The purpose of this study was to characterize the use of HOT in preterm infants with BPD in China based on the Chinese Neonatal Network data, so as to generate beneficial information in the design of interventions and the formulation of appropriate HOT policy.

In our cohort, we identified clinical risk factors associated with higher use of HOT for preterm infants with BPD that were similar to those of previous studies—including small for gestational age at birth, more doses of surfactant needed, shorter NICU stays, and a longer duration on invasive ventilation.

Table 1 shows that low incidence of severe NEC in babies discharged with HOT was noticed. Actually, the percentage of severe NEC and death in infants who met the exclusion criteria was higher than that of the study population, suggesting that low incidence of NEC on HOT may be attributed to early death caused by NEC. Among infants with HOT, the length of stay in NICU in the NEC group (73.5 days) was longer than that of the non-NEC group (58.5 days), which indicated that patients with severe NEC had a chance to receive longer respiratory support in the hospital. As a result, the longer length of stay allowed time for recovery from BPD, and thus lower HOT use was observed.

Although previous studies indicated that PDA ligation and SGA were related to a higher proportion of HOT use [11, 12], in our cohort, these indices were not significantly correlated with higher odds of HOT use after multivariable analysis. Among our patients with HOT, only nine patients underwent PDA ligation and 29 patients were born SGA. Our small sample size may have led to bias in our results; so future studies should require the enrollment of a larger number of infants to explore the potential risk factors involved in producing a higher rate of HOT use in China.

HOT has been indicated for some preterm infants with severe bronchopulmonary dysplasia (BPD) upon discharge since the 1970s [6], and studies have shown that HOT engenders potential familial and economic benefits. For example, HOT can reduce the average length of a hospital stay by 1.5 weeks [11], reduce economic burden [6], prevent the impact of chronic hypoxemia, enhance the overall quality of life in children, and improve the impact of the family environment on family members of infants without increased risk of readmission [27].

There is currently no worldwide consensus for HOT use in preterm infants with BPD. The recent guidelines from the Thoracic Societies in Britain, Australia, New Zealand, and the United States differ markedly on some issues. For example, the delineation of relatively stable newborns who can be discharged with home oxygen remains controversial [28]. According to a survey, there was no national guideline in Germany, and this resulted in a wide range of SpO2 cutoff values indicated for HOT in different facilities that ranged from 80% to 94% [29]. The reported rate of HOT use in other countries varied from 13.2 to 65.2% among BPD preterm infants, with wide variations in institutional proportions from 7% to 95% [8, 11, 12]. In our cohort, HOT was used in 26.8% of preterm Chinese infants with BPD, which was lower than the proportions of HOT use for infants with BPD in the United States and United Kingdom, but higher than those for South Korea, New Zealand, and Australia [8, 9].

With respect to the factors that contribute to the wide variation in institutional proportions of HOT, physician as well as parent preferences regarding HOT for infants with BPD might play a significant role [30]. In our cohort there were more than 1/3 of hospitals with an observed use of HOT that was significantly greater or less than expected, with a proportion of HOT that varied between 0 and 89% among infants with BPD in different institutes. This result is similar to the wide institutional variation noted in other reports [12]. Future study is needed to further address the aforementioned contributions to the use of HOT in China.

In addition, the distribution of social resources occupies an important role in HOT. The premature infants who were discharged with HOT needed more support with respect to community health care. There should be special nursing teams pursuing long-term follow-up, professional nurses for regular sleep research, and medical and health professionals for additional home visits to comprehensively evaluate the family environment. When parents suspect that their baby has dyspnea, pediatric community medical staff can then evaluate the neonates and provide general respiratory support [31].

Regional pediatric community health services vary in China. A survey in 2014 showed that the proportion of children under the age of 15 who used community services was only 2.4%, that the training of community health-institution doctors in children's health care was limited, and that most community-service institutions did not provide inpatient services [32].

Distinct from our initial assumption, we have now recognized that the regional proportion of HOT use was negatively correlated with the level of provincial economic development. Spooner et al. identified key differences that exist across patient and hospital characteristics with respect to discharged against medical advice and found that lack of health insurance (OR 3.78; 95% CI 3.62–3.94) was one of the major predictors [33]. We speculated that the negative correlation in our cohort might be related to the fact that parents in cities with a higher economic level might possess satisfactory insurance coverage or financial support for hospitalization expenses, and thus prefer to stay in the hospital for a longer period of time. In contrast, parents in cities exhibiting a lower economic level might not have sufficient insurance coverage or financial support for their hospital expenses and prefer early discharge with HOT. The causes underlying the relationship between institutional use of HOT and GDP per capita also warrant additional clarification in future studies. Given this intriguing phenomenon, efforts should be made to improve local community medical-service systems to provide sufficient support for preterm infants discharged with HOT. In addition, local—but not national—HOT guidelines are needed in countries that exhibit a wide variation in HOT use and in those with disparate regional economic levels, such as China.

Supplemental Figs. 2 and 3 suggested that more emission of particulate matter was associated with both lower GDP per capital and more HOT use. The air pollution may slow infants’ respiratory recovery at home, which may be related to increase of oxygen use at home. However, this relationship may be biased due to potential confounders, such as smoking status of parents and air quality of home. In future, we will further investigate the association of HOT use with the air pollution due to lack of data on air quality at home.

This is the first study to characterize the use of HOT in preterm infants with BPD in China. The strengths of this study included its large sample size and our valid and reliable data collection system. Data were collected prospectively from 57 tertiary centers located in 27 regions of China and adequately represented the characteristics of HOT use in China, thus avoiding bias to an extent.

However, the findings from our study possess several limitations that require consideration when interpreting the data. First, the survival rate for premature infants who were very small for gestational age or showed very low weight at birth was low. Investigators therefore need to enroll more infants in future to explore the actual use of HOT in such groups of Chinese infants. Second, because treatment of BPD varied at different centers and there are no extant guidelines for HOT, this might have also resulted in bias. Finally, long-term follow-up is required to compare the prognosis of infants discharged with HOT to those without HOT.

In conclusion, we demonstrated that the use of HOT varied across China and was negatively correlated with economic development at the provincial level. These findings suggest that local HOT guidelines are needed in countries such as China that display a wide variation in HOT use and different regional economic levels.