Objective. To determine, in the postsurfactant era, the incidence and clinical characteristics of infants with atypical versus traditionally defined bronchopulmonary dysplasia (BPD) among premature infants with birth weights <1251 g.
Design. Retrospective cohort study.
Setting. A single regional neonatal intensive care unit (level III/IV).
Patients. Two hundred thirty-two premature infants <1251 g at birth consecutively admitted during a 2-year period.
Main Outcome Measure. Incidence of classic BPD and atypical chronic lung disease (CLD) (occurring without preceding respiratory distress or after recovery from respiratory distress).
Results. Among 177 infants <1251 g who survived to 28 days, 27 (15%) had atypical CLD and 61 (34.5%) had classic BPD. Atypical CLD infants were significantly heavier and more mature than classic BPD infants (mean birth weights, 922 ± 152 g vs 854 ± 173 g; and mean gestational age, 26.8 ± 1.3 weeks vs 26.1 ± 1.6 weeks). Median duration of ventilator support (31 days; range, 2 to 127 vs 42 days; range, 4–145 days) and oxygen therapy (30 days; range, 11 to 163 vs 48 days; range, 19–180 days) were shorter in atypical CLD infants than in classic BPD infants.
Conclusion. Atypical CLD comprised 31% of total cases of CLD. Atypical CLD appears to be less severe than classic BPD. These data suggest that initial, acute lung injuries are not the sole antecedents of neonatal CLD.
- CLD =
- chronic lung disease •
- BPD =
- bronchopulmonary dysplasia •
- x-ray =
- chest radiogram •
- NICU =
- neonatal intensive care unit •
- Fio2 =
- fractional concentration of inspired oxygen •
- CGA =
- corrected gestational age •
- RDS =
- respiratory distress syndrome •
- CLDpRDS =
- chronic lung disease after resolved respiratory distress syndrome •
- DCLD =
- delayed onset chronic lung disease •
- NL =
In the current era of routine surfactant replacement and sophisticated neonatal care, increasing numbers of increasingly premature infants are surviving. These very immature infants may be contributing to the static incidence of chronic lung disease (CLD) found by several investigators.1 ,2 In addition, many authors have reported an atypical type of bronchopulmonary dysplasia (BPD) occurring in infants <1000 g at birth with mild or absent initial respiratory distress.3–6 Classic BPD, as originally described by Northway et al,7 excludes infants without respiratory distress. Furthermore, these cases do not meet the criteria of the previously described neonatal CLD syndromes such as Wilson-Mikity syndrome8 and chronic pulmonary insufficiency of prematurity.9 The current definition of BPD is broad and hazy, with several competing definitions in the literature (eg, oxygen supplement at 28 days of age,7oxygen at 28 days of age with at least 21 days of oxygen supplementation and a consistent chest radiogram [x-ray],10 and oxygen at 36 weeks' corrected gestational age (CGA).11)
Few reports have characterized the incidence and clinical hallmarks of atypical presentations of neonatal CLD.3–5 A review of medical records of all premature infants <2000 g at birth admitted to our neonatal intensive care unit (NICU) over 2 years showed a low incidence of CLD (3.6%) in neonates >1250 g at birth.12This report thus focuses on infants with birth weights <1251 g. To understand more fully atypical patterns of oxygen requirement in these premature newborns, we asked the following two questions: 1) In the postsurfactant era, what is the incidence of CLD among infants <1251 g birth weight? and 2) among those infants with CLD, how many have atypical CLD versus classic BPD, and what are their characteristics? To answer these questions, we conducted a cohort study examining all premature infants with birth weights <1251 g admitted to our NICU over a 2-year period.
All premature infants with birth weights <1251 g admitted to a single regional NICU covering a 12-county region (population of 1.2 million), during the 2-year period of July 1, 1991, to June 30, 1993, were identified. Patients with major congenital malformations (chromosomal anomaly, heart defects, pulmonary hypoplasia, and diaphragmatic hernia) and patients admitted beyond 1 week of age were excluded.
Perinatal and neonatal information was collected by a single reviewer (L.C.) directly from the original medical records. Demographic data, date of birth, sex, birth weight, gestational age as determined by the admitting neonatologist, and Apgar scores at 1, 5, and 10 minutes after birth were collected for each infant. Fractional concentration of inspired oxygen (Fio 2) at 1 hour and 4 hours after birth were recorded along with the daily most common Fio 2 administered during each of the first 28 days, as well as the Fio 2 at 36 weeks' CGA or discharge, whichever came first. The daily most common Fio 2 was the Fio 2required for the longest period over 4 hours during a 24-hour period as determined from the hourly recorded Fio 2 values during each 24-hour period. From the daily oxygen requirement individually plotted for all infants, five distinct patterns of respiratory disease, including three patterns of chronic oxygen requirement, were identified. For research purposes, strict definitions of these patterns were generated (Table 1) and then used to study these different patterns of CLD. The following data were also collected: prenatal administration of corticosteroids (at least one dose of betamethasone), administration of any surfactant therapy, duration of any oxygen requirement and ventilator support, dexamethasone therapy for CLD, chest radiograph interpretations by the attending pediatric radiologist at days 1, 3, 10, and 28, and at 36 weeks' CGA when available, and total duration of hospitalization. Duration of hospitalization was taken as the number of days in any hospital before discharge home. For those infants transferred to another facility, the information was obtained directly from medical records of the receiving hospital. Duration of hospitalization beyond 36 weeks' CGA was also analyzed.
In our unit, surfactant therapy has been used since 1985. The management of premature infants <1251 g during the period of this report included administration of prophylactic surfactant (Infasurf, ONY, Buffalo, NY) in the delivery room to infants <29 weeks. Subsequent surfactant (Infasurf or Survanta, Ross Abbott Laboratories, Columbus OH) was given up to a total of four doses, to infants with respiratory distress syndrome (RDS) requiring ≥40% Fio 2 and/or a mean airway pressure of ≥7. Prenatal corticosteroids were used at the discretion of the attending obstetrician (39.8% of infants <1251 g). The goals of respiratory management for RDS were to administer the lowest mean airway pressure and Fio 2 to maintain Pao 2 between 50 and 70 mm Hg, Paco 2 between 45 and 55 mm Hg, and arterial pH >7.30. At the discretion of the attending neonatologist, a 42-day tapering course of dexamethasone13 was initiated at 3 to 4 weeks after birth if the infant remained ventilated and required >30 to 40% Fio 2 despite treatment with diuretics and bronchodilators.
Statistical analysis was done by using the Intercooled StataTM program (Stata Corporation, College Station, TX). χ2 test or Fisher's exact test as appropriate were used to analyze categorical variables. One-way analysis of variance with Scheffé's correction was used to analyze multiple continuous variables. Logarithmic transformation was performed when continuous variables were not normally distributed. To compare classic BPD and atypical CLD, we used the Student's t test to analyze continuous variables, Wilcoxon rank sum test when continuous variables were not normally distributed, and Fisher's exact test for categorical variables. Stepwise multiple regression analysis was done to assess the independent relations of multiple variables with the presence of one or another pattern of CLD. All tests were two sided and P< .05 was considered statistically significant.
Distribution of the Study Population
A total of 232 premature infants with birth weights <1251 g were admitted to the NICU at the University of Rochester's Children's Hospital at Strong during the 2-year period of the study. Twelve infants were excluded (7 had congenital anomalies, 4 were late admissions, and 1 was extremely small for gestational age [405 g]) and 29 died before 28 days of age (Fig 1).
Among the 191 subjects of this study, 14 could not be classified in any of the defined chronic supplemental oxygen patterns. Among those unclassified infants, 11 were on oxygen for >2 weeks but <28 days (8 of these had x-ray changes of BPD at 28 days although they were not still requiring supplemental oxygen). The remaining 3 infants required supplemental oxygen continuously from birth through 28 days but their x-rays were all normal. All 14 unclassified infants were not included in the analysis, because we wanted to compare respiratory courses between strictly delineated groups of infants with atypical CLD and infants with classic BPD. A total of 177 infants survived >28 days and were the subject of further analysis. Of these, 10 infants died after 28 days and 7 of these died after 36 weeks' CGA. The causes of death were respiratory failure with CLD (5 of 10) and sepsis (5 of 10). Among those 10 neonates who died after 28 days of age, 2 infants had atypical CLD and 8 had classic BPD (Fig 1).
Patterns of Neonatal Chronic Lung Disease
Three patterns of chronic oxygen requirement were identified (Fig 2), classic BPD and two atypical CLD patterns, CLD post RDS (CLDpRDS) and delayed CLD (DCLD). Infants with classic BPD required continuous oxygen supplementation from birth until at least 28 days and had an abnormal x-ray consistent with BPD.10 Infants with CLDpRDS recovered from RDS and remained in room air for at least 72 hours, but then required oxygen supplementation at least until 28 days and had an abnormal x-ray. Infants with DCLD had no respiratory disease at birth and started to require oxygen supplementation at or after 6 days of age and subsequently developed an abnormal x-ray. Figure 3 illustrates the ages at the time of new supplemental oxygen requirement of infants with CLDpRDS or DCLD after being on room air and/or after having their RDS resolved. Most of these infants required supplemental oxygen starting at days 7 and 9.
Incidence and Characteristics of Three Patterns of Neonatal CLD
Incidence of CLD in Infants Less Than 1251 Grams
Among the 177 studied infants, 88 (50%) had CLD and 69% of those (61 of 88) had classic BPD. The incidence of atypical CLD during that period was 15% of the entire study population (27 of 177); CLDpRDS accounted for 11% and DCLD for 4%. Half of the 177 patients were classified as normal (NL) (19.8% of total) or having resolved RDS (30.5% of total).
Because infants with CLDpRDS or DCLD had similar patterns of oxygen requirement after the first week of age, regardless of initial pulmonary course, we combined these subjects into one group, the atypical CLD group. For the remainder of this analysis, classic BPD subjects were compared with all atypical CLD subjects.
Among 61 infants with classic BPD, 49 (80%) of these were <1001 g and 50 (82%) were <28 weeks at birth. Of the 27 infants with atypical CLD, 19 (70%) were <1001 g and 19 (70%) were <28 weeks. The birth weight and gestational age distributions of all CLD patterns are shown in Fig 4.
As expected, neonates with any CLD pattern were significantly smaller and more premature than infants in the RDS or NL group (P < .001 for all comparisons except birth weight in atypical CLD vs NL group, P = .02). When comparing atypical CLD and classic BPD infants, atypical CLD infants were significantly more mature (P = .038) and heavier at birth (P = .025). Male/female ratios and 5-minute Apgar scores were similar among all groups (Table 2).
Information about prenatal maternal treatment with betamethasone was available for 117 cases of whom 107 (91.4%) were <30 weeks' gestation. Corticosteroids were given prenatally in 40% who had information available. When comparing infants with any type of CLD and infants without CLD, the frequencies of documented prenatal corticosteroid administration were not different (40.6% vs 37.5%, respectively, P = .7).
Significantly more neonates in the NL or RDS groups were small for gestational age compared with neonates with any CLD pattern (P = .001) (Table 2). However, whether outcome data were analyzed comparing all subjects or only appropriate for gestational age infants in each group, similar results were obtained (data not shown).
Proportions of surfactant therapy, dexamethasone therapy, median durations of oxygen requirement, ventilator support, and hospitalization beyond 36 weeks' CGA are shown in Table 2. By definition, neonates with any pattern of CLD required significantly more ventilator support and oxygen therapy. They were also more likely to be hospitalized beyond 36 weeks' CGA when compared with infants without CLD. Similar proportions of infants with atypical CLD and infants with classic BPD received prophylactic surfactant at birth before any manifestation of lung injury (80% and 73%, respectively) and similar proportions of infants required later surfactant therapy for their respiratory distress. Infants with atypical CLD required oxygen therapy and mechanical ventilation for a shorter period of time when compared with infants with classic BPD. When multiple regression was done, the duration of oxygen therapy was significantly affected by the two groups (P < .008) and this was independent of birth weight and gestational age. In contrast, the duration of mechanical ventilation was significantly affected by birth weight (P < .000) but not the CLD pattern. There was a trend toward a higher proportion of classic BPD than atypical CLD infants needing supplemental oxygen beyond 36 weeks' CGA, but this difference was not statistically significant. The mean CGA of oxygen discontinuation was 39.5 weeks CGA for classic BPD and 37.5 weeks CGA for atypical CLD. A total of 38 patients received a 42-day course of dexamethasone therapy; most were <1000 g (32 of 38) and < 28 weeks' gestation (28 of 38) at birth. Four of these infants died (2 had classic BPD). Classic BPD subjects appeared to receive dexamethasone more frequently than atypical CLD subjects, but this difference was not statistically significant. The median duration of hospitalization beyond 36 weeks' CGA was not different between atypical CLD infants and classic BPD infants. Over the 2-year period studied, only 2 infants required oxygen therapy at home (1 of them had a tracheostomy) and those infants had a classic BPD pattern.
To date there is no satisfactory definition for neonatal CLD, and atypical presentations of CLD are more frequently observed in neonatal intensive care units.4 ,6 There are very few data on the incidence and disease course of these atypical presentations. In this study conducted during the postsurfactant era, three patterns of CLD were identified, ie, the classic BPD pattern, similar to the descriptions of Northway et al7 and Bancalari et al,10 and two atypical CLD patterns. CLDpRDS occurs after resolution of initial RDS and DCLD occurs later without preceding acute lung disease. CLD occurred in 50% of premature neonates <1251 g at birth and surviving beyond 28 days. This incidence is comparable with previously reported figures in this birth weight range.1 14–17 If supplemental oxygen at 36 weeks' CGA (which carries more significant prognostic value)11 was considered, CLD incidence dropped to 11%, and of those, 15% had atypical CLD. Development of CLD was significantly associated with birth weight and gestational age, as previously shown by multiple studies.1 ,16 18–20 Nearly one third of neonates with CLD had atypical CLD patterns. Infants with atypical CLD were heavier and more mature at birth than those with classic BPD and most of them were 26 to 27 weeks' gestation and 750 to 1000 g at birth. Furthermore, previous reports have shown that prenatal betamethasone therapy decreases the incidence of RDS21 ,22 and BPD15 ,23 in premature infants at risk. Our data did not show a difference in prenatal corticosteroids exposure between infants with or without CLD. Finally, although infants with atypical presentations had lengths of hospitalization similar to classic BPD infants, the morbidity related to their lung disease was less. They required less oxygen therapy even after correcting for birth weight and gestation, suggesting at least a milder disease and possibly a different pathophysiology. As the purpose of this study was descriptive, we did not record many factors that could potentially predict the frequency of different patterns of CLD. Recently, the role of sepsis and patent ductus arteriosus (PDA) were studied by Rojas et al4 in surviving premature infants <1000 g at birth. Sepsis, PDA, or the combination of both were associated with an increased risk of developing CLD (defined as the need for supplemental oxygen for 28 days or longer during the first 2 months of age with an abnormal x-ray). Their definition and study population fit into one pattern of atypical CLD (CLDpRDS) we described. Infants were initially on oxygen (median duration of 3 days during the first week) and recovered to room air before they required supplemental oxygen. If the two patterns of atypical CLD share the same pathophysiology, PDA and sepsis may also be associated with the advent of the delayed CLD described in our population. We suspect that these patterns of oxygen requirement may represent entities distinct from classic BPD and therefore may be useful for studying the mechanisms of lung injury and the pathophysiology of CLD. If these patterns have different causes, effective therapeutic interventions and long-term prognosis may differ as well.
Of the total cohort, 14 infants could not be classified into one particular pattern and therefore were not included in the final analysis, because we sought to compare subjects with clear diagnoses of either classic BPD or atypical CLD. The main reason for nonclassification was a duration of oxygen requirement of <28 days. The oxygen requirement patterns of 7 infants were similar to those of classic BPD, and the 7 others had patterns similar to CLDpRDS; in fact 8 infants had BPD changes on x-ray. However, because those 14 unclassified subjects did not meet the strict definition criteria of classic BPD, they were not initially included. When the data were analyzed including these infants either in the RDS group or in the BPD group (7 subjects) and CLDpRDS group (7 subjects), the conclusions were not affected.
Before the surfactant era, Wung et al3 described two types of CLD in premature infants <1500 g at birth. Type I CLD occurred in infants <1000 g at birth, who had no or mild RDS, apnea and bradycardia, and required low Fio 2 (<30%) with minimal ventilator support in the form of continuous positive airway pressure. Type II CLD occurred in infants >1000 g at birth who had severe RDS. This type was consistent with classic BPD. The authors attributed type I CLD to lung immaturity and type II CLD to damage from mechanical ventilation and oxygen toxicity. Our data show that classic BPD and atypical CLD are both seen in infants with birth weights <1000 g as well as in many infants with birth weights between 1000 and 1250 g. Similar to Wung et al,3 we found a group of infants with delayed onset of CLD who had low oxygen requirements during the first week after birth. Thus, immaturity and low birth weight are major factors in CLD development and may be sufficient to put infants at risk for CLD even in the absence of initial RDS or pneumonia.
Le Guennec et al,5 also before the surfactant era, reported a bimodal temporal distribution to the oxygen dependency in surviving premature infants with birth weights between 600 and 1000 g. These neonates experienced a “honeymoon” period, independent of initial RDS, during which they were weaned to room air by the end of the first week of age. Although those patients' mean Fio 2 values were not reported, the authors found that among 76 infants requiring oxygen at day 28 of age, 21 infants (28%) were in room air at day 3 and 33 (43%) were in room air at day 7. Fifty-seven percent of the studied infants did not have RDS at birth; whether they had another acute lung injury is unclear. In the group of infants we studied, we found a similar honeymoon period between day 4 and days 7 to 8 after birth, in patients with atypical CLD (Fig 2). During this period, the mean Fio 2of all infants on any 1 day was <0.25 and each infant, by definition, had at least 3 consecutive days of Fio 2 of 0.21. The resurgence of an oxygen requirement appeared between day 7 and day 9 (Fig 3) and continued to at least day 28. The honeymoon phenomenon may reflect the resolution of acute lung injury followed by the occurrence of new insult(s) leading to chronic lung disease. As the new onset of oxygen requirement was also observed in neonates without initial lung disease in a similar time frame, we hypothesize that the two atypical CLD patterns we have described may have a similar pathophysiology. The narrow window between 7 and 9 days of age may be an interesting target for basic research in the onset of CLD.
We have described the characteristics of infants with two forms of neonatal CLD in the postsurfactant era. The findings of others before the surfactant era are confirmed, suggesting that surfactant may not eliminate the process that leads to atypical CLD presentation. This atypical presentation may reflect developmental alteration after premature use of the lungs. This “developmental” CLD occurs in neonates who are heavier and less premature than the ones who develop classic BPD, it appears around 6 to 10 days after birth, and it usually resolves before the infant's discharge. The distinction between classic BPD and atypical CLD is particularly useful in counseling parents of premature infants with regard to short-term prognosis and discharge plans. Because of the retrospective nature of this investigation, no causal relations between the factors examined and the development of a particular pattern of CLD can be deduced. Further prospective studies are needed to clarify the role of prenatal corticosteroids, the relation between barotrauma, infection, PDA and atypical CLD, and the prognostic significance of a given pattern of CLD. Understanding these patterns should facilitate design and analysis of prospective clinical trials, and guide preclinical investigations into the cause of CLD.
These studies were conducted with the assistance of the General Clinical Research Center at the University of Rochester, funded by Grant RR0004434 from the US National Institutes of Health, and NHLBI SCOR HL 36543.
We thank Drs W.M. Maniscalco, J. T. McBride, and R. H. Notter for their valuable critique and discussion of the manuscript and Dr R. M. Ryan for her assistance with data analysis. Also, the authors would like to thank Hillary Danneman, Denise Houston, and Joan Merzbach for initial data collection.
- Received April 27, 1998.
- Accepted August 28, 1998.
Reprint requests to (D.L.P.) Division of Neonatology, Children's Hospital at Strong, 601 Elmwood Ave, Box 651, Rochester, NY 14642.
- Parker R,
- Lindstrom D,
- Cotton R
- ↵Wilson M, Mikity V. A new form of respiratory disease in premature infants. AMA J Dis Child. 1960;99:119/489–129/499
- Krauss A,
- Klain D,
- Auld P
- Shennan A,
- Dunn M,
- Ohlsson A,
- Lennox K,
- Hoskins E
- Van Marter L,
- Leviton A,
- Kuban K,
- Pagano M,
- Allred E
- Avery M,
- Tooley W,
- Keller J,
- et al.
- Edwards D,
- Dyer W,
- Northway W Jr
- Horbar J,
- McAuliffe T,
- Adler S,
- et al.
- Van Lierde S,
- Smith J,
- Devlieger H,
- Eggermont E
- Copyright © 1999 American Academy of Pediatrics