Bronchopulmonary dysplasia remains the most common morbidity of prematurity and carries a significant disease burden [10]. Throughout the published literature, BPD displays itself as a disease with significant heterogeneity [11,12,13,14]. This is found not only within different “types” of BPD but also within the definition itself; as published data defines it as oxygen at 28 days, 36 weeks or other combinations of factors [15]. Therefore, it is essential to have accurate information for prediction, analysis and treatment. We performed this systematic review and meta-analysis to determine large-scale rates of bronchopulmonary dysplasia, with a subgroup analysis according to two major definitions. To our knowledge this is the largest and most comprehensive study describing BPD prevalence to date.
Our study expands on the 2019 study by Siffel et al. [3] to provide a more complete review of available data. We discovered, reviewed and analyzed data over a 41-year period (versus 11 years), with inclusion of a higher number of studies across more regions. As an additional contrast, we defined BPD (oxygen at 28 days or 36 weeks) and manually extracted data for combined analysis. This allowed us to use pooled data to compare subgroups and pursue further statistical analyses. We were therefore able to provide a more accurate prevalence for each provided outcome, rather than reporting outcomes as a set of ranges from individual studies.
As anticipated, the foremost risk factor for developing BPD was found to be low birth weight, particularly with a weight below 750 g. This trend was evident across both individual subgroup analyses and combined evaluations. Additionally, our observations revealed discrepancies in BPD rates among different gestational age groups, notably between ELGA and VLGA infants. These findings align with existing literature that underscores an inverse association between BPD rates and gestational age/birthweight, further affirming the current understanding in the field [8, 16].
We also compared BPD rates across three decades (1990–1999, 2000–2010 and 2010–2020), which showed no difference between the groups across the definitions of BPD. This is found throughout the literature and highlights the difficulty in preventing and treating this disease. Medical advancements in the care of preterm neonates have led to higher survival, especially in the most industrialized nations [17, 18]. This coincides with the survival of more infants with BPD and accounts for much of the similarity of the prevalence across decades. While our study focuses on reporting BPD rates in decade cohorts, it’s essential to acknowledge the limitations inherent in utilizing these broader definitions of BPD. We recognize that the clinical landscape of BPD management may have evolved over the past 30 years, potentially leading to improvements not fully captured by the BPD28 and BPD36 definitions. Our exclusion of studies using the newer BPD definition by Jensen et al. was indeed mentioned in the methods section, but we acknowledge the importance of reiterating this point here for clarity.
While the incidence of BPD exhibits considerable variation among different countries, current evidence indicates minimal disparities in its prevalence across major continents. Numerous studies have explored BPD incidence and associated risk factors in various regions spanning North America, Europe, Asia, and Australia, generally yielding comparable rates. For instance, research by Jain et al. found no significant divergence in BPD incidence among preterm infants across North America, Europe, and Australia [19]. In contrast, our investigation suggests notable differences in BPD rates among regions or continents, particularly with lower rates observed in Europe and South America. However, it’s noteworthy that South America’s data pool was limited to just 1–2 studies. These findings imply that the risk factors and underlying pathophysiology of BPD may not uniformly align across geographical regions, underscoring the imperative for further investigation to elucidate these distinctions. This prompts consideration as to whether disparities in clinical practices might potentially justify these findings.
The Neonatal Research Network (NRN) in the United States has compiled large retrospective analyses of care practice and patient outcomes among extremely premature infants. They have demonstrated that rates of antenatal steroids and surfactant administration have increased, delivery room intubation has decreased [7]. However, the rates of bronchopulmonary dysplasia (BPD36) in their study ranged from 32 to 45%, which is notably higher than the 21% observed in this study. This difference could be attributed to the varying gestational ages included in the studies, as the NRN’s research comprised newborns between 22 and 28 weeks. In comparison, the Chinese Neonatal Network’s cohort of 8,148 preterm neonates had a BPD36 rate of 29.2%, which is higher than our study’s results, again differences most likely due to their inclusion of neonates 31 weeks and younger whereas our study included neonates of ≤ 32 weeks [20].
The prevalence of BPD varied depending on the study setting, with national cohorts demonstrating the highest rates for both definitions of BPD. These estimates may be more reliable, as they offer a broader representation across multiple institutions, reducing the impact of outliers and the unique management practices of individual hospitals on the results. Furthermore, many of these national studies employed inclusion criteria that targeted younger gestational ages, further enhancing their robustness. Despite the thought that GDP may have an impact on BPD rates, subgroup analyses based on quartiles of a nation’s GDP showed no differences. One possible explanation for this finding is that other factors beyond GDP, such as access to healthcare and neonatal resources, may play a more significant role.
Limitations
Despite conducting an extensive data search employing multiple reviewers and diverse search methods, there remains a possibility that certain available studies may have been overlooked. Our findings reveal considerable heterogeneity across all examined outcomes, with many I2 values approaching 1. Despite efforts to minimize this through meticulous data extraction and analysis, the persistence of heterogeneity underscores the importance of cautiously applying the results to specific disease populations. For example, Bonamy et al. reported low BPD rates as it exclusively classified the condition in individuals with the severe form of the disease. In an attempt to mitigate the observed heterogeneity, we conducted a sensitivity analysis, which yielded rates comparable to those obtained in the initial analysis characterized by high heterogeneity.
Another constraint stems from the limited granularity of the original datasets, owing to the diverse definitions of BPD and the myriad ways in which data can be presented. This limitation restricts our ability to conduct more sophisticated statistical analyses and may lead to unequal weighting of studies where data accessibility varies. Additionally, there is a notable disparity in the amount of data available for some regions, notably North America, Oceania, and Europe, compared to other global populations. It would have been ideal to gather data as comprehensive as that publicly available from Australia and New Zealand, Canada, and Japan. Moreover, handling mortality data was a significant challenge in our analysis. We encountered variations among studies, where some solely included survivors while others reported mortality rates without adjusting them in their BPD rates. Some observed rates may have been exceptionally low, especially if their mortality rates were high. We were unable to solely include survivors due to variations in study methodologies, with some studies including only survivors while others encompassed all patients in their denominator for BPD, regardless of neonatal mortality. Adapting our analysis to account for this disparity without access to patient-level data limited our analyses. To address this limitation, we included mortality rates in the supplementary materials. This allows for transparency regarding the impact of mortality on our findings and provides additional context for interpreting the results. While we hypothesized differences in pathophysiology as a possible cause for national differences, it is essential to acknowledge other potential factors that may influence BPD rates, such as variations in reporting practices, gestational age and birth weight distributions, and early mortality rates. These factors could contribute to the observed regional differences in BPD rates and warrant further investigation. Also, differences in the sophistication of medical treatment across regions impacts survival and eventual diagnosis of BPD, all of which affect overall outcomes and generalizability.
Add Comment