Background

Respiratory distress syndrome (RDS) is a major cause of death amongst preterm neonates worldwide, and a significant contributor to complications associated with preterm birth [1, 2]. The most common supportive treatments for RDS include supplemental oxygen and continuous positive airway pressure (CPAP) [3]. While most infants can be successfully managed on CPAP, the requirement for invasive mechanical ventilation (IMV), or CPAP failure, occurs in approximately 20–40% of preterm infants [4, 5]. CPAP failure has been reported to be higher in areas with lower antenatal corticosteroid administration [6]. Early selective surfactant administration in infants at risk of CPAP failure can help prevent IMV and complications associated with its use [7,8,9]. Early identification of neonates at risk of CPAP failure is not only necessary in highly resourced settings that can provide surfactant and/or mechanical ventilation, but is also important in lower-resourced settings when transfer to a higher level of care with advanced respiratory support is feasible [1].

There are various methods used to identify infants that would most benefit from surfactant administration, including physiologic measures [Fraction of inspired oxygen (FiO2), respiratory severity score or mean airway pressure (MAP): FiO2 product], blood gas parameters (pCO2 and pH), and lung imaging (lung ultrasound, and chest radiograph) [9,10,11,12]. Although a FiO2 of 0.3 to 0.4 has been the most frequently used parameter in practice and in many studies, the optimal threshold remains unknown and likely varies based on gestational age [9, 13,14,15,16]. Clinical prediction scores have also been developed, but many incorporate gestational age or other factors that may be unknown or difficult to obtain in low-resource settings [17]. The Downes score was originally developed in 1970. Although its ability to predict the need for surfactant administration was not studied, it is being used extensively in low-resourced settings to objectively measure respiratory distress, help guide CPAP therapy, and predict CPAP failure [18,19,20,21,22,23,24,25], likely because it correlates well with physiological variables reflecting severe RDS [18]. The Downes score includes cyanosis (and whether or not it is relieved by supplemental FiO2), respiratory rate, retractions, grunting, and decreased air entry on auscultation. Each category is given 0, 1, or 2 points which is aggregated into a total score of 0 to 10. Scores of 4 and greater have been commonly used to guide respiratory support and predict CPAP failure in low-resource settings [21,22,23, 25]. The recently published RDS-NeXT Workshop consensus statements recommended the use of the modified Downes score as a clinical examination scoring tool in addition to FiO2, level of respiratory support, and blood gas parameters (pH or pCO2), if available, to help guide surfactant administration [26].

The aim of this study was to determine if a simplified Downes score obtained from nursing admission data for infants admitted on nasal CPAP would be able to identify, with high sensitivity and specificity, the infants that received surfactant therapy in a highly resourced level III NICU.

Patients and methods

This study was an analysis of existing nursing admission data for infants admitted on nasal CPAP. The primary objective was to evaluate the predictive value of a simplified Downes score in the first hour of life for the receipt of surfactant, with or without mechanical ventilation. The secondary objective was to investigate the predictive value of individual components of the simplified Downes score (FiO2 > 0.21, respiratory rate ≥60, retractions, grunting, decreased air entry on auscultation) for the receipt of surfactant, with and without mechanical ventilation. The University of Vermont Institutional Review Board determined that this study was not human subjects research.

Epic SlicerDicer was used to search the University of Vermont Medical Center (UVMMC) Level III NICU electronic medical record (EMR) for all infants admitted on CPAP between January 1, 2018 and January 31, 2023 who weighed ≤ 2000 grams and were ≥ 25 weeks’ gestation at birth. The weight cutoff was selected with the goal of identifying patients who may be considered preterm based on birthweight alone, in the absence of accurate gestational age dating, such that may occur in lower-resourced settings. The gestational age cutoff was chosen because our institution had protocolized routine delivery room intubation in infants <25 weeks’ gestation during the selected time period.

At UVMMC, nasal bubble CPAP is commonly initiated at 5–6 cm H2O. For infants receiving surfactant, the practice during the study period was to provide an initial dose of 2.5 mL/kg of CUROSURF® (poractant alfa) via the INSURE (intubate-surfactant-extubate) technique, or endotracheal tube (ETT) if intubated for continued IMV. For infants to be designated as having received INSURE in this study, they were not placed on the ventilator and did not have radiographic confirmation of ETT placement. Although a FiO2 > 0.3 is a common threshold for surfactant administration at our institution, clinician discretion allows deviation from this guideline to individualize care in the best interest of patients.

For infants identified from the EMR, chart review was performed starting with exclusionary criteria. Infants were excluded if they were outborn, intubated in the delivery room, or had severe congenital anomalies known prenatally or identified on admission. For infants fulfilling the inclusion criteria, nursing admission flowsheet data was searched for a documented admission physical examination within the first hour of life. Infants without a documented admission physical examination were excluded. In patients for whom there was a documented admission physical examination from nursing admission flowsheet data, the simplified Downes score was calculated using FiO2, respiratory rate, and the documented presence of retractions, grunting, and decreased air entry on auscultation (Fig. 1). UVMMC does not currently use the Downes score in clinical care, and nursing electronic flowsheet data does not routinely discriminate between the severity of grunting (audible by ear versus audible with auscultation), retractions (mild versus severe), or degree of air entry. For this reason, and to obtain information from nursing admission data, a simplified Downes score was designed. The Downes score was simplified by: (1) removing the distinction between mild and severe for the presence of retractions, grunting, and decreased air entry; (2) not discriminating between tachypnea ≥ 60 and > 80; and (3) using any FiO2 > 0.21 instead of the presence of cyanosis. The simplified Downes score removed the “middle” score of a 1, so infants could only obtain a score of 0 or 2 in each category (Fig. 1). Setty and colleagues simplified the Silverman-Andersen score in a similar way to help facilitate use in low-resourced settings [27].

Fig. 1: Simplification of Downes score.
figure 1

The Downes score was simplified by: (1) removing the distinction between mild and severe for the presence of retractions, grunting, and decreased air entry; (2) not discriminating between tachypnea ≥ 60 and > 80; and (3) using any FiO2 > 0.21 instead of the presence of cyanosis (FiO2 : fraction of inspired oxygen).

Full size image

A documented presence of the possible predictors (grunting, retractions, and decreased air entry) was counted as a “true” positive and used to calculate the simplified Downes score. For possible predictors that were not documented in the nursing admission data flowsheet in the first hour of life (grunting, retractions, and decreased air entry), other note types (admission history and physical, nursing note, respiratory therapy note) were searched for the presence of possible predictors to determine if they were truly absent, equating to a “true” negative or simply missing/incomplete. If discordance was found (documented presence found in other note type, with no mention in the nursing admission data flowsheet for ≥2 out of 3 variables), the patient was excluded from the study due to incomplete data. If there was concordance found (no documented presence found in other note type, with no mention in the nursing admission data flowsheet for ≥2 out of 3 variables), then the possible predictors were to be considered true negatives and the patient would be included in the study. For all patients included in the study, demographic and clinical data were also recorded.

To assess the validity and feasibility of using nursing admission documentation to calculate the simplified Downes score, we performed an initial assessment of 20 patient charts using the above assessment for discordance in documentation. A preplanned study stopping threshold was set, where if 25% of charts (5 out of 20) had discordance between admission documentation assessments, the study would be terminated. Due to finding < 20% discordance among the infants in the initial chart review, the study was continued as planned. Additionally, there was no evidence that gestational age (either higher or lower) contributed to discordance.

The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated for both the total simplified Downes score and its individual components for the receipt of surfactant. We also calculated receiver operating characteristic (ROC) curves and the area under the receiver (AUC) operating characteristic curve statistic. SAS 9.4 (Cary, North Carolina, USA) was used for all analyses.

Results

During the study period, 202 infants were admitted on CPAP. Of these 202 infants, 34 infants were excluded due to discordance or incomplete documentation. A simplified Downes score was calculated from nursing admission flowsheet data for 168 infants (final cohort). From the final cohort, 53 (31.5%) infants received surfactant. Demographic and clinical data are presented in Table 1. Five out of 168 infants (2.9%) were greater than 37 weeks’ gestation.

Table 1 Characteristics of included patients.
Full size table

A simplified Downes score of ≥4 predicted the receipt of surfactant with a sensitivity of 90.6%, specificity of 52.2%, PPV of 46.6%, NPV of 92.3%, and accuracy of 64.3% (Table 2). The predictive value of the individual components of the simplified Downes score are displayed in Table 2. Diminished breath sounds by itself had good accuracy at 71.4%. FiO2 by itself showed high sensitivity (81.1%) and NPV (87%), but was not very specific (58.3%). Accuracy of FiO2 on admission was 65.5% for predicting the receipt of surfactant.

Table 2 Predictive value of the total score and individual components of the simplified Downes score for receipt of surfactant (FiO2: Fraction of inspired oxygen).
Full size table

As the simplified Downes score increased, the likelihood of receiving surfactant increased (Fig. 2). The ROC curves are shown in Fig. 3. The AUC was 0.78, but improved to 0.83 after adjusting for gestational age. Of the final cohort, 19.6% of infants admitted on CPAP received escalated respiratory support with invasive mechanical ventilation in the first 72 h of life. This included infants who were intubated for surfactant (not INSURE), intubated for respiratory failure (and did not receive surfactant), and those intubated after initially receiving INSURE.

Fig. 2: Likelihood of receipt of surfactant per simplified Downes score.
figure 2

The likelihood of receiving surfactant increased as the simplified Downes score increased.

Full size image
Fig. 3: ROC Curves for the ability of the simplified Downes score to predict the receipt of surfactant.
figure 3

The top ROC (A) is without adjustment for gestational age and the bottom (B) is with adjustment for gestational age.

Full size image

Discussion

When using the simplified Downes score as an objective screening tool on admission, a score of ≥ 4 demonstrated both high sensitivity and NPV for the receipt of surfactant. The recent RDS-NeXT Workshop consensus statements, designed to help guide surfactant therapy in highly resourced settings, included the Downes score as an objective measure of respiratory distress [26]. Although the Downes score is commonly used in low-resourced settings to help guide respiratory support, to the best of our knowledge, this is the first publication with the aim of studying the ability of the Downes score to predict the need for surfactant. The results of this study suggest utility in using a reliable tool such as the Downes score, or a simplified version as used in this study, to help predict which infants require surfactant. When applying a threshold of 4 or greater, the high NPV of the simplified Downes score could be useful to help reassure practitioners that infants with low scores are less likely to require surfactant. This finding is in line with Downes’ original description where the observation was made that “infants whose score remained less than 4 throughout the first eight hours of life were nearly always normal from a clinical standpoint by age 24 h” [18]. At the higher cutoff of 8, the specificity and PPV of the simplified Downes score were very good at predicting the receipt of surfactant. Not only do these results suggest the benefits of using a simple and reliable tool to help improve the early identification of infants who may require surfactant therapy in highly resourced settings, but there will likely be application as well in translating the results back to low-resourced settings where the score is already being used.

The ability of FiO2 alone to predict CPAP failure and surfactant receipt is complicated, as blenders are not universally available in low-resourced settings and FiO2 requirement may be impacted by respiratory support level and mean airway pressure. Importantly, there is no consensus on the optimal FiO2 at which to provide surfactant when available [9]. With the implementation of less invasive methods of surfactant administration and ever decreasing gestational ages, a lower FiO2 cutoff is sometimes used. This controversy is evident in the RDS-NeXT guidelines where a lower FiO2 was suggested earlier (≤ 2 h of life) and a higher cutoff later (> 2 h of life) in determining the need for surfactant [26]. Despite the unclear evidence regarding the optimal FiO2 threshold at which to provide surfactant, the ability of a FiO2 threshold of 0.29 in the second hour of life to predict CPAP failure has been reported to have an AUC of 0.7 [13, 16, 28, 29]. Two other studies found that FiO2 in the first hours of life (one study combined maximum FiO2 with chest radiograph showing severe RDS) was able to predict CPAP failure with an AUC of greater than 0.8 [10, 11]. Though not directly comparable, the simplified Downes score obtained on admission was able to predict the receipt of surfactant (after adjusting for gestational age) with an AUC of 0.83.

The MAP: FiO2 product, referred to as the respiratory severity score (RSS) has also been evaluated for its ability to predict CPAP failure and the receipt of surfactant, with RSS cutoffs of greater than 1.28 and greater than 1.5 having been reported [12, 17, 30]. However, the optimal RSS cutoff to predict the receipt of surfactant is yet to be determined. In our study, the mean RSS was 1.59 for infants that did not receive surfactant and 2.33 for infants that did receive surfactant, with both groups having mean RSS scores greater than the cutoffs previously suggested in the literature [12]. In addition to needing further validation and exploration, the broad use of this score is limited due to lack of standardization in CPAP modalities as pressure delivered varies between devices and interfaces. The de-emphasis of blood gases and chest radiographs in favor of preemptive surfactant therapy based on work of breathing and FiO2 requirement, as mentioned in the most recent European Consensus Guidelines on the management of RDS, suggest the need for a validated clinical examination tool, such as the Downes score [9].

The Silverman-Andersen score is another clinical examination scoring tool that has been shown to predict the need for additional respiratory support in highly resourced settings [31, 32]. It is also being used to help identify CPAP failure and guide surfactant administration in low-resourced settings [6]. Setty and colleagues further simplified the Silverman-Andersen score with the goal of it being easier to implement in low-resource settings [27]. However, when compared to the Silverman-Andersen score, the modified Downes score has been suggested to be more reliable and easier to obtain [33]. In addition, the modified Downes score has been shown to have a high inter-rater reliability [22].

Our study’s limitations lie in the fact that the Downes score was calculated and simplified from nursing admission data at a single center in a highly resourced setting. Further prospective studies should be performed using the Downes score, or a simplified version as used in this study, to predict CPAP failure and to help guide surfactant administration. Also, this study only investigated the simplified Downes score when applied at a single time point on admission. Although our study suggests utility in a simplified Downes score being completed on admission, due to the progressive nature of RDS, further studies should evaluate the Downes score when used serially as it may improve the prediction of infants requiring escalation of respiratory support and/or surfactant. As these infants are cared for by multi-disciplinary care teams, serial assessments could also be completed by nurses, respiratory therapists, advanced practice providers, physicians, or other team members responsible for serial clinical assessments, and then discussed within the context of family-centered and goal-directed care. As the Downes score has been previously used to assess the response to surfactant, there is likely utility in not only using the score serially before surfactant, but to use it after to assess response, need for repeat dosing, and for weaning support [34].

Another major limitation to our study was the required simplification of the Downes score as our level III NICU does not currently use the Downes score, and nursing electronic flowsheet data does not routinely discriminate between the severity of grunting, retractions, or degree of air entry. Due to the fact that the Downes score was simplified from nursing admission data, it would need to be validated to determine if it maintains the high inter-rater reliability that the Downes score has previously demonstrated. If found to demonstrate high inter-rater reliability, a simpler score may be more feasible for medical professionals with limited experience with neonatal RDS. Additionally, considering the Downes score correlated well with blood gas parameters, the simplified Downes score should be further studied to see if it retains the ability to predict blood gas abnormalities. The use of 2000 grams as a surrogate marker for prematurity proved feasible in our study, with less than 3% of infants (n = 5) in the final cohort being greater than 37 weeks. However, the AUC did improve after adjusting for gestational age.

With the anticipated improved availability and affordability of surfactant in low- and middle-income countries (LMIC), it will be beneficial to have a clinical examination scoring tool that does not rely on potentially unavailable resources, such as chest radiography and blood gases, to help guide clinical decision making regarding surfactant therapy. Early identification of infants with RDS who are at risk for CPAP failure is critical to allow for earlier transport to a center that could provide surfactant therapy and/or IMV. In low-resourced settings, where the Downes score (or Silverman-Andersen score) is already frequently employed, the optimal threshold will need to be determined. A score of ≥ 4 is commonly used to help guide respiratory support or initiation of CPAP. For infants admitted and stabilized on CPAP with a score less than 4, the high NPV suggests that these infants are unlikely to require surfactant therapy. Although many centers in LMIC currently use the Downes score, immediate generalizability for the ability of the score to predict surfactant is difficult due to the lower rates of antenatal corticosteroids in these settings compared to that in our study and other highly resourced settings.

Per the original description of the score by Dr Downes, “To be useful to the community pediatrician who lacks facilities for arterial pH, [arterial] pCO2 and [arterial] pO2 measurements, a clinical scoring system must (1) be simple and easily learned; (2) involve no biochemical determinations; (3) provide some guide to treatment as well as prognosis; and (4) be related within reasonable limits to the physiologic derangements in the infant” [18]. Even 50 years later, the Downes score and its modifications continue to be used by practitioners around the world, and now may have found new utility in guiding surfactant administration.

Conclusion

The simplified Downes score can help predict the receipt of surfactant. The results of this study suggest that the incorporation of a reliable clinical assessment tool, such as the original or simplified Downes score, into normal admission workflows of infants with RDS may help guide clinical decision making regarding surfactant administration. Further studies should be performed prospectively to see if a standardized clinical assessment improves the early identification of infants at risk of requiring invasive mechanical ventilation, and to determine who may benefit from surfactant administration.