- Email: [email protected]
Incidence and risk factors for symptomatic patent ductus arteriosus among inborn very-low-birth-weight infants
39
Early Human Development, 12 (1985) 39-48 Elsevier
EHD 00665
Incidence and risk factors for symptomatic patent ductus arteriosus among inborn very-low-birth-weight infants Jaime A. Furzan *, Joan Reisch, Jon E. Tyson, Pennock Laird and Charles R. Rosenfeld Depariments of Pediatrics, Obstetrics and Gynecology, and Internal Medicine, The University of Texas Health Science Center at Dallas, Southwestern Medical School, Dallas, Texas, U.S.A. Accepted
for publication
1 April 1985
Summary
The incidence of symptomatic patent ductus arteriosus (sPDA) in very-low-birthweight infants has been reported almost exclusively from referral centers. Moreover, the incidence has varied considerably. We prospectively evaluated the incidence and risk factors of sPDA in a totally inborn population of infants I 1500 g birth weight (n = 120) receiving conservative fluid management and examined the role of echocardiography in predicting sPDA before clinical signs are evident. The incidence of sPDA was relatively low (16%) and the age at onset was relatively late (mean f SE. = 14.7 k 3.0 days) when compared to prior reports. Fchocardiographic changes preceded physical and radiographic findings of sPDA. Fluid volume within the range administered to our infants was unrelated to the occurrence of sPDA. However, risk factors identifiable within 24 h of birth were defined: obstetric estimate of gestational age, race, mean fluid volume intake in the initial 24 h, and early treatment with volume expanders. From these risk factors, 79% of infants with sPDA and 79% without sPDA could be predicted by stepwise discriminant function analyses. These findings may be employed in studies needed to assess the benefit of early surgical or pharmacologic therapy for infants at highest risk for sPDA. patent ductus arteriosus; very-low-birth-weight tors; volume expansion
* Dr. Furzan was a Fellow in Neonatal-Perinatal Caribobo, Venezuela. Address all correspondence to: Charles R. Rosenfeld, Boulevard, Dallas, Texas 75235, U.S.A.
037%3782/85/$03.30
infants; echocardiography;
Medicine
and
M.D., Department
0 1985 Elsevier Science Publishers
presently
is at the University
of Pediatrics,
B.V. (Biomedical
risk fac-
Division)
5323 Harry
of
Hines
40
Introduction
The incidence of symptomatic patent ductus arteriosus (sPDA) among very-lowbirth-weight infants (VLBW; _<1500 g birth weight) has been reported almost exclusively from referral institutions [4,5,7,12] and until recently, institutions in which fluids have been administered liberally [4,5]. Incidence rates have been as high as 50% but have differed considerably between centers [4,5,7,12,13]. Despite the use of uniform diagnostic criteria, the incidence varied more than 3-fold (ll-36%) between centers in a recent large collaborative study [6]. In a review of all VLBW infants born at our center between June 1978 and May 1979, we found a relatively low incidence of sPDA (16%). Moreover, the mean age at onset, 20 * 4 (S.E.) days was considerably later than would be expected from published studies. Differences between centers with respect to sPDA might relate to selection biases in the referral of patients, as well as differences in obstetric and neonatal risk factors. For these reasons, we initiated a prospective study in a large inborn population of VLBW infants to determine: 1) if the incidence of sPDA is lower and the onset later than that generally reported from other institutions; 2) if the risk factors that relate to the occurrence of sPDA identified by multivariant discriminant analysis are similar to those reported for a largely outborn population [5]; and finally, 3) if echocardiographic changes precede the clinical diagnosis of sPDA, thus facilitating identification of infants at-risk for sPDA.
Materials and Methods
Sample size and patient population 120 consecutive infants who weighed 1500 g or less at birth and lived at least 72 h were enrolled. These infants were all born at Parkland Memorial Hospital between June 1, 1979 and April 30, 1980. Our sample size was selected prior to study as an appropriate number of infants in order to distinguish major differences between infants with sPDA and those without, assuming an incidence of sPDA similar to that observed in our retrospective study, 16%. The postnatal age at entry to the study, as well as the diagnostic criteria, were selected to allow comparison of our incidence and risk factors with those identified in a referral population by Cotton et al. [5], who performed discriminant function analyses similar to those which we planned. Given the relatively late onset of sPDA observed in our retrospective study, we anticipated that a diagnosis of sPDA would be unlikely prior to 3 days. Patient care procedures 85% of mothers delivering at Parkland Memorial Hospital received prenatal care. Mothers with complications of pregnancy were hospitalized in a high-risk maternity unit. An Obstetrics faculty member remained in the hospital at all hours to supervise care in the Labor and Delivery unit. The delivery of VLBW infants was routinely attended by a senior Pediatric
41
resident. At delivery, the umbilical cord of each infant was routinely clamped without delay. During resuscitation, emphasis was placed on ensuring adequate ventilation. Sodium bicarbonate was used in limited quantities (5 2 mequiv./kg of a 0.5 mequiv./ml solution). Each infant was transferred from the delivery room to the Intensive Care Nursery, which was supervised by 1 of 2 attending neonatologists who evaluated all infants twice daily. Uniform guidelines for the administration of volume expanders and the regulation of fluid balance, infant temperature, and ventilatory therapy were employed. Administration of volume expanders (saline, whole blood or plasma) was encouraged only in the presence of both delayed capillary filling (> 3 set) and hypotension, defined from relatively conservative standards for normal blood pressure in neonates (a systolic blood pressure > 1 S.D. below the mean value reported by Bucci et al. [3] for infants of the same birth weight, gestational age, and postnatal age). Fluid intake was regulated to maintain an isotonic urine and to allow a weight loss of approximately 15% during the first week. Skin temperature of all infants was servo-controlled at 36.5”C. Infants were routinely nursed under a radiant warmer for at least 1-2 days and whenever ventilatory therapy was administered. Intubation was performed to provide continuous distending airway pressure or mechanical ventilation for infants with hyaline membrane disease who required an FiO, > 0.6, or developed hypercapnia (PaCO, > 50-60 mm Hg) or a rapid clinical deterioration. Infants with sPDA, as defined below, were treated with furosemide (1 mg/kg per day) and fluid restriction (I 100 ml/kg per day). Surgical closure was undertaken when sPDA did not improve after at least 48 h of medical therapy. Indomethacin was not used in any infant. Patient evaluation Following study entry, each infant was examined every other day by the same investigator (JAF) for the first 2 postnatal weeks and twice weekly thereafter until discharge. Symptomatic PDA was diagnosed using the criteria of Cotton et al. [4,5] so as to provide comparable data. This diagnosis required a large left-to-right shunt (i.e., at least 4 of the following signs: elevated basal heart rate (> 165 beats/min), hyperdynamic precordium, bounding pulses, characteristic murmur, progressive cardiomegaly (cardiac/thoracic ratio > 0.60) and prominent pulmonary vasculature on chest x-ray, and an abnormal echocardiogram (left atrial/aortic root ratio > 1.4, increased left ventricular and atria1 dimensions)) and signs of congestive heart failure (i.e., at least 2 of the following signs: gallop rhythm, pulmonary edema on chest x-ray, congestive hepatomegaly, increasing episodes of apnea and bradycardia unexplained by prematurity (apneic episodes > 20 set associated with heart rate < 100 bpm), and carbon dioxide retention (PaCO, > 55 mm Hg) unexplained by lung disease). Information on fluid intake, ventilator therapy, weight gain, and days of oxygen therapy were recorded daily following the infant’s physical examination. Bronchopulmonary dysplasia, necrotizing enterocolitis, and other medical problems were diagnosed according to predetermined criteria. Hyaline membrane disease was diagnosed in the presence of progressively increasing oxygen requirements (FiOz >
42
0.40) typical radiographic findings, and a negative blood culture. Acute perinatal stress was recorded in the presence of any of the following: placenta previa, placental abruption, umbilical cord prolapse, breech vaginal delivery, birth asphyxia (Apgar score of 6 or less at 1 or 5 min), or hypovolemia during the first 24 h after birth (i.e., the presence of 3 or more of the following: capillary filling time r 3 set, hypotension as defined above, severe metabolic acidosis (pH < 7.10), evidence of abnormal bleeding, anemia (central hematocrit I 40%) and poor urinary output (< 0.5 ml/kg per h)). For each infant a M-mode echocardiogram was scheduled between the 7th and 11th days of life, an age when sPDA would not be expected from our retrospective study. These studies were performed to determine whether echocardiographic findings preceded clinical findings of sPDA. All echocardiograms were read by the same person (PL) without knowledge of the infant’s clinical course. Assessment included measurements of both ventricles, left atrium, aorta, shortening fraction, left atrium to aortic root ratio, and the determination of two indices: left atrium measured in millimeters divided by birth weight in grams, and left ventricular end-diastolic dimension in millimeters divided by birth weight in grams. Statistical methods Descriptive statistics were calculated for all variables. The numerical variables were summarized using means and standard deviations, while frequency distributions were determined for the categorical variables. The sPDA group and group without sPDA were compared for each variable considered separately. (X-square contingency table analysis or Fisher’s Exact Probability Test were utilized for categorical variables, and Student’s t-tests for continuous variables. To determine whether one or more of all the variables measured was predictive of sPDA, multivariate discriminant analysis was used. The Statistical Package for the Social Sciences discriminant analysis program with Wilks’ stepwise selection method was utilized.
Results
The mean birth weight (f S.D.) of the 120 study infants was 1165 + 233 g; the mean gestational age by obstetric criteria was 29.4 f 3.1 wk. Hyaline membrane disease occurred in 33% of all infants. The mean fluid intake ( f S.D.) was 91.1 f 22.3 ml/kg per day on the first day, 131 f 20.4 during the first week, and 162 f 21.3 during the second week. Volume expanders were administered during the first 24 h after birth to 27% of infants (32/120). The incidence of sPDA was 16% (19/120); another 21% of infants (25/120) had clinical evidence of a left-to-right shunt. The incidence of sPDA was 26% (8/31) among infants < 1000 g and 12% (11/89) in infants between 1000 and 1500 g. Among infants with hyaline membrane disease sPDA occurred in 29% (9/39). The mean age at onset was 14.7 (range 3-35) days. The diagnosis was made before 7 days (‘early onset’ group) in 4 of 19 (21%) infants; 15 (79%) were diagnosed after the
43 TABLE
I
Characteristics
of the study population
Variable
No. of patients Birth weight (g) Gestational age (wks) Sex Male Female Race Negro Caucasian Latin AGA infants SGA infants ’ Acute perinatal stress Birth asphyxia 1 min Apgar 5 min Apgar Prolonged rupture membranes Hypocalcemia Hyaline membrane disease Volume expanders first 24 h Continuous distending pressure first 24 h Fluid intake first 24 h d Fluid intake first week d Fluid intake second week d lntraventricular hemorrhage Necrotizing enterocolitis Bronchopulmonary dysplasia Deaths (between 3 and 60 days) a Values presented are mean f 1 S.D. b Birth weight less than tenth percentile
according
to the presence
With symptomatic
or absence
PDA
19 1048 + 247 29.4 + 2.2
of symptomatic
Without symptomatic
PDA
101 1189+223’ 31.2 f 2.2 ’
10
45 56
9 11 8 0 14 5 16 (84.2%) 13 (68.4%) 3.9+2.6 6.2+1.9 6 (31.6%) 18 (94.7%) 9 (47.4%) 10 (52.6%)
62 23 16 61 40 70 (69.3%) 68 (67.3%) 4.9 f. 2.7 7.2 f 2.1 21 (20.8%) 77 (76.2%) 30 (29.7%) 22 (21.8%) ’
12 (63.2%) 84.3 f 12.3 130* 18.6 169 + 22.4 3 (15.8%) 0
37 (36.6%) 90.3 f 23.5 131* 20.9 152i19.5 8 (7.9%) 8 (7.9%)
4 (21.1%) 4 (21.1%)
4 (4.0%) 11 (10.9%)
for gestational
PDA a
age according
to Colorado
grid.
’ Statistically significant at P < 0.05. d Includes fluids for volume expansion.
first week (‘late onset’ group). Surgical ligation was performed in only 9 infants (7.5% of all infants; 47% of infants with sPDA). Infants with and without sPDA are compared in Table I. Those with sPDA were smaller, more preterm, had a higher incidence of placental abruption, and were more likely to have received volume expanders in the first 24 h. Borderline differences (0.10 > P > 0.05) were found for the incidence of hyaline membrane disease, hypocalcemia, and the use of continuous distending airway pressure within the first 24 h following birth. Of note, mean fluid intake in the two patient groups was virtually identical. The stepwise discriminant analysis procedure used to determine a subset of
44
variables optimal for the separation of the groups with and without sPDA yielded the following variables in the final discriminant function: obstetric estimate of gestational age (weeks), race (Negro or Caucasian), use of volume expanders (yes/no), and fluid intake in the first 24 h after birth (ml/kg per day). The discriminant score was calculated for each subject by multiplying the value of each of the variables in the function by its respective weighting coefficient and summing the resulting products. If a condition was present, it was given a value of 1 in the equation; if absent, the value was zero. If the final score was less than the cutoff value, 0.691, the infant would be predicted as not having sPDA; if the score was greater than the cutoff value, the infant would be predicted as having sPDA. Use of the discriminant function is illustrated below: Consider a Caucasian infant, with an obstetric estimate of gestational age of 32 wks, who did not receive volume expanders in the first 24 h of life, but whose fluid intake in the first 24 h was 100 ml/kg per day. This information is substituted into the final discriminant function, and a score is calculated: Class: 6.030 - 0.217 (Gestational age) + 1.343 (Volume expanders) + 1.933 (Caucasian) + 0.761 (Negro) - 0.011 (Fluid intake) Calculation of the infant’s score: Class: 6.030 - 0.217 (32) + 1.343 (0) + 1.933 (1) + 0.761 (0) - 0.011 (100) = - 0.081 Since the score is less than the cutoff value of 0.691, the infant would be predicted as not developing sPDA. Employing the discriminant function 95 of the 120 infants (79%) included in the study could be correctly predicted. The sensitivity was 79% (15/19 with sPDA were correctly predicted) and the specificity also was 79% (80/101 without sPDA were correctly predicted). Early and late onset sPDA Given the small number of affected patients, no significant differences were found between the early (n = 4) and late (n = 15) onset groups with respect to all variables shown in Table I. Only 1 of the 8 infants < 1000 g and 3 of 11 infants between 1000 and 1500 g with sPDA had an early onset. In the early onset group 2 infants died with massive intraventricular hemorrhage; the remaining 2 infants required surgical ligation at 7 and 18 days and had an uneventful recovery. The mortality in the late onset group was 13% (2/15), and was related to intraventricular hemorrhage and bronchopulmonary dysplasia. In the late onset group 7 infants (47%) required surgical ligation between 18 and 37 days. Medical& and surgically treated infants There were no significant differences between infants considered sufficiently ill to require surgical ligation and other infants with sPDA with respect to birth weight, gestational age, and echocardiographic values obtained prior to the onset of clinical signs of sPDA. Echocardiographic data Of the 120 infants enrolled into the study, 87 had an M-mode echocardiogram
TABLE
11
Echocardiographic
data according
to subsequent
No. of patients Age at echocardiogram (days) Left ventricle systolic measurement (cm) Left ventricle diastolic measurement (cm) Shortening fraction (%) Bight ventricle (cm) Aorta (cm) Left atrium (cm) Left atrium/aortic root ratio Left ventricle (mm)/birth weight (g) index Left atrium (mm)/birth weight (g) index
development
of symptomatic
PDA a
With subsequent symptomatic PDA’
Without subsequent symptomatic PDA
15 8.2 +I.0
70
0.83 kO.14 1.35 *0.21 36.7 k7.1 0.48 + 0.12 ’ 0.61+ 0.06 0.83 &-0.21 ’ 1.42kO.30 ’ 13.67 f 3.1 ’ 8.26 f 2.4 ’
8.7+1.2 0.85 +0.13 1.31 kO.15 34.8 k5.7 0.56 + 0.12 0.62 f 0.07 0.72 f 0.14 1.17*0.26 11.13+2.0 6.13k1.3
’ Values presented are mean Ifr1 S.D. ’ Includes only infants with onset of sPDA after the echocardiogram. ’ Statistically different at P < 0.05.
performed between 7 and 11 days of age; 2 were uninterpretable. The reasons for not obtaining an echocardiogram within the prescheduled period included: 7 deaths before 7 days of age from causes considered to be unrelated to sPDA; 4 infants with the onset of signs of sPDA before 7 days of age; 2 infants were in isolation due to maternal herpes or varicella; and 18 infants did not have it performed for technical reasons. To determine if echocardiographic findings preceded development of sPDA, we excluded from analysis those infants with onset of sPDA before their scheduled echocardiogram. Significant differences (P < 0.05) were found between groups and are presented in Table II. We found no significant differences between echocardiographic measurements obtained in the late onset sPDA infants and in infants who only had signs of left to right shunt without sPDA. When we compared infants with hyaline membrane disease who developed sPDA to those who did not, significantly higher values were found in the former infants for the left atrium (0.86 + 0.15 vs. 0.68 _t 0.12 cm), left atrium/aortic root ratio (1.45 + 0.29 vs. 1.10 + 0.23) left ventricle/birth weight index (13.6 * 3.9 vs. 11.0 k 2.0 mm/g) and left atrium/birth weight index (7.96 k 2.72 vs. 6.04 k 1.16 mm/g). However, infants with hyaline membrane disease who developed sPDA had values similar to those without HMD who developed sPDA.
Discussion
our
It is difficult to interpret the difference between the incidence of sPDA (16%) in study and that reported for VLBW infants in the recent collaborative study
46
(27%) [6]. In the collaborative study, a largely referral population was examined, the mean infant age at study entry was not reported, and more liberal diagnostic criteria for symptomatic PDA were used than in our study. We did not attempt, as did the investigators in the collaborative study, to identify infants affected by a ‘silent’ ductus with significant hemodynamic effects despite minimal or no physical or radiologic signs of sPDA. The reliability and accuracy with which such infants can be identified remains unclear, and their contribution to the total incidence of sPDA may not be great. Despite the difference in diagnostic criteria used in our study and in the collaborative study, comparable echocardiographic measurements were observed for infants categorized as having no significant PDA. For example, the left atrium/aortic root ratio for such infants was 1.17 k 0.26 in our study as compared to 1.13 f 0.23 in the collaborative study [8]. Our study was designed to allow comparison of our findings with those reported by Cotton et al. [5]. The lower incidence of sPDA in our center (16%) than that reported by Cotton et al. (51%) may relate to a number of variables: our entirely inborn population, the large percentage of Negros among our patients [6], or our relatively conservative use of fluid therapy [2] and volume expanders. The use of volume expanders may also relate to the lower incidence of sPDA and later onset in our study than that reported by Lorenz et al. [ll] (16 vs. 25% incidence; 21 vs. 68% of infants with sPDA whose onset occurred before day 7, respectively). Although mean birth weight and fluid intake are similar in the two studies, the infants reported by Lorenz et al. received a mean of 7-8 ml/kg of blood products and up to 30 ml/kg during the first 24 h following delivery. The presence of hypovolemia, rather than treatment with volume expanders per se, may of course increase the likelihood of sPDA. However, Arant [l] has reported that acute volume expansion in the newborn puppy may increase plasma concentration of vasodilating prostaglandins and reopen the previously closed ductus arteriosus. The use of volume expanders has not been reported in most studies of sPDA [2,4,5,10]. The administration of volume expanders has been based on standards for blood pressure measurements derived from infants with delayed clamping of the umbilical cord [9]. These standards recently have been modified to lower values [14]. It is likely that blood pressure is only loosely correlated with blood volume; thus, the administration of volume expanders in the resuscitation of normovolemic infants may inadvertently increase their risk of sPDA. Bell and coworkers have demonstrated that the administration of a high total fluid volume increases the likelihood of sPDA [2]. The increase in fluids administered to our infants after the first postnatal week at a time when use of radiant warmers and phototherapy decreased may account for the late onset of sPDA in some patients. However, from our analyses fluid volume, within the range administered in our nursery, is not an important determinant of sPDA in our population. Although mean fluid volume in the first 24 h was not significantly different between groups when analyzed by univariate analysis, it was identified as a risk factor by multivariate analyses. This reflects how single variables which by themselves do not discriminate between groups may be effective discriminators when considered with other variables.
47
A statistical model for the prediction of sPDA from clinical findings has been attempted, to our knowledge, in only 1 other study, that reported by Cotton and coworkers [5]. The perinatal risk factors identified in the two studies are quite similar. Moreover, application of their discriminant function to our study population correctly classified 73% of all infants (68% of infants with sPDA and 73% without sPDA); this would suggest that the same risk factors can be used to identify infants at highest risk for sPDA in centers with quite different patient populations and incidence of sPDA. The concept that the patent ductus arteriosus, even in the absence of clearly discernible clinical signs, is a major cause of morbidity in the first postnatal week has prompted some investigators to suggest early, aggressive therapy [7,12]. Our echocardiographic findings for infants later diagnosed as having sPDA provide evidence that significant hemodynamic effects preceded recognition of sPDA; however, the value of aggressive early pharmacologic or surgical therapy to prevent or treat PDA in VLBW infants remains to be established in properly designed, randomized trials. The influence of volume expanders on the incidence of PDA and the outcome of high-risk infants has not yet been addressed and also warrants rigorous study. Given substantial differences between centers in the incidence of sPDA, the risk factors identified in discriminant function analyses in studies like ours or that by Cotton and coworkers [5] may be useful in such studies to identify VLBW infants most likely to be affected by PDA.
References I Arant. B.S., Jr. (1981): Nonrenal factors influencing renal function during the perinatal period. Clin. Perinatol. 8, 225-240. D., Stonestreet, B.S. and Oh, W. (1980): Effect of fluid administration on the 2 Bell, E.F., Warburton, development of symptomatic patent ductus arteriosus and congestive heart failure in premature infants. N. Engl. J. Med., 302, 598-604. A., Savignoni, P.G., Mendicini, M., Picece-Bucci, S. and Piccinato, L. (1972): 3 Bucci, G., Scalamandre, The systemic systolic blood pressure of newborns with low weight: A multiple regression analysis. Acta. Pediatr. Stand., Suppl. 229, l-26. 4 Cotton, R.B., Stahlman, M.T., Kovar, I. and Catterton, W.Z. (1978): Medical management of small preterm infants with symptomatic patent ductus arteriosus. J. Pediatr., 92, 467-473. D.P. and Stahlman, M.T. (1981): Early prediction of symptomatic patent 5 Cotton, R.B., Lindstrom, ductus arteriosus from perinatal risk factors: A discriminant analysis model. Acta Paediatr. Stand.. 70, 723-127. 6 Ellison. R.C., Peckham, G.J., Lang, P., Talner, N.S., Lerer, T.J., Lin, L., Dooley, K.J. and Nadas, A.S. (1983): Evaluation of the preterm infant for patent ductus arteriosus. Pediatrics, 71, 364-372. D., Kulovich, M., Kurlinski, J., Merritt, T.A. and 7 Jacob, J.. Gluck, L., DiSassa, T., Edwards, Friedman, W.F. (1980): The contribution of PDA in the neonate with severe RDS. Pediatrics, 96, 19-87. 8 Johnson, G.L., Breart, G.L., Gewitz, M.H., Brenner, J.I., Lang, P., Dooley, K.J. and Ellison, R.C. (1983): Echocardiographic characteristics of premature infants with patent ductus arteriosus. Pediatrics, 72, 864-871. J.A., Phibbs, R.H. and Tooley, W.H. (1969): Aortic blood pressure in normal newborn 9 Kitterman, infants during the first 12 hours of life. Pediatrics, 44, 959-968. J.A., Edmunds, L.H., Jr., Gregory, G.A., Heymann, M.A., Tooley. W.H. and Rudolph, 10 Kitterman,
48
11
12
13
14
A.M. (1972): Patent ductus arteriosus in premature infants. Incidence, relation to pulmonary disease and management. N. Engl. J. Med., 287,473-477. Lorenz, J.M., Kleinman, L.I., Kotagal, U.R. and Reller, M.D. (1982): Water balance in very-lowbirth-weight infants: Relationship to water and sodium intake and effect on outcome. J. Pediatr., 101, 423-432. Mahoney, L., Carnero, V., Brett, C., Heymann, M.A. and Clyman, R.I. (1982): Prophylactic indomethacin therapy for patent ductus arteriosus in very-low-birth-weight infants. N. Engl. J. Med., 306, 506-510. Siassi, B., Blanco, C., Cabal, L.A. and Coran, A.G. (1976): Incidence and clinical features of patent ductus arteriosus in very-low-birth-weight infants: A prospective analysis of 150 consecutively born infants. Pediatrics, 57, 347-351. Versmold, H.T., Kitterman, J.A., Phibbs, R.H., Gregory, G.A. and Tooley, W.H. (1981): Aortic blood pressure during the first 12 hours of life in infants with birth weights 610 to 4220 grams. Pediatrics, 67, 607-613.
Early Human Development, 12 (1985) 39-48 Elsevier
EHD 00665
Incidence and risk factors for symptomatic patent ductus arteriosus among inborn very-low-birth-weight infants Jaime A. Furzan *, Joan Reisch, Jon E. Tyson, Pennock Laird and Charles R. Rosenfeld Depariments of Pediatrics, Obstetrics and Gynecology, and Internal Medicine, The University of Texas Health Science Center at Dallas, Southwestern Medical School, Dallas, Texas, U.S.A. Accepted
for publication
1 April 1985
Summary
The incidence of symptomatic patent ductus arteriosus (sPDA) in very-low-birthweight infants has been reported almost exclusively from referral centers. Moreover, the incidence has varied considerably. We prospectively evaluated the incidence and risk factors of sPDA in a totally inborn population of infants I 1500 g birth weight (n = 120) receiving conservative fluid management and examined the role of echocardiography in predicting sPDA before clinical signs are evident. The incidence of sPDA was relatively low (16%) and the age at onset was relatively late (mean f SE. = 14.7 k 3.0 days) when compared to prior reports. Fchocardiographic changes preceded physical and radiographic findings of sPDA. Fluid volume within the range administered to our infants was unrelated to the occurrence of sPDA. However, risk factors identifiable within 24 h of birth were defined: obstetric estimate of gestational age, race, mean fluid volume intake in the initial 24 h, and early treatment with volume expanders. From these risk factors, 79% of infants with sPDA and 79% without sPDA could be predicted by stepwise discriminant function analyses. These findings may be employed in studies needed to assess the benefit of early surgical or pharmacologic therapy for infants at highest risk for sPDA. patent ductus arteriosus; very-low-birth-weight tors; volume expansion
* Dr. Furzan was a Fellow in Neonatal-Perinatal Caribobo, Venezuela. Address all correspondence to: Charles R. Rosenfeld, Boulevard, Dallas, Texas 75235, U.S.A.
037%3782/85/$03.30
infants; echocardiography;
Medicine
and
M.D., Department
0 1985 Elsevier Science Publishers
presently
is at the University
of Pediatrics,
B.V. (Biomedical
risk fac-
Division)
5323 Harry
of
Hines
40
Introduction
The incidence of symptomatic patent ductus arteriosus (sPDA) among very-lowbirth-weight infants (VLBW; _<1500 g birth weight) has been reported almost exclusively from referral institutions [4,5,7,12] and until recently, institutions in which fluids have been administered liberally [4,5]. Incidence rates have been as high as 50% but have differed considerably between centers [4,5,7,12,13]. Despite the use of uniform diagnostic criteria, the incidence varied more than 3-fold (ll-36%) between centers in a recent large collaborative study [6]. In a review of all VLBW infants born at our center between June 1978 and May 1979, we found a relatively low incidence of sPDA (16%). Moreover, the mean age at onset, 20 * 4 (S.E.) days was considerably later than would be expected from published studies. Differences between centers with respect to sPDA might relate to selection biases in the referral of patients, as well as differences in obstetric and neonatal risk factors. For these reasons, we initiated a prospective study in a large inborn population of VLBW infants to determine: 1) if the incidence of sPDA is lower and the onset later than that generally reported from other institutions; 2) if the risk factors that relate to the occurrence of sPDA identified by multivariant discriminant analysis are similar to those reported for a largely outborn population [5]; and finally, 3) if echocardiographic changes precede the clinical diagnosis of sPDA, thus facilitating identification of infants at-risk for sPDA.
Materials and Methods
Sample size and patient population 120 consecutive infants who weighed 1500 g or less at birth and lived at least 72 h were enrolled. These infants were all born at Parkland Memorial Hospital between June 1, 1979 and April 30, 1980. Our sample size was selected prior to study as an appropriate number of infants in order to distinguish major differences between infants with sPDA and those without, assuming an incidence of sPDA similar to that observed in our retrospective study, 16%. The postnatal age at entry to the study, as well as the diagnostic criteria, were selected to allow comparison of our incidence and risk factors with those identified in a referral population by Cotton et al. [5], who performed discriminant function analyses similar to those which we planned. Given the relatively late onset of sPDA observed in our retrospective study, we anticipated that a diagnosis of sPDA would be unlikely prior to 3 days. Patient care procedures 85% of mothers delivering at Parkland Memorial Hospital received prenatal care. Mothers with complications of pregnancy were hospitalized in a high-risk maternity unit. An Obstetrics faculty member remained in the hospital at all hours to supervise care in the Labor and Delivery unit. The delivery of VLBW infants was routinely attended by a senior Pediatric
41
resident. At delivery, the umbilical cord of each infant was routinely clamped without delay. During resuscitation, emphasis was placed on ensuring adequate ventilation. Sodium bicarbonate was used in limited quantities (5 2 mequiv./kg of a 0.5 mequiv./ml solution). Each infant was transferred from the delivery room to the Intensive Care Nursery, which was supervised by 1 of 2 attending neonatologists who evaluated all infants twice daily. Uniform guidelines for the administration of volume expanders and the regulation of fluid balance, infant temperature, and ventilatory therapy were employed. Administration of volume expanders (saline, whole blood or plasma) was encouraged only in the presence of both delayed capillary filling (> 3 set) and hypotension, defined from relatively conservative standards for normal blood pressure in neonates (a systolic blood pressure > 1 S.D. below the mean value reported by Bucci et al. [3] for infants of the same birth weight, gestational age, and postnatal age). Fluid intake was regulated to maintain an isotonic urine and to allow a weight loss of approximately 15% during the first week. Skin temperature of all infants was servo-controlled at 36.5”C. Infants were routinely nursed under a radiant warmer for at least 1-2 days and whenever ventilatory therapy was administered. Intubation was performed to provide continuous distending airway pressure or mechanical ventilation for infants with hyaline membrane disease who required an FiO, > 0.6, or developed hypercapnia (PaCO, > 50-60 mm Hg) or a rapid clinical deterioration. Infants with sPDA, as defined below, were treated with furosemide (1 mg/kg per day) and fluid restriction (I 100 ml/kg per day). Surgical closure was undertaken when sPDA did not improve after at least 48 h of medical therapy. Indomethacin was not used in any infant. Patient evaluation Following study entry, each infant was examined every other day by the same investigator (JAF) for the first 2 postnatal weeks and twice weekly thereafter until discharge. Symptomatic PDA was diagnosed using the criteria of Cotton et al. [4,5] so as to provide comparable data. This diagnosis required a large left-to-right shunt (i.e., at least 4 of the following signs: elevated basal heart rate (> 165 beats/min), hyperdynamic precordium, bounding pulses, characteristic murmur, progressive cardiomegaly (cardiac/thoracic ratio > 0.60) and prominent pulmonary vasculature on chest x-ray, and an abnormal echocardiogram (left atrial/aortic root ratio > 1.4, increased left ventricular and atria1 dimensions)) and signs of congestive heart failure (i.e., at least 2 of the following signs: gallop rhythm, pulmonary edema on chest x-ray, congestive hepatomegaly, increasing episodes of apnea and bradycardia unexplained by prematurity (apneic episodes > 20 set associated with heart rate < 100 bpm), and carbon dioxide retention (PaCO, > 55 mm Hg) unexplained by lung disease). Information on fluid intake, ventilator therapy, weight gain, and days of oxygen therapy were recorded daily following the infant’s physical examination. Bronchopulmonary dysplasia, necrotizing enterocolitis, and other medical problems were diagnosed according to predetermined criteria. Hyaline membrane disease was diagnosed in the presence of progressively increasing oxygen requirements (FiOz >
42
0.40) typical radiographic findings, and a negative blood culture. Acute perinatal stress was recorded in the presence of any of the following: placenta previa, placental abruption, umbilical cord prolapse, breech vaginal delivery, birth asphyxia (Apgar score of 6 or less at 1 or 5 min), or hypovolemia during the first 24 h after birth (i.e., the presence of 3 or more of the following: capillary filling time r 3 set, hypotension as defined above, severe metabolic acidosis (pH < 7.10), evidence of abnormal bleeding, anemia (central hematocrit I 40%) and poor urinary output (< 0.5 ml/kg per h)). For each infant a M-mode echocardiogram was scheduled between the 7th and 11th days of life, an age when sPDA would not be expected from our retrospective study. These studies were performed to determine whether echocardiographic findings preceded clinical findings of sPDA. All echocardiograms were read by the same person (PL) without knowledge of the infant’s clinical course. Assessment included measurements of both ventricles, left atrium, aorta, shortening fraction, left atrium to aortic root ratio, and the determination of two indices: left atrium measured in millimeters divided by birth weight in grams, and left ventricular end-diastolic dimension in millimeters divided by birth weight in grams. Statistical methods Descriptive statistics were calculated for all variables. The numerical variables were summarized using means and standard deviations, while frequency distributions were determined for the categorical variables. The sPDA group and group without sPDA were compared for each variable considered separately. (X-square contingency table analysis or Fisher’s Exact Probability Test were utilized for categorical variables, and Student’s t-tests for continuous variables. To determine whether one or more of all the variables measured was predictive of sPDA, multivariate discriminant analysis was used. The Statistical Package for the Social Sciences discriminant analysis program with Wilks’ stepwise selection method was utilized.
Results
The mean birth weight (f S.D.) of the 120 study infants was 1165 + 233 g; the mean gestational age by obstetric criteria was 29.4 f 3.1 wk. Hyaline membrane disease occurred in 33% of all infants. The mean fluid intake ( f S.D.) was 91.1 f 22.3 ml/kg per day on the first day, 131 f 20.4 during the first week, and 162 f 21.3 during the second week. Volume expanders were administered during the first 24 h after birth to 27% of infants (32/120). The incidence of sPDA was 16% (19/120); another 21% of infants (25/120) had clinical evidence of a left-to-right shunt. The incidence of sPDA was 26% (8/31) among infants < 1000 g and 12% (11/89) in infants between 1000 and 1500 g. Among infants with hyaline membrane disease sPDA occurred in 29% (9/39). The mean age at onset was 14.7 (range 3-35) days. The diagnosis was made before 7 days (‘early onset’ group) in 4 of 19 (21%) infants; 15 (79%) were diagnosed after the
43 TABLE
I
Characteristics
of the study population
Variable
No. of patients Birth weight (g) Gestational age (wks) Sex Male Female Race Negro Caucasian Latin AGA infants SGA infants ’ Acute perinatal stress Birth asphyxia 1 min Apgar 5 min Apgar Prolonged rupture membranes Hypocalcemia Hyaline membrane disease Volume expanders first 24 h Continuous distending pressure first 24 h Fluid intake first 24 h d Fluid intake first week d Fluid intake second week d lntraventricular hemorrhage Necrotizing enterocolitis Bronchopulmonary dysplasia Deaths (between 3 and 60 days) a Values presented are mean f 1 S.D. b Birth weight less than tenth percentile
according
to the presence
With symptomatic
or absence
PDA
19 1048 + 247 29.4 + 2.2
of symptomatic
Without symptomatic
PDA
101 1189+223’ 31.2 f 2.2 ’
10
45 56
9 11 8 0 14 5 16 (84.2%) 13 (68.4%) 3.9+2.6 6.2+1.9 6 (31.6%) 18 (94.7%) 9 (47.4%) 10 (52.6%)
62 23 16 61 40 70 (69.3%) 68 (67.3%) 4.9 f. 2.7 7.2 f 2.1 21 (20.8%) 77 (76.2%) 30 (29.7%) 22 (21.8%) ’
12 (63.2%) 84.3 f 12.3 130* 18.6 169 + 22.4 3 (15.8%) 0
37 (36.6%) 90.3 f 23.5 131* 20.9 152i19.5 8 (7.9%) 8 (7.9%)
4 (21.1%) 4 (21.1%)
4 (4.0%) 11 (10.9%)
for gestational
PDA a
age according
to Colorado
grid.
’ Statistically significant at P < 0.05. d Includes fluids for volume expansion.
first week (‘late onset’ group). Surgical ligation was performed in only 9 infants (7.5% of all infants; 47% of infants with sPDA). Infants with and without sPDA are compared in Table I. Those with sPDA were smaller, more preterm, had a higher incidence of placental abruption, and were more likely to have received volume expanders in the first 24 h. Borderline differences (0.10 > P > 0.05) were found for the incidence of hyaline membrane disease, hypocalcemia, and the use of continuous distending airway pressure within the first 24 h following birth. Of note, mean fluid intake in the two patient groups was virtually identical. The stepwise discriminant analysis procedure used to determine a subset of
44
variables optimal for the separation of the groups with and without sPDA yielded the following variables in the final discriminant function: obstetric estimate of gestational age (weeks), race (Negro or Caucasian), use of volume expanders (yes/no), and fluid intake in the first 24 h after birth (ml/kg per day). The discriminant score was calculated for each subject by multiplying the value of each of the variables in the function by its respective weighting coefficient and summing the resulting products. If a condition was present, it was given a value of 1 in the equation; if absent, the value was zero. If the final score was less than the cutoff value, 0.691, the infant would be predicted as not having sPDA; if the score was greater than the cutoff value, the infant would be predicted as having sPDA. Use of the discriminant function is illustrated below: Consider a Caucasian infant, with an obstetric estimate of gestational age of 32 wks, who did not receive volume expanders in the first 24 h of life, but whose fluid intake in the first 24 h was 100 ml/kg per day. This information is substituted into the final discriminant function, and a score is calculated: Class: 6.030 - 0.217 (Gestational age) + 1.343 (Volume expanders) + 1.933 (Caucasian) + 0.761 (Negro) - 0.011 (Fluid intake) Calculation of the infant’s score: Class: 6.030 - 0.217 (32) + 1.343 (0) + 1.933 (1) + 0.761 (0) - 0.011 (100) = - 0.081 Since the score is less than the cutoff value of 0.691, the infant would be predicted as not developing sPDA. Employing the discriminant function 95 of the 120 infants (79%) included in the study could be correctly predicted. The sensitivity was 79% (15/19 with sPDA were correctly predicted) and the specificity also was 79% (80/101 without sPDA were correctly predicted). Early and late onset sPDA Given the small number of affected patients, no significant differences were found between the early (n = 4) and late (n = 15) onset groups with respect to all variables shown in Table I. Only 1 of the 8 infants < 1000 g and 3 of 11 infants between 1000 and 1500 g with sPDA had an early onset. In the early onset group 2 infants died with massive intraventricular hemorrhage; the remaining 2 infants required surgical ligation at 7 and 18 days and had an uneventful recovery. The mortality in the late onset group was 13% (2/15), and was related to intraventricular hemorrhage and bronchopulmonary dysplasia. In the late onset group 7 infants (47%) required surgical ligation between 18 and 37 days. Medical& and surgically treated infants There were no significant differences between infants considered sufficiently ill to require surgical ligation and other infants with sPDA with respect to birth weight, gestational age, and echocardiographic values obtained prior to the onset of clinical signs of sPDA. Echocardiographic data Of the 120 infants enrolled into the study, 87 had an M-mode echocardiogram
TABLE
11
Echocardiographic
data according
to subsequent
No. of patients Age at echocardiogram (days) Left ventricle systolic measurement (cm) Left ventricle diastolic measurement (cm) Shortening fraction (%) Bight ventricle (cm) Aorta (cm) Left atrium (cm) Left atrium/aortic root ratio Left ventricle (mm)/birth weight (g) index Left atrium (mm)/birth weight (g) index
development
of symptomatic
PDA a
With subsequent symptomatic PDA’
Without subsequent symptomatic PDA
15 8.2 +I.0
70
0.83 kO.14 1.35 *0.21 36.7 k7.1 0.48 + 0.12 ’ 0.61+ 0.06 0.83 &-0.21 ’ 1.42kO.30 ’ 13.67 f 3.1 ’ 8.26 f 2.4 ’
8.7+1.2 0.85 +0.13 1.31 kO.15 34.8 k5.7 0.56 + 0.12 0.62 f 0.07 0.72 f 0.14 1.17*0.26 11.13+2.0 6.13k1.3
’ Values presented are mean Ifr1 S.D. ’ Includes only infants with onset of sPDA after the echocardiogram. ’ Statistically different at P < 0.05.
performed between 7 and 11 days of age; 2 were uninterpretable. The reasons for not obtaining an echocardiogram within the prescheduled period included: 7 deaths before 7 days of age from causes considered to be unrelated to sPDA; 4 infants with the onset of signs of sPDA before 7 days of age; 2 infants were in isolation due to maternal herpes or varicella; and 18 infants did not have it performed for technical reasons. To determine if echocardiographic findings preceded development of sPDA, we excluded from analysis those infants with onset of sPDA before their scheduled echocardiogram. Significant differences (P < 0.05) were found between groups and are presented in Table II. We found no significant differences between echocardiographic measurements obtained in the late onset sPDA infants and in infants who only had signs of left to right shunt without sPDA. When we compared infants with hyaline membrane disease who developed sPDA to those who did not, significantly higher values were found in the former infants for the left atrium (0.86 + 0.15 vs. 0.68 _t 0.12 cm), left atrium/aortic root ratio (1.45 + 0.29 vs. 1.10 + 0.23) left ventricle/birth weight index (13.6 * 3.9 vs. 11.0 k 2.0 mm/g) and left atrium/birth weight index (7.96 k 2.72 vs. 6.04 k 1.16 mm/g). However, infants with hyaline membrane disease who developed sPDA had values similar to those without HMD who developed sPDA.
Discussion
our
It is difficult to interpret the difference between the incidence of sPDA (16%) in study and that reported for VLBW infants in the recent collaborative study
46
(27%) [6]. In the collaborative study, a largely referral population was examined, the mean infant age at study entry was not reported, and more liberal diagnostic criteria for symptomatic PDA were used than in our study. We did not attempt, as did the investigators in the collaborative study, to identify infants affected by a ‘silent’ ductus with significant hemodynamic effects despite minimal or no physical or radiologic signs of sPDA. The reliability and accuracy with which such infants can be identified remains unclear, and their contribution to the total incidence of sPDA may not be great. Despite the difference in diagnostic criteria used in our study and in the collaborative study, comparable echocardiographic measurements were observed for infants categorized as having no significant PDA. For example, the left atrium/aortic root ratio for such infants was 1.17 k 0.26 in our study as compared to 1.13 f 0.23 in the collaborative study [8]. Our study was designed to allow comparison of our findings with those reported by Cotton et al. [5]. The lower incidence of sPDA in our center (16%) than that reported by Cotton et al. (51%) may relate to a number of variables: our entirely inborn population, the large percentage of Negros among our patients [6], or our relatively conservative use of fluid therapy [2] and volume expanders. The use of volume expanders may also relate to the lower incidence of sPDA and later onset in our study than that reported by Lorenz et al. [ll] (16 vs. 25% incidence; 21 vs. 68% of infants with sPDA whose onset occurred before day 7, respectively). Although mean birth weight and fluid intake are similar in the two studies, the infants reported by Lorenz et al. received a mean of 7-8 ml/kg of blood products and up to 30 ml/kg during the first 24 h following delivery. The presence of hypovolemia, rather than treatment with volume expanders per se, may of course increase the likelihood of sPDA. However, Arant [l] has reported that acute volume expansion in the newborn puppy may increase plasma concentration of vasodilating prostaglandins and reopen the previously closed ductus arteriosus. The use of volume expanders has not been reported in most studies of sPDA [2,4,5,10]. The administration of volume expanders has been based on standards for blood pressure measurements derived from infants with delayed clamping of the umbilical cord [9]. These standards recently have been modified to lower values [14]. It is likely that blood pressure is only loosely correlated with blood volume; thus, the administration of volume expanders in the resuscitation of normovolemic infants may inadvertently increase their risk of sPDA. Bell and coworkers have demonstrated that the administration of a high total fluid volume increases the likelihood of sPDA [2]. The increase in fluids administered to our infants after the first postnatal week at a time when use of radiant warmers and phototherapy decreased may account for the late onset of sPDA in some patients. However, from our analyses fluid volume, within the range administered in our nursery, is not an important determinant of sPDA in our population. Although mean fluid volume in the first 24 h was not significantly different between groups when analyzed by univariate analysis, it was identified as a risk factor by multivariate analyses. This reflects how single variables which by themselves do not discriminate between groups may be effective discriminators when considered with other variables.
47
A statistical model for the prediction of sPDA from clinical findings has been attempted, to our knowledge, in only 1 other study, that reported by Cotton and coworkers [5]. The perinatal risk factors identified in the two studies are quite similar. Moreover, application of their discriminant function to our study population correctly classified 73% of all infants (68% of infants with sPDA and 73% without sPDA); this would suggest that the same risk factors can be used to identify infants at highest risk for sPDA in centers with quite different patient populations and incidence of sPDA. The concept that the patent ductus arteriosus, even in the absence of clearly discernible clinical signs, is a major cause of morbidity in the first postnatal week has prompted some investigators to suggest early, aggressive therapy [7,12]. Our echocardiographic findings for infants later diagnosed as having sPDA provide evidence that significant hemodynamic effects preceded recognition of sPDA; however, the value of aggressive early pharmacologic or surgical therapy to prevent or treat PDA in VLBW infants remains to be established in properly designed, randomized trials. The influence of volume expanders on the incidence of PDA and the outcome of high-risk infants has not yet been addressed and also warrants rigorous study. Given substantial differences between centers in the incidence of sPDA, the risk factors identified in discriminant function analyses in studies like ours or that by Cotton and coworkers [5] may be useful in such studies to identify VLBW infants most likely to be affected by PDA.
References I Arant. B.S., Jr. (1981): Nonrenal factors influencing renal function during the perinatal period. Clin. Perinatol. 8, 225-240. D., Stonestreet, B.S. and Oh, W. (1980): Effect of fluid administration on the 2 Bell, E.F., Warburton, development of symptomatic patent ductus arteriosus and congestive heart failure in premature infants. N. Engl. J. Med., 302, 598-604. A., Savignoni, P.G., Mendicini, M., Picece-Bucci, S. and Piccinato, L. (1972): 3 Bucci, G., Scalamandre, The systemic systolic blood pressure of newborns with low weight: A multiple regression analysis. Acta. Pediatr. Stand., Suppl. 229, l-26. 4 Cotton, R.B., Stahlman, M.T., Kovar, I. and Catterton, W.Z. (1978): Medical management of small preterm infants with symptomatic patent ductus arteriosus. J. Pediatr., 92, 467-473. D.P. and Stahlman, M.T. (1981): Early prediction of symptomatic patent 5 Cotton, R.B., Lindstrom, ductus arteriosus from perinatal risk factors: A discriminant analysis model. Acta Paediatr. Stand.. 70, 723-127. 6 Ellison. R.C., Peckham, G.J., Lang, P., Talner, N.S., Lerer, T.J., Lin, L., Dooley, K.J. and Nadas, A.S. (1983): Evaluation of the preterm infant for patent ductus arteriosus. Pediatrics, 71, 364-372. D., Kulovich, M., Kurlinski, J., Merritt, T.A. and 7 Jacob, J.. Gluck, L., DiSassa, T., Edwards, Friedman, W.F. (1980): The contribution of PDA in the neonate with severe RDS. Pediatrics, 96, 19-87. 8 Johnson, G.L., Breart, G.L., Gewitz, M.H., Brenner, J.I., Lang, P., Dooley, K.J. and Ellison, R.C. (1983): Echocardiographic characteristics of premature infants with patent ductus arteriosus. Pediatrics, 72, 864-871. J.A., Phibbs, R.H. and Tooley, W.H. (1969): Aortic blood pressure in normal newborn 9 Kitterman, infants during the first 12 hours of life. Pediatrics, 44, 959-968. J.A., Edmunds, L.H., Jr., Gregory, G.A., Heymann, M.A., Tooley. W.H. and Rudolph, 10 Kitterman,
48
11
12
13
14
A.M. (1972): Patent ductus arteriosus in premature infants. Incidence, relation to pulmonary disease and management. N. Engl. J. Med., 287,473-477. Lorenz, J.M., Kleinman, L.I., Kotagal, U.R. and Reller, M.D. (1982): Water balance in very-lowbirth-weight infants: Relationship to water and sodium intake and effect on outcome. J. Pediatr., 101, 423-432. Mahoney, L., Carnero, V., Brett, C., Heymann, M.A. and Clyman, R.I. (1982): Prophylactic indomethacin therapy for patent ductus arteriosus in very-low-birth-weight infants. N. Engl. J. Med., 306, 506-510. Siassi, B., Blanco, C., Cabal, L.A. and Coran, A.G. (1976): Incidence and clinical features of patent ductus arteriosus in very-low-birth-weight infants: A prospective analysis of 150 consecutively born infants. Pediatrics, 57, 347-351. Versmold, H.T., Kitterman, J.A., Phibbs, R.H., Gregory, G.A. and Tooley, W.H. (1981): Aortic blood pressure during the first 12 hours of life in infants with birth weights 610 to 4220 grams. Pediatrics, 67, 607-613.