Mechanical ventilation (MV) has been widely employed to enhance outcomes in pediatric intensive care unit (PICU) and respiratory care in very preterm newborns (1, 2). However, weaning has long been a cause of concern. Prolonged MV is related with infant ventilator-associated lung injury and neurodevelopmental impairment, whereas premature weaning is associated with extubation failure (EF) and an increase risk of reintubation (3, 4). Moreover, many studies have linked EF and subsequent reintubation to increased mortality and bronchopulmonary dysplasia (BPD) in premature newborns (5–7). Therefore, choosing the best weaning interval has a substantial clinical impact on the prognosis of newborns with MV.

There are currently no standard guidelines to guide clinical weaning of infants with MV (8), and weaning timing is primarily determined by the attending physician based on clinical evaluation, ventilator parameters, and blood gas results (9), resulting in approximately 30% of premature infants failing extubation (10). As a result, new precise indications for predicting successful weaning timing in children with MV are required.

Because of its non-invasive, convenient, and real-time features, ultrasound is appropriate for bedside assessment of MV patients in the ICU (11, 12). Diaphragmatic function and pulmonary ventilation status are two components of ultrasound assessment of whether a patient can be weaned, with the former proven by the diaphragmatic excursion (DE) and diaphragm thickening fraction (DTF) (13), and the latter by the lung ultrasound score (LUS) (14). These three signs have been demonstrated to be accurate predictors of weaning in adult MV patients (15). Short DE distances and high DTF scores indicate poor diaphragm function, whereas high LUS suggests inadequate pulmonary ventilation, implying a significant risk of EF (16). However, because infant physiology and anatomy differ from those of adults, more research is needed to determine whether ultrasound technology can be used to anticipate weaning in infants with MV.

The purpose of this meta-analysis is to establish the accuracy of three diaphragm and lung ultrasound indicators: DE, DTF, and LUS in predicting weaning outcomes in infants with MV.

Although lung and diaphragmatic ultrasound have been frequently used to predict weaning in adult MV patients (32, 33, 34), their application in newborns has received little attention. In this meta-analysis, we discovered that lung and diaphragm ultrasound were also helpful techniques for predicting newborn weaning outcomes by examining the weaning outcomes of 828 infants who received MV. LUS performed the best overall, with sensitivity and specificity of 0.88 and 0.81, respectively. DTF outperformed DE in the diaphragm ultrasound evaluation index, and AUSROC was 0.86 and 0.79, respectively. It is important to note that the sensitivity of DE and DTF varies greatly. Moreover, some studies were highly susceptible to bias, particularly in patient selection, raising concerns about applicability.

LUS is an important method for evaluating lung ventilation loss, which represents a drop in lung capacity available for gas exchange and can be used to forecast EF (35, 36). Our data also shown a relationship between high LUS and EF, which is consistent with the findings for adults (37, 38). The unique truncation point, however, cannot be identified due to the variety of LUS methods, which include 6-zone, 10-zone, 12-zone, and 14-zone, but LUS has a high AUSROC of 0.90 (95%CI, 0.87–0.92) when assessed using the different cut-off points of each study. Although Jiang et al. (31) believe that prediction accuracy increases with partition size, the results of our subgroup analysis reveal that prediction accuracy for 6-zone and other zones is not significantly different, with AUSROC both above 0.90. We discovered that LUS was more accurate at predicting weaning in newborns than it was at predicting weaning in adults (AUSROC, 0.77) (15). Additionally, similarly to our work, LUS exhibited a good ability to predict weaning in several adult studies, with cut-off levels ranging from 10 to 14 (24, 39). In our subgroup analysis, subgroups with cut-off points higher than 15 had a DOR of 43.54 (95%CI, 21.47–88.28), greater than subgroups with cut-off points lower than 15, which had a DOR of 22.96 (95%CI, 11.67–45.18). The inclusion of groups with high extubation success rates and tight study methods were also sources of variability and variables with increased sensitivity, according to meta-regression.

DE and DTF are two key diaphragm ultrasonography measurements for predicting weaning results in MV patients, with DE being connected to inspiratory volume and DTF being a response measure of diaphragmatic inhalation effort (40, 41). DE and DTF worked well in adult studies, with greatest predictive values of 10–15 mm and 20%–36%, respectively (42). Except for one study, where the DTF cut-off point was chosen at 40.50% because the patients studied were generally healthy (28), the DTF cut-off values in the included studies were similar to those in adults. However, the ideal cut-off for DE in the studies we considered was often less than 10 mm, significantly smaller than the adult figure, with the exception of Gazi et al. (28), who used a cut-off of 12.15 mm. This could be because the diaphragm is a skeletal muscle whose thickness and strength fluctuate with age (43), and the children in Gazi et al. are much older than the children in the other included studies. In addition, infant diaphragms are not the same as adult diaphragms. The neonatal diaphragm, for example, is morphologically flat and implanted into the chest wall at a greater angle than the dome-shaped diaphragm in adults, resulting in a smaller apposition area and a limited range of displacement (44). In contrast to the adult diaphragm, which functions like a piston inside the ribs, the newborn diaphragm moves dorsally, like a bellows (45). Because the newborn diaphragm is structurally made of less anti-fatigue slow twitching (type I) fibers, reserve capacity is minimal and muscle fatigue is likely (46). When the amount of respiratory work rises, the accessory respiratory muscles (scalene and sternocleidomastoid) must be recruited to overcome tiredness (47).

Although the optimal cut-off points chosen by each study differed, our meta-analysis found that DE and DTF had good high sensitivity and moderate specificity for diagnosing infant weaning, with values of 0.85, 0.71, 0.89, and 0.79, respectively. In line with earlier studies, DTF outperformed DE in predicting MV weaning outcomes (48). Unfortunately, we were unable to perform meta-regression or subgroup analysis of DE and DTF results to determine the source of their heterogeneity due to the short number of papers available. We still consider Yao's studies as a significant source of heterogeneity because their sensitivity was 0.45 to 0.57, which was much lower than in other researches. Therefore, we feel that the capabilities of diaphragm ultrasound are underappreciated and that it could play a more prominent role in predicting infant MV weaning outcomes.

Ana M. Llamas-Llvarez et al. (15) found in a systematic analysis of adults that DTF performed better than DE in predicting patient weaning outcomes, and our investigation supported this finding for infants. Even if their findings also explain the power of LUS, they feel that this information should be viewed with caution because they were unable to undertake a thorough investigation of the capabilities of LUS due to limitations in the included literature. We can categorically state that lung ultrasonography is effective at predicting weaning outcomes because we included enough material to enable us to carry out in-depth subgroup analysis. Overall, our study found that both diaphragmatic and lung ultrasound had good accuracy and can be used to forecast the results of newborn weaning.

The lungs and diaphragm are coordinated during breathing motions because the diaphragm acts as a brake at the end of expiration to maintain end-expiratory lung volume (EELV) and prevent lung collapse (49). Patients who receive positive end-expiratory pressure ventilation have their diaphragmatic myofibers remodeled over time with the help of MV, whereas when weaned, sudden changes in EELV when the diaphragm does not have time to respond reduce inspiratory muscle strength and lead to weaning failure (50). Prolonged MV reduces diaphragm thickness, and a study in adults found that a thinner diaphragm is associated with weaning failure (51). A recent agreement on diaphragm ultrasonography in critically ill patients recommends a thickness drop of more than 10% as the cut-off for atrophy (52). Furthermore, excessive diaphragmatic effort can still result in self-inflicted lung injury since spontaneous breathing efforts can cause the lungs to stretch abruptly during inflation, even if tidal volume is not increased (53). Therefore, fresh ideas for lung and diaphragm protective breathing have been put forth (54). Although preliminary clinical trials are still being conducted, we may anticipate further study on lung and diaphragm interactions to increase the likelihood that weaning will be successful in the future.

Our meta-analysis did have certain limitations. First, the infant ages in the research we included varied greatly, thus there could be a consistent effect on the accuracy of the results. Second, differences in diaphragmatic ultrasonography procedures, which might be difficult to identify and harmonize, may have an adverse effect on the outcomes. Moreover, we only made a preliminary conclusion that pulmonary and diaphragm ultrasound are useful instruments for measuring newborn weaning with MV, but there is no consensus on acceptable cut-off values. Finally, because of the small number of studies included, we were unable to investigate the sources of heterogeneity in diaphragm ultrasound measurements.

The findings of this meta-analysis demonstrate that high LUS, low DE, and low DTF all indicate EF, implying that lung and diaphragm ultrasound, particularly LUS performance, is an excellent technique for predicting weaning outcomes in infants with MV. Subgroup analysis of diaphragm ultrasonography is also required to establish its predictive accuracy under different definitions and measurement settings.