Pulmonary Hemorrhage: Clinical Course and Outcomes Among Very Low-Birth-Weight Infants

Objective  To describe the clinical course, neonatal morbidity, and neurodevelopmental outcomes of very low-birth-weight (<1500 g) children who develop pulmonary hemorrhage.

Design  A retrospective case-control study in which 58 very low-birth-weight infants who developed pulmonary hemorrhage during 1990 through 1994, of whom 29 survived, were each matched to the next admitted infant who required mechanical ventilation for respiratory distress syndrome and was of the same sex, race, and birth weight (within 250 g).

Setting  A regional tertiary neonatal intensive care unit and follow-up clinic for high-risk infants at University Hospitals of Cleveland, Cleveland, Ohio.

Main Outcome Measures  Survival, neonatal morbidity, and neurodevelopmental outcome at 20 months' corrected age.

Results  Pulmonary hemorrhage occurred in 5.7% of the total population of very low-birth-weight infants. Despite similar severity of lung disease, significantly more infants who developed pulmonary hemorrhage received surfactant therapy compared with controls (91% vs 69%, P=.005). Infants with pulmonary hemorrhage who died had a lower birth weight and gestational age compared with those who survived (766 g vs 1023 g; 25 weeks vs 28 weeks, P<.001) and more received surfactant therapy (100% vs 83%, P=.05). Survivors with pulmonary hemorrhage did not differ significantly from controls in rates of oxygen dependence at 36 weeks corrected age (52% vs 38%), grade 3 to 4 periventricular hemorrhage (28% vs 17%), or necrotizing enterocolitis (3% vs 7%), but tended to have more seizures (24% vs 3%, P=.05), periventricular leucomalacia (17% vs 0%, P=.06), and patent ductus arteriosus (79% vs 55%, P=.09). There were no significant differences in neurodevelopmental outcomes at 20 months' corrected age, (cerebral palsy, 16% vs 14%; subnormal [<70] Bayley Mental Developmental Index, 59% vs 43%; and deafness, 13% vs 10%).

Conclusion  Although mortality is high, pulmonary hemorrhage does not significantly increase the risk of later pulmonary or neurodevelopmental disabilities among those who survive.

PRIOR TO 1990, neonatal pulmonary hemorrhage occurred mainly as a terminal event among infants who were small for gestational age and asphyxiated; however, after the introduction of surfactant therapy, an increased occurrence of pulmonary hemorrhage was noted among preterm infants with respiratory distress syndrome.^1-5 Depending on birth weight and the criteria used for defining pulmonary hemorrhage, reported rates of pulmonary hemorrhage among very low-birth-weight (VLBW) infants range from 0.5% to 11%.^6-13 Currently, approximately 50% of infants who experience pulmonary hemorrhage survive; however, there is very little information on their subsequent health and neurodevelopmental outcome.^11-15

We sought to describe the clinical course and associated neonatal morbidity and later neurodevelopmental outcome of VLBW (<1500 g) children who were admitted to our neonatal intensive care unit between 1990 and 1994 and who developed pulmonary hemorrhage. We furthermore sought to identify the perinatal predictors of pulmonary hemorrhage and the correlates of death in this population.

One thousand eleven VLBW infants were admitted to the Neonatal Intensive Care Unit at Rainbow Babies & Children's Hospital, University Hospitals of Cleveland, Cleveland, Ohio, between January 1, 1990, and December 31, 1994. Eight hundred fourteen infants (81%) were born at our perinatal center and the remainder were transported from community hospitals; 477 (47%) were male, 407 (40%) were white, 218 (22%) were of multiple birth, and 40 (4%) had major congenital malformations. The mean ± SD birth weight and gestational age was 1056 ± 272 g and 28.1 ± 3 weeks, respectively. Eight hundred thirty-two infants developed respiratory distress syndrome requiring assisted ventilation, of whom 474 (57%) received surfactant therapy and 662 (80%) survived to 20 months' corrected age.

Seventy-one VLBW infants were recorded in our neonatal database as having pulmonary hemorrhage, of whom 61 were considered to have a true pulmonary hemorrhage. This was defined as a nontraumatic gush of bloody secretion from the endotracheal tube associated with clinical deterioration requiring increased ventilatory support.^16,17 Nine children had a suspected pulmonary hemorrhage, which was defined as a persistent blood-tinged secretion suctioned from the endotracheal tube, but not associated with clinical deterioration or the need for increased ventilatory assistance. One additional infant categorized as having a pulmonary hemorrhage was excluded from the study owing to incomplete documentation of the hemorrhage. Of the 61 children with a true pulmonary hemorrhage, 1 had trisomy 18 syndrome, 1 had trisomy 21 syndrome, and 1 later developed acquired immunodeficiency syndrome. These 3 children were excluded from further analysis. The study population thus included 58 children with a true pulmonary hemorrhage. These children constituted 5.7% of the total population of VLBW infants. The rates of pulmonary hemorrhage were similar between inborn and outborn infants (5.8% vs 5.6%, P=.95). Seventeen of the infants weighed less than 750 g at birth, 23 weighed between 750 and 999 g, and 18 weighed between 1000 and 1499 g. They constituted 10%, 9%, and 3% of their respective birth-weight subgroups. Twenty-nine (50%) of the 58 children with pulmonary hemorrhage survived to 20 months' corrected age.

A retrospective case-control method was used to select a comparison group of infants without pulmonary hemorrhage. Each infant with pulmonary hemorrhage was matched to the next admitted infant who had respiratory distress syndrome; required mechanical ventilation; and was of the same sex, race, birth weight within 250 g; and singleton or multiple birth. Because a matched comparison group of surviving children was required to examine the 20-month neurodevelopmental outcomes, survivors and nonsurvivors were matched separately. Control children with major congenital malformations and survivors who had not completed the 20 month follow-up were excluded from the matching process. Fifteen of the infants with pulmonary hemorrhage were of multiple birth (13 twins and 2 triplets), of whom 3 were matched to singletons owing to lack of a multiple-birth match.

The mean birth weight and gestational age of the 58 infants with pulmonary hemorrhage and of the matched controls is presented in Table 1. In both the study and control groups, 35 (60%) of the infants were male and 24 (41%) were white. Sociodemographic characteristics of the survivors as measured by maternal age and education were similar between groups. The mean ± SD age of the mothers of children surviving pulmonary hemorrhage was 25.7 ± 5 years and of the control mothers, 27.7 ± 6 years. Seven (26%) of 27 and 2 (7%) of 28 mothers of the children with pulmonary hemorrhage and controls, respectively, had less than a high school education (difference not significant).

Surfactant therapy became standard practice in our institution in December 1989 and was given as rescue therapy to infants with respiratory distress syndrome who required assisted ventilation and more than 30% inspired oxygen. Both a combination of colfosceril palmitate, cetyl alcohol, and tyloxapol (Exosurf; Burroughs Wellcome Co, Research Triangle Park, NC) and beractant (Survanta; RossLaboratories, Columbus, Ohio) were used during the study period. The combination of colfosceril palmitate, cetyl alcohol, and tyloxapol was administered in doses of 5 mL/kg every 12 hours and beractant at 4 mL/kg every 6 hours. A maximum of 4 doses were given.

Prenatal, intrapartum, and neonatal data were examined as correlates of pulmonary hemorrhage and its outcomes. Data were extracted from the medical records at the time of the infant's discharge from the hospital. Additional details specifically pertaining to pulmonary hemorrhage were later extracted from the medical records by one of us (M.T.). These included details of surfactant administration and neonatal morbidity existing prior to the pulmonary hemorrhage, the time of occurrence and clinical presentation of the hemorrhage, vital signs, and results of arterial blood gases and ventilatory settings recorded within 6 hours prior to and 1 hour after the pulmonary hemorrhage. The oxygenation index was used as a measure of the severity of lung disease. This was calculated as the mean airway pressure times the fraction of inspired oxygen, divided by the PaO₂ times 100. Blood cell counts and results of coagulation studies and blood cultures performed within 72 hours prior to and after the hemorrhage were recorded. Septicemia was defined as a positive blood culture in the presence of clinical indices of infection. Coagulopathy included 2 or more of the following: a prothrombin time of more than 20 seconds, fibrinogen less than 1.5 g/L, D-dimer more than 0.5 mg/L, and thrombocytopenia defined as a platelet count less than 100 × 10⁹/L. Multiorgan failure was defined as the presence of 3 or more of the following: respiratory insufficiency, cardiovascular instability including hypotension and poor peripheral perfusion, renal failure, intracranial hemorrhage, and metabolic abnormalities including metabolic acidosis and glucose intolerance. Intrauterine growth failure was defined as a birth weight less than 2 SDs of the mean for gestational age.¹⁸ Chronic lung disease was defined as an oxygen dependence at 36 weeks' postconceptional age.¹⁹ Additional measures of respiratory outcome included the total number of days of oxygen and ventilator dependence and the postnatal use of steroids for chronic lung disease. Periventricular hemorrhage was defined according to Papile et al.²⁰ Necrotizing enterocolitis was defined according to the criteria of Bell et al.²¹

The primary cause of death was respiratory distress syndrome and/or its complications, such as pulmonary hemorrhage or chronic lung disease, or infection, necrotizing enterocolitis, catastrophic cerebral hemorrhage, sudden infant death syndrome, or other causes.

Follow-up information to 20 months' corrected age was available from the follow-up database and included results from the Amiel-Tison neurological examination²² and the Bayley Scales of Infant Development (BSID).²³ Twenty-two children with pulmonary hemorrhage and 28 control children had a complete developmental assessment performed at 20 months' corrected age. Two of the remaining 7 children with pulmonary hemorrhage and 1 control child had been examined at 8 months' corrected age. Major neurosensory abnormalities were categorized as cerebral palsy (spastic diplegia, quadriplegia, or hemiplegia), hypotonia, hypertonia, shunt-dependent hydrocephalus, unilateral or bilateral blindness, and unilateral or bilateral deafness requiring hearing aids. The children born between 1990 and 1991 (n=19) were assessed using the original BSID, and those born between 1992 and 1994 (n=31) were assessed using the revised BSID (BSID II).²⁴ Because more of the infants were tested using BSID II, which has been reported to give lower scores than the BSID, we used a correction factor for the scores of the children born prior to 1992 tested with BSID.

All data were analyzed with the Statistical Program for Social Sciences (SPSS, Chicago, Ill). Children with pulmonary hemorrhage were compared with their controls in regard to perinatal and neonatal risk factors and outcomes, using 2-tailed t tests for continuous data and χ² analysis for categorical data. Continuous variables with skewed distribution were compared using nonparametric tests (eg, the Mann-Whitney test).

Fifty-seven of the 58 infants with pulmonary hemorrhage had respiratory distress syndrome, of whom 56 required assisted ventilation and 1 oxygen via a hood. Fifty-three (91%) of the infants were treated with surfactant. They received a mean of 2 doses of surfactant (range, 1-4), with the first dose administered at a mean ±SD age of 5 ± 5 hours. The mean oxygenation index was 12 ± 7 immediately preceding the first dose of surfactant and it decreased to 5 ± 3 before the pulmonary hemorrhage. For the infants who did not receive surfactant therapy, the initial oxygenation index was measured at 5 hours of life.

Seventeen infants (29%) presented with prodromal episodes of a frothy reddish secretion suctioned from the endotracheal tube. These were not associated with clinical deterioration and persisted for periods of a few hours to days prior to the acute pulmonary hemorrhage. Seven infants (12%) presented with extrapulmonary bleeding prior to the pulmonary hemorrhage including severe bruising, hematuria, gastric hemorrhage, and oozing around the umbilical catheter and blood puncture sites. Eleven infants (19%) had evidence of multiorgan failure prior to the hemorrhage. Four infants (7%) had septicemia (Haemophilus influenzae, Escherichia coli, Candida albicans, Staphylococcus epidermidis) and 1 had a congenital cytomegalovirus infection.

The acute episode of hemorrhage occurred at a median of 40 hours of life (range, 7-511 hours). This occurred before 48 hours of life in 35 infants (60%), between 48 and 72 hours in 12 infants (21%), and beyond 72 hours of life in 11 infants (19%) (of whom 5 presented after 7 days of life). The presenting clinical signs of the pulmonary hemorrhage, in addition to the acute gush of blood, included cyanosis in all the infants and bradycardia in 35 of 58 infants. As the condition deteriorated, gasping, apnea, or an increased work of breathing with diminished breath sounds on auscultation were noted in 12 infants. An acute drop in blood pressure was observed in 21 infants, extreme agitation in 3, and seizures in 3. All infants were being treated with antibiotics at the time of the hemorrhage. Management of the acute hemorrhage included an increase in ventilatory support and administration of parenteral fluids and medications. Epinephrine was administered via endotracheal tube to 32 infants and intravenously to 8 infants. Packed red blood cell transfusions were given to 41 infants, fresh-frozen plasma to 20 infants, and platelet transfusions to 10 infants. In addition, 17 infants received isotonic sodium chloride solution or a solution of 5% albumin, 12 received vasopressors, 29 received sodium bicarbonate, 37 received sedatives, and 15 received paralytic agents.

The ventilatory parameters prior to and within 1 hour of the pulmonary hemorrhage are summarized in Table 2. The oxygenation index increased significantly after the hemorrhage and hypoxemia and respiratory acidosis persisted despite aggressive therapy. No additional signs of extrapulmonary hemorrhage were noted after the pulmonary hemorrhage; however, coagulopathy was documented in 10 of 24 infants tested, with thrombocytopenia in 9 of 52 infants tested. Despite blood transfusions, the mean hematocrit fell from 0.39 ± 0.7 to 0.32 ± 0.7.

Among the 29 infants who survived, respiratory status gradually improved after the initial therapy and ventilatory support could be weaned to the preexisting ventilator settings within a mean of 5 days (range, 2-15 days). All surviving infants had a persistent bloody secretion suctioned from the endotracheal tube for a mean of 7 days after the acute hemorrhage (range, 2-14 days). This gradually changed from bright red to frothy pink to brown.

Infants with pulmonary hemorrhage who died (n=29) had a significantly lower birth weight and gestational age compared with those who survived (n=29) and tended to be African American (Table 3). They had more severe respiratory disease as evidenced by a significantly higher oxygenation index. More of the infants who died received surfactant (P=.05) and their hemorrhages occurred significantly sooner after the last dose of surfactant than did the hemorrhages in infants who survived. There were no differences in the type of surfactant used or response to this therapy. The combination of colfosceril palmitate, cetyl alcohol, and tyloxapol was given to 18 (75%) of 24 infants who survived and to 16 (55%) of 29 infants who died (P=.23), and beractant to 7 (29%) of 24 who survived, and 13 (45%) of 29 who died (P=.38).

A comparison of the obstetric and delivery complications of mothers of the 58 infants who developed pulmonary hemorrhage and their matched controls revealed no significant differences between groups ( Table 4). There tended, however, to be higher rates of adverse previous pregnancy outcomes and cesarean sections among the mothers of infants who developed a pulmonary hemorrhage (P=.07).

The status of the infants at birth, as evidenced by low Apgar scores and the need for endotracheal intubation in the delivery room, was similar between groups (Table 1). There was no significant difference in the initial severity of lung disease, as measured by the oxygenation index; however, significantly more infants who developed pulmonary hemorrhage received surfactant therapy as compared with controls. There was no difference between groups in the type of surfactant used. The combination of colfosceril palmitate, cetyl alcohol, and tyloxapol was given to 34 (64%) of 53 infants with pulmonary hemorrhage and 30 (75%) of 40 controls (P=.37), and beractant to 20 (38%) of 53 infants with pulmonary hemorrhage and 11 (28%) of 40 controls (P=.42); these data include 1 pulmonary hemorrhage survivor and 1 control survivor who received both the combination of colfosceril palmitate, cetyl alcohol, and tyloxapol and beractant.

When compared with their matched controls, significantly more infants with pulmonary hemorrhage who died did not survive beyond 28 days of life. For both groups, respiratory distress syndrome and its complications were the most common cause of death (Table 5). Pulmonary hemorrhage was, however, considered to be a major factor in contributing to the death of 28 of the 29 infants with pulmonary hemorrhage.

The birth weight–specific mortality at 20 months' corrected age of the total population of VLBW infants and those with and without pulmonary hemorrhage is presented in Table 6. Pulmonary hemorrhage contributed to 28 (15%) of 182 of the deaths of the total VLBW population of infants admitted to the nursery.

A comparison of neonatal morbidity between the infants who survived pulmonary hemorrhage and their matched controls is presented in Table 7. Although there was no significant difference between groups, the duration of ventilator dependence and rates of oxygen dependence at 28 days of life, patent ductus arteriosus, seizures, and periventricular leukomalacia tended to be higher in the pulmonary hemorrhage group. Of the 7 infants with pulmonary hemorrhage who had seizures, 1 had a seizure 2 days prior to the hemorrhage, 4 within 2 weeks after the hemorrhage, and 2 more than 4 weeks after the hemorrhage. The relative risk for oxygen dependence at 28 days among the infants with pulmonary hemorrhage was 3.39 (95% confidence interval, 1.0-11.4); for periventricular leukomalacia, 2.2 (95% confidence interval, 1.64-2.97), for seizures, 8.9, (95% confidence interval, 1.02-77.90), and for patent ductus arteriosus, 3.11 (95% confidence interval, 0.98-0.92).

A comparison of the neurological and developmental outcomes at 20 months' corrected age is presented in Table 8. The mean mental and psychomotor developmental indices were lower among the infants who survived pulmonary hemorrhage; however, the differences were not significant. Five of the 7 infants with pulmonary hemorrhage who had seizures also had a severe (grade 3-4) periventricular hemorrhage. All 7 children had some form of neurodevelopment impairment at 20 months' corrected age (Bayley Mental Developmental Index <70, neurological abnormality, or hearing loss).

Pulmonary hemorrhage, a rare but well-recognized complication of respiratory distress syndrome, occurred in 5.7% of the total population of VLBW infants admitted to our neonatal intensive care unit during 1990 through 1994 and contributed to 15% of VLBW deaths during this time. Significantly more infants who developed pulmonary hemorrhage had received surfactant therapy compared with matched controls, despite a similar severity of lung disease. Most were showing an improvement in their respiratory status prior to the pulmonary hemorrhage. The infants who died had a significantly lower birth weight and gestational age, more severe lung disease, and more received surfactant therapy than those who survived. The survivors, after recovery from the pulmonary hemorrhage, had a neonatal course similar to their matched controls; however, they tended to require longer ventilatory support and have higher rates of patent ductus arteriosus, seizures, and periventricular leukomalacia. There were no significant differences between groups in the rates of neurodevelopmental impairment at 20 months' corrected age.

With the exception of an abstract by Pandit et al,¹⁵ to our knowledge, this is the first report of the neonatal morbidity and early childhood outcomes of infants who survive pulmonary hemorrhage. Pandit et al noted an increased risk of chronic lung disease among infants with moderate to severe pulmonary hemorrhage; however, similar to our findings, the 2-year neurodevelopmental outcome of the survivors did not differ from a control population.

With the exception of surfactant therapy, we could not demonstrate any perinatal or neonatal risk factors associated with the occurrence of pulmonary hemorrhage. In the presurfactant era, perinatal risk factors, which were considered to predispose to the development of pulmonary hemorrhage, included breech delivery, cesarean section for preeclampsia, multiple gestation, low birth weight, intrauterine growth failure, male sex, and birth asphyxia.^1-5 Correlates of pulmonary hemorrhage since the introduction of surfactant therapy have included the severity of respiratory distress syndrome,^7,8 the type of surfactant used,⁸ the time of its administration,⁷ the presence of a patent ductus arteriosus, and maternal antibiotic therapy.^14,17,25 Owing to the retrospective nature of our study and the lack of routine echocardiograms to diagnose a patent ductus arteriosus, we could not examine the association between patent ductus arteriosus and pulmonary hemorrhage in our population. A meta-analysis of the results of the randomized clinical trials of surfactant therapy indicate a 47% increased risk of pulmonary hemorrhage for infants who received surfactant therapy; however, no association with patent ductus arteriosus was demonstrated.⁸

Pulmonary hemorrhage is considered to be a hemorrhagic edema resulting from stress failure of pulmonary capillaries associated with lung overdistension, inadequate protective surface tension, and fragility of the pulmonary capillary wall.^5,26 It is postulated that after surfactant therapy, pulmonary vascular resistance falls and that the increase in left to right shunting and lung compliance predispose to pulmonary hemorrhage.^27,28 The fact that up to 10% of infants with respiratory distress syndrome who did not receive surfactant therapy in the randomized clinical trials developed pulmonary hemorrhage and that 5 of the infants in our population developed the pulmonary hemorrhage after 7 days of life indicates that other yet-unidentified factors may also play a role in the pathogenesis of pulmonary hemorrhage.⁸

The smallest and sickest infants who developed pulmonary hemorrhage died. Thus, it is not surprising that the survivors had neonatal and longer-term outcomes similar to VLBW children who did not develop pulmonary hemorrhage. We did not examine pulmonary function in our population. However, the rate of rehospitalizations was similar between groups. During the first year of life 11 (44%) of 25 of pulmonary hemorrhage survivors on whom we had information were rehospitalized at least once, compared with 10 (34%) of 29 of the controls (P=.66). In both groups, most rehospitalizations were due to pulmonary causes such as respiratory tract infection and reactive airway disease.

A weakness in this study is the small population size, which might not have provided sufficient power to reveal statistically significant differences in outcome. The increased risk of neonatal seizures and periventricular leukomalacia suggested in our results may possibly be due to hypoxic ischemic encephalopathy, which could have occurred during the period of acute decompensation associated with the pulmonary hemorrhage. This might also explain the 8-point difference in the Bayley Mental Developmental Index between children surviving pulmonary hemorrhage and controls that, although not statistically significant, may be clinically meaningful. Despite this consideration, we conclude that although the mortality for infants who develop pulmonary hemorrhage is high, aggressive treatment is justified since the infants who survive do not suffer from excessive neonatal or longer-term morbidities.

Accepted for publication December 16, 1998.

Corresponding author: Maria Tomaszewska, MD, Division of Neonatology, Rainbow Babies & Childrens Hospital, 11100 Euclid Ave, Cleveland, OH 44106-6010.

Editor's Note: What would you suppose the long-term outcome of the surviving neonates would be?—Catherine D. DeAngelis, MD