Predicting dexamethasone-associated extubation success in preterm infants

Abstract

Objective:

Dexamethasone improves respiratory status in some preterm infants with lung disease. Dexamethasone increases heart rate variability, which decreases the heart rate characteristics index (HRCi), a sepsis risk score that reflects inflammation. Respiratory improvement can be measured by the ability to wean support, quantified by a respiratory severity score (RSS). We hypothesized that HRCi and RSS early in dexamethasone treatment are associated with respiratory improvement marked by successful extubation.

Study Design:

We retrospectively reviewed NICU patients born at <32 weeks gestational age (GA) admitted from 2012–2022 who received >3 days of dexamethasone for lung disease while on mechanical ventilation. Daily mean FiO₂, HRCi, and RSS (Mean Airway Pressure × FiO₂) were calculated for the dexamethasone start day and two days before and after. Successful extubation was defined as occurring during the dexamethasone course without reintubation within 7 days. We compared variables between infants with and without successful extubation.

Results:

A total of 65 infants (mean GA 25±1 weeks) were included. HRCi, FiO₂, and RSS significantly decreased by Day 3 of dexamethasone. Successful extubation (n=38) was associated with higher postmenstrual age (PMA), lower FiO₂ and RSS, and being on conventional rather than high-frequency ventilation (all p<0.05). Multivariable analysis found that RSS and PMA, but not HRCi, predicted successful extubation.

Conclusion:

Dexamethasone treatment decreased the HRCi, but this was not associated with extubation success. Higher PMA and lower respiratory support were associated with successful extubation during dexamethasone treatment.

Background

Dexamethasone administration acutely improves respiratory status in many preterm infants.^1–3 Since dexamethasone is associated with neurodevelopmental impairment⁴, predicting at least short-term respiratory benefit and optimizing the timing of therapy could lead to better long-term respiratory and neurologic outcomes. Some have proposed limiting dexamethasone administration to infants beyond 1 week of age² or to those infants with >50% odds of death or moderate to severe bronchopulmonary dysplasia.⁵ Others use dexamethasone therapy for infants unable to wean off mechanical ventilation⁶. The use of standard metrics associated with respiratory morbidity, such as the respiratory severity score (RSS; FiO₂ × Mean Airway Pressure)^7,8 could help to stratify disease severity and guide the use of dexamethasone, but other biomarkers are needed to identify infants likely to benefit from dexamethasone therapy.

Atypical heart rate (HR) patterns from continuous HR monitoring in Neonatal Intensive Care Units (NICU) can inform clinicians about systemic inflammatory conditions and worsening clinical status.^9,10 HR is regulated by the autonomic nervous system, with frequent small accelerations and decelerations reflecting healthy sympathetic and parasympathetic function.¹¹ HR patterns may also be impacted by hypoxemia¹² or respiratory acidosis and can, therefore, be used as a biomarker to track the severity of lung disease and response to therapies. The heart rate characteristics index (HRC index as displayed by the HeRO monitor) combines measures of HR variability and HR decelerations in a predictive model designed for early warning of imminent sepsis.^9,13 The HRC index can also be elevated in other inflammatory conditions without sepsis, including significant acute and chronic lung disease of prematurity. In both an animal model and in preterm infants with chronic lung disease, dexamethasone therapy leads to improved HR variability.^14,15 This led us to hypothesize that a fall in the HRC index in response to dexamethasone could be used as an adjunct to the RSS to predict whether dexamethasone treatment, once started, will lead to successful extubation for preterm infants.

Methods

Study Design, Patient Population, Dexamethasone Dosing

We conducted a retrospective chart review of preterm infants less than 32 weeks gestational age (GA) admitted to the UVA NICU between 2012 and 2022 who were on mechanical ventilation and received at least three consecutive calendar days of dexamethasone for lung disease. Only the first dexamethasone course was analyzed. A clinical database (NeoData, Isoprime; Chicago, IL) and the electronic medical record were used to identify infants who met study criteria and to collect demographic and clinical data. The UVA Institutional Review Board approved this study under a waiver of consent.

Infants were excluded if they had significant congenital or chromosomal anomalies. Infants were also excluded if they were transferred to UVA after receiving a course of dexamethasone at an outside hospital, or if they were >60 days old or died during the first course of dexamethasone.

During the study period, dexamethasone was prescribed using the DART (“Dexamethasone A Randomized Trial”) dosing protocol, with three days of 0.025 mg/kg divided every 12 hours followed by a 7-day taper.¹⁶ We defined the calendar day that the first dose of dexamethasone was given as “Day 1.” The clinical team determined extubation readiness and timing according to guidelines that suggest weaning to a mean airway pressure of <10 cm H2O and FiO₂ <50% before extubation, with clinical judgement determining when to deviate from these guidelines. Support following extubation was either continuous positive airway pressure or nasal intermittent positive pressure ventilation. The decision to reintubate was made by the clinical team based on significant worsening in respiratory status, including a major increase in FiO₂, blood gas CO2, work of breathing, or severe apnea.

The primary outcome under investigation was successful extubation, defined as extubation at any time during the dexamethasone course without reintubation within 7 days. As a secondary outcome, we assessed whether infants had a positive response, which we defined as at least a 5% relative decrease in FiO₂ or RSS by day 3 of dexamethasone therapy.

HRC Index Score, FiO₂, and Respiratory Severity Score

Heart rate characteristics monitoring was routine for all infants using the HeRO System (Medical Prediction Sciences Corporation, Charlottesville, VA), with hourly HRC index recorded in the electronic medical record. We recorded daily mean HRC index, FiO₂, and RSS over a 5-day period relative to the first course of dexamethasone: the first day of treatment and two days before and after. RSS was calculated as hourly FiO₂ × mean airway pressure, averaged over 24 hours. The set positive end expiratory pressure was used to approximate mean airway pressure for RSS calculation for infants on non-invasive respiratory support after extubation.

Statistical Analysis

Categorical data were summarized by frequency and compared using Chi-squared tests; continuous variables were summarized as median [1st, 3rd quartile] and compared between infants with and without successful extubation during dexamethasone using t-tests. We used Wilcoxon paired t-tests to compare HRC index, FiO₂, and RSS between Day 1 and Day 3 for both groups. Statistics were performed using GraphPad Prism 9.

We used logistic regression to evaluate multivariable associations with extubation success at the start of dexamethasone. For this multivariable analysis, we included the HRC index to test our primary hypothesis and variables significantly associated with extubation success in univariate analysis to model the probability of successful extubation during the dexamethasone course. Variables were considered significant if they had a p-value of < 0.05. These analyses were performed using R version 4.2.2.

Results

Clinical Characteristics

Sixty-five infants admitted to the University of Virginia NICU between 2012 and 2022 met inclusion criteria. Demographic characteristics are summarized in Table 1. The median length of the dexamethasone treatment course was 10 days, and 45 infants (69%) were extubated during the dexamethasone course. The median interval from dexamethasone start to extubation was 4.5 days. Seven infants (16%) met our criteria for extubation failure, with 5 infants reintubated within 72 hours, and one each on days 6 and 7 after extubation. Thus, 38/65 (58%) infants were successfully extubated during the dexamethasone course. There was no significant difference in gestational age, birth weight, sex, race, chronological age or weight at dexamethasone start, or duration of dexamethasone treatment between infants successfully extubated or not. Median postmenstrual age (PMA) at the start of dexamethasone was significantly higher in successfully extubated infants (30.4 vs 28.7 weeks, p=<0.05). The proportion of infants with successful extubation was higher for those on conventional ventilation than those on high-frequency ventilation at the start of dexamethasone (32% vs 4%, p=<0.05).

Table 1.

Cohort characteristics overall and grouped by successful extubation status.

	Overall (n=65)	Successful Extubation (n=38)	No Successful Extubation (n=27)
Birth weight (grams)	649 [577, 785]	712 [597, 804]	621 [540, 710]
Gestational age (weeks)	24.6 [24.0, 25.9]	24.9 [24.3, 26.0]	24.4 [23.5, 25.1]
Male Sex	37 (57%)	20 (52%)	17 (63%)
Race
White	46 (71%)	26 (68%)	20 (74%)
Black	17 (26%)	11 (29%)	6 (22%)
Other	2 (3%)	1 (3%)	1 (4%)
Mortality*	6 (9%)	0 (0%)	6 (22%)
Respiratory Support Type at Dexamethasone Start
High-frequency ventilator*	52 (80%)	26 (68%)	26 (96%)
Conventional ventilator*	13 (20%)	12 (32%)	1 (4%)
Chronological age at dex start (days)	32 [29, 46]	34.5 [30, 46]	29 [21, 43]
Post-menstrual age at dex start (weeks)*	30.1 [28.7, 31.1]	30.4 [29.2, 31.3]	28.7 [27.5, 30.6]
Weight at dex start (grams)	1080 [935, 1320]	1120 [985, 1445]	1060 [920, 1280]
Duration of dex treatment (days)	10 [9, 10]	10 [10, 10]	9 [5, 10]
Extubated during dex course*	45 (69%)	38 (100%)	8 (29%)
>5% decrease in mean FiO2 and RSS#	56 (86%)	34 (89%)	22 (81%)

Summarized as median [1st, 3rd quartile] or N (%).

^#Infants with >5% decrease in mean FiO₂ and RSS from Day 1 to Day 3;

^*p<0.05 comparing successful vs no successful extubation.

HRC Index, FiO₂, and RSS Before and After Dexamethasone

HRC index, FiO₂, and RSS significantly decreased by Day 3 of dexamethasone treatment overall and in the subgroups of infants successfully extubated or not (Figure 1). Day 1 and Day 3 FiO₂ and RSS, but not HRC index, were significantly lower in the successfully extubated infants (Table 2, p=<0.01). The change in HRC index, FiO₂, and RSS from Day 1 to Day 3 was not significantly different between the two outcome groups (Table 2).

Figure 1: Change in respiratory and oxygen parameters with dexamethasone treatment.

Comparison between infants who were successfully extubated versus those who were not extubated during the dexamethasone course or who were extubated during the course and then reintubated within seven days (38 successfully extubated, 27 not successfully extubated). Each infant’s average hourly HRC index, FiO₂, and RSS were calculated on the day dexamethasone was started (Day 1) and two days later (Day 3). A line of identity representing 1:1 correlation is drawn on each graph. Points further below this line had a larger decrease in the plotted metric over the first 2 days of dexamethasone treatment for chronic lung disease of prematurity.

Table 2.

HRC Index, FiO₂, and RSS Day 1 and Day 3 overall and grouped by successful extubation status.

Variable	Overall (n=65)	Successful Extubation (n=38)	No Successful Extubation (n=27)
HRC index Day 1	1.74 [1.15, 2.53]	1.55 [1.10, 2.51]	1.84 [1.65, 2.61]
HRC index Day 3	0.86 [0.58, 1.37]#	0.81 [0.51, 1.34]#	1.02 [0.68, 1.46]#
Change in HRC index	−0.59 [−1.37, −0.18]	−0.62 [−1.36, −0.19]	−0.59 [−1.54, −0.07]
FiO2 Day 1*	60 [48, 81]	51 [44, 65]	73 [60, 88]
FiO2 Day 3*	43 [37, 52]#	40 [31, 45]#	51 [42, 63]#
Change in FiO2	−14 [−23, −6]	−12 [−20.25, −5.75]	−18 [−26, −10]
RSS Day 1*	944 [573, 1377]	698 [506, 974]	1257 [944, 1677]
RSS Day 3*	554 [373, 808]#	432 [328, 590]#	767 [612, 987]#
Change in RSS	−303 [−512, −121]	−206 [−390, −112]	−395 [−616, −219]

Summarized as median [1st, 3rd quartile].

^*p<0.05 comparing successful vs non-successful extubation;

^#p<0.05 comparing Day 1 with Day 3.

Using our criteria for a positive dexamethasone response being 5% improvement in FiO2 or RSS, there were only 3/65 (5%) non-responders, none of whom were successfully extubated during the dexamethasone course.

In univariate analysis, Day 1 FiO₂, RSS, and PMA were associated with the extubation outcome. We performed multivariable analysis using the following predictors: Day 1 (first day of dexamethasone) RSS, HRC index, and PMA. We excluded FiO₂ from this analysis because of its collinearity with RSS (RSS = MAP × FiO₂). Results showed that Day 1 RSS and PMA remained statistically significant in their association with extubation during dexamethasone without reintubation within 7 days. HRC index was not an independent predictor of extubation success. Increased RSS by one standard deviation had an odds ratio of 0.15 (95% CI 0.05–0.34) for extubation success, and increased PMA by one standard deviation had an odds ratio of 2.99 (95% CI 1.31 – 8.45) for extubation success.

Discussion

In this single-center retrospective study, we found that most preterm infants treated with dexamethasone had an acute decrease in their heart rate characteristics index (HeRO score), reflecting improved heart rate variability. Most infants also had a decrease in FiO₂ requirement and RSS, and more than half were successfully extubated during dexamethasone treatment. Baseline predictors of successful extubation were older postmenstrual age, lower FiO₂ and RSS, and being on a conventional rather than a high-frequency ventilator. In contrast to our initial hypothesis, neither the HRC index on the day dexamethasone was started nor the decline in the score by day 3 of treatment predicted extubation success.

Many studies have reported predictors of extubation success or failure for extremely preterm infants, but few have studied predictors of what some call “dextubation” success, meaning extubation during dexamethasone treatment.¹⁷ A similar study by O’Connor et al. reported that higher FiO₂ and mean airway pressure (the two components of RSS) predicted the inability to successfully extubate within 2 weeks after the first dose of dexamethasone.¹⁸ They found higher gestational age to be associated with successful “dextubation,” whereas we found higher postmenstrual age to be predictive. Of note, the median chronological age at treatment was 19 days in the O’Connor cohort versus 29 days in ours. The finding of older PMA as predictive of extubation success with dexamethasone has also been reported by Kuschel et. al.¹⁹

We undertook this analysis with the hypothesis that the baseline HRC index or the change in this score after starting dexamethasone would predict extubation success because of the association between inflammation and low heart rate variability. Another important question is whether dexamethasone should be started in the first place, but this was not our objective. In another study, Kaczmarek et al. reported that low heart rate variability was associated with extubation failure for preterm infants.²⁰ The HRC index, developed as an early warning system for sepsis, incorporates not only a measure of low HR variability but also a measure of heart rate decelerations. In prior studies and in the current study, nearly all infants had improvement in HR variability within two days of starting dexamethasone.¹⁴ However, we did not find an association between baseline HRC index or change in HRC index and extubation success, either due to the small sample size or the fact that dexamethasone improved respiratory status by mechanisms other than its anti-inflammatory effect. Infants extubated soon after starting dexamethasone may have had increased apnea and bradycardia spells, which can increase the HRC index despite improved HR variability.

Of note, there were some infants who had no change in FiO₂ or RSS within three of starting dexamethasone, yet the drug was continued for a full 10-day course. Consideration should be given to discontinuing treatment for infants who demonstrate no measurable improvement in respiratory status. This is especially important since total dexamethasone exposure is associated with adverse neurodevelopmental outcomes.⁵ Infants who fail to wean with an early dexamethasone trial may have a more favorable response at a later age when they are on less support and more likely to be successfully extubated and when the harmful effects of dexamethasone on the developing brain may be less.

A strength of this work is the use of two available tools, the HRC index and the RSS, to assess acute physiologic improvement after starting dexamethasone. Several limitations deserve consideration. First, the HRC index was designed to assess heart rate patterns associated with systemic inflammatory conditions including sepsis²¹ and necrotizing enterocolitis²² and not to predict extubation readiness. Also, we did not control for confounding variables that may impact the HRC index, such as severe intraventricular hemorrhage,²³ nor did we assess factors that might be associated with extubation success, including hematocrit, caffeine dosing, and mode and level of post-extubation support. Another limitation is that in a retrospective study over a decade-long period, we cannot account for variability in clinical care over time, including decisions regarding the use of dexamethasone, the timing of extubation, and the need for reintubation. Finally, the cohort is small, and a larger multicenter study would be required to determine criteria predicting “dextubation” success.

The evidence on the risks and benefits of postnatal steroids for severe chronic lung disease continues to evolve, and the optimal type, timing, dose, and duration of therapy has not been established. While concern for neurodevelopmental sequelae remain, a recent meta-regression analysis demonstrated reduced risk of cerebral palsy with dexamethasone treatment when the projected risk of bronchopulmonary dysplasia is high.²⁴ Our work adds to growing literature on biomarkers for extubation readiness, with RSS and PMA predicting successful extubation during dexamethasone treatment. While we did not find that the HRC index predicts successful extubation, we did find a marked improvement in heart rate variability (reflected by a decline in the HeRO score) reflecting a potent anti-inflammatory effect of dexamethasone.

Conclusion

In this cohort of 65 preterm infants who received dexamethasone, the HRC index improved but did not predict the 58% of infants that were successfully extubated. Successful extubation during dexamethasone treatment was associated with higher postmenstrual age, lower FiO₂, and a lower respiratory severity score at the time dexamethasone was started.

KEY POINTS.

• 58% of preterm infants were successfully extubated during dexamethasone treatment.

• Higher PMA and lower Respiratory Severity Scores were associated with extubation success.

• HR characteristics improved with dexamethasone but did not predict extubation success.