%) 108 (34%)
18 (16%) 51 (16%)
25 (22%) 25 (8%)
21 (18%) 15 (5%)
14 (12%) 7 (2%)
SGA = small for gestational age; AGA = appropriate for gestational age. The data are presented as number of total pregnancies, fetal deaths, or neonatal deaths in each gestational age category, and percentage of the total number of pregnancies or deaths in the SGA or AGA group (percentage out of cases in that line).
rate for SGA pregnancies in every gestational age category (Table 2). However, stratification by birth weight (Figure 1B) revealed no significant difference in neonatal mortality rates between SGA and AGA pregnancies either overall or by individual categories (Table 3). Comparison of perinatal morbidity also revealed significant differences. The diagnosis of FHR abnormality was more common in the SGA pregnancies (10.5 versus 8%, OR 1.34 195% CI 1.03-1.7511 and the need for cesarean approached significance (26 versus 23%, OR 1.19 195% CI 0.99-1.421). Apgar score less than 7 at 5 minutes (21.9 versus 19.5%, OR 1.15 [95% CI 0.95-1.3911, birth trauma (1.7 versus 2.6%‘), and neonatal sepsis (7.8 versus 6.2%) did not differ significantly between SGA and AGA infants. Based on the strong relationship between measures of
A
age-size and respiratory disease, we stratified our SGA and AGA pregnancies by both gestational age and birth weight categories for comparison of the rates of hyaline membrane disease (Figure 2). Comparison by gestational age categories revealed a significantly higher rate of hyaline membrane disease in the SGA infants (OR 1.70 [95% CI 1.29-2.231) compared with their AGA peers (Figure 2A). In contrast, comparison by birth weight categories revealed a significantly lower rate of hyaline membrane disease in the SGA pregnancies (OR 0.75 [95% CI 0.58-0.951) (Figure ZR). To evaluate the true net effect of each of the independent variables on outcome, logistic regression analysis was performed individually for fetal death, neonatal death, and perinatal mortality (Table 41. The independent variables analyzed for fetal death were presence or absence of growth retardation, anomalies, chorioamnio-
T--__-
B /’1;
1,000
ISGP IAGP
Figure 1. Neonatal mortality (NNM) rates in small for gestational age (SGA) and appropriate for gestational age (AGA) pregnancies compared by gestational age (A) and birth weight (B). Neonatal death rates in SGA pregnancies were significantly higher by gestational age (P < .05) but not by birth weight.
NBOO M po: 1000 200
2426 27.20 20.30 3142 33.34 35.36
G’ESTATIONAL AGE VOL. 87, NO. 2, FEBRUARY
1996
BIRTHWEIGHT (g) Piper et al
Grozvth Retardation and
Preterm
171
Table 2. Neonatal
Mortality
Rates by Gestational
Age
rate/1000 live births
Gestational age (wk)
SGA
AGA
OR (95% CI)
24-26 27-28 2930 3132 3s34 35-36
947 818 514 316 121 27
659 424 154 39 19 9
9.32 W-51.1) 6.13 (1.9-22) 5.81 (2.7-12.8) 11.40(5.9-22.2) 7.10 (3.4-14.9) 3.02 (1.1-8.3)
SGA = small for gestational age; AGA = appropriate for gestational age; OR = odds ratio; CI = confidence interval.
nitis, abruptio placentae, placenta previa, maternal medical complications, and maternal hypertension. For both neonatal death and perinatal mortality, additional variables entered into the logistic regression model were 5-minute Apgar score less than 7, cesarean delivery, birth trauma, FHR abnormality, and respiratory problems. Only chorioamnionitis proved to be a significant contributor to all three dependent variables; however, several other factors were contributora for one or more logistic regression analyses (Table 4). Growth retardation was a significant risk factor for fetal death and approached statistical significance for neonatal death (P = .06); thus, it was an overall contributor to perinatal mortality (OR 2.17 195% CI 1.8-2.61). Chorioamnionitis also contributed to both fetal and neonatal mortality. Placental complications (abruption and placenta previa) increased risk of fetal death but did not alter neonatal mortality. Low Apgar scores, anomalies, and neonatal respiratory disease were significantly associated with neonatal death. Cesarean delivery was associated with a protective effect against neonatal and perinatal death.
Discussion The key findings of our study to be discussed individually are as follows: 1) Growth-retarded preterm infants Table 3. Neonatal
Mortality
Rates by Birth Weight
Categories SGA
AGA
Birth weight (g)
n
NMR
n
NMR
P
-aQo 500-749 750-999 1000-1249 1250-1499 1500-1999 ZOO0
15 39 40 91 78 327 252
933 821 475 253 167 37 4
3 78 168 198 200 755 1553
795 607 364 150 50 10
NS NS NS NS NS NS
S-GA = small for gestational age; AGA = appropriate for gestational age; NMR = neonatal mortality rate/lWiJ live births; NS = not significant.
172 Piper et al
Growth Retardation and Preterm
have no survival advantage over appropriately grown infants of the same gestational age; 2) growth retardation is an independent predictor of perinatal mortality; 3) the rates of hyaline membrane disease are significantly higher in SGA infants than in appropriately grown infants of the same gestational age; and 4) the rates of hyaline membrane disease are significantly lower in the SGA infants compared with appropriately grown infants of the same birth weight. Our analysis of more than a thousand growthretarded preterm pregnancies revealed significantly higher fetal death rates, neonatal death rates, and perinatal mortality rates in the growth-retarded pregnancies compared with appropriately grown pregnancies. Stratification by gestational age also displayed significantly higher neonatal death rates for the SGA infants. This finding is compatible with prior reports of decreased survival in preterm growth-retarded infants However, comparison of our by gestational age.7,9~‘o~12 data by birth weight categories showed no difference in neonatal mortality between SGA and AGA infants. This finding is also compatible with previous reports of unchanged or improved outcome in preterm SGA infants compared by birth weight only,3” although ours is the first report of morbidity and mortality compared by both gestational age and birth weight in a single large population of preterm infants. The percentage of SGA infants in our study population may seem high (1012 of 4183, 24%), but it is important to recognize that our study cohort spec&ally excludes diabetic pregnancies and large for gestational age infants, but includes fetal deaths and fetuses with congenital anomalies (specifically excluded from birth percentile curve calculations) because of their obvious impact in the consideration of morbidity and mortality. Accumulation of a population of this size required collection over an extended period of time (15 years, 1970-1985); however, analysis by 5-year blocks revealed no significant change in population characteristics over that time. Growth retardation was an independent predictor of fetal death, neonatal death, and perinatal mortality. Other factors asblociated with fetal death were placental problems and intrauterine infection. Additional factors associated with neonatal mortality were chorioamnionitis, neonatal respiratory disease, S-minute Apgar score less than 7, and congenital anomalies. Cesarean delivery was associated with decreased neonatal mortality, probably because of more intensive therapy in these cases. Of interest was our finding of congenital anomalies in 13-14% of our preterm deliveries, whether SGA or AGA, much higher than the 2-4% rates quoted for United States births overall.” We expected congenital anomalies to be more frequent in SGA pregnancies but this proved not to be the case. Undiagnosed anom-
Obstetrics & Gynecology
Figure 2. Hyaline membrane disease (HMD) rates in small for gestational age (SGA) and appropriate for gestational age (AGA) infants compared by gestational age (A) and birth weight (8). Comparison by gestational age revealed an overall significantly higher rate of hyaline membrane disease in the SGA infants compared with the AGA infants (P < .05). In contrast, when compared by birth weight categories, the rate of hyaline membrane disease was lower in the SGA than in the AGA pregnancies (P < .05). 24-26 27-28
29-30 31-32 33-34 35-35
Gestational
Age (weeks)
alies are always a concern, but every infant in this preterm population was examined specifically for malformations and, if anything, one would anticipate that the SGA infants were scrutinized even more scrupulously than their AGA peers. Thus, we conclude that pregnancies delivered preterm are at high risk of congenital anomalies (13-14%), regardless of fetal growth pattern. In our evaluation of neonatal morbidity, comparison by gestational age showed significantly higher rates of hyaline membrane disease in the SGA infants than in the AGA infants. A commonly held belief is that the stress of growth retardation accelerates pulmonary maturation so that preterm delivery is not as great a risk for a growth-retarded pregnancy.lJ2 Our data clearly do not support such a relationship. Indeed, at any given point Table
4. Factors Associated with and Perinatal Death
Abruptio placentae Growth retardation Placenta previa Chorioamnionitis Congenital anomalies 5-min Apgar <7 Respiratory problems Cesarean delivery
Risk of Fetal, Neonatal,
Fetal death
Neonatal death
Perinatal death
4.45 (3.3-6.1) 2.85 (2.3-3.6) 2.42 (1.6-3.8) 1.81 (1.2-2.7) 0.56 (0.4-0.8)
0.87 (0.5-1.4) 1.31 (1.0-1.07) 0.86 (0.5-1.6) 2.05 (1.3-3.1) 1.96 (1.5-2.6)
2.80 (2.1-3.8) 2.17 (1.8-2.6) 1.72 (1.1-2.6) 2.13 (1.6-2.9) 1.22 (0.96-l .5)
18.5 (14.3-24.0) 1.92 (1.5-2.5)
7.12 (5.8-8.7) 1.54 (1.3-1.9)
0.60 (0.4-0.8)
0.66 (0.5-0.8)
Logistic regression, expressed as odds ratio (95% confidence interval).
VOL.
87, NO.
2, FEBRUARY
1996
600
500.750-l&w1250.1500. >1999 749 ss9 124s 1499 lBS@
Birthweight
(grams)
in pregnancy, the SGA infant is placed at higher risk by preterm delivery than a comparable AGA pregnancy because of higher morbidity and mortality. On initial evaluation, our finding of significantly lower rates of hyaline membrane disease in the SGA infants compared by birth weight would appear to create a conflict. However, this is easily explained by the fact that within each birth weight category, the AGA infants were of much lower gestational age than the SGA infants. Thus, the SGA infants were protected by their higher gestational age, whereas stratification by gestational age removed the protection, allowing the full effect of growth retardation to be seen. This highlights the need for comparison by both birth weight and gestational age to draw adequate conclusions regarding the effects of growth retardation on perinatal morbidity. Although comparison by both birth weight and gestational age are essential for research purposes, based on our data, we feel that gestational age is the better predictor of neonatal outcome. Likewise, comparison of neonatal mortality by both gestational age and birth weight is essential to determine the impact of growth retardation. We found significantly higher mortality in SGA infants than in AGA infants when compared by gestational age but no difference when compared by birth weight. Prior reports compared outcome based on either birth weight or gestational age, but not both. Indeed, the reports of higher neonatal survival rates in SGA infants were based on comparison by birth weight only.3,4 As described above for hyaline membrane disease, it is reasonable that comparison by birth weight alone could
Piper et al
Growth Retardation
and Preterm
173
show increased survival for SGA infants because an 800-g SGA infant of 29 weeks’ gestation would be compared with an 800-g AGA infant of 24-25 weeks gestation. Thus, when compared by birth weight, the benefit of being less premature outweighs the risk from growth retardation, whereas comparison by gestational age allows the mortality risk of growth retardation to be observed clearly. The authors who compared by gestational age found either no difIerencesf6 or poorer neonatal outcome7*9S9”0*12 in the SGA infants. However, any stratification was restricted in these studies by limitations of sample size, varying from 19 to 182 preterm growth-retarded infants. The magnitude of the preterm growth-retarded population reported here (1012 pregnancies) allowed significant differences to be obtained after stratification, not only in summary, but also within the individual categories. We can thus report with confidence that growth-retarded premature infants have a uniformly higher risk of death in the neonatal period, as well as a higher risk of fetal death and overall perinatal mortality. This increased risk of neonatal death for preterm SGA infants continues to outweigh the risk of fetal demise at all preterm gestational ages; therefore, elective preterm delivery is not indicated for growth retardation.
References 1. Preterm and post-term pregnancy and fetal growth retardation. In: Cunningham FG, MacDonald PC, Gant NF, Leveno KJ, Gilstrap LC Ill, eds. Williams obstetrics. 19th ed. Norwalk, Connecticut: Appleton & Lange, 19X+853-90. 2. Leake R. Growth disorders. In: Taeusch HW, Ballard RA, Avery ME, eds. Schaffer and Avery’s diseases of the newborn. 6th ed. Philadelphia: WB Saunders:23642. 3. Bhargava SK, Bhargava V, Kumari S, Madhavan S, Ghosh S. Outcome of babies with severe intra-uterine growth retardation: Maternal factors, congenital malformation, mortality and survival pattern. Indian J Med Res 1964$2:367-74. 4. Starfield 8, Shapiro S, McCormick M, Brass D. Mortality and morbidity in infants with intrauterine growth retardation. J Pediatr 1982;101:978-83. 5. Laurin J, Persson PH, Polberger S. Perinatal outcome in growth retarded pregnancies dated by ultrasound. Acta Obstet Gynecol *and 1987;66:337-43.
174
Piper
et al
Growth Retardation and Preterm
6. T&erg AJ, Walther FJ, Pena IC. Mortality, morbidity, and outcome of the small-for-gestational-age infant. Semin Perinatol 1988;12:8494. 7. Balcazar H, Haas JD. Retarded fetal growth patterns and early neonatal mortality in a Mexico City population. Bull Pan Am Health Organ 1991;12:55-63. 8. D&son PC, Abell DA, Be&her NA. Mortality and morbidity of fetal growth retardation. Aust N Z J Obstet Gynaecol 1981;21:6972. 9. Kmps BL, Morgan LJ, Battaglia FC. Neonatal mortality risk in relation to birthweight and gestational age: Update. J Pediatr 1982;101:969-77. 10. Perry CP, Harris RE, DeLemos RA, Null DM Jr. Intrauterine growth-retarded infants: Correlation of gestational age with matemal factors, mode of delivery, and perinatal survival. Obstet Gynecol 1976;48:182-6. 11. Williams RL, Creasy RK, Cunningham GC, Hawes WE, Norris FD, Tashiro M. Fetal growth and perinatal viability in California. Obstet Gynecol 1982;59:624-32. 12. Heionen K, Matilainen R, Koski H, Launiala D. Intrauterine growth retardation (IUGR) in preterm infants. J Perinat Med 1985;12:171-9. 13. Low JA, Boston RW, Pan&am SR. Fetal asphyxia during the intrapartum period in intrauterine growth-retarded infants. Am J Obstet Gym01 1972;113:351-7. 14. Myers SA, Ferguson R. A population study of the relationship between fetal death and altered fetal growth. Obstet Gynecol 1989;74:3%31. 15. Simpson JL. Genetic counseling and prenatal diagnosis. ln: Gabbe SG, Niebyl JR, Simpson JL, eds. Obstetrics normal and problem pregnancies. 2nd ed. New York: Churchill Livingstone, 1991:26998.
Address reprint requests to:
Jemna M. Piper, MD University of TexasHealth ScienceCenter Department of Obstetrics and Gynecology 7703 Floyd Curl Drive San Antonio, TX 78284-7836
Received July 27, 1995. Received in revisedform October 10, 1995.
Accepted October 24, 1995. Copyright 0 1996 by The American College of Obstetricians Gynecologists. Published by Elsevier Science Inc.
and
Obstetrics & Gynecology