Predictive Validity of a Neonatal Extubation Readiness Estimator in Preterm Neonates: A Retrospective, Pilot Analysis in an Inner-city Level-3 Neonatal Intensive Care Unit

HIGHLIGHTS

• Very-low birth weight (VLBW; birth weight ≤1500 gm) infants contribute disproportionately to the burden of neonatal morbidity, mortality, and the cost of intensive care.

• The timing of neonatal extubation is widely based on clinical judgment. There is a need for optimized weaning protocols, methods to determine the best timing of extubation, and clear definition/criteria of extubation failure.

• In this study, we have reviewed a predictive model based on an extubation readiness estimator (ERE) scale to determine extubation readiness in extremely preterm infants.

• A total of 105 VLBW infants born at a gestational age of below <30 weeks, who were intubated and received assisted ventilation were studied. In this cohort, the ERE score using a cut-off of 80% did not accurately predict extubation success.

• We are aware that the study is not adequately powered and so it should be viewed only as a pilot report. However, the comparisons did not harm and have justified a larger, hypothesis-based study.

INTRODUCTION

Very-low birth weight (VLBW, birth weight ≤1500 gm) infants constitute a small proportion of births globally but contribute disproportionately to the burden of neonatal morbidity, mortality, and the cost of intensive care.^1–3 These infants routinely require prolonged, invasive mechanical ventilation, which is associated with multiple comorbidities ranging from bronchopulmonary dysplasia (BPD), retinopathy of prematurity (ROP), hearing disorders, and require a prolonged need for parenteral nutrition, and hospital stay.^4,5 Hence, successful early extubation is a crucial determinant in decreasing overall morbidity and mortality in VLBW neonates.

The timing of neonatal extubation is widely based on clinical judgment, which may involve consideration of various clinical parameters such as ventilator settings, blood gas analysis, physical examination, weight at birth and at the time of evaluation, and the overall clinical stability of the patient.⁶ There is a wide variation in protocols between various neonatal intensive care units (NICUs) for neonatal extubation throughout the world; some use unit-based guidelines, whereas others rely on spontaneous breathing test trials.^7,8 As our ability to salvage ever-smaller infants improves, the need for optimized weaning protocols, methods to determine the best timing of extubation, and clear definition/criteria of extubation failure remains. In this context, we reviewed the predictive model developed by Gupta et al.⁹ to determine extubation readiness in extremely preterm infants. They have developed an ERE scale for easy guidance.

In this pilot study, we tested this aforementioned ERE scale for safety and reliability to assess extubation readiness of VLBW infants who were being treated with assisted ventilation. Our thinking was that it showed promise, we will test it in larger studies and aim to eventually develop protocols to replace our current practice of extubation based on subjective, clinical assessment of their readiness. We also explored the significance of additional parameters associated with successful extubation, morbidities associated with failed extubation, and the correlation between the mode of respiratory support after extubation and failure or success.

METHODS

We conducted a retrospective chart review including all intubated infants gestational age <30 weeks with a birth weight ≤1500 gm, who were admitted to our level-3 inner-city NICU during the period between January 2016 and December 2020 and extubated within the first 60 days. The parameters of the original Gupta ERE were recorded.⁹ To minimize confounding factors, infants with major congenital malformations and those intubated for elective surgical procedures were excluded. The local institutional ethics board approved this retrospective analysis and exempted the need for informed consent.

We recorded demographic variables including maternal and neonatal characteristics. Maternal data included prolonged premature rupture of membranes, gestational diabetes mellitus, gestational hypertension, preeclampsia, histopathologically confirmed chorioamnionitis, and administration of antenatal steroids. Neonatal variables included birth weight, gestational age, age at extubation, ventilator settings, preextubation blood gas results, respiratory severity score (RSS) at 6 hours after birth [mean airway pressure (MAP) × fraction of inspired oxygen (FiO₂)], surfactant administration, caffeine use, and weight at extubation. Other morbidities such as spontaneous intestinal perforation (SIP), necrotizing enterocolitis (NEC), late-onset sepsis, ROP ≥stage 2, and chronic lung disease/bronchopulmonary dysplasia (CLD/BPD) [National Institute of Child Health and Human Development (NICHD) classification] were compared in two groups formed based on different ERE scores. Similar to the original ERE study, an extubation attempt was recommended for a score with ≥80% probability of success whereas extubation was not recommended for a score <80% (Table 1).⁹

Our unit currently does not have set criteria for extubation; most neonates are extubated based on subjective clinical assessment, oxygen requirements, and ventilatory settings. Assessment for extubation readiness varies among neonatology care providers. After extubation, most infants receive noninvasive respiratory support with nasal continuous positive airway pressure (nCPAP), noninvasive mechanical ventilation (NIMV), or noninvasive neural-adjusted ventilatory assist (NAVA). For study purposes, extubation failure was defined as neonates requiring reintubation within 5 days postextubation for respiratory support.

Statistical Package for the Social Sciences (SPSS), version 28.0 (IBM, USA) software was used to compare maternal and neonatal characteristics in both groups. Parameters such as caffeine use, steroid use, and RSS at 6 hours of life were compared using univariate and multivariate analyses. The association between risk factors and extubation failure was reported as odds ratio (OR) and 95% confidence intervals (CIs) using the Chi-square test. Risk factors with a p < 0.1 were included and adjusted for in the multiple logistic regression model; p ≤ 0.05 was considered statistically significant. We produced receiver-operating characteristic curves and calculated the area under the receiver-operating characteristic curve (AUROC) as a measure of predictive probability for reintubation.

RESULTS

We studied a total of 105 infants with gestational age below 30 weeks and weight ≤1500 gm, who were intubated and received assisted ventilation during the study period. Forty-one intubated infants were excluded from the study because of death prior to extubation or the nonavailability of complete data. The remaining 64 neonates were assessed for the probability of successful extubation. Fifty-three neonates were extubated successfully, while 11 required reintubation within 5 days of the first attempt. Forty infants had an ERE score below 80% and 6 of them required reintubation. Twenty-four had an ERE score ≥80% and 5 required reintubation.

In those with an ERE score of below 80%, there was no significant difference in major morbidities among infants who were reintubated (Table 1). Gestational age was significantly lower among infants with an ERE score ≥80% who failed extubation, but the birth weight did not differ significantly. The incidence of CLD and ROP ≥ stage 2 was significantly higher among those reintubated who had the ERE score ≥80% (Table 2). Lower gestational age and RSS at 6 hours after birth were significant risk factors for reintubation (Table 3). The ERE model was not accurate in our sample (AUROC = 0.49; Fig. 1) and logistic regression indicated no significant utility in predicting reintubation. In our study, sensitivity and specificity at probability scores of ≥80% were 36 and 54%, respectively.

Of 34 neonates who were extubated successfully and had an ERE score below 80%, 8 were extubated to CPAP, 23 to NIMV, and 3 to NAVA. Out of six neonates who failed extubation in this group, two received support with bubble CPAP and four were placed on NIMV. In those with ERE ≥80%, 19 were successfully extubated; 9 received nCPAP and 10 were treated with NIMV. Among the five infants who failed extubation in this group, four were placed on NIMV and 1 on NAVA (Table 4).

DISCUSSION

To our knowledge, this is the first study that evaluated the performance of a neonatal ERE in VLBW babies. This scale is based on demographic and clinical parameters are available to the authors; the analysis was adjusted following the first elective attempt, and included higher gestational age, chronologic age at extubation, higher blood gas pH before extubation, and lower preextubation FiO₂, along with lower values for highest RSS in the first 6 hours after birth (as per the timing of RSS used in the original calculator).⁹ In our study, the ERE score using a cut-off of 80% did not accurately predict extubation success. It is possible that applying the ERE in clinical practice may have resulted in prolonged intubation and invasive mechanical ventilation in our population. These data differ from those of a different, previously reported cohort.¹⁰ There is a need for further studies in larger cohorts.

In our study, we have been able to successfully extubate neonates at a lower percentage score. The sensitivity and specificity based on the ERE were lower in our population. Our data showed that sensitivity and specificity at an ERE cut-off of ≥80% were 36 and 54%, respectively. These thresholds were significantly lower (54 and 81%) than the original study that was used to formulate the estimator.^9,11 Finally, the AUROC in our population was 0.49, and logistic regression showed no significant utility of the ERE in predicting reintubation or successful extubation.

In many institutions, a trial of spontaneous breathing with endotracheal CPAP is used to evaluate the readiness for extubation. In the meta-analysis of Teixeira et al.,¹² 6 studies evaluated intubated infants with spontaneous breathing trials (SBTs) using endotracheal CPAP for 3–5 minutes. Failure was defined as significant desaturations and bradycardia. Although SBTs showed high pooled sensitivity (0.97, 95% CI) to correctly identify neonates as “ready” for successful extubation, pooled specificity was still low at 0.40 at 95% CI. Hence, although SBT in premature infants can accurately predict extubation success, there is still a lack of evidence to support its use as an independent predictor for extubation failure in preterm neonates.¹²

The RSS is computed as a product of MAP and the FiO₂. Oxygen supplementation and mechanical ventilation are essential for the survival of premature infants at the expense of long-term morbidities such as CLD and the need for frequent hospitalizations, especially during the first few years of life.¹³ RSS has been used to make a decision for extubation in preterm infants. We found lower RSS at 6 hours after birth to predict successful extubation in both univariate and multivariate analysis. Surfactant use, antenatal steroids, and caffeine use did not differ between groups, but these need further evaluation in larger cohorts.

Prolonged mechanical ventilation is associated with severe morbidities in premature infants, and hence, it is essential to limit the duration of invasive mechanical ventilation. Strategies to preserve spontaneous respiration, such as patient-triggered ventilation may reduce the incidence of these complications. The use of respiratory stimulants such as caffeine and weaning modes of respiratory support may help in early extubation. A study by Shi et al. showed that extubating preterm neonates to NIMV from higher ventilator settings may be helpful.^14,15 Although not statistically significant, our data suggested that compared to nCPAP, more babies were extubated to NIMV support of ventilation in the low-probability group and remained successfully extubated.

Antenatal steroids decrease the length of stay by enhancing lung maturity and reducing oxygen demand and ventilator dependence. The antenatal administration of steroids may be important as a variable for developing an objective probability estimator. In a systematic review conducted by McGoldrick et al.,¹⁶ which included 27 studies from 20 countries (with 11,272 randomized women and 11,925 neonates), continued use of a single course of antenatal corticosteroids reduced the risk of perinatal death, neonatal death, respiratory distress syndrome (RDS), and intraventricular hemorrhage (IVH) by enhancing the lung maturity regardless of the resource setting.¹⁶ Our sample size was small to explore this correlation or determine the effect of the timing of steroid administration on the success of extubation. Similarly, surfactant use may also be an important variable in predicting the success or failure of extubation in newborns.¹⁷ Most infants in our study had received surfactant, and it remains unclear whether this could be deemed a significant factor for successful extubation. Gestational age and weight at extubation may be two important confounding variables that might alter the effect of surfactant use.

The incidence and severity of ROP are inversely related to gestational age and birth weight, and the risk increases with hyperoxia and prolonged ventilation.^18–20 In our study, the incidence of ROP ≥stage 2 was significantly higher in the reintubation group with a probability score ≥80% category. Similarly, CLD is one of the common sequelae of preterm birth. The incidence of CLD increases with lower gestational age and birth weight. Additional important risk factors include intrauterine growth restriction, sepsis, and prolonged exposure to mechanical ventilation and supplemental oxygen.²¹ Sucasas Alonso et al.²² studied 202 newborns (mean gestational age 29.5 ± 2.1 weeks); CLD was independently associated with gestational age (p < 0.001; OR = 0.44 with 95% CI = 0.30–0.65) and the need for mechanical ventilation on the first day after birth (p = 0.001; OR = 8.13 with 95% CI = 2.41–27.42).²² Our study showed a similarly increased risk of CLD in neonates with prolonged intubation in the setting of extubation failure and being dependent on mechanical ventilation.

Prolonged mechanical ventilation may increase the length of stay and increase hospital costs associated with it. Russell et al.²³ reported that the mean hospital costs for preterm infants with common morbidities of prematurity were 4–7 times higher than their gestational age-equivalent healthy controls. With prolonged intubation, the risk of comorbidities increases exponentially which are potentially associated with increased hospital costs in managing low birth weight babies. In a review study conducted in 2007 prolonged intubation-associated comorbidities were associated with an increase in direct hospital costs by $13,500 with the presence of brain injury, $17,000 with NEC, $31,500 with CLD, and $11,000 with late-onset sepsis.^23,24

Gestational age and birth weight are the most important factors for successful extubation. The lower the gestation, the higher the risk of extubation failure. Brix et al.²⁵ showed that a 2-week lower gestational age increased the odds of failure of the INtubation-SURfactant-Extubation (INSURE) procedure by a factor of 1.8. Predictors for INSURE failure were low gestational age and hemoglobin <8.5 mmoL/L. Predictors for mechanical ventilation for >24 hours were as follows: Gestational age, Apgar at 5 minutes <7; oxygen need >50%, CO₂ pressure >7 kPa (~53 mm Hg), pH <7.3, lactate >2.5 mmoL/L, need for inotropes, and surfactant administration shortly after birth.²⁵ In our study, lower birth weight and gestational age were important variables to be considered while deciding on extubation. Finally, the AUROC in our population was 0.49, and logistic regression showed no significant utility of the ERE in predicting reintubation or successful extubation.

Available data do not provide clear-cut tools to help clinicians decide to successfully extubate a preterm infant. Limitations of our study were a retrospective design, a relatively small sample size, and being a single-center study. We are aware that the study is not adequately powered and so this should be viewed only as a pilot report. Considering that the total number of cases with successful extubation was 53 while 11 required reintubation, the AUROC of the neonatal ERE score in predicting successful extubation was 0.452. We used the R-Studio package pROC for these estimations.²⁶ The code used in these estimations was the power.roc.test (number of cases in the two groups as 53 and 11, the area under the curve = 0.452, and significance level = 0.05. Using these calculations, the power of the study was only 7.34%. The 2-sided α-error was consistent with these calculations. Having said this, the study showed that the comparisons did no harm, and justified a larger, hypothesis-testing study. All clinicians who provide care to extremely premature infants have battled with difficulties in determining the optimum timing for extubation.

CONCLUSION

We believe that our pilot study served its purpose. The primary purpose of these studies is to understand (A) the clinical relevance of fundamental questions; (B) feasibility; and (C) the need for specific modifications while designing larger, ensuing hypothesis-testing studies. If the objectives are clearly understood, this exercise can be beneficial.

In our study, extubation based on the ERE tool was safe but not predictive of successful extubation. If we had used clinical judgment for extubation as earlier, most of these infants would have received a trial of extubation in our NICU at probability scores ≤60%. The ERE in our population showed a lower predictive sensitivity and specificity than the cohort in the original study that was used to formulate the estimator. A large multicenter prospective study is needed to develop a more robust and accurate calculator to predict successful extubation in extremely preterm infants.

ACKNOWLEDGMENT

The authors sincerely thank Sergio Golombek, Professor of Pediatrics and program Director of Neonatal–Perinatal Fellowship Program at State University of New York (SUNY) Downstate Health Sciences University, Brooklyn, New York USA, who provided invaluable support and guidance for our research and writing process.

AUTHORS’ CONTRIBUTIONS

KB: Conceptualization, methodology, data collection, data analysis, wrote the manuscript.

BT: Data collection, data analysis, and writing of the manuscript.

FS: Conceptualization, data analysis, and writing of the manuscript.

RT: Data collection and writing of the manuscript.

MMG: Data Analysis, review, and revision of the manuscript.

CT: Data analysis, review, and revision of the manuscript.

SR: Conceptualization, data analysis, review, and revision of the manuscript.

REFERENCES

1. Chawanpaiboon S, Vogel JP, Moller AB, et al. Global, regional, and national estimates of levels of preterm birth in 2014: A systematic review and modelling analysis. Lancet Glob Health 2019;7(1):e37–e46. DOI: 10.1016/S2214-109X(18)30451-0.

2. Patel RM. Short- and long-term outcomes for extremely preterm infants. Am J Perinatol 2016;33(3):318–328. DOI: 10.1055/s-0035-1571202.

3. Stoll BJ, Hansen NI, Bell EF, et al. Trends in care practices, morbidity, and mortality of extremely preterm neonates, 1993–2012. JAMA 2015;314(10):1039–1051. DOI: 10.1001/jama.2015.10244.

4. Choi YB, Lee J, Park J, et al. Impact of prolonged Mechanical Ventilation in Very Low Birth Weight Infants: Results From a National Cohort Study. J Pediatr 2018;194:34.e3–39.e3. DOI: 10.1016/j.jpeds.2017.10.042.

5. Escobar V, Soares DS, Kreling J, et al. Influence of time under mechanical ventilation on bronchopulmonary dysplasia severity in extremely preterm infants: A pilot study. BMC Pediatr 2020;20(1):241. DOI: 10.1186/s12887-020-02129-2.

6. Al-Mandari H, Shalish W, Dempsey E, et al. International survey on periextubation practices in extremely preterm infants. Arch Dis Child Fetal Neonatal Ed 2015;100(5):F428–F431. DOI: 10.1136/archdischild-2015-308549.

7. Kamlin CO, Davis PG, Morley CJ. Predicting successful extubation of very low birthweight infants. Arch Dis Child Fetal Neonatal Ed 2006;91(3):F180–F183. DOI: 10.1136/adc.2005.081083.

8. Chawla S, Natarajan G, Gelmini M, et al. Role of spontaneous breathing trial in predicting successful extubation in premature infants. Pediatr Pulmonol 2013;48(5):443–448. DOI: 10.1002/ppul.22623.

9. Gupta D, Greenberg RG, Sharma A, et al. A predictive model for extubation readiness in extremely preterm infants. J Perinatol 2019;39(12):1663–1669. DOI: 10.1038/s41372-019-0475-x.

10. Dryer RA, Salem A, Saroha V, et al. Evaluation and validation of a prediction model for extubation success in very preterm infants. J Perinatol 2022;42(12):1674–1679. DOI: 10.1038/s41372-022-01517-z.

11. Shalish W, Latremouille S, Papenburg J, et al. Predictors of extubation readiness in preterm infants: A systematic review and meta-analysis. Arch Dis Child Fetal Neonatal Ed 2019;104(1):F89–F97. DOI: 10.1136/archdischild-2017-313878.

12. Teixeira RF, Carvalho ACA, de Araujo RD, et al. Spontaneous breathing trials in preterm infants: Systematic review and meta-analysis. Respir Care 2021;66(1):129–137. DOI: 10.4187/respcare.07928.

13. Northway WH Jr, Rosan RC, Porter DY. Pulmonary disease following respirator therapy of hyaline-membrane disease. Bronchopulmonary dysplasia. N Engl J Med 1967;276(7):357–368. DOI: 10.1056/NEJM196702162760701.

14. Shi Y, Muniraman H, Biniwale M, et al. A Review on non-invasive respiratory support for management of respiratory distress in extremely preterm infants. Front Pediatr 2020;8:270. DOI: 10.3389/fped.2020.00270.

15. Abdel–Hady H, Nasef N, Shabaan AE, et al. Caffeine therapy in preterm infants. World J Clin Pediatr 2015;4(4):81–93. DOI: 10.5409/wjcp.v4.i4.81.

16. McGoldrick E, Stewart F, Parker R, et al. Antenatal corticosteroids for accelerating fetal lung maturation for women at risk of preterm birth. Cochrane Database Syst Rev 2020;12(12):CD004454. DOI: 10.1002/14651858.CD004454.pub4.

17. Garib M, Salama N, Deraz S. Early versus late extubation after surfactant replacement therapy for respiratory distress syndrome. Egypt Pediatric Association Gaz 2015;63(1):1–5. DOI: 10.1016/j.epag.2015.02.003.

18. Eckert GU, Filho JBF, Maia M, et al. A predictive score for retinopathy of prematurity in very low birth weight preterm infants. Eye (Lond) 2012;26(3):400–406. DOI: 10.1038/eye.2011.334.

19. Goncalves E, Nasser LS, Martelli DR, et al. Incidence and risk factors for retinopathy of prematurity in a Brazilian reference service. Sao Paulo Med J 2014;132(2):85–91. DOI: 10.1590/1516-3180.2014.1322544.

20. Holmstrom G, Broberger U, Thomassen P. Neonatal risk factors for retinopathy of prematurity: A population-based study. Acta Ophthalmol Scand 1998;76(2):204–207. DOI: 10.1034/j.1600-0420.1998.760216.x.

21. Jensen EA, Schmidt B. Epidemiology of bronchopulmonary dysplasia. Birth Defects Res A Clin Mol Teratol 2014;100(3):145–157. DOI: 10.1002/bdra.23235.

22. Alonso AS, Diaz SP, Soto RS, et al. Epidemiology and risk factors for bronchopulmonary dysplasia in preterm infants born at or less than 32 weeks of gestation. An Pediatr (Engl Ed) 2022;96(3):242–251. DOI: 10.1016/j.anpede.2021.03.006.

23. Russell RB, Green NS, Steiner CA, et al. Cost of hospitalization for preterm and low birth weight infants in the United States. Pediatrics 2007;120(1):e1–e9. DOI: 10.1542/peds.2006-2386.

24. Johnson TJ, Patel AL, Jegier BJ, et al. Cost of morbidities in very low birth weight infants. J Pediatr 2013;162(2):243.e1–249.e1. DOI: 10.1016/j.jpeds.2012.07.013.

25. Brix N, Sellmer A, Jensen MS, et al. Predictors for an unsuccessful INtubation-SURfactant-Extubation procedure: A cohort study. BMC Pediatr 2014;14:155. DOI: 10.1186/1471-2431-14-155.

26. Robin X, Turck N, Hainard A, et al. pROC: An open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics 2011;12:77. DOI: 10.1186/1471-2105-12-77.