Tachycardia–Desaturation Episodes in Neonatal Intensive Care Unit Patients with and without Bronchopulmonary Dysplasia

Abstract

Background

Much attention has been paid to measuring physiological episodes of bradycardia–oxygen desaturation (BDs) in the neonatal intensive care unit (NICU). NICU patients also have spells of tachycardia–desaturation (TDs), but these have not been well-characterized. We hypothesized that TDs would be more common among infants with bronchopulmonary dysplasia (BPD).

Objectives

We aimed to quantify daily TDs compared to BDs in NICU patients across a range of gestational and postmenstrual ages (GA and PMA) and determine whether TDs are associated with BPD.

Study Design

We analyzed every 2-second heart rate (HR) and peripheral saturation of oxygen (SpO₂) throughout the NICU stay of all infants with 24 to 39 weeks GA admitted to a single, level IV NICU from 2012 to 2015. BDs were defined in our prior work (HR <100 bpm for 2:4 seconds with concurrent SpO₂ <80% for 2:10 seconds) and TDs as a 20% increase in HR from the previous 2-hour mean baseline and concurrent SpO₂ <80% for 2:10 seconds. We calculated the median daily BDs and TDs across a range of GAs and PMAs. For infants < or = 32 weeks GA, we compared TDs for those with and without BPD at 36 weeks PMA and discharge on supplemental oxygen.

Results

We analyzed 782,424 hours of HR and SpO₂ data from 1,718 infants, with a median of 271 hours analyzed per infant. TDs frequency increased with increasing PMA across all GAs. BDs occurred most frequently in infants <29 weeks GA and decreased as infants approached term equivalent age. For infants with < or = 32 weeks GA, one or more TD per day from 33 to 35 weeks PMA was associated with BPD and home oxygen.

Conclusion

Episodes of TD at the thresholds defined in this analysis occurred more frequently at later PMA and were more common in infants with BPD and those requiring home oxygen.

Cardiorespiratory instability is common among infants in the neonatal intensive care unit (NICU).¹ Understanding clinical associations with acute changes in heart rate (HR) and oxygenation could lead to strategies to minimize these episodes. Intermittent hypoxemia (IH) events occur extremely frequently in premature infants with respiratory distress syndrome, apnea of prematurity, and bronchopulmonary dysplasia (BPD).^2–4 High IH burden increases mortality risk and worsens neurodevelopmental outcomes and therefore has been the subject of much research.^5,6 IH commonly occurs together with bradycardia, and our group developed automated algorithms to quantify bradycardia– oxygen desaturation events (BDs) and describe their associations with clinical variables including sex,⁷ gestational and postmenstrual age (GA and PMA),⁴ medications, mode of respiratory support, length of NICU stay, diagnosis of BPD, and need for respiratory support after NICU discharge.⁸

Clinicians caring for NICU patients recognize that spells of tachycardia with desaturation (TDs) also occur, but these events and their clinical associations have not previously been quantified or characterized. As a first step toward understanding the clinical significance of TDs, we sought to describe the frequency of these spells across a range of GA and PMA. We hypothesized that TD frequency would increase with increasing PMA for preterm infants as control of breathing matures and apnea-associated BDs become less frequent. We also hypothesized that TDs would be more common in preterm infants with BPD.

Materials and Methods

Study Design and Cohort

We conducted a retrospective cohort study of infants admitted to the University of Virginia NICU from 2012 to 2015. The unit is a level IV regional referral center (Vermont Oxford Network Type C). All NICU patients have continuous bedside vital sign monitoring for HR from electrocardiogram and peripheral saturation of oxygen (SpO₂) from pulse oximetry. We included infants with GA 2:24 weeks who had at least 24 hours of archived, concurrent HR and SpO₂ data available between PMAs of 24 and 39 weeks. This study was approved by the University of Virginia Institutional Review Board with a waiver of consent due to the noninterventional nature of the work.

Data Collection, Clinical Practices, and Bronchopulmonary Dysplasia Definitions

We collected demographic and clinical data routinely extracted from the electronic medical record into a relational clinical database (NeoData, Isoprime, Chicago, IL). The target SpO₂ range for infants on supplemental oxygen was 88 to 95% until 36 weeks PMA and 92 to 98% beyond 36 weeks PMA. During the study period, caffeine was routinely given to infants born at less than 32 weeks GA and discontinued beyond 32 weeks PMA once infants were off continuous positive airway pressure and did not have significant central apnea. BPD was defined as a requirement for supplemental oxygen or positive pressure support at 36 weeks PMA.^9,10 Infants who could not be assessed for BPD due to death or transfer before 36 weeks PMA were excluded from the BPD analysis.

Heart Rate and Oxygen Saturation Data Collection

HR from electrocardiogram and SpO₂ from pulse oximetry were collected from all NICU bedside monitors using BedMaster (Hillrom, Chicago, IL). SpO₂ was monitored with Masimo technology using an 8-second averaging time. HR and SpO₂ readings were sampled every 2 seconds (0.5 Hz). We used a previously published automated algorithm to identify BD events.⁴ Brady-cardia was defined as HR <100 for 2:4 seconds and desaturation as SpO₂ <80% for 2:10 seconds. For the TD analyses, the same definition of desaturation was used for BDs, and tachycardia was defined as an increase in HR of 20% above the previous 2-hour mean HR. Algorithm-detected desaturation tags were labeled as BD or TD events if the desaturation occurred within 30 seconds of the bradycardia or tachycardia tag, irrespective of which came first. BD or TD events were joined together if any component (B, D, or T) occurred within 30 seconds of a nearby event. Figure 1 shows a representative example of a BD and a TD. We quantified the number of BD and TD events and normalized the counts to the amount of available data to create a rate of the events per day of data.

Fig. 1.

Representative examples of bradycardia–oxygen desaturation (BD) and tachycardia–desaturation (TD) events. Panels (A, B) show a BD and a TD event, respectively. The event components (bradycardia, tachycardia, and desaturation) are shown as solid lines above the plotted heart rate (HR) and peripheral saturation of oxygen (SpO₂) time series. The threshold for desaturation is shown as a dotted line. Panel (A) shows the bradycardia threshold as a dashed line and panel (B) shows the tachycardia threshold as a solid line.

Statistical Analyses

We summarized continuous variables as median and inter-quartile range^Q7 (IQR) and categorical variables as number and percent unless otherwise noted. BD and TD event rates were examined using heat maps to display rates in discrete GA and PMA combinations. We aggregated data by GA categories (<28, 28–32, >32 weeks) and analyzed BD and TD rates by PMA across the NICU stay.

For the subgroup of preterm infants with 24 to 32 weeks GA, we used logistic regression to assess the associations between TD events as infants approached term-corrected age and respiratory outcomes. Since some infants were discharged before 36 weeks PMA, we performed this analysis from 33 to 35 weeks PMA. We compared the rate of TD events per day of data in this time frame for infants with or without a diagnosis of BPD and with or without the requirement for oxygen after discharge home from the NICU. For the logistic regression models, TD event rate was used as a binary predictor, stratified by <1 or 2:1 TD event per day of data, in two distinct models with BPD and discharge with home oxygen as binary outcomes. We evaluated outputs from univariate logistic regression models and multivariable models to account for sex, GA, and birth weight. Analyses were performed in R version 4.2.2 with statistical significance considered as p < 0.05.

Results

Cohort Characteristics

During the study period, 2,294 infants with 24 to 39 weeks GA were admitted to the NICU. Of these, 576 (25%) had <24 hours of HR or SpO₂ data available between PMAs of 24 and 39 weeks and were therefore excluded, leaving 1,718 infants with data available for analysis. The final cohort had a median GA of 34 (IQR: 31, 37) weeks and a median birth weight of 2,231 g (IQR: 1,600, 2,970 g). Table 1 summarizes the cohort characteristics for infants overall and within each GA category used for TD analysis.

Table 1.

Cohort characteristics

	Overall (n ¼ 1,718)	GA <28 weeks (n ¼ 154)	GA 28–32 weeks (n ¼ 390)	GA >32 weeks (n ¼ 1,174)
GA (wk)	34 (31, 37)	25 (25, 27)	31 (29, 32)	36 (34, 38)
Birth weight (g)	2,231 (1,600, 2,970)	815 (700, 970)	1540 (1,260, 1,790)	2,683 (2,180, 3,240)
Male	991 (58%)	85 (55%)	212 (54%)	694 (59%)
Race
White	1,220 (71%)	106 (69%)	281 (72%)	833 (71%)
Black	322 (19%)	40 (26%)	77 (20%)	205 (17%)
Other	87 (5%)	3 (2%)	14 (4%)	70 (6%)
Hispanic ethnicity	80 (5%)	4 (3%)	16 (4%)	60 (5%)
Length of stay (days)	19 (9, 37)	93 (74, 116)	35 (25, 49)	12 (7,21)
BPD	116/501 (23%)	88/128 (69%)	28/373 (8%)	NA
Home on supplemental oxygen	86/501 (17%)	64/128 (50%)	22/373 (6%)	NA
Died before NICU discharge	56 (3%)	11 (7%)	5 (1%)	40 (3%)

Abbreviations: BPD, bronchopulmonary dysplasia; GA, gestational age; NA, not applicable; NICU, neonatal intensive care unit.

Variables are summarized as median (first and third quartiles) or number (%) for infants overall and grouped by GA category. The denominator is shown for rows summarizing the proportion with BPD and discharged home on supplemental oxygen to show the total number of infants < or = 32 weeks GA assessed for these outcomes at 36 weeks postmenstrual age. These variables are reported as not applicable for infants 32 weeks GA and above.

Frequency of Bradycardia–Oxygen Desaturation and Tachycardia–Desaturation Events Based on Gestational and Postmenstrual Ages

We analyzed >89 infant years of data and detected 94,494 BDs and 43,406 TDs between PMAs 24 and 39 weeks. Representative examples of a BD and a TD event are shown in Fig. 1. The rate of BD and TD events per day of data varied by GA and PMA (Fig. 2). BDs were more frequently detected overall, particularly at lower GA and PMA. The TD event rate at each week of GA increased as PMA increased. Figure 3 further stratifies the BD and TD event rate analysis in three GA categories (<28, 28–32, >32 weeks). Most premature infants (GA <28 weeks) had a low median rate of TD events (<1 TD per day of data) until PMA 36 weeks, and the median rate then rose with each subsequent week of age. Infants born at GA >32 weeks had a higher median rate of TD than BD events per day of data at PMA 36 weeks (0.73 vs. 0, p < 0.05).

Fig. 2.

Daily bradycardia–oxygen desaturation (BD) and tachycardia–desaturation (TD) events by gestational age (GA) and postmenstrual age (PMA). The median number of daily BDs (top) and TDs (bottom) by GA and PMA. The heat map color scale goes from lowest in blue (0 events/day) to highest in red (6 events/day). The number of infants with data analyzed at each GA and PMA is indicated by text in the boxes.

Fig. 3.

Median bradycardia–oxygen desaturation (BD) and tachycardia–desaturation (TD) events per day by postmenstrual age (PMA) among infants grouped by gestational age (GA) <28, 28–32, and >32 weeks. TD events occurred more often at higher PMA, especially among infants born at GA >32 weeks. For infants with GA <28 weeks, the BD event rate peaked at 4 weeks of age, whereas the TD event rate peaked at 12 weeks.

Association of Tachycardia–Desaturation Episodes with Respiratory Outcomes

Of 544 infants born at GA ::: 32 weeks, 501 were assessed for BPD at 36 weeks and 481 had >24 hours of HR and SpO₂ data available from PMA 33 to 35 weeks. Of these infants with data available, 105/481 (20%) had BPD. The median (IQR) of TDs per day of data during this time frame was 0.86 (0.23, 2.87) for infants with BPD and 0.25 (0.05, 0.68) for those without BPD (p < 0.05). As shown in Fig. 4, the median TD event rate during each PMA week was higher for infants with BPD than for those without. After dichotomizing the event rate, we found that 377 (78%) had <1 TD and 104 (22%) had 2:1 TD per day of data between 33 and 35 weeks PMA. Table 2 sum-marizes clinical variables stratified by TD event rate for these infants. TD events in this PMA range were associated with BPD, both in univariate analysis (OR ¼ 3.55, 95% CI: 2.22–5.66) and adjusted for GA, birth weight, and sex (OR ¼ 2.36, 95% CI: 1.25–4.44). In addition to the association with diagnosis of BPD, the occurrence of TD events in this PMA range was even more strongly associated with the requirement for supplemental oxygen after discharge home from the NICU (unadjusted OR ¼ 4.28, 95% CI: 2.64–6.97; adjusted OR ¼ 3.05, 95% CI: 1.65–5.68).

Fig. 4.

Box-and-whisker plots of the tachycardia–desaturation (TD) rate per day of data for infants with and without bronchopulmonary dysplasia (BPD), shown for postmenstrual age (PMA) 33 to 35 weeks. These weeks show differences in TD rate during the weeks preceding BPD diagnosis at 36-week PMA, defined as supplemental oxygen or respiratory support with positive pressure. The upper and lower box margins represent the third and first quartile, the horizontal line represents the median, the whisker top indicates the 95th percentile and points above this are outliers.

Table 2.

Select demographics and outcomes for infants < or = 32 weeks gestational age, stratified by tachycardia–desaturation event rate

	<1 TD per day of data (n ¼ 377)	2:1 TD per day of data (n ¼ 104)
Median TD ratea (events per day of data)	0.19 (0.00, 0.38)	2.50 (1.44, 3.80)
BPDa	60 (16%)	45 (43%)
Discharged with home oxygena	56 (15%)	45 (43%)
PMA at discharge (weeks)	36 (35, 37)	37 (36, 40)
GA (weeks)a	30 (28, 31)	28 (25, 31)
Birth weight (g)a	1,400 (1,070, 1,721)	1,180 (785, 1,570)
Male	203 (54%)	60 (58%)
NICU mortality	5 (1%)	2 (2%)

Abbreviations: BPD, bronchopulmonary dysplasia; GA, gestational age; NICU, neonatal intensive care unit; PMA, postmenstrual age; TD, tachycardia–desaturation.

Results are shown for infants with at least 24 hours of concurrent heart rate and peripheral saturation of oxygen (SpO₂) data for analysis between 33 and 35 weeks PMA.

^aIndicates p < 0.05.

Discussion

In a large, observational, cohort study of NICU patients with continuous cardiorespiratory monitoring data available, we quantified events of severe oxygen desaturation (SpO₂ <80% for at least 10 seconds) accompanied by either decline or rise in HR (BD or TD, respectively). We found that TD events were less common than BD events, occur at later PMA, and are associated with BPD and the requirement for home oxygen.

In prior work, we characterized BDs in preterm infants born less than 35 weeks GA and found that infants with more of these events throughout the NICU stay have a higher risk of BPD. In the current work, we turned our attention to spells of tachycardia coupled with severe oxygen desaturation. We undertook this analysis because of the clinical observation that infants with chronic lung disease seemed to have spells of severe oxygen desaturation with a rise in HR that had not previously been characterized. As expected, we found that BD events are more common at earlier PMA, when infants have more apnea. Apnea of prematurity, whether central, obstructive, or mixed, is typically associated with a decline in HR but may have mixed HR responses, including accelerations.¹¹ Individuals born preterm with sleep-disordered breathing in infancy and childhood have been reported to have a rise in HR during the arousal phase following airway obstruction, likely due to sympathetic nervous system activation in response to hypoxia.^12,13 We were unable to determine whether some of the TDs we found in older preterm infants may be caused by similar pathophysiology, as this would require a more complex analysis to evaluate for airway obstruction.

Consistent with our hypothesis, we found that TDs increased with rising PMA, as BDs decreased with the maturation of control of breathing. We also found that TDs were more common from 33 to 35 weeks PMA among infants with BPD and with the requirement for home oxygen. Due to the retrospective study design, we could not determine the pathophysiology of TD events. However, we speculate that some may be due to agitation, such as those associated with NICU care interventions,¹⁴ which may lead to poor air exchange. Investigators recently demonstrated that, for extremely preterm infants after 32 weeks PMA, IH events were less likely to be associated with central apnea and more likely due to abdominal muscle contraction, indicating active exhalation or breath holding.¹⁵

This study is the first report quantifying episodes of TD in NICU patients. A strength of this work is the large number of infants with frequently sampled HR and SpO₂ data stored over the entire NICU stay. We also note several limitations. TDs were defined using thresholds based on clinical judgment and were not empirically derived. Quantifying events with less severe hypoxemia or a larger or smaller HR change may produce different results. Also, SpO₂ was measured with 8-second averaging, which is the default setting on the pulse oximeter. We have previously reported that 8-second averaging leads to underestimating the frequency and overesti-mating the duration of hypoxemia events.¹⁶ Additionally, we did not analyze granular data on the level of invasive or noninvasive respiratory support, which could impact TD event frequency, or data on caffeine discontinuation, which impacts BD event frequency. Finally, the retrospective nature of this analysis did not allow us to determine clinical associations at the time of TD episodes, such as feeding, crying, or airway obstruction. Further work to determine whether TDs are provoked or occur spontaneously could inform how clinicians respond to and mitigate such events. We speculate that, for some infants, TDs are an indicator of agitation. As such, they may be a useful biomarker to determine response to therapies to improve the comfort of NICU patients with BPD.

Conclusion

TDs are more common among NICU patients of greater PMA, in those with BPD, and in those requiring home oxygen. Determining precipitating factors might lead to strategies to mitigate these spells.

Key Points.

• Desaturation episodes occur often in preterm infants.

• Bradycardia or tachycardia can coincide with desaturation.

• TD occurs later and with BPD.