|Home | About | Journals | Submit | Contact Us | Français|
Pulmonary disease among infants of <28 weeks' gestation (extremely low gestational age newborns) often has the following pattern: the infant starts out with little need for supplemental oxygen and ventilatory support in the first postnatal week but then has pulmonary deterioration in the second postnatal week, with an increased need for supplemental oxygen and respiratory support. We evaluated the antecedents and correlates of patterns of early lung disease, with particular emphasis on pulmonary deterioration, in a large cohort study (the Extremely Low Gestational Age Newborn [ELGAN] study).
We examined data collected prospectively on 1340 infants born between 2002 and 2004 at 23 to 27 completed weeks of gestation and who survived to 14 days. Pulmonary deterioration was defined as receipt of fraction of inspired oxygen <0.23 on any day between days 3 and 7 and receipt of fraction of inspired oxygen ≥ 0.25 on day 14.
One fifth (20%) of the infants had consistently low fraction of inspired oxygen, approximately two fifths (38%) had pulmonary deterioration, and the remaining approximately two fifths (43%) had consistently high fraction of inspired oxygen (early and persistent lung dysfunction). Compared with infants who had consistently low fraction of inspired oxygen, infants who experienced pulmonary deterioration had lower gestational ages and lower birth weights, had higher scores for neonatal acute physiology, and received more intensive modes of respiratory support. Gender, multifetal pregnancy, cesarean delivery, antenatal steroids, chorioamnionitis, and funisitis were not associated with pulmonary deterioration. The incidence of chronic lung disease, defined as oxygen therapy at 36 weeks' postmenstrual age, was 17% in the consistently low fraction of inspired oxygen group, 51% in the pulmonary deterioration group, and 67% in the early and persistent pulmonary dysfunction group. The incidence of death in these 3 groups before 36 weeks' postmenstrual age was 1%, 3%, and 5%, respectively.
Nearly 40% of extremely low gestational age newborns experience pulmonary deterioration in the first 2 postnatal weeks, and half of these infants develop chronic lung disease. Indicators of developmental immaturity and illness severity were associated with both pulmonary deterioration and chronic lung disease. Studying the antecedents of pulmonary deterioration might provide new insights about chronic lung disease pathogenesis.
Nearly all infants born before 28 weeks' gestation (extremely low gestational age newborns [ELGANs]) require ventilatory assistance immediately after birth; most often with mechanical ventilation and supplemental oxygen.1 During the first 2 postnatal weeks, several distinct patterns of lung disease emerge.2–6 Some ELGANs have relatively little lung disease and rapidly recover, whereas others have early and persistent pulmonary dysfunction (EPPD) characterized by the need for mechanical ventilation and high concentrations of supplemental oxygen during the weeks after birth. A third group has resolution of their initial lung disease during the first week when mechanical ventilation may no longer be necessary and the need for oxygen is markedly reduced, only to deteriorate during the second week resulting in the need for increased supplemental oxygen and resumption of mechanical ventilation. We describe this group as having pulmonary deterioration (PD).
Rarely do infants with PD die; however, approximately one half develop chronic lung disease (CLD), also known as bronchopulmonary dysplasia.7 Therefore, PD seems to be an important early identifier of infants likely to develop CLD. The antecedents of PD have not yet been identified, although late surfactant deficiency might be a contributing factor.8 Bacteremia, pneumonia, and patent ductus arteriosus (PDA) are sometimes suspected as the cause of PD, but most often no cause is identified. Because PD is associated with CLD, acquiring a better understanding of PD might provide important insights about CLD.
A major advantage to focusing on PD, rather than the more established outcome, CLD, is that PD can be identified early in life. The events that may confound our understanding of PD are fewer in number than those that occur between birth and the diagnosis of CLD at 36 weeks' postmenstrual age. Finally, it is possible that risk factors for PD might be modified or mitigated, and that reducing exposure to these risk factors might reduce the likelihood of CLD. Therefore, we sought to identify the antecedents and modifiers of 3 patterns of early lung disease, with particular emphasis on PD, in a large cohort of ELGANs.
The study population included infants enrolled in a multicenter epidemiologic study to identify characteristics and exposures that increase the risk of structural and functional neurologic disorders in ELGANs (the ELGAN Study).9 From March 2002 to August 2004, women delivering before the 28th week of gestation at 1 of 14 participating institutions were asked to enroll in the study. Individual institutional review boards approved the enrollment and consent processes. Of 1506 infants enrolled, 1340 survived the first 2 postnatal weeks and were thus eligible for the analyses presented here. Historical, demographic, and clinical data were abstracted from maternal and infant charts, and histologic and microbiologic studies of the placentas were performed to identify evidence of inflammation and infection.
ELGANs were classified into 3 mutually exclusive groups: those with consistently low fraction of inspired oxygen (Fio2) (an Fio2 consistently <0.23 on all days between 3 and 7 postnatal days and receiving Fio2 ≤ 0.25 on day 14), those with PD (an Fio2 < 0.23 on any days between 3 and 7 days and receiving Fio2 > 0.25 on day 14), and those with EPPD (an Fio2 consistently ≥ 0.23 on all days between 3 and 7 postnatal days and receiving Fio2 > 0.25 on day 14). For individual infants, the Fio2 assigned for each day of life was the mode Fio2 for that day. Day 0 began at the time of birth and continued through 24 hours of age. Day 1 began at the end of the first 24 hours of age and ended at midnight of that calendar day. Thus, day 1 varies in length. After day 1, all days began and ended at midnight.
We examined candidate risk factors for each respiratory outcome defined above including gender, gestational age, birth weight z score (based on an external standard10), maternal race, receipt of antenatal steroids, chorioamnionitis, funisitis, pregnancy complications, and multifetal pregnancy. Gestational age was determined by an a priori hierarchy based in order of decreasing preference on: fetal ultrasound before 14 weeks (62%), ultrasound between 14 and 18 weeks (17%), ultrasound after 18 weeks with a consonant maternal report of the beginning of the last menstrual cycle (13%), maternal report of the beginning of the last menstrual cycle only (7%), and gestational age recorded in the log of the NICU (1%).
The presence of chorioamnionitis (defined by polymorphonuclear leukocytes in the chorion or chorioamnion) and funisitis (defined by polymorphonuclear leukocytes in the wall of a blood vessel in the umbilical cord or the chorionic plate) was determined by a pathologist at each institution who first engaged in training procedures to minimize interobserver variability, was masked to maternal history, and used predefined operational definitions.10,11
To further identify evidence of infection, we cultured chorion/decidua biopsies of the placenta. Immediately after delivery, the chorioamnionic membranes were removed from an area on the fetal surface of the placenta. Specimens of placenta were then collected under sterile conditions, flash-frozen, stored locally −80°C, and later shipped to a central laboratory where they were processed in a standardized manner.12,13
We examined bacteriologic information from the infant-placenta dyads according to mode of delivery, first, considering all deliveries, and second, limiting analyses to infants born by cesarean section. The latter approach has the disadvantage that it excludes a moderate number of deliveries initiated by factors associated with intrauterine infection (eg, preterm labor and rupture of membranes). On the other hand, it has the advantage of excluding placentas contaminated during passage through the vagina.
We examined the occurrence during the first 2 postnatal weeks of the following factors: Score for Neonatal Acute Physiology II (SNAP-II),14 mode of ventilation, diagnosis and treatment of PDA, diagnosis of bacteremia, and medication use. Mode of ventilation was defined as the highest level of support on each day and ranged from no support, supplemental oxygen by hood or nasal cannula, nasal continuous positive airway pressure, and conventional mechanical ventilation to high frequency ventilation. We did not record the devices used. Presumed bacteremia was defined as an episode of suspected bacteremia during which the infants were treated with antibiotics for more than 72 hours, but the blood culture was negative. Confirmed bacteremia required a positive blood culture. The diagnosis of PDA was assigned by clinicians without uniformity of definition; echocardiographic confirmation was not required. CLD was defined as oxygen therapy at 36 weeks' adjusted gestation. Medications were recorded if given on any day during the first week and included surfactant, analgesics (ie, morphine, fentanyl, or methadone), sedatives (ie, lorazepam, midazolam, or chloral hydrate), vitamin A, and steroids (ie, hydrocortisone and dexamethasone). We did not collect data about the dosages.
Infants in the EPPD and PD groups were compared in univariate analyses to those in the low Fio2 group to identify the characteristics that distinguished them. Characteristics that distinguished infants who developed PD from infants who remained in the low Fio2 group with a P value of ≤ .30 were then included in logistic regression analyses to assess the strength of association of each characteristic/exposure to the risk of PD, adjusted for other factors included in the regression. We also created separate logistic regression models of the risk of EPPD based on characteristics and exposures that distinguished children who developed EPPD from infants who remained in the low Fio2 group.
Gestational age and birth weight z score15,16 were included in every multivariable model. Multivariate logistic regression models were developed using manual backward selection, dropping the least significant variable followed by rerunning the model with the remaining variables. A model was complete when all of the variables that remained were associated with the outcome at a level of P < .1. The dropped variables were individually reintroduced to see if they now contributed to the complete model.
The cohort comprised 1340 infants: 249 (20%) infants had consistently low Fio2, 484 (38%) infants had PD, and 576 (43%) infants had EPPD (Fig 1).
Compared with the consistently low Fio2 group, infants in the PD group tended to have a lower gestational age, lower birth weight, and a higher SNAP-II (Table 1). Compared with the EPPD group, infants in the PD group had a higher gestational age, higher birth weight, and a lower SNAP-II. Gender, multifetal pregnancy, antenatal steroids, initiator of delivery, cesarean delivery, and histologic indicators of inflammation did not distinguish one group from another.
In samples from the entire cohort and from those delivered by cesarean section, no one organism was identified as having a statistically significant association with pulmonary outcomes. Mycoplasma species were recovered most frequently from the placentas of infants in the EPPD group, and least frequently from the placentas of infants in the consistently low Fio2 group (Table 2). No other species showed this gradient. Compared with infants whose placenta did not harbor a vaginal organism, infants whose placenta did harbor a vaginal organism were at modestly greater risk of PD (45% vs 36%).
With increasingly severe and persistent respiratory dysfunction (from the low Fio2 group, to the PD group, to the EPPD group), the frequency of undesirable characteristics/exposures increased, including high SNAP-IIs, presumed bacteremia, confirmed bacteremia, echocardiogram-documented PDA, medical and surgical treatment of PDA, and high frequency ventilation on days 0 and 7 (Table 3). The frequency of receipt of the following medications also increased: surfactant, hydrocortisone, analgesics, sedatives, and vitamin A (Table 3).
In addition, the incidence of CLD was 17% (41/243) among children in the low Fio2 group, 51% (238/469) among children in the PD group, and 67% (358/535) among children in the EPPD group. Among all infants with CLD, 37% had an antecedent history of PD. On day 0 through day 7, the frequency of mechanical ventilation was lowest in the low Fio2 group, and highest in the EPPD group (Fig 2).
In multivariate analysis comparing the PD group to the low Fio2 group, low gestational age, low birth weight z score, and mode of ventilation on postnatal day 7 were associated with increase risk of PD (Table 4). The increased risk associated with a diagnosis of PDA approached, but did not reach, nominal statistical significance (P = .07).
When comparing the EPPD group to the low Fio2 group, low gestational age, low birth weight z score, mechanical ventilation on day 7, confirmed bacteremia, receipt of vitamin A and surfactant, and treatment of PDA were associated with increased risk of EPPD (Table 5). The P value associated with a diagnosis of PDA nearly achieved nominal statistical significance (P = .06). Histologic chorioamnionitis and funisitis were not associated with PD or EPPD in these models (data not presented).
The importance of EPPD and PD is that infants who have these respiratory patterns are at very high risk of CLD. Early descriptions of CLD noted its association with severe early lung disease that would be categorized as EPPD in our study. More recently, an additional pattern of early disease preceding CLD has been reported. Among infants who developed CLD, Charafeddine et al2 observed 2 groups of infants who did not have antecedent EPPD. Infants in these groups were either well throughout the first week of life or had recovered from their acute lung disease by the end of the first postnatal week, but then developed an oxygen requirement during the second postnatal week. The authors were the first to describe this experience as “atypical CLD.”2 These infants experienced what we call PD.
Streubel et al6 identified a group of infants who had an Fio2 of <0.25 for 3 days after recovery from respiratory distress syndrome (RDS) and a subsequent increase in Fio2 to at least 40%; these infants were classified as having atypical CLD. All infants in this atypical CLD group developed bronchopulmonary dysplasia (BPD). Compared with infants with no BPD, infants with atypical CLD were smaller, less mature, and had a higher rate of sepsis and pneumothorax. In a similar study, Panickar et al3 reported that ~40% of all who developed BPD had an antecedent history of PD. We confirmed this relationship, finding that >50% of children with EPPD or PD developed CLD compared with <20% of children in the low Fio2 group.
The terms in current use in the literature are confusing. We feel that our terms (PD and EPPD) describe a clinical pattern of respiratory disease, defined by Fio2, which may or may not develop into CLD. Previous authors identified patients as having CLD (usually at 36 weeks' adjusted gestation), and then looked retrospectively to identify the pattern of lung disease (eg, “atypical RDS”2). We do not believe that infants with PD or EPPD necessarily have RDS. The advantage of using our terms is threefold: first, we, and other investigators, will be able to relate events that occur after the pattern of lung disease to these patterns (eg, CLD and, in future studies, neurodevelopmental outcomes); second, a clinician is now able to distinguish these patterns by the 14th postnatal day and place infants into appropriate risk categories of morbidities associated with prematurity (namely, CLD); and third, therapeutic modalities designed to reduce the incidence of morbidities associated with prematurity (eg, CLD) can use these patterns as markers of risk.
In our cohort of ELGANs, PD occurred in 38% of infants, a prevalence similar to that of EPPD (43%). Only 19% of ELGANs were in the consistently low Fio2 group, having little or no evidence of lung dysfunction throughout the first 2 postnatal weeks. We found that infants with PD had a risk factor profile that was intermediate between that of infants with consistently low Fio2 and infants with EPPD. This risk factor profile included birth weight, gestational age, and a variety of treatments and complications that are associated with low gestational age, such as higher SNAP-II values, treatment with mechanical ventilation, patent ductus arteriosus, treatment with hydrocortisone, and confirmed bacteremia.
To our knowledge, the entity of PD is unique to ELGANs. The peculiar predisposition of ELGANs to PD could be explained on the basis of processes that are developmentally regulated resulting in vulnerability by virtue of birth at extreme prematurity (eg, adrenal glucocorticoid synthesis and antioxidant production).17–19 This possibility is supported by our observation of the association between PD and early gestational age.
ELGANs may also be vulnerable because of a unique environmental exposure that is distinct to this population. One possibility is an intrauterine exposure. The most common feature that distinguishes the intrauterine environment of ELGANs from more mature neonates is the far greater likelihood of intrauterine infection.20 Although not a consistent finding, intrauterine infection has been associated with the development of CLD, either as a risk factor by itself21 or when antenatal infection is associated with the need for prolonged mechanical ventilation.22
Neither EPPD nor PD had an increased frequency of histologic expressions of inflammation in the placenta as chorioamnionitis and funisitis. Therefore, we explored the possibility that placental infection with a specific organism might be associated with risk for PD. In previous work, postnatal recovery of Ureaplasma urealyticum from tracheal samples has been associated with the development of CLD.23 Although not statistically significant, among our study subjects Mycoplasma species, including Ureaplasma, was detected more frequently among the placentas of infants who developed EPPD.
Besides gestational age and birth weight, the most consistent risk factor for PD was treatment with mechanical ventilation (conventional or high-frequency ventilation) on postnatal day 7. It is not clear whether this association reflects a causal link between ventilation mode and PD or whether ventilation mode identifies infants with more severe lung disease. Treatment with oxygen during the first week of life in the low Fio2 and PD groups was similar, as was the frequency of the diagnosis of apnea at any time during the study period.
Two neonatal events, confirmed bacteremia and persistent PDA, have been associated with adverse outcomes, including CLD.24 However, prophylactic treatment of the PDA has not been demonstrated to reduce the risk of CLD.25 We examined both of these antecedents in relation to patterns of early lung disease. Bacteremia was a risk factor for EPPD but not for PD. This finding suggests that the etiology of persistent, severe lung disease may be different from the etiology of PD. The relationship between EPPD and both intrauterine exposure to specific organisms and postnatal infection implicates an infectious/inflammatory exposure in the pathogenesis.21
The diagnosis of PDA at any time during the first 2 weeks of life may be a risk factor for both EPPD and PD as this association approached statistical significance for both entities. A cause-and-effect relationship between shunting through a PDA and lung disease has biological plausibility.15,16 It is possible that our finding an association between PDA and both EPPD and PD is explained on the basis of an acquisition bias, in that infants with more severe lung disease were more likely to be clinically suspected of having a PDA and to be subjected to diagnostic testing (ie, echocardiography). Alternatively, PDA may be causally related to PD. Our data about indomethacin and/or ligation for PDA do not allow us to assess if these therapies reduced the risks of EPPD or PD. Determining the contribution of the PDA is important, because it may be a modifiable risk factor.
We examined the patterns of lung dysfunction in the first 2 weeks of life and the risk factors associated with these patterns in a large cohort of ELGANs. Infants with PD and EPPD tended to have lower gestational ages and lower birth weights than infants with consistently low Fio2. The only statistically significant potentially modifiable risk factor unique to infants in the PD group was receipt of mechanical ventilation (conventional or high-frequency ventilation) on postnatal day 7, although PDA might also be important. Future studies that characterize the inflammatory milieu of the lung by measuring biomarkers in tracheal fluid may suggest pathologic mechanisms for PD and strategies for prevention. Additional studies are also needed to characterize the relationship between these early patterns of lung dysfunction and subsequent health and development during infancy.
This study was supported by a cooperative agreement with National Institute of Neurological Disorders and Stroke grant 5U01NS040069-04. Dr Bose was supported by the Thrasher Research Fund.
We acknowledge our ELGAN Study colleagues: Olaf Dammann (Tufts Medical Center, Boston, MA), Bhavesh L. Shah (Baystate Medical Center, Springfield, MA), Camilia Martin (Beth Israel Deaconess Medical Center, Boston, MA), Robert Insoft (Brigham and Women's Hospital, Boston, MA), Karl Kuban (Boston Medical Center, Boston, MA), Francis Bednarek (UMass Memorial Health Center, Worcester, MA), John Fiascone (Tufts Medical Center, Boston, MA), Richard Ehrenkranz (Yale-New Haven Children's Hospital, New Haven, CT), T. Michael O'Shea (Wake Forest University/Baptist Medical Center, Winston-Salem, NC), Stephen C. Engelke (University Health Systems of Eastern Carolina, Greenville, NC), Carl Bose (University of North Carolina, Chapel Hill, NC), Mariel Poortenga and Ed Beaumont (DeVos Children's Hospital, Grand Rapids, MI), Nigel Paneth (Sparrow Hospital, Lansing, MI), Michael D. Schreiber (University of Chicago Hospital, Chicago, IL), Daniel Batton (William Beaumont Hospital, Royal Oak, MI), Greg Pavlov (Frontier Science and Technology Research Foundation, Amherst, NY), and Deborah Hirtz, our project officer.
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
ELGANs experience different patterns of respiratory disease in the first 2 postnatal weeks.
What This Study Adds
We examined the antecedents and modifiers of 3 a priori–defined patterns of respiratory disease in ELGANs.
The online version of this article, along with updated information and services, is located on the World Wide Web at: http://www.pediatrics.org/cgi/content/full/123/4/1124
Reprints Information about ordering reprints can be found online: http://www.pediatrics.org/misc/reprints.shtml