This study investigated the critical postnatal periods for the development of severe ROP influenced by oxygen supplementation or invasive ventilation in the setting of a target saturation of 90–95% using an institutional protocol. TWAFiO2 or invasive ventilation was associated with ROP treatment in the first 7 weeks PNA. In terms of PMA, invasive ventilation was associated with ROP from PMA 26–31 weeks, whereas no association between TWAFiO2 and treated ROP was found.

One of the notable findings elucidated in the present study pertains to the association between invasive ventilation and severe ROP even after adjusting for the level of oxygen given to the patients. While the existing literature has underscored the link between mechanical ventilation and ROP, prior investigations have not scrutinized the independent impact of invasive ventilation, distinct from the quantity of oxygen concentration delivered [15,16,17]. Although non-invasive ventilation can effectively support patient’s respiratory support, there are disadvantages of oxygenation supplementation compared with invasive ventilation in several aspects [18]. Patient-dependent factors, such as minute ventilation, amount of mouth breathing, leakage around the patient-device interface, and the relative duration of inspiration and expiration, can influence the fractional inspired oxygen concentration, thereby compromising the intended oxygen delivery efficacy [19, 20].

As the present study investigated the level of oxygen given to the patients with regard to the development of severe ROP, considering the mode of ventilation that affects effective oxygen delivery was crucial in this study. Our findings were also supported by the results of an ad hoc analysis, comparing the incidence of severe ROP between patients receiving invasive ventilation and non-invasive ventilation in the subgroup stratified by weekly TWAFiO2 (< 0.3 and ≥ 0.3). Across both subgroup delineations, there was a higher incidence of severe ROP in the group receiving invasive ventilation early in life (Supplementary Table 2, Additional file 1).

After the AAP stated that a lower target saturation strategy for the prevention of ROP could not account for the higher incidence of death in preterm infants, several studies have attempted to demonstrate the effect of a graded saturation target strategy in the prevention of ROP without increasing mortality in preterm infants [10, 12, 13]. The hypothesis that the target saturation varies with age is based on the pathophysiology of ROP, which comprises two phases. In phase I, early hyperoxia induces attenuation of retinal vascular growth and vaso-obliteration. This is followed by the hypoxia-induced phase II vasoproliferation in the avascular areas of the retina [7, 8]. However, these studies evaluated specific periods of oxygen restriction based on the PMA rather than on the PNA. A retrospective study in 2016 applied a graded SpO2 target based on PMA (83–89% until 326/7 weeks, 90–94% until 356/7 weeks, and > 94% at ≥ 36 weeks PMA) and showed decreased rates of severe ROP and laser surgery without increasing mortality [12]. Another study by Shukla et al. compared a graded SpO2 target (85–92% until 336/7 weeks PMA and 95% at ≥ 34 weeks PMA) with a constant target (91–95%) [13]. The results of the present study also showed that setting the period before and after PMA to 32–33 weeks might be rational, as severe ROP was associated with higher level of oxygen supplementation or invasive ventilation before 32 weeks PMA. Moreover, our study investigated the period of interest in terms of PNA. PNA-based evaluation could provide additional information, as preterm infants of various GAs were included in the study. As the oxygen concentration in the atmosphere is higher than that in the intrauterine environment, it can be assumed that virtually all preterm infants are exposed to oxygen stress from the moment they are born.

In recent years, there has been an increasing focus in preclinical and clinical research on target therapies aimed at decoupling oxygenation and vascularization [21, 22]. Given the variability in VEGF levels between phase I and phase II of ROP pathogenesis, it is valuable to identify specific target periods where therapeutic interventions can maximize their efficacy in decoupling the effects of oxygen on vascular development. Hence, the results of the current study may provide insights into determining the optimal timing of these therapeutic agents in future applications.

The first limitation of the present study is that different oxygen saturations were not compared. Instead, we sought to evaluate the effect of the actual amount of oxygen supplementation and invasive ventilation on severe ROP. The analysis of these factors is a different approach from previous studies that set different oxygen saturation targets [10, 12, 13]. However, SpO2 can be affected by other systemic conditions such as pulmonary hypertension, blood pressure, or lung disease. In fact, the treated ROP group in this study tended to have lower mean SpO2 values, even though they were supplemented with higher oxygen levels (Supplementary Table 3, Additional file 1). Moreover, because oxygen can be delivered both invasively and noninvasively, and the level of oxygen can be determined accordingly, the association of invasive ventilation at a specific PNA or PMA with the development of severe ROP was also considered. Considering the graded oxygen saturation target strategy, the first seven weeks of life or up to PMA 31 weeks could be a candidate period for a lower saturation target. As infants who died before discharge were excluded from the study population, the concerns regarding mortality in the lower saturation target group, as seen in previous studies such as the SUPPORT trial, could not be addressed in this study. The primary objective of the present study was to investigate the vulnerable period in the development of severe ROP in preterm infants.

Another limitation of this study is that we did not calculate the actual amount of oxygen delivered during non-invasive ventilation. However, the aim of this study was not to determine the exact amount of oxygenation that increases the risk of severe ROP but rather to investigate the period of significance affected by oxygen supplementation and invasive ventilation. Therefore, we adjusted for invasive ventilation in each period.