Pulse Rate Variability as a Predictor for Length of Stay for Patients with Bronchiolitis in the Pediatric Intensive Care Unit

doi:10.21203/rs.3.rs-4505039/v1

Download PDF

Article

Pulse Rate Variability as a Predictor for Length of Stay for Patients with Bronchiolitis in the Pediatric Intensive Care Unit

https://doi.org/10.21203/rs.3.rs-4505039/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Patients admitted to pediatric Intensive Care Unit (PICU) due to bronchiolitis have unpredictable length of stay (LOS). The aim of this study is to observe the difference in the relationship between pulse rate variability (PRV) and heart rate variability (HRV) for patients with bronchiolitis admitted to the PICU and its association with LOS. The first 12 hours of physiologic data after PICU admission were used for analysis. Electrocardiogram (ECG) and photoplethysmography (PPG) were divided into non-overlapping 5-minute segments, and R-peak and PPG-peak were obtained to calculate PRV and HRV. Correlation was calculated between HRV and PRV for each PICU LOS group and was tested with Fisher Z-transformation. The student’s t-test, chi-square test was performed on other independent variables, including age, gender, prematurity and ventilation type. A total of 119 patients were included in this study. For both LOS groups, PRV and HRV parameters were significantly different. However, the correlations between PRV and HRV parameters were significantly higher in the short-stay group compared to the long-stay group. This study demonstrates that the correlation between the PRV and HRV is lower in patients with longer length of stay, suggesting this can be a potential metric for LOS in PICU.

Health sciences/Biomarkers

Health sciences/Medical research

Health sciences/Medical research/Paediatric research

Bronchiolitis is a leading cause of hospitalization in the pediatric population. 75,000 to 125,000 pediatric patients with bronchiolitis are hospitalized each year ^1,2 and around 25,000 to 42,000 are admitted to a pediatric intensive care unit (PICU), where they require close observation and often non-invasive or invasive mechanical ventilation^3–5. While morbidity and mortality are low, the burdens on patients and families of a lengthy PICU stay are considerable⁶. Predictors of lengthy PICU stays would enable appropriate healthcare resource allocation, plan for adequate patient nutrition, and prognostication to support caregiver expectations⁷. Longer days in the hospital for a patient result in high healthcare cost, loss of working days for caregivers, and psychological burden for caregivers and their children⁸, making length of stay (LOS) a metric for quality of care and operational efficiency in critical care.

Cardiorespiratory parameters, including telemetry and photoplethysmogram (PPG) (also known as pulse oximetry), are routinely captured for patients with bronchiolitis in the PICU. Heart rate variability (HRV) is exquisitely sensitive to disease and has been associated with general wellbeing⁹, stress assessment^10,11, and mortality^12,13. Time domain HRV measures of the electrocardiography (ECG) signal reflect sympathetic and parasympathetic activity¹⁴. Pulse rate variability (PRV) is analyzed from the PPG signal, the pulse wave captured in peripheral tissue, and has been proposed as a surrogate for HRV ^15–18. Interestingly, PRV is not perfectly aligned with HRV, and the difference may be dependent on underlying physiologic conditions^19–21. The QRS electrical potential seen on the ECG is the result of ventricular contraction, which ejects a particular volume of blood to propagate down the arterial tree. The pulse wave generated from the ejection of the blood from the heart is then observed via the photosensors of the pulse-oximeter, which comprises the PPG waveform²². The subtle differences between HRV and PRV could be leveraged to provide greater insight into patients’ cardiorespiratory physiology.

In this study, we hypothesized that the PRV and HRV signals would be different regardless of LOS; however, we postulated that the relationship between the two signals may be dependent on patient underlying physiology and illness, which could be reflected in PICU LOS. This retrospective observational cohort study examined the differences in PRV and HRV relationship between a short-stay group and long-stay group.

Participants

This study was approved by the Institutional Review Board of Columbia University (Protocol # AAAU5398) and conducted in accordance with the principles embodied in the Declaration of Helsinki and accordance with local statutory requirements. We performed a single-center, retrospective cohort study of patients who were admitted to the PICU at NewYork-Presbyterian Morgan Stanley Children’s Hospital at Columbia University with a clinical diagnosis of bronchiolitis between February 2020 to April 2023. We chose patients with bronchiolitis as they represent a large population of patients who are admitted to the PICU with highly variable lengths of stay. The study was approved by the Institutional Review Board. Informed consent was waived. Patients were identified based on an admitting or discharge ICD-10 diagnosis of “bronchiolitis”, and then subsequently screened by a clinical provider (BW). Patients who were older than 36 months were excluded. We included children up to 36 months of age, given the highly variable practice patterns observed in children with bronchiolitis. Additionally, patients who required intubation or who had preexisting tracheostomies were excluded, as these patients, given their degree of illness and or chronic conditions, will have increased length of stay.

Data Collection

Patient characteristics were collected from the electronic medical record. Clinical data including age, prematurity, past medical history, viral testing results, and presence and degree of respiratory support were collected. Description of respiratory support was extracted manually from the medical record. At our institution, patients who require High Flow Nasal Cannula (HFNC), Continuous Positive Airway Pressure (CPAP) greater than 6 cm H₂O and the use of Bilevel Positive Airway Pressure (BiPAP) require PICU admission.

Physiologic data were acquired from Phillips Intellivue MX700 multiparametric patient monitors (Amsterdam, Netherlands) from time of PICU admission until PICU discharge. Telemetry (or continuous ECG) and PPG waveform data were extracted for each patient at a sampling rate of 256 and 128 samples per second, respectively.

Data processing and statistical analysis

All data processing and statistical analysis were performed using MATLAB 2020b (Math Works, Massachusetts). The first 12 hours of data after PICU admission were used. Parts of waveforms with artifact were manually selected and removed. ECG and PPG were divided into non-overlapping 5-minute segments. R-peaks of ECG were determined using the Pan-Tompkins algorithm and peaks of PPG were determined using Multi-Scale Peak and Trough Detection method²³. Time-domain and non-linear measures were calculated for each 5-minute segment for both ECG and PPG: Standard deviation of the difference between successive RR intervals (SDSD), standard deviation of NN intervals (SDNN), root mean square of successive RR interval difference (RMSSD), percentage of successive RR intervals that differ by more than 50 ms (pNN50), Poincaré plot standard deviation perpendicular the line of identity (SD1), Poincaré plot standard deviation along the line of identity (SD2), and ratio of SD1-to-SD2 (SD1/SD2). All segments were averaged per patient. Two groups were defined depending on the PICU length of stay (LOS): PICU less than or equal to 2 days (short-stay group) and PICU greater than 2 days (long-stay group).

Student’s t-test was performed between HRV and PRV for each group. Pearson’s correlation was calculated between HRV and PRV for each PICU LOS group, and correlation’s coefficients was compared by Fisher Z-transformation. The Bonferroni correction was applied to adjust for multiple comparisons. Student’s t-test and chi-squared tests were performed on other independent variables between short-stay and long-stay group, including age, sex, prematurity, and ventilation type. Additionally, analysis of variance (ANOVA) was performed comparing LOS among the ventilation type.

A total of 119 patients were included in the study. There were 66 patients in the short LOS and 53 patients in the long LOS group. The mean age was 10.8 months with a standard deviation of 7.9 months. 50 (42%) participants were female. 20 (16.8%) of the patients were premature with a mean gestational age of 31.2 weeks. The highest level of respiratory support types during the PICU admission were 81 (68.1%) BiPAP, 32 (26.9%) CPAP, 2 (1.7%) HFNC, 3 (2.5%) NC. One patient in this study never required supplemental oxygen but required the PICU for frequent suctioning and monitoring. Patient demographics are shown in Table 1.

Table 1

Patient Demographics
Variable	Total Patients ( n = 119)	Long Stay Patients: < 2 days (n = 66)	Short Stay Patients: ≥2 days (n = 53 )	p-value (test done)
Age, Mean (standard deviation)	10.8 (7.9)	11.1(7.8)	10.3(8.0)	0.57 (t-test)
Female Sex	50 (42%)	29(43%)	21(40%)	0.64 (chi-square test)
Respiratory Support				0.128 (analysis of variance)
BiPAP (%)	81 (68.1)	39	42
CPAP (%)	32 (26.9)	23	9
HFNC (%)	2 (1.7)	0	2
NC (%)	3 (2.5)	3	X
RA (%)	1 (0.8)	1	x
Number of Premature Patients (%)	23 (19.3)	8(12%)	12(22%)	0.13 (chi-square test)
Gestational age of premature patients, weeks (standard deviation)	31.5 (3.4)	30.5(4.1)	31.7(3.6)	0.42 (Mann-Whitney U-test)

Mean and standard deviation of HRV and PRV for all patients and for each LOS group are shown in Table 2 and represented in boxplot in Figs. 1 and 2. The correlations between HRV and PRV for each group are shown in Table 3 and Fig. 3. The time-domain variabilities: SDSD, SDNN, RMSSD, SD1 and SD2 were significantly different between HRV and PRV in both ICU cohorts (ICU stay less than 2 days and ICU stay greater than or equal to 2 days). The correlation between PRV and HRV was greater in the short-stay group for all parameters except for SD1/SD2.

Table 2

HRV and PRV values for all, short-stay and long stay group
Feature	Mean HRV and PRV (n = 119)		PICU stay < 2 days (n = 66)			PICU stay ≥ 2 days (n = 53)
	HRV	PRV	HRV	PRV	p-value	HRV	PRV	p-value
SDSD	13.72(7.88)	5.88(3.38)	13.72(9.22)	6.22(3.97)	< 0.001	13.01(5.85)	5.46(2.41)	< 0.001
SDNN	10.24(5.41)	4.83(2.6)	10.24(6.19)	4.97(2.94)	< 0.001	9.94(4.3)	4.66(2.13)	< 0.001
RMSSD	13.72(7.88)	5.88(3.38)	13.72(9.22)	6.22(3.97)	< 0.001	13.01(5.85)	5.46(2.41)	< 0.001
pNN50	0.77(0.11)	0.75(0.1)	0.77(0.11)	0.75(0.1)	0.386053	0.76(0.11)	0.74(0.09)	0.195054
SD1	9.7(5.57)	4.16(2.39)	9.7(6.52)	4.4(2.81)	< 0.001	9.2(4.13)	3.86(1.71)	< 0.001
SD2	10.6(5.36)	5.28(2.87)	10.6(5.94)	5.36(3.1)	< 0.001	10.5(4.6)	5.19(2.59)	< 0.001
SD1/SD2ratio	0.85(0.15)	0.85(0.21)	0.85(0.17)	0.87(0.19)	0.662175	0.86(0.13)	0.83(0.22)	0.37269

Table 3

Correlation between HRV and PRV
	ICU stay 0–2	ICU stay 3–5	p-value
SDSD	0.721	0.293	< 0.005
SDNN	0.755	0.391	< 0.005
RMSSD	0.721	0.293	< 0.005
pNN50	0.820	0.675	0.075
SD1	0.721	0.293	< 0.005
SD2	0.769	0.457	< 0.01
SD1/SD2ratio	0.069	0.108	0.838

None of the additional independent variables were significantly different between the two LOS groups. P-value for age, gender, and prematurity when compared between the two LOS groups were 0.57, 0.64, 0.13 respectively. When comparing the LOS among respiratory support types (BiPAP, CPAP, HFNC, and NC), we did not see any significance (p = 0.128).

The aim of this study was to observe whether the relationship between PRV and HRV in patients with bronchiolitis was associated with PICU LOS. There was no statistically significant difference between length of stay and the type of non-invasive support required (i.e. BiPAP, CPAP, etc) nor with other independent variables such as age, gender, prematurity. Both groups had significant differences in PRV and HRV. PRV and HRV were more highly correlated in the short-stay group. PRV and HRV, while derived from the same cardiac cycle origin, are measuring different physiology. The loss of correlation between the two is identifiable early (first 12 hours of PICU stay) and may be able to be leveraged to predict longer ICU admission.

HRV has been used as a marker for parasympathetic and sympathetic activity for a long time and has been studied as a biomarker in a broad range of fields. PRV on the other hand has gained attention as a surrogate for HRV, as PPG signal is economically efficient and easier to obtain compared to ECG. However, Constant et al.²⁰ has shown that pediatric patients whose heart rate was controlled by an implanted pacemaker still had PRV despite lacking any HRV. Charlot et al.¹⁹ has shown exercise and posture modulate PRV and HRV differently. Yuda et al. has explained that it takes 6 steps for the electric potential of the ECG R-peak to be transferred to the detected PPG waveform: R-peak, cardiac contraction, generation of pressure impulse, propagation of the pressure wave, displacement of intravascular blood volume and alteration in light intensity as detected of photodiode. Each step can be modulated by autonomic activities, respiration, blood pressure and disease. Thus, we hypothesized PRV should be treated as an independent biomarker instead of a simple surrogate for HRV and focused on analyzing the difference between two signals in PICU patients with bronchiolitis.

PRV and HRV were compared using Student’s t-test and linear correlation. The results show that PRV and HRV are different in both groups, with significantly higher correlation in the short-stay group. These differences are maintained across both time-series (SDSD, SDNN, RMSSD) and non-linear parameters (SD1, SD2). On the other hand, Fisher Z-transformation shows the correlation between PRV and HRV in both time-series and non-linear parameters is significantly lower in the long-stay group. These results indicate that there was more modulation in transformation from electrical excitement of the R-peak in the ECG compared to the light-intensity of PPG in the long-stay group, which may reflect physiologic indications for their longer stay in PICU. We did not see any correlation between LOS and type of non-invasive respiratory support, suggesting that the degree of respiratory distress and need for higher level of non-invasive respiratory support is not as sensitive as the relationship of PRV and HRV. Additionally, the lack of correlation between non-invasive respiratory support and LOS highlights the need for more precise markers to predict LOS. Perhaps the unique relationship between PRV and HRV captures additional elements influencing LOS in the PICU and can be a useful tool to detect underlying illness-induced electromechanical changes and associate them with meaningful patient outcomes.

There are limitations in this study. First, the study has a relatively small sample size and is single-centered, which could affect the statistical power. A study with a larger sample size to validate our findings would be required. Although we looked at different types of respiratory support, we did not use any patient illness severity scores in comparison between these patients. Secondly, this study did not consider medication used. Future studies could observe the effect of medications such as how beta agonists and sedation (namely albuterol and dexmedetomidine), which alter heart rate and likely affects the correlation between PRV and HRV. Lastly, per standard recommendations for HRV analysis we used 5-minute nonoverlapping segments of waveform for analysis, whereas longer segments (i.e., 24-hour segments) may provide different insights ²⁴.

In conclusion, this study re-demonstrated that PRV and HRV are independent biometrics in PICU patients with bronchiolitis. As far as we are aware, we are the first to show that the correlation between the two (measured early) is lower in patients with longer length of stay, suggesting this can be a metric for LOS in PICU.

Author Contribution

SK and BW wrote the main manuscript text. SK performed data analysis. MM and YK helped with analysis. BW, AG, SP, SM, BR provided clinical analysis. DN, TA, IL prepared Figures and Tables. AG, YK and SP supervised the study. All authors provided critical feedback and helped shape the research, analysis and manuscript.

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request. Due to privacy and ethical restrictions, access to patient-related data is limited. Non-patient-related data and aggregated results are available for researchers who meet the criteria for access to confidential data.

Boyce TG, Mellen BG, Mitchel Jr EF, Wright PF, Griffin MR. Rates of hospitalization for respiratory syncytial virus infection among children in medicaid. The Journal of pediatrics. 2000;137(6):865–870.
Shay DK, Holman RC, Newman RD, Liu LL, Stout JW, Anderson LJ. Bronchiolitis-associated hospitalizations among US children, 1980–1996. Jama. 1999;282(15):1440–1446.
Mansbach JM, Piedra PA, Stevenson MD, et al. Prospective multicenter study of children with bronchiolitis requiring mechanical ventilation. Pediatrics. 2012;130(3):e492-e500.
Wang EE, Law BJ, Stephens D. Pediatric Investigators Collaborative Network on Infections in Canada (PICNIC) prospective study of risk factors and outcomes in patients hospitalized with respiratory syncytial viral lower respiratory tract infection. The Journal of pediatrics. 1995;126(2):212–219.
Willson DF, Horn SD, Hendley JO, Smout R, Gassaway J. Effect of practice variation on resource utilization in infants hospitalized for viral lower respiratory illness. Pediatrics. 2001;108(4):851–855.
Clark ME, Cummings BM, Kuhlthau K, Frassica N, Noviski N. Impact of pediatric intensive care unit admission on family financial status and productivity: a pilot study. Journal of intensive care medicine. 2019;34(11–12):973–977.
Khanna AK, Labeau SO, McCartney K, et al. International variation in length of stay in intensive care units and the impact of patient-to-nurse ratios. Intensive and Critical Care Nursing. 2022;72:103265.
Masarweh K, Gur M, Leiba R, et al. Factors predicting length of stay in bronchiolitis. Respir Med. Jan 2020;161:105824. doi:10.1016/j.rmed.2019.105824
Tan JPH, Beilharz JE, Vollmer-Conna U, Cvejic E. Heart rate variability as a marker of healthy ageing. International journal of cardiology. 2019;275:101–103.
Kim H-G, Cheon E-J, Bai D-S, Lee YH, Koo B-H. Stress and heart rate variability: a meta-analysis and review of the literature. Psychiatry investigation. 2018;15(3):235.
Villareal RP, Liu BC, Massumi A. Heart rate variability and cardiovascular mortality. Current atherosclerosis reports. 2002;4(2):120–127.
Castaldo R, Melillo P, Bracale U, Caserta M, Triassi M, Pecchia L. Acute mental stress assessment via short term HRV analysis in healthy adults: A systematic review with meta-analysis. Biomedical Signal Processing and Control. 2015;18:370–377.
de Castilho FM, Ribeiro ALP, Nobre V, Barros G, de Sousa MR. Heart rate variability as predictor of mortality in sepsis: A systematic review. PloS one. 2018;13(9):e0203487.
Rajendra Acharya U, Paul Joseph K, Kannathal N, Lim CM, Suri JS. Heart rate variability: a review. Medical and biological engineering and computing. 2006;44:1031–1051.
Lu G, Yang F, Taylor J, Stein JF. A comparison of photoplethysmography and ECG recording to analyse heart rate variability in healthy subjects. Journal of medical engineering & technology. 2009;33(8):634–641.
Selvaraj N, Jaryal A, Santhosh J, Deepak KK, Anand S. Assessment of heart rate variability derived from finger-tip photoplethysmography as compared to electrocardiography. Journal of medical engineering & technology. 2008;32(6):479–484.
Gil E, Orini M, Bailon R, Vergara JM, Mainardi L, Laguna P. Photoplethysmography pulse rate variability as a surrogate measurement of heart rate variability during non-stationary conditions. Physiological measurement. 2010;31(9):1271.
Weinschenk SW, Beise RD, Lorenz J. Heart rate variability (HRV) in deep breathing tests and 5-min short-term recordings: agreement of ear photoplethysmography with ECG measurements, in 343 subjects. European journal of applied physiology. 2016;116:1527–1535.
Charlot K, Cornolo J, Brugniaux JV, Richalet J-P, Pichon A. Interchangeability between heart rate and photoplethysmography variabilities during sympathetic stimulations. Physiological measurement. 2009;30(12):1357.
Constant I, Laude D, Murat I, Elghozi J-L. Pulse rate variability is not a surrogate for heart rate variability. Clinical Science. 1999;97(4):391–397.
Lin W-H, Wu D, Li C, Zhang H, Zhang Y-T. Comparison of heart rate variability from PPG with that from ECG. Springer; 2014:213–215.
Yuda E, Shibata M, Ogata Y, et al. Pulse rate variability: a new biomarker, not a surrogate for heart rate variability. Journal of physiological anthropology. 2020;39(1):1–4.
Bishop SM, Ercole A. Multi-scale peak and trough detection optimised for periodic and quasi-periodic neuroscience data. Springer; 2018:189–195.
Malik M. Heart rate variability: Standards of measurement, physiological interpretation, and clinical use: Task force of the European Society of Cardiology and the North American Society for Pacing and Electrophysiology. Annals of Noninvasive Electrocardiology. 1996;1(2):151–181.

No competing interests reported.

Download PDF

Reviews received at journal
05 Sep, 2024
Reviewers agreed at journal
28 Aug, 2024
Reviewers agreed at journal
16 Aug, 2024
Reviewers invited by journal
18 Jul, 2024
Editor assigned by journal
18 Jul, 2024
Editor invited by journal
11 Jul, 2024
Submission checks completed at journal
09 Jul, 2024
First submitted to journal
30 May, 2024

You are reading this latest preprint version

Pulse Rate Variability as a Predictor for Length of Stay for Patients with Bronchiolitis in the Pediatric Intensive Care Unit

Status:

Version 1

Abstract

Figures

Introduction

Method

Participants

Data Collection

Data processing and statistical analysis

Result

Discussion

Declarations

Author Contribution

Data Availability

References

Additional Declarations

Status:

Version 1