Objective. In preterm infants, the rapid and accurate determination of the presence of a hemodynamically significant patent ductus arteriosus (PDA) is extremely important, but this is often difficult. Plasma B-type natriuretic peptide (BNP) measurement has been reported to be a helpful aid in the diagnosis of hemodynamically significant PDA in preterm infants. The aim of our study was to investigate the usefulness of a rapid BNP assay as a diagnostic marker of symptomatic PDA (sPDA) in preterm infants.
Methods. Sixty-six preterm infants, ranging from 25 to 34 gestational weeks of age, underwent clinical and echocardiographic examinations for PDA every other day from the third day of life until the disappearance of ductal flow. Blood samples were collected and plasma BNP concentrations were measured simultaneously using a commercial kit, (Triage BNP test kit; Biosite Diagnositics, San Diego, CA). When ≥2 clinically significant features of PDA were noted, and a large ductal flow was confirmed by color Doppler echocardiography, sPDA was diagnosed and treated with indomethacin.
Results. On the third day after birth, the mean BNP concentration in the sPDA group (n = 23) was significantly higher than in the control group (n = 43) (2896 ± 1627 vs 208 ± 313 pg/mL). Seventeen infants (74%) in the sPDA group became asymptomatic after an initial course of indomethacin and their BNP levels concomitantly decreased. Moreover, BNP concentrations were significantly correlated with the magnitudes of the ductal shunt, such as the ratio of left atrial to aortic root diameter and the diastolic flow velocity of the left pulmonary artery (r = 0.726 and 0.877). The area under the receiver operator characteristic curve for the detection of sPDA was high: 0.997 (95% confidence interval: 0.991–1.004). The best cutoff of BNP concentration for the diagnosis of sPDA was determined to be 1110 pg/mL (sensitivity: 100%; specificity: 95.3%).
Conclusion. In preterm infants, the circulating BNP levels correlated well with the clinical and echocardiographic assessments of PDA. Although not a stand-alone test, the rapid BNP assay provides valuable information for the detection of infants with sPDA that require treatment. Moreover, serial BNP measurements may be of value in determining the clinical course of PDA in preterm infants.
Rapidly and accurately determining the indications of therapeutic closure of a hemodynamically significant patent ductus arteriosus (hsPDA) in preterm infants is extremely important.1–3 The currently used tools, such as the clinical findings, which include heart failure and the typical echocardiographic features of hsPDA, have unsatisfactory sensitivities and/or specificities.4–8 A simple blood test that could easily, rapidly, and accurately help in diagnosing hsPDA would be of great use, especially in low birth weight infants. However, currently no accepted blood test is available to aid in the diagnosis and management of preterm infants with PDA.
Plasma B-type natriuretic peptide (BNP) measurement has been proposed as an aid in the diagnosis of hsPDA in preterm infants.9,10 Preliminary evidences have demonstrated the potential usefulness of sequential plasma BNP measurements in monitoring these infants with a rapidly changing hemodynamic status due to PDA.11,12 These 2 previous studies demonstrated the potential usefulness of sequential plasma BNP measurements in monitoring preterm infants with PDA. However, several confounders that may affect circulating BNP levels in preterm infants have not been evaluated and discussed.
However, the techniques for determining hormone levels are problematic, as are the time and amount of sample blood required for hormone analysis. The recently Food and Drug Administration-approved point-of-care, rapid whole-blood assay for BNP gives the clinician an opportunity to explore its potential usefulness. Moreover, after addition of 250 μL of whole blood to the sample port of the disposable device, the assay is completed in 10 minutes.
In our previous studies13 using this commercially available bedside test, we observed the interrelationships between hemodynamic changes of ductal shunts and BNP concentrations in healthy preterm infants without evidence of respiratory distress. We found that BNP concentrations are related to the magnitude of the ductal shunt and that they serve as an indicator of spontaneous closure of ductus arteriosus in preterm infants. Therefore, it was expected that BNP measurements might be useful when deciding on the necessity and the appropriate time for the medical and surgical management of preterm infants with hsPDA.
In the present study, to investigate whether plasma BNP in premature infants can identify hsPDA requiring treatment, plasma BNP concentrations were measured prospectively in premature infants with and without symptomatic PDA (sPDA) and compared with other echocardiographic criteria of hemodynamic significance. Moreover, we examined whether repeated plasma BNP measurements might be of value in determining the clinical course of PDA during the first 7 days of life. Finally, to investigate the diagnostic utility of a rapid BNP assay for the differentiation of sPDA, the best cutoff value of BNP was determined.
For Korean infants who had been admitted into the neonatal intensive care unit of Ansan Hospital, Korea University Medical Center, the entry criteria for this study were: (1) born at a gestational age of 25 to 34 weeks, estimated from the last menstrual period; (2) admitted within 24 hours after birth; (3) no major congenital malformation, including cardiac abnormalities except for patent foramen ovale; and (4) an echocardiographic study for PDA started within 72 hours after birth. The study protocol was approved by the local research ethics committee of Ansan Hospital, and written parental informed consent was obtained in all cases.
Eighty-five infants were initially enrolled during a 16-month period (from August 2002 to November 2003). Parental consent was refused for 5 infants, and 11 eligible infants were not studied, because the investigator (B.M.C.) was unavailable. Of 39 infants receiving surfactant, 3 infants with complications (pneumothorax, n = 2; pulmonary hemorrhage, n = 1) were excluded.
After these exclusions, 66 infants remained in the study and were eligible for evaluation. Their mean gestational age was 29.9 weeks (range: 25–34 weeks) and the mean birth weight was 1356 g (range: 625–2180 g). Thirty-one (47.0%) infants were male, 49 (74.2%) had received some antenatal steroids, and 63 (95.5%) were born at our hospital (the others were transferred after delivery).
The information collected included demographic details and the use of antenatal steroids and exogenous surfactant. Oxygen requirements, ventilatory settings, and intraarterial blood pressures were recorded at the time of each echocardiographic scan.
We prospectively evaluated infants for PDA within the first 3 days of life using a variety of diagnostic methods and administered indomethacin if PDA became clinically apparent (but before it became symptomatic).
An Acuson Cypress ultrasound machine (Acuson Corporation, Mountain Drive, CA) with a 7-MHz transducer incorporating color-flow, pulsed-wave, and continuous-wave Doppler was used. Echocardiography was performed every other day from the third day (48–72 hours) of life on preterm infants who fulfilled the entry criteria. The structural normality of the heart was established during the course of the scanning.
Initially, ductal shunt flow was visualized in a high left parasternal view using color-flow Doppler echocardiography. Shunting was graded by assessing color Doppler flow in the main pulmonary artery: (1) “mild,” if a narrow jet was detected at the pulmonary end of the ductus by color Doppler echocardiography without disturbed flow at the level of the pulmonary valves; (2) “moderate,” if a jet was wider and passed well into the pulmonary artery but not quite into the pulmonary valve; and (3) “large,” if a jet was very wide, reached the pulmonary valve, and swirled down into the pulmonary artery.
To estimate the magnitude of left-to-right shunting across PDA, 2 representative echocardiographic markers were evaluated. The ratio of the left atrial to aortic root diameter (LA/AO) was determined in the parasternal long-axis view at the level of the aortic valve using a standard method.14 Three measurements were taken and averaged.
To obtain the diastolic flow velocity of the left pulmonary artery (DFLPA), a pulsed Doppler sample volume was placed on the left pulmonary artery, just after the bifurcation of the main pulmonary artery. The pulsed Doppler flow pattern of the left pulmonary artery was then recorded and the DFLPA was measured from the screen. The DFLPA was estimated from the maximum velocity during the diastolic period.15
When ductal flow was confirmed to have disappeared by color Doppler echocardiography, no subsequent echocardiographic examination was performed. Throughout the study, echocardiography was performed by 1 investigator only (B.M.C.) to avoid any interobserver variability. Results were recorded on videotape for subsequent analysis. Hemodynamic variables were analyzed in terms of their possible correlation with BNP levels.
Diagnosis and Management of PDA
Clinical diagnoses of sPDA were made by individual attending pediatricians based on: (1) the presence of a systolic or continuous murmur; (2) a bounding pulse or a hyperactive precordial pulsation; (3) a difficulty in maintaining blood pressure; (4) a worsening ventilatory status; and (5) chest radiographic evidence, ie, pulmonary congestion or cardiomegaly (a cardiothoracic ratio > 60%) with increased pulmonary flow. We defined sPDA as the presence of 2 of these 5 signs with a confirmation of a large left-to-right ductal flow by color-flow Doppler echocardiography.
After confirming the diagnosis of sPDA, indomethacin was administered intravenously if not contraindicated. The initial dose was 0.2 mg/kg followed by 2 additional doses of 0.2 mg/kg at intervals of 12 hours. A subsequent clinical and echocardiographic assessment of sPDA was performed 48 hours after the initial course of indomethacin. If the diagnostic criteria for sPDA were no longer fulfilled, no additional indomethacin was administered and the infant was observed closely. When the criteria for sPDA were fulfilled again, an additional second course of indomethacin was administered.
In our unit, total fluid intakes of preterm infants were increased in a stepwise manner: 60 mL/kg per day on day 1, 80 mL/kg per day on day 3, and 100 mL/kg per day on day 7. However, after diagnosis of sPDA had been confirmed, fluid intakes were restricted to low maintenance to prevent congestive heart failure and worsening pulmonary edema.
Measurement of Plasma BNP
At the time of performing the Doppler echocardiographic measurements, blood samples for the measurement of BNP were collected by radial artery or umbilical artery catheter aspiration. BNP concentrations were detected using a commercial kit, (Triage BNP test; Biosite Diagnositics, San Diego, CA), by fluorescence immunoassay. The measurable range of BNP concentrations by the Triage assay was 5 to 1300 pg/mL (from August 2002 to December 2002) or 5 to 5000 pg/mL (from January 2003 to November 2003).
Categorical data are presented as numbers (%), and continuous data as mean ± SD. We used the χ2 or Fisher's exact test to compare categorical variables and the Student's t test, the paired t test, or the Mann-Whitney's rank sum U test to compare continuous variables. Correlations between BNP concentrations and hemodynamic variables were calculated by Spearman rank correlation analysis. Receiver operator characteristic (ROC) analysis was performed to determine the best cutoff values for BNP to detect sPDA as defined above and to assess the discrimination ability (by calculating the area under the ROC curve) of BNP to recognize the preterm infants with sPDA. Data were analyzed using SPSS 10.0.7 for Windows (SPSS Inc, Chicago, IL). A P value of <.05 was considered significant.
Characteristics of Preterm Infants
Of the 66 enrolled infants, 43 infants did not develop sPDA and the ductus closed spontaneously within 5 days of life (control group, mean gestational age: 30.3 weeks; mean birth weight: 1396 g). The remaining 23 infants developed sPDA, which required treatment with indomethacin within 3 days of life (sPDA group, mean gestational age: 29.1 weeks; mean birth weight: 1283 g). The characteristics of these 2 groups are summarized in Table 1. No significant differences were observed in terms of gestational age, birth weight, gender, delivery type, or the use of antenatal steroid. However, Apgar scores at 1 and 5 minutes, the use of surfactant, and the use of dopamine were statistically different in these 2 groups. In the control group, echocardiographic evaluations and blood samples for BNP were taken at 56.7 ± 11.8 hours of life, compared with 52.4 ± 14.4 hours in the sPDA group. Age at the first evaluation of the control group was similar to that of the sPDA group.
At the first echocardiographic evaluation (3 days after birth), only 14 infants in the control group had a left-to-right ductal shunt. By the color-flow Doppler method, infants with a large ductal shunt were more frequent in the sPDA group (23 of 23) than in the control group (7 of 43 [16.3%]).
BNP Levels in the Control Group
On the third day after birth, only 14 infants (32.5%) in the control group had a left-to-right ductal shunt but did not develop sPDA (asymptomatic PDA [asPDA] group) (Fig 1). The mean BNP concentration of asPDA infants was significantly higher than that of closed ductus arteriosus infants (469 ± 443 vs 82 ± 73 pg/mL, respectively; P < .001), and BNP fell to the adult levels after spontaneous ductal closure 5 days after birth (32 ± 24 pg/mL; P = .003).
BNP Levels in the sPDA Group
On the third day after birth, the mean BNP concentration for those who developed clinically significant PDA (sPDA group; Fig 1) was significantly higher than that of the control group (2896 ± 1627 vs 208 ± 313 pg/mL; P < .001). On the fifth day after birth, 17 infants in the sPDA group became asymptomatic after the administration of 3 doses of indomethacin. The ductus of 7 of these infants was particularly closed by echocardiographic examination. There was a statistical difference in BNP concentrations before and after indomethacin administration in these 17 infants (2807 ± 1524 vs 585 ± 714 pg/mL, respectively; P < .001). In the other 10 infants with asPDA, who had a reduced ductal shunt at posttreatment measurement, ductal closure was confirmed 2 days later.
On the other hand, in the remaining 6 infants in the sPDA group, PDA symptoms continued and no statistical difference was observed in BNP concentrations before and after indomethacin administration (3150 ± 2027 vs 2747 ± 1814 pg/mL, respectively). These 6 infants were treated with a secondary course of indomethacin treatment. Two days later, 4 of these infants became asymptomatic, and their mean BNP concentration was reduced significantly to 141 ± 58 pg/mL (P = .004). The ductus of 2 infants did not close after the second indomethacin administration, and no significant difference was found in their mean BNP concentration.
Correlation Between BNP Level and the Magnitudes of Ductal Shunt by Echocardiography
At the first echocardiographic evaluation, a significant correlation was found between BNP concentrations and the LA/AO ratio (r = 0.726; P < .001; Fig 2). In addition, BNP concentration and DFLPA, as determined using the pulsed-wave Doppler method, were found to be correlated (r = 0.877; P < .001; Fig 3).
Association Between BNP Levels and a Diagnosis of sPDA
The area under the ROC curve for the detection of sPDA was high: 0.997 (95% confidence interval: 0.991–1.004; P < .001; Fig 4). The best cutoff BNP concentration for the diagnosis of sPDA was found to be 1110 pg/mL. At this cutoff level the sensitivity was 100% (23 of 23), the specificity was 95.3% (41 of 43), and positive predictive value was 92.0% (23 of 25).
BNP is one of a family of structurally similar peptide hormones that also includes atrial natriuretic peptide and C-type natriuretic peptide.16,17 BNP, which is produced by the cleavage of a precursor protein into BNP and the biologically inactive peptide N-terminal precursor protein, causes natriuresis, diuresis, vasodilatation, and smooth muscle relaxation.18,19 Unlike the atrial natriuretic peptide, which is stored in the cardiac atria and ventricles, the cardiac ventricles are the major source of plasma BNP, suggesting that BNP may be a more sensitive and a more specific indicator of ventricular disorders than other natriuretic peptides. Moreover, BNP release seems to be in direct proportion to ventricular volume expansion and pressure overload.20,21
Plasma concentrations of BNP increase in various pathologic states, particularly those involved in increased cardiac chamber wall stretch and expanded fluid volume (eg, in cases of heart failure, renal failure, or primary hyperaldosteronism), or reduced peptide clearance (eg, in case of renal failure). BNP seems to have clinical utility in terms of excluding the diagnosis of heart failure in patients with symptoms of breathlessness or fluid retention and may provide prognostic information about those with heart failure or other cardiac diseases.22–27 Also, there is some evidence that it may be useful for monitoring heart failure therapies.28,29
Yet, little data are available to suggest that BNP may be useful in a pediatric clinical setting. In pediatric clinical practice, BNP measurement has been suggested to be a means of determining the severity of congenital heart disease. Ootaki et al30 reported that BNP levels in plasma correlate well with biventricular volume, particularly with left ventricular volume in various congenital heart disease such as ventricular septal defect, atrial septal defect, and tetralogy of Fallot and several cyanotic heart diseases. Suda et al31 reported that plasma BNP reflects pressure and volume loading of the pulmonary artery and the right ventricle and suggested that BNP determinations may help to identify children with ventricular septal defect complicated by pulmonary hypertension, which demands early intervention. These results suggest that the measurement of plasma BNP may add clinically useful information relevant to the management of children with congenital heart disease.
In healthy newborn infants, plasma BNP concentrations are relatively high and vary greatly over the first few days. However, they reduce rapidly during the first week of life, suggesting that BNP has a physiologic regulatory role in the cardiovascular hemodynamic changes that occur during the postnatal period.32,33 Moreover, perinatal circulatory changes lead to an increase in ventricular volume and pressure load, which may stimulate BNP synthesis and secretion in the ventricle and increase circulatory BNP concentrations shortly after birth. Such increased levels of plasma BNP may act to alleviate the increased ventricular load after birth and may also support the heart function with a decreased preload in the first days of life. This may be because a raised circulatory BNP increases sodium and water excretion by the kidneys, suppresses renin and aldosterone secretion, and leads to venous and arterial dilatation.
Plasma BNP concentrations in preterm infants are higher than in healthy term infants for the first few days after birth.11,12 In particular, when PDA is present in preterm infants, the BNP concentration is abnormally high and reflects the magnitude of shunting through the PDA. Holmstrom et al11 suggested that the magnitude of shunting through the PDA is a major determinant of plasma BNP in premature infants, based on clinical and echocardiographic assessments of shunt severity. In our previous studies,13 plasma BNP levels in healthy preterm infants with a ductal shunt were higher than in preterm infants without ductal shunt 24 hours after birth and then significantly decreased within 72 hours after birth according to the reduction in ducal flow.
The present study also demonstrates that plasma BNP levels in premature infants are strongly related to the hemodynamic influence of the PDA. On the third day after birth, the mean BNP concentration of the sPDA group, members of which developed clinically significant PDA, was significantly higher than that of the control group. These findings are consistent with a previous report12 and suggest that BNP is a useful diagnostic marker of sPDA requiring treatment.
With the increasing use of the recently approved rapid BNP assay, it is important to determine diagnostic cutoff values. We found that a BNP value > 1110 pg/mL on the third day after birth strongly predicted sPDA and thus indicates the need for intervention. Because circulating BNP is influenced considerably by several factors such as demographic factors, clinical situations, medical managements, and hemodynamic effects, large-scale randomized, clinical trials are necessary if we are to definitively determine the cutoff value in sPDA that best indicates the need for treatment in preterm infants.
In addition, the present study indicates that serial BNP measurements may add clinically useful information to the management of preterm infants with hsPDA. Plasma BNP levels in the sPDA group were found to be significantly lower after pharmacologic PDA closure, which concurs with the finding of a study by Holmstrom et al,11 who demonstrated comparable BNP changes after pharmacologic and surgical closure. Our results show a distinct fall of BNP, indicating hemodynamic improvement in hsPDA. Repeated BNP measurement may offer a simple and objective tool for following these preterm infants.
In the clinical management of preterm infants with hsPDA, it is important to quantify the shunt size and identify patients needing intervention. However, there is no generally accepted technique based on objective numerical values for the determination of ductal flow. The present study indicates that plasma BNP can reflect left-to-right ductal flow volume and may be helpful in identifying preterm infants with hsPDA that require early intervention. In fact, BNP was found to be significantly and positively correlated with the LA/AO ratio and DFLPA, which reflects the magnitude of ductal shunt.
Left atrial dilatation can be semiquantified by comparing its diameter with the diameter of the aortic root. Although the degree of left atrial enlargement depends on a number of factors in addition to the size of the ductal shunt (eg, the size of the foramen ovale, the state of hydration, and the presence of mitral regurgitation), it has been suggested that an LA/AO ratio of >1.4 signifies a large left-to-right ductal shunt.7,34 According to this suggestion, the best cutoff of BNP concentration for the detection of hsPDA (large ductal shunt) is 1295 pg/mL; using this cutoff, the sensitivity was 95.2% (20 of 21) and the specificity was 93.3% (42 of 45).
In the branch pulmonary arteries there is an abnormally high antegrade diastolic flow as the continuous stream of blood from the aorta pours into them via the ductus. In particular, DFLPA is considered to reflect the left-to-right ductal flow volume more directly than other echocardiographic methods. Suzumura et al15 reported that a DFLPA of >30 cm/second, exceeding the mean control value by + 2 SD on the third day after birth, suggests a large shunt. According to this result, the best cutoff of BNP concentration for the detection of hsPDA is 1110 pg/mL; using this cutoff, the sensitivity was 88.0% (22 of 25) and the specificity was 97.6% (40 of 41).
Echocardiography is the main method used to detect PDA in newborns, and several markers of hsPDA have been identified using this method, including the LA/AO ratio, DFLPA, ductal size, left ventricular output and stroke volume, and reversed diastolic flow of the descending aorta. Echocardiography is required to confidently diagnose PDA, but the hemodynamic effects of PDA may be difficult to determine even with these typical echocardiographic features; also, the course of PDA cannot be reliably anticipated.7,8 Based on the results of the present study, it is believed that BNP measurement can be used as an adjunct to echocardiography in PDA. Although it is not a stand-alone test, the BNP assay provides valuable information quickly, concerning the detection of an infant requiring intervention for PDA, and for determining the clinical course of PDA in a preterm infant.
Overall, in this study, circulating BNP levels were found to correlate well with the clinical and echocardiographic assessments of PDA in preterm infants, and changes in BNP levels were considered to indirectly represent ductal shunt changes. Particularly, a cutoff BNP level of 1110 pg/mL differentiated well between preterm infants with and without sPDA. Therefore, it is expected that the plasma BNP level is a useful diagnostic marker of sPDA in premature infants that may require treatment. Moreover, serial plasma BNP measurements may be of value in determining the clinical course of PDA in preterm infants.
The rapid assay for BNP measurement in blood seems to be a sensitive and specific test for differentiating preterm infants with and without sPDA in the neonatal intensive care setting. If additional studies validate these exploratory findings, it is possible that BNP will prove to be an additional cost-effective diagnostic armamentarium of attending pediatricians in neonatal intensive care units.
We thank the doctors and nurses at the neonatal intensive care unit, Department of Pediatrics, Ansan Hospital, Korea University Medical Center, for their enthusiastic support and cooperation.
- Accepted November 8, 2004.
- Reprint requests to (C.S.S.) Department of Pediatrics, Anam Hospital, Korea University Medical Center, #126-1, 5-Ga, Anam-Dong, Seongbuk-Gu, Seoul, 136-705, Korea. E-mail:
No conflict of interest declared.
- ↵Fowlie PW, Davis PG. Prophylactic indomethacin for preterm infants: a systematic review and meta-analysis. Arch Dis Child Fetal Neonatal Ed.2003;88 :F464– F466
- Urquhart DS, Nicholl RM. How good is clinical examination at detecting a significant patent ductus arteriosus in the preterm neonate? Arch Dis Child.2003;88 :85– 86
- ↵Lee JY, Seo WH, Choi BM, et al. Changes of plasma brain natriuretic peptide according to the changes of ductal shunt in healthy preterm infants. J Korean Pediatr Cardiol.2003;7 :128– 137
- ↵Sahn DJ, DeMaria A, Kisslo J, Weyman A. Recommendations regarding quantitation in M-mode echocardiography: results of a survey of echocardiographic measurements. Circulation.1978;58 :1072– 1083
- ↵Yasue H, Yoshimura M, Sumida H, et al. Localization and mechanism of secretion of B-type natriuretic peptide in comparison with those of A-type natriuretic peptide in normal subjects and patients with heart failure. Circulation.1994;90 :195– 203
- ↵Nakagawa O, Ogawa Y, Itoh H, et al. Rapid transcriptional activation and early mRNA turnover of brain natriuretic peptide in cardiocyte hypertrophy. Evidence for brain natriuretic peptide as an “emergency” cardiac hormone against ventricular overload. J Clin Invest.1995;96 :1280– 1287
- ↵Tsutamoto T, Wada A, Maeda K, et al. Attenuation of compensation of endogenous cardiac natriuretic peptide system in chronic heart failure: prognostic role of plasma brain natriuretic peptide concentration in patients with chronic symptomatic left ventricular dysfunction. Circulation.1997;96 :509– 516
- McClure SJ, Caruana L, Davie AP, Goldthorp S, McMurray JJ. Cohort study of plasma natriuretic peptides for identifying left ventricular systolic dysfunction in primary care. BMJ.1998;317 :516– 519
- Nagaya N, Nishikimi T, Uematsu M, et al. Plasma brain natriuretic peptide as a prognostic indicator in patients with primary pulmonary hypertension. Circulation.2000;102 :865– 870
- ↵Yoshibayashi M, Kamiya T, Saito Y, et al. Plasma brain natriuretic peptide concentrations in healthy children from birth to adolescence: marked and rapid increase after birth. Eur J Endocrinol.1995;133 :207– 209
- ↵Koch A, Singer H. Normal values of B type natriuretic peptide in infants, children, and adolescents. Heart.2003;89 :875– 878
- ↵Silverman NH, Lewis AB, Heymann MA, Rudolph AM. Echocardiographic assessment of ductus arteriosus shunt in premature infants. Circulation.1974;50 :821– 825
- Copyright © 2005 by the American Academy of Pediatrics