Definition
Comments
Clinical
Only receipt of supplemental oxygen required, without specification for level of respiratory support
May be specified to require supplemental oxygen for each of the first 28 days
36 weeks’ PMA [7]
Only receipt of supplemental oxygen required, without specification for level of respiratory support
May be modified to classify all infants on assisted ventilation as BPD (similar to physiologic definitions), but this is not specified in most publications
May be modified to include status at discharge to home or transfer to another hospital prior to 36 weeks’ PMA
Should reflect infant’s usual level of support at 36 weeks’ PMA
28 days of FiO2 >0.21 for at least 12 h per day prior to 36 weeks’ PMA required, but some have modified the definition to require FiO2 >0.21 at 28 days of age (rather than 28 days of supplemental oxygen), or to drop this requirement (infants without supplemental oxygen at 36 weeks’ PMA are No BPD)
Infants on assisted ventilation are classified as severe
“Effective FiO2” may be used in some cases to determine severe versus moderate classification
Categories may be collapsed to moderate-to-severe versus none
Physiologic
Determined at 36 ± 1 weeks’ PMA with ongoing need for supplemental oxygen determined by physiologic challenge
“Effective FiO2” <0.30 may be used to determine eligibility for challenge
Categories may be collapsed to moderate-to-severe versus none
Determined at 36 ± 1 weeks’ PMA as ongoing assisted ventilation or ongoing need for supplemental oxygen (FiO2 >0.21)
As the impact of the physiologic challenge test on rates of BPD has decreased, failure rates for the physiologic challenge test have increased. In the initial multicenter experience, 44 % (101/227) of infants passed the physiologic challenge [21]. In contrast, only 30 % (80/266) of infants in the recent Prematurity and Respiratory Outcomes Program (PROP) study passed the challenge [28]. It is important to note that these studies both used the same oxygen saturation cutoff for failure (<90 %) [21, 29]. Higher oxygen saturation cutoffs will result in lower rates of passing the physiologic challenge; therefore, overall rates of BPD by physiologic definitions will vary based on oxygen saturation targets [20]. In practice, lower clinical oxygen saturation targets result in fewer infants remaining on supplemental oxygen and eligible for the challenge, while lower test saturation targets result in fewer infants failing the physiologic challenge, together resulting in lower rates of BPD.
Given consistency in oxygen saturation criteria, the results of the physiologic challenge have been validated with respect to baseline level of support. Infants on low flow nasal cannula (0.02–2.00 LPM) with effective FiO2 0.21–0.49 who underwent physiologic challenge were more likely to pass the challenge with a lower baseline effective FiO2 (0.23 ± 0.03 versus 0.26 ± 0.05) [23].
An additional potential benefit of the physiologic challenge is for the classification of infants receiving respiratory support without supplemental oxygen. This scenario is not explicitly addressed by any of the clinical definitions for BPD, with the exception of the use of assisted ventilation for support in infants with severe BPD. Walsh and colleagues described infants on low flow nasal cannula (0.13–2.00 LPM) with FiO2 0.21; 15/22 (68 %) of infants passed the challenge [23]. Similarly, in PROP, 55/81 (68 %) of infants on nasal cannula support without supplemental oxygen passed the physiologic challenge [28]. Use of newer respiratory support strategies, such as humidified high flow nasal cannula (HHFNC), which provides variable degrees of positive pressure respiratory support (that are not measured in most clinical settings), is particularly challenging to classify (with or without supplemental oxygen) [30]. In addition to the inability to assess its contribution to stabilizing an infant’s respiratory status due to inconsistent provision of positive pressure, the use of HHFNC adds further variability to a clinical diagnosis of BPD as it is not currently employed in a consistent manner within and across centers [31–33].
Another aspect of the respiratory status of former preterm newborns to be considered when evaluating the need for ongoing respiratory support is dysmaturity of control of breathing. Infants born more preterm are more likely to have persistent episodes of periodic breathing and/or desaturation at and beyond 36 weeks’ postmenstrual age and term [34–37]. Coste et al. demonstrated that control of breathing was an important factor in passing the physiologic challenge test for BPD at 36 weeks’ postmenstrual age [38]. Sixteen of 17 preterm infants who experienced increased periodic breathing during the challenge subsequently failed the challenge due to desaturation. This does not preclude the coexistence of lung disease, however. In healthy former preterm infants off respiratory support, both frequency of apnea and proportion of apneic events with desaturation were inversely related to functional residual capacity (FRC), and desaturations were also inversely related to minute ventilation [39]. Consistent with these findings, Coste and colleagues found that infants failing the physiologic challenge averaged higher baseline nasal cannula flows than those passing the test (1.1 LPM versus 0.4 LPM), whereas FiO2 was not a significant discriminator of the challenge result [38]. Thus, positive pressure provided by nasal cannula support could stabilize an infant’s FRC, decreasing periodic breathing and desaturation, and account for the ongoing prescription of respiratory support in the clinical setting, as well as failure to maintain oxygenation during the physiologic challenge [30, 40]. This could be true even in the absence of supplemental oxygen, which would explain why approximately one third of infants on nasal cannula flow without supplemental oxygen are reported to have failed the challenge test [23, 28].
Interestingly, with respect to the variable physiology of BPD that is reflected in the persistence of respiratory support with or without oxygen supplementation in extremely premature newborns, Ballard and colleagues prospectively planned for repeat clinical or physiologic assessments at 40 weeks’ postmenstrual age (term) in two trials of investigative therapies to prevent BPD [24, 26, 41]. Both of these studies showed substantial decreases in the diagnosis of BPD in survivors between 36 and 40 weeks (66 % versus 38 % in the more recent Trial of Late Surfactant) [26].
Validation of the Diagnosis of BPD: Respiratory Morbidity Outcomes
Although there are multiple studies demonstrating that various definitions of BPD are significantly associated with respiratory morbidity at follow-up, there are few studies that explicitly evaluate validity of the BPD diagnosis for prediction of later pulmonary outcomes [42–44]. As previously noted, Shennan and Davis and their colleagues evaluated evolving clinical definitions of BPD for accuracy in predicting various pulmonary outcomes at 1–2 years corrected age, while Ehrenkranz et al. also provided important validation data [7, 10, 17]. Findings are summarized in Table 2. Parad et al. evaluated the utility of BPD diagnoses at 28 days and 36 weeks’ postmenstrual age for respiratory morbidity outcomes defined at 2 years corrected age (cough, wheeze, respiratory medication use, and hospitalization for respiratory indication) [45]. They found poor utility for both BPD definitions (area under the receiver-operator curve 0.50–0.62) in both development and validation cohorts. They also compared the performance of the BPD classifications to prevalence of respiratory symptoms or medications use at 1 year corrected age, demonstrating that 1-year respiratory morbidity was significantly better at predicting 2-year outcomes than BPD (area under the receiver-operator curve 0.64–0.78) [45]. Taken together, these studies suggest that sensitivity is higher than specificity at 28 days, while the reverse is true at 36 weeks’ postmenstrual age, although this does not necessarily translate into greater accuracy for outcome prediction. Also, BPD may be a better predictor of composite morbidity outcomes (which incorporate home respiratory support and respiratory medications and hospitalizations, with or without symptoms) than it is of individual respiratory morbidity domains (home support, medications, hospitalization, or symptoms).
Table 2
Sensitivity, specificity, and utility of clinical definitions of bronchopulmonary dysplasia for various respiratory morbidity outcomes at 1–2 years corrected age
Study | Patient population | Corrected age at assessment | Outcome | Outcome incidence | BPD definition | Sensitivity | Specificity | Accuracy/ discrimination |
|---|---|---|---|---|---|---|---|---|
Shennan (1988) | ≤1500 g, born 1981–1985, n = 605 | 2 years | Composite of various respiratory morbidities (including medication use and hospitalizations) | 20 % | 28 days | 79 % | 69 % | 71 % correctly classified |
36 weeks’ PMA | 63 % | 91 % | 85 % correctly classified | |||||
Davis (2002) | 500–999 g, born 1996–1998, n = 945 | 18 months | Composite of various respiratory morbidities (including medication use and hospitalizations) | 54 % | 28 days | 67 % | 54 % | 61 % correctly classified |
36 weeks’ PMA | 46 % | 82 % | 63 % correctly classified | |||||
40 weeks’ PMA | 33 % | 96 % | 62 % correctly classified | |||||
Ehrenkranz (2005) | ≤1000 g, <32 weeks’, born 1995–1999, n = 3848 | 18–22 months | Bronchodilators or diuretics | 35 % | 36 weeks’ PMA | 54 % | 62 % | NA |
Hospitalization for respiratory indication | 30 % | 36 weeks’ PMA | 52 % | 60 % | NA | |||
Stevens (2014) | 24–27 weeks’, born 2005–2009, n = 918 | 6–22 months | Persistent cough or wheeze | 61 % | 36 weeks’ PMA | 44 % | 64 % | NA |
Inhaled steroids | 26 % | 56 % | 64 % | NA | ||||
Hospitalization for respiratory indication | 31 % | 51 % | 64 % | NA | ||||
Parad (2015) | 23–28 weeks’, ventilated on first day of life, born 1998–2001, n = 76 (development cohort) and n = 227 (validation cohort) | 2 years | Respiratory symptoms and/or medications | NA | 28 days | NA | NA | AUROC 0.53 |
36 weeks’ PMA | NA | NA | AUROC 0.53–0.62 | |||||
Hospitalization for respiratory indication | NA | 28 days | NA | NA | AUROC 0.50–0.55 | |||
36 weeks’ PMA | NA | NA | AUROC 0.54–0.55 |
Ehrenkranz also demonstrated significant increases in rates of use of respiratory medications (diuretics and bronchodilators) and hospitalization for respiratory indication with increasing severity of BPD, by the NIH Workshop definition [17]. I have been unable to identify published reports of the relationship of the physiologic definitions of BPD and later pulmonary outcomes.
Validation of the Diagnosis of BPD: Pulmonary Function
The classification of BPD based on physiologic definitions demonstrates that, at least for brief observation periods, infants without BPD do not require respiratory support (with or without supplemental oxygen) to maintain adequate oxygen saturations. The differences leading to the ongoing need for respiratory support are likely related to lung structure and function, with some contribution of variable maturation of control of breathing. Kaempf and colleagues demonstrated that infants with BPD (as classified by both clinical and physiologic definitions) have higher capillary PCO2 at 36 weeks’ postmenstrual age than infants without BPD, indicating poorer respiratory system function at the time of the diagnosis of BPD [46]. We have recently reviewed pulmonary function in former preterm infants in infancy and childhood [47]. Generally, pulmonary function tests demonstrate that lung function in infants and young children is decreased following a diagnosis of BPD (by various definitions, although most commonly a clinical diagnosis at 36 weeks’ postmenstrual age) compared to full-term controls. Former preterm infants without BPD also often demonstrate lower lung function than controls but better function than infants with BPD. Airway obstruction, with or without differences in lung volumes, is the most common abnormality in lung function. Former preterm children with and without BPD may have evidence of relative decline in lung growth, before compensatory growth is seen, although there can be a persistence of decreased lung function at school age and beyond. Filbrun and colleagues demonstrated that one factor associated with improving lung function in children with a diagnosis of BPD is “catch-up” growth, which exceeds expected growth rates [48]. Other work has shown that children with a diagnosis of BPD experience less compensatory lung growth than children without BPD [49].
Balinotti and colleagues have shown that infants with BPD have decreased alveolar-capillary surface area, relative to lung volume, compared to full-term controls [50]. Follow-up work showed this difference is due to proportional decreases in both the pulmonary diffusion membrane and capillary blood volume components [51]. These data are consistent with enlarged alveoli with decreased septation. In addition, both of the components (capillary membrane and capillary bed) were directly related to body length, providing a pathway by which enhanced somatic growth in former preterm infants could result in improved lung structure and normalization of lung function, as well as an explanation for increasing or persistent decrements in lung function in childhood, particularly in children with a diagnosis of BPD.
Validation of the Diagnosis of BPD: Nonpulmonary Outcomes
Generally, the diagnosis of BPD correlates with other neonatal morbidities of prematurity. This convergent validity has been demonstrated for both clinical and physiologic definitions of BPD [17, 25, 28]. Ehrenkranz et al. evaluated a modification of the NIH Workshop definition, demonstrating increased rates of severe intraventricular/intracranial hemorrhage, sepsis, and necrotizing enterocolitis with increasing severity of BPD [17]. Similarly, Poindexter and colleagues evaluated two clinical definitions of BPD determined at 36 weeks’ postmenstrual age—modified versions of the definition proposed by Shennan et al. and the NIH workshop definition. Infants with BPD by either definition were significantly more likely to have severe intraventricular/intracranial hemorrhage, sepsis, and severe retinopathy of prematurity [28]. With respect to the binary physiologic definition of BPD at 36 weeks’ postmenstrual age, Natarajan et al. showed that BPD was significantly associated with increased rates of severe intraventricular/intracranial hemorrhage, sepsis, necrotizing enterocolitis requiring surgery, and severe retinopathy of prematurity [25]. Thus, the level of illness and immaturity that lead to the diagnosis of BPD, also carry risks of these other morbidities, which, in some cases, may directly increase the risk of BPD due to associated inflammation (particularly sepsis and necrotizing enterocolitis) [52]. Similarly, neurologic injury is associated with delayed brain development, which may influence maturation of respiratory control [53–55].
The diagnosis of BPD has also been associated with poor somatic growth. Poindexter and colleagues showed increased rates of growth failure (weight <10th percentile) at 36 weeks’ postmenstrual age with both of their clinical definitions of BPD [28]. In infants with severe BPD, the prevalence of growth failure increased with advancing postmenstrual age (36–48 weeks’) among infants who remained hospitalized [56]. Rates of growth failure (<10th percentile) at 18–22 months corrected age also increased with increasing severity of BPD by the NIH Workshop definition and the diagnosis of BPD by the physiologic definition, with length and head circumference more variably affected [17, 25].
Multiple neonatal morbidities are associated with adverse neurodevelopmental outcomes in extremely preterm infants [57–59]. Among the nonneurologic morbidities, the diagnosis of BPD has been consistently associated with poorer neurodevelopmental status, at least doubling the odds of developmental disability [57, 59]. Schmidt and colleagues used a clinical definition of ongoing receipt of supplemental oxygen at 36 weeks’ postmenstrual age for these analyses. Other investigators have also shown associations between clinical diagnoses of BPD and neurological outcomes, although some studies have focused on more severe definitions of BPD (e.g., including only those infants discharged on supplemental oxygen) [10, 60, 61]. However, Laughon and colleagues were unable to show a significant relationship between BPD and neurodevelopmental outcome at 2 years corrected age after adjusting for various antenatal factors considered to be antecedents of BPD [62]. Ehrenkranz and colleagues did show a significant association with severity of BPD using the clinical NIH Workshop definition and various neurodevelopmental outcomes [17]. And, Davis et al. showed similar utility of both 36- and 40-week clinical definitions of BPD for neurodevelopmental outcomes at 18 months corrected age [10]. Finally, Natarajan and colleagues demonstrated significant associations between the binary physiologic definition of BPD and various cognitive and motor outcomes at 18–22 months corrected age [25].
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
