Healthcare burden of bronchopulmonary dysplasia among very preterm infants in China: a cohort study Authors: Zhicheng Zhu, Lin Yuan, Jie Yang, Siyuan Jiang, Yun Cao, Jiao Yang, Yun Zhang, Yan Mo, Chao Chen, Huayan Zhang, Jianguo Zhou, Shoo K Lee Published: 2026 ABSTRACT AbstractObjectiveTo assess the healthcare burden of bronchopulmonary dysplasia (BPD) among very preterm infants in China.DesignA prospective cohort study between 2022 and 2023.SettingChinese Neonatal Network (CHNN) participating centres.PatientsInfants with gestational age <32 weeks admitted to CHNN neonatal intensive care units.Main outcome measuresA composite rate of BPD or mortality at 36 weeks’ postmenstrual age (PMA), major comorbidities, clinical resources utilisation and outcome at discharge. BPD severity was classified by Jensen et al’s criteria.ResultsAmong 17 793 eligible infants, 568 (3.2%) infants died before 36 weeks’ PMA, 1729 (9.7%) were discharged against medical advice before 36 weeks’ PMA, 9895 (55.6%) were classified as no BPD, 2751 (15.5%) developed Grade 1 BPD, 2634 (14.8%) developed Grade 2 BPD and 216 (1.2%) developed Grade 3 BPD. Infants with BPD had significantly longer hospital stays than those without BPD (median (IQR), 67 (52–84) vs 44 (34–56) days) and incurred higher total hospitalisation charges (median (IQR), 127 (91–177) vs 73 (52–103) thousand CNY) and charge per day (median (IQR), 1975 (1638–2361) vs 1714 (1409–2048) CNY). Mortality at discharge increased with BPD severity, with rates of 0.2% (18/9895) for infants without BPD, 0.5% (15/2751) for Grade 1 BPD, 2.1% (56/2634) for Grade 2 and 26.9% (58/216) for Grade 3. Similarly, the rates of major comorbidities and the need for home oxygen therapy increased with BPD severity.ConclusionsGreater BPD severity was associated with increased comorbidities, higher in-hospital mortality and greater utilisation of healthcare resources. These findings emphasised the ongoing need to develop cost-saving strategies to reduce the risk and severity of BPD in this vulnerable population and improve overall care. WHAT IS ALREADY KNOWN ON THIS TOPIC Several studies in high-income countries have reported a substantial burden of bronchopulmonary dysplasia (BPD) on healthcare systems. The pressures might be exacerbated in low- and middle-income countries, where resource constraints limit access to specialised care and worsen outcome disparities, while critical gaps persist in understanding the full scope of BPD’s burden, particularly in large populations such as China. Several studies in high-income countries have reported a substantial burden of bronchopulmonary dysplasia (BPD) on healthcare systems. The pressures might be exacerbated in low- and middle-income countries, where resource constraints limit access to specialised care and worsen outcome disparities, while critical gaps persist in understanding the full scope of BPD’s burden, particularly in large populations such as China. WHAT THIS STUDY ADDS We found that more than 40% of very preterm infants either died before 36 weeks’ postmenstrual age (including discharge against medical advice with advanced life support) or developed BPD in China. Greater BPD severity was associated with increased comorbidities, higher in-hospital mortality and greater utilisation of healthcare resources during initial hospitalisation. We found that more than 40% of very preterm infants either died before 36 weeks’ postmenstrual age (including discharge against medical advice with advanced life support) or developed BPD in China. Greater BPD severity was associated with increased comorbidities, higher in-hospital mortality and greater utilisation of healthcare resources during initial hospitalisation. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Insights into the healthcare burden of BPD can assist in medical resource allocation, parental counselling and policy development, emphasising the ongoing need to develop cost-saving strategies to reduce the risk and severity of BPD. Insights into the healthcare burden of BPD can assist in medical resource allocation, parental counselling and policy development, emphasising the ongoing need to develop cost-saving strategies to reduce the risk and severity of BPD. Introduction Bronchopulmonary dysplasia (BPD) is one of the most prevalent and severe complications for very preterm infants (VPIs). Globally, the reported incidence of BPD varied widely (ranging from 10% to 89%) depending on diagnostic criteria, gestational age (GA) and regional healthcare practices.1 Although modern perinatal care had improved survival rates of these infants,2 BPD incidence remained high or even increased in some cases.35 A double increase in the rate of BPD in extremely preterm infants from 2010 to 2019 in China had been reported.6 BPD is characterised by disruptions in the development of alveolar and vascular compartments, leading to a variety of comorbidities and adverse long‐term outcomes.7 8 This chronic lung disease gives rise not only to reductions in quality of life but also to increased resource use and healthcare costs, thus exerting a substantial burden on families and societies. In high-income countries, first-year healthcare expenditures for severe BPD could exceed $440 000 per infant, with additional costs accruing from frequent readmissions, home oxygen therapy and long-term rehabilitation.9 These economic pressures might be exacerbated in low- and middle-income countries, where resource constraints limit access to specialised care and worsen outcome disparities. Data from a Chinese single-centre study revealed that initial hospitalisation charges for BPD infants were 1.4 times higher than those of uncomplicated controls, driven by prolonged neonatal intensive care unit (NICU) stays and multidisciplinary care needs.10 Despite its clinical and socioeconomic significance, critical gaps persist in understanding the full scope of BPD’s burden, particularly in large populations such as China, where nationwide epidemiological and cost data remain scarce. Given these challenges, a comprehensive assessment of BPD’s healthcare burden is urgently needed to guide medical care resource allocation, parental counselling and policy development. This study aimed to assess the healthcare burden of BPD among VPIs in China using a large cohort from the Chinese Neonatal Network (CHNN). Methods Study design and data source This cohort study used data collected prospectively in the CHNN, a standardised national database established in 2019 to monitor care practices and outcomes for preterm infants born before 32 weeks’ GA or weighing less than 1500 g across NICUs in China.11 Both inborn and outborn infants were included and followed until discharge or death from NICUs. The CHNN expanded gradually from 57 NICUs in the initial phase to 89 NICUs by 2022 and subsequently increased to 102 NICUs by 2023. All participating centres were authorised tertiary referral hospitals with level III NICUs, known for their expertise in caring for high-risk neonates. Collectively, they provided care for approximately 10% of all VPIs in China and were selected to represent neonatal care in different areas of the country.12 Data collection from patient medical records was conducted following standardised protocols by trained abstractors, who used operational manuals to enter information into a customised database featuring built-in error checking. Deidentified patient data were electronically transmitted to the coordinating centre at Children’s Hospital of Fudan University, where rigorous quality control measures were implemented.13 In 2022, CHNN updated the data collection protocols to include new variable entries on hospitalisation expenses. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline. Study population Eligible infants were born less than 32 weeks’ GA who were admitted to participating NICUs of CHNN between 1 January 2022 and 31 December 2023. There was no lower limit set for GA or birth weight. Infants who were admitted after 72 hours of life, had major congenital anomalies and those with missing or implausible data of survival status or respiratory support at 36 weeks’ postmenstrual age (PMA) were excluded. Outcomes and definitions Our primary outcome was a composite rate of BPD or mortality at 36 weeks’ PMA. Secondary outcomes included the severity of BPD, mortality at discharge, major comorbidities and clinical resources utilisation among survivors at 36 weeks’ PMA. In this study, we applied a modern, evidence‐based definition of BPD proposed by Jensen et al. in 2019.14 This criterion defined and categorised BPD severity according to the mode of respiratory support administered at 36 weeks’ PMA or discharge to home, if earlier: no BPD, no respiratory support; Grade 1, nasal cannula ≤2 L/min; Grade 2, nasal cannula >2 L/min or noninvasive positive airway pressure and Grade 3, invasive mechanical ventilation (IMV). Variables used in this study were defined in accordance with the protocols stipulated in the CHNN abstractor’s manual.12 The Papile criteria were used to grade intraventricular haemorrhage.15 Periventricular leukomalacia (PVL) was defined as the presence of periventricular cysts on cranial ultrasonography or MRI. Necrotising enterocolitis was defined as stage II or above, in accordance with the Bell criteria.16 Sepsis was defined as a positive blood or cerebrospinal fluid culture and antibiotic therapy or intent of antibiotic therapy for 5 days.17 Surgical treatment of patent ductus arteriosus included both ligation and cardiac catheter occlusion. Reported treatments for retinopathy of prematurity were retinal cryosurgery, laser surgery or anti-vascular endothelial growth factor drug therapy. Pulmonary hypertension was diagnosed in preterm infants at or beyond 36 weeks’ PMA by the presence of one or more of the following echocardiographic findings: bidirectional or right-to-left shunt at the patent foramen ovale or ductus arteriosus, estimated systolic pulmonary arterial pressure >40 mm Hg and/or >50% of the systemic systolic blood pressure, every degree of end-systolic flattened or left-deviated of the interventricular septum.18 IMV included conventional mechanical ventilation and high‐frequency ventilation via endotracheal tubes. Noninvasive ventilation (NIV) included high‐flow nasal cannula, nasal continuous positive airway pressure (NCPAP), bi‐level positive pressure, nasal intermittent positive pressure ventilation and nasal high‐frequency ventilation. Under the fee-for-service payment system in China, the government sets the fee structures and controls the pricing of each medical service item in public hospitals. The basic health insurance coverage (Urban-Employee Basic Medical Insurance and Urban-Rural Resident Basic Medical Insurance) has reached over 95% of the Chinese population nowadays. However, in the basic medical scheme, only about half of the total NICU charges can be reimbursed.19 Total hospitalisation charges in this study referred to the sum of direct medical expenditures paid by the patient’s family during hospitalisation according to a government-set fee schedule, including out-of-pocket expenses and later reimbursement from all kinds of insurances and subsidies. Previous studies had demonstrated high rates of discharge against medical advice (DAMA) among VPIs in China, which significantly impacted mortality and morbidity rates.20 Therefore, in our study, we specifically addressed this issue by separately listing the data for DAMA patients. We were unable to determine the survival of these infants after discharge due to the lack of follow-up data. We used the predefined criteria to make assumptions about survival of DAMA infants. If DAMA with advanced life support, which referred to requiring invasive or noninvasive mechanical ventilation, inotropes infusion or total parenteral nutrition on the day of discharge, we assumed that these infants would not survive after discharge.11 Statistical analysis Descriptive statistics were presented as frequencies (%) for categorical variables. Normally distributed continuous variables, such as birth weight and GA, were presented as mean (SD), while skewed data were presented as median (IQR), including length of stay, duration of ventilation and charges. Healthcare burden was compared between infants with and without BPD and across different BPD severity levels. Group comparisons between infants with and without BPD were made using Pearson’s χ2 test for categorical variables and Student’s t-test for normally distributed continuous variables or Mann-Whitney U test for variables with non-normal distributions. Within the BPD-graded population (including Grade 1, Grade 2 and Grade 3), trends across BPD grades were assessed by Cochran-Armitage trend test for categorical variables and by linear or median regression for continuous variables. To assess the association between BPD and key burden outcomes (including length of stay, hospital charges, death and DAMA), the generalised estimating equation models (adjusted by confounders) with random effect on site were applied, and OR with 95% CI or mean difference (MD) with 95% CI were estimated. All models were adjusted for GA, small for GA, male, multiple pregnancy, caesarean delivery, antenatal steroids and inborn status. All analyses were based on non-missing values of covariates and outcomes. Missing data were not imputed. P values <0.05 were considered statistically significant. Analyses were completed using SAS statistical software V.9.4 (SAS Institute). Results Patient characteristics After exclusion of ineligible infants (figure 1), the study cohort consisted of 17793 VPIs (GA range, 21+0–31+6 weeks) with a mean (SD) birth weight of 1268 (332) g, of whom 9998 (56.6%) were male. Maternal and infant characteristics of the excluded cohort were shown in online supplemental eTable 1. Of the enrolled infants, 568 (3.2%) died before 36 weeks’ PMA, 1729 (9.7%) were DAMA before 36 weeks’ PMA. Among survivors at 36 weeks’ PMA, 9895 (63.9%) were classified as no BPD, 2751 (17.8%) developed Grade 1 BPD, 2634 (17.0%) developed Grade 2 BPD and 216 (1.4%) developed Grade 3 BPD. Overall, mortality before 36 weeks’ PMA (including DAMA with advanced life support) or BPD was 42.5% (7566/17 793) during the study period. Table 1 shows the demographics and perinatal characteristics of the study cohort, stratified by BPD severity. Study cohort flow diagram. BPD, bronchopulmonary dysplasia; CHNN, Chinese Neonatal Network; DAMA, discharged against medical advice; PMA, postmenstrual age. Table 1Maternal and infant characteristics of the study cohortCharacteristics*No BPDBPDGrade 1 BPDGrade 2 BPDGrade 3 BPDDied before 36 weeks’ PMADAMA before 36 weeks’ PMATotal(n=9895)(n=5601)(n=2751)(n=2634)(n=216)(n=568)(n=1729)(N=17 793)GA, mean (SD), wk29.6 (1.4)28.5 (1.8)28.8 (1.7)28.2 (1.9)27.9 (2.0)26.8 (2.4)27.7 (2.2)29.0 (1.9)Birth weight, mean (SD), g1386 (297)1149 (315)1214 (282)1085 (288)1044 (288)951 (312)1092 (358)1268 (332)Small for gestational age578/9820 (5.9)732/5554 (13.2)282/2727 (10.3)408/2613 (15.6)42/214 (19.6)60/564 (10.6)157/1718 (9.1)1527/17 656 (8.6)Male5366/9820 (54.6)3319/5554 (59.8)1598/2727 (58.6)1579/2613 (60.4)142/214 (66.4)342/564 (60.6)971/1718 (56.5)9998/17 656 (56.6)1 min Apgar score ≤3277/9809 (2.8)400/5543 (7.2)154/2723 (5.7)219/2604 (8.4)27/216 (12.5)138/561 (24.6)241/1695 (14.2)1056/17 608 (6.0)5 min Apgar score ≤341/9770 (0.4)75/5477 (1.4)28/2693 (1.0)42/2570 (1.6)5/214 (2.3)46/552 (8.3)48/1661 (2.9)210/17 460 (1.2)Maternal age >35 years1788/8350 (21.4)1029/4599 (22.4)477/2276 (21.0)508/2151 (23.6)44/172 (25.6)94/466 (20.2)244/1425 (17.1)3155/14 840 (21.3)In vitro fertilisation1970/9835 (20.0)1250/5546 (22.5)626/2720 (23.0)577/2611 (22.1)47/215 (21.9)181/560 (32.3)351/1710 (20.5)3752/17 651 (21.3)Multiple pregnancy2850/9895 (28.8)1790/5601 (32.0)866/2751 (31.5)856/2634 (32.5)68/216 (31.5)220/568 (38.7)585/1729 (33.8)5445/17 793 (30.6)Caesarean delivery6655/9888 (67.3)3596/5600 (64.2)1759/2750 (64.0)1697/2634 (64.4)140/216 (64.8)279/568 (49.1)902/1728 (52.2)11432/17 784 (64.3)Premature rupture of membrane >24 hours1849/9890 (18.7)826/5589 (14.8)452/2747 (16.5)348/2626 (13.3)26/216 (12.0)80/568 (14.1)256/1727 (14.8)3011/17 774 (16.9)Chorioamnionitis1438/9024 (15.9)753/4925 (15.3)387/2428 (15.9)334/2318 (14.4)32/179 (17.9)86/498 (17.3)214/1529 (14.0)2491/15 976 (15.6)Hypertensive disorder complicating pregnancy2118/9823 (21.6)1375/5533 (24.9)615/2719 (22.6)707/2601 (27.2)53/213 (24.9)86/561 (15.3)311/1705 (18.2)3890/17 622 (22.1)Gestational diabetes mellitus2378/9822 (24.2)1176/5543 (21.2)593/2727 (21.7)540/2602 (20.8)43/214 (20.1)89/559 (15.9)266/1705 (15.6)3909/17 629 (22.2)Antenatal steroids8327/9560 (87.1)4405/5279 (83.4)2179/2577 (84.6)2068/2501 (82.7)158/201 (78.6)384/534 (71.9)1259/1635 (77.0)14375/17 008 (84.5)Inborn8229/9895 (83.2)4184/5601 (74.7)2046/2751 (74.4)1984/2634 (75.3)154/216 (71.3)409/568 (72.0)1386/1729 (80.2)14208/17 793 (79.9)*Data were presented as no./total no. (%) unless otherwise indicated. Denominators (total no.) varied according to the number of missing data for each variable. Information was missing as follows: small for gestational age, 137 (0.8%) infants; male, 137 (0.8%) infants; 1 min Apgar score, 185 (1.0%) infants; 5 min Apgar score, 333 (1.9%) infants; maternal age, 2953 (16.6%) infants; in vitro fertilisation, 142 (0.8%) infants; caesarean delivery, 9 (0.1%) infants; premature rupture of membrane, 19 (0.1%) infants; chorioamnionitis, 1817 (10.2%) infants; hypertensive disorder complicating pregnancy, 171 (1.0%) infants; gestational diabetes mellitus, 164 (0.9%) infants; antenatal steroids, 785 (4.4%) infants.BPD, bronchopulmonary dysplasia; DAMA, discharged against medical advice; GA, gestational age; PMA, postmenstrual age. Maternal and infant characteristics of the study cohort Data were presented as no./total no. (%) unless otherwise indicated. Denominators (total no.) varied according to the number of missing data for each variable. Information was missing as follows: small for gestational age, 137 (0.8%) infants; male, 137 (0.8%) infants; 1 min Apgar score, 185 (1.0%) infants; 5 min Apgar score, 333 (1.9%) infants; maternal age, 2953 (16.6%) infants; in vitro fertilisation, 142 (0.8%) infants; caesarean delivery, 9 (0.1%) infants; premature rupture of membrane, 19 (0.1%) infants; chorioamnionitis, 1817 (10.2%) infants; hypertensive disorder complicating pregnancy, 171 (1.0%) infants; gestational diabetes mellitus, 164 (0.9%) infants; antenatal steroids, 785 (4.4%) infants. BPD, bronchopulmonary dysplasia; DAMA, discharged against medical advice; GA, gestational age; PMA, postmenstrual age. The proportion of infants who survived to 36 weeks’ PMA and did not develop BPD of any grade increased with GA (figure 2). The rates of mortality before 36 weeks’ PMA (including DAMA with advanced life support) or Grade 3 BPD decreased from 69.7% among infants born at less than 24 weeks’ gestation to 3.6% among those born at 31 weeks’ gestation (p<0.001). Number and rates of death or DAMA before 36 weeks’ PMA and of BPD, stratified by GA. BPD, bronchopulmonary dysplasia; DAMA, discharged against medical advice; GA, gestational age; PMA, postmenstrual age. Major comorbidities As shown in table 2, infants with BPD exhibited higher rates of major comorbidities compared with those without BPD. A consistent severity gradient was evident across evaluated major comorbidities, with Grade 3 BPD infants having approximately 3–6 times higher rates than Grade 1–2 BPD cases. Table 2Major comorbidities of very preterm infants among survivors at 36 weeks’ PMA, stratified by BPD severityComorbidities*No BPDBPDP value†Grade 1 BPDGrade 2 BPDGrade 3 BPDP trend†(n=9895)(n=5601)(n=2751)(n=2634)(n=216)Tracheomalacia or bronchomalacia34/9895 (0.3)53/5601 (0.9)<0.00121/2751 (0.8)21/2634 (0.8)11/216 (5.1)<0.001Pulmonary hypertension112/9895 (1.1)244/5601 (4.4)<0.00174/2751 (2.7)138/2634 (5.2)32/216 (14.8)<0.001Surgically treated PDA19/9895 (0.2)164/5601 (2.9)<0.00128/2751 (1.0)116/2634 (4.4)20/216 (9.3)<0.001PVL‡198/9131 (2.2)215/5042 (4.3)<0.00193/2527 (3.7)111/2314 (4.8)11/201 (5.5)0.037IVH (grade III–IV)‡254/9579 (2.7)412/5368 (7.7)<0.001176/2659 (6.6)214/2501 (8.6)22/208 (10.6)0.002Surgically treated NEC121/9609 (1.3)177/5515 (3.2)<0.00159/2702 (2.2)90/2597 (3.5)28/216 (13.0)<0.001Treated ROP§160/9352 (1.7)481/5378 (8.9)<0.001142/2665 (5.3)293/2499 (11.7)46/214 (21.5)<0.001Late-onset sepsis504/9891 (5.1)632/5596 (11.3)<0.001215/2748 (7.8)359/2632 (13.6)58/216 (26.9)<0.001*Data were presented as No./total No. (%) unless otherwise indicated. Denominators (total No.) varied according to the number of missing data for each variable.†P values were determined for differences between infants with and without BPD, and P trend values were assessed for trend differences across BPD grades.‡Total proportion is among infants who underwent cranial imaging (ie, cranial sonography, MRI or CT).§Total proportion is among infants who received ROP examination.BPD, bronchopulmonary dysplasia; IVH, intraventricular haemorrhage; NEC, necrotising enterocolitis; PDA, patent ductus arteriosus; PMA, postmenstrual age; PVL, periventricular leukomalacia; ROP, retinopathy of prematurity. Major comorbidities of very preterm infants among survivors at 36 weeks’ PMA, stratified by BPD severity Data were presented as No./total No. (%) unless otherwise indicated. Denominators (total No.) varied according to the number of missing data for each variable. P values were determined for differences between infants with and without BPD, and P trend values were assessed for trend differences across BPD grades. Total proportion is among infants who underwent cranial imaging (ie, cranial sonography, MRI or CT). Total proportion is among infants who received ROP examination. BPD, bronchopulmonary dysplasia; IVH, intraventricular haemorrhage; NEC, necrotising enterocolitis; PDA, patent ductus arteriosus; PMA, postmenstrual age; PVL, periventricular leukomalacia; ROP, retinopathy of prematurity. Resources utilisation Table 3 reveals significant, severity-dependent increases in NICU resources utilisation associated with BPD. In terms of medications, treatment variations were observed, with BPD infants being more likely to use inhaled nitric oxide, systemic corticosteroids, bronchodilators, diuretics and anti-reflux medications. A similar increasing trend was identified for the evaluated drug therapies as BPD severity increased, with the exception of caffeine (P trend=0.31) and anti-reflux drugs (P trend=0.62). Table 3Resources utilisation of very preterm infants among survivors at 36 weeks’ PMA, stratified by BPD severityResources*No BPDBPDP value†Grade 1 BPDGrade 2 BPDGrade 3 BPDP trend†(n=9895)(n=5601)(n=2751)(n=2634)(n=216)Medications during hospitalisation Surfactant5246/9895 (53.0)4035/5601 (72.0)<0.0011882/2751 (68.4)1995/2634 (75.7)158/216 (73.1)<0.001 Caffeine8012/9895 (81.0)4945/5601 (88.3)<0.0012434/2751 (88.5)2329/2634 (88.4)182/216 (84.3)0.311 Inhaled nitric oxide47/9895 (0.5)177/5601 (3.2)<0.00158/2751 (2.1)95/2634 (3.6)24/216 (11.1)<0.001 Systemic corticosteroids1237/9895 (12.5)2283/5601 (40.8)<0.001973/2751 (35.4)1178/2634 (44.7)132/216 (61.1)<0.001 Bronchodilators89/9895 (0.9)175/5601 (3.1)<0.00152/2751 (1.9)96/2634 (3.6)27/216 (12.5)<0.001 Diuretics829/9895 (8.4)1728/5601 (30.9)<0.001670/2751 (24.4)953/2634 (36.2)105/216 (48.6)<0.001 Anti-reflux drugs198/9895 (2.0)264/5601 (4.7)<0.001134/2751 (4.9)111/2634 (4.2)19/216 (8.8)0.620Respiratory support during hospitalisation Any IMV3409/9895 (34.5)3891/5601 (69.5)<0.0011681/2751 (61.1)1994/2634 (75.7)216/216 (100.0)<0.001 HFV549/9895 (5.5)1231/5601 (22.0)<0.001464/2751 (16.9)674/2634 (25.6)93/216 (43.1)<0.001 CMV3217/9895 (32.5)3632/5601 (64.8)<0.0011535/2751 (55.8)1882/2634 (71.5)215/216 (99.5)<0.001 Any NIV9405/9895 (95.0)5506/5601 (98.3)<0.0012674/2751 (97.2)2634/2634 (100.0)198/216 (91.7)0.004 NHFV333/9895 (3.4)537/5601 (9.6)<0.001251/2751 (9.1)276/2634 (10.5)10/216 (4.6)0.871 NIPPV/BI‐PAP4259/9895 (43.0)3682/5601 (65.7)<0.0011750/2751 (63.6)1786/2634 (67.8)146/216 (67.6)0.002 NCPAP7930/9895 (80.1)4708/5601 (84.1)<0.0012150/2751 (78.2)2412/2634 (91.6)146/216 (67.6)<0.001 HFNC3883/9895 (39.2)3165/5601 (56.5)<0.0011353/2751 (49.2)1705/2634 (64.7)107/216 (49.5)<0.001 Cumulative duration of IMV, median (IQR), days0 (0–2)5 (0–14)<0.0013 (0–8)7 (1–19)47 (18–74)<0.001 Cumulative duration of NIV, median (IQR), days16 (8–29)38 (24–52)<0.00130 (18–42)48 (35–61)36.5 (19.5–54)<0.001 Cumulative duration of respiratory support, median (IQR), days24 (12–37)59 (44–77)<0.00154 (41–68)66 (49–85)90 (68–120.5)<0.001Treatment at discharge to home among survivors‡ Requirement of feeding tube6/9478 (0.1)13/4763 (0.3)0.0015/2503 (0.2)6/2123 (0.3)2/137 (1.5)0.083 Requirement of home oxygen therapy25/9478 (0.3)1102/4763 (23.1)<0.001677/2503 (27.0)380/2123 (17.9)45/137 (32.8)<0.001 Low-flow oxygen11/9478 (0.1)733/4763 (15.4)<0.001463/2503 (18.5)242/2123 (11.4)28/137 (20.4)<0.001 NIV4/9478 (0.04)87/4763 (1.8)<0.0011/2503 (0.04)74/2123 (3.5)12/137 (8.8)<0.001 HFNC2/9478 (0.02)57/4763 (1.2)<0.0010/2503 (0)46/2123 (2.2)11/137 (8.0)<0.001 NCPAP/NIPPV2/9478 (0.02)30/4763 (0.6)<0.0011/2503 (0.04)28/2123 (1.3)1/137 (0.7)<0.001*Data were presented as No./total No. (%) unless otherwise indicated. Denominators (total No.) varied according to the number of missing data for each variable.†P values were determined for differences between infants with and without BPD, and P trend values were assessed for trend differences across BPD grades.‡Total proportion is among infants who survived at discharge to home.BI‐PAP, bi‐level positive pressure; BPD, bronchopulmonary dysplasia; CMV, conventional mechanical ventilation; HFNC, high‐flow nasal cannula; HFV, high‐frequency ventilation; IMV, invasive mechanical ventilation; NCPAP, nasal continuous positive airway pressure; NHFV, nasal high‐frequency ventilation; NIPPV, nasal intermittent positive pressure ventilation; NIV, noninvasive ventilation; PMA, postmenstrual age. Resources utilisation of very preterm infants among survivors at 36 weeks’ PMA, stratified by BPD severity Data were presented as No./total No. (%) unless otherwise indicated. Denominators (total No.) varied according to the number of missing data for each variable. P values were determined for differences between infants with and without BPD, and P trend values were assessed for trend differences across BPD grades. Total proportion is among infants who survived at discharge to home. BI‐PAP, bi‐level positive pressure; BPD, bronchopulmonary dysplasia; CMV, conventional mechanical ventilation; HFNC, high‐flow nasal cannula; HFV, high‐frequency ventilation; IMV, invasive mechanical ventilation; NCPAP, nasal continuous positive airway pressure; NHFV, nasal high‐frequency ventilation; NIPPV, nasal intermittent positive pressure ventilation; NIV, noninvasive ventilation; PMA, postmenstrual age. Among BPD infants, 69.5% received IMV whereas 98.3% received any form of NIV at some point during hospitalisation, with the highest utilisation rate of NCPAP. The median duration of respiratory support increased with greater BPD severity: 54 (41–68) days for Grade 1; 66 (49–85) days for Grade 2; 90 (68–120.5) days for Grade 3. Respiratory support requirements among survivors discharged home varied by BPD severity level (table 3). Only 0.3% (25/9478) of infants without BPD were treated with supplemental oxygen, whereas 22.8% (1057/4626) of those with Grade 1–2 BPD and 32.8% (45/137) of those with Grade 3 BPD received oxygen therapy at the time of discharge. As shown in table 4, among survivors at 36 weeks’ PMA, BPD infants had significantly longer hospital stays than those without BPD (adjusted MD, 11.7; 95% CI 9.2 to 13.8; p<0.001). They also incurred higher total hospitalisation charges (thousand CNY; adjusted MD, 3.2; 95% CI 2.7 to 3.8; p<0.001) and charge per day (CNY; adjusted MD, 171.2; 95% CI 80.5 to 262.0; p<0.001). Length of stay and hospital charges increased with BPD severity (P trend <0.001), and similar variations were found after adjusted confounders such as GA. Length of stay and hospital charges were further described stratified by GA in online supplemental eTable 2. Trend differences in daily charges across BPD severity levels were specifically observed in infants born at 25–31 weeks’ gestation. Table 4Length of stay, hospital charges and outcomes of very preterm infants among survivors at 36 weeks’ PMA, stratified by BPD severityResourcesNo BPDBPDBPD vs no BPDGrade 1 BPDGrade 2 BPDGrade 3 BPDGrade 2 vs Grade 1Grade 3 vs Grade 1(n=9895)(n=5601)Adjusted OR/MD (95% CI)*(n=2751)(n=2634)(n=216)Adjusted OR/MD (95% CI)*Adjusted OR/MD (95% CI)*Length of stay, median (IQR), days44(34–56)67(52–84)11.70(9.20 to 13.80)61(48–73)74(58–92)95(74.5–121)6.54(3.57 to 9.50)33.80(25.30 to 42.20)Total charges, median (IQR), thousand CNY73(52–103)127(91–177)3.24(2.69 to 3.80)107(80–142)149(110–202)244(180–301)2.81(1.99 to 3.64)11.50(7.43 to 15.50)Charge per day, median (IQR), CNY1714(1409–2048)1975(1638–2361)171.20(80.50 to 262.00)1812(1516–2174)2098(1769–2505)2430(1952–2955)321.30(188.40 to 454.20)561.90(339.70 to 784.10)Death, n (%)†11/9895(0.1)75/5601(1.3)8.43(3.96 to 18.00)8/2751 (0.3)30/2634(1.1)37/216(17.1)3.47(1.95 to 6.19)60.90(24.80 to 149.40)DAMA, n (%)†167/9895(1.7)321/5601(5.7)3.57(2.30 to 5.53)120/2751(4.4)162/2634(6.2)39/216(18.0)1.55(1.15 to 2.08)5.75(3.12 to 10.60) DAMA with advanced life support, n (%)7/9895(0.1)54/5601(1.0)10.60(4.41 to 25.50)7/2751(0.3)26/2634(1.0)21/216(9.7)3.59(1.91 to 6.76)39.80(14.00 to 113.10) DAMA without advanced life support, n (%)160/9895(1.6)267/5601(4.7)2.91(1.87 to 4.52)113/2751(4.1)136/2634(5.2)18/216(8.3)1.24(0.93 to 1.65)2.07(1.11 to 3.88)*The generalised estimating equation models with random effect on site were applied and odds ratio (95% CI) or mean difference (95% CI) were estimated. All models were adjusted by confounders (GA, small for gestational age, male, multiple pregnancy, caesarean delivery, antenatal steroids and inborn).†Total proportion is among infants who survived at 36 weeks’ PMA.BPD, bronchopulmonary dysplasia; CNY, China Yuan; DAMA, discharged against medical advice; GA, gestational age; MD, mean difference; PMA, postmenstrual age. Length of stay, hospital charges and outcomes of very preterm infants among survivors at 36 weeks’ PMA, stratified by BPD severity The generalised estimating equation models with random effect on site were applied and odds ratio (95% CI) or mean difference (95% CI) were estimated. All models were adjusted by confounders (GA, small for gestational age, male, multiple pregnancy, caesarean delivery, antenatal steroids and inborn). Total proportion is among infants who survived at 36 weeks’ PMA. BPD, bronchopulmonary dysplasia; CNY, China Yuan; DAMA, discharged against medical advice; GA, gestational age; MD, mean difference; PMA, postmenstrual age. Outcome at discharge Among infants who survived at 36 weeks’ PMA, the mortality rate at discharge (including DAMA with advanced life support) increased with BPD severity (P trend <0.001), with rates of 0.2% (18/9895) for infants without BPD, 0.5% (15/2751) for Grade 1 BPD, 2.1% (56/2634) for Grade 2 and 26.9% (58/216) for Grade 3 (table 4). Discussion In this large national contemporary cohort of VPIs in China, we provided a comprehensive description of benchmark data on the healthcare burden of BPD defined according to the severity‐based diagnostic criteria proposed by Jensen et al.14 BPD was first described by Northway et al in 1967.21 Over subsequent decades, the diagnostic criteria have undergone significant evolution to accommodate changes in the characteristics of high‐risk infants, further understanding of the pathophysiology, and development of contemporary neonatal respiratory care with new modes of ventilation.22 The reason we adopted the Jensen BPD grading system was that this evidence-based definition could best predict late death or serious respiratory morbidity at 18–26 months’ corrected age, and it was straightforward to apply.14 23 Moreover, Kurihara et al24 found that the BPD severity grading by Jensen had the best prediction of health economics impact through the first year of life as a surrogate of disease complexity and resource utilisation in VPIs. BPD remains one of the most common and significant complications of VPIs. In total, 11% of infants in our cohort died before 36 weeks’ PMA (including DAMA with advanced life support), and 31.5% developed BPD. The majority of BPD infants were classified as having a severity of Grade 1–2, while 1.2% developed Grade 3. The rate of BPD in our cohort was similar to that reported in other multicentre studies in China,6 11 25 where BPD was defined as oxygen use at 36 weeks’ PMA. A retrospective cohort study of infants born from 22 to 29 weeks’ gestation from the Vermont Oxford Network showed that 10.3% of infants died before 36 weeks’ PMA in the USA, and 40.6% met the criteria for BPD (36.9% developed Grade 1–2; 3.7% developed Grade 3) according to the same Jensen definition.26 The differences might be partly explained by the variations in the GA of the infants studied. The rates of death before 36 weeks’ PMA or Grade 3 BPD decreased dramatically with increasing GA: a result that was reported consistently in studies of VPIs.4 26 27 In the present study, BPD infants had substantial healthcare burden in their hospital of birth, including longer hospital stay, higher hospital charges, greater use of medical resources, and more frequent comorbidities. Previous studies reported that infants with BPD required hospital stays 2–4 weeks longer than those without BPD.27 28 Another CHNN study showed that BPD was associated with increasing length of stay with adjusted mean ratio 1.25 (1.23–1.28) among VPIs.29 We found that infants with severe BPD frequently spent over 3 months in NICUs. Their PMA at discharge was postponed by more than 6 weeks compared with infants without BPD. Prolonged stays in the NICU were associated with other problems such as delayed establishment of developmental milestones, disturbed familial emotional bonding, increased risk of hospital-acquired nosocomial infections, and substantial socioeconomic burdens on families and societies.29 It was estimated that the actual total number of bed days occupied by BPD infants in our study exceeded the rated number of bed days, which meant that hospitals had opened extra beds to accommodate these infants, greatly increasing the workload of medical staff and consuming medical resources. Analysis of the total charges to treat BPD infants showed results in line with those for length of stay. We found that the initial hospitalisation charges for infants with BPD were approximately 3.3 times the national average disposable income per capita during the study period. Given that CHNN participating centres provided care for approximately 10% of all VPIs in China, it was estimated that the total annual initial hospitalisation charges for BPD infants across the country amounted to approximately 3.5 billion CNY. In a systematic review of the healthcare costs associated with BPD, most studies reported high costs during the initial hospitalisation after birth (range: Int$21 392-Int$1 094 509 per child, in 2019).9 Data from California Perinatal Quality Care Collaborative revealed that BPD was associated with a 1.52-fold higher cost for initial hospitalisation compared with infants without BPD.30 Prior studies showed that after controlling for birth weight, GA, and sociodemographic characteristics, BPD was associated with the highest increase in costs in comparison to other prematurity morbidities.10 30 31 After we adjusted for possible confounding factors that might contribute to the outcomes, the large hospital charge differences persisted and remained statistically significant. In addition, we analysed charge per day to account for varying length of stay, and it was found to be significantly higher for Grade 3 in comparison to Grade 1–2 or no BPD. This finding suggested that the increased charges are not merely related to prolonged length of stay, but also associated with increased resource use, which was evident in our analysis of NICU resources utilisation categories. Healthcare burden varied dramatically according to BPD severity. However, few studies had evaluated the correlation with disease severity,26 32 not to mention applying the latest evidence-based definition. Our data confirmed that infants with Grade 3 BPD experienced substantial mortality risk and a high burden of prognostically important neonatal comorbidities. The significant disparities in outcomes and healthcare utilisation rates across BPD severity grades underscored the necessity for future studies to present BPD incidence stratified by severity level, rather than merely distinguishing between its presence or absence. The health and financial impact of having BPD extended far beyond the initial birth hospitalisation,33 34 although the initial hospitalisation accounted for the overwhelming majority (84%–96%) of the total costs up to 5 years of age.35 In a large population-based cohort study (EPIPAGE-2), Tréluyer et al found that even at age 5–6 years, BPD was independently associated with more frequent rehospitalisation and developmental support.36 Our findings also revealed the complex post-discharge care requirements associated with the increasing severity of BPD. In our cohort, home oxygen therapy was used in 32.8% of Grade 3 BPD infants. As a result, caregivers might encounter challenges in managing numerous medicines and medical devices, and they were required to use various healthcare services, including visits to specialists and emergency care.37 The available data underlined the economic strain BPD placed on healthcare resources, which was further exacerbated by the emotional and physical toll it took on families.33 38 Thus, it is crucial to implement standardised, regular follow-up schedules. Strengths and limitations To our knowledge, this was the first study to characterise the short-term healthcare burden of VPIs with BPD in China using a large, high-quality source of data collected prospectively. This study was also unique in providing comprehensive burden estimates for different severities of BPD using the latest definition. Our findings highlighted the need for efficient resource allocation and the development of cost-effective interventions to improve outcomes in developing countries. These quantitative outcomes could also serve as a basis for future cost-effectiveness and cost-utility analyses. However, several limitations warrant mention. First, our data were hospital‐based from tertiary NICUs rather than population‐based, potentially limiting generalisability. The mortality rate might be underestimated because infants who died in the delivery room were not included. Second, as an observational study, we were limited in identifying causal relationships. Third, we lacked follow-up data, thus information regarding long-term medical burden such as readmissions, outpatient visits, medication use and rehabilitation after discharge was not reported. Fourth, the reported health economic outcomes were limited to hospital charges only, as it was difficult to derive the precise cost for each patient from the medical records. So far, there has been no generally recognised charge-to-cost conversion method in the Chinese medical payment system. Moreover, our results might underestimate the actual economic burden because the direct non-medical costs and indirect costs were not taken into account.39 Conclusion In conclusion, more than 40% of VPIs in China either died before 36 weeks’ PMA or developed BPD. Greater BPD severity was associated with increased comorbidities, higher in-hospital mortality and greater utilisation of healthcare resources during initial hospitalisation. These findings emphasised the ongoing need to develop cost-saving strategies to reduce the risk and severity of BPD and improve overall care. Supplementary material